AT
Jude S [{
eh iv
IH
Yor
lod
CAT. FOR
PUBLIC HEALTH
1964
Nev)
: : noi-b ar
Binders fifle: JS Nat | Conter for Healdh stuhstics,
x Health €Xam . Slr vey
/
i.
a
R_UY
TTR AS AA
See inside of back cover
for catalog card.
Public Health Service Publication No. 1000-Series 11-No. 1
For sale by the Superintendent of Documents, vernment Printing Office
.C., 20402 - Price 30 cents
NATIONAL CENTER| Series 11
For HEALTH STATISTICS | Number 1
VITALand HEALTH STATISTICS
DATA FROM THE NATIONAL HEALTH SURVEY
Cycle | of the
Health Examination Survey:
Sample and Response
United States. 1960-1962
The first of a series of publications of the results
of the "first cycle" of the Health Examination Sur-
vey, describing the sampling procedures and esti-
mating techniques employed, the similarity be-
tween the sample and the universe it represents,
and the impact of nonresponse.
Washington, D.C. April 1964
U.S. DEPARTMENT OF
HEALTH, EDUCATION, AND WELFARE Public Health Service
Anthony J. Celebrezze Luther L. Terry
Secretary Surgeon General
CAT. [OR
PUBLIC HEALTH
NATIONAL CENTER FOR HEALTH STATISTICS
Forrest E. Linder, Ph.D., Director
Theodore D. Woolsey, Deputy Director
Oswald K. Sagen, Ph.D., Assistant Director
(for Professional Relations)
Walt R. Simmons, M.A., Statistical Advisor
Alice M. Waterhouse, M.D., Medical Advisor
James E. Kelly, D.D.S., Dental Advisor
Louis R. Stolcis, M.A., Executive Officer
DIVISION OF HEALTH EXAMINATION STATISTICS
Arthur J. McDowell, Chief
COOPERATION OF THE BUREAU OF THE CENSUS
In accordance with specifications established by the National
Health Survey, the Bureau of the Census, under a contractual
agreement, participated in the design and selection of the sample,
and carried out the first stage of the field interviewing and certain
parts of the statistical processing.
Public Health Service Publication No. 1000-Series 11-No, 1
Introduction -----
CONTENTS
The Target Population---eeeeeeeec cme m mmm mm mm
Selection of the Sample«cecceecc ec mmmmmc nem ——ne en ————
Sample Response-
The Sample Design and Estimating Techniques---------=----
Weighting and
Estimation-=--cce momo cece
The Sample Drawn and the Target Population----=-cceecee--
THE PEL SaITIDI@ i rt mm oe mm
Sample vs. Nonsample Persons-----=-eeommccmaamcacaaon
Sample Estimates and the Population----=-ceceoeeaaaa___
Respondents vs. NONTeapPONIenis =m mw mmm ww wm ww wenn mn ms
Reasons for Nonresponse-----==ee comm momcmmcmceeeeeee
Selected Demographic CharacteristiCS-=--==eceeocacaano
Medical Conditions Reported on Interview---—--ccoceeeaaeoo
Inquiry of Physicians -----eeeoeoom oo ______
Some Aspects of the Examination Process---==cecececccaano
DLAIAEA TABLES wos wr mms ps i 38590 0
Appendix I. Comparison of Examined and Nonexamined
Sample Persons
055
wn
© 00 NJ
Oo
18
20
23
26
36
SYMBOLS
Data not available-----=ccmmmmmmmmmmceeeee
Category not applicable---------cccucmemen
Quantity Zero=----===m-mmmmmmmmmmmoomoooo
Quantity more than O but less than 0.05-----
Figure does not meet standards of
reliability or precision-----=-c=cecmceme--
CYCLE | OF THE
HEALTH EXAMINATION SURVEY
SAMPLE AND RESPONSE
INTRODUCTION
The National Health Survey uses three meth-
ods for obtaining information about the health of
the U.S. population. The first is a household in-
terview in which persons are asked togive infor-
mation related to their health or to the health of
other household members. The second method is
the utilization of available health records. The
third method is direct examination, which the
Health Examination Survey administers by draw-
ing samples of the civilian, noninstitutional pop-
ulation of the United States and, by means of med-
ical and dental examinations and various tests and
measurements, undertakes to characterize the
population under study.
The overall plan of the Health Examination
Survey (HES) is to conduct successive, separate
cycles of examinations in specific segments of the
national population. The plan and initial program
of the HES have been described in another re-
port.! Data are collected by actual examinations
of, and tests upon, the individuals selected in the
sample. Such examinations and tests can yield
morbidity information unobtainable through other
programs of the National Health Survey (NHS).
They can provide information about diagnosed
conditions including those which persons may fail
to report or may be incapable of reporting in a
This report was prepared by Tavia Gordon and Henry W. Miller
of the Division of tiealth Examination Statistics.
survey based upon individual interviews. They
can also reveal previously undiagnosed, unattend-
ed, and nonmanifested chronic diseases. In addi-
tion to serving this primary purpose of determin-
ing prevalence of specified diseases, the exami-
nations are intended to obtain baseline data on
certain physical and physiological measurements.
Such measurement data on a defined population
are needed for understanding departures from
normal, as well as for assisting in planning cer-
tain specific programs dependent upon human
engineering information.
Another key characteristic of the Health
Examination Survey—one which is shared with
other National Health Survey programs—is the
use of a nationwide probability sample of the pop-
ulation. This makes it possible to obtain the de-
sired statistics efficiently and in such a manner
that the statistical reliability of results is de-
terminable. These factors, together with the fact
that the measurement processes are highly stand-
ardized and closely controlled, are essential in-
gredients of any survey that sets out to describe
the entire population of the United States on the
basis of a relatively small sample.
Furthermore, in the process of defining the
sample group, information about all sample per-
sons and their households is obtained prior to
examination, by means of a household interview.
The first cycle of the Health Examination
Survey was the examination of a sample of adults.
It was directed toward the collection of statistics
on the medically defined prevalence of certain
chronic diseases and of a particular set of dental
findings and physical and physiological measure-
ments. The probability sample consisted of 7,710
of all noninstitutional, civilian adults in the age
range 18-79 years in the United States. Altogether,
6,672 persons were examined during the period of
the Survey which began in October 1959 and was
completed in December 1962.
This report is the first of a series describ-
ing and evaluating the plan, execution, and find-
ings of the first Health Examination Survey. While
a number of previous publications have dealtwith
specific methodological investigations undertaken
as part of the survey, this series will describe
the results of the survey.
Although this initial report does notdeal with
the survey findings as such, it does consider the
frame against which the findings are to be pre-
sented, describes the sampling procedures, the
sample drawn and the group examined, and indi-
cates how the survey data will be converted into
estimates for the general population. The report
includes a few comparisons with the population
from which the sample was drawn. Thus itdemon-
strates the similarity of the sample and the pop-
ulation it represents with respect to a number of
characteristics not specifically controlled in the
sample design and explores the impact of non-
response on the survey findings.
Obviously, the success of a program as large
and complex as this one was possible because of
the efforts of many staff members as well as the
cooperation of a large number of outside individ-
uals and organizations, State and local medical,
dental, and osteopathic societies and health offi-
cers, and the staff of the Bureau of the Census.
THE TARGET POPULATION
The target population of the first cycle of the
Health Examination Survey consisted of all nonin-
stitutional, civilian adults in the age range 18-79
years in the United States. A more complete
specification of this target should include several
qualifications.
1. Alaska and Hawaii are not included.
2. The survey period is centered on October
_ 1961, but should be considered as the in-
terval 1960-62.
3. "Noninstitutional' is defined by excluding
residents of several types of places. In
particular, among persons out-of-scope
are inmates of correctional institutions,
resident hospitals, nursing homes, and
homes for the aged. Resident staff of these
places and persons in local jails are, how-
ever, included in the population surveyed.
4. Members of the crews of vessels are ex-
cluded from coverage.
5. Civilian personnel residing at a military
base and Indians on reservations are in-
cluded in the target.
6. Aliens are included if they have a place of
residence in the United States; U.S. cit-
izens residing overseas are excluded.
There are other categories of persons who
are included in at least a conceptual target, but
who because of the nature of the survey can
scarcely be considered to be effectively repre-
sented by the group of persons finally examined.
Worthy of mention in this category are two groups:
(1) persons who die or otherwise move out-of-
scope between the date of first contact at their
place of residence and the time at which they were
to have been examined (this was a very small
group), and (2) persons who are manifestly unable
to be examined, such as those gravely ill in short-
stay hospitals or elsewhere, or persons with
severe physical disabilities or impairments whose
conditions prevented their being transported to
the place of examination.
SELECTION OF THE SAMPLE
The size of the sample was keyed to the num-
bers necessary to yield reliable data on the con-
ditions to be studied. Actually, the determina-
tions of the size of sample and of the conditions
to be studied are interrelated and interdependent,
and a factor in these determinations was the num-
ber of examinations which could be accomplished
within 3 years. Also determining the sample size
were the budget and the statistical design and
structure of the examining process.
The selection process providedthatthe sam-
ple be stratified with respect to broad geographic
locations and the size of place of residence. For
purposes of the Health Examination Survey, the
1,900 primary sampling units (PSU's) which ac-
count for the 48 contiguous States and whichwere
originally designed for use in the Health Interview
Survey were grouped into 42 strata.” These strata
were formed so that they were as equal as possi-
ble with respect to population size, each with ap-
proximately 3.5 million persons aged 18-79 years
in 1950, and so that there were an approximately
equal number of strata in each of the five popula-
tion-density classes in each of the three geo-
graphic locations (table A).
Using a modified Goodman-Kish controlled-
selection technique one PSU was drawn from each
of the 42 strata. The sampling within PSU's was
carried out in several steps beginning witha ran-
dom selection of geographically clustered seg-
ments containing approximately six households.
A systematic selection was made usually of four
of these households. Within each selected house-
hold a roster was made of eligible adults (civilian,
noninstitutional persons aged 18-79 years). Every
alternate eligible adult within an interviewed
household was a sample person. The alternation
began with the first person in an ordered sequence
in one subsegment and with the second person in
the next subsegment. This alternation prevenis
bias in relation to the head of the household who
was usually listed first. The number of segments
selected varied somewhat from stand to stand, as
they were chosen to yield an expected 150-160
persons. The design was essentially self-weight-
ing, although operating efficiencies required some
variation in sampling rates among PSU's, and
occasionally among segments within a PSU.
The overall sampling process yielded an
initial listing of 9,035 households from 2,174 seg-
ments. Of these households, 1,221 were vacant,
belonging to persons having a usual residence
elsewhere or to members of the Armed Forces
on regular active duty. Another 163 households
consisted of units which were demolished, outside
segment boundaries, never intended for residen-
tial use, nonexistent, unoccupied and unfit for
human habitation, converted to business or stor-
age, or merged with another unit. Thus there
were 7,651 households which formed the sample
and which contained persons eligible for inter-
view. No interview was obtained for 125 or 1.6
percent of the sample households because of re-
fusal, because no one was home despite repeated
Table A. Distribution of strata by geographic location and population density: Health
Examination Survey, United States, 1960-62
Geographic location
Population density u.3,
total North
est South West
All strata----------cecemccccccccemeaao 42 14 14 14
Giant metropolitan areas====-===c-c--ccaaoooo 9 6 - 3
Other very large metropolitan areas---------- 6 2 2 2
Other standard metropolitan statistical
AreaS=mmm meme ccc 9 3 3 3
Other urban areas--------==--=---- m————memeeeeo 8 2 4 2
RUral=mm mm meen ee eee 10 1 5 4
NOTE: The States included in the Northeast are Maine, Vermont, New Hampshire, Massachusetts, Connecticut, Rhode Island, New York,
Pennsylvania, Ohio, and Michigan. The States included in the South are Delaware, Maryland, District of Columbia, West Virginia, Virginia,
North Carolina, South Carolina, Georgia, Florida, Kentucky, Tennessee, Alabama, Mississippi, Arkansas, Louisiana, Oklahoma, and Texas.
The States included in the West are Washington, Oregon, California, Idaho
, Nevada, Montana, Utah, Arizona, Wyoming, Colorado, New Mexico,
North Dakota, South Dakota, Nebraska, Kansas, Minnesota, Iowa, Missouri, Wisconsin, Illinois, and Indiana. This division of the United
States was especially made for the design of the HES sample.
calls, or because the sample person was absent
during the period of the survey. From the 7,526
households in which an interview was obtained,
15,038 eligible persons were identified from
which 7,710 were selected as sample persons.
(In three stands the within-household sampling
rate exceeded 1 in 2.)
It will be shown that the distribution of per-
sons in the sample closely parallels the distribu-
tion for the United States for a number of demo-
graphic characteristics. No major feature of the
U.S. population can be said to be seriously dis-
torted.
Obviously, this is not true for minor features.
In an area sample, such as was used for the
Health Examination Survey, a geographically
clustered population group (or in more abstract
terms, a localized health characteristic) might
be either completely missed or oversampled.
Since the population is probably closer to a con-
glomerate than to a uniform mixture, this limita-
tion in minor detail—a limitation which the Health
Examination Survey shares with all similar sur-
veys—is by no means trivial. Even where the
sample includes exactly the expected number of
persons from some specified subgroup, if this
group is numerically small it is quite out of the
question to describe any of its health character-
istics with assurance. A group of 100,000 would
have an expected representation of 6 examined
persons in the Health Examination Survey, and
even for larger groups, suchas American Indians
or persons 75-79 years of age, the sample size
is so small that statements of findings for the
groups must be limited both in number and in
confidence.
SAMPLE RESPONSE
Of the 15,038 eligible persons in sample
households, 7,710 sample persons were identified
and 6,672 were examined. The examination or re-
sponse rate based on these figures is 86.5 per-
cent. Since the sample was based on households,
however, another type of rate should also be con-
sidered. This rate, which might be termed the
"nmet'' examination or response rate, as contrasted
to a gross rate, provides an adjustment for pre-
sumed eligible sample persons in the 125 non-
interviewed households. The adjustment assumes
that the ratio of sample persons in these house-
holds is the same as for the interviewed house-
holds. Using this adjustment the number of sam-
ple persons would increase by 128 to a total of
7,838. The ''met' examination or response rate
would be 85.1 percent (table 8).
Whichever one of these percentages is con-
sidered, 86.5 or 85.1, the fact is that nonresponse
in this cycle of the Health Examination Survey
program was remarkably low. At the outset of the
planning for this program it was feared that per-
haps as much as one-third of the sample selected
might end up as "unexamined." Like other pro-
grams of the National Health Survey, this is based
on legislation which specifies that the required
information will be secured ''on a noncompulsory
basis." A number of voluntary surveys involving
health examinations had been made in local areas
in the United States, > 4 5and despite intensive
persuasion, only about two-thirds of the sample
had been examined in each of them. The fact that
the first cycle program has only about one-half
the percentage of nonrespondents of the earlier
surveys greatly minimizes the possibility of
bias in the survey results due to nonresponse.
The following describes briefly the measures
adopted to ensure maximum cooperation. Prior
to the beginning of the survey proper, methodolog-
ical studies into the motivations and attitudes in-
volving willingness to participate in a health
examination survey were made. 67 The survey
design incorporated some techniques from these
studies, some from earlier surveys, and some
from the pilot tests made of the Health Examina-
tion Survey plans. Thus, the request to consent to
the health examination was made only directly of
the sample person. An attempt was made toiden-
tify the possible ''moncooperator'' early and to
handle his case in an individualized way. An effort
was made to minimize personal inconvenience to
the examinee in loss of time, amount of travel,
and the nature of the examination. The findings of
the examination were made available to the exami-
nee's physician (or in the case of dental findings
to his dentist) when the examinee wanted this done.
An attempt was made to allay the fears and doubts
that might stem from lack of knowledge about just
what was involved by providing information
through various means—pamphlets, newspaper
stories, etc. Above all, a resourceful, skilled, and
highly motivated staff persisted in their efforts to
explain the program to all whowere inanyway in-
volved, doing this out of the conviction that, for
the most part, persons who really understood the
program would be favorably inclined toward it.
THE SAMPLE DESIGN AND
ESTIMATING TECHNIQUE
The essence of probability sampling can be
expressed rather simply. Suppose there exists a
population of N individuals in which x; is the value
of a specified measurement for the = individual.
The average of these measurements for all per-
sons in the population might be defined as X. A
probability survey design in this universe might
be described as a process in which:
1. A sample or subset of n of the N individ-
uals is drawn in such a fashion that every
individual in the population has a known
nonzero probability of inclusion in the
sample.
SH. : i
2. For each i individual included in the sam-
ple, the measurement X; is obtained.
’
3. An estimating equation is adopted which
converts the measurement xX; into anesti-
mate X' of the population Xx.
Probability designs have many features of
which two are of special interest in the present
context. The first of these is that a good design
will be technically unbiased or nearly so; there-
fore, over repeated trials of the survey the aver-
age value of the estimate x' would be equal to the
true value of X.
The second attribute is that procedures exist
whereby the variance of the estimate X' can be
calculated. The variance permits the calculation
of an interval around the estimated mean within
which the true mean X lies with a given probabil-
ity. It is the yardstick for determining the pre-
cision of the estimate x’.
The HES sampling method is an unbiased one.
The variance of a statistic depends notonlyon the
design, but on the statistic itself; the variance is
higher for measurements which are highly vari-
able from one individual to another, and lower for
measurements which are less variable in the
population. HES publications will include esti-
mates of variance for principal statistics which
are presented.
Weighting and Estimation
In this survey the selection of an individual
sample person is the outcome of four sampling
procedures. In order to convert data froma sam-
ple person into estimates of population param-
eters, therefore, weights relating to the selec-
tion of a sample person by each of the four pro-
cedures must be applied. Briefly these weights
are the reciprocals of the probability of selection
of:
1. The PSU from the stratum from which it
was drawn.
2. The segments from chosen PSU's.
3. The households in a segment from the total
number of households within that segment.
4. The sample person from the eligible adults
in the household.
The master design is essentially self-weight-
ing with respect to persons; that is, each person
has about the same probability of falling into the
sample. In addition, the HES employs three tech-
niques which enhance the representativeness of
the survey. Of these, two are technical adjust-
ments which leave the process still unbiased, but
which increase precision by bringing survey re-
sults into closer alignment with the target popu-
lation with respect to geography, population-den-
sity, age, and sex.
One is called a first-stage ratio adjustment.
Computationally, using population controls from
the 1960 census, this adjustment multiplies every
observation by such a factor as would make the
42 first-stage primary sampling units reflect
exactly the total 1960 population, if the PSU's had
been enumerated completely, for each of eight
geographic sectors of the Nation. The required
multiplication factors are shown in table B.
The second of these two adjustments is a
poststratification by age and sex. In it, a multi-
plier for each observation is utilized which has
the effect of (a) obtaining most of the advantage
Table B. First-stage ratio adjustment factors: Health Examination Survey, United
States, 1960-62
Nonself-representing areas
Self-
Geographic location representing
areas! Standard metropolitan |Other
statistical areas areas
NortheaSt=========-cccom emcee meme m 1.00 0.97 0.98
South=====ecceccmc ccc c ccc c ccc cc ccc — cc ec ———— - 1.09 0.88
WeStrmmmmmmrr mmc c rem — ec ———————————————————— 1.00 0.88 1.04
INew York, Chicago, Los Angeles. Philadelphia, Detroit, and I30ston sample areas represented only themselves.
Table C. Second-stage age-sex adjustment
factors: Health Examination Survey,
United States, 1960-62
Multiplier factor
Age
Male Female
18-24 years---==-=--- 1.153 1.05
25-34 years-=-======-- 0,97 0.96
35-44 years-====-===-= 1.00 0.97
45-54 years-=--===-==-- 1.16 0.95
55-64 years--======-- 1.08 1.11
65-79 years-=----=--=--- 1.14 1.25
which would have been attained if the original
sample had been drawn from a population strati-
fied by age and sex; and (b) making the final sam-
ple estimate of population agree precisely within-
dependent population controls prepared by the
Bureau of the Census for October 1,1961,in each
of 12 age-sex classes. These second-stage ad-
justment factors are shown in table C.
The third adjustment is entirely different.
Its function is to minimize the impact of nonre-
sponse on final estimates. Unlike the other two
adjustments, it can have a biasing effect, although
the intention was to reduce bias that might have
been introduced because 13.5 percent of the sam-
ple persons were not examined.
It must be recognized that when data for a
specified person are missing there is no tech-
nique which can completely remove the conse-
quent difficulty. All techniques which have been
or might be used involve either explicit or im-
plicit imputation for the missing data. From the
experience of previous surveys involving health
examinations, it had been feared that the nonre-
sponse rate in the HES might be as great as 35
percent. Much energy was expended in cutting
nonresponse to 13.5 percent of the designated
sample persons. Residual nonresponse was treat-
ed by imputing tononrespondents the characteris-
tics of "'similar'' respondents. The "similar" re-
spondents were those in the same cell. The choice
of cell boundaries was guided by three principles:
1. The nonresponse rates should be different
for different cells.
2. Key substantive statistics might vary from
cell to cell.
3. The cells should be large enough to pre-
vent the frequent occurrence of cells with
very few respondents.
These principles resulted in the choice of bound-
aries for seven age-sex groups within each stand,
yielding 294 separate cells with average mem-
bership of about 25 sample persons each. The ad-
justment resulted in substituting for the nonre-
spondent the characteristics of respondents of a
similar age and sex in the same general com-
munity. The mechanism of the adjustment was to
multiply data of respondents by factors which
made total persons in the cells equal to the de-
signed number. Multipliers in the cells were dis-
tributed as shown in table D.
Table D. Nonresponse adjustment factors:
Health Examination Survey, United States,
1960-62
Number of
: 11s i
Size of multiplying factor py ys
was used
All cells-====mmmmec-- 294
1.00-1.10--====mmmemm————— 127
1.11-1,20-===mmmmmmc meme 63
1.21-1.40=====mmmmmc cme 66
1.41-1.50====mmmmmmccc em 12
1.51-2.00=====mmmmemmceme 23
2,01-2.10====mmmmmm mmm mm 3
THE SAMPLE DRAWN AND THE
TARGET POPULATION
From the structural point of view, bypassing
questions relating to the validity of the measure-
ment itself, the degree of confidence which one
can place in a particular survey statistic from a
probability design is most simply measured by
its standard error. Standard errors will be pub-
lished in future reports for the principal sub-
stantive findings of the Health Examination Survey.
It cannot be safely assumed that any sample
design will be infallible in design or execution.
There are certain to be some mistakes made in
carrying out a large-scale survey. In addition,
any given probability sample can be counted on to
differ from expected values, slightly for large or
controlled sectors of the population and more for
small segments or for characteristics not con-
trolled in selection. Because of these facts it is
useful, in assuring that the total survey plan work
well, to look at some comparisons between the
population that actually was involved and that
which might have been 'expected' from the de-
sign.
These comparisons bear on two issues. The
first is the execution of the sampling process. A
close correspondence between the HES sample and
other sources makes it likely that the sampling
scheme was faithfully executed. These compari-
sons also bear on questions of exposition. While
it is always possible (for known characteristics)
either to control the estimating process or to pre-
sent the data in sufficient analytic detail to com-
pensate for recognized deviations from the uni-
verse, it is sometimes awkward to do so. Obvi-
ously, exposition of the data becomes much sim-
pler, and is also more trustworthy, if the sample
is essentially the same as the universe for known
characteristics.
One important warning is worth repeating.
Comparisons of the kind just described in this
section cannot be used to determine whether the
design is sound or what its precision may be. If
the comparisons are close in one or ina dozen re-
spects, it is not proof that the sample is efficient.
Nor is it proof that the sample is inefficient if in
some respects the comparisons reflect large dif-
ferences. On the other hand, the procedure would
be suspect if there were little semblance between
the observed and expected results. Itis suggested
that the evidence presented in this report en-
courages confidence in the survey results since
the sample does exhibitdemographic characteris-
tics quite similar to those of the total U.S. popu-
lation.
To this, obviously, must be added the warning
that a sample faithful for demographic represen-
tation may not provide a faithful representation
of the prevalence of chronic disease or of the
physical and physiological attributes that were
the object of this cycle of the Health Examination
Survey. However, future reports of findings will
include comparisons with findings from other
population surveys, as well as giving the sampling
errors for the published statistics.
The PSU Sample
As mentioned previously, the 1,900 primary
sampling units constituting the United States
were grouped into 42 strata with further control
by three geographic locations and five population-
density classes. One primary sampling unit was
then selected from each stratum with a probabil-
ity proportionate to its 1950 population.
If each primary sampling unit is weighted by
the reciprocal of its probability of selection from
a stratum and adjusted by the first-stage adjust-
ment, then the sum of the 42 units becomes an es-
timate of the total 1960 population of the United
States. If the selection of the PSU's is not faulty,
then such weighted and adjusted distributions of
various demographic characteristics of the PSU's
should approximate those of the United States,
with some allowance for sampling variability.
A brief comparison between the weighted dis-
tributions of the PSU's and the United States with
respect to several selected demographic charac-
teristics follows. While there are many character-
istics that might be used for comparative pur-
poses, these are considered most important in
judging the success of this first stage in the sam-
pling and estimating procedure.
The sources of the statistics used for this
comparison were the final reports of 1960 Census
of Population (primarily the Population Census—
B, C, and D series). Data for the New England
stands of Boston and Providence where the PSU's
were made up of minor civil divisions were ob-
tained from tract reports of the U.S. Censuses of
Population and Housing. The data include the in-
stitutional population and members of the Armed
Forces.
The race distribution presented in table 1
shows that the sample of PSU's would yield a
population which was 87.5 percent white; the Cen-
sus figure was 89.9 percent. A part of this dif-
ference may be attributed to the selection of sev-
eral rural areas in the South withunusually heavy
Negro populations and an area in the West with a
high proportion of American Indians. Regarding
age, the estimates obtained from the weighted
PSU's differ slightly from those of the Census,
with 57.1 percent of the population in the 18-44
year age group as compared with 55.4 percent in
the Census. Comparison of the distributions of
four social characteristics—marital status, years
of school completed, type of occupation, and fam-
ily income —did not reveal any differences of con-
sequence.
All in all, according to these characteristics,
the population of the PSU's did not differ signifi-
cantly from that of the total U.S. population.
Sample vs. Nonsample Persons
As previously mentioned, 15,038 adults were
identified as eligible for inclusion in the Health
Examination Survey panel. Except in three stands
(Philadelphia, Pa., Valdosta, Ga., and Winslow,
Ariz., where in some or all of the segments every
eligible adult was sampled), every alternate eli-
gible person became a potential sample person.
Table 2 provides comparative distributions of
various demographic characteristics of sample
and nonsample persons from the stands where
alternate selection was used. The most notable
difference occurred in the sex distribution; slight-
ly more sample females (and consequently less
sample males) were selected than had been ex-
pected. If the expected number of females is the
product of the proportion of females in the eligi-
ble adult population times the number of sample
persons, then there were approximately 100 more
females selected through this sampling procedure
than had been expected. No other differences
which might indicate the introduction of bias
through this step in the sampling process were
observed.
Sample Estimates and the Population
Tables 3, 4, 5, and 6 present comparisons of
several sample estimates with the target popula-
tion. These estimates will not appear in future
reports, but are a special set compiled from data
for all sample persons, whether examined or not.
They are intended to reflect what a particular es-
timating procedure would have yielded from the
sample cases if there had been no nonresponse.
The distribution of the sample estimates by age
and sex in table 3 is derived by applying to the
sample persons the four sample selection weights
and the first-stage ratio adjustment previously
described. The estimating procedure used for
tables 4-6 includes all sampling weights, the
first-stage ratio adjustment, and a poststratifica-
tion to control on age and sex.
There is a minor difference between the es-
timated population and the target population in
the proportion of white persons, 88.0 percent as
compared with 89.6 percent (table 3). A similar
difference has been noted previously in a com-
parison of race differences between the weighted
PSU's and the United States. In both instances,
this difference is due to the chance selection of
several PSU's atypical with respect to race from
the strata from which they were drawn. One
PSU is responsible for the fact that the estimated
proportion of the "'other' race in the sample pop-
ulation is almost three times as great as that of
the target.
The age-sex distributions of the two popula-
tions differed very little. The largest observed
difference occurred in the 65-79 year age group
where the estimated percent in the sample popu-
lation was 11.1 compared with 12.6 percent in the
target. While even this difference is trivial, the
second-stage poststratification adjustment will
compensate for it and like differences.
Similar distributions of marital status, years
of school completed, and occupation of employed
presented in tables 4, 5, and 6, respectively, re-
vealed only slight differences between the esti-
mated and target populations. These may be at-
tributed to the age groups upon which the distri-
butions of the target population were based. The
Census distributions for marital status and years
of school completed were based on a population
18 years of age and over and the distribution of
occupation based on an age group 14 years of age
and over. The HES sample did not include any
persons under 18 or over 79 years of age.
RESPONDENTS VS.
NONRESPONDENTS
In any health examination survey, after the
sample is identified and the sample persons are
requested to participate in the examination, the
survey meets one of its more severe problems.
Usually, a sizable number of sample persons will
not participate in the examination. If the nonpar-
ticipants are essentially identical with those who
do participate, the loss in numbers is trivial in
consequence, except as it reduces the effective
sample size. If participants and nonparticipants
differ, however, the problems resulting from non-
response may be quite serious.
Reasons for Nonresponse
When examinations were completed atagiven
location, the field representatives of the Health
Examination Survey responsible for contacting
sample persons attempted todetermine the reason
each nonrespondent failed to come in for exami-
nation. For most of the 1,038 persons not exam-
ined, "no special reason'' was designated. There
were 94 cases in which medical reasons were
adduced, including hospitalization, disability, and
3 deaths, While examinations were being conducted,
61 persons were out of the area during all or part
of the period, and 38 were either very difficult to
contact or inaccessible for religious or other
reasons. There were examinations of six persons
which were so grossly incomplete that these per-
sons were considered nonexamined. Most of the
persons not examined were simply unwilling to
be examined or to make the personal arrange-
ments necessary to come in.
Altogether, 13.5 percent of the identified sam-
ple were not examined. This response loss
carries two serious risks. It may distort the
demographic frame against which the examination
findings are referred. Such a distortion would
arise only if the response level varied according
to the demographic class, and the effect of such
differential response tends to be minimized by
specificity in analysis. More serious and more
difficult to evaluate is the possible distortion of
the actual examination findings; that is, the pos-
sible difference between examined and nonexam-
ined persons with respect to the characteristics
under examination. It is difficult to remedy de-
fects of this kind. Both aspects of nonresponse
will be discussed: its possible effect on the frame
and on the findings. (If the larger part of the dis-
cussion is concerned with the frame rather than
the findings, it is not because it is the more im-
portant subject but because more and better in-
formation is available on the demographic than on
the morbidity characteristics of the nonrespond-
ent group.)
Comparison of the examined and nonexamined
persons is made possible by the fact that, with
few exceptions, all sample persons were included
in a regular household interview survey. The in-
terview procedure and the forms used were es-
sentially those of the Health Interview Survey of
the National Health Survey. In fact, the method
used for selecting the sample persons for exami-
nation included the conducting of a household in-
terview at each residence falling in the sample
area. As already noted, the household interview
produced not only a large amount of demographic
information about each sample person but also
morbidity information of various kinds. It should
be realized that since the informant was not
necessarily the sample person, information could
be collected even for nonexamined persons absent
from the household during the examining period.
However, a uniform household interview is
not assurance that all sample persons are equally
willing or able to supply the desired information.
It is possible that the household interview infor-
mation for nonexamined persons is, in some re-
spects, not so good as for the examined. This
seems likely on a priori grounds, since unwilling-
ness to cooperate ought to manifest itself during
the household interview, just as it did when the
person was asked to make an appointment for
examination.
But we need not rely on presumption. For
most items of the household interview, the pro-
portion of missing or recognizably defective re-
plies was greater for nonexamined than examined
persons. For some items, the difference was
trivial; for others it was not (table 7). The fol-
lowing are illustrative of items where the per-
centage of missing information is substantially
greater for nonexamined than examined persons:
Examined Nonexamined
Class of worker-- 2.8 5.4
Education-------- 2.2 5.4
Family income-- - 9.0 17.1
The poorer quality of information for non-
examined persons is not to any marked degree
due to the particular person supplying the infor-
mation. More specifically, the respondent to the
household interview was equally likely to be the
sample person himself whether or not the person
was subsequently examined. It is true that the
likelihood of having the sample person as in-
10
formant differs greatly by sex and, for men
especially, varies with age, but even if the com-
parison is made on an age-sex-specific basis,
there is little difference between the examined
and the nonexamined groups in the percentage
self-respondent to the interview.
The essential difference between examined
and nonexamined persons appears to be simply a
willingness to cooperate. Of the reasons adduced
for nonresponse by field representatives of the
Health Examination Survey, ''refusal, unqualified"
accounted for 81 percent of all nonrespondents.
As had been expected, uncooperativeness was
manifested, in those persons who subsequently
joined the nonexamined group, as the interview
progressed toward an actual request for exami-
nation.
At the end of the household interview each
respondent was requested by the interviewer to
sign an authorization permitting access to his
medical records. In the households of nonexamined
persons, one out of four of the respondents re-
fused to sign. The refusal rate for the examined
group was only 5 percent. Failure to sign the re-
quested authorization was, in fact, one of the best
indicators of ultimate nonresponse, even more
dependable than failure to make an examination
appointment at the time of the initial interview.
Selected Demographic Characteristics
Place. —The largest variability in response
level was that associated with place. The percent
examined at the various stands ranged from 65.5
at Philadelphia, Pa., to 97.8 at Eufala, Ala.
Philadelphia was the first area surveyed and itis
possible that better results would have been at-
tained at this stand if it had been scheduled later,
when the staff had acquired more experience in
obtaining cooperation. On the other hand, Phila-
delphia is one of the places where cooperation
tended to be below average. To a large extent,
response differentials fell into patterns by popu-
lation size and location. The larger the place the
poorer the response, ranging down from 92.0 per-
cent in rural areas to 77.7 in giant metropolitan
areas (table E). This was anticipated from the
results of an earlier methodological study.® The
Northeast, with a heavier concentration of large
Table E.
of sample
sity:
States, 1960-62
Number of
places
and percent
examined by population den-
Health Examination Survey, United
places than the South and West, had an overall
poorer response rate—81.8 compared with 89.8
and 88.2 percent for the South and West—but for
places of the same size there was, in general,
little difference in response from one location to
: : Number of | Percent :
Population density places examined another (fig. 1, table F).
The source of this differential response by
place, as distinguished from its possible conse-
All scrote 2 86.3 quences, can only be guessed at from the data
Giant metropolitan collected in this survey, but it is Possible that it
AL CHS wise mmm ies mmm 9 99.7 reflects chiefly some cultural attribute of the
Other very large local populations. It definitely does notarise from
in wea 6 85.8 special distributions of other demographic vari-
BEBE mira i 9 87.5 ables defined by the survey, such as age, race,
Other urban areas--- 8 90.7 or sex, and for this reason the response level of
Rural aveas-=-==mwn= 10 92.0 places has been made an analytical variable in
some of the later discussion. The 42 stands have
100 100
®
@ = HES STAND ¥ °
ee * $ |
$ .
§ . °
eS
90 I~ ° —{ 90
o ® ° °
w ® ° ® ®
®
°
g 80 — . * — 80
5 °
c ¢
Bk . .
¥
°
70 |- 70
f= ® ef
04 . L| 1 11 | ok L] 1 11 i 4 11 1 1 1 50
oC ] ] 1] 1 ] I TI TT Lo] rr ] To
Nort |souTh| west [NORTH south | west "nor |souTh| west || NORTH | south | west or south | west
Giant Metropolitan Areas
Other Large Metropoli-
Area:
tan
Small Metropolitan Areas
"Other Urban Areas Rural
Figure 1. Percent examined by stand according to geographic location and size of place.
1
Table F. Percent of sample examined by
geographic location and population den-
sity: Health Examination Survey, United
States, 1960-62
Geographic location
Population density
North-
east South | West
All strata---- 81.8 | 89.8 [88.2
Giant metropolitan
areag-===-========-- 77.5 - | 78.9
Other very large
metropolitan
areas--===========- 87.4 | 83.5 86.8
Small metropolitan
areas=-=-========== 84.2 | 87.6 [90.6
Other urban areas--- 88.8 | 90.3 | 94.0
Rural places-------- 79.4 | 93.8 | 93.0
been divided into five response groups defined as
follows:
Response Number of stands Porest
group ——— examined
1 8 94-98
2 12 89-93
3 10 84-88
4 6 79-83
5 6 66-78
While there is a strong correlation between
the population size of a place and its response
level, so that all places in group 1 are either
rural or "other urban' places, and all places in
group 5 are giant metropolitan areas, the corre-
lation is by no means complete. Therefore it is
preferable to consider response level and popu-
lation size as different analytical axes.
Up to this point, the 42 stands have been con-
sidered as demographic entities. In some re-
spects, of course, they are not. For example,
persons living in a standard metropolitan statis-
tical area may reside within the central city or
12
Table G. Percent of sample examined by
age and sex: Health Examination Survey,
United States, 1960-62
Age Total || Male | Female
Total, 18-79
years======= 86.5 || 88.3 85.0
18-24 years-====---- 90.2 || 91.7 89.0
25-34 years-==-==--= 89.5 | 92.0 87.5
35-44 years-------- 88.7 90.6 87.0
45-54 years===-==--=- 86.7 87 +2 86.3
55-64 years=====-==-- 81.1 || 81.5 80.7
65-74 years====-==- 80.5 || 85.2 76.7
75-79 years-==-=--=-- 74.3 || 80.0 69.3
100
Male
20
rr
bwin
2
2 ;
a .,
“.,
5 \,
2 ,
w S,
a \
.
70 1
60
ol 4 L111
0 20 30 40 50 60 70 80
AGE (in years)
Figure 2. Percent examined by age and sex.
in the suburbs, and the socioeconomic differences
between these two groups are often quite con-
siderable. However, their response rates were
quite similar, with 82.7 percent for persons in
the central cities and 83.7 percent for persons
outside the central city. Obviously, with an overall
rate of 91.6 percent, places not located in the
standard metropolitan areas have noticeably high-
er response levels.
Similarly, there are substantial socioeconom-
ic differences between persons living on farms
and rural residents not living on farms. Here the
response rates differed somewhat, with 93.7 per-
cent for rural farm and 89.0 percent for rural
nonfarm. The response rate for urban places
averaged 84.6 percent, though it is worth noting
that for the smallest urban places (2,500-9,999
population) it was decidedly higher —92.0 percent.
Age and sex.—The examination rate was 88.3
percent for men and 85.0 percent for women. It
was higher for men in every age group, but at
ages under 65 years the sex differential was
trivial. After 65 years of age, women were sub-
stantially less willing than men to come in for
examination (table G, fig. 2). By age, the response
rate was highest for the youngest persons and
diminished with increasing age. In the age group
18-24 years, 90.2 percent were examined. In the
age group 75-79 years, the examination rate was
only 74.3 percent, or 80.0 percent for men and
69.3 for women.
Race.—The response rate was higher in the
nonwhite population than in the white (table H,
fig. 3). Overall, it was 85.8 percent for the white
population and 91.4 percent for the Negro. The
difference was not due to a confounding of place
with race; there was no special concentration of
Negroes in places of high response. In fact, re-
sponse was higher for the Negro population in
most age groups and at most places. There was
little difference in response rates between Negro
and other nonwhite races (chiefly Indian).
Other demographic variables. — For a number
of other demographic variables, there was some
slight variation noted in the percent examined.
The response rate varied slightly with marital
status, from 81.3 percent for widowed persons to
87.2 percent for married and 88.3 percent for
separated persons (table I). It varied slightly
Table
H. Percent of
sample examined of
the white and Negro populations, by age
and sex: Health Examination Survey,
United States, 1960-62
White Negro
Age
Male | Female | Male | Female
Total, 18-79
years=----- 87.8 84.1 91.8 90.7
18-24 years-- | 91.2 87.8 96.2 94.0
25-34 years-- | 91.5 86.5 93.4 92.9
35-44 years-- | 91.2 86.5] 86.6 91.0
45-54 years-- | 86.2 85.71 9%.1 89..9
55-64 years-- | 79.5 80.2 94.7 8545
65-74 years-- | 85.0 75.6 90.0 85.7
75-79 years-- | 80.2 66,3 77.8 90.0
100
90
80
PERCENT EXAMINED
70
0
White
female
White Negro
male female
91.8
907
87.8
| :
Figure 3. Percent examined by race and sex.
Table I. Percent of sample examined by
marital status and sex: Health Examina-
tion Survey, United States, 1960-62
Table K. Percent of sample examined by
family income and sex: Health Examina-
tion Survey, United States, 1960-62
Marital status Total Male | Female Family income Total Male | Female
Single-r==m==mu=-==-- 85.3 85.6 84.9 Under $2,000------- 89.3 90.6 88.3
Married!----------- 87.2 || 89.2| 85.5 $2,000-3,999------- 86.6 | 87.7 85.8
Separated=======-=-=-- 88.3 87.0 89.4 $4,000-6,999=-====~ 88.6 90.1 87.1
Widowed=======m===- 831.3 84.4 80.8 $7,000+====mmmmmum 86.2 88.5 84.1
Divorced-========== 86.6 84.4 87.9 Unknown--========-- 77.2 79.6 15.5
IExclusive of separated.
Table J. Percent of sample examined by
years of school completed and sex:
Health Examination Survey, United States,
1960-62
Table L. Percent of sample examined by
usual activity status and sex: Health
Examination Survey, United States, 1960-
Beans Of epal Total Male | Female Usual ootiviey Total Male | Female
Under 5 years------ 89.2 90.5 88.0 Usually working---- 88.3 89.2 86.4
5-8 years-=======--- 85.2 87.9 82.8 Keeping house------ 84.8 ite 84.8
9-12 years--=--==-=-=-- 86.6 | 89.1 84:8 Retired--=====mm==- 81.8 84.3 61.53
13+ years--===-===-- 89.5 | 89.4 89.7 Other---===m=mmm——- 86.1 86.6 85.1
Unknown---=-=====-=- 72.1 71.7 12.5
Table M. Percent of sample examined by occupation and sex: Health Examination Survey,
United States, 1960-62
Occupation Total Male Female
Professional, technical, and kindred workers------------ 89.6 89.1 90.2
Farmers and farm managers---------==--=--cecccc-coe-—--- 92.4 92.6 90.0
Managers, clerical, and sales workers-------=-=---------- 86.5 88.0 84.9
Craftsmen, foremen, and kindred workers-------=--------- 89.5 89.7 82.4
Operatives and kindred workers===----=-e-c---ccccccao--- 88.8 90.3 85.7
Private household and service workers--=---=---cccee-ac--- 90.5 91:3 90.1
Laborers (except mine) =----------cccccccceccmanamannoa—- 93.3 92.9 95.1
Occupation not reported===--=-mmm--eceeecccccae eee eee————— 83.2 81.6 83.6
NOTE: Omits stands 01 and 02.
14
with the number of years of schoolcompleted,
from 85.2 to 89.5 percent (table J), and with in-
come, from 86.2 to 89.3 percent (table K). In
neither case was there any definite pattern. In
terms of usual (economic) activity, the response
rate ranged from 81.8 percent for retired persons
to 88.3 percent for usually working persons
(table L). Similarly, there was a slight variation
in response rates by occupation (table M).
Crude response rates, however, are insuffi-
cient for describing which of these differentials
are meaningful and which are not. Itis well known
that most demographic features, such as educa-
tion, income, and occupation, vary with age and
sex. Since variation in response is generally more
marked by age and sex than by these other demo-
graphic characteristics and since age and sex are
well defined for the entire sample, it is appro-
priate to consider response rates for these vari-
ables after controlling by age and sex.
To do this, the following procedure is used.
The number of examined persons with some
specified characteristic, such as the number of
married persons, is counted. This number is com-
pared with the number of examined persons ex-
pected to have this characteristic. The expected
number for an age-sex group will be considered
to be PX, where p; is the proportion of the sam-
.th .
ple in the it age-sex group that was examined
and xX; is the number of sample persons in that
age-sex group having the specified characteristic.
The expected number for several age groups is
the sum of the expected numbers for these groups.
The sampling characteristics of this statistic are
discussed in Appendix I. Those cases in which the
person had not been adequately characterized for
a specific variable are omitted from the total, so
that totals differ from one characteristic toanoth-
er. As already noted, the percentage of suchcases
is higher in the nonexamined than in the examined
group for all demographic variables, and there is
little point in rediscovering that fact by an addi-
tional route. On the other hand, the following dis-
cussion may be incorrect if the persons for whom
there is no information on the specific variable
differ from those persons who are well char-
acterized on that variable.
|
+10 +20 +30
EXCESS OF ACTUAL OVER EXPECTED NUMBER OF EXAMINEES
Figure 4. Excess of actual over expected number of examinees
by marital status and sex.
When this approach is used, it is seen that
response did vary somewhat by marital status
and by education (figs. 4 and 5) and this varia-
tion is statistically significant. There were slight-
ly more examined than had been expected in the
married group and slightly less in the never-
married group. In other words, married persons
were more cooperative than single persons;
whereas response differed little from expectation
for widowed, divorced, or separated persons. It
should be noted that the differential response that
was observed for widowed persons when crude re-
sponse rates were calculated has disappeared for
age-adjustment, and, in fact, the difference in
crude rates arose from the greater likelihood of
being widowed at older than at younger ages. Per-
sons with no education or who had completed no
more than the first four grades of elementary
15
L
9-12 years
y 1
Less than 5 years
13 years or more
| | |
-40 -30
EXCESS OF ACTUAL OVER EXPECTED NUMBER OF EXAMINEES
-20
-10
0 +10 +20 +30
+40
Figure 5. Excess of actual over expected number of examinees
by years of school completed and sex.
school were more likely to come in for examina-
tion than expected, whereas persons who had gone
to high school but not to college were less likely
to do so.
Variation in response by income, while sta-
tistically significant, followed no clear pattern.
Response was greater than had been expected for
persons from families with incomes less than
$2,000 per year and less than had been expected
for persons from families with annual incomes of
$7,000 or more. Variation in response by kind of
usual (economic) activity, by (economic) activity
in the 2 weeks preceding the interview, and by
occupation, was within the range of chance fluc-
tuation.
Similar calculations were performed for sev-
eral characteristics for each of the five response
groups of places. For family income and medical
authorization, response varied from expected
values more than chance when all 42 stands were
considered together. However, no consistent pat-
tern for family income carried over from one
group of stands to another (fig. 6). Ofall the vari-
ables investigated for each of the five response
EXAMINATION
RATE (%)
TOTAL
94-98
89-93
84-88
79-83
66-78
INCOME
UNDER $2,000
INCOME INCOME INCOME $ 7,000 EXAMINATION
$2,000-3,999 $4,000-6,999 AND OVER RATE (%)
| 1 T T
g 94-98
%,
%;
bY |
7 84-88
/ 79-83
f 66-78
er 0
+10
+20 + +40 -10
0
+10 -10 0 +10 -40 -30 -20 -10 0 +10
EXCESS OF ACTUAL OVER EXPECTED NUMBER OF EXAMINEES
16
Figure 6. Excess of actual over expected nunber of examinees by fanily imcome for places grouped according to their examination rate.
groups, only the signing of a medical authoriza-
tion was consistently associated with differential
response in areas of high and low response alike.
It may be more meaningful to consider, in-
stead of the response rates, the comparison of
the demographic configuration of the examined
group with that of the total sample. If these com-
parisons are made one characteristic at a time,
it becomes evident that nonresponse produces
some differences in the distributions by age, race
and sex, population size of place, and location; but
it is also apparent that characteristics such as
family income, occupation, usual (economic) ac-
tivity, education, and marital status appear to
have the same lineaments in the examined groups
as in the total sample. This was to be expected,
of course. For these variables the differences be-
tween the examined and nonexamined groups were
not large; also the examined group constituted
more than 85 percent of the total sample. Beyond
this, the estimation process used for the Health
Examination Survey includes adjustments for dif-
ferential response by age, sex, and place of ex-
amination. These adjustments tend to compensate
for distortions arising from differential nonre-
sponse. Differential nonresponse is not unimpor-
tant, but its effects on ‘the demographic picture
are apparently not serious.
The demographic differences may be looked
at from another point of view. Data from the first
Health Examination Survey will ordinarily be
presented either in an age-sex-specific form or
with adjustments for the known distribution of age
and sex in the population. Thus, the question is
how noticeably the demographic characteristics
of each age-sex group are altered by nonre-
sponse. A few examples may serve to illustrate
how small the effect is (table N). The median
family income, while consistently lower among
examined women than in the entire group of sam-
Table N. Selected demographic characteristics of total sample and examined persons,
by sex and age: Health Examination Survey, United States, 1960-62
Median income Median education | Percent working Percent married
Sex and age Sample Sample Sample Sample
Total Examined | Total Examined | Total Examined | Total Examined
Male
18-24 years---- | $3,778 $3,695 10.7 11.7 67.4 67.6 35:9 37.2
25-34 years---- | $4,207 $4,212 10.9 11.0 93.0 93.0 79.0 81.0
35-44 years---- | $4,548 $4,573 10,2 10.2 95.4 96.3 88.8 89.6
45-54 years---- | $4,521 $4,473 9.6 944 93.6 93.4 87.9 88.5
55-64 years---- | $3,653 $3,550 8.0 7+9 80.3 80.9 82.1 81.8
65-74 years---- | $1,674 $1,614 72 152 33.2 32.8 75.6 75.8
75-79 years---- [$ 879 $ 886 6.4 6.4 7.8 5+6 70.0 70.8
Female
18-24 years---- | $3,429 $3,343 10.6 10.6 33.9 33+3 57.0 56.0
25-34 years---- | $4,192 $4,173 10.5 10.5 30.5 29.4 83.5 83.9
35-44 years---- | $4,468 $4,463 10.3 10.3 32.2 32.2 86.2 87.1
45-54 years---- | $3,812 $3,712 9+3 93 42.6 43.8 76.4 76.3
55-64 years---- | $2,887 $2,812 8.8 89 32.8 34.8 66.8 65.9
65-74 years---- | $1,510 $1,395 7.4 7.4 9.8 9.7 47.7 45.8
75-79 years---- |$ 982 $§ 925 7.6 7.6 5.9 5.7 35.6 37.1
17
ple women, is only slightly lower. For men, there
is no consistent difference. The median number
of school years completed for the examined group
is indistinguishable in the various age-sex groups
from the median for the entire sample, although
this agreement may well conceal some slight
compensating differences in distribution. Simi-
larly, the percentage working was about the same
in each age-sex group and in the percentage
married there was little noticeable difference be-
tween the total sample and the examined group.
This relatively unruffled reflection of the sample
group in the examined group is, as already noted,
only to be expected on the basis of the facts al-
ready adduced.
MEDICAL CONDITIONS REPORTED
ON INTERVIEW
Reliance on household interview reports of
illness for comparing examined and nonexamined
persons is limited by two factors. The first is
that the reporting of chronic diseases in the house-
hold interview is selective and incomplete for
cases involving minor amounts of medical care.
This was an important reason for the establish-
ment of the Health Examination Survey. The sec-
ond limiting factor is that persons unwilling tobe
examined may be assumed to be less willing to
supply health information than persons who agree
to a medical examination. The household inter-
view information whichnonexamined persons gave
tended to be less complete and satisfactory in a
number of respects than the information given by
examined persons. Nonetheless, the health infor-
mation from the interview is highly pertinent and
ought to be considered.
Table O presents for selected conditions a
comparison between the number of conditions re-
ported on the household interview for nonex-
amined persons and the number that would have
been expected if reports for nonexamined persons
were the same as reports for the sample group
as a whole. Figure 7 presents similar data for
places grouped according to their examination
rate. It will be noted that for those chronic dis-
ease categories of special interest to the first
Health Examination Survey—diabetes, the cardio-
18
Table O. Actual and expected prevalence
of selected conditions reported on the
household interview for nonexamined per-
sons: Health Examination Survey, United
States, 1960-62
Number of conditions
Selected conditions
Actual Expected!
Diabetes=======c=c== 19 21.2
Cerebrovascular
accidents====-====== 15 8.7
Cardiovascular
diseases other
than cerebrovascu-
lar accidents------ 100 130.4
Heart disease----- 45 54.2
Hypertension------ 66 85.9
Arthritis and
rheumatism=---=----- 109 144.8
Visual defects------ 49 58.4
Hearing defects----- 52 74.2
Paralysis==========- 13 11.4
Deliveries, dis-
orders of preg-
nancy, etc--------- 42 36.8
NCTI: Conditions are not mutually exclusive.
If r is the proportion not examined in a specific ace-
sex group and t; is the number of sample nercon= in the age
group reporting the condition, then P; t. is the expected
i
value. For deliveries, etc., only women 18-44 years of ace
were considered in computing the expected value.
vascular conditions, and arthritis and rheuma-
tism—the reported prevalence for nonexamined
persons was less than expected. For diabetes the
difference was trivial, but for both the cardiovas-
cular diseases and for arthritis and rheumatism
the reported deficit was about 25 percent (table
0). The deficit in the number of hearing defects
reported was even larger, and there was also a
sizable deficit of reports of visual defects in the
nonexamined group. On the other hand, nonex-
amined persons reporting cerebrovascular acci-
dents or paralysis were more numerous than ex-
pected and there was a slight excess in the num-
ber of women reporting deliveries, etc. However,
EXAMINATION Cerebrovascular Cardiovascular EXAMINATION
RATE (%) Diabetes accidents disease Heart disease Hypertension RATE (%)
94-98 94-98
89-93 89-93
84-88 84-88
79-83 79-83
66-78 66-78
-5 0 +5 -5 0 +5 -10 -5 +5 -5 0 +5 -5 0 +5
EXCESS OF ACTUAL OVER EXPECTED NUMBER OF EXAMINEES
EXAMINATION Arthritis ond EXAMINATION
RATE (%) rheumatism Visual defects Hearing defects Paralysis Deliveries, etc. RATE (%)
94-98 94-98
89-93 89-93
84-88 84-88
79-83 79-83
66-78 66-78
-10 -5 0 +5 -5 0 +5 -5 +5 -5 0 +5 -5 0 +5
EXCESS OF ACTUAL OVER EXPECTED NUMBER OF EXAMINEES
Figure 7. Fxcess of actual over expected number not examined by condition reported on household interview for places grouped according
to their examination rate.
these latter differences probably resulted from
sampling fluctuation and in any eventare numeri-
cally too small to produce any noticeable bias in
the survey.
There was a slight deficit in the number of
nonexamined persons hospitalized during the pre-
ceding year (112 rather than the expected 122)
and in the number of episodes that were surgi-
cally treated. Again, according to the interview
reports, there was less evidence of serious ill-
ness in the nonexamined group than in the ex-
amined. It appears likely that the reported def-
icits of serious illness in the nonexamined group
reflect, in considerable measure, an unwillingness
to reveal the presence of serious illness. Their
unwillingness to be examined may be merely
another manifestation of the same attitude, Even if
the reported deficits were accurate estimates of
the difference in disease prevalence between ex-
amined and nonexamined persons, the biasing
effect would be quite small. Thus, assuming a
prevalence 25 percent lower in the nonexamined
than in the examined group, an estimate which
imputed to nonexamined persons the same preva-
lence as found for examined persons would over-
state the true figure by only 4 percent,
19
INQUIRY OF PHYSICIANS
When a substantial part of the sample is not
examined, it is most important that the persons
not examined be similar in terms of the char-
acteristics under study to the persons examined.
If all of the 13.5 percent of the sample not ex-
amined during the first Health Examination Sur-
vey had heart disease, for example, the preva-
lence figures for heart disease derived from the
examined group would represent a serious under-
statement of these statistics. Even less extreme
differentials could lead to considerable bias in
survey results. In general, a lower prevalence
in the nonexamined than in the examined group is
a matter of less concern, since the magnitude of
possible bias from a lower prevalence is usually
more limited.
In the absence of concrete evidence, itisim-
possible to say whether the nonresponse bias for
any specific characteristic is large or small,
positive or negative, or, indeed, whether itexists
at all. One source of information available to us
is the household interview. This is supplemented
by a program which was developed to obtain
auxiliary information from the person's own phy-
sician. During the household interview each sam-
ple person was asked to give the name and ad-
dress of his personal physician and to indicate
how long it had been since he had last seen him.
In each household the respondent was asked to
sign a form authorizing his physician to release
medical information to the National Health Survey.
If a nonexamined person signed such a medical
release and gave the name of a personal physi-
cian whom he had seen inthe preceding 2 years,
an inquiry was sent his physician. If the person
did not sign a release, the inquiry form was sent
to him with a request that he forward it to his
physician for completion. A similar inquiry form
was sent to an examined person from the same
place who was of the same sex and, as nearly as
possible, the same age. This program was under-
taken too late to include examined persons from
the first two stands, since their physicians had
already received reports of findings from the
Health Examination Survey and the evaluation was
to be based only on the expressed judgments of
the personal physician.
20
The inquiry form is shown in figure 8. The
request for information is brief, simple, and cat-
egorical. No criteria were offered to, or re-
quested from, the physician for any diagnosis.
Replies were tabulated as received, with no fol-
lowup to clarify obscure entries or to complete
those forms that were incomplete.
If the person said he had not seen a physi-
cian within the last 2 years, no inquiry was sent.
About 15 percent of the nonexamined persons fell
into this category. While there are a large num-
ber of reasons for not seeing a physician, includ-
ing suspected ill health, it seems reasonable to
assume that this group had a smaller than aver-
age proportion of persons with serious health
problems. Another 33 percent either gave no
verifiable physician's name or did not sign a re-
lease. As already indicated, these persons were
sent a copy of the inquiry with a request to for-
ward it to their physicians. Inthe remaining cases
the inquiry was sent directly to the physician. If
no reply to an inquiry was received, one follow-
up letter was sent; at this point the investigation
was terminated. In short, the program was con-
ducted under very low pressure.
Returns were received from 419 nonexamined
persons, or 45 percent of the nonexamined persons
from the 40 stands included. Some of the returns
indicated that the physician either did not know
the person or had relatively little of the informa-
tion requested. There were 312 returns complete
enough to be used: 272 from the 475 inquiries sent
directly to the physician and 40 from the 310 in-
quiries sent to the sample person to be forwarded
to his physician.
Total returnS -=-==-=eemmccemc eee ————— 419
Patient
Reply usable----mmmmecm meee 312
Reply unusable----mmemmcmm meee 60
Not a patient------mme-ecermemne ence ————— 47
In other words, inquiries made of personal physi-
cians yielded usable medical information for only
34 percent of the nonexamined population.
Returns for the examined persons included in
the inquiry were greater. The chief reason for
Confidentiality has been assured the individual as set forth in 22 FR 1687
PHS-3504 Form Approved Bureau of the
$60 PHYSICIAN INQUIRY Budget No. 68-R620-54
HES-211 Expires 6-30-63
SERIAL NUMBER
PATIENT'S NAME, ADDRESS, AGE, AND SEX
1. When did you last see this patient?
2. What did you treat him for at that time?
3. In general, would you describe the patient's health at that time as:
[J excellent [J Good Fair [J poor
4. Did the patient have any of the following conditions? (Please check the appropriate block)
Yes, Don't know
CONDITION Yes, possible No (Have no Information
definite or bearing on this
tentative condition)
a. Hypertension
b. Peripheral vascular disease
c. Coronary heart disease
d. Hypertensive heart disease
e. Rheumatic heart disease
f. other heart disease (Please specify)
g. Diabetes
h. Arthritis or rheumatism
5. If in your record, please specify the following measurements and the date latest
measurement was taken:
a. B8lood pressure
(Date)
b. Height Er ———————————————————; TETTE!
Cc. Weight ——————————————————————————— Pate]
(Signature of physician) (Date)
Note: Please use other side for additional information or comments
“igure & Physician in~uiry form.
21
this, of course, was the fact that all examined
persons, excluding a few accidental omissions,
signed a medical release and, in any case, ex-
amined persons were included in the inquiry only
if they had given a verifiable physician's name
and reported having seen a physician within the
last 2 years. Among those inquiries sent directly
to the personal physician the percentage of re-
turns was nearly as high for nonexamined as for
examined persons. The returns on inquiries sent
directly to the personal physician are given in the
following table:
Examined Nonexamined
Inquiries sent------- 767 475
Inquiries returned--- 656 373
Patient-=-=ceeun -- 563 326
Reply usable-- - - 489 272
Reply unusable-— 74 54
Not a patient---- -= 93 47
For both examined and nonexamined persons,
where the inquiry was sentdirectly to the person's
physician and he acknowledged that the person was
his patient, 86 percent of the returns were com-
plete enough to be used. Undoubtedly, there were
instances in which the physician consulted with
the patient before completing the inquiry. It seems
likely, however, that response or nonresponse
ordinarily reflected the cooperativeness of the
physician rather than that of his patient. Coinci-
dentally, the rate of return of forms sent to the
physician and the examination rate for the survey
were quite similar.
Examination of the returns indicated that in
190 cases there was a usable return for both a
nonexamined person and his specific match in the
examined group. This left 122 unmatched usable
returns for the nonexamined group and 299 for
the examined group. Since there was nothing to
distinguish matched from unmatched returns
either in the distribution of subjects by age and
sex or in the medical conditions reported, it was
decided to combine the two groups and compare
all usable returns for nonexamined persons with
those for examined persons.
So far as can be judged from these data the
prevalence of the cardiovascular diseases, arth-
ritis and rheumatism, and diabetes was the same
in the examined as in the nonexamined group
(table P). Furthermore, the two groups were in-
distinguishable in average height, weight, and
blood pressure (table Q). The general health of
persons in either group, in the judgment of their
personal physicians, was similar (table R).
Needless to say, neither the physician in-
Table P. Prevalence of conditions reported by personal physicians for examined and non-
examined persons:
Health Examination Survey, United States, 1960-62
Number of conditions Rate per 100 persons
Condition
Examined | Nonexamined q ]
persons persons Examined Nonexamined
Hypertension=-===-=c-ccccmccacaancaaaaaoo 99 66 20.4 21.3
Peripheral vascular disease--=---------- 59 33 12.2 10.6
Coronary heart disease----=---ccocae---- 40 32 843 10.3
Hypertensive heart disease-======-c-ce--- 54 34 11.3 11.0
Rheumatic heart disease----==----------- 15 4 3d 1.3
Other heart disease--===-=mecccmccccccaa- 12 9 2.5 2.9
Diabetes=====vccecccccncaa-o mmm meme 27 15 5.8 5.0
Arthritis or rheumatism--------cececee--- 76 43 15.8 14.2
NOTE: There were 489 examined and 312 nonexamined returns that were entirely, or almost entirely, complete. For a spe-
cific category the total may be slightly less. Conditions are not exclusive.
22
Table Q. Mean blood pressure, height,
and weight reported by personal physi-
cians for examined and nonexamined per-
sons: Health Examination Survey, United
States, 1960-62
Measurement Number Mean
Blood pressure (inmm/hg)
Examined persons==---=-- 439 | 134/80
Nonexamined persons---- 269 | 135/81
Height (in cms)
Examined persons------- 196 | 165.4
Nonexamined persons---- 116 167.5
Weight (in pounds)
Examined persons------- 313 | 149.1
Nonexamined persons---- 193 148.6
NOTE: There were 489 examined and 212 nonexamined
returns that were entirely, or almost entirely, complete. For
a specific category the total may he slightly less.
quiry, nor the morbidity information reported on
the household interview, rules out the possibility
that examined and nonexamined persons do differ
with respect to some of the characteristics eval-
Table R.
for examined and nonexamined persons:
1960-62
uated by the first Health Examination Survey.
Obviously, the information available for nonex-
amined persons is less complete and reliable
than that for examined persons; this is manifested
in the household interview and, toa larger extent,
in the physician inquiry. What does seem im-
probable at this point, however, is that the ex-
amined and nonexamined groups differ greatly;
that is, it seems unlikely that the nonresponse
has introduced a really serious bias in the find-
ings of the Health Examination Survey. This as-
surance, imprecise as it is, adds greatly to the
usability of the data.
SOME ASPECTS OF THE
EXAMINATION PROCESS
It must be stressed that the sampling aspects
of a survey are not restricted to choosing the
sample persons and persuading them to report for
examination. The conducting of the survey itself,
and of the examination, has numerous sampling
features, some of which are now mentioned,
chiefly in a cautionary vein.
In the course of the first Health Examination
Survey, 42 different places were visited. Examina-
tions were performed by 62 different physicians
and 5 dentists. There were 20 different techni-
Percent distribution of general health status reported by personal physicians
Health Examination Survey,
United States,
Number of persons Percent distribution
General health status
‘ : Examined | Nonexamined
Examined | Nonexamined per sons —
Total==mmemmcc cece ccc ccc cme mam 466 293 100.0 100.0
Excellent=====-meemeccc ccm c cee 120 73 25.8 24.9
GOOd==mmm emcee cece mmm mee 237 145 50.9 49.5
Faire=-cececccm ccc ccmcccccccmc cee 94 58 20.2 19.8
POOr=-mmmm ccc 15 17 3.2 5.8
NOTE: There were 489 examined and 312 nonexamined returns that were entirely, or almost entirely, complete. For a spe-
cific category the total may be slightly less.
23
cian-observers responsible for the anthropomet-
ric and audiometric examinations.
Ideally, each of the 6,672 examinees should
have been assigned to examination on a random
basis—random, that is, with respect to time,
place, and examiner. Obviously, this was impos-
sible. Sample persons were examined in their usu-
al area of residence. All examinations at a given
place were completed within the time span of 3
or 4 weeks. They were performed by a specific
examining team, which ordinarily included only
2 of the 62 physicians, 2 of the 20 technicians,
and 1 of the 5 dentists. If there were any difficul-
ties in the equipment or the environment at that
location, if there were any peculiarities in the
conducting of the examination by any one of the
examining physicians, if there were any wavering
in the laboratory standards at that time, or in the
interpretative standards used in X-ray or elec-
trocardiographic reading, this would be likely to
be reflected in the examination findings and would
appear as a place peculiarity. For these reasons,
apparent place differences in health found by the
first Health Examination Survey must be ex-
amined critically, taking into consideration other
factors which may possibly have produced dif-
ferences.
This has more ramifications than first ap-
pear. Places vary with respect to a large number
of demographic characteristics, so that to some
extent examination differences from place to place
tend to produce some effect on other demographic
comparisons. Most of these are probably negligi-
ble. The one possible exception is race, since
40.5 percent of all Negroes examined came from
5 of the 42 stands and 1 stand accounted for 73.0
percent of the other nonwhite examinees.
Place differences are confounded not only
with possible examination differences but also
with seasonal differences. The scheduling of
stands was deliberately arranged so that the North
would be avoided in winter and the South in sum-
mer (fig. 9). To the extent that any characteris-
tic under study varies with the season, regional
comparisons for that characteristic will be dif-
ficult to interpret. For example, if people in all
parts of the country weigh more in winter than in
summer, the survey would tend to understate the
weight of northerners and overstate the weight
24
GEOGRAPHIC AREAS
NORTHEAST BR
sou
vest
Figure 9. Health Examination Survey stands by geographic
location and date of examir.ation.
of southerners. Bias may also be introduced into
racial comparisons of suchcharacteristics unless
these are made specific by place.
Finally, the association of age and sex with
the time of day at which the person was examined
should be discussed (fig. 10). The time of exami-
nation was fitted to the convenience of the ex-
aminee; this was related to such factors as em-
ployment, and so to sex and age. Women were
more likely than men to come in during the morn-
ing or afternoon and less likely to come in during
the evening. Young people were less likely than
old to appear for examination in the morning. The
sex differentials, while consistent for the various
age groups, were not large. Some of the age dif-
ferentials, however, were quite marked. More
than 40 percent of the men under 55 years of age
came in for examination after 5 p.m., while rela-
tively few persons over 65 years old were ex-
amined in the evening.Only one-fourth of the ex-
aminees aged 18-24 years came in before noon,
AGE AND SEX
18-24 years
2am-4pm.
Male 8-11a.m.
Female
45-54 years
Male
Female
65-74 years
Female
0 20 40 60 80 100
PERCENT EXAMINED BY TIME OF DAY
Figure 10. Percent distribution of examinations by time of day
according to selected age groups and sex.
as compared with half of those aged 75-79 years.
Hence, if any variable under examination has a
marked diurnal variation this can easily appear
as an age differential in the findings for that vari-
able, even though no real age differential exists.
Small diurnal variation—with a range, for exam-
ple, of less than 10 percent—can probably be ig-
nored in analysis. Larger variation probably can-
not be ignored.
Aside from these factors, there exists in
most examination measurements a degree of var-
iability, which is often termed ''measurement
variance' to distinguish it from bias. Various
measures and much effort were devoted to prob-
lems of standardization of observations, valida-
tion of measurement processes, and other as-
pects of quality control. Some of the measures
used to assure standardization during the exam-
ining process involved developing a sound proto-
col and carefully selecting, training, and retrain-
ing an examining staff, Also worth mentioning is
the periodic use of consultants to observe and
comment on specific aspects of the examination.
Certain blood chemistry tests and measurements
were made at outside laboratories—serologic
tests for syphilis, serum bentonite flocculation
tests, and blood glucose and serum cholesterol
measurements. In addition to the quality controls
of the testing laboratories, replicate measure-
ments of blood glucose and serum cholesterol
were taken for a sample of cases. Electrocardio-
grams, chest X-rays, and X-rays of the hands
and feet also were evaluated outside the clinic.
Each of these was evaluated in replicate determi-
nations.
The factors discussed in this section consti-
tute possible sources of bias in the survey find-
ings. They are pointed out not because they are
unusual, but because they may not be so obvious
to the reader as they are tothe survey staff. Pre-
liminary analyses of Health Examination Survey
findings generally indicate that these factors are
not a matter for serious concern, but obviously
they should be considered in the analysis of spe-
cific data.
REFERENCES
1yy.s. National Health Survey: Plan and Initial Program of the
Health Examination Survey. Health Statistics, Series A-4, PHS
Publication No. 584-A4. Public Health Service, Washington, D.C.,
1962.
2U.S. National Health Survey: The Statistical Design of the
Health-Household Interview Survey. Health Statistics, Series A-2,
PHS Publication No. 584-A2. Public Health Service, Washington,
D.C., 1958.
3Commission on Chronic Illness in 1953-54: Chronic Illness in
a Large City: The Baltimore Study (Chronic Illness in the United
States, Vol. IV). Harvard University Press, 1957.
4Commission on Chronic Illness: Chronic Illness in a Rural
Area: The Hunterdon Study (Chronic Illness in the United States,
Vol. III). Harvard University Press, 1959.
3Chen, E. and Cobb, S.: “'Further Study of the Nonparticipation
Problems in a Morbidity Survey Involving Clinical Examination’
Journal of Chronic Diseases, 7:321-331, 1958.
U.S. National Health Survey: Co-operation in Health Examina-
tion Surveys. Health Statistics, Series D-2, PHS Publication No.
584-D2. Public Health Service, Washington, D.C., 1960.
7U.S. National Health Survey: Attitude Toward Co-operation in
a Health Examination Survey. Health Statistics, Series D-6, PHS
Liason No. 584-DG. Public Health Service, Washington, D.C.,
1961.
} 8\Mantel, N. and Haenszel, V.: ‘Statistical Aspectsof the Analy-
sis of Data from Retrospective Studies of Disease.’’ Journal of the
National Cancer Institute, 22:719-748, 1959.
000
25
Table 1.
26
DETAILED TABLES
Percent distribution according to sex, by selected demographic characteristics
reported in the Census and as estimated from the primary sampling units: United
States, 1960--=-=-mecemccccceee meee m meee meee meee eee ee meee sees ees ————————————
Number and percent distribution according to sample and nonsample eligible per-
sons, by selected demographic characteristics: United States, 1960-62----=-====--
Percent distribution by age and race, according to sex: Census and weighted HES
sample--=-===--c-memmemeeee meee meee meee meee meee eee em seme sme ssemsses——— ooo
Percent distribution by marital status, according to sex: Census and weighted HES
SAMPLEmmm mmm mmm mmm mn ee ee ee eee eee meme m mmm mm mmm mm—meeomoe—oomoeee
Percent distribution by years of school completed, according to sex: Census and
VELZhted HES SAMPLaw rw me ew mmm mn mom mm on oo or om om 20 0 tc 0 0 0 0 000 2 om on 0
Percent distribution by occupation group, according to sex: Census and weighted
HES sample-=======mmemceececc cece ccc cece meee eee meee eee e esses sms sss oss coe
Percent of information unknown or incomplete for selected items for examined and
nonexanined sample persons: United States, 1960=62 == mimi wisi suum m mmm m mw swe
Selected sample data=====rremrrrr rrr rrr scr rr rrr
Page
27
29
31
31
32
32
33
34
Table 1.
Percent distribution according to sex, by selected demographic characteristics reported
in the Census and as estimated from the primary sampling units: United States, 1960
Both sexes Male Female
Characteristic
Census | Estimate Census | Estimate | Census | Estimate
AGE
All races Percent distribution
Total-18-79 years-=======-cecc-e--- 100.0 100.0 100,0 100.0 100,0 100,0
18-24 years-=-=-=mm=mmmmmmmemeee mmm ————— 13.8 14.2 13.9 14,2 13.7 14,1
25m Bl YORE ees er ape eo 20.2 21.1 20.4 21.3 20.1 20.9
35-44 JOOUBmmmmmmetm mm mato mom wm ti om mm 21.4 21.8 21.3 21.9 21.3 21.6
45-54 years-======-=-----smeemccsccsecoe——— 18.2 18.0 18.5 18.3 18.0 17.7
5500, | yen Rrg eee ostium ee 13.8 13.3 13.8 13.3 13.9 13.3
5S wl] YO OIE i ah ae SO 9.8 9.1 9.4 8.7 10.2 9.6
T5
The presence of glu-
cose in the urine was evaluated in the mobile
clinic itself on a semiquantitative test scale
(Testape).
DIABETES ON MEDICAL HISTORY
A history of diabetes was considered defi-
nitely diagnostic if the examinee reported the use
of insulin or an oral hypoglycemic agent. If the
disease was reported to have been diagnosed by a
physician but the person was not on medication,
the case was accepted as definite known diabetes,
unless the blood glucose level was below 138 mg.%
without challenge or 148 mg.% with challenge. In
most instances the levels were substantially high-
er (Appendix II). Most of the diabetes reported in
the medical history met the tests for definite
known diabetes. If a person reported diabetes
which did not satisfy the criteria for a definite
diagnosis but had seen a physician for the disease
within the last 6 months and had a followup med-
ical appointment scheduled within the next 6
months, he was considered a questionable case.
Otherwise no diabetes was diagnosed. Less than
9 percent of the examinees reporting diabetes
failed to meet the criteria for a definite or ques-
tionable diagnosis.
The prevalence of definite known diabetes in
adults (18-79 years) was 1.8 percent (table A), or
approximately 2 million persons. A history of
diabetes was rarely reported under 45 years of
age but was more common at older ages. Although
the specific rates reported for each age group
Table A. Number of adults with definite
known diabetes per 100 persons, by age
and sex: United States, 1960-62
Age BOLE Men | Women
Number of diabetics
per 100 persons
Total-18-79
years-===--- 1.8 1.3 2.1
18-24 years------ 0.3 0.2 0.5
25-34 years-=---- 0.4 0.2 0.6
35-44 years------ 0.9 1.1 0.8
45-54 years------ 2.0 L.1 2.9
55-64 years------ 3.3 3.3 3.2
65-74 yearg------ 4.8 32 6.1
75-79 years------ 4.7 2.7 6.7
have a high sampling variability, there is clearly
a gradient by age. There is also a sex differen-
tial, with more women than men reporting dia-
betes. While the data are insufficient for delineat-
ing the sex differential with great precision, they
are not inconsistent with a slight excess of dia-
betes in women at younger ages and a larger ex-
cess at older ages.
THE GLUCOSE TOLERANCE
TEST—TECHNIQUES
As already noted, a glucose challenge was
given each examinee without regard to the time or
content of the previous meal, and 1 hour later a
venous blood specimen was taken. This proce-
dure, while differing in many respects from a
standard clinical test for glucose tolerance, has
been shown to provide a satisfactory equivalent
(Appendix 1) In any event, the standard testwas
clearly impractical for use in the Health Exami-
nation Survey and some reasonable alternative
had to be devised.
A blood specimen of 3 ml. was collected in
prelabeled B-D ''vacutainers' containing 30 mg.
of sodium fluoride. Specimens were promptly re-
frigerated and twice a week the accumulated spec-
imens were shipped on water ice to the Diabetes
Field Research Unit in Brighton, Massachusetts
for determination of glucose concentration by the
Somogyi-Nelson method.
A considerable effort went into attempts to
control and measure the technical variability of
blood glucose determination during the Survey.
This is a much more difficult enterprise than is
generally realized, but worth the effort, since the
most carefully designed survey can easily be de-
graded by careless laboratory work. The Survey
was especially fortunate in having an excellent
laboratory available for blood glucose determi-
nations.
All the standard controls were used in the
laboratory work at Brighton, and as the study
progressed additional controls were devised. Also
several methods were used by the Survey staff to
keep informed of what was happening. The first
was a comparison of blood glucose levels for dif-
ferent places. While it is impossible to distin-
guish real place differences from laboratory
fluctuations, any systematic change with time
was regarded with suspicion. In one instance, a
series of reporting errors was uncovered by this
means and corrected.
Variations from stand to stand are shown in
figure 1. Expected values are calculated for each
stand allowing for differences in age-sex distri-
butions but assuming that the levels reported at
each age group for all stands combined are what
should be expected at any individual stand. Both
mean blood glucose levels and the prevalence of
a trace or more of urine glucose vary, stand by
stand, from expected values. The fact that these
two measures tend to deviate in the same direc-
tion suggests that most of the observed stand
variation reflects differences in the persons ex-
amined. When these two measures deviate in op-
posite directions, the discordance can be attrib-
uted partly to the fact thaturine glucose and blood
glucose levels are not exactly correlated and
partly to technical variability. The fact that such
discordances do not persist over a series of stands
suggests that long-term technical variability is
probably not an important factor in this Survey.
The second method consisted of drawing two
blood specimens from the same person, sending
the original to Brighton as a routine specimen and
sending the duplicate to an independent laboratory.
Again, this provides no absolute check, since good
laboratories do differ and it is seldom possible
to agree on which is "right," but gross deviations
serve to alert a laboratory to possible unsus-
pected difficulties. The third method was to pro-
vide blind aliquots to the Brighton laboratory to
determine the consistency of the work of the
laboratory.
Some of the details from these comparisons
are given in Appendix IV. The conclusions drawn
from the various comparisons may be summa-
rized briefly.
Laboratories and technicians in the same
laboratory tend to operate at different levels.
Thus, in a series of 272 comparisons undertaken
in February and March 1961 the Brighton labora-
tory determinations averaged 7.8 mg.% higher
than those from the comparison laboratory,and
on comparisons made between June 1961 and May
1962 the. Brighton laboratory averaged 0.3 mg.%
lower than another laboratory on 103 specimens,
although if one highly aberrant series were elim-
inated the Brighton laboratory would be 2.0 mg.%
higher. While each technician tended to be highly
consistent in a single laboratory run, two tech-
nicians might on occasion differ in the level of
their blood glucose determinations by as much as
7 mg.% despite the regular use of control speci-
mens and standard techniques. So far as can be
determined, this difference may arise in the
handling of whole blood without appearing atall in
determinations made on the usual aqueous or
serum controls.
Changes within a laboratory over a period of
time are exceedingly difficulttodiscover and con-
trol. During a study conducted between January
and May 1962, it was concluded that the effective
technical variation did not exceed 5 mg.% at the
Brighton laboratory. Included in this were vari-
ability in the work of individual technicians,
among technicians, and between different labora-
tory runs over the entire time period. This is a
remarkably low figure and no claim is made that
it covers the entire span of the first cycle of the
Health Examination Survey, although another se-
ries of comparisons for a longer period of time
suggests that the long-term variability is not
much greater.
Finally, no evidence was found that the pre-
servative used, the varying length of time be-
tween drawing a specimen and measuring it, or
the methods of transporting specimens between
the field and the laboratory produced any signifi-
cant effects on the blood glucose level reported.
These observations summarize the findings
from the various quality checks made on blood
glucose determinations. No similar checks were
made of urine glucose determinations. When rea-
sonable attention to instructions is given, the
technique used in this Survey has been shown to
be quite reliable.>® The high correlation of urine
+15
+10
+5
i
Ty
Excess (mg. %)
o
Mean blood glucose
+10
| | re :
0
Prevalence of a trace or more of urine glucose
+15
|
8 fo
+15 +15
S tio +10
e
: ” |]
Q *
2 5 [Te [1] | L =
2 i. -
} Pep oT Tr
< 5 _5
@
8
= -lo -10
-15 -15
Figure 1. Excess of actual over expected levels of blood and urine glucose, by stand: Health Examination Survey, 1960-62.
Table B. Mean blood glucose levels after Table C. Number of adults with urine glu-
challenge in adults, by age and sex: cose after challenge per 100 persons, by
United States, 1960-62 age and sex: United States, 1960-62
Both Both
Age Senes Men | Women Age SEES Men | Women
Mean blood glucose Number of adults with
levels in mg.% urine glucose per
100 persons
Total-18-79
years======= 12..3 115.7 126.4 Total-18-79
years======= 4.3 17.9 10.8
18-24 years=----= 99.7 94.6 104.1
25-34 years=====-- 105.7 101.5 109.5 18-24 years===--- 8.7 11.4 6.3
35-44 years====-= 116.5 115.2 117.6 25-34 years====== 11.1 13.8 8.7
45-54 years====--- 125.8 118.2 133.1 35-44 years===-=- 15.3 20.0 10.9
55-64 years------ 137.8 130.1 145.2 45-54 years===---- 16.2 20.4 12.0
65-74 years===--- 150.7 139.8 159.7 55-64 years==-=-=--- 14.8 17:3 12.4
75-79 years==---- 166.3 154.4 178.7 65-74 years===--- 19.3 26.4 13.4
75-79 years------ 33.0 21.3 24.4
glucose determinations with blood glucose levels
for the same persons, independently determined,
tends to corroborate the precision of both meas-
urements.
FINDINGS OF THE
GLUCOSE TOLERANCE TEST
On the average, the blood glucose concentra-
tion after challenge was higher the older the ex-
aminee (table B, fig. 2). Overall, the level for
women was about 10 mg.% higher than that for
men, being somewhat less at younger ages and
somewhat greater at older. The age gradient for
either sex was quite steep. For men 18-24 years
of age, the mean glucose level was 94.6 mg.%;
for men 75-79 years, it was 154.4 mg.%. There
was a similar age gradient for women.
The indicated shift in mean levels by age
corresponds to a shift in the distribution curves,
with the appearance of an increasing number of
high glucose values at older ages (fig. 3). Exclud-
ing known diabetics, only 0.8 percent of men under
35 years of age had levels in excess of 200 mg.%,
whereas 9.7 percent of men 65-79 years had
values this high. For women, the comparable fig-
ures were 0.9 and 14.0 percent.
It should be noted that the levels reported are
those obtained after challenge. Most diabetics,
NOTE: Urine glucose was considered present if a trace or more
of glucose was found in the urine.
180
60} —
140
120
Blood glucose level (mg. %)
LL Co]
20 30 40 50 60 70 80
Age in years
Figure 2. lean blood glucose levels after challenge in adults, by
age and sex: United States, 1960-62.
Percent of persons in each age group
Men AGE GROUP Women
201 -20
18-24
|
10} | Hi0
0 —_ 0
20 | -120
- 25-34 |
|
10 | Im : dio
0 Ll 0
201 20
35-44
10} | | dio
| |
| | |
0 0
20 420
| 45-54 | |
] | | |
10 I] 10
Mh all ore
ol 0
20 | -20
| 55-64 |
10k | | dio
0 he | TL, a 0
20F 420
65-74 | |
|
| |
10 — | | | ~10
| hn |
0 0
20 20
75-79 | ]
of 410
0 [1 ! ed 1 0
50 100 150 200 250 300 50 100 150 200 250 300
Blood glucose levels (mg.%)
Figure 3. Percent distribution of blood glucose levels in adults, by age and sex: United States, 1960-62.
then, are not included in the population covered by
tables B and C, since most diabetics were not
given a glucose challenge. These diabetics would
have blood glucose levels after challenge sub-
stantially higher than those for nondiabetics —100
mg.% higher is probably a conservative estimate—
and practically all of the male diabetics and the
majority of the female diabetics could be expected
to have positive urine specimens after a glucose
challenge. Therefore, if all diabetics had been
given a glucose challenge, the values shown in
tables B and C would have been slightly higher
than those reported, especially in age groups over
45 years.
The prevalence of urine specimens with a
trace or more of glucose, on the other hand, showed
weaker differentials by age and sex (table C).
While the proportion of urine specimens with a
detectable amount of glucose increased with age,
the increases were not comparable with the in-
creased proportion of high blood glucose levels.
There was roughly a twofold increase from ages
18-24 years to 65-79 years in the percentage of
persons with a trace or more of glucose in their
urine after a glucose challenge—from 11.4 to
25.4 percent for men, from 6.3 to 15.4 percent
for women. This implies, of course, that the prob-
ability of "spilling" glucose at any given concen-
tration of blood glucose decreases with age, and
this is indeed what was found. (Defining glucose
as present only if the urine concentration was 1
plus or greater would reduce overall prevalence
some 30 percent but would not significantly alter
the reported differentials by age and sex. A defi-
nition based on a concentration of 2 plus or more
would lead to an overall prevalence of only 45
percent of that reported.)
It is interesting to note that women, despite
higher blood glucose levels, were distinctly less
likely to have glucose in their urine than were
men. In fact, overall, only 10.8 percent of the
women, as against 17.9 percent of the men, had a
trace or more of glucose in their urine speci-
mens. This sex differential, as might be expected,
holds for all levels of blood glucose concentra-
tion. In particular, when the blood glucose con-
centration was 220 mg.% or more, 92.4 percent of
the men, while only 64.1 percent of the women,
showed evidence of urine glucose. There does not
seem to be any precedent for this finding, although
it is implicit in the reports of several other
studies, and unpublished data from at least one
study show an even larger sex differential.
COMPARATIVE DATA
While many medical surveys include tests for
diabetes, to our knowledge there has been only one
systematic canvass of a well-defined population
group using techniques comparable with those of
the Health Examination Survey.’ Glucose tol-
erance data from this survey have not yet been
published. There have been, of course, numerous
screening programs for unrecognized diabetes
and some of these have involved canvassing well-
defined population groups; but it is exceedingly
difficult to compare their results or to define a
table of equivalents between their glucose tol-
erance findings and those from the Health Exami-
nation Survey.
A number of surveys, however, have meas-
ured the prevalence of known (diagnosed) diabetes
in well-defined populations. A few examples are
of interest for comparative purposes. By putting
the reported statistics, so far as possible from
the published data, on a population base com-
parable to that used in this Survey (the civilian,
noninstitutional U.S. population aged 18-79
years as of October 1, 1961), the following prev-
alence figures per 1,000 may be cited:
Men Women
United States (1960-62) 13 21
Oxford, Massachusetts
(1946-47)8 14 22
Newmarket, South Por-
cupine, and Hawkesbury,
Canada (1951, 1953)9,10 14 17
Bergen, Norway (1956)11 9 5
Ibstock, Great Britain
(1958) 12 5 16
Birmingham, Great
Britain (1962)13 7 7
It is seen that the United States and Canadian
surveys reported essentially the same prevalence
of known diabetes, while the Norwegian and
British surveys reported a lower prevalence.
Another source of information for the preva-
lence of known diabetes is the Health Interview
Survey of the National Health Survey, which de-
rives its information from household morbidity
interviews. The health interview appears to lead
to a net understatement of the number of persons
with diagnosed diabetes (Appendix II), but itis not
unreasonable to assume that the amount of under-
statement is approximately the same for every
age-sex group. At least, the available evidence
from the Health Examination Survey, scanty
though it is, is not inconsistent with such a con-
clusion. Since the sample used for the Health In-
terview Survey is so much larger than that used
for the Health Examination Survey, the Health
Interview Survey reports probably constitute the
best source for information on differentials of
diagnosed diabetes by age and sex for the United
States. !4
Unlike reports on the prevalence of known
diabetes, there are few usable data on blood glu-
cose levels in general population groups. The
Oxford survey used as its test of glucose toler-
ance a venous blood specimen (and urine speci-
men) obtained about 1 hour after the midday or
evening meal, without any additional glucose load-
ing. So far as can be inferred from the published
data, not more than two persons inevery. hundred
aged 18-79 years who were supposedly free of
diabetes were found to have blood glucose levels
of 170 mg.% or more. The laboratory method
used by the Oxford survey yielded blood glucose
levels roughly 20 mg.% higher than those obtained
by the laboratory method in use by the Health
Examination Survey.!’
Allowance must also be made for differences
in the methods of challenge, since each examinee
of the Health Examination Survey was given a
drink of 50 grams of glucose 1 hour before a blood
specimen was drawn. A study undertaken by the
Health Examination Survey * suggests that this
would yield specimens withlevels roughly 10mg.%
higher than those obtained by the challenge used
in the Oxford survey. (This assumes that in the
Oxford survey blood specimens were always taken
1 hour after a meal. Were they frequently taken at
a longer interval a slightly larger difference—say
15 mg.%—should be allowed.)
Thus, the screening level used in the Oxford
survey corresponds approximately to a level of
160 mg.% in the Health Examination Survey. The
Health Examination Survey found 16 percent of the
persons without known diabetes to have blood glu-
cose levels after challenge at least that high, as
contrasted with the 2 percent found by the Oxford
survey. In fact, 2 percent of the persons in the
Health Examination Survey had blood glucose
levels of 200 mg.% or more.
Without going into similar detail for other
studies, it appears that even when due allowance
is made for differences in technique, the number
of persons with elevated blood glucose levels is
greater in the Canadian surveys cited than in the
Oxford survey and greater in the Bergen survey
than in the Canadian. In none of these surveys,
however, are there as many persons withelevated
blood glucose levels as were found by the Health
Examination Survey.
These differences are very puzzling. The
methodological study of the HES as well as the
work of others indicates that a regular meal con-
stitutes a reasonable equivalent to a standard
glucose challenge, if due allowance is made for
differences in absolute levels. Nonetheless, even
after making this allowance there remains a con-
siderable gulf between HES findings and findings
previously published. The most careful check of
HES data is convincing that the data are valid and
reliable, and unpublished data from other sources
suggest that HES findings may, in fact, be con-
servative. It would appear that there are some
factors involved in the usual screening survey
that tend to lead to unreasonably conservative
results.
That this possibility should be seriously con-
sidered is suggested by data from the Bergen sur-
vey. Capillary blood levels 1 to 2 hours after a
meal for persons over 30 years of age were ap-
proximately 108 mg.%. Asa methodological check,
standard glucose tolerance tests were performed
on a sample of persons who were negative in the
initial screening. Their levels 1 hour after chal-
lenge were approximately 152 mg.%. When an
allowance for a difference of 25 mg.% between glu-
cose concentrations in capillary and venous spe-
cimens is made © these levels are notgreatly dif-
ferent from those found in the Health Examination
Survey. On the other hand, the difference between
levels after a meal and levels after a standard
challenge is much greater in the Bergen group
than would be expected. This subject merits fur-
ther investigation.
UNKNOWN DIABETES
It is obvious that by current standards there
are a large number of people in the United States
who have elevated blood glucose levels after chal-
lenge (tables 1-4). The translation of these find-
ings into estimates of the prevalence of unknown
diabetes is another matter, however. Most inves-
tigators would hesitate to make a diagnosis of
diabetes without more extended tests of glucose
tolerance than were undertaken by the Health
Examination Survey. On the other hand, the higher
the blood glucose level after challenge, the great-
er the likelihood that diabetes would be found to
be present on a more extended medical workup.
Thus, the probability that diabetes is present is
very low if the blood glucose level after challenge
is 100 mg.% and very high if it is 300 mg.%.
Opinions would differ, however, as to the propor-
tion of persons at each blood glucose level who
should be diagnosed as diabetic—if for no other
reason than that the criteria for diagnosis are
variable.
The following data, however, will serve to
give some idea of the very large number of per-
sons in the United States who have some evidence
of impaired glucose tolerance. There are more
than an estimated 4 million persons aged 18-79
years in the United States who have blood glucose
levels of 200 mg.% or greater after challenge. Of
these, 2.9 million have urine glucose as well.
There are an estimated 6.4 million persons with
findings of glucose in the urine who have blood
glucose levels of 170 mg.% or more after chal-
lenge (tables 1,3).
Does this mean that there are a large num-
ber of undiagnosed diabetics in the United States?
Perhaps it does. Or perhaps it means that cur-
rent standards for a normal blood gluccse level
are unrealistically low. In the practice of medi-
cine this is an important question. The use of
routine tests for diabetes and of screening sur-
veys means that the suspicion of disease is
raised in persons who have no presenting signs or
symptoms other than an elevated blood glucose
level on challenge. Under these circumstances it
is especially important to have a realistic meas-
ure of the usual blood glucose level ina symptom-
free population. It is hoped that the data from the
Health Examination Survey will serve that pur-
pose.
Obviously, this Survey cannot answer many of
the questions it raises. For example, it is seen
that blood glucose levels after challenge rise with
age even in a symptom-free population. Is this
"mormal'' or is it a mark of anincreasing amount
of hidden pathology? Women have distinctly higher
blood glucose levels than men, and the distinction
is especially marked at older ages. Is this a
"normal" sex difference? If so, it may mean that
the blood glucose levels used in diagnosing dia-
betes should be higher for women than for men; in
other words, the reported sex difference in diag-
nosed diabetes is based on unrealistic standards.
Alternatively, it may mean that women actually
do have diabetes more than men.
Clearly it is outside the scope of this Survey
to answer these questions, but it is within the
scope of the Survey to raise them. In any case, it
is hoped that the findings presented in this report
may stimulate and guide other studies in the field
of glucose tolerance.
SUMMARY
Approximately 2 million persons in the United
States have definite evidence of diabetes and know
they have it.
Blood glucose levels after challenge and the
prevalence of findings of urine glucose after chal-
lenge increase with age.
Blood glucose levels after challenge are high-
er for women than for men; the prevalence of
urine glucose findings is lower for women than
for men.
The likelihood of urine glucose being mani-
fested at a given level of blood glucose is less at
older ages than at younger and less for women
than for men.
The number of persons withwhatis generally
considered evidence of ''unknown diabetes" is sub-
stantially greater than the number of known
diabetics.
REFERENCES
1y.s. National Health Survey: Plan and initial program of the
Health Examination Survey. Health Statistics. PHS Pub. No. 584-
A4. Public Health Service. Washington, D.C., May 1962.
National Center for Health Statistics: Cycle I of the Health
Examination Survey, sample and response. Vital and Health Sta-
tistics. PHS Pub. No. 1000-Series 11-No. 1. Public Health Serv-
ice. Washington, D.C., Apr. 1964.
3Nelson, N. A.: Photometric adaptation of the Somogyi method
for the determination of glucose. J. Biol. Chem. 153:375, 1944.
“National Center for Health Statistics: The one hour oral glu-
cose tolerance test. Vital and Health Statistics. PHS Pub. No.
1000-Series 2-No. 3. Public Health Service. Washington, D.C., July
1963.
SBell, W. N., and Jumper, E.: Evaluationof Testape as a quan-
titative indicator. J.A.M.A. 166:2145-2147, Apr. 1958.
607 Sullivan, J. B., Kantor, N., and Wilkerson, H. L. C.: Com-
parative value of tests for urinary glucose. Diabetes 11:53, 1962.
"Napier, J. A.: Field methods and responserates in the Tecum-
seh Community Health Study. Am. J. Pub. Health 52:208-216, Feb.
1962.
8Wilkerson, H. L. C., andKrall, L. P.: Diabetes ina New Eng-
land town, a study of 3,516 persons in Oxford, Massachusetts.
J.A.M.A. 135:209-216, Sept. 1947.
10
9Kenny, A. ]., Chute, A. L., and Best, C. H.: A study of the
prevalence of diabetes in an Ontario community. Canad. M.A.].
63:233, 1931.
10g nny, A. J., and Chute, A. L.: Diabetes in two Ontario com-
munities. Diabetes 2:187, 1953.
Hysrde, R.: The Diabetes Study in Bergen, Norway, 1956. Ber-
gen, 1962.
Rwalker, J. B., and Kerridge, D.: Diabetes in an English Com-
munity. Leicester University Press, 1961.
Report of a Working Party Appointed by the College of Gen-
eral Practitioners: A diabetes survey. Brit. M.]. 1:1497-1503, June
1962.
141s. National Health Survey: Diabetes reported in inter-
views, United States, July 1957-June 1959. Health Statistics.
PHS Pub. No. 584-B21. Public Health Service. Washington, D. C.,
Sept. 1960.
ISwilkerson, H. L. C., Cohen, A. S., Kantor, N., and Francis,
J. O’S.: A comparison of blood sugar analyses by the Folin-Wu
and Somogyi-Nelson procedures. Diabetes 11:204, 1962.
16Kjime, C. R., Wolff, F. W., Silverman, C., and Conant, J.:
Calibration of a simplified cortisone glucose tolerance test. Dia-
betes 10:351, 1961.
000
Table
1.
DETAILED TABLES
Number of adults accordin to blood glucose level after challenge, by age and
sex: United States, 1960-62--====m-mccmem occ cece cence mme meena
Percent of adults according to blood glucose level after challenge, by age and
sex: United States, 1960-62-=====mmemmmcm emcee cece en
Number of adults with a trace or moreof urine glucose according to blood glucose
level after challenge, by age and sex: United States, 1960-62===--==ccceceecacaaa-
Percent of adults with a trace or more of urine glucose according to blood glu-
cose level after challenge, by age and sex: United States, 1960-62-======cee=e=-
Page
12
13
14
Table 1. Number of adults according to blood glucose level after challenge, by age and sex:
United States, 1960-62
[Fxcludes known diabetics—definite or auestionable]
Blood glucose level Tost 18-24 25-34 35-44 45-54 55-64 65-74 75-79
in mg.% years years years years years years years
years
Both sexes Number of adults in thousands
All levelsl-=ee-a- 111,087 15,568 21,573 | 23,697 20,576 15,638 11,164 2,871
150 OF MOLE===mmmmmmmmmn 23,220 903| 2,097 | 3,778| 4,649] 5,590| 4,790 | 1,414
160 or more----------=--- 17,202 525 1,332 2,011 3,392 4,130 4,021 1,191
170 or more-=--=--------- 12,900 330 903 1,891 2,472 3,269 3,040 992
180 or more=====-==----- 9,372 173 630 1,242 1,695 2,352 2,436 846
190 or more------------- 6,612 98 418 744 1,166 1,846 1,698 643
200 or more===========-- 4,277 73 247 442 756 1,149 1,040 571
210 or more==-========-- 3,037 43 144 287 447 875 842 399
220 or more------------- 2.151 30 93 134 300 661 676 257
230 or more----=-==-==-- 1,757 15 93 94 266 523 535 231
Men
All levelsl-=====-- 52,744 7,139 10,281 11,373 10,034 7,317 4,972 1,428
150 or more======-====-=-- 8,604 191 784 1,700 1,712 2,040 1,665 512
160 or more======--=-==-= 6,219 91 514 1,250 1,321 1,332 1,360 353
170 or more=====-=======- 4,717 32 387 873 988 1,024 1.102 313
180 or more==========--- 3,353 32 286 657 569 781 749 279
190 or more-------ec-cc--- 2,303 32 178 367 314 625 541 247
200 or more===---=----=--- 1,525 16 139 192 186 416 345 231
210 or more=====-===--=-- 1,05) 16 71 129 124 296 259 1535
220 or more=========-==-- 756 16 41 70 62 233 226 109
230 or more===-=---=----- 640 - 41 41 47 199 213 100
Women
All levelsle-meann 58,343 8,430 11,2010 12,325 10,542 8,121 6,192 1,443
150 or more--=-----=----- 14,616 712 1,313 2,078 2,936 3,330 3,125 902
160 or more---=--======= 10,982 434 819 | 1,361| 2,071] 2,798| 2,661 840
170 or more-=-=--===e=a- 8,183 301 516 1,018 1,485 2,246 1,939 679
180 OF MOL@====wnmesnuns 6,019 141 344 585 1,126] 1,571] 1,686 567
190 or more------------- 4,309 66 240 877 853 1,221 1,156 396
200 Or more=====-=--ce--- 2,753 57 108 250 570 732 695 340
210 or more=======-===-- 1,986 28 73 157 323 579 583 243
220 Or more==========-=- 1,396 15 52 64 238 428 450 148
230 or more=====-=====-- 1,117 15 52 54 220 324 322 131
IP ntries for “All levels” are counts of the total population in the specified age-sex group, including diabetics. Percentages displayed in
table 2 are computed with these counts as the base.
NCTE: Many numbers in this table have large sampling errors, entries in the smallest cells being so unstable that they should be interpreted
only as indicating that the true number is small. They have been printed, nonetheless, in the belief that by so doing a better overall pattern is
reflected. See table III for illustrative sampling errors.
12
Table 2.
Percent of adults according to blood glucose level after challenge, by age and sex:
United States, 1960-62
[xcludes known diaketics—definite or questionable]
Blood glucose level Tools 18-24 25-34 35-44 45-54 55-64 65-74 75-79
in mg.% years years years years years years years years
Both sexes Percent of adults
150 or more============= 20.9 5.8 9.7 15.9 22.6 35.7 42.9 49.3
160 or more============= 15.5 3.4 6.2 11.0 16.5 26.4 36.0 41.5
170 or more===========-- 11.6 2:1 4.2 8.0 12.0 20.9 27.2 34.6
180 or more============- 8.4 1.1 2.9 5.2 8.2 15.0 21.8 29.5
190 or more=====-======- 6.0 0.6 1.9 3.1 5.7 11.8 15.2 22.4
200 or more-============ 3.9 0.5 1.1 1.9 3.7 7:3 9.3 19.9
210 or more====-==ec====-- 2.7 0.3 0.7 1.2 2.2 5.6 7.5 13.9
220 Or more-============-= 1.9 0.2 0.4 0.6 1.5 4.2 6.1 2.0
230 Or more============= 1.6 0.1 0.4 0.4 1-3 3.3 4.8 8.0
Men
150 or more~-=-========= 16.3 2.7 7.6 14.9 17.1 27.1 33.5 35.9
160 or more==-========-- 11.8 1.3 5.0 11.0 13.2 17.7 27.4 24.7
170 or more-======-===-=-- 8.9 0.4 3:8 7-7 9.8 13.6 22.1 21.9
180 or more-=-=-==-==-=-- 6.4 0.4 2.8 5.8 5.7 10.4 15.1 19.5
190 or more===========-- 4.4 0.4 1.7 3.2 3.1 8.3 10.9 17.3
200 or more============- 2.9 0.2 1.4 1.7 1.9 5.5 6.9 16.2
210 or more==-=========== 2.0 0.2 0.7 1.1 1.2 3.9 5.2 10.9
220 or more============- 1.4 0.2 0.4 0.6 0.6 3.1 4.5 7.6
230 or more===========-- 1.2 - 0.4 0.4 0.5 2.6 4.3 7.0
Women
150 or more==========--- 25.1 8.4 11.6 16.9 27.9 43.7 50.5 62.5
160 or more-===-===-=---= 18.8 5:1 7:3 11.0 19.6 34.5 43.0 58.2
170 or more==========-== 14.0 3.6 4.6 8.3 14.1 27.7 31.3 47.1
180 or more-==-======-=- 10.3 1:7 3.0 4.7 10.7 19.3 27.2 39.3
190 or more============- 7.4 0.8 2.1 3.1 8.1 15.0 18.7 27.4
200 or more======mem=m==- 4.7 0.7 1.0 2.0 5.4 9.0 11.2 23.6
210 or more===========-- 3.4 0.3 0.6 1.3 3.1 7-1 9.4 16.8
220 or more==========-=-- 2.4 0.2 0.5 0.5 2.3 5.3 7:3 10.3
230 or more-=-====m=————— 1.9 0.2 9.5 0.4 2.1 4.0 5.2 9.1
NOTE: See footnotes on table 1.
Table 3.
Number of adults with a trace or more of urine glucose according to blood glucose level
after challenge, by age and sex: United States, 1960-62
[Fxcludes known diabetics—~definite or questionable]
Blood glucose level Toesl- 18-24 25-34 35-44 45-54 55-64 65-74 75-79
in mg.% years years years years years years years years
Both sexes Number of adults in thousands
All levelsl--eee-- 15,114 1,334 2,386 3:38 Syise 2,180 1,936 565
150 or more--=-=-=====-- 8,389 362 955 1,789 1,637 1,591 1,527 527
160 or more-=-----=-===-- 7,277 247 708 1,531 1,483 1,401 1,435 472
170 or more-=----=-==-=-- 6,375 158 603 1,201 1,328 1,316 1,317 451
180 or more-=------=-==--- 5,054 110 444 911 945 1,089 1,136 418
190 or more--==-===-=-=-- 3,947 82 293 592 713 938 972 356
200 or more-==-=-======- 2,886 73 166 414 524 645 724 340
210 or more-=====-=====- 2,174 43 99 258 366 496 638 273
220 or more-=-=--======= 1,582 30 48 119 258 369 560 199
230 or more--=--====-==== 1,419 15 48 79 235 325 518 199
Men
All levelsl------- 9,119 813 1,419 2,182 2,009 1,244 1,194 258
150 or more===========-- 4,220 109 471 966 814 764 877 220
160 or more==-=====-=-=-- 3,507 39 327 836 714 619 807 165
170 or more-----=======-- 3,042 16 274 655 623 582 729 165
180 or more============= 2,350 16 208 540 382 464 575 165
190 or more---=========- 1,741 16 130 286 232 428 485 165
200 or more---========-- 1,241 16 102 176 139 339 319 149
210 or more---=-=======- 954 16 59 114 109 266 259 132
220 or more---=========- 688 16 27 54 62 203 226 100
230 or more-============ 589 ~ 27 25 47 176 213 100
Women
All levelsl--=-ee-- "5,995 521 967 1,399 1,123 936 742 307
150 or more-------==m==-- 4,169 253 484 824 824 827 650 307
160 or more--========-==- 3,769 208 381 695 768 782 628 307
170 or more-----=======- 3,333 143 329 548 705 734 588 287
180 or more-----===-===-- 2,704 95 237 371 564 625 561 253
190 or more------=-====- 2,206 66 163 306 481 510 488 192
200 or more-==-=mme==--- 1,645 57 63 238 384 306 405 192
210 or more---=========- 1,220 28 40 144 257 230 379 142
220 or more===========-- 894 15 20 64 197 166 333 99
230 or more------=======- 830 15 20 54 188 149 305 99
I¥ntries for *‘All levels’ are counts of the total population in the specified age-sex group, including diabetics. Percentages displayed in
table 4 are computed with ‘‘All levels’’ counts from table 1 as the base.
NOTE: Many numbers in this table have large sampling errors, entries in the smallest cells being so unstable that they should be interpreted
only as indicating that the true number is small. They have been printed, nonetheless, in the belief that by so doing a better overall pattern is
reflected. See table III for illustrative sampling errors.
Table 4. Percent of adults with a trace or more of urine glucose according to blood glucose level
after challenge, by age and sex: United States, 1960-62
[Fxcludes known diabetics—~definite or questionable]
Total-
Blood glucose level 18-79 18-24 25-34 35-44 45=54 55-64 65-74 75-79
in mg.% years years years years years years years years
Both sexes Percent of adults
150 or more===-===-===--- 7.6 2.3 4.4 7.5 8.0 10.2 13.7 18.4
160 or more=-=========-- 6.6 1.6 3.3 6.5 7:2 9.0 12.9 16.4
170 or more====--======-- 5.7 1.0 2.8 5.1 6.5 8.4 11.8 15.7
180 or more---==-=====-- 4.5 0.7 2:1 3.8 4.6 7.0 10.2 14.6
190 or more===========--- 3.6 0.5 1.4 2.5 3.5 6.0 8.7 12.4
200 or more===========-=- 2.6 0.5 0.8 1.7 2.5 4.1 6.5 11.8
210 or more-===========- 2.0 0.3 0.5 1:1 1.8 3.2 5.7 9.5
220 or more============= 1.4 0.2 0.2 0.5 1.3 2.4 5.0 6.9
230 or more============- 1:3 0.1 0.2 0.3 1.1 2.1 4.6 6.9
Men
150 or more---=========-= 8.0 1.5 4.6 8.5 8.1 10.2 17.6 15.4
160 or more--========m==- 6.6 0.5 3.2 7.4 7.1 8.2 16.2 11.6
170 OF MOr@wwwmmws wenn 5.8 0.2 2.7 5.8 6.2 7.7 14.7 11.6
180 or more--=========-= 4.5 0.2 2.0 4.7 3.8 6.2 11.6 11.6
190 or more---======-=-- 3.3 0.2 1.3 2.5 2.3 5.7 9.8 11.6
200 or more============- 2.4 0.2 1.0 1.5 1.4 4.5 6.4 10.4
210 or more============- 1.8 0.2 ‘0.6 1.0 1.1 3.5 5.2 9:2
220 or more============- 1.3 0.2 0.3 0.5 0.6 2.7 4.5 7.0
230 or more============= 1.2 = 0.3 0.2 0.5 2.3 4.3 7.0
Women
150 or more-===========- 7.1 3.0 4.3 6.7 7.8 10.2 10.5 21.3
160 or more--=-========= 6.5 2.5 3.4 5.6 7.3 9.6 10.1 21.3
170 or more-=-===m===n=- 5.7 1.7 2.9 4.4 6.7 9.0 9.5 19.9
180 or more=-=-========- 4.6 1.1 2.1 3.0 5.4 7.7 9.1 17.5
190 or more---========-= 3.8 0.8 1.4 2.5 4.6 6.3 7.9 13.3
200 or more-===========- 2.8 0.7 0.6 1.9 3.6 3.8 6.5 13.3
210 or more=-=-=========- 2.1 0.3 0.4 1.2 2.4 2.8 6.1 9.8
220 or more============= 1.5 0.2 0.2 0.5 1.9 2.0 5.4 6.9
230 or more============= 1.4 0.2 0.2 0.4 1.8 1.8 4.9 6.9
NOTE: See footnotes on table 3.
15
16
APPENDIX |
ITEMS ON THE MEDICAL HISTORY RELATING TO GLUCOSE TOLERANCE
1. a. Do you have any reason to think that you may have diabetes,
sometimes called sugar diabetes or sugar disease? ry
(1F_YES or 7) Bia A
b. Did a doctor tell you that you had diabetes? YES
c. How long ago did you start having it?
[ i voor | ( 1-5 years | over 5 years |
d. Do you take insulin? [xo]
e. (IF TAKE INSULIN:) How many units a day?
f. Do you take any medicine by mouth for diabetes? YES [vol
g. Do you know the name of the medicine?
(name)
h. When did you last visit your doctor for diabetes?
(date)
i. When is your next appointment to visit your doctor for your diabetes?
(date) [J] No appointment
2. a. When did you have your last meal? Time AM [TODAY
PM [YESTERDAY]
YES
b. Did you have meat or fish . . +. + + « « « « « + « + «
Co EQOS OF CHEESE + 4 + v « ¢ ¢ + + « + 4 4 6 a 5 5 +
d. Bread, cereal, potatoes . « + «uv + + + + + 4 + 4.»
O0oca
0000s
e. Cake, pie, sweet rolls, ice cream . . . + + « + o « +
”
w
3. a. Have you had anything to eat or drink since that meal?
(1F YES) What was it? b. Coffee? . . . .. .....
Withcream? . « + « » + « »
With sugar? . . . . ..
00000
00000s
c. Other (Specify) . . . ...
7. a. Have you ever had any children of your own (not
including adopted children)? YES
(LE YES)
b. Did any of your children weigh more than 10 1bs at birthe [ves]
El El
=
Have you had any recent increase in being thirsty
(drink a lot of water)?
70.
Have you had any recent increase in urination
(pass a lot of water)?
71. a. Have you lost any weight recently (without trying to)?
|F YES:
b. How much weight have you lost?____ Tbs.
c. Over what period of time have you lost this weight?
72. a. Has any of your relatives ever had diabetes?
IF YES:
b. Please aive relationship of this person or these persons
to you:
APPENDIX II
DIABETES—DOCUMENTATION
Of the 6,672 sample persons examined, 114
were diagnosed as having definite known dia-
betes and 11 as having questionable known dia-
betes. The persons with questionable known
diabetes gave a history of disease and reported
that the diagnosis had been made by a physician.
Furthermore, all reported having seen a physi-
cian for the disease within the previous 6 months.
None, however, were taking any hypoglycemic
medication. Because they reported they were un-
der close medical supervision, none were given
a glucose challenge. Their blood glucose levels
ranged from 74 mg.% to 122 mg.%. Only one was
found to have urine glucose and his blood glucose
level was 74 mg.%. These cases are excluded
from tables A-C and 1-4.
The 114 persons diagnosed as having defi-
nite known diabetes either reported they were on
medication or were found to have elevated blood
glucose levels. Of the total, 82 were using hypo-
glycemic agents of some sort, 33 using insulin
alone, 5 using both insulin and an oral hypogly-
cemic, and 44 using only an oral hypoglycemic. Of
the 32 persons not on hypoglycemic medication,
24 received a challenge and 8 did not. Blood glu-
cose levels for these 32 persons ranged from 148
to 412 mg.% with challenge and from 138 to 364
mg.% without challenge. Five cases of definite
known diabetes were persons who gave a history
of diabetes but denied that it had been diagnosed
by a physician. Since their blood glucose levels
ranged from 218 to 412 mg.% it was assumed that
these cases had, infact, been medically diagnosed.
The distribution of blood glucose levels in
mg.% in persons having definite known diabetes
was as follows:
Total Legs chen 150-169 | 170-199 | 200-299 | 300+
Not on hypoglycemic
medication
Challenge========-=- 24 1 2 6 il 4
No challenge-=-=---- 8 1 2 3 1 i
On hypoglycemic
medication
Challenge----=-=---- 1 8 1 - - 2 5
No challenge====-=-- 74 34 4 2 18 12
Lor 4 persons no specimen was available.
In order to identify the sample group for the
Health Examination Survey, a household interview
was conducted at each sample household. This
made available a large amount of information both
for persons subsequently examined and for sam-
ple persons who were not examined. Included in
that information were data derived from a mor-
bidity questionnaire.
Reports of diabetes from the household inter-
view are in close correspondence with the final
diagnoses made from the health examination.
Altogether 107 examined persons were reported
to have diabetes on the household interview, as
compared with 125 with definite or questionable
known diabetes on the examination. In 96 cases
the two sources agreed. There were 29 cases
found on examination but not reported on house-
hold interview and 11 cases reported on the inter-
18
view but not diagnosed on the examination. Of the
latter, 2 persons gave a history of diabetes on the
examination but the diagnoses could not be con-
firmed by the evidence available, while 9 persons
gave no such history on the examination. Although
the two sources yield comparable information on
diabetes, the household interview can be con-
sidered as providing a net understatement of the
prevalence of known diabetes in the population.
This is in accord with a previous study of this
subject, which found 88 cases of diabetes reported
by household interview for every 100 identified
from medical sources (National Health Sur-
vey: Health Interview Responses Compared With
Medical Records. Series D-5, PHS Publication
No. 584-D5, Public Health Service, Washington,
D.C., June 1961).
APPENDIX lI
CASUAL ASPECTS OF THE GLUCOSE TOLERANCE TEST
The glucose tolerance test used in the Health
Examination Survey required that the examinee be
given a challenge of 50 grams of glucose shortly
after beginning the examination and that 1 hour
later a venous blood specimen be taken. In that
sense the glucose tolerance test was standardized.
There were a number of respects, however, in
which the glucose tolerance test was not stand-
ardized. For example, an examinee might appear
for examination at any time of the day, from early
morning until late in the evening. Or, he might
arrive either just after eating or many hours after
his last meal. And the content of his last meal, as
well as his usual diet, was entirely uncontrolled
by the Survey. Given all these variables it might
well be asked, "How standardized was the glucose
tolerance test used by the Health Examination
Survey?"
To answer this question, at least in part, the
Health Examination Survey, with the help of staff
members of the Tecumseh Community Health
Study, instituted a special study to investigate the
effect on blood glucose levels of differences in
the size of the glucose challenge, time of day, and
time since last meal.* The study was undertaken
with a group of 24 prisoners who were given a
series of glucose tolerance tests under a variety
of conditions, extending over a period of 16 weeks.
It was found that with a challenge of 50 grams of
glucose the blood glucose level 1 hour after chal-
lenge was affected to no discernible extent by the
time between the last meal and challenge, but that
levels after the midday meal were higher than
levels after the morning meal. It was also found
that any standard test procedure yielded results
comparable to any other standard procedure. Re-
sponse to any given procedure, as with most bio-
logical behavior, tended to vary from one time to
the next.
In part, the same factors canbe examined on
the basis of the examination findings themselves.
All examinees were asked when they had last
eaten. The time of challenge was noted. Mean
blood glucose levels are presented in table I by
sex, in broad age groups, according to the time of
day that the examinee was given the glucose
drink, and according to the interval between his
last meal and the glucose drink. These data are
for examined persons only and do not constitute
estimates for the population of the United States.
The data are restricted to persons who came in
for examination 1 to 4 hours after the meal, since
such persons account for the majority of all ex-
aminees. When differences in blood glucose level
associated with time of challenge and interval
since last meal are measured againstdifferences
between people, the following conclusions are
reached:
1. Persons given 50 grams of glucose 2 to 3
hours or 3 to 4 hours after the morning
meal had higher blood glucose levels after
challenge than persons given the same glu-
cose challenge between 1 and 2 hours after
the same meal.
2. So far as can be judged from the data, no
similar effect is discernible for the mid-
day or evening meals.
3. The blood glucose level after challenge also
varied with time of day. Levels were high-
er after the midday meal than after the
morning or evening meals.
Except for the effect on blood glucose levels
of time after the morning meal, these findings are
consistent with those from the special study4 and
may be considered extensions from the very
restricted and special group of 24 male prisoners
to the population as a whole.
It is of interest to examine table Ifor age and
sex differentials on the possibility that differences
between the various age-sex groups in time of
appearance for the examination may somehow in-
troduce an artifact when the data are consolidated.
This is not the case. Even in data specific for
time of day and time since last meal there is
strong gradient by age and a definite, though
weaker, sex differential, just as there is in the
consolidated data.
19
Table I.
tervals between
vey, 1960-62
last meal
and challenge,
sex,
and age:
Mean blood glucose levels, by time of day challenge was given, specified in-
Health Examination Sur-
Interval between last meal and chal-
lenge, sex, and age
Time of day challenge was given
8-11 a.m. 12-5 p.m. 6 p.m. or later
60-119 minutes in mg.%
Men
18-39 years==--=---mmm-ececcccc ccc 87.4 95.8 98.1
40-54 years===-==-mmememmccccccecce————— 108.5 127.3 110.0
55+ years====m==mmmmmmmmmcee meee ———— 130.6 155.1 106.5
Women
18-39 years--=--=-==mm-cecccccecaaa—- 92.7 105.7 100.5
40-54 years-====m-=mmmmccecccmccce———— 107.7 125.1 116.)
55+ years-=====mmmmmmmm——— er ——————— 118.1 148.5 131.3
120-179 minutes
Men
18-39 years--==-=-=mmmmemceecc cea 103.3 101.0 97.4
40-54 years--=-=-===---ccecccccccce——— 115.1 116.5 113.6
55+ years=====------mmececccme cece 129.1 135.0 117.3
Women
18-39 years-=-=-=---mmmmmeceeecec————— 102.1 108.8 98.8
40-54 years===-===---emececcccccccc———— 121.0 118.7 119.7
55+ years-====mmmmmmmmm——————————————— 146.6 153.5 135.4
180-239 minutes
Men
18-39 years------=---mcccccccccc ee ———— 107.7 99.0 94.5
40-54 yearS§====m=mmmmmemccececccccce———— 119.5 113.6 110.7
35F yeargm =m mmm mm on 130.5 142.8 137.8
Women
18-39 years-=---=---emcccccccccec—————— 110.1 114.0 100.0
40-54 years======m=mmm-ececccceecc———— 144.2 131.5 101.8
55+ years----memmmmmmmm meee ———— 150.3 148.8 124.8
NOTE: Values in this table do not constitute estimates for the population of the United States.
20
APPENDIX IV
QUALITY OF BLOOD GLUCOSE DETERMINATIONS
During the period between January and May
1962, in the course of conducting a special study
of glucose tolerance tests, the Health Examination
Survey instituted a series of quality checks on the
work of the laboratory responsible for the blood
glucose determinations of the Survey —the labora-
tory of the Field Research Unit, Diabetes and
Arthritis Branch, Division of Chronic Diseases,
Bureau of State Services, U.S. Public Health
Service, at Brighton, Massachusetts. The results
of these checks were highly favorable. Full de-
tails are available in the report of that study,*
Both before and after this period, quality
checks of the laboratory determinations at
Brighton had been undertaken in connection with
the routine field collection of specimens. The
first series of checks occurred during the period
between February 9 and March 3, 1961. Aliquots
were obtained of 272 specimens collected rou-
tinely during the field work at San Jose and San
Francisco, California. One aliquot was treated as
a regular specimen and shipped to the Brighton
laboratory for determination. The other was sent
to a special laboratory of the Metabolic Unit of
the University of California by special arrange-
ment with Dr. Peter Forsham. The technicians
at the Brighton laboratory were unaware that a
comparison study was in progress, arrangements
having been made through Dr. Hugh Wilkerson
for this undertaking. As a subsidiary inquiry, 60
specimens were obtained in triplicate, one aliquot
going to the Brighton laboratory, the second going
promptly to the San Francisco laboratory, and the
third being held and sent to the San Francisco
laboratory 6 to 9 days later. The conclusions from
these comparisons were as follows:
1. There was no definite evidence that any
artifacts were introduced in the measure-
ment of blood glucose by HES methods of
transporting the specimens or by the de-
lay between drawing the blood and meas-
uring it.
2. Blood determinations by a single technician
on a single run were highly consistent, in
a sense to be specified later.
3. There were differences in the levels be-
tween technicians, runs, and laboratories;
in other words, the measurement of blood
glucose on the 272 specimens inthis com-
parison was not fully standardized.
4. No change in glucose concentration was
demonstrated even when the specimen was
kept as long as 6 to 9 days before being
measured.
There was a distinct difference in levels be-
tween the two laboratories. The mean glucose
concentrations for the 272 specimens were 117.0
mg.% at the Brighton laboratory and 109.2 mg.%
at the San Francisco laboratory. During this peri-
od, two technicians were working on these speci-
mens at the Brighton laboratory. One tended to
measure close to the level of the San Francisco
laboratory, whereas the other tended to be dis-
tinctly higher; the apparentdifference between the
levels for the two technicians was about 5 mg.%.
The first 104 measurements by one of the
technicians at the Brighton laboratory were com-
pared with measurements on the same specimens
by the San Francisco laboratory. These determi-
nations represented six runs at the Brighton lab-
oratory and nine runs at the San Francisco labora-
tory. If every measurement at San Francisco
were increased by 6.3 percent, 9 out of 10 of the
Brighton measurements would come within 5
mg.% of the San Francisco measurement; that is,
if a fixed difference in measurement level is as-
sumed, there is a remarkably high consistency
between (and consequently within) the measure-
ments at the two laboratories.
The basis for the difference in laboratory
levels was never satisfactorily elucidated. Both
laboratories used essentially the same laboratory
techniques. Both were well controlled. There were
21
no obvious criteria for choosing between them.
Control specimens were sent the two laboratories
and for these the determinations made by the
Brighton laboratory were closer to the alleged
glucose concentrations. On the other hand, the
levels obtained by the San Francisco laboratory
on these specimens tended to be slightly higher
than those obtained by the Brightonlaboratory.
In other words, the comparisons between the lab-
oratories were in the opposite direction from
those that were obtained during the rest of the
series and only confuse the issue.
Interlaboratory comparison is a harsh testof
any laboratory. The general conclusion was that
blood glucose determinations are not so well
standardized as is commonly thought and thatad-
ditional work in standardization is highly desirable.
Although the results of this series were in some
respects equivocal, by any realistic standards
the laboratory work being done on specimens
from the Health Examination Survey was quite
reliable.
22
Between June 1961 and May 1962, a series of
aliquots from specimens collected in the field
were sent at regular intervals to the laboratory
of the Framingham Heart Study, by arrangements
with Dr. Thomas R. Dawber, Director. Except for
one aberrant set of comparisons the Brighton
laboratory averaged slightly higher than the
Framingham. Of more interest, perhaps, is the
variability of measurement. This may be repre-
sented by the statistic s =/W, where w=
2n
4, being the difference between determinations by
the Boston and Framingham laboratories on the
same specimen, and n being the number of speci-
mens. The overall value of s was 8.0 mg.%, or
5.9 mg.% if the one aberrant set were omitted.
When it is considered that this figure includes
variability arising from differences between lab-
oratories, between technicians within laborato-
ries, and between laboratory runs over a period
of 1 year, the results are very encouraging.
APPENDIX V
SURVEY DESIGN, MISSING DATA, AND VARIANCE
The Survey Design
The Health Examination Survey is designed
as a highly stratified multistage sampling of the
civilian, noninstitutional population of the con-
terminous United States, aged 18-79 years. The
first stage of the plan is a sample of 42 primary
sampling units (PSU's) from among 1,900 such
geographic units into which the United States has
been divided. A PSU is a standard metropolitan
statistical area or one to three contiguous coun-
ties. Later stages result in the random selection
of clusters of about four persons from a small
neighborhood within the PSU. The total sample in-
cluded 7,710 persons in the 42 PSU's in 29 differ-
ent States. The detailed structure of the design
and the conduct of the Survey have been de-
scribed in previous reports.!2
Reliability of Probability Surveys
The Survey draws strength from the fact that
it is a probability sample of its total target pop-
ulation, and from the fact that the measurement
processes which were employed were highly
standardized and closely controlled. This does
not mean, of course, that the correspondence be-
tween the real world and survey results is exact.
Data from the Survey are imperfect for three im-
portant reasons: (1) results are subject to sam-
pling error, (2) the actual conduct of a survey
never agrees perfectly with the design, and (3) the
measurement process itselfis inexact, even when
standardized and controlled.
The faithfulness with which the study design
was carried out has been analyzed in a previous
report? Of the 7,710 sample persons, the 6,672
who were examined—a response rate of over 86
percent—give evidence that they are a highly
representative sample of the adult civilian, non-
institutional population of the United States. Im-
putation for the nonrespondents was accomplished
by attributing to nonexamined persons the char-
acteristics of comparable examined persons. The
specific procedure used? consisted of inflating the
sampling weight for each examined person to
compensate for sample persons at that stand and
of the same age-sex group who were not examined.
While it is impossible to be certain that the
prevalence of diabetes is the same in the ex-
amined and the nonexamined groups, the available
evidence indicates that it is. One source of infor-
mation on this question is a household interview
obtained for every sample person. The preva-
lence of diabetes reported for nonexamined per-
sons on the household interview agreed very
closely with that reported for examined persons
of the same age and sex. Another source of infor-
mation is a special inquiry sent to the physicians
of nonexamined persons and to the physicians of a
matching set of examined persons. Again, the
diabetes prevalence reported for the examined
and nonexamined groups was in very close agree-
ment.
In addition to persons not examined, there
were some persons whose examinations were in-
complete in one particular or another. Age and
sex were known for every examined person, but
for a number of people either a blood or urine
specimen was not available. Most of the losses
were accidental. The extent of missing informa-
tion is indicated in table II.
The method for dealing with this missing infor-
mation in tables 1-4 was to attribute to a person
for whom a blood or urine determination was not
available the information available for a com-
parable person with such a- determination. For
example, if a urine specimen was determined but
a blood specimen was not, a person of the same
age and sex and with the same urine glucose find-
ing was selected at random and his blood glucose
determination was used for the missing value. If
a blood specimen was available but a urine find-
ing was not, a person of the same age-sex group
23
with the same blood glucose level was chosen as
a substitute.
In other tables the mean of known values was
used. This assumes that missing values have the
same mean as the present values.
Sampling and Measurement Error
In this report and its appendices, several
references have been made to efforts to evaluate
both bias and variability of the measurement
techniques. The probability design of the Survey
makes possible the calculation of sampling er-
rors. Traditionally the role of the sampling er-
ror has been the determination of how imprecise
the survey results may be because they come
from a sample rather than from measurement of
all elements in the universe.
The task of presenting sampling errors for a
study of the type of the Health Examination Survey
is difficult for at least three reasons: (1) Meas-
urement error and ''pure'' sampling error are
confounded in the data; it is not easy to find a
procedure which will either completely include
both, or treat one or the other separately. (2) The
survey design and estimation procedure are com-
plex and accordingly require computationally in-
volved techniques for calculation of variances.
(3) Thousands of statistics come from the survey,
many for subclasses of the population for which
there are small numbers of sample cases. Esti-
mates of sampling error are obtained from the
sample data and are themselves subject to sam-
pling error, which may be large when the number
of cases in a cell is small. or even occasionally
when the number of cases is substantial.
As variances are estimated for larger num-
bers of statistics from the Health Examination
Survey, it is hoped that an increasing amount of
information can be presented in published re-
ports. In the present report, estimates of ap-
proximate sampling variability for selected sta-
tistics are presented in table III. These estimates
have been prepared by a replication technique
24
which yields overall variability through observa-
tion of variability among random subsamples of
the total sample. The method reflects both "pure"
sampling variance and a part of measurement
variance.
In accordance with usual practice, a 68 per-
cent confidence interval may be considered that
range within one standard error of the tabulated
statistic and a 95 percent confidence interval that
range within two standard errors. An overesti-
mate of the standard error of a differenced=x -
y of two statistics x and y is given by the for-
mula s, = B Vi + y2 Vv? % where V?_ and
d x y , X
, are relvariances respectively of x and y or
the squares of the relative errors shown in table
III. For example, table B shows x = 115.70 mg.%
for men and y=126.35 mg.% for women, while
from table III relvariances are found to be:
Vv = 0.000064 and Le" = 0.000049. The formula
yields the estimate of standard error of the dif-
ference (d =10.65) as 8y=1.3 mg.%. Thus, as the
observed difference is more than eight timesits
sampling error, it can be concluded with near cer-
tainty that the evidence from this Survey| shows
that blood glucose is higher among females than
males.
Small Numbers
In some tables magnitudes are shown for
cells for which sample size is so small that the
sampling error may be several times as great as
the statistic itself. Obviously in such instances
the statistic has no meaning in itself exceptto in-
dicate that the true quantity is small. Such num-
bers, if shown, have been included to convey an
impression of the overall story of the table.
Table II. Number of examined persons, by challenge and diagnosis, and completeness of
glucose tolerance data: Health Examination Survey, 1960-62
Complete glu- Partial geass tolerance
Challenge and diagnosis Total cose tolerance
data Total Blood | Urine Blood
only only | and urine
Total--====cceee-- 6,672 6,410 262 124 134 4
Challenged-=-=======c-==- 6,570 6,314 256 120 132 4
Unchallenged-======c=c=- 102 96 6 4 2 -
Diabeticl-==mwmmmmmmenm 93 88 5 4 1 -
Nondiabetic--=======-=- 9 8 1 - 1 -
Definite or questionable.
Table III. Approximate relative standard errors for selected statistics on glucose tolerance:
Health Examination Survey, 1960-62
Approximate relative standard error (in percent) RC pos ne
Age and sex Uri 1
Diabetes prevalence | Mean blood glu- Tine g Meese
(table A) cose (table B) Pigvalenss Table 2 Table 4
Both sexes-- 8.0 0.4 5.0 8.0 10.0
Men----=====-u-- 12.0 0.8 5.0 10.0 10.0
18-24 years------- (1) 1.5 20.0 (1) (1)
25-34 years==------ (1) 1.5 10.0 40.0 40.0
35-44 years==-==== 40.0 1.5 10.0 30.0 30.0
45-54 years---=---- 40.0 1.3 10.0 30.0 30.0
55-64 years===--=-- 40.0 1.5 10.0 30.0 30.0
65-74 years--=----= 40.0 2.5 10.0 30.0 30.0
75-79 years=--=--- (1) 2.3 20.0 40.0 50.0
Women==========- 12.0 9.1 10.0 12.0 15.0
18-24 years------- (1) 1.5 20.0 (1) (1)
25-34 years=------ (1) 1.5 15.0 30.0 30.0
35-44 years=-=----- 50.0 1:5 15.0 30.0 30.0
45-54 years==-=---=-= 30.0 1.5 15.0 30.0 30.0
55-64 years===-=---- 30.0 1.5 15.0 20.0 30.0
65-74 years==------ 30.0 1.5 15.0 20.0 30.0
75-79 years------- 50.0 4.0 20.0 40.0 40.0
(1) Not estimated.
25
%* U.S. GOVERNMENT PRINTING OFFICE : 1964 O - 727-522
©)
q EE EN al
P= i Te
TEA —=: TF
N - - ws mt
El
A
EHS BE
LIBRARY
Vos 2.
NATIONAL :
rh RUC
For HEALTH Number 3
STATISTICS
Binocular Visual Acuity
i Adults
United States. 1960 - 1962
U. S. DEPARTMENT OF
HEALTH, EDUCATION, AND WELFARE [¢
Public Health Service
See inside of back cover
for catalog card.
Public Health Service Cublication No. 1000-Series 11, No 3
For sale by the Superintendent of Documents, Government Printing Office
Washington, D.C., 20402 - Price 25 cents
NATIONAL CENTER| Series 11
For HEALTH STATISTICS | Number 3
VITALand HEALTH STATISTICS
DATA FROM THE NATIONAL HEALTH SURVEY
Binocular Visual Acuity
of Adults
United States. 1960-1962
Vision testing methods and binocular visual acuity
findings, by age, sex, and race among adults aged
18-79 years.
Washington, D.C. June 1964
U.S. DEPARTMENT OF
HEALTH, EDUCATION, AND WELFARE Public Health Service
Anthony J. Celebrezze Luther L. Terry
Secretary Surgeon General
NATIONAL CENTER FOR HEALTH STATISTICS
Forrest E. Linder, Ph.D., Director
Theodore D. Woolsey, Deputy Director
Oswald K. Sagen, Ph.D., Assistant Director
(for Professional Relations)
Walt R. Simmons, M.A. Statistical Advisor
Alice M. Waterhouse, M.D., Medical Advisor:
James E. Kelly, D.D.S., Dental Advisor
Louis R. Stolcis, M.A., Executive Officer
DIVISION OF HEALTH EXAMINATION STATISTICS
Arthur J. McDowell, Chief
COOPERATION OF THE BUREAU OF THE CENSUS
In accordance with specifications established by the National
Health Survey, the Bureau of the Census, under a contractual
agreement, participated in the design and selection of the sample,
and carried out the first stage of the field interviewing and certain
parts of the statistical processing.
Public Health Service Publication No. 1000-Series 11-No. 3
CONTENTS
Page
Introduction =--=-==-eccccmm cme remem mmm mem 1
Visual Examination---------mccmmmmmm cece eee mmm 1
The Testing INStrument-------==- e-em mme mmm 9
Testing MethodS-===-=mm mm mmm mmm 3
Quality Control--- em eeeeeeemeeeeem em 3
Findings === === moomoo mmm mem ome 3
Uncorrected Distance Visual Acuity------=--cemoeeooooo- 3
"Corrected" Distance Vision-=--=--e-mcmmcmcccmmcccaaan 4
Near ViSion-----emom mmm eee - 4
Age-Sex Differences-------c-commmmmmmmm meee 6
Racial Differences-------=-eememceemcccemcecc cece mmm 6
Comparision With Other StudieS---=-=-=mmcmmomcmommae oo 7
SUMMATY == mmm mmm mmm meee eee eee 11
Detailed TableS-----c-ccmmmmmme ee rere eee 13
Appendix I. Target Specifications and Items on Medical
History Related to Vision Used in This Report-------------- 22
Appendix II. Some Technical Notes on the Vision Test------ 24
Appendix III, Survey Design, Response, and Sampling Varia-
bility Used in This Report--====-momommmmmm cme eee oo 25
The Survey Design---=-c--ccmmmomm meee 25
Reliability in Probability Surveys---==---=--ccmmacmomanaoo 25
Sampling and Measurement EXror-----=--=ceemmeemommaa-- 26
Small Categories=-=-=-mmommom mmm 26
SYMBOLS
Data not available-----cccmmomommmooo
Category not applicable----meecocmamoaaoo
QUATILILY ZEIT O == =m mem mm oe mr mm on mmm mm em
Quantity more than O but less than 0,05-----
Figure does not meet standards of
reliability or precision--------c-cceame--
BINOCULAR VISUAL ACUITY
OF ADULTS
Jean Roberts, Division of Health Examination Statistics
INTRODUCTION
This is one in the series of reports de-
scribing and analyzing the plan, conduct, and
findings of the first cycle of the Health Examina-
tion Survey. This report presents the Survey
results for binocular visual acuity.
The Health Examination Survey from which
these data derive was organized as part of the
National Health Survey to obtain statistics on the
health status of the population of the United States
through direct examination,
The plan and initial program of the Health
Examination Survey have already been recounted
in substantial detail.! A first report on the sur-
vey findings described the demographic com-
position of the sample, the possible effects of
nonresponse on the findings, and the inflation
process used to convert examination findings into
estimates for the adult population of the United
States from which the sample was drawn. 2
In this first cycle, the Health Examination
Survey concentrated on the collection of statistics
for certain of the more prevalent chronic diseases
and on selected physical and physiological meas-
urements among the adult civilian, noninstitutional
population of the United States 18 through 79 years
of age. This phase of the Survey was started in
October 1959 and completed in December 1962.
Out of the defined sample of 7,710 persons, 6,672
(or more than 85 percent) were examined.
A standardized single-visit examination was
given each examinee by medical and other staff
members in the specially designed mobile units
used for the Survey. Prior to the examination,
data comparable to those collected by the Health
Interview Survey were obtained from the house-
holds of the sample persons. In most of the sample
households, every second eligible adult was chosen
for the examination,
Previous reports indicate that no major
feature of the adult population of the United States
can be said to be seriously distorted inthe sample
and that the effects of nonresponse on the demo-
graphic picture are apparently not serious.’
Fewer visual defects were reported oninter-
view among the nonexamined part of the sample
than expected if the prevalence rate of such defects
in the examined and nonexamined groups were
actually the same. On that assumption, these
defects were underreported by about 16 percent
for the nonexamined group. If the same differential
exists for severely defective vision and other eye
conditions obtained from the examination, then
the survey estimates for the total will overstate
the true prevalence figures by only 2 percent.
VISUAL EXAMINATION
Central visual acuity for distance and for near
vision was measured for each examinee as partof
the standardized examination in the first cycle of
the Health Examination Survey. In addition the
Survey staff physician recorded any gross defects
found during the limited examination of the eye.
These physical findings together with the medical
history and household interview information for
the examinee constitute the data on vision available
from this cycle of the Survey.
These data on visual acuity are the first to
be collected for a national probability sample of
the adult population in the United States. Previous
surveys have been limited to information obtained
on interview or from testing of specific population
groups such as insurance policy holders, em-
ployees of large industries, or those in certain
geographic areas.
This report contains estimates of the levels of
binocular distance and near visual acuity by age,
sex, and race. Results are given for tests with
and without corrective lenses in the following two
forms: ,
1. "uncorrected" or ''unaided' acuity refers
to the scores attained without glasses or
other corrective lenses, and
2. "corrected" acuity refers to scores at-
tained with corrective lenses for persons
tested with their glasses together with
scores without correction for those tested
only without glasses either because they
did not bring them to the examination or
do not wear them.
The Testing Instrument
Space and time limitations for the examina-
tion were determining factors in selecting a
commercial instrument, the Sight-Screener, for
testing visual acuity in the Survey. This device,
shownin figure 1, uses the stereoscopic principle
to achieve the optical equivalent of 20 feet for
testing at distance. Near vision is tested at 14
inches without the interpositioning of lenses.
Monocular acuity is measured under conditions
of binocular seeing with the examinee unaware
of which eye is being tested.
The Sight-Screener allows for rapid testing
under controlled conditions of lighting and target
distance from the examinee. The effective illu-
mination on the target and the contrast between
target letters and background were maintained
within the optimum limits for such tests. ?
The acuity target contains three lines dif-
fering ‘only in the sequence of the letters—one
line each for testing the right eye, the left eye,
and binocular vision. Targets for the optical
equivalent of distance and for near vision were
identical. The letters are arranged in blocks or
steps of from one to four letters. The size of the
letters becomes progressively smaller from one
block to the next when reading from left to right,
The unserifed letters of the target follow the
Figure 1. The Sight-Screener.
Snellen principle with their height as well as their
width being five times the width of the lines in
the letters.
Like similar commercial devices the Sight-
Screener is designed for screening purposes —for
pass or fail at certain critical levels, It does not
measure as accurately across the entire range of
vision as would be possible with a good wall chart
or cards. The acuity scale on the targetis coarse
at the poorer levels from 20/200 to 20/50 with
only four steps and few letters, while at levels
critical to qualification (i.e., for service in the
Armed Forces or for certain types of civilian
employment or licenses—20/50 through 20/10),
there are five steps with four letters at each level.
The testing levels on the targets were as follows:
Distance—20,/200, 20/100, 20/70, 20/50,
20/40, 20/30, 20/20, 20/15, 20/10.
Near—14/140, 14/70, 14/49, 14/35, 14/28,
14/21, 14/14, 14/10.5, 14/7.
Despite these limitations, test results on the
Sight-Screener and on Sloan Charts (an im-
proved Snellen-type) were found to be in good
agreement, although slightly lower on the former
because of the coarser scale at the poorer acuity
levels.”
Testing Methods
Right eye, left eye, and binocular vision were
always tested inthatorder. However, the sequence
of near and distance tests was alternated for suc-
cessive examinees—the first started with near
tests, the second with the distance tests. Test
order was so randomized as to minimize any con-
sistent bias for either test series due to fatigue,
practice, or learning of target letters. The meth-
odological study gave no indication that these
factors had a demonstrable effect in test results.
To "pass' or be able to read at a particular
level no errors were allowed if the block con-
tained fewer than four letters and only one error
in steps of four letters. The visual acuity level or
"score'' for an examinee is that which corresponds
to the smallest letters he was able to read with
no more than the allowable number of errors.
Quality Control
After joining the examining staff, each of the
five dentists employed during the cycle was given
training and practice in vision testing techniques
to insure the consistency of test results. Further
practice in testing was obtained during the "dry
run’ examinations which preceded the start of
the regular examinations at each of the 42 areas
in which the mobile Health Examination Centers
were located.
During the survey, two of the examining
dentists carried out a pretest with a group of 144
boys at the National Training School for Boys both
to assess the effect of the standard dental light on
the vision test scores and to determine the com-
parability of their vision test results. The group
was tested by both dentists, half before the dental
examination and half immediately afterward. The
pretest gave no indication that exposure to the
dental light prior to the vision test affected the
acuity scores. Hence, it was assumed that testing
of vision immediately after the dental examination,
as was done throughout this survey, did not ap-
preciably affect visual acuity scores. Acuity test
results obtained by the two dental examiners were
also found to be in good agreement. Comparison
of results obtained by each tester at the stand
locations further indicate that testing hdd re-
mained consistent throughout the cycle, The pro-
portion rated as having normal or better vision
showed essentially no differences attributable to
the testers when the age-sex differences among
examinees at the various stands were removed.
FINDINGS
Uncorrected Distance Visual Acuity
Health Examination Survey findings indicate
that more than half (54 percent) of the civilian,
noninstitutional population of the United States
aged 18 through 79 years have at least normal
central binocular visual acuity at distance when
tested without corrective lenses as shown in
tables A, 1, and 2. About 30 percent have better
than normal vision, attaining levels of 20/15 or
20/10 in Snellen notation (''mnormal' distance
vision in this notation is generally considered to
be 20/20).
The median unaided visual acuity is at the
20/19 level. Hence, half of the adult population
are able to read at 20 feet letters of a size that
persons with normal vision could be expected to
read at 19 feet.
One-fourth of the adults have moderately
defective vision without glasses, reading at best
Table A. Proportion reaching or exceeding
the test levels for distance vision:
United States, 1960-62
Proportion for distance
vision
Test level
Un- " "n
corrected Corrected
20/10 or better-- L.1 1.5
20/15 or better-- 30.3 40.0
20/20 or better-- 53.9 72.9
20/30 or better-- 69.3 90.6
20/40 or better=-- 75.8 95.1
20/50 or better=-- 80.4 96.8
20/70 or better-- 83.9 97.7
20/100 or better- 93.5 99.2
20/200 or better=- 97.6 99.6
no further than the 20/30, 20/40, or 20/50 level.
The majority of these persons (15 percent) score
just short of normal, at the 20/30 level.
The remaining one-fifth of the adults test at
the 20/70 level or less. Included with this latter
group are an estimated 2.6 million or 2 percent
who have binocular distance acuity of less than
20/200.
“Corrected” Distance Vision
As used in this report, "corrected" vision
denotes functional acuity or the level at which the
adults are actually seeing with whatever cor-
rection they are using.
Forty-four percent of the examinees were
tested at distance with and without their glasses.
This represents essentially all persons who stated
they wore glasses for distance vision. Glasses
improved acuity for 76 percent while 19 percent
tested the same with glasses as without, and 5 per-
cent did better without their glasses. A few of
this latter group were inadvertently tested at
distance with refraction intended for near vision.
The remaining 56 percent of the examinees
tested at distance only without glasses had acuity
scores distributed over the entire test range.
Substantially more of them had at least normal
vision than was true for persons with glasses
(when tested without correction)—76 percent com-
pared with 30 percent.
The resultant improvement in acuities with
correction is clearly evident in tables A, 3, and 4,
particularly for those with defective, unaided
vision of 20/70 or better.
Survey findings as shown indicate thatnearly
three-fourths (73 percent) of the adult population
have normal or above normal vision with whatever
correction they are using. The median score was
20/16.5 compared with 20/19 for uncorrected
acuity.
Over 90 percent reached the 20/30 level or
better with '""correction' compared with 69 percent
for unaided vision.
The proportion unable to read at the 20/200
level (0.4 percent) is too small to give a reliable
estimate for this segment of the population. Yet
it can be said with a fair degree of certainty that
the actual proportion in the adult population prob-
ably does not exceed 1 percent. This group will
include the legally blind as well as those whose
vision could be corrected to normal or near nor-
mal. However, neither the testing nor the exam-
ination procedures in this cycle were sufficient
to provide the basis for making a more precise
estimate of the prevalence of blindness.
Near Vision
Near acuity, both 'corrected" and uncor-
rected, was more deficient among these adults
than their distance vision—a finding to be expected
because of the known physiological effects of aging
on the normal eye. In May's Manual of the Dis-
eases of the Eye it is stated that the power of
accommodation needed to bring near objects into
clear focus gradually diminishes with age, due
chiefly to loss of elasticity of the lens, The phys-
iological change becomes most pronounced when
nearing the age of 45. Distance vision is not
similarly affected.
Forty-five percent had at least normal un-
aided near vision and 63 percent tested at the 14/35
level or better (tables B and 5-8) as compared
with the 54 percent and 80 percent reaching similar
levels at distance without refraction (table A).
(Normal near vision in Snellen notation as used
here is generally considered to be 14/14.)
Over half of the examinees (52 percent) were
tested both with and without glasses for near
vision. (An additional 4 percent stated they wore
Table B. Proportion reachingor exceeding
the test levels for near vision: United
States, 1960-62
Proportion for near
vision
Test level
Un- 1" "n
corrected Corrected
14/7 or better--- 1.0 |
14/10.5 or better 24.0 29.8
14/14 or better-- 44,7 64.9
14/21 or better-- 53.6 84.7
14/28 or better-- 58.3 90.9
14/35 or better-- 62.7 93.7
14/49 or better-- 68.2 95.6
14/70 or better-- 83.9 98.6
14/140 or better- 95.7 99.6
20/100
20/70
20/50
20/40
20/30
20/20
20/15
20/10
14/70
14/49
14/35
14/28
14/21
14/14
14/10.5
14/7
DISTANCE -UNCORRECTED
Median for total group
— Male (20/185)
smmmm= Female (20/20)
Age (in years)
20 40 60 80
—— NEAR-UNCORRECTED
8
\
Median for total $roup NS
= — Male (14/16 ¢ *%
mummmms Female (14/20) ¢
4
4
! -—
4
4
J
4
J wy
i)
J
I —
J
4 a)
[/
|= =
|
20 40 60 80
20/100
20/70
20/50
20/40
20/30
20/20
20/15
20/10
14/70
14/49
14/35
14/28
14/21
14/14
14/10.5
14/7
DISTANCE - CORRECTED
Median for total group
— Male (20/16)
=u: Female (20/17)
=,
2
20 40 60
NEAR - CORRECTED
Median for Sctal ou
— — Male |
mamma Female (14/13)
80
20 40 60
Age (in years)
80
Figure 2. Median binocular acuity among adults, by age and sex.
730-774 O - 64 - 2
glasses for near work butdid not bring them to the
examining center.) Of those tested with glasses or
contact lenses, 83 percent had improved acuities
with correction, 14 percent were unchanged, and
3 percent did less well with than without their
glasses.
As for distance vision, substantially more of
those tested only without correction had at least
normal unaided near vision—74 percent com-
pared with 30 percent for those withglasses when
tested without them.
With "correction," as defined for this report,
65 percent reached at least the normal level of
14/14 or better—10 percent less than for ''cor-
rected" distance vision.
Age-Sex Differences
Survey findings show relatively better unaided
distance and near vision for men than for women.
With "correction," the differences are essentially
eliminated (fig. 2).
Significant differences are evident at the ex-
tremes of the range, accounting for the divergent
medians shown in the charts. More men than
women exceeded normal, testing without cor-
rection at 20/15 or 20/10 for distance and 14/10.5
or 14/7 for near, Conversely, women outnumbered
men at the poorer levels of 20/70 or less and 14/49
or less (tables 1 and 3).
The decline of acuity with age is clearly
evident in these charts for both men and women.
The proportion with at least normal vision starts
dropping rapidly after 45 years of age, with the
percentage of men at this level exceeding women
in each age group.
With distance vision, the proportion testing
normal or better without correction falls from
70 percent for men and women under 45 years of
age to less than 10 percent for those 65 years and
over. A similar pattern is evident in the ''cor-
rected’ scores.
The regression with age started a little
earlier (between 35 and 44 years) in uncorrected
near vision, Here, a more precipitous decline was
found than for distance, and few persons over
age 55 were able to attain normal vision without
correction,
At the other extreme (20/70 or less), the
proportion with poorer distance acuity increases
with age and remains consistently greater for
women than men. Less than 10 percent have such
defective vision under the age of 45, while by the
age of 65 more than 35 percent of the men and over
50 percent of the women tested no better than 20/70
without glasses.
Near vision scores show an abrupt change be-
tween ages 35 and 45. In this age span, the pro-
portion with no better than 14/49 vision acceler-
ates from less than 15 percent toabout 60 percent
for both men and women. The sex difference by
age was less pronounced for near than for distance
vision.
Racial Differences
Comparisons are limited here to acuity
findings for Negro and white persons since the
sample was too small to allow for adequate
representation of other nonwhite races.
No consistent racial differences were found
in the prevalence of normal or better unaided
vision either at distance or near as shown in
tables C and D. The median scores attained by
Negro and white persons are also similar through-
out the age range for both men and women.
If the lower extreme of the range of distance
vision is considered, then white men and women
would be found to have relatively more with poor
distance vision (20/70 or less) at each age—the
pattern more pronounced for men than women as
evident in figure 3. A similar trend does not exist
for near vision. On these latter tests the proportion
of white males with such defects exceeded Negro
males at 25-34 and 55-79 years, while among
women an excess of Negroes was found at 45-54
and an excess of whites at 65-79 years.
No such consistent pattern may be seen atthe
normal end of the range. Moreover, there are
noticeable dissimilarities between men and women
in what trend does exist, Relatively more white
than Negro men ages 18-24 and 35-44 years have
at least normal distance vision, while Negro men
are in excess at ages 25-34, 45-54, and 65-79
(fig. 4). Among women with normal distance vision,
there are a disproportionate number of white
women ages 25-34 and 45-64, while more Negro
women than would be expected were found in the
ages 18-24 and 65-79,
Racial differences are less marked and even
less consistent for near unaided vision.
Table C. Distribution of adults reaching
or exceeding specified acuity levels for un-
corrected binocular distant vision, by sex, age, and race: United States, 1960-62
Acuity level
Sex and age 20/20 or better | 20/50 or better | 20/200 or better
White Negro White Negro White Negro
Men
Total-18-79 years======--- 57.3 539.9 83.7 93.2 98,5 99, 1
18-24 year§=-=-mmmmmmmm_-————————— 80.2 75.4 92,2 97.8 98.8 100.0
25-34 year smmemmmmm————— 79.3 85,6 90.4 98.5 99, 1 100.0
35-44 Year Smmmmmmm mm —————————— 80.5 76.1 93.4 94.4 99..3 98.2
45-54 yearSe--m-mmmmmmmm——————— 49,5 55.7 85.9 97.4 99,1 100.0
55-64 yearsS==mmmmmmmmm—————————— 23.1 23.0 72.0 84.9 I 08.1
65-74 year§-==-mmmmmmmm————————— 8.8 15.3 58.4 74.6 95,7 100.0
75-79 years---m-mmmmmmm————————— 1.3 53.1 78.8 93.2 91.3
Women
Total-18-79 years========= 50.4 52.9 75.5 84.5 96.9 96.3
18-24 years-=-mmmmmm———————————- 71.3 78.9 88.0 96.3 97.4 100.0
25-34 years=mmmmmmmmm——————————— 76.2 71.5 90.6 93.4 96.6 97.0
35-44 yearS==mm=mm=mm-=m———————-= 74.1 73.4 91.5 96.0 98.3 99.3
45-54 year§==mmmmmmmm——————————— 40.6 27.7 76.9 81.7 98.0 95,7
55-64 yearS-=mmmmmmmme-—————————— 17.8 12,8 53.3 56.1 98.4 90.6
65-74 yearSe=-mmmmmmmmm————————— 2.4 10.2 38.7 30.5 91.1 89.4
75-79 year fmm mmmmm mm —————————- 1.8 - 30.4 58.7 91.0 87.3
Corrected acuities were significantly better
for whites than Negroes among both men and
women on distance and near vision. On distance
tests, 74 percent of the whites as compared with
62 percent of the Negroes rated normal or better
with their corrected’ vision. Scores with "cor-
rected! vision or near tests were normal or better
for 66 percent of the whites as compared with 53
percent of the Negroes. The proportion with at
least normal vision among whites exceeded Ne-
groes at each age for distance and from 35 years
on for near vision.
COMPARISON WITH OTHER STUDIES
While many surveys have been undertaken in
which a determination of the distribution of visual
acuity was attempted, they have all been limited
to selected groups of the population—industrial
employees, life insurance policy holders, selected
groups of older persons, and Armed Forces per-
sonnel, to mention a few. In addition, measure-
ment techniques used inthe various studies differ.
The present survey is the first in which
measurements of visual acuity were obtained for
a probability sample of the entire adult civilian,
noninstitutional population of the United States
under the age of 80. As indicated, testing was
done under as near optimum conditions of target
illumination, end-point or scoring criteria, and
target distance as possible. The methodological
study showed that with the survey methods used,
the scores attained on the Sight-Screener were in
general comparable to those obtained on Sloan
Charts (an improved Snellen-type chart).
Table D. Distribution of adults reaching or exceeding specified acuity levels for un-
corrected binocular near vision,by sex, age, and race: United States, 1960-62
Acuity level
Sex and age 14/14 or better 14/35 or better 14/140 or better
White Negro White Negro White Negro
Men
Total-18-79 years==-======- 47.4 47.7 63,5 67.6 96.5 97.4
18-24 years===-=mm-mmmmemeeeen——— 87.5 89.3 96.3 100.0 99.5 100.0
25-34 yearS====mmmmmemememmem————— 85.9 93.7 94.0 98.5 99:3 100.0
35-44 yearS-=-m=mremmmmemmm———————— 74.9 76.3 93.4 94.4 99.7 99,2
45-54 yearS-===--mmmmmmmm————————— 13,2 3.4 40.9 36.2 9753 98.1
55-64 yearS-m==mmmmmmmmm——————— 0.9 - 16,7 25.6 92.2 96.5
65-74 yearS-=-mmmmmmm————————— - - 14.0 25.6 86.0 85.1
75-79 years~==mmmmmmmm—m—————— 7.6 - 27.8 40.2 86.8 78.8
Women
Total-18-79 years====-=---- 41.7 45.6 60.6 65.7 94.9 94.3
18-24 years=-=-=mm=mmmmemce————— 79.4 85,7 96.0 29.0 29.5 100.0
25=30 YJEAT Swen 82.8 72:9 95.0 93.7 99.1 98.0
35-44 yearS==mmmmmmrmmmmm——————— 63.3 63.0 86.7 85.8 99,3 99.3
4554 Year gnmmmmmmmmmmmmen nn 7+3 4,7 37.3 32.5 95,1 89.8
55-64 yeargmmmmmmmmmm mmm 0.8 - 15.8 12.7 89.9 89.5
65=74 years=mmmmmrmmmmm——-——— - - 13.3 20.0 81.7 77.8
75-79 yearsS==m=mmmmmmm - - 2.8 9.1 83.3 72,3
Comparison is made here with findings from
a few of the larger studies.
Hirsch? obtained measurements of visual
acuity on nearly 1,700 persons age 40 through 80
and over in a sample selected from private
practice in a small urban-rural California com-
munity supplemented by some 50 blind pensioners
and other patients with subnormal vision. Roughly
200 persons were included for each of the seven
S-year age groups from 40 through 74 and about
130 in each of the older age groups—75-79 and
80 and over. Published reports donot describe the
testing techniques in detail, but apparently Snel-
len-type charts were used in determining the best
corrected distance vision. As indicated below the
acuities obtained for Hirsch's series are sub-
stantially better than those from the National
Health Examination Survey:
HES data
Paea 24 "Corrected"
Acuity level acuities
(45-79
ears) (45-79
y years)
Percent distribution
20/20 or better-- 73 53
20/30 to 20/50--- 20 41
20/70 or less---- 7 6
This difference would be expected since the
present survey obtained acuities with the exam-
inee's present correction whereas the patients
from private practice were tested with the best
possible refraction.
Wilson and McCormick” obtained the pro-
portion with corrected acuities of less than 20/40
in each eye for over 10,000 employees of the B.F.
Goodrich Company ranging in age from under 21
to over 60 years. As in the present survey, the
Sight-Screener instrument was used for testing.
In the Goodrich study 29 percent of the men and
23 percent of the women tested less than 20/40.
MEN
100 ———— DISTANCE-UNCORRECTED
Percent for total group
80 — — White 16.3 %
sm Negro 6.8 %
Age (in years)
60 |— —]
+
2
w
oO
«©
w
a
0 (ET wt |
20 40 60 80
100 —— NEAR-UNCORRECTED
Percent for total group
80 — mm White 36.5 % eel
emm Negro 32.4 %
60 — wr}
=
z
w
©
x
w
a
40 — _—
[J
20 — ’ s—y
=
0 lh mmm==—" | |
20 40 60 80
WOMEN
100 ———— DISTANCE-UNCORRECTED
Percent for total group
80 — — White 24.5 % =
wmm Negro 15.5 %
remem mms?
ok L
20 40 60 80
100 —— NEAR-UNCORRECTED
Percent for total group
80 |— mw White 39.4 %
sme Negro 34.3 %
60 —
40 youd
20 — —
0 |e | |
20 40 60 80
Age (in years)
Figure 3. Percent of adults with 20,70 or less (or 14/49 or less) binocular acuity, oy age, sex, and race.
Present survey findings show only 5 percent of
both men and women of this age range unable to
reach that level with "correction." Even when
comparison is made with monocular acuity scores,
Health Examination Survey findings for the entire
adult population show substantially better acuities
in general than were found among Goodrich em-
ployees. More restrictive scoring criteria in the
industrial survey may account for part of this
difference.
MEN
100 ——————— DISTANCE - UNCORRECTED
- Percent for total greup
80h Fadi” — White 57.3 % |
’ =m mmm Negro 599 %
60 —
—-
z
wl
o
@
w
a
40 —
20
ol
20
100 —————— NEAR-UNCORRECTED
Percent for total group
80 — — White 474 % id
we Negro 47.7 %
60 =
fie
z
w
o
@
&
40 | —
\
20 + \ -
\
\
1
\ J
o LI | Se i.
20 40 60 80
Age (in years)
WOMEN
100 DISTANCE - UNCORRECTED
Percent for total group
80 | sm White 50.4 % _
So emus NEQro 529 %
60 (—
40 —
20
oll
20 40 60 80
100 NEAR-UNCORRECTED
.,
\ Percent for total group
80 — — White 41.7 % sr
\ smmiNegro 456 %
\
60 —
40
20
o Ld
20
Age (in years)
Figure 4. Percent of adults with at least normal binocular visual acuity (20/20 or better on distance and 14/14 or better for near) by age,
sex, ond race.
10
Collins and Pennell’ reported on the extent of
defective vision (less than 20/20 among 112,000
white life insurance policy holders. He found that
45 percent at ages 30-34 did notobtain 20/20 with
each eye and that the percentage increases most
rapidly at about age 45, then tends to level off at
about 80 percent atage 60. A differentpattern may
be seen in the current survey findings. Here only
25 percent of those aged 30-34 tested less than
20/20 without correction, and the percentage con-
tinues to increase steadily from ages 45 through
79 with no leveling off near age 60. If comparison
was made with monocular findings from the
present survey, the differences in the percentages
at ages 30-34 would have been reduced somewhat.
However, this would not account for the dissimilar
trends with age.
In his analysis of racial differences for visual
acuity among 273,000 Selective Service regis-
trants in 1957 and 1958, Karpinos' found better
vision for Negroes than whites, in contrast with
the findings from the present survey as indicated
below:
Karpinos' number per 1,000
male examiness 18-26 years
Acuity level (in at least one eye)
White Negro
20/20 or better==--===----- 780 888
20/50 or better=--==-=-=-- 889 969
20/200 or better==--==-=---- 961 993
HES number per 1,000 male
examinees
Acuity level (binocular vision)
(18-24 yrs.) (25-34 yrs.)
White Negro White Negro
20/20 or better===----=---- 802 754 795 856
20/50 or better=----- “=== 922 978 904 985
20/200 or better==-=------- 988 1,000 991 1,000
It is apparent that if acuities from the present
survey were tabulated for ages 18-26 there would
be less difference between the two races, and the
proportion with at least normal vision among the
Negroes would not exceed that for the whites.
SUMMARY
Health Examination Survey results from test-
ing visual acuity show that among the United
States civilian, noninstitutional population aged
18 through 79 years:
1. Over half have normal or better distance
vision without correction and more than
three-fourths with whatever refraction
they were using at the time of the survey.
2. Near vision tends to be more deficient
than distance vision, as expected because
of the known physiological effects of aging
on the normal eye.
3. Men have better unaided vision than
women at both distance and near.
4, Visual acuity declines with age from about
45 years on, with the percentage of men
with normal or better vision exceeding
women throughout the age range.
5. Regression with age starts alittle earlier
with near than with distance vision.
6. No consistent racial differences were
found in the prevalence of normal or better
unaided vision either at distance or near
for men or women throughout the age
range. However, corrected near and dis-
tance acuities were significantly better
for white men and women than for Negro
men and women.
11
REFERENCES
1u.S. National Health Survey: Plan and initial program of the
Health Examination Survey. Health Statistics. PHS Pub. No. 584-
A4. Public Health Service. Washington, D.C., May 1962.
2National Center for Health Statistics: Cycle I of the Health
Examination Survey, sample and response. Vital and Health Sta-
tistics. PHS Pub. No. 1000-Series 11-No. 1. Public Health Service.
Washington, D.C., Apr. 1964.
3Sloan, L. L.: Measurement of visual acuity. A.M.A. Arch.
Ophth. 45:704-725, June 1951.
Sloan, L. L.: New test charts for the measurement of visual
acuity at far and near distances. Am. |. Ophth 48(6):807-813, Dec.
1959.
SNational Center for Health Statistics: Comparison of two vi-
sion-testing devices. Vital and Health Statistics. PHS Pub. No.
1000-Series 2-No. 1. Public Health Service. Washington, D.C.
June 1963.
6perera, C. A., editor: May’s Manual of the Diseases of the
Eye. 22d edition. Baltimore, Md. Villiams and Wilkins Co., 1957.
Hirsch, M. J., and Wick, R. E.: Vision of the Aging Patient.
Philadelphia and New York. Chilton Co., Book Division, 1960.
8Wilson, R. H., and McCormick, W. E.: Visual acuity--results
of a survey of 10,000 persons. Ind. Med. and Surg. 23:64-72, Feb.
1954.
collins, S. D., and Pennell, E. H.: The use of the logistic
curve to represent the prevalence of defective vision among persons
of specific ages above 30 years. Human Biol. 7:257-266, May 1935.
10Karpinos, B. D.: Racial differences in visual acuity. Pub.
Health Rep. 75(11):1045-1050, Nov. 1960.
Hy ythgoe, R. J.: The measurement of visual acuity. Medical
Research Council, Special Report Series No. 173. London. His
Majesty’s Stationery Office, 1932.
000
Table 1.
: DETAILED TABLES
Number of adults reaching specified acuity levels for uncorrected distance vision,
by age and sex: United States, 1960=62========m eco ommeeccmeeeeem
Percent distribution of adults reaching specified acuity levels for uncorrected
distance vision, by age and sex: United States, 1960-62--=--=-ccccommcmccccacaaax
Number of adults reaching specified acuity levels for 'corrected" distance vision,
by age and sex: United States, 1960-62-======mcmcomm mecca em
Percent distribution of adults reaching specified acuity levels for 'corrected"
distance vision, by age and sex: United States, 1960-62--------ccocmmccmcccccaaa-
Number of adults reaching specified acuity levels for uncorrected near vision, by
age and sex: United States, 1960-=62-======m==mom comme emcee
Percent distribution of adults reaching specified acuity levels for uncorrected
near vision, by age and sex: United States, 1960=62===-===cccecccmmccmmmccacaaaa.
Number of adults reaching specified acuity levels for ''corrected'" near vision, by
age and sex: United States, 1960=62==== === =o moomoo eee
Percent distribution of adults reaching specified acuity levels for 'corrected'
near vision, by age and sex: United States, 1960=62==---ccececcmmmmmcccccccccaaaa
Page
14
15
16
17
18
19
20
2)
13
Table 1. Number of adults reaching specified
acuity levels for uncorrected distance vision, by
age and sex: United States, 1960-62
Total
7 3 18-24 25-34 35-44 45-54 55-64 65-74 75-79
Sex and acuity level RS years years years years years years years
Both sexes Number of adults in thousands
Total--=====ce==== 111,087 15,569 21,572 23,698 20,576 15,637 11,164 2,871
20/10 or better========-= 1,236 277 460 356 143 - - -
20/15-=====mmmmmmm mmm mm 32,286 7,516 10,364 10,043 3,603 658 102 -
20/20=======mmmmmmm—— me 26,168 3,847 5,371 7,800 5,440 2,627 541 42
20/30-===-==mmmmmm mmm 17,080 1,623 2,077 2,576 4,851 3,418 2,160 375
20/40======mcmemmmm mmm 7,298 567 534 743 1,606 1,852 1,490 506
20/50====mmmmmm mmm 5,125 292 319 455 1,294 1,302 1,110 353
20/70=====-mmmm mmm ——— 3,898 163 337 349 788 1,106 767 388
20/100-=====memmmem mmm 10,742 560 605 653 1,758 3,153 3,208 805
20/200========cmmmm me 4,592 465 507 446 777 1,200 1,025 172
Less than 20/200--=--==-- 2,662 259 498 277 316 321 761 230
Men
Total-======e=e==-= 52.744 7,139 10,281 11,373 10,034 7,317 4,972 1,428
20/10 or better=====-==--- 764 126 332 199 107 - - -
20/15-======cccmmmm me 17,792 3,975 5,489 5,577 2,305 387 59 -
20/20--=-====-cmmmm————— 11,77) 1,376 2,376 3,292 2,630 1,477 402 18
20/30-========ccmmm————— 8,185 674 765 995 2,296 2,039 1,164 252
20/40===m=mcmmmm mmm me 3,658 197 231 341 849 881 802 357
20/50=======mmmm mm ————— 2,432 92 156 220 568 706 531 159
20/70=====cmmmmmmmm mm 1,673 83 173 182 332 474 347 82
20/100-=======-mcmmm———— 4,159 186 383 314 586 1,116 1,145 429
20/200=======cccmmm————— 1,495 157 245 165 279 297 321 31
Less than 20/200-======-- 815 73 131 88 82 140 201 100
Women
Total=======ce===- 58,343 8,430 11,291 12,325 10,542 8,120 6,192 1,443
20/10 or better==-======--= 472 151 128 157 36 - - -
20/15======mcemcmmme——— 14,494 3,541 4,875 4,466 1,298 271 43 -
20/20-======cccmmmmm———— 14,397 2,271 3,495 4,508 2,810 1,150 139 24
20/30-==-=-=memmmmmm———— 8,895 949 1,312 1,581 2,555 1,379 996 123
20/40=====mmmmmmmc mmm 3,640 370 303 402 757 971 688 149
20/50-====-mmmmm mmm mmm 2,693 200 163 235 726 596 579 194
20/70===-mmmmmmm mmm 2,225 80 164 167 456 632 420 306
20/100-=======cmcmmm—— me 6,583 374 222 339 1,172 2,037 2,063 376
20/200-===-==mmmmmm— 3,097 308 262 281 498 903 704 141
Less than 20/200=======-- 1,847 186 367 189 234 181 560 130
Table 2.
1960-62
Percent distribution of adults reaching specified acuity levels for uncorrected distance
vision, by age and sex: United States,
Total
. 20’ 18-24 25-34 35-44 45-54 55-64 65-74 75-79
Sex and acuity level igo years years years years years years years
Both sexes Percent distribution
Total=====meeccaa- 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
20/10 or better--==-=-=--- 1.1 1.3 2.1 1.5 0.7 - - -
20/15======mmmmmmmm emma 29.2 48.3 48.0 42.3 17:5 4.2 8.2 -
20/20=====mmmmmmmmm mma 23.6 24.7 27.2 32.9 26.4 16.8 4.8 1.5
20/30=======mmmmmmmmmmme 15.4 10.4 9.6 10.9 23.7 2).8 19.3 13.1
20/40=======cmmm meme 6.5 3.6 2.5 3.1 7.8 11.8 13.3 17.6
20/50=======mmmmmmm mma 4.6 1.9 1.5 1.9 6.3 8.3 9.9 12,3
20/70========mcmmmmmeeam 3.5 1.0 1.6 1.5 3.8 Zul 6.9 13.5
20/100=======ccmmmmmemam 9.6 3.6 2.8 2.3 8.5 20.2 28.9 28.0
20/200========mmmmmmmmn 4,1 3.0 2.4 1:9 3.8 2537 9.2 6.0
Less than 20/200-------- 2.4 13.7 2.3 1.2 1.5 2.1 6.8 8.0
Men
Total-====memmee—a- 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
20/10 or better=--------- 1.4 1.3 3.2 17 Low - - -
20/15======mmmmmmmme ema 33.9 554.6 53:53 49.0 23.0 5.1 1.2 -
20/20=======mmmmmmemmmnm 22.4 22.1 23.1 29.0 26.1 19.6 8.1 1.2
20/30=======cmmeemmmmme 15.5 9.4 7.4 8.8 22.9 27.2 23.4 17.6
20/40=====mmmmmmmm meee 6.9 2.3 2:2 3.0 8.5 11.7 16.1 25.0
20/50=====mmmmmmmmm meee 4.6 1.3 1.5 1.9 547 9.4 10.7 11.1
20/70========mcmmmmmman 3.2 1.2 1.7 1.6 3.3 6.3 7.0 5.8
20/100========mmmmmmmmmm 7.8 2.6 3.7 2.8 5.8 14.8 23.0 30.1
20/200-=======mmmmmm——n 2.8 2.2 2.4 1:4 2.8 4.0 6.5 2.2
Less than 20/200-------- 1.5 1.0 1.3 0.8 0.8 1.9 4.0 7.0
Women
Total=====ceeme——- 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
20/10 or better==--==-=--- 0.8 1.8 1.1 1.3 0.4 - - -
20/15=======mcmmmmcm mma 24.9 42.0 43.2 36.2 12.3 3.3 0:7 -
20/20=====mmmmmmmmm meen 24.8 26.9 31.0 36.5 26.7 14.2 2.2 1.6
20/30======mmmmmmm mm mmmm 15.2 11.3 11.6 12.8 24.2 17.0 16.1 8.5
20/40=====mmmmmm mmm 6.2 4.4 2,7 3.3 7.2 12.0 11.1 10.3
20/50=======cmmcmmmm meee 4.6 2.4 1.4 1.9 6.9 7+3 9.4 13.4
20/70=====mmmmmmm mmm mmm 3.8 1.0 1.4 l.4 4.3 7.8 6.8 21.2
20/100-======mcmmmmmmmmm 11.2 4.4 2.0 2.8 11.1 25.1 33.3 26.1
20/200-=======mmmmmmmmmm 5.3 3.6 2.3 2.3 4,7 11.1 11.4 9.8
Less than 20/200-=------- 3.2 2.2 3.3 1:3 2.2 2.2 9.0 Po
15
Table 3. Number of adults
reaching specified acuity levels
age and sex: United States, 1960-62
for "corrected" distance vision, by
Total,
¥ 18-24 25-34 35-44 45-54 55-64 65-74 75-79
Sex and acuity level xz years years years years years years years
Both sexes Number of adults in thousands
Total-=====ceece==- 111,087 15,569 21.572 23,698 20,576 15,637 11,164 2,371
20/10 or better=-==-=-=-=--- 1,635 373 553 414 194 76 25 -
20/15=====mmmmmmmm mmm 42,638 8,999 12,472 12,483 6,403 1,747 487 47
20/20======cmmm mmm ———— 36,505 4,277 6,314 8,421 8,001 6,022 3,084 386
20/30=======mmmm———————— 19,774 1,588 1,605 1,677 4,222 5,233 4,344 1,105
20/40-====mcmmmmmem mm 5,004 215 319 342 869 1,334 1,407 518
20/50=====--ccmmmmm me 1,971 64 57 180 313 402 687 268
20/70======mmmm mmm mmm 969 - 89 78 105 246 225 226
20/100=====mmmmmm mmm mmm 1,702 7 106 49 332 391 590 227
20/200-=======cccmmm—— me 423 29 42 21 75 114 104 38
Less than 20/200--=----=-- 466 17 15 33 62 72 211 56
Men
Total-========e==- 52,744 75:39 10,281 11,373 10,034 7,517 4,972 1,428
20/10 or better==-=-=--=-=--- 1,001 209 411 225 144 12 - -
20/15-=====ccmmmmmm mmm 22,442 4,373 6,525 6,627 3,734 913 223 47
20/20---====-cmmmm—m—— = 16,355 1,815 2,564 3,516 3,676 2,947 1,613 224
20/30-=======ccmmmmi eae 8,355 611 554 664 1,806 2,440 1,737 543
20/40====-=mm mmm meen 2,254 116 125 162 317 659 593 282
20/50=====-=cmmmm mmm 905 15 25 61 193 174 290 147
20/70=====mmmmmm mmm mem 394 - 11 62 32 104 140 45
20/100-=======cmmmmmm me 764 - 42 12 87 211 293 119
20/200=======mmmmm—————— 165 - 24 21 45 18 57 -
Less than 20/200-----=--- 109 - - 23 - 39 26 21
Women
Total-====m=meee=—- 58,343 8,430 11,291 12,325 10,542 8,120 6,192 1,443
20/10 or better=-=--=--=--- 634 164 142 189 50 64 25 -
20/15-====--cmmmmmmee me 20,196 4,626 5,947 5,856 2,669 834 264 -
20/20=====--=mmmmm mmm 20,150 2,462 3,750 4,905 4,325 3,075 1,471 162
20/30=====cmmem mmm 11,419 977 1,051 1,013 2,416 2,793 2,607 562
20/40 meme ————— 2,750 99 194 180 552 675 814 236
20/50-====ccmcmm mmm 1,066 49 32 119 120 228 397 121
20/70======ccmcmmm mmm 575 - 78 16 73 142 85 181
20/100====m=mmmmm mmm mm 938 7 64 37 245 180 297 108
20/200-=====-===ccmcm=—- 258 29 18 - 30 96 47 38
Less than 20/200---=-===-- ; 357 17 15 10 62 33 185 35
16
Table 4. Percent distribution of adults reaching specified acuity levels for 'corrected'" distance
vision, by age and sex: United States, 1960-62
Total, ;
18-24 25-34 35-44 45-54 55-64 65-74 75-79
Sex and acuity level Sy years years years years years years years
Both sexes Percent distribution
Total-=-=-========= 100.0 100,0 100,0 100.0 100.0 100,0 100,0 100.0
20/10 or better==-==----- 1:3 2.4 2.6 1.7 0.9 0.5 0.2 -
20/15=====mmmmmcemmm——— 38.4 57.8 57.7 52.8 31:1 11.2 4.4 1.6
20/20-=====mmccmmm———eam 32.9 27.5 29.3 35.5 39.0 38.4 27.6 13.4
20/30======-mcmmmmmm mem 17.7 10.2 7.4 2.1 20.5 33.5 38.9 38.6
20/40===mmmmmmmm meme mmm 4.5 1.4 15 1.4 442 8.5 12.6 18.0
20/50======-=cmmmmm———— 1.8 0.4 0.3 0.8 1.5 2:6 6.2 9.3
20/70=======cmmmmmmm—mem 0.9 - 0.4 0.3 0.5 1.6 2.0 7.9
20/100=====mcmmcmmmmmeem 1.5 0.0 0.5 0.2 1.6 2.5 5.3 7.9
20/200========-mmmmmmmae 0.4 0.2 0.2 0.1 0.4 0.7 0.9 1=3
Less than 20/200-------- 0.4 0. 0.1 0.1 0.3 0.5 1.9 2.0
Men
Total-==========--= 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
20/10 or better==-==-===-=-= 1.9 2.9 4.0 2.0 1.4 0.2 - -
20/15===mmmmmmmmm mmm ae 42.8 61.3 63.5 58.4 37:3 12.2 4.3 3.2
20/20======mmmmmmmm mmm 31.0 25.4 25.0 30.9 36.6 39.1 32.4 15.7
20/30========ccmemmmean= 15.8 8.6 5.4 5.8 18.0 32.5 35.0 38.0
20/40=====mmmmmmm meme mm 4.2 1.6 1.2 i) 3.2 8.8 11.9 19.8
20/50======cmmmmmmm meee 1.7 0.2 0.2 0.5 1.9 2.3 5.8 10.3
20/70======mcmm mmm meee 0.7 - 0.1 0.5 0.3 1.4 2.8 3.2
20/100======mmemmm—————— Lots - 0.4 0.1 0.9 2.8 5.9 8.3
20/200-======mmmmm—————— 0.3 - 0.2 0.2 0.4 0.2 1.2 -
Less than 20/200=====-==- 0.2 - - 0.2 - 0.5 0.5 1.5
Women
Total=======eeme=- 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
20/10 or better===-==--=-- lal 1.9 1:3 1.5 0.5 0.8 0.4 -
20/15-=====ccmmmmm—————— 34.7 54,9 52.6 47.5 25.3 10.3 4.3 =
20/20=======cmmmmmm————— 34.6 29.2 33.2 39.8 41.1 37.8 23.8 11.2
20/30===-==mmmmmmmm————- 19.5 11.6 9.3 8.2 22.9 34.4 42.1 39.1
20/40======cmmmemm meee 4.7 1:2 1:7 1:5 5.2 8.3 13.1 16.4
20/50======-mmmmmmmm mn 1.8 0.6 0.3 1.0 1.1 2.8 6.4 8.4
20/70=======mmmmm——————— 1.0 - 0.7 0.1 0.7 1.8 1.4 12.5
20/100--===-==mmcm—————— 1:6 0.1 0.6 0.3 2.3 2.2 4.8 74
20/200=======mmmmm—————— 0.4 0.3 0.2 - 0:3 1.2 0.7 2.6
Less than 20/200-====-== 0.6 0.2 0.1 0.1 0.6 0.4 3.0 2.4
Table 5. Number of adults reaching specified acuity levels for uncorrected near vision, by age and
sex: United States, 1960-62
Total,
5 18-24 25-34 35-44 45-54 55-64 65-74 75-79
Sex and acuity level SS years years years years years years years
Both sexes Number of adult in thousands
Total==m==emeee——— 111,087 15,569 21,372 23,698 20,576 15,637 11,164 2,871
14/7 or better=--=--===-- 1,076 261 457 338 20 - - -
14/10.5======mcmemmeenun 25,480 72,513 10,105 7,451 393 18 - -
14/1h=mmmmmm mmm emcee ee 22,897 5,163 7,482 8,485 1,551 116 - 100
14/21=mmmmmmmmmmm mmm eee im 9,817 1,497 1,800 3,429 2,330 482 245 34
14/28==-=mmmmmmmme eee 5,249 411 332 1,033 1,884 885 482 222
14/35==mmmmmmmmm meee em 4,872 162 217 532 1,735 1,104 859 213
14/49=mmmmmmmmmmm meee 6,156 68 241 527 2,582 1,694 801 243
14/70=====mmmmmmmm mee mm 17,556 331 363 1,276 6,077 5,096 3,440 973
14/140======mcmmmmmee eam 13,148 98 418 485 3,146 4,875 3,487 639
Less than 14/140-------- 4,836 65 157 142 808 1,367 1,850 447
Men
Total=====cmeeee=- 52,744 7,139 10,281 11,373 10,034 7,517 4,972 1,428
14/7 or better----=------ 844 189 369 266 20 - - -
14/10,5-=====mmcmmmceean 14,146 4,003 34312 4,354 259 18 - -
14/1b4mmmm mmm meen emma em 10,042 2,033 2,999 3,904 947 59 - 100
14/2]=mmmmmmmmmmm cme 4,052 528 507 1,438 1,223 223 118 15
14/28==mcmmmmmmmm meme e em 2,373 124 198 400 828 408 215 200
14/35===mmcmmmcmm cman 2,321 28 94 253 818 631 403 94
14/49==cmcmmcmmmmce ema 3,254 27 153 194 1,272 963 497 148
14/70=====cmmmmcmcc meee 8,642 137 190 414 3,128 2,646 1,662 465
14/140======ccmcmmmc meen 5,258 39 217 94 1.301 2,022 1,376 209
Less than 14/140=======- 1,812 31 42 56 238 547 701 197
Women
Total======mmce——— 58,343 8,430 11,291 12.325 10, 542 8,120 6,192 1,443
14/7 or better=-==-====- 232 72 88 72 - - - -
14/10,5-======mccmmmeean 11,334 35510 4,593 3,097 134 - - -
14/14mmmmmm mm mmm mmm ee 12,855 3,130 4,483 4,581 604 57 - -
14/2]==mmmmmmmmc ce mmee em 5,765 969 1,293 1,991 1,107 259 127 19
14/28===mmemmmmm meme eam 2,876 287 134 633 1,056 477 267 22
14/35====mmmmmmm mm mme em 2.55) 134 123 279 967 473 456 119
14/49mm mmm meme mm cee 2,902 41 88 333 1,310 731 304 95
14/70=====cmmmmmmmmem em 8,914 194 173 862 2,949 2,450 1,778 508
14/140-======== tmmm————— 7,890 59 201 391 1,845 2,853 2,111 430
Less than 14/140=======-- 3,024 34 115 86 570 820 1,149 250
Table 6.
Percent distribution
of adults reaching
specified acuity levels for uncorrected near
vision, by age and sex: United States, 1960-62
Total,
- 18-24 25-34 35-44 45-54 55-64 65-74 75-79
Sex and acuity level Se years years years years years years years
Both sexes Percent distribution
Total=======c=e=== 100,0 100,0 100,0 100,0 100,0 100,0 100,0 100,0
14/7 or better=--------- 1.0 1.7 2.1 1.4 0.1 - - -
14/10. 5=======memmmcnean 23.0 48.4 46.9 31.4 1.9 0.1 - -
14/1b4mmmmm mmm mmm eee 20.7 33.2 34.8 35.9 ZS 0.7 - 3.5
14/2]=mmmmm mmm meme em 8.9 9.6 8.3 14.5 11.3 3.1 2.2 1.2
14/28=====mmmmmmmm meme 4.7 2.6 1.5 4.4 9.2 5.7 4.3 7.7
14/35= mmm mmm mmm ee 4.4 1.0 1.0 2.2 8.7 7-1 i 0 7.4
14/49===mmmmmmmmmmeee em 54.5 0.4 1.1 2.2 12.5 10.8 2.2 8.5
14/70=====mmmmmmmmmmme a 15.7 2.1 1.7 5.4 29.6 32.6 30.8 33.8
14/140-====mmmmmm meme 11.8 0.6 1.9 2.0 15.3 31.2 31.2 22.3
Less than 14/140===----= 4.3 0.4 0.7 0.6 3.9 8.7 16.6 15.6
Men
Total=======eeea=- 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
14/7 or better---------- 1.6 2.6 3.6 2.3 0.2 - - -
14/10.5=======mmemmmmean 26.9 56.2 53.7 38.4 2.6 0.2 - -
J A EE EE EE 19.1 28.5 29.2 34.4 9.4 0.8 - 7.0
14/21-=mmmmmmm meme meme Te? 7.4 4,9 12.6 12.2 3.0 2.4 1X
14/28==cmmmmmmmmmcmme ee 4.5 1.7 1.9 3.3 8.3 5.4 4.3 14.0
14/35===mmmmm mmm mmeam 4.4 0.4 0.9 2.2 8.2 8.4 8.1 6.6
14/49=mmmmmmm meme eee ee 6.2 0.4 1.5 Lu 12.7 12.8 10.0 10.3
14/70===-==mmm mmm mem 16.3 1.9 1.8 3.6 31.0 35.2 33.4 32.6
14/140=====mmmmm mmm mm em 9.9 0.3 2.1 0.8 13.0 26.9 27.7 14.6
Less than 14/140-------- 34 0.4 0.4 0.5 2.4 7:3 14.1 13.8
Women
Total=-======c==-= 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
“14/7 or better==-==--=--- 0.4 0.8 0.8 0.6 - - - -
14/10. 5=======mmmmmeeeam 19.6 41.7 40.7 25.1 1.3 - - -
14/14mmmmmm mmm meee 22.1 37.1 39.7 37.1 5.7 0.7 - -
14/2]l==mmcmmmm emma 92.9 11.5 11.4 16.2 10.5 3.2 2.0 1.3
14/28-==mmmmmmmm meme 4.9 3.4 1:2 5.1 10.0 5.9 4.3 1.5
14/35===-mmcmmmmmmm ee 4.4 1.6 de 2.3 92.2 5.8 7.4 8.3
14/49=mmmcmmmmm emma 5.0 0.5 0.8 2.7 12.4 9.1 4.9 6.6
14/70==mmmmmmmm mmm meen 15.2 2.3 1.5 7.0 28.0 30.2 28.7 35.2
14/140====mcmmmcmmcceeam 13.4 0.7 1.8 3.2 17.5 35.0 34.1 29.8
Less than 14/140=-=====-- 0 0.4 1.0 0.7 5.4 10.1 18.6 17.3
Table 7.
Number of adults reaching
specified acuity levels
for "corrected near vision, by age
and sex: United States, 1960-62
Total,
18-24 25-34 35-44 45-54 55-64 65-74 75-79
Sex and acuity level i years years years years years years years
Both sexes Number of adults in thousands
Total--=======c=== 111,087 15,569 21,572 23,698 20,576 15,637 11,164 2,871
14/7 or better---=-=-=-=-=--=- 1,262 208 519 394 110 31 - -
14/10 ,5-======mcmcenenu= 31,624 8,335 11,455 8,835 1,890 856 253 -
14/1b4==mmmmmmm meme mmm em 38,964 5,616 7,841 10,218 7,013 5,251 2,652 373
14/21-==mmmmmmmmmem mee 22,143 1,198 1,385 2,896 6,118 5,191 4,494 861
14/28--=--ccmmmmmmmee em 6,882 119 130 669 2,131 1,744 1,495 594
14/35-=====cmmmmmmmmmmmm 3,162 54 68 250 1,028 840 651 27
J 2,086 16 37 133 787 438 394 281
14/70=====mmmmmmmm mmm m em 3,389 10 96 233 1,188 778 751 333
14/140-=====mmmemmmmmmm 1,124 13 41 25 225 413 359 48
Less than 14/140-=====-- 451 - - 45 86 95 115 110
Men
Total--=======c=== 52,744 7,139 10,281 11,373 10,034 7,517 4,972 1,428
14/7 or better-=--=---=--=-- 979 161 431 294 77 16 - -
14/10.5-======cccmmeemu- 17,281 4,280 6,176 4,981 1,128 555 161 -
14/1b=mmmm mmm mmm eee 16,989 2,089 3,043 4,437 3,511 2.472 1,194 243
A EE 9,116 528 403 1,258 2,611 2,068 1,870 378
14/28====mmmmm mmm mmm em 2,931 53 119 165 821 785 667 321
14/35-==mmmmmm mmm 1,718 28 36 66 499 570 360 159
14/49==cmmmmmmmmcmmee 1,175 - 23 51 536 268 167 130
14/70=====mcmmm mcm meen 1,798 - 42 77 713 479 355 132
14/140-=====mmmmmmmmmm mm 593 - 8 9 126 270 159 21
Less than 14/140======-- 164 - - 35 12 34 39 44
Women
Total======meece==- 58,343 8,430 11,291 12,325 10,542 8,120 6,192 1,443
14/7 or better----=----- 283 47 88 100 33 15 - -
14/10 .5======memecmmm——— 14,343 4,055 5,279 3,854 762 301 92 -
14/1b4=mmmm mm mmm mmm mem 21,975 3,527 4,798 5,781 3,502 2,779 1,458 130
14/21====mmm mmm mmmmm em 13,027 670 982 1,638 3,507 3,123 2,624 483
14/28====mmmmmmmm meme 3,951 66 11 504 1,310 959 828 273
14/35====cmccmmmmcemmee 1,444 26 32 184 529 270 291 112
14/49-=mmmmmmmmmme eee 911 16 14 82 251 170 227 151
14/70====mmmmmmm mmm 1,591 10 54 156 475 299 396 201
14/140-====mcmmmmmm mma 531 13 33 16 929 143 200 27
Less than 14/140-======- 287 - - 10 74 61 76 66
20
Table 8.
Percent distribution
of adults reaching
vision, by age and sex: Un
specified acuity levels
ited States, 1960-62
for "corrected" near
Total,
: - 18-24 25-34 35-44 45-54 55-64 65-74 75-79
Sex and acuity level RS years years years years years years years
Both sexes Percent distribution
Total======eeeaaa- 100,0 | 100,0 100,0 100,0 100,0 100,0 100.0 100,0
14/7 or better==-=====-- 1.1 1.3 2.4 1.7 0.5 0.2 - -
14/10,5-====ccmmmcccaaa 28.6 53.4 53.1 37.3 9.2 5.3 2:3 -
14/14mmmm mmm emma 35.2 36.2 36.4 43.0 34.1 33.5 23.8 13.0
14/21==mmm mmm 19.8 7:7 6.4 12.2 29.7 33.2 40.3 30.0
14/28====mccmmcmmc eee 6.2 0.8 0.6 2.8 10.4 11.2 13.4% 20.7
14/35=====mcmmmmm emma 2.8 0.3 0.3 l.1 5.0 5.4 5.8 9.4
14 [49mm mmm cmc 1.9 0.1 0.2 0.6 3.8 2.8 3.5 9.8
14/70=====cmmmmmmmmccea 3.0 0.1 0.4 1.0 5.8 5.0 6.7 11.6
14/140-===cmmcmmmcccccan 1.0 0.1 0.2 0.1 1.1 2.6 3.2 1.7
Less than 14/140==-====- 0.4 - - 0.2 0.4 0.6 1.0 3.8
Men
Total=====e=eecaa- 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
14/7 or better==-======- 1.9 2.3 4,2 2.6 0.8 0.2 - -
14/10 ,5-====ccmmmcceeeem 33.0 59.9 60.0 43.9 11.2 1.4 3.2 -
14/14=mmmmmmmme meee 32.2 29.3 29.6 39.0 35.0 32.9 24.0 17.0
14/21=====mmmmmmeeee em 17.2 1.4 3.9 11.0 26.0 272.5 37.6 26.4
14/28-===--ccmmmmeemeee 5.5 0.7 1.2 1.4 8.2 10.4 13.4 22.4
14/35=====-ccmmmmcemeeae 3.2 0.4 0.4 0.6 5.0 7.6 7.2 11.2
EE 2.2 - 0.2 0.4 5:3 3.6 3.4 9.1
14/70======cmcmmmmmceeem 3.4 - 0.4 0.7 7.1 6.4 7.2 9.3
14/140======mmmmmmmeeee 1.1 - 0.1 0.1 1.3 3.6 3.2 1.5
Less than 14/140--=-==== 0.3 - - 0.3 0.1 0.4 0.8 3)
Worien
Total=======meeca- 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
14/7 or better---------= 0.5 0.6 0.8 0.8 0.3 0.2 - -
14/10 ,5-=====c=mccmcaan 24.7 48.0 46.7 31.3 7.2 3.7 1.5 -
14/ 14mm m mmm emcee 37.8 41.9 42.5 46.8 33.3 34.2 23.5 9.0
14/21-=--mmmmmmmmmm eee 22.2 7.9 8.7 13.3 33.3 38.4 42.4 33.5
14/28---cmmcmcmmcccee ee 6.7 0.8 0.1 4.1 12.4 11.8 13.4 18.9
14/35=====-=mmmmmceeee 2.5 0.3 0.3 1.5 5.0 3.3 4.7 7.8
14/49-==mcmm mmm eeee ee 1.5 0.2 0.1 0.7 2.4 2.1 3.7 10.4
14/70==-=====mcmmmmeeeem 2.7 0.1 0.5 1.3 4.5 3.7 6.4 13.9
14/140-====cmcmmmceeeee 0.9 0.2 0.3 0.1 0.9 1.8 3.2 1.9
Less than 14/140-=--=--- 0.5 - - 0.1 0.7 0.8 1.2 4.6
21
APPENDIX 1
TARGET SPECIFICATIONS AND ITEMS ON
MEDICAL HISTORY RELATED TO VISION USED IN THIS REPORT
The three lines on the Sight-Screener target used for testing distance and near vision:
t
Specifications of letter sizes and numbers of letters on Sight-Screencr targets for testing distance
and near vision.
Snellen ratios for
: . . : p letter sizes Number
Visual angle in minutes Decimal equivalent sed af of
subtended at standard of Snellen ratios L 1
test distance (reciprocal of
(20 ft. or 14 in.) visual angle) Oistance You? sach level
(20 fc.) (14 in.)
10.00 .1000 20/200 14/140 1
5.00 .2000 20/100 14/70 1
3.50 .2859 20/70 14/49 2
2.50 .4000 20/50 14/35 4
2.00 .5000 20/40 14/28 4
1.50 .6667 20/30 14/21 4
1.00 1.0000 20/20 14/14 4
v3 1.3333 20/15 14/10.5 4
+50 2.0000 20/10 14/7 4
LThis is the size of the visual angle of resolution in minutes of arc subtended by the width of the lines in the test letters used at each
threshold level.
22
Selected Medical History Questions
(Excerpts from HES-204, Medical History—Self Administered)
46. a. Do you wear glasses?
_
m
wn
—<
m
w
Zr =
ofl o
If YES b. Do you wear them all the time?
If you don't wear them all the time, check below
when you do wear them
For seeing at a distance
Q oO
. For reading
®
. For watching TV
f. At other times When?
47. a. Do you have serious trouble with seeing, even when wearing
glasses?
—<||=
m m
unllw
=ji=i}=
Silo) IS
If YES b. Have you had this trouble in the past 12 months?
c. Have you ever seen a doctor about it?
-
m
w
23
APPENDIX I
SOME TECHNICAL NOTES ON THE VISION TEST
The visual acuity test used in this survey is
in effect a subjective examination of the form
sense of the examinee or the ability which the eye
possesses to perceive the shape or form of ob-
jects.
Experimental evidence has shown that, in
addition to the distance from the target, the com-
plexity of the form of the target letters, the ef-
fective illumination used, the target contrast
between letters and background, and the end-point
or scoring criteria will all affect the level ob-
tained in such testing.’ 11
The range of 20 feet is theusual one selected
for distance testing since rays of light from this
distance are practically parallel. When in a state
of rest, the eye is adapted for parallel rays
coming from a distant object. To focus objects
closer than 20 feet, as needed in near vision, the
light rays from the object have to be bent so that
they come together on the retina. The muscles of
the eye accommodate for this by increasing the
convexity of the lens and thus its refractive
power, 6
Binocular vision requires a further muscular
adjustment not involved in monocular seeing. This
is termed convergence or the directing of the
visual lines from both eyes to a near point.”
Both the ability of the normal eye to converge
and to accommodate will tend to decrease with age,
but not necessarily at the same rate. Hence, some
differences may be expected in the decrease of
monocular and binocular acuity with age.
000
24
APPENDIX lI
SURVEY DESIGN, RESPONSE,
The Survey Design
The Health Examination Survey is designed
as a highly stratified multistage sampling of the
civilian, noninstitutional population of the con-
terminous United States, aged 18-79 years. The
first stage of the plan is a sample of the 42 pri-
mary sampling units (PSU's) from among some
1,900 such geographic units into which the United
States was divided. A PSU is a standard metro-
-politan statistical area or one to three contiguous
counties. Later stages result in the random se-
lection of clusters of typically about four persons
from a small neighborhood within the PSU. The
total sample included approximately 7,700 persons
in the 42 areas in 29 different States. The detailed
structure of the design and the conduct of the Sur-
vey have been described in other reports!”
Reliability in Probability Surveys
The Survey draws strength from the fact that
the measurement processes which were employed
were highly standardized and closely controlled.
This does not mean, of course, that the corre-
spondence between the real world and survey re-
sults is exact. Data from the survey are imper-
fect for three important reasons: (1) results are
subject to sampling error; (2) the actual conduct
of a survey never agrees perfectly with the de-
sign; and (3) the measurement process itself is
AND SAMPLING VARIABILITY
inexact, even when standardized and controlled.
The National Center for Health Statistics, both in
special studies and in regular operations, tries
to evaluate its surveys and to present the findings
to consumers,
One part of this effort was reported which
dealt largely with an analysis of the faithfulness
with which the design was carriedout. This study
noted that of the 7,700 sample persons, the ap-
proximately 6,670 who were examined (a response
rate of over 86 percent) give evidence that they
are a highly representative sample of the civilian,
noninstitutional population of the United States.
Imputation for the nonrespondents was accom-
plished by attributing to nonexamined persons the
characteristics of comparable examined persons.
The specific procedure used has been described
in another report.” It amounted to inflating the
sampling weight for eachexamined person tocom-
pensate for sample persons at that stand and of
the same age-sex group who were nonexamined.
In addition to persons not examined at all,
there were some persons whose examination was
incomplete in one particular or another. Age, sex,
and race were known for every examined person,
but for a number of persons onc or more of the
vision tests with or without glasses was not avail-
able. Most of the omissions were accidental. The
extent of missing information for binocular tests
is indicated in table I.
Table I. The extent of missing binocular vision data: Health Examination Survey,
1960-62
Number of
Type of test examinees
Total exXamineeS-=-======-=-- cco mmmm meee em emm--——-——————-——— 6,672
Distance and near tests without glasses completed---==-=-=-=-==-=------co-——o-- 6,531
Only distance tests without glasses completed----=-=----------=----=---o-oo--- 3
Distance and near test done only with glasses---==-====---=mmeccccaccannnnn- 8
Only near rests with ZlagsSes COMPLETEG mim mic immimimm mm mwes u mt mmm mow mmm ow mm mmm ms ow 3
Not tested at distance or near with or without glasgeg-—===-=—=wweeecnun 127
25
To estimate scores for the 14 individuals for
whom at least one vision test was completed, a
"regression-type' decision was made subjectively
on the basis of the existing scores and test results
for other persons of the same age, sex, and race.
For the 127 persons not given any of the
vision tests, a probability selection was made of
a respondent from the same age-sex-race group
and his scores assigned to the nonrespondent.
Sampling and Measurement Error
In the present report and its appendices,
several references have been made to efforts to
evaluate both bias and variability of the measure-
ment techniques,
The probability design of the survey makes
possible the calculation of sampling errors.
Traditionally, the role of the sampling error has
been the determination of how imprecise the sur-
vey results may be because they come from a
sample rather than from measurement of all
elements in the universe.
The task of presenting sampling errors for
a study of the type of the Health Examination Sur-
vey is difficult for at least three reasons:
(1) measurement error and ''pure' sampling error
are confounded in the data; it is not easy to find
a procedure which will either completely include
both or treat one or the other separately,
(2) the survey design and estimation procedure
are complex and accordingly require computa-
tionally involved techniques for calculation of
variances, (3) from the survey will come thousands
of statistics, many for subclasses of the population
for which there are small numbers of sample
cases. Estimates of sampling error are obtained
from the sample data and are themselves subject to
sampling error, which may be large when the
number of cases in a cell is small, or even oc-
casionally when the number of cases is sub-
stantial. Estimates of approximate sampling vari-
ability for selected statistics used in this report
are presented in table II. These estimates have
been prepared by a replication technique, which
yields overall variability through observation of
variability among random subsamples of the total
sample. The method reflects both ''pure' sam-
pling variance and a part of measurement
variance.
26
In accordance with usual practice the interval
estimate for any statistic may be considered to
be the range within one standard error of the
tabulated statistic, with 68 percent confidence;
or the range with two standard errors of the
tabulated statistic, with 95 percent confidence.
An overestimate of the standard error of a
difference d = x - y of two statistics x and y is
given by the formula S4 = [x° v2 + vv y
X y
where Vo and Ve arethe relative sampling errors,
’
respectively of x and y. For example, tables 1 and
2 show x = 17,792,000 or 33.9 percent for men
and y = 14,494 000 or 24.9 percent for women
testing at distance without glasses at the 20/15
level. Table II shows relvariances relative sam-
pling errors of Vy = .04 and Ve = .04 for the re-
spective percentages. The formula yields the
estimate of the standard error of the difference
(d = 9.0 percent) as 84° 1.68 percent. Thus the
observed difference is more than five times its
sampling error and hence significant.
A further example from table 2 shows
x = 109,000 or 0.2 percent for men and y = 357,000
or 0.6 percent for women testing less than 20/200
with whatever correction they were using. Table
II shows relative sampling errors of ¥, = 0.18
and Y, = 0.09 for the respective percentages. The
formula yields the estimate of the standard error
of the difference (d = 0.4 percent as 84° 0.07
percent. Here the observed difference is more
than five times its sampling error and hence
significant.
Small Categories
In some tables magnitudes are shown for cells
for which sample size is so small that the sam-
pling error may be several times as great as the
statistic itself, Obviously in such instances the
statistic has no meaning in itself except to indi-
cate that the true quantity is small. Such numbers,
if shown, have been included in the belief that they
help to convey an impression of the overall story
of the table.
Table II. Relative sampling error for proportion of persons with specified visual
acuity,! by sex, race, and age: United States, 1960-62
Visual acuity
Sex, race, and
age 20/10 Less
or 20/15 | 20/20 | 20/30 | 20/40 | 20/50 | 20/70 | 20/100 | 20/200 than
better 20/200
Both sexes 0.16 | 0.02] 0,02] 0.04] 0.05] 0.06 | 0.10 0.04 0.06 0.08
Male
Total-====~- 0.18 0.04 0.05 0.05 0.06 0.09 0.10 0.06 0.10 0.18
White-------- 0.16 0.04 | 0.06 | 0.06 0.06 | 0.09 | 0.10 0.06 0.09 0.15
Negro-=-=-=-=--= in 0.07 0.10 0.18 0,25 0.24 0.20 0.25 0.60 -—
Age
18-24 years-- 0.30 0.05 0.10 0.08 0.18 0.60 0.50 0.30 0.40 -—
35-44 years-- 0.50 0.05 0.06 0.12 0.30 0.25 0,22 0.25 0.30 0.60
65-74 years-- ion. 0.70 0.153 0,12 0.16 0.22 0.28 0.15 0.22 0.15
Female
Total--==~-- 0.18 | 0.04 | 0.02] 0.06 0.05] 0.06 | 0.16 0.05 0.05 0.02
White-===»==~- 0.18 0.04 0.03 0.06 0.06 0.06 0.18 0.06 0.06 0.10
Negro=======- ——— 0.12 0.05 0.06 0.15 0.20 0.22 0.25 0.30 0.20
Age
25-34 years-- 0.35 0.05 0.05 0.18 0.25 0.40 0.50 0.25 0.30 0.30
45-54 years-- mt 0.07 0.08 0.09 0.13 0.14 0.15 0.10 0.10 0.24
75-79 years-- im —— nm 0.50 0.25 0.30 0.50 0.26 0.30 0.60
lEstimated relative sampling errors are siown in the table as computed. for a considerable number of specific cells. It suouid pe under-
stood in anv instance in whicn the estimated error for a particular cell differs markedly from tuose for other sinilarcells that the discrepancy
may be a reflection of a real phenomenon, but might be the consequence of the fact that the estimated sampling error is itself subject to saw-
pling variation.
000
27
U. S. GOVERNMENT PRINTING OFFICE : 1964 O - 730-774
NATIONAL A
Ro
For HEALTH
STATISTICS WLLL uy
Blood Pressure
of Adults
CRE TR
United States. 1960-1962
U.S. DEPARTMENT OF
HEALTH, EDUCATION, AND WELFARE
Public Health Service
See inside of back cover
for catalog card.
Public Health Service Publication No. 1000-Series 11-No. 4
For sale by the Superintendent of Documents, U.S. Government Printing Office
Washington, D.C., 20402 - Price 35 cents
NATIONAL CENTER| Series 11
For HEALTH STATISTICS | Number 4
VITALand HEALTH STATISTICS
DATA FROM THE NATIONAL HEALTH SURVEY
Blood Pressure
of Adults
by Age and Sex
United States. 1960-1962
Blood pressure measurement, and distributions and
mean levels by age and sex.
Washington, D.C. June 1964
U.S. DEPARTMENT OF
HEALTH, EDUCATION, AND WELFARE Public Health Service
Anthony J. Celebrezze Luther L. Terry
Secretary Surgeon General
NATIONAL CENTER FOR HEALTH STATISTICS
Forrest E. Linder, Ph.D., Director
Theodore D. Woolsey, Deputy Director
Oswald K. Sagen, Ph.D., Assistant Director
(for Professional Relations)
Walt R. Simmons, M.A., Statistical Advisor
Alice M. Waterhouse, M.D., Medical Advisor
James E. Kelly, D.D.S., Dental Advisor
Louis R. Stolcis, M.A., Executive Officer
DIVISION OF HEALTH EXAMINATION STATISTICS
Arthur J. McDowell, Chief
COOPERATION OF THE BUREAU OF THE CENSUS
In accordance with specifications established by the National
Health Survey, the Bureau of the Census, under a contractual
agreement, participated in the design and selection of the sample,
and carried out the first stage of the field interviewing and certain
parts of the statistical processing.
Public Health Service Publication No. 1000-Series 11-No, 4
CONTENTS
Page
INtroduCtion =====-c cement ee 1
Blood Pressure Measurement--=-===-=cemeeemmmeeeeeeeanax 1
Blood Pressure Variation —=------e-cccmmmmmmcmmcm cee neeee 3
Special Sources of Variation----==--eceeeemmmmememem eee 3
Blood Pressure by Age and Sex----=-===-ememmmmmmcanananan 6
Comparisons and Analysis----=-=--ceccmmmmmmmm mmc cco 9
Discussion ======---mmmm meee 12
SUMMATY======= mmm mmm meme meee mmm meme mm 15
Detailed Tableg======m-mmmm meme ee 17
Appendix 1. Blood Pressure Measurement====m==mmmmm————— 35
Appendix II, Survey Design, Missing Data, and Variance----- 36
The Survey Design-=--=--=-==ccmcmomcmmm cece mom 36
Reliability in Probability Surveys---=--==-ceccccecmanaaan 36
Sampling and Measurement Error--------==--ceceeeacau- 36
Small Numbers===-c-cccmmm mmm meme eee meee em 37
Appendix III, Observer Variability in Blood Pressure
Measurement-------- meee mmm mm mmm 38
SYMBOLS
Data not available--==--mcceommmmmm eee
Category not applicable---=-c-ccocmcaa_o
Quantity Zero-==-==m- ccm meee
Quantity more than O but less than 0,05-----
Figure does not meet standards of
reliability or precision-------c-ceceeao--
BLOOD PRESSURE OF ADULTS
BY AGE AND SEX
Tavia Gordon, Division of Health Examination Statistics
INTRODUCTION
The first cycle of the Health Examination
Survey was undertaken to obtain information on
the prevalence of certain chronic diseases, on
dental health, and on the distribution of a number
of anthropometric and sensory characteristics in
the civilian, noninstitutional population of the
United States. A sample of 7,710 persons aged
18-79 years was drawn, and of these 6,672 were
examined. Each personreceived a standard exam-
ination, lasting about 2 hours, performed by medi-
cal and other staff members of the Survey in
specially designed mobile clinics. The study de-
sign and execution have been previously de-
scribed, ! and a description of the sample and
response has been published.’
This report presents data on blood pressure
by age and sex. It describes the pertinent parts of
the examination, specifies the techniques used,
and compares the information obtained in this
Survey with that obtained in other surveys. The
relationship of blood pressure with other findings
of the examination or with demographic variables
other than age and sex is not dealt with in this
report.
BLOOD PRESSURE MEASUREMENT
The measurement of blood pressure was part
of a cardiovascular examination, which included,
in addition to a medical history, an electrocardio-
gram, a chest X-ray, auscultation of the heart,
examination of the peripheral arteries, and fun-
duscopy. Some details of this examination have
been described in a previous report.! Upon
entering the Mobile Examination Center the ex-
aminee was greeted by a receptionist-interviewer,
who obtained a limited number of personal and
medical particulars from him. The examinee then
completed a self-administered medical history.
Since this routine was invariant, at least 45
minutes passed, in most cases, before the ex-
aminee saw a physician. In some instances hehad
already completed part, or all, of the other ex-
amination procedures and had been in the Center
more than an hour and a half before receiving his
physical examination.
The blood pressure of each examinee was
measured three times during the course of the .
physical examination. The first measurement
was taken just after the physician met the
examinee. The second was taken midway in the
examination, after auscultation of the heart inthe
sitting position and before the arthritis examina-
tion. The examinee had just had an electro-
cardiogram taken by the nurse and had been
allowed a few moments after sitting up for the
effects of postural hypotension to disappear. The
third measurement was taken at the end of the
physical examination.
A venipuncture was usually made during the
physical examination, although the specific point
at which it was taken varied from one examinee
to another.
Blood pressure measurements were taken on
the left arm with the examinee sitting on the exam-.
ining table. The nurse placed the middle of the
cuff over the bulge in the upper left arm. The
cuff was left on the arm between the first and
second measurements, was removed after the
second, and returned for the third. The physician
held the arm at the level of the atrium, with the
nurse raising the Baumanometer to the physician's
eye level. Using the bell of his stethoscope, the
physician noted the pressure when the sound was
«first heard, when it first became muffled, and when
it disappeared, recording all three measurements.
In this report, the point at which the Korotkoff
sounds disappeared is given as the diastolic
pressure. If the sounds did not disappear, the
point of muffling, if distinctly heard, is given.
Since the Baumanometer is scaled in intervals of
22 mm., measurements were so recorded. The
background of these procedures is discussed
briefly in Appendix I.
There is a tendency to choose certain end
digits in measurement, with particular preference
for 0 or 5. Table A gives the distribution of end
digits used in reporting systolic and diastolic
pressures on the first blood pressure measure-
ment. The preference for numbers ending in O is
quite marked, and a comparable preference for
the end digit 5 is strong enough in some cases
to overcome the instruction to use only even
Table A. Distribution of end digits on
blood pressure measurement: Health Ex-
amination Survey, 1960-62
First Average of 3
measurement measurements
End digit
Sys- | Dia- Sys- Dia-
tolic | stolic | tolic | stolic
Number of examinees
O-mmmmmmmm 2,169 2,299 560 652
lommm————— - 1 894 836
RR taal 1,073 895 444 480
Fim mimmimim ion - 2 885 829
fommmmmmme 1,200 933 440 429
Smmmmm mmm 66 69 856 814
b= 1,005 1109 409 393
Jmmmmm——— 2 - 910 828
FE 1,153 1,289 453 484
Qemmmmmmmm 2 1 819 927
Missing--- 2 174 E
Ish phase.
numbers in recording. If all three blood pressures
.are averaged, a set of artificial end digits re-
sults which are more uniformly distributed, al-
though the averaging of three even numbers re-
sults in more odd than even quotients. At least
it is possible to group blood pressures ending in
digits O through 4 and those ending in digits 5
through 9 without great irregularity in the re-
sulting distributions.
The preference for certain end digits would
merely be an item of human frailty were it not
also associated with disease judgments. The lower
limit for definite hypertension often used is 160,
and it will be noted that on the first blood pressure
measurement there was definite preference for a
, reading of 160 over a reading of 158 (table B).
A similar situation can be observed in reading
diastolic pressures around 90 mm.hg., which is
frequently used as a lower bound for borderline
hypertension. On the second and third measure-
ments these strong preferences seemed to di-
minish. Averaging all three values, of course,
tends to transform these reading preferences and
to obscure them.
Table B. Number of blood pressure measure-
ments at specified levels, by order of
measurement: Health Examination Survey,
1960-62
Measurement
Blood pressure
(mm. hg.) ]
First | Second | Third
Systolic
144mm mmmmmm emma 126 120 100
146---mmmmmmmm mmm 84 94 97
LE aa 98 79 77
158--=-=mmmmmm 56 53 59
160 -=-==mmm mmm mmm 124 116 88
162--=-====mmmmmm 50 32 41
Diastolic
88--mmmmm mmm mmm 277 284 253
90 --mmmm mmm mmm 390 348 341
tate 124 154 128
Ofc mmm meme mam 106 116 117
9p mmmmm mmm mmm mem 113 133 109
98-mmmm mmm mmm 107 105 90
BLOOD PRESSURE VARIATION
Blood pressure may vary considerdbly over
a short period of time even under relatively.
standard conditions. For half the persons ex-
amined during this cycle of the Health Examina-
tion Survey the difference between thehighestand.
lowest systolic readings was 10 mm. hg, or more.
In half the cases the difference between the highest
and lowest diastolic readings was at least 6 mm..
hg. Similar variation has been noted in other
studies. In the measurements made by the Health
Examination Survey, variation was about the same
for men and women, but both for men and for women
it increased with age. These observations refer, .
of course, only to variation observed during a
single physical examination. If variation is meas-
ured over a longer period of time the median
range becomes greater. In one study where sub-
jects had determinations of resting blood pressure
made six times every weekday for 3 weeks, the
median range over the 3-week period was 30 mm.,
hg. for systolic and 22 mm. hg. for diastolic
pressures.’ Even in hospital studies where an
effort is made to obtain basal blood pressures
under carefully controlled conditions, blood pres-
sures for an individual vary from one time to
another, although less than with Casual pres-
sures
Because blood pressure fluctuates it seemed
reasonable to average the three blood pressure-
measurements obtained for each individual and to
use this average as the best measure of his blood
pressure, It is these average measurements that
are tabulated in this paper (excluding those in
tables A and B). Such average figures do not
necessarily eliminate the recording problems,
even though they probably reduce the effect of
reading preferences. A distribution of average
values is shown in figure 1. It covers only the
range from 130 to 199 mm. hg. systolic and 80 to
109 mm. hg. diastolic. The averaging procedure
leads to an excess of odd-end digits, which pro-
duces a sawtooth effect in the figure, but in addition
some irregularity is probably introduced by a
tendency to shift readings as boundary values are
approached, that is, values which traditionally are
regarded as those separating hypertensive from
normotensive levels,
SPECIAL SOURCES OF VARIATION
Two characteristics of the Health Examina-
tion Survey merit special attention because of their
possible effect on the blood pressure data. The
first is that during the physical examination a
venipuncture was made. The second is that persons .
were examined at different times of the day.
A venipuncture is disturbing to many people,
and although a blood pressure measurement was
never taken immediately after a venipuncture, it
is possible that some delayed reaction to the veni-
puncture might alter the blood pressure level.
If this occurred with sufficient frequency and if
the changes were large enough and tended to be
in the same direction, the mean blood pressure
level would be discernibly altered by the veni-
puncture. Even if this did not happen it is still
possible that the venipuncture increased the
variability of measurement to some extent,
It must be emphasized that no direct measure-
ment of the effect of venipuncture on blood pres-
sure was undertaken during the Health Examina-
tion Survey. However, the three successive blood
pressure measurements on each individual were
recorded, as was the time of the venipuncture,
and it should be possible to discern the effect of
venipuncture from these data—if the effect is
marked. The problem may be approached in the
following way. On the average, systolic pressure
tended to decrease slightly from the first to the
third measurements, whereas the diastolic pres-
sure remained about the same on successive
measurements. For some persons a venipuncture
was made before the first blood pressure meas-
urement, for others between the first and second,
and so on. The question is whether the relation-
ship among successive blood pressure measure-
ments differed in some consistent fashion ac-
cording to the time of venipuncture.
The answer to this question is complicated
by the fact that younger persons, who have lower
and less variable blood pressures, proceeded
through the examination more rapidly than older
persons. Since the venipuncture was timed to
occur a little more than 1 hour after the be-
ginning of the examination, regardless of age,
younger persons were more apt than older to be }
further advanced in the examination at the time
Number of persons
SYSTOLIC
130-199 mm. hg.
150
100
0
Cec bcc bev cc bce bev a ber
130 135 145 165 165 175 185 195
Blood pressure (mm. hg.)
DIASTOLIC
80-102 mm. hg.
300
200 — —
100 + -
Slr bparalen ir snaole eens stb snr sda]
65 80 90 100 110
Blood pressure (mm. hg.)
Figure 1. Distribution of specified blood pressures: Health Examination Survey.
of venipuncture. Therefore, it is necessary to
compute differences between successive blood
pressures that are age specific. In table C these
are summarized as age-adjusted differences.
If blood pressure were affected in a con-
sistent manner by venipuncture, the three num-
bers in any column of table C would vary in the
same pattern as the three numbers in any other
column, The data do not suggest this; in fact,
what variation there is in each column is trivial.
It can therefore be assumed that the venipuncture
had no discernible effect on the blood pressure’
levels reported by this Survey.
A possible diurnal variation in blood pressure
is another concern because sample persons came
for examination at their convenience rather than
at random. Older people were more likely to come
early in the day than younger.’ This difference,
while definite enough, would be important only if
mean blood pressure had a marked diurnal varia-
tion and the data from the Health Examination :
Survey do not indicate this.
Table D presents age-adjusted blood pressure
levels according to the time of day at which per-
sons arrived for the examination. Blood pres-
sures were taken approximately an hour later.
These calculations are for the age range 18-74
years, as data for the age group75-79 years were
too scanty to be included. The tabled values do
not constitute estimates for the population of the
United States.
Table D. Age-adjusted blood pressure, by
time of day: Health Examination Survey,
1960-62
Time of day! Systolic Diastolic
Blood pressure inmm. hg.
8 a.m--mmmumm- 128.7 78.7
9 a.m=--------- 129.6 79.5
10 a.m-======-- 127.6 75.0
11 a.m=-=-=-=====-~ 129.8 78.2
12 p.m=-=-=-=-=-- 130.6 78.7
1 pem=-ennen-- 130.1 78.5
2 pom---=====- 129.8 79.5
3 pum====m=nm= 131.6 79.9
4 p.m----=-=-- 132.0 80.4
5 PumMm======== 133.2 81.1
6 p.m------==- 132.1 78.8
7 Pum mmmm——— 131.1 79.2
1
“Time of day”’ is the time the examinee began his exunination.
Blood pressures were usually measured about an hour later.
NOTL: These values are obtained by weighting mean values by
age and sex for each time of day by the age-sex distribution of the
total U.S. population. They do not constitute estimates for the
United States.
The data are consistent with a slight tendency
for blood pressures to rise in the afternoon and
it is possible that atest of significance (which was
not undertaken) would demonstrate this in statis-
tical terms. However that may be, it is unlikely
that this variation is great or that it constitutes
Table C.
Changes in blood pressure associated with time of venipuncture: Health Exami-
nation Survey, 1960-62
Difference in mm. hg. between
Ti £ . First and second Second and third
ime of venipuncture measurements measurements
Systolic | Diastolic | Systolic | Diastolic
Before either measurement---------=---co--- 2:62 0.40 0.67 -0.28
Between the two measurements----------—-==-- 3.01 -0.05 1.10 -0.39
After both measurements------===--cccaeeaa-o 2.96 0.29 0.75 -0.19
732-721 O - 64 - 2
<
pling variability, but when considered overall, they
present a consistent picture of the relation be-
tween systolic and diastolic pressures. Estimating
Table E. Mean blood pressure in adults, by age and sex: United States, 1960-62
Systolic Diastolic
Age
Both Men Women Both Men Women
sexes sexes
Mean blood pressure in mm. hg.
All ages-18-79 years---- 130,9 132, 129.9 78.7 79.4 78.1
18-24 years-----===-=--cee-oo- 116.4 121.7 111.8 70.4 71.6 69.4
25-34 years-==-----eccomeeonooo 119.9 124.7 115.6 74.6 76.4 72.2
35-44 years---=-m-mmmmmmmmmne- 125.6 128.6 122.8 79.3 80.7 78.0
45-54 years-=-mmmmmmmmeme—m——-- 133.8 133.8 133.8 82.6 83.2 82.0
55-64 years----=------meenmnea- 143.6 140.3 146.6 84.0 83.1 84.9
65-74 years---mmmmmmmmm——e———- 154.8 148.0 160.2 82.5 81.0 83.7
75-79 years----==---c-meemnann 155.5 154.3 156.6 79.4 79.4 79:3
a complicating feature in the analysis of the
data; it is obviously only a minor source, if any, =
of variation,
180
— Men
smmms Women
BLOOD PRESSURE BY AGE AND SEX
0,
oo “a
Mean blood pressures by age and sex are
given in table E and figure 2. These show a tend- 140
ency for systolic blood pressurestorise with age ~ SYSTOLIC
over the age range 18-79 years while diastolic £
-blood pressures rise until age 45-54 years for £ i
men and age 55-64 years for women, after which g Leene®”
they decline. At younger ages blood pressures are 3
higher for men than for women; atolder ages this 2 100
is reversed. 3
With increasing age there is a tendency for DIASTOLIC
the distribution of blood pressures to be dis-
placed toward higher values (figs. 3, 4). Con-
currently, the relationship between systolic and
diastolic blood pressures is altered. Distributions 50 =
of systolic and diastolic blood pressures are given 5 [ | | | | | ]
in tables 1-17 for eachage-sex group. Many of the 20 30 40 50 60 70 80
numbers presented in these tables have high sam- Age (in years)
Figure 2. Mean blood pressure in adults, by age and sex: United
States.
PERCENTAGE OF SPECIFIED AGE-SEX GROUP
30
20
40
30
20
30
20
30
20
30
20
30
20
30
20
MEN
Years
18-24
A
:
25-34
h
35-44
Al
45-54
All
55-64
All
65-74
100
150
200
: 75-79
wo]
250
oe
Sa
WOMEN
ho
he
le
All
il
1
er
Hin
100
Systolic blood pressure (mm.hg.)
250
40
30
20
30
20
30
20
Figure 3. Distribution of systolic blood pressure of adults, by age and sex: United States.
PERCENTAGE OF SPECIFIED AGE- SEX GROUP
30
20
30
20
30
20
30
20
30
20
nl }
Years
18-24
WOMEN
25-34
35-44
45-54
55-64
| all.
65-74
fh
75-79
50 75
—
100 125
Diastolic blood
50
pressure (mm.hg.)
75
125
Figure 4. Distribution of diastolic blood pressure of adults, by age and sex: United States.
techniques and reliability are discussed in Ap-
pendix II.
Differences in mean values, of course, are
only part of the story. The mean blood pressures
for adults aged 18-79 years were 130.9 systolic
and 78.7 diastolic. However, 16 percent had blood
pressures below both 120 systolic and 70 diastolic,
while the same percentage had either a systolic
pressure of at least 160 or a diastolic pressure
of at least 95. For young men aged 18-24 years
the proportions were much greater at the lower
end of the scale and smaller at the upper end: 26
percent had blood pressures below 120/70 while
only about 2 percent were as high as 160 systolic
or 95 diastolic. For women 75-79 years of age the
distribution was reversed, 2 percent with blood
pressures less than 120/70 and 46 percent with at
least 160 systolic or 95 diastolic. The percentage
of persons with high blood pressures by sex and
age is given in table F.
COMPARISONS AND ANALYSIS
If HES findings for the United States are
compared with findings from other surveys, the
salient features may be more obvious. Three
surveys of general populations were chosen for
comparison. One was a survey of a sample of the
adult population aged 29-62 years in Framingham,
Massachusetts, in which 4,469 persons were ex-
amined.’ The second was a survey made of the
population aged 15 years and over of the town of
Bergen, Norway, in which some 68,000 persons ,
were examined’ The third was a survey of two
districts in Taipeh, Formosa, in which about 9,700
Taiwanese and ''mainland'' Chinese were ex-
amined.” The measurement techniques in all three
surveys were essentially the same as those used
by the Health Examination Survey, although inthe
Formosan survey blood pressure measurements
were obtained at home rather than at a clinic.
Table F. Percent of adults with blood pressure of at least 160 systolic or 95 dia-
stolic, by sex and age: United States, 1960-62
Systolic at Diastolic at Systolic at least
Sex and age least least 160 mm. hg. or dia-
160 mm. hg. 95 mm. hg. stolic 95 mm. hg.
Percent of adults
Both sexes-18-79 years---------- 11.3 10.0 15,9
Men
Total-18-79 years-------=-====e==- 9.3 10.5 15.0
18-24 years-=----=memmmcemcmm meee 0.2 « 1.6 1.6
25-34 years-=---m-m--m-memememe———————— 1.0 4.5 4.8
35-44 years-=------=mmmmmmmmmmmm mmm meno Sd2 12.6 13.4
45-54 year§--===-==----mmemmmmm— mmm 8.9 15.7 18.9
55-64 years---=-=-m=---mcmmmem—————— 17.1 13.6 23.3
65-74 yearS=--==-----memmemmmmm em mem 29.0 14.5 30.3
75-79 years------m---mmmcmmmmm— mmm mmo 40.7 13.8 41.6
Women
Total-18-79 years-----==--=-====== 13,0 2.6 16,7
18-24 year§-====mmmrmmmmcme nm —————————— 0.1 1.1 lil
25-34 years----==mm-mmmeemmecee———————— 1.1 3.0 341
35=4l YJEATrSmmmmm mmm mn mm on 3.8 7.5 8.4
45-54 years----====-emmcmm-mmmee ma 12.8 13.4 18.2
55-64 years-=-===--m-mmmmmeem—————————— 26.1 18.3 31.8
65-74 yearS===rrmrmmmmmmmem— meee —————— 46.9 18.9 49.9
75-79 years=---------mcmmcmmmmmmmeeee 44.0 13.0 45.9
The trends by age reported by the three sur-
,veys were similar to those reported for the United
States by the Health Examination Survey (figs. 5,
6). The resemblance to the Bergen findings is
especially striking. Figure 7 shows the percentage
increase in mean blood pressures from one age
group to the next; these changes, especially for
systolic pressure, are nearly the same for the
two populations. The one exception arises from a
reported drop in the systolic pressure for women
“in the United States between the age groups 65-74
and 75-79 years; it is entirely possible that this
discordance is a result of the small number of
persons aged 75-79 years examined by the Health
Examination Survey. The 95 percent confidence
interval for the mean systolic pressure for women
aged 75-79 years has as its upper bound a value
consistent with a rise in blood pressure from ages
MEN
180
— ).S. 1960-62
=mmmmms Framingham, Mass.
mum Bergen, Norway de
= =m mm w Taipeh, Formosa 7°
al
7
SYSTOLIC
140
&
=
£
- Rd
£ -—— —
”
”
o
a
©
8 100
@
DIASTOLIC
LoL LL] p— —
mannenn———————_ an = il # - — — i.
-.
60 i
ol JL
0 L | 1 | | | |
20 30 40 50 60 70
Age (in years)
Figure 5. Mean blood pressure, by age for men, 18-79 years: four
surveys.
10
i ——— eos eee
WOMEN
— U.S 1960-62
=mmmms Framingham, Mass.
==ms= Bergen Norway
1mm mm: Tgipeh,Formosa
180 - 7
SYSTOLIC y
® 140
E
E
o
5
3
2
@
5
° ———
3
o
100 pb
—
DIASTOLIC mr " —
»' -
% _
m——
60 L_ _
o LL Lo]
20 30 40 50 60 70 80
Age (in years)
Figure 6. Mean blood pressure, by age for women, 18-79 years:four
surveys.
65-74 to 75-79 years. Although it would be rash
to assert that this is, indeed, the fact for the popu-
lation of the United States, it would be equally rash
to accept without question the finding that systolic
blood pressure for women begins to decrease after
75 years of age.
Not only did systolic pressure increase with
age for persons 18-79 years but for most of the
age span the rate of increase was greater the
‘older the person (fig. 7). Whether this applies to
the entire age range or whether itis true only until
age 60 for men and age 50 for women, as the Ber-
gen data suggest, is impossible to determine, in
view of the sample size used in the Health Ex-
amination Survey. With diastolic pressure the rate
of increase was less the older the individual, and
MEN
+15
HO b—— — - -.
r | |
30 40 50 60 70 80
MEN
+0
Percentage increase from preceding age group
-10
30 40 50 60 70 80
Age (in years)
SYSTOLIC
——— |).S. 1960-62
Framingham, Mass |
mn = Bergen Norway
tmmrumm= Toipeh, Formosa
DIASTOLIC
WOMEN
-
L
-
-10 | |
30 40 50 60 70 80
+0
-10 |
30 40 50 60 70 80
Age (in years)
Figure 7. Percentage increase in mean blood pressure, by age for men and women: four surveys.
after age 64 for men and 74 for women diastolic
pressure began to decrease with age.
It must be emphasized that what are reported
here as changes associated with age are not de-
rived from observation of individuals as they get *
older. The Health Examination Survey undertook
only to examine persons at one point in time and
the data reported here are cross-sectional. It is
conceivable that data from one-time surveys
understate the tendency of blood pressure to in-
crease as people get older, since young persons
with high blood pressure are less apttosurvive to
an older age than young persons with low blood
pressure.
Neither is it argued that parallel findings in
different populations demonstrate that the phe-
nomenon of higher blood pressures at older ages
is an essential human characteristic. It has been
argued on the basis of findings in certain primitive
groups that there is no inherent tendency of blood
pressure to rise with age.8. 9 For a number of
reasons—the small number of persons in primi-
tive groups, the difficulty of ascertaining age, and
,the strong selective factor of a high mortality—
such evidence must be regarded with considerable
reservation. However, the Health Examination
Survey has not collected any evidence to distinguish
between biological and cultural factors related to
blood pressure differences.
With respect to sex differences, all four sur-
veys indicate higher blood pressures among young
men than among young women, whereas older men
‘have lower blood pressures than older women
(fig. 8). The age at which the reversal occurs
varies somewhat, According to the findings of the
SYSTOLIC
+20
— U.S. 1960-62
=mam: Framingham, Mass.
= «mm Bergen, Norway
=== un Tgipeh, Formosa
+10
©
=
£
E
o o I
Qo
$
2
bh
3
-10
201 | |
20 30 40 50 60 70 80
DIASTOLIC
+10
©
££
£
E
= #
g
®
8
=
8
-10 | | 1
20 30 40 50 60 70 80
Age (in years)
Figure 8. Mean differénce between blood pressures for men and
women: four surveys.
NOTE: Mean blood pressure formen minus mean blood pressure
for women.
12
Health Examination Survey, blood pressures are
higher for women in the United States than for men
only in age groups 55-64 years and older. The
Bergen and Framingham surveys show this shift
to be a decade earlier. The broad age groups used
in this report somewhat exaggerate the difference
between these surveys. However if the shiftarose
as -the function of some relatively fixed event,
such as the onset of menopause in women, one
would expect greater agreement,
DISCUSSION
Data in this report are based on casual blood
pressures measured indirectly, primarily be-
cause this is the blood pressure determination
most easily made. Although this is no trivial
advantage, there are others. For one thing, this
measurement is readily accepted by examinees,
with the result that a blood pressure measurement
was obtained for every person examined by the
Health Examination Survey with the exception of
one woman who was too obese to be measured with
the apparatus inuse. Any attempt to measure blood
pressure directly—by inserting a catheter into an
artery—would surely have entailed some sample
loss because of refusal or technical failure, as
would have an effort to obtain blood pressures
involving hospital confinement.
Another advantage of casual blood pressures
is that they are immediately referable to clinical
experience. As part of an effort by the Health
Examination Survey to evaluate the possibility of
bias arising from nonresponse, inquiries were
_sent to the physicians of nonexamined persons
asking, among other things, for a report of blood
pressure measurements, if available. Similar
inquiries were sent to the physicians of a match-
ing set of examined persons. Not only was the
average blood pressure measurement reported for
each of these two groups similar—134/80 for ex-
amined and 135/81 for nonexamined—but for
examined persons the average measurement re-
ported by their physicians agreed exactly with
their average measurement obtained by the Health
Examination Survey—134/80 in both instances.
While the advantages of indirect pressures
are numerous, it is necessary to note one of the
disadvantages. This is the possibility (not defi-
nitely proved) that such measurements are af-
fected by differences in upper arm girth. Ragan
and Bordley in a study of S51 young adults found
that for persons with arm girths of about 28 cm. |
the direct and indirect systolic pressures were
nearly the same.” With smaller arms the in-,
direct pressure tended to be too low; with larger
it tended to be too high. In the measurement of
diastolic pressures (fourth phase) the indirect
method tended to give results slightly too great
even with small arm girths, and the disparity be-
came greater as the arm girth increased. Since
the majority of American adults have upper arm
girths in excess of 28 cm. it would follow that the
absolute levels reported for the U.S. population
are higher than a set of direct measurements of
blood pressure would show them to be.
Arm girths tend to increase with age. It might »
therefore be surmised that indirect blood pressure
measurements would exaggerate the true rate at
which mean blood pressures increase with age,
and some studies have introduced ''corrections’
for this effect. Figure 9 suggests that these efforts
are hardly justified. While mean blood pressures
are higher for larger arm girths than for smaller,
the rate of increase of blood pressure with age
seems practically the same for persons of any
specified arm girth as for all persons combined.
Obviously this cannot be completely so, but itdoes
suggest that survey data hardly lend themselves
to such refined analysis.
What makes this measurement artifact es-
pecially unfortunate is that fatter people tend to
have larger arm girths. To what extent the higher
blood pressure associated with a greater arm
girth really is a consequence of a positive
association of blood pressure with obesity has
never been accurately determined. The data from
Ragan and Bordley 1° and from other studies,
while suggesting that for a given direct blood
pressure the indirect blood pressure tends to rise
as arm girth increases, are still too scanty to
provide accurate estimates of the numerical ex-
tent of this effect, or indeed to prove that such
an effect exists.
For this and other reasons, differences be-
tween surveys in the absolute levels of blood
pressure reported are very difficult to interpret.
The difficulty is clearly delineated by Bee et al.
in reporting the data from the Bergen survey. ©
This was a complete survey of the population of
Bergen done in conjunction with a compulsory
X-ray examination. Some 68,000 persons were
732-721 O - 64-3
measured, Between January and June 1950 the
Northern District of Bergen was surveyed, and
from January to May 1951 the Southern District
was surveyed. The Southern District had systolic
pressures for the various age groups 5 to 7 per-
cent lower than the Northern and diastolic pres-
sures for most age groups 1 to 2 percent
higher. (The data from the Northern District are
used in figures 5 and 6. Had data from the Southern
District been used instead, the systolic pres-
ures would have been close to those reported by the
Health Examination Survey, while the diastolic
would have been slightly higher.) Since the popu-
lations in these two areas differed relatively little
by any of the usual indices, the most logical ex-
planation for the reported difference in blood
pressure levels was some minor difference inthe
circumstances of the examination or the measure-
ment technique. Bge ef al. concluded: "The results
seem to emphasize that one should not attach too
much importance to absolute figures and give
warning that it may be dangerous to compare in-
vestigations ...."
The point that emerged from the examination
of the Bergen data was that despite differences in
absolute levels, the trend of blood pressure levels
with increasing age and the sex differentials were
practically identical in the two districts.’ This
basic agreement is not surprising since, for
all practical purposes, both groups were large
samples of the same population. When this is not
the case, and particularly when the populations
are special groups-——such as employed groups,
military personnel, or insured persons—it be-
comes difficult to judge whether the reported
differences reflect selective factors or are pro-
duced by some other means.
Nor can the effects of selectivity be assessed
on an a priori basis. A standard reference for
clinicians for many years has been the data on
blood pressure reported by Master etal.'! These
were obtained from a sample of industrial popu-
lations and civilian employees at military bases
during World War II and appear to derive largely
from pre-employment physical examinations. De-
spite the ostensible peculiarities of this sample,
the findings correspond closely to those from the
Health Examination Survey.
Another factor to consider in judging survey
results is the setting in which the blood pressure
was observed. There is some evidence that blood
13
~
MEN ARM GIRTH WOMEN
180
=e — 36-38 cm.
sysToLIC um SysvoLIC
immmunans 30 -32 cm.
smmmms 27 29 cm.
— 04-26 CM.
160
140 -140
Blood pressure (mm. hg.)
DIASTOLIC DIASTOLIC
pen 0 — —
* * "—~—
7 277
PLT
Ce
»
100 — — - —100
|
|
|
60 =
ol 1
20 30 40 50 60 70 80
Age (in years) Age (in years)
| | To
20 30 40 50 60 70 80
Tr
NOTE: Means based on less than ten measurements omitted.
Figure ©. Mean blood pressure, by age for men and women with specified arm girths: Health Examination Survey.
14
pressure measurements taken ina clinical setting
tend to be higher than measurements taken at
home.!21? Of the four surveys cited in this re-
port only one was conducted at home and this
survey reported generally lower blood pressures
than the others.
Re-examination may partly dissipate the
effect of a clinic setting. In one study, blood”
pressures measured 3 weeks to 4 months after
the initial survey were less by 3.9 mm.hg. for
systolic pressures and 3.6 mm.hg. for diastolic. 4
In another study pressures measured 40 to 80
hours later averaged 5.2 mm.hg. lower for systolic
and 1.5 mm.hg. lower for diastolic pressures. !?
In the Framingham Heart Study’ blood pressure
levels in the sample group were less by 3.8 mm.hg.
systolic and 2.9 mm.hg. diastolic when measured
2 years after the initial survey and the level de-
creased again (although by a lesser amount) at the
next biennial examination. A group of volunteers
who were included in the same survey, and ex-
amined in exactly the same fashion as the sample
persons, did not exhibit this trend. Since the Health
Examination Survey performed only a single ex-
amination and did not accept volunteers for
examination, it is reasonable to assume that in
terms of the circumstances under which they were
obtained the blood pressure data from the Health
Examination Survey are comparable with those
from the first examination at Framingham.
Still another factor influencing the blood
pressure data from the Health Examination Sur-
vey was the use of a large number of physicians.
Altogether 62 physicians were employed, each
examining about 80 persons. It is clear thatthere
was a measurable difference among physicians
in their blood pressure determinations. This
difference presumably has two causes. The first
is what Ayman and Goldshine !5 called "the
pressor effect of the physician's presence’ onthe
patient, an effect which may be assumed to vary
from one physician to another. The second arises
from differences in measurement technique. When
a measurement depends upon one's hearing
changes in sound while simultaneously observing
the level of a rapidly moving column of mercury,
it must be taken for granted that, other things
being equal, different observers will make dif-
ferent determinations. The extent of such differ-
ences is discussed in Appendix III, There is little
indication that this observer variation has biased
the blood pressure findings of the Health Examina-
tion Survey, but it does decrease their precision.
SUMMARY
1. Mean systolic pressure in the U.S. popu-
lation rises over the age range 18-79
years, the rate of rise tendingto increase
with age. Mean diastolic pressure rises
until 45-54 years of age for men and 55-64
years for women; atolder ages it declines.
2. Under age 45 blood pressures are higher
for men than for women; over age 54 blood
pressures are higher for women than for
men.
3. Findings for other population groups are
generally similar to those for the United
States.
4, A larger arm girth is associated with
higher blood pressures. For any specified
arm girth, however K mean blood pressures
rise with age. There is little diurnal varia-
tion in mean blood pressure.
5. Blood pressure levels presented in this
report seem comparable with those ob-
tained in the usual clinical situation and
are similar to the standards presently
in use in the United States.
15
REFERENCES
9.5 National Health Survey: Plan and initial program of the
Health Examination Survey. Health Statistics. PHS Pub. No. 584-
A4. Public Health Service. Washington, D.C., May 1962.
National Center for Health Statistics: Cycle I of the Health
Examination Survey: sample and response. Vital and Health St1-
tistics. PHS Pub. No. 1000-Series 11-No. 1. Public Health Serv-
ice. Washington, D.C., Apr. 1964.
3Glock, C. Y., Vought, R. L., Clark, E. G., and Schweitzer,
M. D.: Studies in hypertension. II. Variability of daily blood pres-
sure measurements in the same individuals over a three-week period.
J. Chron. Dis. 4(5):469-476, Nov. 1956.
4
Edwards, J. C.: Management of Hypertensive Diseases. St.
Louis. C. V. Mosby Company, 1960.
Kagan A., Gordon, T., Kannel, W. B., and Dawber, T. R.:
Blood pressure and its relation to coronary heart disease in the
“ramingham study. Hypertension. 7:53, 1959.
6hge, J., Humerfelt, S., and Wedervang, F.: The blood pressure
in a population, blood pressure readings and height and weight de-
terminations in the adult population of the city of Bergen. Acta
med. scandinav. Supplementum 321. Bergen, 1957.
7140, T-Y, Hung, T-P, Chen, C-M, Hsu, T-C, and Chen, K-P:
A study of normal and elevated blood pressures in a Chinese urban
population in Taiwan (Formosa). Clin. Sc. 18(2):301, 1959.
Bays, R. P., and Scrimshaw, N.S.: Facts and fallacies regard-
ing the blood pressure of different regional and racial groups. Circu-
lation. 8(5):655, 1953.
16
000
9 Epstein, F. H.: Epidemiological studies on the nature of high
blood pressure. Proceedings of the Fifteenth Annual Conference
on the Kidney. To be published.
1054520, C., and Bordley, J.: The accuracy of clinical meas-
urements of arterial blood pressure, with a note on the ausculta-
tory gap. Bull. Johns Hopkins Hosp. 69(6):504, 1941.
11
Master, A. M., Dublin, L. I., and Marks, H. H.: The normal
blood pressure range and its clinical implications. J. A.M. A. 143
(17): 1464-1470, Aug. 1950.
12 man, D., and Goldshine, A. D.: Blood pressure determina-
tions by patients with essential hypertension. I. The difference
between clinic and home readings before treatment. A.].M.Sc. 200
(4):465, 1940.
Bcomstock, G. W.: Anepidemiological study of blood pressure
levels in a biracial community in the southern United States. Am.
J]. Hyg. 65(3):271, 1957.
L4amileon, M., Pickering, G. W., Roberts, J. A. F., and Sow-
ry, G. S. C.: The aetiology of essential hypertension, the arterial
pressure in the general population. Clin. Sc. 13(1):11, 1954.
15
Ayman, D., and Goldshine, A. D.: Blood pressure determina-
tions by patients with essential hypertension. II. The difference
between home and clinic readings during and after treatment. Am.
J.M.Sc. 210(2):157, 1941.
Table
Ls
oF,
10.
11.
12.
13.
14.
15.
16.
17.
Number
DETAILED TABLES
of adults aged 18-79 years, by specified systolic and diastolic blood
pressures; United SLATER; 1980m00wimmmimmiomim sm imionim mim iis sin ii mw ude mi odie mo of 101 mf ioe 8 rt
Number of men aged 18-79 years, by specified systolic and diastolic blood pres-
BUTES: UNLLd SCALES; 16052 wm ww mm mmm mm mmm ow mmm mmm mm mm mt tm tt mm re et
Number of women aged 18-79 years, by specified systolic and diastolic blood pres-
Sues] Unlted SEATS, NOBU =B2 mm mimmm smi mm mim imtoo otrmsim ras erie so oo 060 0 1 a tp
Number of men aged 18-24 years, by specified systolic and diastolic blood pres-
SUrEs: United SCates, LIB0=02~ =i im=iemer imikimi sr hmm ie io oe ed ok kit ndinsoh oon oho 8 fo ok 8 wl
Number of men aged 25-34 years, by specified systolic and diastolic blood pres-
sures: United SLates, LO60m062= mmm mmmmim mmm mim im mim mmm sim mo mr mm i mr ee
Number of men aged 35-44 years, by specified systolic and diastolic blood pres-
SUres: United STATES, 1LT60m02m mmr mm mm mm mmm mm mim mom mom om mim mt mt mm mo mm em mt ts
Number of men aged 45-54 years, by specified systolic and diastolic blood pres-
SUres: United STATES, LOO0 m0 =r mmm mmm mimim mim kimi sh ms ss simim oind we mm oe ob mm om ei 4
Number of men aged 55-64 years, by specified systolic and diastolic blood pres-
ures: United STates, 1980m582m =m mmmmm im mim mim mimi mim mim mm mt mt mt ttm ttm tt tf tm
Number of men aged 65-74 years, by specified systolic and diastolic blood pres-
SUres: Uniied STates, LOB m6 mmr mmimm mm mimmim mm mimi mms mmm mses mms me ion mm mm 00 at 1 0 to 0 ste
Number of men aged 75-79 years, by specified systolic and diastolic blood pres-
SUred: United STates, 1980-00 = sm mim mmm minimis mms edi ood oh ie oe sn 0, bo a 0 1 a 6
Number of women aged 18-24 years, by specified systolic and diastolic blood
pressures? United States, L960-52=== =r imma sim mms miso moms emi is min mss om mk ie
Number of women aged 25-34 years, by specified systolic and diastolic blood pres-
sures: United SCALES, 1960=60 = == mm mimi imine. riod ioind find dnindiom eisai ue som so mei 0 0 ws et 10 0
Number of women aged 35-44 years, by specified systolic and diastolic blood pres-
ures: United SCALES, 19B80=00 =m mmm =immimisimmimintiotimiotiimlin imino mon iti mm oh 10 00 of i 40 56 4
Number of women aged 45-54 years, by specified systolic and diastolic blood pres-
sures: United Sates ;~ 1T60=02= = mmm mmm mimi mim meio moo io a at cc we
Number of women aged 55-64 years, by specified systolic and diastolic blood pres-
SUES! United STATES, LOB0 mB m mmm mmm mr mm mm mm mrmiom ms cmon so mead os ao eh tm 0 8
Number of women aged 65-74 years, by specified systolic and diastolic blood pres-
sures: United States; 1OB0m0T = rmimmimmrimimumneiiesns me ums ried sme oe ee m oe ie mm i
Number of women aged 75-79 years, by specified systolic and diastolic blood pres-
SUresS! UnLLed STATES, 10B0=02 =m mmm mim mrmimic mn mim mimo ce orimm iat de a sm
Page
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Table 1. Number
of adult
Ss
aged 18-79 years, by specified
States,
1960-62
systolic and diastolic
blood pressures: United
Systolic blood
Diastolic blood pressure (mm. hg.)
pressure (mm. hg.) Undez
Total 50 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89
Number of adults in thousands
Total------ 111,087 898 1,124 2,661 6,664 13,290 16,984 | 21,078 16,977 12,453
259 43 52 102 12 18 24 9 - -
3,248 160 239 537 1,006 843 280 127 56 -
12.849 254 374 1,006 2,153 3+631 3,187 1,769 384 71
23,32) 184 266 644 1,947 4,216 6,510 5,831 2,715 835
22,883 80 125 291 1,019 2,448 3,869 6,243 5,206 2,667
17,844 30 55 14 242 1,114 1,693 3,906 4,369 3,604
11,073 - - 25 133 417 494 1,665 2,210 2,435
7,076 45 14 27 60 212 505 767 814 1,209
4,499 58 - 15 34 211 142 397 690 586
3,021 44 - - 16 143 62 165 222 424
1,926 - - - 41 - 72 56 219 286
1,472 - - - - - 105 80 32 218
774 - - - - - 26 26 - 60
399 - - - - 18 - 21 - 26
194 - - - - - 16 - - 32
135 - - - - - - 16 - -
11 - - - - - - - - -
32 - - - - - - - - -
Z3 - - - - - - - - -
Diastolic blood pressure (mm. hg.) —Con.
Systolic blood
pressure (mm.hg.)
90-94 95-99 100-104 | 105-109 | 110-114 | 115-119 | 120-124 | 125-129 | 130-134 135+
Number of adults in thousands—Con.
Total-==---- 7,764 4,995 2,597 1,654 665 607 268 171 81 154
98 14 - - - - - - - -
157 177 - - - - - - - -
1,758 771 210 46 32 - - - - -
2,002 1,094 430 139 29 - - - - -
1,202 1,330 S11 271 56 52 - - - -
799 640 413 303 118 66 29 - - -
515 462 398 312 90 131 38 - - -
351 221 256 163 130 99 31 - - -
149 218 198 223 68 67 93 - 21 -
101 67 137 43 74 74 52 66 18 29
16 - 44 48 54 91 16 32 32 -
16 - - 43 16 - - 60 - 11
- - - 63 - - 8 12 10 25
- - - - - - - - - 11
- - - - - 13 - - - 18
- - - - - 13 - - - 60
SEITE
Table 2. Number of men aged 18-79 years, by specified systolic and diastolic blood
1960-62
pressures: United States,
Systolic blood
Diastolic blood pressure (mm. hg.)
pressure (mm.hg. Under ?
Total 50 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89
Number of men in thousands
Total====== 52,744 546 423 986 2,614 5,748 7,49) 10,640 8,044 6,642
Under 90---====-=-- 43 21 - 23 - - - - - -
DOGG nmin 696 59 34 92 187 143 99 44 39 E
LOU=10 mswmm mime 4,137 133 111 300 743 911 | 1,087 690 90 71
L101 1S www mswmm 10,157 150 141 © 349 791| 1,912| 2,644 | 2,568 | 1,165 407
120-129 ===m=mmmm= 12,375 80 89 208 559 | 1,588 | 1,934 | 3,476| 2,588] 1,328
130-130 mmm 10,268 30 47 # 154 674 967 | 2,313| 2,110 2,071
LEO=Lh Gms 6,194 = - 14 95 288 311 800 | 1,131] 1,456
LBO~LBG rm smite 3,960 30 - - 37 114 315 549 355 721
160-188 meme 2,053 44 = = 15 75 53 133 400 286
L7G -Y Tn wnnmmms 1,309 - = = 16 44 20 66 152 178
180-180 mmmmmmns 604 - . = 17 - 30 - 14 125
LILI rcs meres re 501 - - - - - 2 2 = »
200-209 == =mm =m === 248 - 2 - - - 14 - ”
FIG EG remem 74 = - = * - - - - -
RID= Gremio 77 - - - - = 16 2 # 2
230m 23 mmm msm mm 27 - 2 = . 5 - - " »
260-249 mmm mm mmm mm # # - - - - » - - n
250-259 === mmmum 18 - - - - - - - - -
Diastolic blood pressure (mm. hg.) —Con.
Systolic blood
pressure (m.hg) | g4.9, 95-99 | 100-104 | 105-109 | 110-114 | 115-119 | 120-124 | 125-129 | 130-134 | 135+
Number of men in thousands —Con.
Total=====-= 4,050 2,695 1,222 800 375 228 107 25 49 58
Under 90-======== - - - - - - - - - -
90-99 -mmmmmmmmman - - - - - - - - - -
100-109 ========== = - = - - - = - - =
LUOLLD mmm mm mmm 31 - - - - ® = ” -
120-129 ===mmmmmm= 383 141 = - 5 - - - = -
130-130 me smmmin 1,181 558 134 28 - - - - - -
LUO LL rms mmmemm se 1,128 558 279 107 29 = - * - "
150-15 wm mmm ee 658 751 221 130 43 36 * - * -
160-169 ====mn= === 298 307 167 139 98 26 13 - - -
L70=L7G mmm mim 134 198 152 200 72 49 28 - - -
180-180 mmm mie 105 S54 70 70 47 54 16 - # =
190-199 =m mmmmmmmm 77 104 153 69 20 34 24 - 21 -
200m 209 mmm mm mm 56 24 47 14 25 29 10 - - 29
ZU0=21 mm mmm mime - = - - 40 - 16 # 18 ®
SAGES wom - 2 5 26 = " = 25 ® 11
230-239«--==-==-= - - - 17 - - - - 10 -
240-249 mmm mmmmm - - - - - - - - - -
250-259 ~====mmmnn - - - - - - - - - 18
Table 3. Number of women aged 18-79 years, by specified systolic and diastolic blood pressures: United
States, 1960-62
Diastolic blood pressure (mm. hg.)
Systolic blood
pressure (mm. hg.) Under
Total 50 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89
Number of women in thousands
Total-=---- 58,343 352 701 1,675 4,050 7,542 9,493 10,4391 8,934 5,811
216 22 52 7° 12 18 24 9 - -
2.551 102 205 446 819 700 181 82 17 -
8,712 121 262 706 1,409 2,740 2,100 1079 294 -
13,163 34 125 295 1,156 2,304 3,866 3,263 1,610 428
10,508 - 36 83 460 861 1,934 2,767 2,619 1,339
7,576 - 7 14 88 441 726 1,594 2,259 1,533
4,879 - - 10 39 129 133 865 1,079 979
3,116 15 14 27 24 98 190 218 460 488
2,446 14 - 15 19 136 88 264 290 300
1,713 44 - - - 99 41 98 70 247
1,322 - - 24 - 42 56 204 161
971 - - - - - 105 80 32 218
526 = = - - - 12 26 - 60
324 - - - - 18 - 21. - 26
116 - - - - - - - - 32
108 - - - - - 16 - -
11 - - - - - - - - -
13 - - - - - - - - -
13 - - - - - - - -
Diastolic blood pressure (mm. hg.) —Con.
Systolic blood
pressure (mm.hg.)
90-94 95-99 100-104 | 105-109 | 110-114 | 115-119 | 120-124 | 125-129 | 130-134 135+
Number of women in thousands—Con.
Total------ 3,714 2,299 1,375 854 290 379 161 146 33 96
67 14 - - - - - z - -
374 36 - - - - - - - -
577 213 76 17 32 - - - - -
875 536 151 32 - - = - - 5
545 578 290 141 12 16 - - - -
501 333 246 164 20 41 16 - - -
381 264 246 112 17 82 10 - - -
246 167 187 93 82 44 15 - ¥ =
72 114 45 153 48 33 69 - - -
45 43 90 29 49 45 42 66 18 -
16 - 44 48 13 91 - 32 14 -
16 - - 18 16 - - 35 - -
- - 46 - - 8 12 - 25
- - - - = - - - - 11
- - - = ~ 13 - ~ - -
- - - - - 13 - - - 60
20
Table 4. Number
of men aged 18-24 years, by specified
United States,
systolic and diastolic
1960-62
blood pressures:
Diastolic blood pressure (mm. hg.)
Systolic blood pressure (mm.hg.)
Total under 50-54 | 55-59 | 60-64 | 65-69 | 70-74
Number of men in thousands
Total-=mmemcem ccm —————— 7,139 358 256 336 914 951 1,205
Under 90----==-reemcrcceccee————— 21 21 - C - - -
O00 1D mm mm 155 31 34 48 - - 16
100-109======mmmmm mmm mcm mmm 999 82 39 112 254 197 144
110-119====mmmmm emcee mem mmm 2,178 95 108 122 316 431 495
120-129=====mmmmm mmm mmc 1,896 68 61 54 174 199 346
130-139=====mmmmmm mmm mmm 1,197 30 13 - 107 117 130
140-149 == mmmmmre mmc cme ————— 521 - - - 26 13 34
150-159 =m mmm mmm mmc mn ———————— 156 30 - - 37 - 38
160+ ===mmmmmmmmm meme meme 15 - - - - - -
Diastolic blood pressure (mm. hg.)=—Con.
Systolic blood pressure (mm. hg.)
75-79 80-84 85-89 90-94 95-99 100-104
Number of men in thousands--Con.
Totalemm mmm ——— 1,494 726 400 371 96 23
Under 90===mm=rmmmmes simi = o - = - -
90=99-==mmmmmm mmm meme cm ———— - 26 - - - -
100-109 =m mmmmmmmm— ec ——————————— 105 10 57 - - -
110-119 = mmm mmm mee ————————— 416 149 46 - - -
120-129====mmmmmm mcm mmm ——— 648 255 90 - - -
130-139=====mmmmmmm mmm mmm 246 241 55 193 63 -
140-149====mmmmemmmc mem —— mm 42 45 139 178 33 10
L500 mm 1.5 Gow mm mm mm 37 - 14 - w -
160+ mm om mee om om om mm om mm en - - - - - 13
21
Table 5. Number
of men aged 25-34 years, by specified
United States,
systolic and diastolic
1960-62
blood pressures:
Systolic blood pressure (mm. hg.)
Diastolic blood pressure (mm. hg.)
Total Under 50-54 | 55-59 | 60-64 | 65-69 | 70-74 | 75-79
Number of men in thousands
Total--=--===mcmcmmccenmnenn 10,281 129 51 165 768 1,494] 1,773] 2,275
Under 90====mmmmremenmm mmm ——— - - » - -
90-99--=-mmmmm mmm meme eee 203 28 - - 113 62 - -
100-109=====mmmmmmmm amc c mcm —————— 1,129 50 7 76 152 289 258 283
110-119-=====-mmm mmm mmm mmm mmm 2,390 24 14 89 217 506 602 571
120-129-mm=m- momen 3,187 12 13 - 257 440 489 958
130-139===mmmmmm mmc mmm mmm 2,025 - 17 - 28 156 327 354
140-149=mmmmmm mmm cme mmm mee 927 - - - - 40 55 54
150159 mmm mm mmm me mm on mm om me me 311 - - - - - 42 56
160-169=====mmmmmmm mmm mmm mm 88 14 - - - - - -
170-179==mmmmme mcm cmc ccm mmm mm 8 - - - - - - -
180+-mmmmmmmre mmm ——————————— 13 - - - - - - -
Diastolic blood pressure (mm. hg.)=--Con.
Systolic blood pressure (mm.hg.)
80-84 | 85-89 |90-94 [95-99 [100-104 |105-109 110-114 | 115+
Number of men in thousands--Con.
Total====smmmmee cece cen 1,706 | 1,029 424 250 65 116 - 37
Under 90-====m=rmemmmm mmm nee ——— - - - - - - - -
90-99 mm mmm m mmm —————————— - - - - - - - -
100-109=-mr=mmmm meme — a ——————— 14 - - - - - - -
110-119===m=mmmmmc mcm mcm mm 318 48 - - - - - -
120-129--===mmmmm mmc ————————— 630 260 116 11 - - - -
130-139--mmmmmm mmm mm ————————— 453 421 129 133 7 - - -
140-149==mmmmmmm mmc ccm ccm 252 255 147 52 13 60 - -
150-159= =m mmmmm mmm mm ——————— 26 38 31 53 11 43 - 11
160-169-=mmmmmm mmm mmm mm ———— 13 8 - - 26 13 - 13
170-179===mmmmemm mmc ce cee ———— - - - - 8 - - -
180+-~=mmmmmmm mmc —————— - - - - - - - 13
22
Table 6.
Number of men aged 35-44 years, by specified
systolic and diastolic
United States, 1960-62
blood pressures:
Systolic blood
Diastolic blood pressure (mm. hg.)
PhpaSTS (0, 50) Total Under | 50.54 | 55-59 | 60-64 | 65-69 | 70-74 | 75-79
Number of men in thousands
Total=meme=emecmaaaan 11,373 14 86 205 408 922 1,736] 2,449
Under 90-====-omcmmcccaaa= - - - - - - - -
90-9 mmm mm mmm mmm ————— 95 - - 18 34 . 31 11
00 L0G wm wins simi imi i 1,012 - 52 76 175 127 373 142
L10-110-= snmmmmmamsnsmmmn 2,755 14 19 56 122 344 771| 901
120-129=--=mmmmmcc ccm em 2,894 - 15 55 77 346 380 700
130-139-=-cccmmmmmmemmeeam 2,153 - - - - 84 115 497
140-149 == =mmmmmmmmmmmmmmmm 1,171 - - - - - 14] 148
L50= 158 wm wmummmmmrn mmm 703 . - - - 21 20 39
160-169---==c-mmmmcceccan- 267 - - - - - 31 12
LTO LT Dw mimmumimnimmm minim 240 = = - = - » =
180-189 ====m==mmmmmmmmmemm 21 - - - - - - -
190-199-=--memmmmmmmmm eam 12 - - - - - - -
BOQ reese si ss sr em sme 50 - - - 2 - : -
Diastolic blood pressure (mm. hg.) —Con.
Systolic blood
pressure (mm. hg.)
80-84 | 85-89 | 90-94 95-99 | 100-104 | 105-109 | 110-114 | 115-119 | 120+
Number of men in thousands—Con.
Total====mmmmmeeeau- 1,756 | 1,426 929 797 | 295 175 94 13 67
Under 90--==mcecemeceanoa- - - - - - - - - -
90-99 --mmmmmmm mmm eee - - - - - - - - -
LO0= 0mm mms as wes 52 15 - - - - = - -
LAD L LD mmm imation 360 | 138 31 . - - - . .
120-120 mnmnmmnmnmnnnanunns 711| 417] 130 63 - - - = -
LDL IG wmmmmmmnn sins anes 400 | 525) 298 181 41 11 - = s
EDL mmm minim win a 206 | 248| 252) 161 118 24 - = .
150-159=====mmmmmmmmeenana 13 69 194 216 74 13 43 - -
LOD LG mmm wm sme sm womans se 13 » 7 79 32 52 41 . .
LTO LTD wm mits re rs sms . 15 17 97 9 62 9 13 17
180 = L8G we em srs sr im - - - - 21 - - - -
190-199---====cmcmmmmmmma - - - - - 12 - - -
DOs ism mania isos - . . - - . . - 50
23
Table 7. Number of men aged 45-54 years, by specified systolic and diastolic blood pressures: United
States, 1960-62
Diastolic blood pressure (mm. hg.)
Systolic blood
pressure (mm. hg.) Under
Total 50 50-54 | 55-59 | 60-64 | 65-69 | 70-74 | 75-79 | 80-84 85-89
Number of men in thousands
Total========n-=- 10,034 30 - 70 139 798 | 1,176 1,871] 1,794 1,726
90-99 --mmmmm mmm ee 120 - - 17 12 21 24 33 13
100-109 ====mmmmmmmema 526 - - 37 56 175 203 42 14 =
110-119-==memmmmmeee 1,523 - - - 54 380 401 383 219 85
120-129-=cmmmmmmcceeem 2,200 - - 17 - 90 398 566 593 381
130-139===mmmmmmm meme 2,575 - - - - 30 107 626 658 626
L401 40mm vm mim mmm cons om 1,258 - - - 17 88 17 125 183 304
LOL TD mm msm wom mo 941 - - - - 15 27 95 56 205
160-169 -===m=memuanum- 467 30 - - - - - - 59 88
170-179 ====mmmcmmmmm mm 197 - - - - - - - - 14
180-189 --mmmmm mmm mmm 133 - - - - - - - - 23
190-199 -==mmmmcmmmnnam 56 - - - - - - - - -
200-209 -======mmmmmmm 38 - - - - - - - - =
Diastolic blood pressure (mm. hg.) —Con.
Systolic blood
pressure (mm. hg.)
90-94 | 95-99| 100-104| 105-109 | 110-114 | 115-119 | 120-124 | 125-129| 130-134 | 135+
Number of men in thousands—Con.
Total=======-=-- 847 701 358 223 133 95 51 - 8 13
90-99 -mcmmmmme meee - - - - - - - = - -
100-109 === mmm meme mmm mm - - - - - - - - - -
110-110 vem mm emma msn | - * = = = = = - # =
120-129-===cmmmmmmcaan | 110 46 - - - - - - - -
130-139-==mmmmmmm mmm 298 167 46 17 - - - - - -
140-149 -==m mmm mmm mm mm 220 158 108 9 29 - - - - -
150-159 -==-=mmmmmmmmam | 108 258 104 48 - 23 - - - -
160-169-==-=mmmeemuan- 111 53 24 41 47 16 - - - -
170-179--======nmmmam= - 20 42 31 32 28 11 - - -
180-189 -=mmmmmmmm mmm - - 20 57 - 17 16 - - -
190-199--=-=-mmmmmmm - - 15 - - 8 24 - 8 -
200-209 -=======cmmmm mm - - - - 25 - - - - 13
24
Table 8. Number of men aged 55-64 years,
by specified
systolic and diastolic
blood pressures: United
States, 1960-62
Diastolic blood pressure (mm. hg.)
Systolic blood
pressure (mm. hg.) Under
Total 50 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89
Number of men in thousands
Total-=========- 7,517 - 13 80 161 652 726 1,444 1,111 1,305
90-99---=--mmmmmemm 41 - - 2 - 32 - - - -
100-109---== mm mmm ————- 307 - 13 - 63 78 73 81 - -
110119 mmm mim mm mmm meme 893 - - 23 28 184 288 223 88 59
120-129-===mmmm mmm ———— 1,361 - - 33 36 267 162 380 278 157
130-139 =wmmmr mm ———— 1,29) w - - - 80 91 347 209 307
140-149 -===mmmmmmmm mm 1,394 - - 14 19 12 80 180 240 374
150= 159 mmm wm om mim wm we 945 - = = - - 18 172 100 233
160-169 ==mmmmmm— mmm 580 - = - 15 - - 61 133 126
170-179-==mmmmm mm ————— 327 - - - - - - - 63 18
180-189 -===mmmmmmmmmmm 149 - - - - - 14 - - 31
190-199-===mmuu= —————— 115 - - - - - - - - -
200-209-===smmmmmmm mmm 66 - - - - - - - - -
210-219-=--==emmmmmmmmm 18 - - - - - - - - -
220-229 mmm mm mmm em 29 - - - - - - - - -
230-239 -==mcmmmm mmm - - - - - - - - - -
240-249 mmmm mmm mmm - - - - - - - - - -
250-259 -=mmmmm mmm ———— - - - - - - - - - -
Diastolic blood pressure (mm. hg.) —Con.
Systolic blood
pressure (mm. hg.)
90-94 | 95-99 | 100-104 | 105-109 | 110-114 | 115-119 | 120-124 | 125-129 | 130-134 | 135+
Number of men in thousands—Con.
Total==mmmmmm——-— 999 513 226 146 47 47 = - 18 29
90-99--===mmmmmmm————— - - - - - - - - - -
100-109--=mmmm mmm - - - - - - - - - -
110-119-==mmmmmmmm - - - - - - - - - -
120-129-===-cmmemmm mmm 27 21 - - - - - - - -
130-139-==-mmmmmmm mm 204 14 40 - - - - - - -
140-149 ---cmmmm mmm mmm 309 121 30 14 - - - - - -
150-159-=====-mmmmmmu- 187 188 32 14 - - - - - -
160-169--====m===-mmu- 117 85 33 - - - - - -
170-179-=====cmmmmmmm = 79 - 42 86 31 - - - -
180-189-=mmmmmmm mma 35 37 - - 16 16 - - - -
190-199--===mmmmmmean-" - 46 37 32 - - - - - -
200-209-===mmmmmmmm——— 40 - 1) - - 14 - - - -
210-219---=mm mmm mmm - - - - - - - - 18 -
220-229 -==mmmmm mm mmm == - - - - - - - - - 11
230-239-==mmmmmmm mmm - - - - - - - - - -
240-249 mm mmm mm mmm - - - - - - - - - -
250-259 -==~mmmmmmmmm mm - - - - - - - - - 18
25
Table 9. Number of men
aged 65-74 years, by specified
United States, 1960-62
systolic and diastolic
blood pressures:
Systolic blood
Diastolic blood pressure (mm. hg.)
pressure (mm. hg.)
Total 55-59 60-64 65-69 70-74 75-79 80-84 85-89
Number of men in thousands
Total--------- 4,972 93 102 710 789 898 677 623
23 23 - - - - - -
42 - - 15 27 - ” ”
141 - 23 45 35 38 - -
309 41 14 67 86 52 17 32
726 29 15 232 159 201 65 24
858 - 18 94 198 223 149 118
749 - 16 108 111 198 | 150 113
681 - - 78 85 101 124 139
474 - - 49 23 38 103 44
414 - 16 22 20 46 69 115
150 - - 16 - - 39
253 - - - - - - -
53 - - - 14 - - -
56 - - - - - - -
16 - - - 16 - - -
27 - - - - - - -
Diastolic blood pressure (mm. hg.) —Con.
Systolic blood
pressure (mm. hg.)
90-94 | 95-99 | 100-104 | 105-109 | 110-114 | 115-119 | 120-124 | 125-129 | 130-134
Number of men in thousands-—Con.
Total-------=-- 357 250 205 99 81 62 16 - 10
Under 90----------== - - - - - = = = -
90-99----=cmmmmmee om - - - - - - - - -
100-109----=--==---- - - - - - - - - -
110-119------===-=-- - - - - - - - - -
120-129-------anauun - - - - - - - - -
130-139-==-c-mmoueu- 58 - - - - - - - -
140-149-==cmcmcmcenan 21 32 - - - - - - -
150-130 = msm 120 23 - 11 - - 4 - -
160-169-----ceeeu-== 63 76 37 33 9 - - - -
170-179-===mcmeeceeu- 38 52 36 - - = - - -
180-189~==c-memeemam - - 30 13 31 22 - - -
190-199----ccmceceeae-- 56 58 88 25 = 25 - - -
200-209-----=cccee-- - 9 14 - - 15 - - -
21042) Dm mmwmuninin ni # z = = 40 = 16 = =
220-229 mmmmmmmee mem - - - - - - - - -
230 = 23 wm mmm mm 2 - - 17 - = - - 10
26
Table 10. Number
of men aged 75-79 years, by specified
United States,
1960-62
systolic and diastolic
blood pressures:
Systolic blood
Diastolic blood pressure (mm. hg.)
presnme (wi be.) Total ger | 50-54 | 55-59 | 60-64 | 65-69 | 70-74 | 75-79
Number of men in thousands
Total===-====e=n=-u= 1,428 16 17 38 121 214 86 209
90-99---cmmmcmmmm eee eam 40 - - - 28 13 - -
100-109-===cmc emma 2) - - - 21 - - -
110-119 =mmmmmmmm mmm em 109 16 > 18 38 - - 22
120-129-===meemmemennnan 111 - - 19 - 14 - 23
130-139-=-====mmmmmm mmm 168 - 17 - - 113 - 19
I Ee 174 - - - 17 27 - 53
150-159--=cmmmmmmmm nema 223 - - - - - 86 49
160-169--=====-mmmmmee—m 178 - - - - 26 - 23
170-179--=====m=mmmmmmm- 107 - - - - 21 - 20
180-189-=-=mmcmmmmm mmm 151 = - - 17 - - "
190-199 === =mmmmmmmmnnae= 53 - = z - 2 - -
200-209-=========mmmmmu= 66 - - - - - - -
210-219--====--=mcmmmmm= - - - - - - - -
220-229---==-=mmmmmmmmm 26 - - - - - - -
Diastolic blood pressure (mm. hg.) —Con.
Systolic blood
pressure (mm. hg.)
80-84 85-89 90-94 95-99 100-104 105-109 110-114
Number of men in thousands—Con.
Total============= 274 132 122 89 49 40 20
90-99-=mmmmm mmm mma - - - - - - -
100-109---=-memmmeecnnam - - - - - - -
110-119-===-==eemmmmmen 15 - - - - - -
120-129-==-mmmmmmmcme mm 54 - - - - - -
130-139-=-===-mmmmmmmmmm - 19 - - - - -
140-149 -=====mmmmmmmm mmm 55 22 - - - - -
150-159 -====cmmmmmmm mmm 35 23 16 15 - - -
160-169-======-mmemmmm 80 20 - 14 15 - -
170-179--===mmmmmmm mmm 20 16 - 29 - - -
180-189 -=-==mmmmmmm mm mmm 14 32 74 17 = - -
190-199--==-=-mmmmmmmmmam - - 20 - 13 - 20
200-209--=======m=mmmmu- - - 16 15 21 14 -
210-219 ===m=mmman———— - = = - - - -
220-229-----mmmmmm mm mmm - - - - - 26 -
27
Table 11.
1960-62
Number of women aged 18-24 years, by specified systolic and diastolic blood pressures:
United States,
Systolic blood pressure (mm. hg.)
Diastolic blood pressure (mm. hg.)
Total| U39€¥| 50-54| 55-59| 60-64] 65-69| 70-74
Number of women in thousands
Total===r===-reeece ccc; cece ———— 8,430 177 357 656 1,316] 1,728 1,804
Under 90----=m-e-mmecececc cece cece ————— 67 10 18 30 - - -
90-99 mmm mmm meee eee 1,031 75 75 186 409 194 57
100-109 =mmmm mmm mmm meme meme 2,773 93 113 276 493 868 676
110-119 mmm mm mmm mm meme eee 2,508 - 109 147 323 448 686
120-129 ==mmm mmm mmm eee m———mmm eo 1,497 - 36 17 61 185 346
130-139--rm-rmemmmce remem mm ———————————— 448 - 7 - 29 34 40
140-149 mmm mm mmr nm mm mn 64 - - - - - -
150-159 mmm mmm mm nn mn mm mm mm mn mn 34 - - - - - -
160-169==-=mmrmmemmce cmc mmm ———————————— 7 - - - - - -
Diastolic blood pressure (mm. hg.)=--Con.
Systolic blood pressure (mm. hg.)
75-79 80-84 85-89 90-94 | 95-99 100+
Number of women in thousands=——Con.
Total-===-rmmmccccc ccc cece cee ————————— 1,202 798 219 75 79 12
Under 90-===mmeremeccecc ccc ccc —————————— 9 - - - - -
90-99--qmmmrmmmce nce meme meee ————————— 37 - - - - -
100-109==rmmmmce mmc meee ——————————————— 177 78 - - - -
110-119-=-mmmmrm mmm ——— 458 234 84 19 - -
120-129=mmmmmr mmm cnn mm ————————————————— 423 304 73 38 15 -
130-139 mmm mmm on om om om on om om mm nm 99 164 39 - 35 -
140-149 === mm mmm meme eee meee mem - 17 - 7 28 12
1507 159mm mmm mmm om mmm om om om mmm nm mmm mm mn nme - - 23 11 - -
160-169==mmmmmmmm mcm c meee — mem ———————————— - - - - - 7
28
Table 12.
1960-62
Number of women aged 25-34 years, by specified systolic and diastolic blood pressures:
United States,
Systolic blood pressure (mm. hg.)
Diastolic blood pressure (mm. hg.)
Total Wider 50-54 | 55-59 | 60-64 | 65-69 | 70-74 | 75-79
Number of women in thousands
TOTAL mmm mow om on om om 1.291 74 275 529 1,286 2,098 2,327) 2,052
Under 90--===mmmrmmm_——————————— 102 13 12 49 12 - 16 -
90-99=mmmmmmmmnnnemn——n—————— 928 27| 110 | 168 253 326 29 16
100-109 === mmm m—— ce ————————— 2,440 - 136 216 503 697 537 259
110-119 vm mmmem mm. ——————————— 4,174 34 16 75 389 787 1,289 | 1,028
120-129 mm mm mmm mmm mm mm om om om mm 2,255 - - 21 112 234 427 497
L300 130 mmc sem om mt wr 0 935 - = - 17 54 30 202
140-149-mmmmmemmm mmm mc ee ————— 222 - - - - - - 51
150-159 == mmm mmm 116 - - = - = - -
160= 169 === mm mmm mmm mm mmm mm mmm 73 . - = - - . -
170-179=====mmmmmmcc ecm 34 - - - - - - -
180+-==mmmmm mmm mmm 14 - - - - - - -
Diastolic blood pressure (mm. hg.)=--Con.
Systolic blood pressure (mm. hg.)
80-84 85-89 | 90-94 | 95-99 | 100-104 | 105-109 | 110-114 | 115+
Number of women in thousands-—Con.
Totalm====m=====- ymmmmm———— 1,399 670 241 163 106 18 17 37
Under 90-==mmmmmmmm mmm - - - - - - - -
G0=gF= mmm mmm mm mm mmm mm mmm mmm mm - - - - = - - -
100-109 ==mmmnenmmmmm————————— 90 - - - - - . -
110-119=-m mmm ————————————— 447 88 7 14 - - - -
L200 L2G wm se re rm 582 | 300 70 12 = - - -
130-139 mmm emmee mmm men ————— 263| 215| 108 20 - 9 17 -
140-149 mmmmmmmmmc emcee ———— - 66 31 73 - - - -
150-159===mmmemmmmamnana——————— 16 - 14 43 17 9 - 16
160-169 =m mmmmmmm mmm ————————— - - 11 - 62 - - -
LT 0m 1.7m mm om mm - » - - 27 - - 7
TL 0 rs ese sm - - - - - - - 14
29
Table 13. Number of women aged 35-44 years, by specified systolic and diastolic blood pressures: United
States, 1960-62
Diastolic blood pressure (mm. hg.)
Systolic blood pressure (mm. hg.)
Total Water 50-54 | 55-59 | 60-64 | 65-69 | 70-74 | 75-79 | 80-84
Number of women in thousands
Totale====-memmmm mmm cme m em 12,325 44 3s 233 652] 1,666 | 2,199 | 2,509] 2,061
Under 90-==rmmmeemmcer— cc ———————————— 16 - 8 - - = 8 v >
90-99 mmm mmm mmm mmm mmm meme mmmmem men 391 - 12 37 118 97 89 30 8
THY TYE) nmin so 6 i 2,134 28 13 137 270 609 593 387 97
110-119==== =m mmm mm mmm mmm mmm 3,571 - - 59 210 706 973] 1,031 452
120-129-===-mmmmmm meme mmm mmm m meee 2,79 - - - 55 206 351 747 822
130-139===mmmmcmmmme mmm m mem ———— 1,761 - - - - 48 164 269 513
140-149=====mmmmmcmmmm mmm mm mmm em 747 - - - - - 21 12 103
150-159 = nme nn mmm nn ——————————————— 436 - - - - - - 34 66
160-169 =m mmm mm mm mm mm mm mm mm mmm 245 - - 2 # = = 5 5
170-179 == =mmemmam mm ——n———————— 87 16 - = $ = = s =
180-189===mmmmmmmmmmm mmm meme mmm mm ————— 102 - - - - - - - -
190 1.9 Gm em mmm mm 2 21 - - - - - - - -
200m mmm mmm mm ——————————— 20 - - = g 2 # = =
Diastolic blood pressure (mm. hg.)-—Con.
Systolic blood pressure (mm. hg.)
85-89 | 90-94 | 95-99 | 100-104 | 105-109 [110-114 [115-119 | 159°
Number of women in thousands--Con.
Total====mmmceem ccc cece cme —— 1,271 727 386 280 124 59 64 16
Under G0 mmm mmm m mmm mow ow om on ar mn om se mm mm om ae = - = - - #
G0=9Ymmmmmmmmmm mmm —————————————— - - - - - = - -
100-109=mnmnmnmmeemmane———————————————— - - - - - - - -
L10= 119 = mmm mmm mm mm mmm mm i i 117 27 = - = = - -
120-129=m=menmecmmemnncann mem ———————— 434 180 - = - - - -
130 L3G mm mm mmm mm me 472 174 76 24 8 15 - -
140-149- === mmmmmmmmmc mmm —————————————— 175| 233] 133 70 - - - -
150-159====mmmmmmcmm—————————————————— 56 36 104 98 42 - - -
L160 L6G mmm mmm mm tm mm em 10 47 66 39 53 - 14 16
170-179 cn nemmmnn mmm mm —————— 8 13 - 14 - 7 29 -
180-189-=-mmmmmmm mmm ——————— - 22 7 36 - 38 - -
LTO GD im im im om ee i ie = ~ . = 21 - - -
ZO0Hm mm mmm em mm mm mt mm mm - - = - - - 20 -
30
Table 14.
Number of women aged 45-54 years, by specified
States, 1960-62
systolic and diastolic blood pressures: United
Systolic blood
Diastolic blood pressure (mm. hg.)
pressure (mm. hg.) Under
Total 50 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89
Number of women in thousands
Total===-===-=-= 10,542 28 8 75 260 956 1,378 1,975 2,157 1,360
18 - - - - 18 - - - -
118 - 8 25 - 69 7 - 9 -
1,046 - - 36 110 455 209 217 20 -
1,895 - - 14 98 226 555 501 373 128
2,078 - - - 45 122 372 574 606 284
2,143 - - - - 67 211 445 733 424
1,264 - - - - - - 131 293 297
625 - - - - - 14 36 14 56
466 - - - - - 11 23 66 76
353 28 - - - - - 28 - 45
260 - - - 7 - - - 43 36
49 - - - - - - - - -
96 - - - - - - - - 14
73 - - - - - - - - -
10 - - c - - - - - -
12 - - - - - - - - -
13 - - - = - - - - -
24 - - - - - - - - -
Diastolic blood pressure (mm. hg.) —Con.
Systolic blood
pressure (mm. hg.)
90-94 95-99 100-104 | 105-109 | 110-114 | 115-119 | 120-124 | 125-129 | 130-134 | 135+
Number of women in thousands—Con.
Total=======- 925 654 289 173 87 77 80 37 - 24
U#der 90----======-= = - - - - - - - - %
90-99 --ccmmmmmeemam - - - - - - - - - -
100-109-==--==-==== - - - - - - - - - -
1L0~119 = mmm me - - - # - = = - - -
120-129--=-==mmmumn 66 8 - - - - - - - -
130-139-===cemcumu- 193 47 24 - - - - - - -
140-149 -===mcmmcanmn 301 167 34 20 - - - - - -
150-159-=--==--=-==-- 233 186 56 18 12 - - - - -
160-169---=======-- 13 84 67 47 20 - - - - -
170-179-===-==-omm= 43 98 46 36 11 8 10 - - -
180-189--=--=-=--==-- 15 48 56 32 7 - 15 - - -
190-199 --===mmmmnun - - 7 - 17 13 13 - - -
DOD = P20 im mw ime - 16 - 7 7 11 42 - - -
210-219---==-=--=-= - - - 13 13 33 - 14 - -
220-229---=cmmmmman - - - - - - - 10 - -
230-239 --=cccmmnann - - - - - - - 12 - -
240-249 -==ccmmeennn - - - - - - - - - -
250-259 -=====mmmun- - - - - - 13 - - - -
260+-====mmmmmm mmm - - - - - - - - - 24
31
Table 15.
Number of women aged 55-64 years,
by specified
States, 1960-62
systolic and diastolic blood pressures: United
Systolic blood
pressure (mm. hg.)
Diastolic blood p
ressure (mm. hg.)
Total | UBSeT | 50-54 | 55-59 | 60-64 | 65-69 | 70-74 | 75-79 | 80-84 | 85-89
Number of women in thousands
Total swesnnsns 8,121 15 w 70 231 394 950 | 1,431| 1,403] 1,154
90-99mmmmmm mmm mm mmmm 42 “ ® - 28 14 - - = *
00-10 = snesenmemanin 250 - - e 34 100 86 23 8 =
110-11 emmmm meme 714 s = - 87 99 216 179 105 10
L040 enn nnn mesma 1,242 2 - 45 63 44 285 419 233! 133
USO Fim mmm mre 1,370 - . . - 80 196 329 385 | 278
WAL wma 1,587 - . . - 21 47 423 337 | 368
LOLS wm asian 794 15 - 10 - 21 33 a 112] 199
LEO BD emma mame 757 - - 15 19 v 60 23 106 92
170-179 === m mmm mm mmmm 454 - = ’ % 17 . . 70 30
180-189 = mm mmm mmmmm me 306 g . = ” - . 21 15 42
190-100 sre anmanns 314 - . ” - - 28 15 32 .
BO <2 mame 113 . . = 7 - - ® . *
OIF smmmme mn mnnn 56 - - . - - “ ” ” -
200-220 mnnennnannwnn 23 - - 7 " = - . - :
230 23T mmm mmmemmmmnn 50 % ® ® . ” " - - -
Pw irre isasitsi 11 - - - w - - - g ”
WHO wD wim ar sien - - - - - - s v . 3
Tibor nn 36 - - - - " ’ . - -
Diastolic blood pressure (mm. hg.) --Con.
Systolic blood
pressure (mm. hg.) |g4 o. | 95.99 | 100-104 [105-109 [110-114 [115-119 [120-124 [125-129 [130-134 135+
Number of women in thousands--Con.
Ot iimemnnvinn 987 | 578 276 273 69 113 19 68 18 72
90-99 mm mm mmm mmm mmm 2 2 = - . “ - - - g
VOD Pw scimmiesion sesimnisn - . . " . - - - - .
LOD 19 = - - - - - v F -
120-20 nenmnn manu 20 - - - - - . x ” “
130-139 ==mmmmmmm mmm 63 10 28 " . . . . - -
140-149 = mm mmm mm mmmmm 238 | 135 18 - . - % - - 2
150-150 smi aman 131] 175 | 45 54 - - - - : =
LOO wen nnn 203 | 121 54 45 - 19 : # . »
LID LT Gmmmmmr mmm mre 147 62 70 43 . 15 . . " -
180 18Tw mmm mmm 109 18 C19 25 38 19 . . - -
LOD 1G mmm rims min 56 56 . 64 31 20 11 - - -
POY Bis esses w # 42 » . 28 - 25 18 ”
FULLY Disa soi senso - . - 25 - 12 - 18 ® .
BEGDIm mmsecmin - - - - - 3 . 25 - .
230-239 mmm mmm mmm mmm - - ” 17 ” - 8 . - 25
240 = 24) wars rn - ® - - - - 2 - 11
FIO EEG mmm * - ” - - = ” . - v
Yi mn mms sme - = < - - - . . “ 36
32
Table 16. Number of
women
aged 65-74 years, by specified systolic and diastolic blood pressures:
United
States,
1960-62
Systolic blood
Diastolic blood pressure (mm. hg.)
pressure (mm. hg.)
Total 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89
Number of women in thousands
Total======== 6,192 28 81 219 500 597 1,078 949 930
Under 90============ 14 14 - - - - - - -
90-99-====mmmmmm———— 42 - 30 11 - - - - -
100-109=====mmmmmu=- 52 - 40 - 11 - - - -
110-119=====mmmm mm 252 - - 25 39 122 66 - -
120-129====mmmmmmn-- 506 - - 62 47 136 87 72 101
130-139-==mmmmmm 755 - - 41 88 61 250 201 73
140-149 ==mmmmmm mm ——— 767 - 10 39 78 83 176 235 73
150-159==mmmmmmm———-— 897 14 - 24 46 63 123 233 140
160-169=====m=mmmm-- 733 - - - 110 - 190 108 100
170-179==mmmmmmmm ma 566 - - - 62 - 22 - 106
180-189==m==mmmmm = 518 - - 17 - 42 35 99 57
190-199-===mmmmmmmu- 533 - - - - 78 65 - 208
200-209-==mmmmm————— 267 - - - - 12 26 - 46
210-219==m=mmmmm———— 196 - - - 18 - 21 - 26
220-229 ==mmmmmm———— 50 - - - - - - - -
230-239 mmm mmm ————— 45 - - - - - 16 - -
Diastolic blood pressure (mm. hg.)--Con.
Systolic blood
pressure (mm. hg.)
90-94 95-99 | 100-104 | 105-109 | 110-114 | 115-119 | 120-124 | 125-129 | 130-134
Number of women in thousands--Con.
Total-======= 639 349 387 214 59 94 32 23 14
Under 90=========== - - - - - - - - -
90-99 =mmmmmm————— - - - - - - - - -
100-109=======m===-= - - - - - - - - -
110-119==m=mmmmm——— - - - - - - - - -
120-129=== mmm mm——- - - - - - - - - >
130-139====mmmmmm—- 16 24 - - - - - - -
140-149 ==mmmmmm———— 43 - 29 - - - - - -
150-159===-mm=m=mum 119 42 75 18 - - - - -
160-169=====m==m==-= 145 37 25 18 - - - - -
170=179=====mmmm=u= 167 87 76 24 - 22 - = -
180-189====mmmmmm—- 57 73 76 36 - 26 - - -
190-199====mmmmmmn= 16 59 38 38 - - 32 - -
200-209============ 45 27 24 22 43 - - 23 -
210-219===m=mmm———— 16 - 44 10 - 46 - - 14
220-229==mmmmmm———— 16 - - 18 16 - - - -
230-239= mm mmmm———— - - - 29 - - - - -
33
Table 17. Number of women
aged 75-79 years, by specified systolic and diastolic blood pressures:
United States,
196
0-62
Diastolic blood pressure (mm. hg.)
Systolic blood pressure
gms. Be. Total || UBder! so.s4| 55-59) 60-64 65-69| 70-74| 75-79 80-84] 85-89
50
Number of women in thousands
Total=======cceeeann 1,443 14 = 32 86 200 237 191 168 206
100-109=m==mmmmmmm—————— 17 - - - - - - 17 - -
110-119====mmmmm mmm mmm em 49 - - - 25 - 25 - - -
L20« 120mm mmm mmm mm mmm mmm 136 - - - 62 22 18 20 - 14
130-139==m=mmcmm mmm meee 164 - - 14 - 71 25 - - 32
140-149 =m mmm mmmm mmm mm 229 - - - - 30 32 51 94 -
150-159=m=m=mmmmmmmm——————- 213 - - x7 - 30 80 26 18 14
160-169=mmmmmmmmmm———————— 165 14 - - - 26 18 28 10 22
170=179===mmmmmc meme cemme 219 - - - - 21 41 48 - 57
180-189=mmnmmmmmm———————— 136 - - - - - - - 47 26
190-199=m=mmmmmmm mmm mmm mm 41 - - - - - - - - 10
200-209 == =mmmmm—————— 42 - - - - - - - - -
210-219=mmmmmmm—————— - - - - - - - - - -
220-229==m=mmmmmm meme me mem 32 - - - - - - - - 32
Diastolic blood pressure (mm. hg.)=-Con.
Systolic blood pressure
(mm. hg.
90-94 95-99 100-104 105-109 110+
Number of women in thousands--Con.
Total-====c=mceeena- 120 91 38 40 19
100-109=====mmmcme mmm cme - - - - -
110-119=mmmmm mmm meme meme mm - - - - -
120-129=mmmmmmmmm————————— - - - - -
130-139-mmmmmmemm———————- 22 - - - -
140-149=mmmmmmm mmm emcee a 21 - - - -
150-159====mnmmmmmnnmmnnn - 28 - - -
160-169======mmmmmmm mean 22 25 - - -
170-179==mmmmmmmmm meee mem 10 18 14 10 -
180-189=mm=mmmmm ccm meme mm 43 21 - - -
190-199==m=mmcmmcm meee ee a - - - 30 -
200-209======mmmmm— mm meee - - 24 - 19
210-219 mmmmem—————— mae - - - - -
220-229==mmmmmmm mmm ————— - - - - -
34
APPENDIX |
BLOOD PRESSURE MEASUREMENT
The techniques for measuring blood pressure used
by the Health Examination Survey follow procedures
suggested in the Report of the Conference on Longi-
tudinal Cardiovascular Studies, National Heart Insti-
tute, Bethesda, Maryland, 1957 (the "Beaconsfield
Report') which essentially follows the lines of the
American Heart Association recommendations. This
does not constitute a definitive specification, however,
since in a number of particulars alternative suggestions
are offered, and there isnoreally satisfactory basis for
choosing between them. For example, in the recording
of diastolic pressure some investigators prefer to use
the fourth phase and some the fifth. It would have been
possible for the Health Examination Survey to have
reported both diastolic pressures, since both were
tabulated, but this would simply have complicated the
presentation without any apparent gain in utility.
The sphygmomanometers used in this Survey were
standard instruments (Baumanometer). They are very
durable and relatively trouble-free. There is some
reason to believe, however, that occasionally these
instruments—usually through unnoticed spillage of
mercury—were slightly out of calibration, and it would
have been desirable to have checked the instruments
more frequently than was done.
In this report the average of the three readings was
tabulated. Although the report of the Beaconsfield Con-
ference permits this procedure, many persons with
clinical training think it an unwarranted innovation.
The fact is, of course, that the blood pressure for any
individual is a constantly altering value, with periods
when it is low and occasions when it is unusually high.
Presumably if only one figure is to be used to charac-
terize the blood pressure of an individual it should ideal-
ly integrate his total experience. If this cannot be ob-
tained, an average of several readings probably serves
better than a single casual pressure, however standard-
ized the circumstances of measurement for that single
measure seem to be. Certainly for describing population
groups it seems thatan average of several blood pressure
measurements is the preferable statistic, among the
various possible alternatives, despite the obvious re-
luctance to use it.
There is no standard environment for taking a blood
pressure measurement. The usual procedure is to try
to have the examinee calm and rested before measure-
ment but the specific program for arriving at this state
is highly variable. Essentially, the procedure used in
the Health Examination Survey might be considered as
approximating the usual situation in clinical practice,
with the blood pressure being measured, without special
preparation, during the course of a physical examination.
Other investigators, arguing that the home is a person's
usual environment, prefer taking the blood pressure
measurement there. Whatever the possible advantages
to this technique, the difficulty of conducting an extended
medical examination in a standardized fashion at home
ruled this out for the Health Examination Survey.
The efforts that have been made to standardize
blood pressure measurement, while highly useful, must
ultimately be limited by the great lability of this measure.
For survey purposes there is little real difference be-
tween the various acceptable alternative procedures.
However, if the results of one survey are to be com-
pared with those from another, it would be desirable to
make the circumstances and techniques of measurement
of both as similar as possible. In any case, there is a
remarkable resemblance among the blood pressure
findings of various surveys, despite recognized and un-
recognized differences in procedure.
O00
35
APPENDIX II
SURVEY DESIGN, MISSING DATA, AND VARIANCE
\
The Survey Design
The Health Examination Survey is designed as a
highly stratified multistage sampling of the civilian, non-
institutional population (aged 18-79 years) of the con-
terminous United States. The first stage of the plan is
a sample of 42 primary sampling units (PSU's) from the
1,900 geographic units into which the United States has
been divided. A PSU is a county, twoor three contiguous
counties, or a standard metropolitan statistical area.
Later stages result in the random selection of clusters
of about four persons from a small neighborhood within
the PSU. The total sample included 7,710 persons in the
42 PSU's in 29 States. The detailed structure of the de-
sign and the conduct of the Survey have been described
in previous reports. >
Reliability in Probability Surveys
The Survey draws strength from the fact that it is
a probability sample of its total target population, and
from the fact that the measurement processes which
were employed were highly standardized and closely
controlled. This does not mean, of course, that the
correspondence between the real world and survey
results is exact. Data from the Survey are imperfect
for three important reasons: (1) results are subject to
sampling error, (2) the actual conduct of a survey never
agrees perfectly with the design, and (3) the measure-
ment process itself is inexact, even when standardized
and controlled. The faithfulness with which the study
design was carried out has been analyzed in a previous
report. 2
Of the 7,710 sample persons, the 6,672 who were
examined—a response rate of over 86 percent—give
evidence that they are a highly representative sample
of the adult civilian, noninstitutional population of the
United States. Imputation for the nonrespondents was
accomplished by attributing to nonexamined persons the
characteristics of comparable examined persons. The
specific procedure used ° consisted of inflating the
sample weight for each examined person to compensate
36
for sample persons at that stand and of the same age-
sex group who were not examined.
While it is impossible to be certain that the blood
pressures are the same in the examined and the non-
examined groups, the available evidence indicates that
it is. One source of information on this question is u
special inquiry sent to the physicians of nonexamincd
persons and to the physicians of a matching set of ex-
amined persons. The mean blood pressures reported
for the examined and nonexamined groups were inexact
agreement. Further details on this subject appear in a
previous report. ”
Sampling and Measurement Error
In this reportand its appendices, several references
have been made to efforts to evaluate both bias and
variability of the measurement techniques. The proba-
bility design of the Survey makes possible the calculation
of sampling errors. Traditionally the role of the
sampling error has been the determination of how im-
precise the survey results may be because they come
from a sample rather than from measurement of all
elements in the universe.
The task of presenting sampling errors for a study
of the type of the Health Examination Survey is difficult
for at least three reasons. (1) Measurement error and
"pure' sampling error are confounded in the data; it is
not easy to find a procedure which will either completely
include both or treat one or the other separately. (2)
The survey design and estimation procedure are complex
and accordingly require computationally involved tech-
niques for calculation of variances. (3) Thousands of
statistics come from the survey, many for subclasses
of the population for which there are small numbers of
sample cases. Estimates of sampling error are obtained
from the sample data and are themselves subject to
sampling error, which may be large when the number
of cases in a cell is small, or even occasionally when
the number of cases is substantial.
As variances are estimated for larger numbers of
statistics from the Health Examination Survey, it is
hoped that an increasing amount of information can be
presented in published reports. In this report, estimates
of approximate sampling variability for selected sta-
tistics are presented in tables I and II. These estimates
have been prepared by a replication technique which
yields overall variability through observation of vari-
ability among random subsamples of the total sample.
The method reflects both pure’ sampling variance and
a part of measurement variance.
Table I. Relative standard error of the mean blood
pressure of adults, by age and sex: United States,
1960-62
Systolic Diastolic
Age
Men Women Men Women
Relative standard error
in percent
Total-18-79 years- 0.3 0.5 0.7 0.6
18+24 yearg-====nmmmn= 1.0 1.0 1.0 1.0
25-34 years----===m==-- 1.0 1.0 1.0 1.0
35-44 years---=mmmm——- 1.0 1.0 1.0 1.0
45-54 years-=mmmmmm——- 1.0 1.0 1.0 1.0
55-64 years----===m=== 1.0 1.5 1.0 1.0
65-74 years-=-mmmm=m-- 1.5 1.5 1.5 2.0
75-79 years-===mmmmm=- 2.0 1.5 1:5 2.0
In accordance with usual practice a 68 percentcon-
fidence interval may be considered that range within
one standard error of the tabulated statistic and a 95
percent confidence interval that range within two stand-
ard errors. An overestimate of the standard error of a
difference d= x — y of two statistics x and y is given by
the formula s,= [x2vi + yi] % , where V% and Vy
are relvariances respectively of x andy, or the squares
of the relative errors shown in table I. For example,
table E shows systolic x= 132.1 for men and y=129.9
for women, while from table I relvariances are found
to be V?,=0.00001 and V?,=0.00002. The formula
yields the estimate of standard error of the difference
(d = 3.0) as sq=0.71. Thus, as the observed difference
is more than four times its sampling error, it can be
concluded with near certainty that the evidence from
this Survey is that systolic blood pressure is higher
among men than among women.
Small Numbers
In some tables magnitudes are shown for cells for
which sample size is so small that the sampling error
may be several times as great as the statistic itself.
Obviously in such instances the statistic has nomeaning
in itself except to indicate that the true quantity is small.
Such numbers, if shown, have been included to convey an
impression of the overall story of the table.
Table II. Relative standard error of number of adults with blood pressure of at least 160 systolic
or 95 diastolic, per 100 persons, by sex and age: United States, 1960-62
Systolic at |Diastolic at Systolic at least
Sex and age least least 160 mm. hg. or dia-
160 mm. hg. 95 mm. hg. stolic 95 mm. hg.
Relative standard error in percent
Both sexes-18-79 years------==--=cececcea-- 7 8 4
Men
Total-18-79 years-=--======-mcmmmmcece cc ccn—an 10 9 5
18-24 years--===---mmmremmeeee eee ——— mm ——————— * * *
25-34 years-=--=mm--mmmmmemememeeeee seme ee m————————— * 20 25
35-44 JEATS = mmm mmm mr mmm mm mm 20 15 10
45-54 YeaArS=mmmmmm mmm mmm mm mm 20 15 10
55-64 yearS----mmmemmeemeeme seem emma ——————— 15 15 10
a ET 10 20 10
75279 FOALS = mem comm mm maim mn ms om on mom om om om om om om om me 20 * 20
Women
Total-18-79 yearS-====mm-mmeeecceeeccccccccnaa— 6 8 5
18-24 years==wemmrrmmrmcm— ee —e—e———————————————— * * *
25-34 yearSemmmmmmmmmmmm mmm ———————————— * 25 25
35-44 years==-==mmm-mmmmmeecmemeeeccemmmemameeeen 25 15 15
45-54 yearS--=-----msecmmmme meme mme meme mmm momo 15 15 15
55«64 YearS==mmurmemmmommme seem ——————— 10 15 10
65-74 yearS===-m-mmmmmmmmmmemeeem eee eee————————— 10 15 10
75-79 years--==-==mm--mmmmmmmem mmm eemme———————— 15 * 15
37
APPENDIX II
OBSERVER VARIABILITY IN BLOOD PRESSURE MEASUREMENT
Blood pressure measurement is subject to con-
siderable variation from observer to observer. Part of
this may be considered technical, arising from differ-
ences in the method of inflating and deflating the cuff,
uncertainties in recognizing the Korotkoff sounds, the
problems of reading scales, and so forth. Partmay arise
from the reaction to the observer of the person being
measured, and this may vary according to the age, sex,
race, or income of the person being measured.
The Health Examination Survey can be considered
to have obtained an unbiased clinical measure of blood
pressure. It has been shown for a subsample of the ex-
aminees that the average of blood pressure measure-
ments reported by their private physicians was identical
with the average measurement obtained by the Health
Examination Survey physicians. In individual cases,
however, there frequently were large divergences be-
tween measurements from these two sources, but then
it is equally true that there were large divergences
among the three blood pressure measurements taken
by a single physician from an individual during his ex-
amination. The essentially unbiased clinical measure
obtained by the Survey as a whole may be considered to
reflect the averaging effect arising from the use of a
large number of physicians.
For individual physicians, there is ample internal
evidence from the Survey of significant differences in
levels of measurement. Table III shows the extent that
the average blood pressure measurement for each phy-
sician differs from the average for all physicians.
Since the persons examined by a given physician may
differ considerably from the general population in their
distribution among the various age-sex groups, the
mean of the blood pressure measurements for a given
physician is compared with an expected value obtained
by weighting the age-sex specific blood pressures for
the total sample by the percentages in the various
age-sex groups examined by that physician, Specifically,
for a given examiner,
Let N; be the number of persons inthe ith age-sex
group examined by the examiner (sum of N;=N)
38
Let X,; be the mean blood pressure obtained by that
examiner for age-sex group 1.
Let X; be the mean blood pressure for the sth age-
sex group as measured by all examiners. Then
= d XX
= NaN (X%-X)
D is a summary measure of the deviation of this
physician from the average physician and is the sta-
tistic tabulated.
There were 42 stands at which examinations were
conducted. At most stands there were two physicians who
took examinees alternately. On the average there were
about 160 examinees at a stand, with roughly half being
examined by each physician. While the persons at a
specific stand may have blood pressures which deviate
from the average for the United States, two physicians at
the same stand should have examined a random sample
of the population at that stand, and these physician
samples should not be expected to differ from each other
more than chance. Hence, it is appropriate to compare
the deviation at the stand for each of the two physicians
there.
Table III presents statistics for systolic and dia-
stolic deviations, specific for stand and for physician.
The presentation is slightly simplified. Data for the phy-
sicians atone stand where the assignment of examinees
was clearly not random are omitted. This eliminated
160 examinees. All cases where the physician examined
fewer than 32 persons at a stand were also omitted.
This accounts for an additional 135 omissions. Four
quasi-stands were constructed to replace four of the 42
actual stands. This was done in such a way as to retain
a random assignment of the stand populations together
with a pairing of physicians.
While a full analysis of these data is not undertaken
in this report, the tabular material suggests the scale
of the physician's impact on results. Physician dif-
ferences are, of course, linked with place differences.
A formal separation of the components of variation
would be an involved matter. It would have to take into
account the complex sample design and estimation pro-
Table III.
physician: Health Examination Survey, 1960-62
Deviation of actual from expected mean blood pressure, by stand number and
Stand number
Systolic blood pressure
Diastolic blood pressure
Physician A
Physician B
Physician A
Physician B
Mean deviation from
~3.
ew
+0
.02
5.
.03
27
.03
34
71
.02
.53
.81
.61
.95
.25
.09
.26
2
PY
a
2. 5
67
98
38
53
197
+92
.59
13
.26
2.16
24
.27
.97
«99
0.70
wll
.38
«29
.60
.66
.92
.93
.10
.28
+11
.61
.95
2.38
.56
«78
.12
.15
.08
.24
.88
+23
02
.64
22
+729
.04
.82
«31
.02
«62
.96
.70
.56
«20
.96
+23
.60
«56
.61
«73
+91
wal
42
.18
.85
.36
.31
«17
«39
.69
«91
.00
45
expected
2
-2
2g
9 3
1-8
a0
-5
6
35.
1.
1.0
1.2.
WB
-2
1
5.
23.
hi
8
-0
15
16
en
8.
6.
91.
21a.
3.82
(mm.
+L
.00
.67
.34
v27
.34
.73
«23
66
21
34
40
.93
.60
25
67
.16
.69
25
68
35
66
53
.63
32
.61
.08
.06
«27
«35
.05
03
14
.01
73
we
we
"33
46
.35
hg.)
Ipseudo stands.
99, on
2-22%Gme physician.
cated by the superscript 2.
For example, one physician conducted examinations at stands 23, 4, and 6 and his deviation are indi-
39
cedure used in the Health Examination Survey. It would
have to allow for deep primary stratification, ratio
estimation, poststratification, multistage selection, and
other departures from simple random sampling. Pre-
liminary investigations indicate that such an analysis
would show that between-physician variation is by no
means trivial and indeed contributes a substantial pro-
portion of overall total survey variance. Since this com-
ponent decreases directly with an increasing number of
physicians taking measurements, it is much smaller in
the Health Examination Survey thanitwouldbe in a sur-
vey with only a few persons taking the blood pressure
measurements.
A more serious concern than increased variability
from interphysician differences is the possibility that
interphysician differences complicate the analysis of
blood pressure data from the Health Examination Survey.
Granting a slight attenuation introduced into compari-
sons between subgroups of the population, what is the
risk of bias being introduced into such comparisons?
The answer to this question must be that such a risk is
practically nonexistent. For studies in which only a few
observers measure the blood pressure such risks are
obviously present, particularly if examinees are not
assigned randomly to observers. In the HES, however,
the large number of examiners and the relatively small
number of persons examined by any one physician re-
duce this risk to an indiscernible level.
Finally, while there is a general interest in the
magnitude of physician differences, and while observer
variance is a significant part of total variability, total
variability is small for most estimates in this report.
For most categories the relative standard error is
only a fraction of what a physician would accept as a
substantive tolerance.
000
40
U.S. GOVERNMENT PRINTING OFFICE : 1964 O - 732-721
VITALand HEALTH STATISTICS
DATA FROM THE NATIONAL HEALTH SURVEY
NATIONA
STATISTICS ————
Blood Pressure
of Adults
by Race and Area
United States-1960-1962
iL. .
CENTER Series 11
For HEALTH Number 5
ZEEE 2
U. S. DEPARTMENT OF [o/s \*
w 4 iy -
HEALTH, EDUCATION, AND WELFARE A \ J
Public Health Service \%
See inside of back cover
for catalog card.
Public Health Service Publication No. 1000—Series 11 —No. 5
For sale by the Superintendent of Documents, U.S. Government Printing Office
Washington, D.C., 20402 - Price 25 cents
NATIONAL CENTER| Series 11
For HEALTH STATISTICS | Number §
VITALand HEALTH STATISTICS
DATA FROM THE NATIONAL HEALTH SURVEY
Blood Pressure
of Adults
by Race and Area
United States-1960-1962
N.ean blood pressure by race and area.
Washington, D.C. July 1964
U.S. DEPARTMENT OF
HEALTH, EDUCATION, AND WELFARE Public Health Service
Anthony J. Celebrezze Luther L. Terry
Secretary Surgeon General
NATIONAL CENTER FOR HEALTH STATISTICS
Forrest E. Linder, Ph.D., Director
Theodore D. Woolsey, Deputy Director
Oswald K. Sagen, Ph.D., Assistant Director
(for Professional Relations)
Walt R. Simmons, M.A., Statistical Advisor
Alice M, Waterhouse, M.D., Medical Advisor
James E. Kelly, D.D.S., Dental Advisor
Louis R. Stolcis, M.A., Executive Officer
DIVISION OF HEALTH EXAMINATION STATISTICS
Arthur J. McDowell, Chief
Public Health Service Publication No. 1000-Series 11-No. 5
CONTENTS
Page
Introduction = === ec ce memo 1
ATCA mmm mm meme mmm e em 1
RACE ===cmmm meee eee ere eee 3
Race and Region------cccmmcmmmm cme ee emcee 3
Discussion =======c cmon eee 5
SUMMAry ===-ccemo mmm emma eee 7
Detailed Tables-====ccm mcm cm mmc cee eee 9
Appendix I. Statistical NoteS====cemcmmommc cmc 15
The Survey Design-----coceocmmmmm omic meee 15
Reliability in Probability Surveys------eccecmcamcaaaaax 15
Sampling and Measurement Error--------cccocceceea-- 1S
Small Numbers--=--cece mom cece em 16
Tests of Significance------c-coecmmmommmmmc eee 16
Expected Values------ccmcmccmm meme cmcccccccmeee eee 17
Stand Variation------ccmccccm meee 18
Appendix II. Demographic Terms------=-ccccccmmmcacanoo 19
SYMBOLS
Data not available-=+===mccmcamcccceeeaeo
Category not applicable-==c-emmmmmcmaaaao .
Quantity zero---===-cecmmmm meee
Quantity more than 0 but less than 0,05----
Figure does not meet standards of
reliability or precision-===-=mceeeceana-
BLOOD PRESSURE OF ADULTS
BY RACE AND AREA
Tavia Gordon, Division of Health Examination Statistics
INTRODUCTION
The National Health Survey uses three meth-
ods for obtaining information about the health of
the U.S. population. The first is a household
interview in which persons are asked to give
information relating to their health or to the
health of other household members. The second
is the collection of data from available health
records. The third is direct examination. The
Health Examination Survey (HES) was organized
to use the third procedure, drawing samples of the
population of the United States and, by medical ex-
amination and with various tests and measure-
ments, undertaking to characterize the population
under study.
The initial enterprise of the Health Exami-
nation Survey was the examination of a sample
of adults. Its purpose was to obtain information
on the prevalence of certain chronic diseases,
on dental health, and on the distribution of a
number of anthropometric and sensory charac-
teristics. A nationwide probability sample of
7,710 persons aged 18-79 years was drawn.
Altogether, 6,672 persons were examined during
the course of the Survey which was begun in
October 1959 and completed in December 1962.
Sample persons received a standard examination,
lasting about 2 hours, performed by medical and
other staff members of the Survey in specially
designed mobile clinics.
This is one of a series of reports describing
and evaluating the plan, conduct, and findings of
the first cycle of the Health Examination Survey.
The general plan has already been described; ! a
description of the sample population and response
has been published. ?
This report continues the presentation of
data on blood pressure. The first report on this
subject described the context in which these data
were collected and the techniques used in meas-
urement.3 '""Casual'' blood pressure readings were
taken. It was shown that these were comparable
in level with readings obtained in ordinary clinical
practice. Possible artifacts arising from the
conduct of the Survey or the examination pro-
cedures were examined and it was concluded
that none of these constituted complicating factors.
Estimates for the United States were given of the
distribution of blood pressure and of mean levels
by age and sex. In the present report the relation-
ship of blood pressure to area and race is con-
sidered. Comparison of racial differences is
limited to findings for white and Negro persons
since the sample was too small to permit adequate
representation of other nonwhite races.
AREA
The sample design of the Health Examination
Survey resulted in the selection of 42 primary
sampling units or ''stands.' The selection pro-
cedure was stratified by region and size of place.
As a consequence, relatively efficient estimates
by region and by size of place can be made, al-
though the sample size will necessarily result in
estimates with relatively sizable variances. This
section presents estimates of mean blood pressure
by region and by size of place. In addition, in-
dividuals were assigned to demographic areas
within the stand, with these areas designated as
urban or rural, and mean blood pressures are
presented for these more specific areas.
In making comparisons between areas, al-
lowance must be made for the fact that there are
differences from place to place inthe distribution
of the population by age and sex and that mean
blood pressure varies by age and sex. Because
the sampling variability of age-sex-specific val-
ues for individual areas is verylarge, a summary
comparison was thought preferable to the presen-
tation of mean values specific by age and sex. For
this reason, the actual mean systolic blood pres-
sure for an area is compared with an expected
mean value. The expected value is obtained by
weighting age-sex-specific means for the total
United States by the age-sex distribution for the
area. Presented in tables A, B, and C are the
actual and expected mean systolic pressures and
the differences between them. The obvious meaning
can be attached to these differences, with the
understanding that small differences may arise
by chance. A positive difference, for example,
indicates that the mean blood pressure for the
area is higher than expected. In general, where
there is no statistically significant difference
between the actual and expected values for an
area, differences for individual age-sex groups
exhibit only random fluctuations.
Table A. Actual and expected mean systolic
blood pressures of adults, by region:
United States, 1960-62
Ex- Differ-
Region Actual pected RCE
Mean blood pressure in
mm. hg.
Northeast--=---- 132.6 § 131.0 1.6
South====cceeaaa 129.7 130.1 -0.4
West=ewemcanaaa-" 130.2 131.5 -1.3
Table B. Actual and expected mean systolic
blood pressures of adults, by population-
size group: United States, 1960-62
Population-size Ex- | Differ-
group Actual pected | ence
Mean blood pressure in
mm, hg.
Giant metropol-
itan areas----- 132.5] 131.4 1.1
Other very large
metropolitan
areas==--==--- 129.1 130.7 -1.6
Other standard
metropolitan
statistical
areas-----=--- 130.2 130.4 -0.2
Other urban
areas-==~-=---- 129.9 130.1 0.2
Rural areas----- 132.1 131.8 0.3
Table C. Actual and expected mean systolic
blood pressures of adults, by Place of
residence: United States, 1960-6
Residence Actual pertad Piffer-
Mean blood pressure in
mm. hg.
Urban----- 130.7 130,7 0.0
SMSA! —in
central city---| 131.1| 131.6 -0.5
SMSAl—outside
central city---| 130.7 | 130.2 0.5
Not in SMSA!----| 129.7 | 130.1 -0.4
Rural--=-=- 131.9 131.7 0.2
Farme==-=eceece--- 133.2 132.5 0.7
Nonfarme----=e-- 131.4] 131.4 0.0
lstandard metropolitan statistical area.
The systolic pressure was chosen for this
comparison rather than the diastolic for two
reasons: (1) a smaller proportion of its varia-
bility arises from measurement error than is
the case with diastolic pressure; and (2) systolic
pressure has a distinct gradient with age, so
that differences in age distribution are more
clearly reflected in the expected values than
would have been the case if diastolic pressures
were used. A higher expected mean systolic pres-
sure indicates an older population group.
The findings in tables A, B, and C can be
briefly summarized. There are only slight dif-
ferences between regions with respect to mean
systolic pressure. These differences are sta-
tistically significant. Mean pressures are essen-
tially the same for persons living in giantmetro-
politan areas, other large metropolitan areas, or
any other areas specified by population size.
Blood pressures are the same for residents of
standard metropolitan statistical areas in the
central city as for residents outside the central
city, are the same in urban as inrural areas, and
are the same in rural-farm areas as in rural-
nonfarm areas. It cannot be said that no area
differences exist, but those that do exist are
probably small.
RACE
Blood pressure was higher for Negro adults
than for white adults—5.6 mm. hg. higher for
Table D. Mean blood pressure for white
and Negro adults, by sex: United States,
1960-62
Systolic Diastolic
Sex
White | Negro | White | Negro
Mean blood pressure
in mm. hg.
Both sexes-| 130.6 | 136.2 | 78.3 83.3
systolic pressure and 5.0 mm. hg. higher for
diastolic (table D). This relationship appeared
in almost every age group (table 1, figs. 1 and
2). The difference was small in the age group
18-24 years, and for men in this age group the
pressure was higher in the white population than
in the Negro. The racial contrast became greater
in the age groups 25-34 and 35-44 years and
remained at this higher level at older ages. It
was less for men than for women. These dif-
ferences in mean levels were reflected in similar
differences in the number of persons withelevated
blood pressures (tables 3-5).
RACE AND REGION
Slightly more than half of the Negro population
of the United States is concentrated in the South.
Therefore, it may be useful to consider the
131.8
129.4
136.2 | 79.0 | 83.4
136.3 | 77.5| 83.2
180
MEN
— White
snssnasnns Negro
SYSTOLIC
140
100
DIASTOLIC
Mean blood pressure (mm. hg.
.
anes?
a
oe
oo
30 40 50 60 70
AGE
Figure 1. Mean blood pressure for white and Negro men, by age.
180
WOMEN
— White
wenn Negro
SYSTOLIC
140
100
Mean blood pressure (mm. hg.)
DIASTOLIC
sem
susan Hn,
an o,
wert
Ol
LLL]
20 30 40 50 60 70 80
AGE
Figure 2. Mean blood pressure for white and Negro women, by age.
racial contrast separately for the South. Similarly,
the percentage of Negroes is higher in the South
than in the rest of the United States. Hence it is
desirable to consider the regional contrast sepa-
rately for the white population.
The racial difference in mean blood pressure,
although small, is associated with a large dif-
ference in the percentage of persons withelevated
blood pressures. Thus, 18.8 percent of all Negro
adults had systolic pressures in excess of 160
mm. hg. and 22.0 percent had diastolic pressures
in excess of 95 mm. hg., while the corresponding
percentages for white adults were 10.5 and 8.7.
The implication of these facts, which are closely
connected with a consideration of hypertension
and its pathological concomitants, will not be
discussed here. In terms merely of the distribu-
tions given in tables 4 and 5, the effect of adding
to each blood pressure measurement of white’
adults the mean difference between the races
might be considered. This would move the distri-
bution curve of the white population without any
change in shape but with a change of the mean
value to equal that for the Negro population. The
curve for systolic pressures at levels about 160
mm. hg. would then overlay the curve for the
Negro population and what differences there were
could be attributed to sampling variability. Dia-
stolic blood pressures above 95 mm. hg. however,
would present a somewhat different picture, since
there would still remain a clear excess of high
diastolic pressures in the Negro population, es-
pecially at levels above 110 mm. hg. In short,
the differences noted between the races are not
fully described by the differences in mean values.
Unfortunately, the relatively small number of
Negro examinees makes it difficult to proceed
with a detailed analysis of the differences in
blood pressure findings for white and Negro
adults. One observation might be made, however,
respecting the relation of systolic and diastolic
blood pressures. For both races the mean pulse
pressure (the difference between systolic and dia-
stolic pressures) rises with age, with an especially
sharp rise after 55 years of age. In this respect
there was very little difference between the
pressures of white and Negro persons, and while
the pulse pressure was slightly greater for Negro
than for white women, it was practically the same
for Negro and white men, except for what appears
to be sampling variation.
Table E. Mean blood pressure for white
and Negro adults in the southern region,
by sex: United States, 1960-62
Systolic Diastolic
Sex
White | Negro | White | Negro
Mean blood pressure
in mm. hg.
Both sexes-| 127.6 | 138.1 78.2 84.3
Men-========= 129.11] 137.7 719.11 84.6
Women=-====== 126.4 | 138.4 77.5 | 84.0
Table E presents mean blood pressures for
the South by race and sex. (Table 2 gives these
data by age.) Although there are differences in
detail for the United States as a whole, most of
these can be explained by the large sampling
variability of regional statistics. Overall, the
racial differences in blood pressure level seem
about the same in the South as in the remainder
of the country.
In table F, regional comparisons are pre-
sented for the white population only. When the
values in this table are compared with those in
table A, it is obvious that the mean systolic
blood pressure is definitely, if only slightly, lower
in the South than in the rest of the country. The
expected mean values for table F would be prac-
tically the same as those used in table A.
DISCUSSION
The failure to find any but minor difference
in blood pressure level associated with area in
the United States is one of the more striking
findings of the Health Examination Survey. Small
differences no doubt do exist, although these may
be presumed of little meaning from an epidemi-
ological or public health point of view. Whether the
Health Examination Survey provided a sufficiently
precise instrument for measuring such small
differences may be questioned. The itinerary for
the Survey was designed to avoid the South in
summer and the North in winter, so that if there
are (as there may well be) slight fluctuations in
mean blood pressure with the seasons, slight
regional differences might be either lost or ac-
centuated as a consequence.
Table F. Mean systolic blood pressure of
Thife gaules, by region: United States,
1960-
Mean blood pres-
Region sure in mm. hg.
NortheasSt-========mm== 132.6
South=======mmee————— 127.6
WeSLmrmmmmm meme ———— 130.5
The difference found between white and Negro
blood pressures was, of course, expected. There
is evidence from a large number of surveys that
blood pressure is higher among the Negro race
than among the white, both in this country and in
the West Indies. Figures 3 and 4 compare the
findings from the Health Examination Survey
with those from three other surveys—two in this
country and one in Nassau.® > ¢ While all these
surveys (as well as at least one unpublished sur-
vey 7 ) show higher levels for the Negro population
than for the white, the Health Examination Survey
in general shows a smaller differential than the
others and also contrasts with these surveys in
showing little or no racial difference in pulse
pressure.
One difference between the surveys cited for
the United States and the Health Examination Sur-
vey is that the Health Examination Survey ex-
amined a random sample of the population whereas
the others examined generally rural populations.
It is conceivable that the racial contrast is
greater in rural areas than in the country as a
whole and that this accounts for the difference
between the findings of the Health Examination
‘Survey and other surveys in the United States.
However, the data from the Health Examination
Survey are too sparse to permit firm conclusions
on such details, but so far as can be judged from
these data, there is no evidence that this is so.
The racial contrast appears to be about the same
for the rural South as for the remainder of the
country.
The mean difference between the races found
by the Health Examination Survey was approxi-
mately 5 mm. hg., both in systolic and diastolic
pressures. This is not a very large difference,
and if the levels for the different races had been
obtained by different surveys it would be im-
possible to assert that the races differ in mean
‘blood pressure. Since the data are from one sur-
vey, however, the difference is clearly significant
statistically. It is not equally definite that this
statistically significant difference could not have
resulted (at least in part) from an artifact in the
examination process. It is conceivable that the
examination situation generally represents great-
er stress for Negro examinees than for white,
either because of differences in social status
between these groups or because of the fact
SYSTOLIC
MEN
+30
EES JS. 1960-62
MINI Rural US. (Gover)
I BR Muscogee Co, Go.
EV. 4m Nossou, Bahamas
+20 |—— 21 eo]
+o ——4f
Difference (mm. hg)
20 30 40 50 60 70 80
AGE
WOMEN
+30
+20
+10
-10 || | | | | |
20 30 40 50 60 70 80
AGE
Figure 3. !iean difference between systolic blood pressures for white and for Negro persons, by age and sex: four surveys.
NOTE: Mean systolic blood pressure for Negro persons minus mean systolic blood pressure for white persons.
that medical procedures and a medical setting
are less familiar to the average Negroexaminee.
It is, of course, well known that differences in
attitude toward an examination can affect the
blood pressure level by an amount as great as
that found between the Negro and white races in
this Survey.
This factor of tension would be difficult to
investigate and the Survey did not undertake to
do so. However, there is some indirect evidence
in the data collected. If the blood pressure is
measured repeatedly while the examinee is other-
wise undistracted, it tends to fall to a 'mear
basal” level.” The blood pressure of each ex-
aminee was measured three times, once at the
beginning of the physical examination, once near
the middle, and once at the end. While the physi-
cal examination did not provide an undistracted
setting, the examinee was probably less tense at
the end of the examination than at the beginning.
This was reflected in the blood pressure. On the
average, systolic blood pressure was lower on
the third measurement than on the first (diastolic
pressure remained unchanged). If there was a
greater drop during the physical examination in
the blood pressure for Negro than for white ex-
aminees, this would indicate a greater initial
tension on their part; if the drop was the same it
would argue that there is no racial difference in
tension.
MEN
+30
EE U.S. 1960-62
Hnnnm Rural U.S. (Gover)
IN Bm Muscogee Co.,Ga.
EV am Nassou, Bahamas
+20
+0 | - fp « \ —]
ole
Difference (mm hg)
-1o LL | | | |
20 30 40 50 60 70 80
AGE
DIASTOLIC
WOMEN
+30
+20
+10 (Nyy an pid — \
-10 LL | | | | |
20 30 40 50 60 70 80
AGE
Figure 4. Mean difference between diastolic blood pressures for white and for iiegro persons, by age and sex: four surveys.
NOTE: Mean diastolic blood pressure for Negro persons minus mean diastolic blood pressure for white persons.
On the first blood pressure measurement the
average blood pressures by race were as follows:
Dif-
fer-
Negro White ence
Systolic-========- 138.56 132.68 5.88
Diastolic========= 83.44 77.35 6.09
On the third measurement the mean pressures
were:
Dif-
fer-
Negro White ence
Systolic=========- 133.94 129.07 4.87
Diastolic-»=====-~ 82.72 77.50 5.22
In short, the decrease was greater for the Negro
examinee than for the white. It is therefore
reasonable to suggest that some of the difference
in mean blood pressure between Negro and white
examinees found by the Health Examination Sur-
vey reflects a greater tension by Negroexaminces
at the time of the examination.
SUMMARY
1. There were only slight differences between
regions of the United States in mean blood pres-
sure level, but these differences were sta-
tistically significant.
2. No differences in blood pressure level were
demonstrable between places (PSU's) of differ -
ent population size or between urban and rural
areas or between subdivisions of such areas.
3. The blood pressure of Negro adults was
greater than the blood pressure of white adults,
by 5.6 mm. hg. systolic and 5.0 mm. hg. dia-
stolic. The comparison was about the same if
the South is considered separately.
4, The HES data suggest that part of the recorded
racial difference in blood pressure readings
may arise from greater tension on the part
of Negro examinees at the time of examination.
REFERENCES
ys. National Health Survey: Plan and initial program of the
Health Examination Survey. Health Statistics. PHS Pub. No. 584-
A4. Public Health Service. Washington, D.C., ay 1962.
INational Center for Health Statistics: Cycle I of the Health
Examination Survey, sample and response. Vital and Health Statis-
tics. PHS Pub. No. 1000-Series 11-No. 1. Public Health Service.
Washington, D.C., April 1964.
3National Center for Health Statistics: Blood pressure of
adults, by age and sex, United States, 1960-1962. Vital and Health
Statistics. PHS Pub. No.1000-Series 11-No. 4. Public Health Serv-
ice.. Washington, D.C., June 1964.
“Gover, M.: Physical impairments of members of low-income
farm families--11,490 persons in 2,477 rural families examined by
the Farm Security Administration, 1940. VII. Variation of blood
pressure and heart disease with age; and the correlation of blood
pressure with height and weight. Pub. Health Rep. 63(34):1083-
1101, Aug. 1948.
Comstock, G. W.: An epidemiologic study of blood pressure
levels in a biracial community in the southern United States.
Am. ].Hyg. 65(3):271, 1957.
Gy olinison, B. C., and Remington, R. D.: A sampling study of
blood pressure levels in white and Negro residents of Nassau,
Bahamas. J. Chron. Dis. 13(1):39, 1961.
"McDonough, J. R., Garrison, G. E., and Hames, C. G.: Blood
pressure and hypertensive disease among Negroes and whites in
Evans County, Georgia. To be published.
8Veale, A. M. O., Hamilton, M., Irvine, R. O. H., and Smirk,
F. H.: Population study of casual and near-basal blood pressures:
with comments on survey techniques. New Zealand M.]. 61(354):
65-77, Feb. 1962.
O00
Table 1.
DETAILED TABLES
Mean blood pressures for white and Negro populations, by sex and age: United
States, 1960-62---cncemcrmrem rere r cree m meee eee meee eee esses esse esses sse—————
Mean blood pressures for white and Negro populations in the southern region, by
sex and age: United States, 1960-62--===w--memeceecccc ccc cece ccc cece —— eee ea —————
Percent of white and Negro adults with blood pressures of at least 160 systolic
or 95 diastolic, by sex and age: United States, 1960-62-----=memmeeeemecemeaene——
Number and percent distribution of systolic blood pressures for white and Negro
adults, by sex: United States, 1960-62«=wmremecrrrercnrer ese ee cere ee ————
Number and percent distribution of diastolic blood pressures for white and Negro
adults, by sex: United States, 1960~62-=m===rr>mr=mrmeemmeeeme esc n cece meee re —e———
Page
10
11
12
13
14
Table 1.
States, 1960-62
Mean blood pressures for white and Negro populations, by sex and age: United
Sex and age
Systolic
Diastolic
White | Negro
White | Negro
Total-18-79 years===-=-==ceemmcc ecco
18-24 years===m=mm mmm eee eee eee
25-34 years-==---==== mmm eee
35-44 years======m mmm eee eee
45-54 years=m=mm mm mmm mm meen
55-64 years=m==m=m mmm eee emcee
a ro TT TTT ——
75-79 years-=----=-==mm meee emcee
18-24
25-34
35-44
45-54
55-64
65-74
Total-18-79 years===-=m=m- meee eee cee m
75-79 years===-== mmm eee eee eee
Mean blood pressure
in mm, hg.
130.6 | 136.2 | 78.3| 83.3
131.8 | 136.2 | 79.0 83.4
122.3] 119.0 | 71.6 72.5
124.5] 127.4 | 76.0 79.8
127.9 | 134.7 | 80.2 84.4
133.2] 139.0 | 82.7 87.1
139.7 | 148.3 | 82.6 89.3
147.1] 158.3 | 80.5 86.9
154.1 | 156.5 | 78.9 84.9
129.4 | 136.3 | 77.5 83.2
111.6 | 114.6 | 69.1 71.5
115.2 | 119.7 | 72.5 76.6
121.6 | 132.1 | 77.0 85.3
132.2 | 147.8 | 81.1 89.9
145.8 | 155.7 | 84.2 91.9
159.2 | 175.2 | 83.3 89.7
156.5 | 162.8 | 79.1 82.9
10
Table 2. Mean blood pressures for white and Negro populations in the southern region,
by sex and age: United States, 1960-62
Systolic Diastolic
Age and sex
White | Negro | White | Negro
Mean blood pressure
in mm. hg.
Both sexes=18-79 years=======-cccmcommomcommaeaaan 127.6 | 138.1 | 78.2 84.3
Men
Total-18-79 years=======-emeecccece ccc ccc ccm —————— 129,11 137.7 79,1 84.6
18-24 years========--ecemeccccccce cece ce ccc e cee ————— 120.3 | 121.1 | 71.9 73.9
25-34 yearS====m=-memememmme mec cme eee ee eee esse ee ———————————— 123.6] 126.2 | 76.1 80.3
35-44 years 123.6] 141.1 79.6 88.1
45-54 years 131.6 | 143.5 | 83.1 90.4
55-64 years 138.5] 143.0 | 83.3 87.0
65-74 years 146.9 | 157.7 | 81.5 88.2
75-79 yearS=--==m=mmmmeeedcccccee cece ee cee eee eee. ————— 149.7 | 142.7 79.4 80.2
Women ’
Total-18-79 years---=-=cemeccccec ccc ccc ccc cee 126.4] 138.4 | 77.5 84.0
18-24 years=====mmmemeccmecccccc cece eee e eee cece —————— 110.3 | 115.4 | 70.7 72.0
25-34 years--===mmmmmmmcce ccc cm eee eec eee ee, —————— 113.6; 119.9 72.8 78.6
35-44 years=--=mmmmmemmmmcecmce ccc e eee e eee ————— 120.1 | 133.4 | 76.7 86.1
45-54 year S===mmm mmm eee eee eee 129.31 152.3 | 81,2 91.4
55-64 yearS-=m==mmmmmmmmemem ems —e sees esses eee e—————————— 143.3 | 148.6 | 84.0 88.0
65-74 yearS==-===mmmmcemcmcccece emcees ce eeeece ee —————— 159.1 | 177.3 | B4.4 93.0
75-79 years=====m=mmmmmmem mmm m see ————————————————————— 157.2 | 162.3 78.5 79.6
1
Table
3. Percent of white and Negro adults with blood pressures of at
least 160 sys-
tolic or 95 diastolic, by sex and age: United States, 1960-62
Systolic at least
Diastolic at least
Systolic at least 160
160 mm. hg. 95 mm. hg. mm. hg. or diastolic
Sex and age 95 mm. hg.
White Negro White Negro White Negro
Percent
Both sexes-18-79
years======mmm———- 10.5 18.8 8.7 22.0 14.7 27.6
Men
Total-18-79 years---- 8.6 16.8 9.1 22,6 13.6 27.6
18-24 years-=====mmm==——-—- 0.2 - 1,7 1.9 1,7 1.9
25-34 years-mmmmmmnemnac 0.7 4.6 3.4 11.5 3.7 12.5
35-44 years-=-=m=mmm————- 3+9 16,2 10.9 25.9 11.8 26.5
45-54 years=mmmmmmmmm———— 8.7 10.8 13.8 29.3 17.3 30.8
55-64 yearS-mmmmmmmm————- 15.9 29.4 11.9 31.6 21.4 44.6
65-74 years=====mmmmmm—-=- 26.1 63.2 12.3 40.5 27.3 66.0
75-79 years---===mmmmm———- 39.1 59.8 13,3 21,2 40.2 59.8
Women
Total-18-79 years---- 12.3 20.4 8.3 21.5 15.6 27.6
18-24 years--==mmmmmm———— iy 0.7 0.8 3.4 0.8 3.4
25-34 years====mmmmmm————- 0.7 3.4 2.) 8.5 2.3 8+5
35-44 years--mmmmmmmmm———- 2.3 14,3 3.3 24,1 6,2 25,6
45-54 years--=-rmmmmmm==- 10.7 30.8 10.9 34.3 15.5 41.9
55-64 years====mmmwmm=m=—— 25.3 33.8 16.4 36.7 31.0 41.0
65-74 years-=--=mmmm————- 45.4 68.5 7.9 32.1 48.6 71.0
7579 years==memwmmsawenwme 42.7 69.4 12.0 26.3 44.9 69.4
12
Table 4. Number and percent distribution of systolic blood pressures for white and
Negro adults, by sex: United States, 1960-62
Men Women Men Women
Pressure in mm. hg.
White Negro | White Negro | White | Negro | White | Negro
Number of persons in thousands Percent distribution
Total-mmmmmm———————— 46,561 | 5,195 | 51,184 | 6,219 | 100.0 | 100.0 | 100.0 100.0
Under 90-~=m=mmmmeeecne——— 43 - 167 18 0.1 & 0.3 0.3
90-99-=mmmmmm———————————— 584 99 2,258 196 1.3 1.9 4.4 3.1
100 L0G mmm me mmm mm om mm om mm en 3,517 434 7,566 825 7.6 8.4 14.8 13.3
110-119=mmmmmmm mm ———— 8,866 955 § 11,655 1,333 19:0 18.4 22.8 21.4
120-129 = mmmm mmm mm mmm owen 11,287 920 9,432 919 24.2 12.7 18.4 14.8
130-139 =mmmmmm mmm 9,290 814 | 6,813 698 20.0 15.7 13.3 1.2
140m 149 mm ome om om mm os om mn mm ow 5,558 571 | 4,296 536 11.9 11.0 8.4 8:6
1507 159 =m m mmm mmiom mom oot on mm mm om om 3,382 522 2,676 420 743 10.1 542 6.8
160-169-===mmmmmm mmm em 1,734 319 2,047 370 3.7 6.1 4.0 6.0
1707 179mm mmm mt om om om mm om om 1,060 249 1,467 246 2.3 4.8 2.9 3.9
180-189 mmm mm om mom mm mm om 447 157 1,085 236 1.0 3.0 Zk 3.8
190% 199» mmm mm mmm mow srs on om onion on om 416 86 843 127 0.9 1.6 1.6 2.0
200-209 == mmmmmm—————————— 214 34 465 61 0.5 0.2 0.9 1.0
210-219 mmm mmm ——————— 74 - 172 152 0.2 - 0.3 2.4
220-229- =~ mmmmmm—————————— 53 25 91 25 0.1 0.5 0.2 0.4
230-239 = mmm mm —————————— 27 - 88 19 0.1 - 0.2 0.3
240-249 mm mmmm———————————— - - 11 - - - 0.0 -
250-259 mmm mmmm————————— 9 9 13 - 0.0 0.2 0.0
260+- mmm mm mmm —————————— =" = 36 36 i - 0,1 0.6
13
Table 5. Number and percent distribution of diastolic blood pressures for white and
Negro adults, by sex: United States, 1960-62
Men Women Men Women
Pressure in mm. hg.
White Negro White Negro White | Negro | White| Negro
Number of persons in thousands Percent distribution
Total-==memmccc cee 46,561 5,195| 51,184| 6,219] 100.0 100.0 100.0 100.0
Under 50-==--mmmmemeeen—n—— 490 43 314 25 1.1 0.8 0.6 0.4
50-54 =m mmmmmm mmm me 367 42 552 23 0.8 0.8 1.1 1.5
55-59 m-mec em 846 108 1:573 74 1.8 2:1 3.1 1.2
60-64 m=mmmmmm mmm 2,362 167 3,619 279 5) 3.2 7:1 4.5
65-69 mmm 5,094 461 6,698 714 10.9 8.9 13.1 11.5
70-74 amma 6,689 686 8,636 680 14.4 13.2 16,9 10.9
75-79 mmmmm mmm ————————— 9,807 722 9,364 976 21.1 13.9 18.3 15,7
80-84==mmmmmm meme eee 153191 686 715923 891 15.4 13.2 15.5 14.3
85-89-=mmmmmm mcm 5,936 598 4,999 706 22.7 11.5 9.3 11.4
QO = Oly mm em ms mm ew 3,520 507 3.250 441 7.6 9.8 6.3 7.1
RR EE EE 2,023 638 1,819 457 4.3 12.3 3.6 7.4
100-104=mmmmmm mm ne 990 182 1,132 243 2.1 3.5 2.2 3.9
105 LOD == mm mimi om os am om om ime 663 108 616 221 1.4 2.1 1.2 3.6
110-114==-mmmmmmce ee me 243 132 191 99 0.5 2.5 0.4 1.6
115-110 mmm mmm msm on ono oon mn 162 35 234 145 0.3 1.1 0.5 2.3
120-124==m=mmmemmm mcm 80 27 109 52 0,2 0.5 0.2 0.8
125-129==mmmmmcccem mem - 25 64 83 - 0.5 0.1 1.3
130134 mmm mon omom om m wm o 49 18 14 0.1 - 0.0 0.2
135tm mmm mmm me —————— 48 9 72 24 0.1 0.2 0.1 0.4
14
APPENDIX |
STATISTICAL NOTES
The Survey Design
The Health Examination Survey is designed as a
highly stratified multistage sampling of the civilian,
noninstitutional population, aged 18-79 years, of the
conterminous United States. The first stage of the plan
is a sample of the 42 primary sampling units (PSU's)
from 1,900 geographic units into which the United States
has been divided. A PSU is a county, two or three
contiguous counties, or a standard metropolitan sta-
tistical area. Later stages resultin the random selection
of clusters of about four persons from a small neighbor-
hood within the PSU. The total sample included 7,710
persons in the 42 PSU's in 29 different States. The
detailed structure of the design and the conduct of the
Survey have been described in previous reports. 12
Reliability in Probability Surveys
The methodological strength of the Survey derives
especially from its use of scientific probability sampling
techniques and of highly standardized and closely con-
trolled measurement processes. This does not imply
that statistics from the Survey are exact or without
error. Data presented are imperfect for three im-
portant reasons: (l) results are subject to sampling
error, (2) the actual conduct of a survey never agrees
perfectly with the design, and (3) the measurement
process itself is inexact, even when standardized and
controlled. The faithfulness with which the study design
was carried out has been analyzed in a previous report?
Of the total of 7,710 sample persons, 86 percent
or 6,672 were examined. Analysis indicates that the
examined persons are a highly representative sample
of the adult civilian, noninstitutional population of the
United States. Imputation for the nonrespondents was
accomplished by attributing to nonexamined persons the
characteristics of comparable examined persons. The
specific procedure used ° consisted of inflating the
sampling weight for each examined person tocompensate
for nonexamined sample persons at the same stand and
of the same age-sex group.
While it is impossible to be certain that the blood
pressures are the same in the examined and the
nonexamined groups, the available evidence indicates
that they are. One source of information on this
question is a special inquiry sent to the physicians
of nonexamined persons and to the physicians of a
matching set of examined persons. The mean blood
pressure reported for the examined and for the non-
examined groups was in exact agreement. For further
details on this subject see Vital and Health Statistics,
Series 11, No. 4.
Sampling and Measurement Error
In this reportand its appendices, several references
have been made to efforts to evaluate both bias and
variability of the measurement techniques. The proba-
bility design of the Survey makes possible the calcu-
lation of sampling errors. Traditionally the role of the
sampling error has been the determination of how im-
precise the survey results may be because they come
from a sample rather than from measurement of all
elements in the universe.
The task of presenting sampling errors for a study
of the type of the Health Examination Survey is com-
plicated by atleast three factors. (1) Measurementerror
and ''pure' sampling error are confounded in the data;
it is not easy to find a procedure which will either
completely include both or treat one or the other
separately. (2) The survey design and estimation pro-
cedure are complex and accordingly require compu-
tationally involved techniques for calculation of vari-
ances. (3) Thousands of statistics come from the sur-
vey, many for subclasses of the population for which
there are small numbers of sample cases. Estimates
of sampling error are obtained from the sample data
and are themselves subject to sampling error, which
may be large when the number of cases in a cell is
small, or even occasionally when the number of cases
is substantial.
In the present report, estimates of approximate
sampling variability for selected statistics are pre-
sented in tables I and II. These estimates have been
prepared by a replication technique which yields over-
all variability through observation of variability among
random subsamples of the total sample. The method
15
Table I. Relative standard errors for estimated
actual mean systolic blood pressure of adults,
by region, population size, and place of resi-
dence
Relative
Item standard
error
Region
Northeast=======rmermemcreecc cae —c——————— 0.020
South=------eccecm ccm ce meme 0.020
West======m=mmeceemem cmc cm ec ——ee————— 0.020
Population size
Giant metropolitan areag--—=-=====-==== 0.010
Other very large metropolitan
areas----==----mm----cmmemceeo——ooa- 0.010
Other standard metropolitan sta-
tistical arease=emmrmwmmm—————————— 0.010
Other urban areas-=--===m==m==-==—————- 0.025
RUral areas=e=wwmwemmmmmmmmmee sm em -- 0.010
Residence
Urban-==~===e-ereee rc ————————— 0.005
SMSA—in central city--======m======- 0.010
SMSA==~outside central city-==-=-=------- 0.010
Not in SMSA=wemremereme eee. ———————— 0.015
Rural--====-e-ecece cmc c mmm ee 0.010
Farme====ee-eeemmccec ceca ccae een 0.015
Nonfarmeeee mere m mm mer mm. —————————— 0.010
reflects both "pure' sampling variance and a part of
measurement variance.
In accordance with usual practice, a 68 percent
confidence interval may be considered the range within
one standard error of the tabulated statistic and a
05 percent confidence interval the range within two
standard errors.
An overestimate of the standard error of a dif-
ference d=x-y of two statistics x and y is given by the
formula s fv svi] where v' and v? are
d X yi X y
relvariances respectively of x and y, or the squares
of the relative errors shown in table I. For example,
table D shows systolic x=130.6 for white adults and
y=136.2 for Negro, while from table II relvariances
are found to be: V = 0.000025 and vey = 0.000225.
The formula yields the estimate of standard error of
the difference (d = 5.6 ands, = 2.14 mm. hg.). Thus, as
the observed difference is more than two times its
sampling error, it can be concluded that systolic
blood pressure is higher among Negro adults than
among white.
Table II. Relative standard errors of mean blood
pressure for the United States and for the South,
by race, sex, and age
Systolic Diastolic
Area, sex, and age
White | Negro | White | Negro
United States
Both sexes-18-79
years--=--mmmm--- 0.0051 0.015) 0.005} 0.010
Men-18-79 years------ 0.005| 0.015] 0.005{ 0.015
Women-18-79 years----| 0.005] 0.020 | 0.005| 0.010
Men-55-64 years------ 0.010 | 0.025] 0.010 | 0.020
Women-35-44 years----| 0.005] 0.020| 0.010 | 0.015
South
Both sexes-18-79
years=-====m==== 0.0101 0,0201 0,010) 0,020
Men-18-79 years------ 0.010] 0.020 0.010 | 0.025
Women-18-79 years----| 0.015] 0.025| 0.010 | 0.025
Men-55-64 years 0.025] 0.040 | 0.020 | 0.040
Women-35-44 years----| 0.020 | 0.035] 0.020 | 0.030
Small Numbers
In some tables magnitudes are shown for cells for
which sample size is so small that the sampling error
may be several times as great as the statistic itself.
Obviously in such instances the statistic has nomeaning
in itself except to indicate that the true quantity is
small. Such numbers, if shown, have been included to
convey an impression of the overall story of the table.
Tests of Significance
As shown above, the difference in mean blood
pressure between Negro and white adults was submitted
to a formal test of significance and found to be sig-
nificantly different from zero. This difference could
have been examined in other ways. It might have been
more meaningful, for instance, to ask whether the
blood pressure for Negro adults was higher than
(rather than "different from') the blood pressure for
white adults. There is much evidence indicating this,
and the test for a one-sided hypothesis is more
powerful than the test for a two-sided hypothesis.
Alternatively, the question might have been, "Is the
blood pressure higher for Negro adults than for white
adults if age is held constant?" Conceivably, the age-
sex-specific means could be identical for the two
groups but a larger proportion of older people in one
group could lead to an overall higher blood pressure
for that group.
This last version of the hypothesis can be tested
directly from table 1, with the use of a table for the
binomial variable. Mean diastolic pressure is higher
for Negro adults in every age-sex group and mean
systolic in all but one age-sex group. The chances
of 14 heads out of 14 tosses of a true coin are 0.00006
and this corresponds to the case for diastolic pressure,
The chances of 13 or more heads out of 14 tosses are
0.00098 and this corresponds to the case for systolic
pressure. A more powerful statistical procedure could
be suggested but is obviously unnecessary.
Differences among areas are of course confounded
by age-sex differential composition. Accordingly a
review was made for regions of age-specific means
(not published). These show that in 12 of the 14 age-
sex groups, the mean systolic pressure was higher in
the Northeast than in the South. On the null hypothesis
of no difference, the probability of such a contrasting
result is about 0.01. In 13 age-sex groups it was higher
in the Northeast than in the West. If the data for white
adults are considered, the regional differentials are
even stronger. On the other hand, a comparison of
South and West shows 7 age-sex groups for which
systolic pressure is higher in one region than in the
other and 7 in which the comparison is reversed. It
can therefore be concluded that when these two latter
regions are compared the mean systolic pressure is
not consistently higher in either. Ranking the mean
systolic pressure for the three regions in each age-
sex group leads to the same average rank for South
and West, corroborating the supposition of no dif-
ference between these two regions.
Expected Values
In tables A-D, expected mean systolic blood pres-
sures are computed and the actual blood pressures for
the area are compared with the expected. The com-
putation of expected values was done as follows:
Suppose that in an area (say the Northeast) the
Health Examination Survey estimates that there are N,
persons in the ith age-sex group (i=1, 2,..., 14;
sum of N = N).
Table III. Excess of actual over expected blood
Pressure: by stand: Health Examination Survey,
1960-62
Blood pressure
Stand number
Systolic Diastolic
Mean deviation (mm. hg.)
[mmm mmm mi 0.6
2-mmmmmemmm—————— -0.5
-1.8
-0.5
2.3
-1.3
-5.7
“2.1
"a
NS
1
FHEOUNWLWOANVLWOUOULWOONNULUNEFUOPRLUNOODULDEWODSOO®
1
OHWOFRNFUVOOFOFFHFPWARNNONFFENONONFFOULWWLULFEFERULON
«+ a a
—
[e)]
1
1
ese ew eee eee.
www
Sw
i$ 38
Tm
BE A |
i 00
3
1 1
1 1
i 9a
va
1-4 3
I ¢
3's 3
$1
3 6
1 '
1 '
1 1
1 I
11
«ee
17 0
“oso.
REE
COHN ENOUNNOONNPENONNOPWNOARPUNWWOO
NOH OWNOFHHFOOCOONRPHRAFFRFNUFRDPWONENO
liree stands combined
Suppose the Health Examination Survey estimates
that the systolic pressure for the United States in the
ith age-sex group is X.
i.
Then the expected mean systolic pressure for the
area is
1 —
_ N X
N i i 1
If the N; are considered tobe a set of constants, the
variance of the expected value for an area will ordinarily
be negligible when compared with the variance of the
estimated actual value. This means that as a first
approximation the variance for the difference between
the actual and the expected value can be taken as equal
to the variance for the actual value. This should be
considered as indicative, only, since the actual and
expected values are not independent.
Stand Variation
The discussion of area differences may be supple-
mented by a consideration of differences among PSU's.
To do this, the following computation is performed:
Suppose that N, persons are examined at a stand
Let x, be the mean blood pressure obtained at this
stand for age-sex group i.
Let X, be the mean blood pressure for the ith age-
sex group as measured at all stands combined. Then
Pa) ZN & -%)
N ii i
D is a summary measure of the deviation of this stand
from the average stand and it is the statistic tabulated
in table III.
There were 42 stands at which examinations were
conducted. At most stands there were two physicians
who took examinees alternately. It was shown in the
first report on blood pressure } that blood pressure
measurements vary significantly from one examiner
to another. Since physicians were ordinarily engaged
for only one or two stands, differences between physi-
cians will be confounded with differences between
stands. The separation of these two components of
variance is a difficult undertaking in view of the com-
plexity of the sampling design and estimation procedures
used in the Health Examination Survey and is not at-
tempted in this report. However, preliminary analysis
indicates that there is a measurable component of
variation attributable to stand variation. The differences
presented in table III, however, considerably overstate
from the ith age-se i=1,2,..1 f . . :
N. = N) age-sex group, i=1, 2, ..., 14 (sumo the amount (because they include interexaminer vari-
i ’ ation).
O00
APPENDIX II
DEMOGRAPHIC TERMS
Age.—~The age recorded for each person is the age
at last birthday. Age is recorded in single years.
Race.—Race is recorded as ''White," "Negro," or
"Other." "Other" includes American Indian, Chinese,
Japanese, and so forth. Mexican persons are included
with "White" unless definitely known to be Indian or
other nonwhite race.
Population density.—The five classes comprising
this characteristic were derived from the design of the
sample which accomplished a stratification of the pri-
mary sampling units by population density in each of
three broad geographic locations. Because the Survey
was started in 1960, the primary sampling units within
each of the five population density classes were neces-
sarily based on populations and definitions of the 1950
census. The name of each selected primary sampling
unit within each population density class and geographic
location, along with selected sample data are presented
in an earlier report.”
The definitions for each of the five population density
classes are as follows:
Giant metropolitan areas.—This class includes
nine primary sampling units, defined in the 1950 census
as a standard metropolitan statistical area (SMSA) and
having a population of 3,000,000 persons or more.
Other very large metropolitan areas.— Included in
this class are six standard metropolitan statistical
areas with a population of 500,000 to 3,000,000 as
defined by the 1950 census.
Other standard metropolitan statistical areas.—
This class includes nine other SMSA's selected as
primary sampling units. With one exception—Provi-
dence, R.[.—all had less than 500,000 population.
Other urban.—This includes eight primary sam-
pling units which were highly urban in composition but
were not defined in 1950 as standard metropolitan areas.
Rural.—This includes 10 primary sampling units
which were primarily rural in composition according
to 1950 census definitions.
Region.— For the purpose of classifying the popu-
lation by geographic area, the United States was di-
vided into three major regions. This division was
Region
especially made for the design of the HES sample.
The regions and the States included are as follows:
States Included
Northeast------- Maine, Vermont, New Hampshire,
Massachusetts, Connecticut, Rhode
Island, New York, Pennsylvania,
Ohio, and Michigan.
South -=-ceeaeaa Delaware, Maryland, District of
Columbia, West Virginia, Virginia,
North Carolina, South Carolina,
Georgia, Florida, Kentucky,
Tennessee, Alabama, Mississippi,
Arkansas, Louisiana, Oklahoma,
and Texas.
West ==ceceean- Washington, Oregon, California,
Idaho, Nevada, Montana, Utah,
Arizona, Wyoming, Colorado,
New Mexico, North Dakota, South
Dakota, Nebraska, Kansas,
Minnesota, Iowa, Missouri,
Wisconsin, Illinois, and Indiana.
Location of residence terms.—This term refers
to urban or rural place of residence of the sample per-
sons. For the first six primary sampling units at
which examinations were conducted, the definition of
urban and rural was the same as that used in the 1950
census. These locations were Philadelphia, Pa.,
Valdosta, Ga., Akron, Ohio, Muskegon, Mich., Chicago,
[11., and Butler, Mo. For the remainder of the sampling
units, the 1960 census definitions were used.
The change from 1950 to 1960 definitions is of
small consequence in the Survey since only six loca-
tions were affected and the major difference is the
designation in 1960 of urban towns in New England and
of urban townships in New Jersey and Pennsylvania.
According to the 1960 definition, the urban popu-
lation comprises all persons living in (a) places of
2,500 inhabitants or more incorporated as cities,
19
boroughs, villages, and towns (except towns in New
England, New York, and Wisconsin); (b) the densely
settled urban fringe, whether incorporated or unin-
corporated, of urbanized areas; (c) towns in New
England and townships in New Jersey and Pennsylvania
which contain no incorporated municipalities as sub-
divisions and have either 25,000 inhabitants or more, or
a population of 2,500-25,000 and a density of 1,500
persons or more per square mile; (d) counties in
States other than the New England States, New Jersey,
and Pennsylvania that have no incorporated munici-
palities within their boundaries and have a density of
1,500 persons or more per square mile; and (e)
unincorporated places of 2,500 inhabitants or more not
included in any urban fringe. The remaining population
is classified as rural.
Size of place.—In this Survey the urban population
is classified as living "in the central city’ or "outside
the central city" of an SMSA. The remaining urban
population is classified as "not in SMSA."
The definitions and titles of standard metropolitan
statistical areas are established by the U.S. Bureau
of the Budget with the advice of the Federal Com-
mittee on Standard Metropolitan Statistical Areas.
The definition of an individual standard metropoli-
tan statistical area involves two considerations: first,
a city or cities of specified population to constitute
the central city and to identify the county in which it is
located as the central county; and, second, economic
and social relationships with contiguous counties which
are metropolitan in character, so that the periphery
of the specific metropolitan area may be determined.
Persons "in the central city" of an SMSA are
therefore defined as those whose residency is in the
city appearing in the stand and metropolitan statis-
tical area title. Persons residing in a SMSA but not
in the city appearing in the SMSA title are considered
to reside ''outside the central city."
Rural farm - nonfarm residence.—The rural popu-
lation may be subdivided into the rural-farm population,
which comprises all rural residents living on farms,
and the rural-nonfarm population, which comprises the
remaining rural population. The farm population in-
cludes all persons living in rural territory on places
of 10 or more acres from which sales of farm products
amounted to $50 or more during the previous 12 menths
or on places of less than 10 acres from which sales
of farm products amounted to $250 or more during the
preceding 12 months. Other persons living in rural
territory were classified as nonfarm. Persons were
also classified as nonfarm if their household paid rent
for the house but their rent did not include any land
used for farming.
C00
20
REPORTS FROM THE NATIONAL CENTER FOR HEALTH STATISTICS
Public Health Service Publication No. 1000
Series 1. Programs and collection procedures
Origin, Program, and Operation of the U.S. National Health Survey. 35 cents.
Health Survey Procedure: Concepts, Questionnaire Development, and Definitions in the Health Examination Survey. 45 cents.
Comparison of Two Methods of Constructing Abridged Life Tables. 15 cents.
Acute Conditions, Incidence and Associated Disability, United States, July 1961-June 1962. 40 cents.
Family Income in Relation to Selected Health Characteristics, United States. 40 cents.
Length of Convalescence After Surgery, United States, July 1960-June 1961. 35 cents.
Current Estimates From the Health Interview Survey, United States, July 1962-June 1963. 35 cents.
Impairments Due to Injury, by Class and Type of Accident, United States, July 1959-June 1961. 25 cents.
Disability Among Persons in the Labor Force, by Employment Status, United States, July 1961-June 1962.
Types of Injuries, Incidence and Associated Disability, United States, July 1957-June 1961. 35 cents.
Medical Care, Health Status, and Family Income, United States. 55 cents.
. Acute Conditions, Incidence and Associated Disability, United States, July 1962-June 1963.
. Cycle I of the Health Examination Survey: Sample and Response, United States, 1960-1962. 30 cents.
. Blood Pressure of Adults, by Race and Area, United States, 1960-1962.
No. 1.
No. 2.
Series 2. Data evaluation and methods research
No. 1. Comparison of Two Vision-Testing Devices. 30 cents.
No. 2. Measurement of Personal Health Expenditures. 45 cents.
No. 3. The One-Hour Oral Glucose Tolerance Test. 30 cents.
No. 4.
Series 3. Analytical studies
No. 1. The Change in Mortality Trend in the United States. 35 cents.
No. 2. Recent Mortality Trends in Chile. 30 cents.
Series 4. Documents and committee reports
No reports to date.
Series 10. Data From the Health Interview Survey
No. 1.
No. 2.
No. 3.
No. 4. Disability Days, United States, July 1961-June 1962. 40 cents.
No. 5.
No. 6.
No. T.
No. 8.
No. 9.
No. 10
Series 11. Data From the Health Examination Survey
No. 1
No. 2. Glucose Tolerance of Adults, United States, 1960-1962. 25 cents.
No. 3. Binocular Visual Acuity of Adults, United States, 1960-1962.
No. 4. Blood Pressure of Adults, by Age and Sex, United States, 1960-1962.
No. 5
Series 12. Data From the Health Records Survey
No reports to date.
Series 20. Data on mortality
No reports to date.
Series 21. Data on natality
No reports to date.
Series 22. Data on marriage and divorce
No reports to date.
Series 23.
Data from the program of sample surveys related to vital records
No reports to date.
Catalog Card
U.S. National Center for Health Statistics.
Blood pressure of adults by race and area, United States, 1960-1962. Mean blood
pressure by race and area. Washington, U.S. Department of Health, Education, and
Welfare, Public Health Service, 1964.
20 p. diagrs., tables. 27cm. (Its Vital and health statistics, Series 11, no. 5)
U.S. Public Health Service. Publication no. 1000, Series 11, no. 5.
1. Blood - Pressure - Statistics. I. Tite. (Series. Series: U.S. Public Health Service.
Publication no. 1000, Series 11, no. 5)
Cataloged by Department of Health, Education, and Welfare Library.
40 cents.
% U., S. GOVERNMENT PRINTING OFFICE : 1964 O - 735-762
iL
ate FORT TTC DE PE
aw
Sr
=
8
=
waa ir pts pg ep le Je mi lS
3 § A
Ce -
’
— N E A
— - w i
= i uf =
.
=
t
.
N
=
#
-
a
4
»
-
=
®
-
.
a
-
B
i
+
a
n
-
o
_n
.
= a
-
E
=
N r
-
p=
H
.
.
-
1
’
-
Ry
=
P
-
i)
:
NATIONAL vv
CENTER Series 11
For HEALTH Number 6
SYN RE aes]
Heart ILE
THT
United States. 1960-1962
U.S. DEPARTMENT OF /s
HEALTH, EDUCATION, AND WELFARE
Public Health Service
See inside of back cover
for catalog card,
Public Health Service Publication No. 1000-Series 11, No. 6
For sale by the Superintendent of Documents, Government Printing Office
Washington, D.C., 20402 - Price 35 cents
NATIONAL CENTER| Series 11
For HEALTH STATISTICS | Number 6
VITALand HEALTH STATISTICS
DATA FROM THE NATIONAL HEALTH SURVEY
Heart Disease
in Adults
United States. 1960-1962
A description of the examination and diagnostic pro-
cedures with major findings by age, sex, and race.
Washington, D.C. September 1964
U.S. DEPARTMENT OF
HEALTH, EDUCATION, AND WELFARE Public Health Service
Anthony J. Celebrezze Luther'L. Terry
Secretary Surgeon General
NATIONAL CENTER FOR HEALTH STATISTICS
Forrest E. Linder, Ph.D., Director
Theodore D. Woolsey, Deputy Director
Oswald K. Sagen, Ph.D., Assistant Director
(for Professional Relations)
Walt R. Simmons, M.A., Statistical Advisor
Alice M. Waterhouse, M.D., Medical Advisor
James E. Kelly, D.D.S., Dental Advisor
Louis R. Stolcis, M.A., Executive Officer
DIVISION OF HEALTH EXAMINATION STATISTICS
Arthur J. McDowell, Chief
COOPERATION OF THE BUREAU OF THE CENSUS
In accordance with specifications established by the National
Health Survey, the Bureau of the Census, under a contractual
agreement, participated in the design and selection of the sample,
and carried out the first stage of the field interviewing and certain
parts of the statistical processing.
Public Health Service Publication No. 1000-Series 11-No. 6
CONTENTS
The Cardiovascular Evaluation----------cemcmemmmmcnone—o- 1
The Medical HiStOry----===c=c=-mmmocmcmm mmm mmm mo 2
The Cardiac Examination-------cc-mcmmmmmemmmcccmcem em 2
Blood Pressure Measurement---------=ce--oc-memomoun—x 2
Other Parts of the Examination----===-cccccmcoc-neoa-o- 3
Comparison With Clinical Examination-------==-=c------- 3
Heart Disease Diagnosis--------=--c-c-mmmmmmmmmemmmonnao
Interpretation of the X-ray and Electrocardiogram---------
Classification and Criteria---==ceeeccmcmcccc ccc
Diagnosis =--=====m cmon meme meme
ON U1 WW
Age =m mmm mmm meme meme mmmmmm moomoo
Multiple Diagnosis----=-e=--m-cmomcmmm mimeo
Other Heart Disease------=--=mcmmmmoommmmmecmmmecmme oo
Heart Findings----==mmmmmmocmmmo comme emo em
N= \O 00 © 3
—
SUMMAYY =====m =m mmm mmm oem m oom cme mm mmm mmm mmm mom 12
Detailed TableS-=-==-ccmmmmmmm emcee erm emma 14
Appendix I. Medical History Questions Related to Cardio-
vascular Disease-------cmmmm mmm mmc eee 17
Appendix II. Forms Used in Recording Findings onthe Physical
Examination ==--ceememm mmm mmm mmm eee 21
Appendix III. Electrocardiographic Readings------=----=--- 28
Criteria and Classification----=-=-c-cemmmmmcmmcccaceooo 28
ECG Code Sheet=-=-cm-mmmmcmcm cece meme mmm mm em 32
CONTENTS—Continued
Appendix IV. Interpretation of Chest X-ray--------------=--
Form Used in Pulmonary Reading--------------ccccuuuo-
Form Used in Cardiovascular Reading-----------ccec-oo-
Pulmonary Readers----=--cmmmmm como ooo
Cardiovascular ReglerB-——=—====~snmmmnmmmcme mm =m www
Final Evaluation----=--cecmm mcm cece eee
Appendix V. Diagnostic Review----c- moomoo
Appendix VI. Statistical Notes----==-m-c-ccommm meee
The Survey Design----=--=-ccmmmmmm meee
Reliability in Probability Surveys=------eeceeccm ccc
Sampling and Measurement EXror------c-cecommmmaaooano
STAT] INTL DICT Siw mim 0 mi im 0 im 0 0 0 0
Tests of Significance-----=--=-cm-mmmmmmmcce meee eee
Demographic Terms------c-ecmmmm mom eee
SYMBOLS
Data not available------=-=ccecmmeeme eee mm
Category not applicable----eeemeocmaouano
Quantity BIBL Qin wm sw 0 8 rn
Quantity more than O but less than 0,05-----
Figure does not meet standards of
reliability or precision-----=--cceccaca-o
39
41
41
41
41
42
42
43
HEART DISEASE IN ADULTS
Tavia Gordon, Division of Health Examination Statistics
The National Health Survey uses three
methods for obtaining information about the
health of the U.S. population. The first is a
household interview in which persons are asked
to give information relating to their health or
to the health of other household members. The
second is the collection of data from available
health records. The third is direct examination.
The Health Examination Survey was organized to
use the third procedure, drawing samples of the
population of the United States and, by medical
examination and with various tests and measure-
ments, undertaking to characterize the population
under study.
The initial enterprise of the Health Examina-
tion Survey was the examination of a nationwide
probability sample of 7,710 persons aged 18-79
years. Its purpose was to obtain information
on the prevalence of cardiovascular disease,
arthritis, diabetes and certain other chronic
diseases, on dental health, and on the distribution
of a number of anthropometric and sensory
characteristics. Altogether, 6,672 persons were
examined during the course of the Survey which
was begun in October 1959 and completed in
December 1962. Sample persons received a
standard examination, lasting about 2 hours,
performed by medical and other staff members of
the Survey in specially designed mobile clinics.
This is one of a series of reports describing
and evaluating the plan, conduct, and findings of
the first cycle of the Health Examination Survey.
The description of the general plan! and of the
sample population and response 2 has been pub-
lished. These provide general background for all
reports of findings. In this report the cardiovascu-
lar examination is outlined and those parts of the
examination relating to the diagnosis of heart
disease are discussed. An account is given of the
method of evaluating the findings and of the pro-
cedures used in arriving atdiagnoses. The preva-
lence of heart disease in adults is summarized,
for total heart disease and for specific diagnoses.
THE CARDIOVASCULAR
EVALUATION
The cardiovascular evaluation included the
following:
1. A medical history
2. A cardiovascular examination performed
by a fellow or first-year resident in
internal medicine with
a. Three measurements of blood pres-
sure
b. Examination of the ocular fundi with
an ophthalmoscope
c. Examination of the neck for venous
engorgement
d. Inspection and palpation of the pe-
ripheral arteries
€. Examination of the extremities for
evidence of edema
f. Examination of the heart by auscul-
tation and palpation for thrills, heart
sounds, or murmurs
g. Other observations which might con-
tribute to differential diagnosis, such
as a set of serological tests for syphi-
lis and evidences of thyroid enlarge-
ment, congenital malformations, phy-
sical impairments, and residuals of
cerebrovascular accidents.
3. A 12-lead electrocardiogram
4. A chest X-ray—14 by 17 inches in size,
taken at a 6-foot distance
The Medical History
The cardiovascular examination began witha
self-administered medical history. After a brief
interview by a receptionist, the examinee was
asked to complete a medical history form. The
receptionist remained available to provide the
examinee with any assistance necessary. Included
among the questions were some concerning cardi-
ovascular symptoms or disease. These are shown
in Appendix I. The examinee was then offered a
drink which included 50 grams of glucose, unless
he was under treatment for diabetes, and after
completing the self-administered history was
asked a few additional questions by the recep-
tionist. These included questions about physical
handicaps, major health problems, and operations
and were designed to elicit relevant medical in-
formation that had not appeared in response to
the more specific questions on the history. The
receptionist, at the same time, reviewed the his-
tory both for completeness and for consistency
and queried the examinee further where any
deficiencies were evident.
The examining physician reviewed the medi-
cal history before beginning the physical exami-
nation. He attempted to correct any incomplete-
ness or inconsistency remaining in the record and
where the examinee had been uncertain in his
answer attempted to arrive at a definite ''yes'
or 'mo" by further questioning. In some cases
he could not. For most of the cardiovascular
questions the physician was instructed to ask for
further information if an answer of ''yes' or "?"
had been checked, or if the examinee had indi-
cated that he did not know the answer. A series
of standard probes were used (Appendix I) and
the answers to these were recorded. When these
probes were completed the physician was free
to further question the examinee until he was
satisfied that he had all the relevant information
that could be obtained in a single session.
Among the cardiovascular questions two were
of especial importance for the diagnosis of heart
disease—questions 21 and 22 (Appendix I). These
dealt with chest pain and heart pain. It was on
the basis of the response to these questions and
the associated probes that a diagnosis of angina
pectoris was made. Responses to the other
cardiovascular questions on the medical history
form were also of assistance in, although not
sufficient in themselves for, heart disease diag-
nosis.
The Cardiac Examination
After reviewing the medical history, the
physician began the physical examination. In-
cluded in this was a standardized examination
of the heart, undertaken without exercise. The
precordium was palpated for thrills with the
examinee first sitting upright, then leaning for-
ward. This was first done with the examinee
breathing normally and then repeated with the
examinee holding his breath in expiration. Aus-
cultation was done with a stethoscope, using both
the bell and the diaphragm, and proceeded from
the apex upward along the left sternal border
and then to the pulmonic and aortic areas. It
was done with the examinee upright, first breath-
ing normally and then holding his breath in ex-
piration. Next, palpation and auscultation were
repeated with the examinee supine. Finally, he
rolled over on his left side and was examined
with the bell and palpated for thrills.
Findings from this examination were re-
corded on a standard form (Appendix II). If a
murmur was noted it was described in specific
terms, as to intensity, time, pitch, quality, and
duration.’ Intensity was graded on a five-point
scale, from very faint (grade 1) to very loud
(grade 5).
Blood Pressure Measurement
Three blood pressure measurements were
made, the first just after the physician met the
examinee; the second midway in the examination,
after completing the auscultation of the heart
in the sitting position; and the third at the end
of the examination. Blood pressures were taken
while the examinee was sitting on the examining
table. The nurse placed the middle of the cuff over
the bulge in the upper left arm. The cuff was
left on the arm between the first and second
measurements, removed after the second, and
returned for the third. The physician held the
arm at the level of the atrium, with the nurse
holding the Baumanometer at the physician's
eye level. Using the bell of his stethoscope,
the physician noted the pressure when the sound
first was heard, when it first became muffled,
and when it disappeared. All three measurements
were recorded. The point at which the Korotkov
sounds disappeared was taken as the diastolic
pressure. If the sounds did not disappear, the
point of muffling, if distinctly heard, was used.
Since the Baumanometer is scaled in intervals
of 2 mm., measurements were so recorded.
Some results from this examination have already
been reported.’
Other Parts of the Examination
For the chest X-ray, a posterior-anterior
view was taken at a 6-foot distance and recorded
on a 14 by 17 inchfilm. The exposure was taken in
inspiration but was not timed for a fixed phase
of the heart cycle. The electrocardiogram was
obtained by a Twin Visomachine (model 60-1300).
Twelve leads were recorded: I, II, III, AVR, AVL,
AVF, V,-v 6
The other aspects of the cardiovascular
examination, while not leading to the diagnosis
of heart disease as such, were helpful either in
evaluating the signs of heart disease or in
determining a specific etiology. Thus, the pres-
ence of congenital abnormalities might contribute
to the differential diagnosis of congenital heart
disease. The finding of a positive serological
test for syphilis was required in order to make
a diagnosis of syphilitic heart disease.
Comparison With Clinical Examination
The uniform, single-visit examination used
for the Health Examination Survey differed in
both objectives and procedures from the usual
clinical examination. In clinical practice the
objectives are evaluation and medical manage-
ment of the individual patient. Usually the patient
is being studied because of some complaint for
which he has sought medical advice. If the diag-
nosis or treatment seems obvious on clinical
grounds, the workup may be minimal. On the
other hand, if the diagnostic clues are equivocal,
there may be an extended series of tests and
consultations and the patient may be under obser-
vation for an appreciable period before diagnosis.
Diagnosis may be modified by the patient's re-
sponse to treatment, by his subsequent clinical
history, or by new findings. There is, in short, a
variable diagnostic workup and an extended oppor -
tunity to confirm or reject the original impres-
sions.
On the other hand, the purpose of the Health
Examination Survey is to characterize a popu-
lation group. The cardiovascular examination
was designed to provide reliable diagnostic in-
formation insofar as such information could be
obtained during a single visit. Since there was no
responsibility for patient care, persons with
medical complaints need not be diagnosed as
having disease if the findings were equivocal or
nonspecific. Since persons did not present
themselves for medical care but because they
were members of a population sample, the
absence of complaints gave no assurance that there
was no disease. Therefore, a standardized exami-
nation was given to every examinee.
Prior to beginning the firstcycle of the Health
Examination Survey, a special study was under-
taken under the direction of Dr. Jeremiah Stam-
ler.¢ Its purposes were to design a single-visit
cardiovascular examination which would yield
diagnoses in accord with current survey practice,
to compare diagnoses obtained by this examination
with diagnoses obtained for the same individuals
by a replicate of this examination, and tocompare
diagnoses made by the single-visit examination
with diagnoses arrived at in clinical practice.
The single-visit examination developed for this
study was later adopted, with minor modifications,
by the Health Examination Survey for use in its
examination of adults.
While there is a distinct contrast between the
standardized single-visit examination anda clini-
cal examination, the study did not find large dif-
ferences between the two in diagnostic results.
The chief discrepancies were with respect to
coronary heart disease. The diagnosis of angina
pectoris was more common on the single-visit
examination than on the clinical, whereas minor
electrocardiographic abnormalities were more
likely to lead to a diagnosis of coronary heart
disease on the clinical examination than on the
single-visit examination.
HEART DISEASE DIAGNOSIS
Several intermediate steps were involved
in progressing from examination findings to
heart disease diagnoses. The first step was
interpreting the chest X-ray film and the elec-
trocardiographic tracing. The second was con-
structing a set of diagnostic criteria. The third
was developing a procedure for translating the
findings from the examination and the interpre-
tation of the X-ray and electrocardiogram into
specific diagnoses. How these steps were taken
for the Health Examination Survey is discussed
in the following sections.
Interpretation of the X-ray
and Electrocardiogram
Both the electrocardiogram and the chest
X-ray were interpreted independently by several
specialists. These interpretations were made
without any other information about the examinee.
The electrocardiogram was read independ-
ently by three cardiologists according to criteria
agreed upon in advance. These criteria are
specified in Appendix III, which also contains a
reproduction of the precoded form on which the
findings were entered. For all major findings
allowance was made for designating any electro-
cardiographic abnormality observed by the elec-
trocardiographic reader even though the specified
criteria for that abnormality were not satisfied.
After completion, the three independent determi-
nations were compared. Where they all agreed,
the unanimous decision was used for subsequent
diagnosis. In the event that there was any disa-
greement, the three met with Dr. Michael A.
Corrado, who served as coordinator for this work,
and together they came to a final decision. This
final decision was the one used in such cases.
The evaluation of the chest X-ray was a some-
what more complicated undertaking. Initially,
arrangements were made to have the X-ray films
interpreted by radiologists specializing in pul-
monary disease. In addition to noting evidence of
pulmonary disease, the ''pulmonary readers"
were requested to record evidence of distinct
cardiovascular abnormality. As had been antici-
pated, this led to an estimate of the prevalence of
cardiovascular abnormalities which was much
lower than is ordinarily found in cardiovascular
surveys. Another group of radiologists was there-
fore employed to reexamine the films for evidence
of cardiovascular abnormality. These 'cardi-
ovascular readers' were chosen on the basis of
standards set by Dr. Lloyd E. Hawes, radiologist
for the Framingham Heart Study. A set of films
from the Health Examination Survey was read
first by him and then by a number of different
radiologists. Three were found to employ about
the same standards as Dr. Hawes and were chosen
to read the Health Examination Survey films for
cardiovascular abnormalities. Each was given a
random third of the films toread. The forms used
in recording the radiological findings for both the
"pulmonary readers' and the ''cardiovascular
readers'' are reproduced in Appendix IV.
The reading procedure was designed as fol-
lows. A finding of general cardiac enlargement or
left ventricular hypertrophy, definite or possible,
was considered "positive." All films were read by
two pulmonary readers and one cardiovascular
reader. The determination of the two pulmonary
readers provided a preliminary evaluation. If
‘both considered the film ''positive'' a decision of
enlargement was made whatever the findings of the
cardiovascular reader. If they disagreed and the
cardiovascular reader considered the film posi-
tive, the decision was that enlargement was pres-
ent; otherwise a second cardiovascular reader
interpreted the film and his decision was final.
If the two pulmonary readers considered the film
"negative'' and the cardiovascular reader agreed
with them, the decision was that no enlargement
was present; otherwise a second cardiovascular
reader examined the film and his decision was
binding. All decisions were made independently
and no reconciliation of differences was under-
taken.
The rationale for this procedure is too com-
plicated to be discussed at this point. It is partly
explained in Appendix IV. The effect was to
produce reading results which conformed well,
both in level of abnormalities found and in
attributions to specific individuals, with the
standards of the Framingham Heart Study.
Classification and Criteria
After extensive consultation the Health Ex-
amination Survey arrived at the following diag-
nostic categories and criteria for hypertension
and heart disease. Ultimately, they were derived
from definitions of the New York Heart Associ-
ation’ but were modified to fit the circumstances
of population surveys in general and of the
Health Examination Survey in particular,’ 8
Hypertension:
Hypertension.—160 mm. hg. or over sys-
tolic or 95 mm. hg. or over diastolic
Borderline hypertension.—Below160 mm. hg.
systolic and below 95 mm. hg. diastolic, but not
simultaneously below both 140 and 90 mm. hg.
Normotension.— Below both 140 mm. hg. sys-
tolic and 90 mm, hg. diastolic
(When aortic insufficiency is present or the heart
rate is under 60, hypertension or borderline hy-
pertension must be defined by the diastolic
pressure.)
Hypertensive Heart Disease
Definite.— One of the following:
1. Hypertension plus left bundle branch block
or left ventricular hypertrophy (LVH) by
ECG. (By voltage criteria when 35 years
of age or over. If under 35 years left
ventricular or subendocardial ischemia
must be present in addition to LVH by
voltage criteria. No person under 35 had
hypertension or borderline hypertension
with this combination of ECG findings.)
2. Hypertension plus LVHor general cardiac
enlargement (GCE) by X-ray.
3. A history of hypertension currently on
medication for hypertension, and LVH or
GCE by X-ray and/or LVH by ECG.
Suspect.—One of the following:
1. Borderline hypertension plus LVHby ECG
and/or LVH or GCE by X-ray.
741-678 O - 64 - 2
2. Borderline hypertension plus LVHor GCE
by X-ray.
Rheumatic Heart Disease
Definite.—One of the following:
1. Any diastolic murmur in the absence of
evidence of a congenital or syphilitic
etiology.
2. If there is no history of rheumatic fever
or chorea, a grade 4 pansystolic murmur
at the apex in the absence of other evidence
of congenital heart disease.
3. History of rheumatic fever or chorea and
a grade 3 pansystolic murmur at the apex.
No Suspect Category
Syphilitic Heart Disease
Definite.— Positive serology and a diastolic mur-
mur at the base.
No Suspect Category
Coronary Heart Disease
Definite.— One of the following:
1. Myocardial infarction (MI) on ECG and/or
definite angina (judgment of examining
physician). Angina will not be ascribed to
coronary heart disease if aortic stenosis
or syphilitic heart disease is present.
2. History of myocardial infarction in judg-
ment of examining physician and either
left ventricular ischemia on the ECG or
myocardial infarction on ECG outside
criteria.
Suspect.—One of the following:
1. History of myocardial infarction in judg-
ment of examining physician with no
evidence of myocardial infarction or left
ventricular ischemia on the ECG.
2. Suspect angina (judgment of examining
physician).
Congenital Heart Disease
Individual case review—no suspect category
Other Heart Disease
Definite.—One of the following:
Medical examination
1. Aortic stenosis (a systolic ejection mur-
mur accompanied by a thrill at the base
with diminished or absent Ay in the
absence of other etiology).
ECG findings
Left bundle branch block
Complete heart block
Atrial fibrillation
LVH including left ventricular ischemia
or subendocardial ischemia
as WN
Suspect.—One of the following:
X-ray findings
1. Cardiomegaly (LVH or GCE)
ECG findings
Right bundle branch block
Partial A-V block
Atrial flutter
Right ventricular hypertrophy
Isolated left ventricular ischemia
(complete)
sw
There are some omissions from this list.
Because sample persons had to visit the mobile
center for examination two manifestations of
heart disease were automatically omitted. The
first were acute clinical episodes. The second
were episodes which run a rapid fatal course—
in particular, coronary heart disease first mani-
festing itself as ''sudden death." Moreover, past
manifestations which left only equivocal evidence
were also undiagnosed.
In addition, the Survey chose to ignore two
important clinical manifestations of heart dis-
ease. It was thought that the examination was not
adequate for diagnosing cases of congestive
heart failure. To be sure, most such cases were
diagnosed as some form of heartdisease, anyhow,
because of other findings in the examination,
but a few cases did elude diagnosis. Cor pulmonale
was also omitted, again because it was decided
that the examination was inadequate for differ-
ential diagnosis. Again a few cases of heart dis-
ease were not diagnosed because of this omission.
In diagnosing rheumatic, congenital, and
syphilitic heart disease no provision was made
for a category of suspect disease, chiefly because
such diagnoses depend on quite subtle differ-
entiations of heart sounds and murmurs. It was
thought that without verification of the examining
physician's impressions it would be unwise to
place great weight on such evidence; therefore
the Survey considered it preferable to omit
suspect categories for these diseases.
Diagnosis
After all the findings were available, the
final step was to arrive at a diagnosis. Even in
favorable circumstances this is a difficult proc-
ess to standardize. In the Health Examination
Survey, it was more difficult than usual. There
were 62 different examining physicians. To rely
on their consistent use of the same diagnostic
standards and criteria was clearly impossible.
What is more, they did not have available the
specialist judgments on the electrocardiographic
tracing and the chest X-ray or the findings from
the serological tests for syphilis. Thus, though
the examining physician was requested to enter
his diagnostic impressions, these were used only
as indicators; the final diagnoses were made by
the permanent staff of the Survey, with consultant
help in difficult cases.
The first step in this procedure was to supply
a set of rules suitable for diagnosis by computer,
which would convert the coded information from
the medical record and from the interpretation
of the X-ray film and the electrocardiogram into
a diagnostic decision. An example of the computer
output is given in Appendix V. Some of these
decisions were then subject to review. For the
first few hundred cases all computer diagnoses
were reviewed by Dr. Alice M. Waterhouse,
medical advisor to the National Center for
Health Statistics. These reviews made it evident
that many diagnostic decisions did not require a
special medical review and the classes of cases
subject to review were finally narrowed to the
following:
1. Cases with significant murmurs.
2. Cases with a diagnosis of angina pectoris.
3. Cases where the diagnosis depended on a
history of hypertension or a history of
myocardial infarction.
4. Cases with electrocardiographic findings
of myocardial infarction outside of cri-
teria or of left ventricular ischemia,
where a diagnosis of definite coronary
heart disease had not been made.
5. Cases diagnosed as having heart disease
by the examining physician but not by
the computer.
This omitted from review those cases witha clear
and definite diagnosis of heart disease on the
available evidence and those cases where there
was no possibility of diagnosing heart disease
from the available evidence.
In most cases where the computer diagnosis
was reviewed, the diagnostic decision made by
the computer was unaltered. In a few instances,
however, there was a diagnostic change on the
basis of review. Where a review decision seemed
to require specialist judgment the case was re-
ferred to Dr. Abraham Kagan of the Framingham
Heart Program for a final decision. The discussion
of the details of these decisions is not feasible,
but in general equivocal evidence of heart dis-
ease was treated as nondiagnostic, althoughitwas
recognized that some of these cases would
warrant medical supervision.
The review procedure did more than arrive
at final diagnoses. It also submitted the diagnostic
Table A. Prevalence of definite
criteria to repeated scrutiny. In the balance they
appear to be both reasonable and conservative.
MAJOR FINDINGS
Of the 111.1 million adults in the United
States, some 14.6 million had definite heart dis-
ease and nearly the same number had suspect
heart disease. Of every 100 persons aged 18-79
years, 13.2 had definite heart disease while an
additional 11.7 had suspect heart disease (table A).
Among the specific forms of the disease, the
one most commonly encountered was hypertensive
heart disease. More than 10 million adults had
definite hypertensive heart disease; nearly 4.8
million had suspect hypertensive heart disease.
Numerically, coronary heart disease was next
in importance, with 3.1 million definite and 2.4
million suspect cases. Other forms of heart dis-
ease accounted for substantially fewer cases.
Sex
Definite heart disease was more prevalentin
women than in men, while suspect heart disease
was more prevalent in men than in women (tables 1
and B). The relationship varied with the diagnosis.
Women were more likely to have definite hyper-
tensive heart disease; men were more likely to
have definite coronary heart disease or heart
and suspect heart disease in adults, by heart disease
diagnosis: United States, 1960-62
Definite
, di 3 i
Heart disease diagnosis weatt disease
heart disease
Definite
heart disease
Suspect
heart disease
Suspect
Number of adults in thousands
Percent of all adults
Total-===-==mmemee——— 14,621 12,979 13,2 11,7
Hypertensive-=============- 10,499 4,759 9.5 4.3
Coronary=================== 3,125 2,410 2.8 2,2
Rheumatic==========c==c==-= 1,270 PR 1.1 “oie
Congenital--====mmmeee—m=-=- 244 0.2 wine
Syphilitic-====m==m=m=m=mmn- 147 is 0.1 ove
Other----===--eeecomomcoaan- 292 7,330 0.3 6.6
NOTE: Counts for "Other" exclude persons with any of the specified heart diseases.
Counts for the specified heart diseases, on the other hand, are not exclusive. The cri-
teria do not provide
heart disease.
for "possible" categories of rheumatic, congenital, or syphilitic
Table B. Prevalence of definite and suspect heart disease in men and women, by heart
disease diagnosis: United States, 1960-62
Heart disease diagnosis Men Women Men Women
Definite heart disease Number of adults in thousands Percent of all adults
Total-===-=eceeece eau 6,652 7,970 12.6 13.7
Hypertensive--=-=--===-ce=ucea-x 4,050 6,449 7.7 11.1
Coronary-=====-====ce-eecea=-= 1,945 1,180 3.7 2.0
Rheumatic========ceceeceuau-= 608 662 L+2 1.1
Congenital----===mrerme mm ——— : 160 84 0.3 0.1
Syphilitic--====mmmmemcc eeu 921 57 0.2 0.1
Other=-===-eeecmce ccc cece 128 164 0.2 0.3
Suspect heart disease
Total-==-===receeccna— 75315 5,663 13.9 9,7
Hypertensive-=-==-====ccecceax 2,518 1,914 4.8 3.3
Coronary======-=memec eee aea= 1,136 1,274 2,2 2.2
Other--=-==--eeceece ccm eee e 4,122 3,208 7.8 5.5
NOTE: Counts for "Other" exclude persons with any of the specified heart diseases.
Counts for specified heart diseases, on the other hand, are not exclusive. The criteria
do not provide for ''possible"
disease.
disease of congenital or syphilitic origin. On the
other hand, suspect hypertensive heart disease
was more common in men than in women, while
suspect coronary disease was more common in
women than in men. The significance of these
differentials will not be discussed in this report.
Race
Heart disease was more common in Negro
than in white adults (tables 2 and C). (Comparison
of racial differences is limited to findings for
white and Negro persons since the sample was too
small to permit adequate representation of other
nonwhite races.) Some 24.4 per 100 Negro adults
had definite heart disease as contrasted with 12.0
per 100 white adults. For suspect heart disease,
the prevalence rates were 14.8 and 11.3 per
hundred, respectively. This racial difference,
evidenced by both men and women, arose from
the marked racial contrast in the prevalence of
hypertensive heart disease. For definite hyper-
categories of rheumatic, congenital, or syphilitic heart
tensive heart disease, the prevalence was nearly
3 times as great for Negro men as for white
and 2.2 times as great for Negro as for white
women. A similar, but much smaller, difference
was noted for suspect hypertensive heartdisease.
With other heartdisease categories, where preva-
lence rates were lower, it is difficult to be sure
whether there was a racial difference in the preva-
lence of disease. For coronary and rheumatic
heart disease, specifically, there is no evidence
from this Survey of a racial difference in preva-
lence.
Age
The prevalence of heartdisease rose sharply
with age. In the age group 18-24 years, less than
2 percent had definite heart disease. By age
75-79 years, 39 percent of the men and 46 per-
cent of the women had definite heart disease
(table 1). A similar, althoughless steep, gradient
with age was observed for suspect heart disease.
Table C. Prevalence of definite and sus-
pect heart disease in white and Negro
adults, by heart disease diagnosis:
United States, 1960-62
Heart disease diagnosis| White | Negro
Percent of
Definite heart disease specified race
Total-=======n===- 12.0 24.4
Hypertensive----======== 8.2 20.8
Coronary--==============- 2.9 2.6
Rheumatic-~======m=mmw=- 1.1 1.7
Congenital====-=========- 0.2 0.2
Syphilitic-=-=======""=~ 0.1 0.7
Other====-==c-cecceceaaa- 0.3 0,2
Suspect heart disease
Total-=====mmmmm=- 11.3 14.8
Hypertensive----======-= 3.9 4.9
Coronary========mm=mmmmn= 242 2.6
Other----====mmemem em ——— 6.4 8.3
NOTE: Counts for "Other" exclude per-
sons with any of the specified heart dis-
eases. Counts for the specified heart
diseases, on the other hand, are not ex-
clusive. The criteria do not provide for
"possible" categories of rheumatic, con-
genital, or syphilitic heart disease.
The majority of all persons in the age group
75-79 years had heart disease of some form,
with more persons manifesting definite than
suspect evidence of such disease. Heart disease
was more common in men than in women until
age 55 years and more common in women at older
ages. In other words, heart disease prevalence
rose with age more rapidly for women than for
men.
The tendency for heart disease prevalence to
increase sharply with age can be observed in both
definite and suspect hypertensive and coronary
heart disease (tables 3 and 4, figs. 1, 2, and 3).
The curves exhibiting prevalence by age tended to
be steeper for women than for men in definite
hypertensive heart disease and suspect coronary
heart disease. For definite coronary heart disease
the curves for men and women are closely
parallel, while for suspect hypertensive heart
disease a sex difference is moot.
The prevalence of rheumatic heart disease
also increased with age (table 5, fig. 4). For
syphilitic and congenital heart diseases there
were too few sample cases for any judgments
to be made about differentials by age, sex, or
race.
Multiple Diagnosis
In a substantial number of cases a diagnosis
of heart disease, while appearing under one
rubric, could have been made on more than one
basis. Thus, the weight of evidence of heart
disease is really greater than has been indicated
up to this point. At the same time the relative
frequency of multiple heart disease findings
complicates the discussion considerably. Among
persons with suspect hypertensive heart disease,
for example, are some with only marginal evidence
of heart disease and others with very definite
evidence of heart disease, perhaps with a grossly
enlarged heart and distinct electrocardiographic
abnormality, but with only marginal evidence ofa
hypertensive etiology. This is equally true of other
diagnostic categories. The following examples
will make this clearer.
Of the persons with definite hypertensive
heart disease fully 89 percent would be considered
to have heart disease even in the absence of
hypertension (table D). Some 16 percent had
coexisting coronary heart disease or some other
specific form of heart disease. About 7 percent
more would be considered to have definite heart
disease on the basis of their electrocardiograms
and for another 64 percent a finding of heart en-
largement on X-ray would, by itself, have led to
a diagnosis of suspect heart disease. For suspect
hypertensive heart disease the comparable per-
centages are equally impressive. Some 14 percent
had another specific heart disease, another 2 per-
cent had definite heart disease on the basis of
their electrocardiogram, while 69 percent more
would be considered to have suspect heart disease
on the basis of evidence of heart enlargement by
X-ray.
With coronary heart disease the situation is
similar (table E). About 41 percent of all defi-
nite cases of coronary heart disease had co-
existing hypertensive heart disease or some other
specific heart disease and another 15 percent had
some other evidence of heart disease. For suspect
800 ~ 800
WHITE WHITE
S—— Men ————— Men
mmmms Women =mme-e Women
NEGRO
a
a 2
3 &
2 5
©
z
Zz 100 o 100
E <
= <
3 >
z a
o o
5 a
a
s w
i &
S oO
&
& &
w uw
© 10 o 10
= =
Zz 2
w w
o oO
[+4 [+4
w w
a a
ol | | | | | ol 1 | | | |
20 30 40 50 60 70 80 20 30 40 50 60 70 80
AGE AGE
Figure I. Prevalence of definite hypertensive heart disease for white and Figure 2. Prevalence of suspect hypertensive heart disease for white and
Negro adults, by age and sex. Negro adults, by age and sex.
NOTE: Because of the small sample size, rates for the Negro population are subject to very high sampling errors.
80.0 80.0
DEFINITE CHD —— Men
— Men mmmm= Women
----- Women
SUSPECT CHD
Men
tereesnenans Women a
2
2 °
2 s
x
o 10.0 10.0
x
x w
by 5
? 3
& <
<<
o @
w w
ht o
Oo w
8 a
a n
uw
w
© 10 o
-
= z
u o
x x
w w
a a
ol | 1 | | 1 o.l 1 | 1 1 |
20 30 40 50 60 70 80 20 30 40 50 60 70 80
AGE AGE
Figure 3. Prevalence of definite and suspect coronary heart disease, by Figure 4. Prevalence of definite and suspect rheumatic heart disease, by
age and sex. age and sex.
10
Table D. Prevalence of definite and sus-
pect hypertensive heart disease in
adults with and without other heart
disease: United States, 1960-62
Table E. Prevalence of definite and sus-
pect coronary heart disease in adults
with and without other heart disease:
United States, 1960-62
Hypertensive heart Coronary heart
Other heart disease disease diagnosis Other heart disease disease diagnosis
diagnosis diagnosis
Definite Suspect Definite Suspect
Number of adults Number of adults
in thousands in thousands
Total-========- 10,499 4,759 Total========- 3,125 2,410
None============e=-- 1,205 724 None-=======m=m==me=u- 1,361 1,308
Coronary heart Hypertensive heart
disease disease
Definite--====-=--=- 917 337 Definite-======--- 917 623
SUSPECL ~~ === mem 623 178 Suspect=-========- 337 178
Other specified Other specified
heart disease------- 149 138 heart disease------ 30 47
Heart disease, other-| 7,605 3,382 Heart disease, other- 480 254
NOTE: The category '"Other specified NOTE: The category ''Other specified
heart disease' consists of rheumatic, con-
genital, or syphilitic heart disease.
Countsare exclusive.Categories are listed
in order of descending priority.
coronary heart disease 35 percent of the persons
had another specific heart disease and 11 percent
more had some other evidence of heart disease.
Furthermore, an appreciable number of cases
of hypertensive heart disease were manifest on
both the X-ray and the electrocardiogram (fully
21 percent of all definite cases and 11 percent of
all suspect cases). Multiple evidence of coronary
heart disease also was common.
In short, heart disease is very often a com-
‘plex, multifaceted disease entity, inadequately
displayed by the rubrics in current use.
Other Heart Disease
Some examination findings, while clearly in-
dicating heart disease, did not satisfy the cri-
teria for hypertensive, coronary, rheumatic,
congenital, or syphilitic heart disease. These
were incorporated into a miscellaneous category
"Other heart disease'; those persons who had
such findings but none of the specified heart dis-
heart disease' consistsof rheumatic, con-
genital, or syphilitic heart disease.
Countsare exclusive. Categories are listed
in order of descending priority.
eases were so categorized. This category, then,
unlike the others, was used only if a person
could not be categorized as having heart dis-
ease on other grounds.
Two kinds of evidence were considered
indicators of definite "Other heart disease."
The first was aortic stenosis. The second were
certain electrocardiographic findings, the most
common of which were left bundle branch block
and atrial fibrillation. It was rare that either of
these indicators was found in persons who didnot
have heart disease defined on some other basis.
In fact, a total of only 271,000 adults was esti-
mated to have definite ''Other heart disease,"
most of these on the basis of the electrocardio-
gram.
Suspect "Other heart disease' was diagnosed
if none of the specified heart diseases were diag-
nosed and definite "Other heart disease'' was not
present but if heart enlargement was noted on the
X-ray. Heart enlargement without a defined
etiology was very common: it is estimated that
11
6,910,000 adults had this finding using the Survey
standards. Electrocardiographic findings indi-
cating suspect ''Other heart disease' were much
less common. Most cases diagnosed on these
grounds had either right bundle branch block or
left ventricular ischemia, with the cases being
evenly divided between these two categories. As
with other findings included in "Other heart dis-
ease,’ left ventricular ischemia was much more
commonly found with other evidence of specific
heart disease than it was as an isolated finding.
Heart Findings
Up to this point the discussion has focused
on persons with cardiac findings satisfying the
Table F. Number of adults with specified
cardiac findings but without diagnosed
heart disease: United States, 1960-62
Number of
adults in
thousands
Electrocardiographic findings
Myocardial infarction
outside of criteria-=-======-- 163
Left ventricular hypertrophy-- 2,644
Subendocardial ischemial?2----- 567
Nonspecific T-wave
abnormalities! 3-cemcccacaaaa 1,857
Incomplete right bundle
branch block or I-V block=---- 503
Tachycardia or abnormal
nodal rhythm--=-=--ccceccaca- 185
Miscellaneous abnormalities
of the Q or P waves---===-=---- 383
Left axis deviation with
specified history-----ceue-- 793
First degree A-V block with
specified history’-==-=-==--=--- 161
Physical examination findings
Significant systolic murmur--- 3,476
Tnside or outside criteria.
With or without digitalis effect.
3Includes left ventricular ischemia
outside criteria.
4Chest pain, heart pain, high blood
pressure, or heart trouble.
Rheumatic fever, chorea, high blood
pressure, or heart trouble.
NOTE: Counts are not exclusive.
12
diagnostic criteria of this Survey. Between such
persons and persons who clearly and certainly
gave no evidence of heart disease was a group of
persons with possibly serious but nondiagnostic
heart findings. If they were under a physician's
care it is probable that some of these would be
designated as having heart disease. At the very
least, they would be reexamined at regular inter-
vals or submitted to additional diagnostic tests.
Two kinds of findings may be mentioned
(table F). The first was a miscellaneous set of
electrocardiographic findings. These range from
nonspecific T-wave abnormalities to electrocar-
diographic tracings which fall just short of the
rather severe Survey criteria for myocardial
infarction. Fully 6.4 percent of all adults had
such findings in the absence of diagnosed heart
disease. The second was a significant systolic
murmur, which another 2.8 percent had. Alto-
gether 9.2 percent of all persons had at least one
of these findings but were not diagnosed as having
heart disease.
SUMMARY
There were about 14 million adults in the
United States with definite heart disease and
nearly the same number with suspect heart dis-
ease. The most common form of heart disease
was hypertensive.
Definite heart disease was more frequent
among women, and suspect heart disease was
more frequent among men.
Hypertensive heart disease was more com-
mon in Negro than in white adults.
The prevalence of heart disease rose steeply
with age. This age trend was evident for hyper-
tensive, coronary, and rheumatic heart disease,
although the rate of rise varied with the diag-
nosis. Altogether, less than 2 percent of persons
in the age group 18-24 years had definite heart
disease, while 39 percent of all men aged 75-79
years and 46 percent of all women in this age
group had definite heart disease.
A large number of persons with heart dis-
ease had more than one manifestation of the dis-
ease. A large number of persons without diag-
nosed heart disease had possibly serious cardiac
findings.
REFERENCES
ys. National Health Survey: Plan and initial program of the
Health Examination Survey. Health Statistics. PHS Pub. No. 584-
A4. Public Health Service. Washington. U.S. Government Printing
Office, May 1962.
National Center for Health Statistics: Cycle I of the Health
Examination Survey, sample and response. Vital and Health Statis-
tics. PHS Pub. No. 1000-Series 11-No. 1. Public Health Service.
Washington. U.S. Government Printing Office, Apr. 1964.
3New York Heart Association: Nomenclature and Criteria for
Diagnosis of Diseases of the Heart and Blood Vessels. New York
Heart Association, 1955.
“Nasional Center for Health Statistics: Blood pressure of
adults'by age and sex. Vital and Health Statistics. PHS Pub. No.
1000-Series 11-No. 4. Public Health Service. Washington. U.S.
Govern ment Printing Office, June 1964.
3 tional Center for Health Statistics: Blood pressure of
adults by race and area. Vital and Health Statistics. PHS Pub. No.
000
1000-Series 11-No. 5. Public Health Service. Washington. U.S.
Government Printing Office, July 1964.
6
U.S. National Health Survey: Evaluation of a single-visit
cardiovascular examination. Health Statistics. PHS Pub. No. 584-
D7. Public Health Service. Washington. U.S. Government Printing
Office, Dec. 1961.
7
National Heart Institute: Reportofthe Conference on Longi-
tudinal Cardiovascular Studies. Bethesda, Md., 1957.
Bp ollnck, H., and Kreuger, D. E., eds.: Epidemiology of Cardi-
ovascular diseases—hypertension and arteriosclerosis. Supplement
to Am. |. Pub. Health, Vol. 50, No. 10, 1960.
Hilbish, T. F., and Morgan, R. H.: Cardiac mensuration by
roentgenologic methods. Am. J. M. Sc. 224(5):586-596, Nov. 1952.
13
DETAILED TABLES
Page
Table 1. Prevalence of definite and suspect heart disease in men and women, by age: United
States, 1960-62----=--m-mmmee meee m mem mmeeeeeeee———ao 14
2. Prevalence of definite and suspect heart disease for white and Negro adults, by
age and sex: United States, 1960-62-------rereccrcc cece; ——— ce; ————————— 15
3. Prevalence of definite and suspect hypertensive heart disease for white and Negro
adults, by age and sex: United States, 1960-62-------emreecccmcccc cece —————— 15
4. Prevalence of definite and suspect coronary heart disease for white and Negro
adults, by age and sex: United States, 1960-62 mmmmmmmmmmm mmm mmm mmm mms wm = mm mm 16
5. Prevalence of rheumatic heart disease in men and women, by age: United States,
1960=62======= cm meme meee eee meee meee ecemecee—ee--- 16
Table 1. Prevalence of definite and
suspect heart disease in men and women, by age: United States,
1960-62
Definite heart disease Suspect heart disease
Age
Both Both
Genes Men Women foxes Men Women
Percent of specified age-sex group
Total-18-79 years-======m=== 13.2 12.6 13.7 Tl:7 13:9 9.4
18-24 yearg-=-==rmmmmennmmremann—— 1.2 Li 1.1 4.0 6.4 2.0
25-34 years-mmmmrememssmmemmmn——-— 2.4 2.9 2.0 4.9 6.6 3.3
3S lils JODY Grwmimimiminim' mio ve mm min mmm 0% 6.7 7.4 Gl 8.8 11.4 6.4
45-54 yearg---==mrmmmmmmmmemem————— 13.2 13.8 12.5 15.3 18.3 12.4
55-64 years-------mmmmmceemeee———— 25.3 24.2 26.2 19.4 18.5 20.1
65-74 years-----------mmmmccea———- 39.9 33.2 45.2 20.7 25.3 17:1
75-79 years~====mmermmmemnmmnm——-——— 42.3 38.8 45.8 25.2 27.1 23.3
14
Table 2. Prevalence of definite and suspect heart disease for white and Negro adults, by age and
sex: United States, 1960-62
Age
Definite heart disease
Total~18-79 years=====rmerommmmmm mmm.
18-24 yearg-===mmmmmmmmmmme meen enn mmm ——————
25-34 years--
35-44 years--
45-50 JEAU Smmmmmm mmm mmm mm nm on on mm nn nm
55-64 years=-==mmmmmmmmemmmmmee mmm mee mm ———————————
65-74 YearS-mmmmmmmmme mmm mmm ———————————————
75-79 YORE == sn ee Se NS we i Se i SSR we Se
Suspect heart disease
Total-18-79 years--==m-mmm=rereecee mec e emcee ———————
Men Women
White Negro
Percent of specified population group
11.5 12.5 24.8
1.4 «9 0.8 3.2
2.3 .9 1.4 6.8
6.1 .1 4.9 14.0
1.3 +0 9.6 36.6
Le .6 23.7 52,2
31.3 .9 43.5 70.1
39.3 +3 44.8 69.5
13.5 +5 9.3 12.6
6.3 ny a2 8.3
5.5 «9 2.6 7:8
10.6 «7 5.4 13.0
18.4 +2 11.8 14.8
17.6 +2 20.3 20.3
26.4 9 17.3 16.2
25.3 .3 23.4 14.2
Table 3. Prevalence of definite and suspect hypertensive heart disease for white and Negro adults,
by age and sex: United States, 1960-62
Age
Definite hypertensive heart disease
Total-18-79 years-=-==-=mmemememeeeeeeee cece ee —————
18-24
25-34
35-44
45-54
55-64
65-74
15-79
Suspect hypertensive heart disease
Total-18-79 years=====s---cecccccccccccccc cence
18-24
25-34
35-44
45-54
55-64
65-74
75-79
Men Women
White Negro
Percent of specified population group
6.5 5.8 22 0
0.2 1.9 - 1.6
1.1 5.2 0.7 4.7
4.0 15.2 2.7 14.0
7.7 24.4 6.8 31.5
11.7 33.1 19.5" 46.4
16.3 50.2 37.5 66.4
24.0 32.3 37.1 69.5
5.0 Sud 4.7
1.5 1,5 - -
1.2 7 0.7 -
4.0 6.2 0.8 3.6
4.3 10.5 3.4 5.9
7.3 13.8 8.5 15.0
13.8 - 8.4 10.3
15.7 21.4 10.7 14.2
15
Table 4. Prevalence of definite and suspect coronary heart disease for white and Negro adults, by
age and sex: United States, 1960-62
Men Women
Age
White Negro White Negro
Definite coronary heart disease Percent of specified population group
Total-18-79 year§---========-=-eemeeeccesoe=—e=a==- as 310 2.) 2.0
18720 JOR Gm re mr mm ow v6 0 0 2 0 tt - - - -
D530) YG Ts wt eros sew io mt 9.0010, 0 Sp fw 0.1 3.1 0.2 -
35-44 year§=--=-=emmmmmememm-se—sseseses ee ——————————— 1.2 - 0.4 1.0
45-54 years--===memmmmmmmme-mm memes esse ——————————————— 3.0 7.4 1.3 3.9
55m64 YRArS-====mmm mm mmm on 10.3 5.7 4.7 5.3
BT TL JOO Given im 0 51001 ht 0 ym 12.2 3.4 8.2 51
75m TD JEL ve wwii rm om ho a 9.8 - 5.1 -
Suspect coronary heart disease
TOA 18-79 YEATES ww mw mmm mim mmm or we ww om om 2 rw 20 2 iL Suk 2.2 2.2
18-24 yearg-========-m--e---cosmos-msoomoooeoooooooosssossos - - - -
25-34 year§==-==--m--mm==-----e--ee-eoe—eoooo---essosssso- - - 0.2 -
35-44 Joars~m=mmmisse wenn 1.0 3.5 0.3 0.9
45=54 year gr=mmmsnien www 3.5 2.8 2.4 4.1
SSL); FEI ir mm 0 00 mt 0 4.2 77 5.3 4.3
BST GOAL Grew mmo rm oF 1 4 20 a 5.1 7+5 6,2 9.0
75079 VRAIS == were om mmm mm mm 0 0 0 2 4.1 - 2.3 -
Table 5. Prevalence of rheumatic heart disease in men and women, by age: United States, 1960-62
Both
sexes Men Women
Age
Percent of specified
age-sex group
1.1 1.2 1.1
18-24 0.5 0.4 0.5
25-34 0.5 0.5 0.6
35-44 1.1 1.1 1.0
45-54 1.5 1.1 1.8
55-64 1.3 1.3 1.3
65-74 2.2 3.0 1.5
75-79 3.3 3.8 2.9
APPENDIX I.
MEDICAL HISTORY QUESTIONS RELATED TO CARDIOVASCULAR DISEASE
(Excerpts From HES-204, Medical History-Self Administered)
1. a. In the past few years have you had any headaches?
If YES b. How often? [Every few days | [Less often | Probes A,B
c. Do they bother you [quite a bit] [ust a little]
2. a. In the past few years have you had any nosebleeds?
If YES b. How often? [Every few days | [Less often | Probe A
c. Do they bother you [quite a bit | {just a little]
3. a. At any time over the past few years, have you ever noticed ringing
in your ears or have you been bothered by other funny noises
Probes A,B
In your ears?
If YES b. How often? { Every few days] [Less often]
c. Do they bother you [auite a bit | [just a little |
4. a. Have you ever had spells of dizziness? Price 1
robe
If YES b. How often? | Every few days | [ Less often |
c. Do they bother you [quite a bit | [just a little |
5. Have you ever fainted or blacked out?
6. a. Have you ever had a stroke?
If YES b. Have you had a stroke in the past 12 months?
c. Have you ever seen a doctor about it?
7. Has any part of your body ever been paralyzed?
9. Was there anytime in your life when you had a lot of bad sore
throats?
16. a. Have you ever been bothered by shortness of breath when climbing
stairs? Probes A,D
If YES b. How often? [ Almost everytime | [Less often |
c. Does it bother you | quite a bit just a little
17
18
Probe A
Probe A
Probes A,B
17.
18.
19.
20.
. Have you ever been bothered by shortness of breath when doing
physical work or exercising?
|f YES b. How often? ( Almost everytime | [ Less often |
c. Does it bother you quite a bit just a little
. Have you ever been bothered by shortness of breath when you were not
doing physical work or exercising?
If YES b. How often? [Every few days | [Less often]
c. Does it bother you [auite a bit | [just a little |
. Have you ever been bothered by shortness of breath when you are
excited or upset about something?
If YES b. How often? [ Almost everytime | [ Less often |
c. Does it bother you | quite a bit just a little
. Have you ever waked up at night because you were short of
breath?
If YES b. How often? [Every few nights | [ Less often |
c. Does it bother you | quite a bit just a little
In the past few years, have you ever had any pain, discomfort,
or tightness in your chest?
IF YES, please answer questions b through j below.
b. How often? [ Every few days | [ Less often |
c. Does it bother you [auite a bit | [ just a little
d. Where does it bother you? (Check every place it bothers you.)
To aie WadTe
Somewhere else State where
e. Does it usually [stay in one place] [move around | [7]
f. How long does the pain usually last?
{Just a few minutes | | Few minutes to an hour | [More than an hour |
a. Does it usually come [ When you take a lot of exercise | or
when you are quiet or
is there no difference
h. Does it usually come or
[ doesn't this make any difference |
ij. Do you take any pills or medicine for it?
22.
In the past few years, have you ever had any pain, discomfort,
or trouble in or around your heart?
IF YES, please answer questions b through j below.
b. How often? | Every few days | [Less often |
c. Does it bother you (quite a bit [ just a little
d. Where does it bother you? (Check every place it bothers you.)
[ Somewhere else | State where
e. Does it usually [ stay in one place | [ move around | [7]
f. How long does the pain usually last?
[ Just a few minutes | [ Few minutes to an hour | [More than an hour
a. Does it usually come [ When you take a lot of exercise or
when you are quiet or
is there no difference
h. Does it usually come when you are upset or
[ doesn't this make any difference |
23.
24.
25.
26.
62. a.
j. Do you take any pills or medicine for it?
. Sometimes, our hearts "act funny" (odd) like missing a beat,
or beating real fast, or seem to turn over. Have you ever
noticed your heart do anything like that? Probes A,B
If YES b. How often? [ Every few days | Less often |
c. Does it bother you [quite a bit | [Just a little |
. Have you ever been bothered by your heart beating hard?
If YES b. How often? [ Every few days] [Less often] Probes A,B
c. Does this bother you [ quite a bit | [Just a Tittle]
. Are your ankles ever swollen at bedtime?
Probe A
If YES b. Is the swelling gone by morning?
. When you walk, do you have pains or cramps in your legs?
If YES b. How often? [ Every few days] [Less often] Probe A
c. Does it bother you [quite a bit | [ just a little |
Has a doctor ever said you had rheumatic fever (inflammatory
rheumatism)
If YES b. Have you had it in the past 12 months?
c. Are you taking any pills or medicine for it?
If YES d. What is it?
19
63. Has a doctor ever said you had chorea or St. Vitus' Dance?
65. a. Has a doctor ever told you that you have hardening of the
If YES b. Have you had this condition in the past 12 months?
66. a. Have you ever had any reason tc think you may have high blood
Did a doctor tell you it was high blood
pressure?
. How long ago did you first start having it?
[1year] [1-5 years] [over 5 years |
. Have you had it in the past 12 months? [ves] [no] [2]
. Do you take any pills or medicine for it?
Give name of the medicine
67. a. Have you ever had any reason to think you may have heart
Probe C
arteries?
pressure?
if YES or 2 b.
c
Probe C
d
e
if YES fF.
trouble?
if YES or 7 b.
Probe C
Did a doctor tell you that you had heart
trouble? [ves] [no]
If YES, what did he call it?
. How long ago did you first start having it?
( 1 year | 1-5 years | [over 5 years |
d. Have you had it in the past 12 months? YES [ves][vo]f[=]
e. Do you take any pills or medicine for it? Des ol]
If YES f. Give name of the medicine
Probes: A. Do you have any idea what causes your 7
B. Tell me how it feels.
C. In what way does it bother or affect you?
D. How many flights?
These questions were used, where indicated, if the examinee
answered either
20
"yes" or npn
000
APPENDIX II.
FORMS USED IN RECORDING FINDINGS
ON THE PHYSICAL EXAMINATION
Confidentiality has been assured the individual as set forth in 22 FR 1687
PHS-3034 Health Examination Survey
REV.. 4=41 PHYSICAL EXAMINATION
HES-205
BLOOD PRESSURE - LEFT ARM
DIASTOLIC 2
OCULAR FUNDI RIGHT | LEFT | REMARKS CODE
4. Normal
5. Fundus not Visualized
6. Globe Absent
7. Increased Light Reflex
8. Narrow Arterioles
9. Tortuous Arterioles
10. AV Compression
11. Hemorrhage
12. Exudate
13. Venous Engorgement
Ww. Papilledema
15. Disc Abnormal
16. Lens opacities
17. Iritis
18. Other (Specify)
19. kw erade [0] [1 [2] HD [E
741-678 O - 64 -3 _
EARS RIGHT | LEFT REMARKS CODE
20. Normal
21. Drum not Visualized
22. Malformation
23. Exudate
24. perforated Drum
25. Scarred Drum
NECK
26. Venous Engorgement (upright)
RA fo
PERIPHERAL ARTERIES = inspection and palpation
27. All Normal ]
RIGHT SIDE NORMAL SCLEROTIC | TORTUOUS NOT DONE® CODE
28. Superficial Temporal
29. Brachial
30. Radial
LEFT SIDE NORMAL SCLEROTIC | TORTUOUS NOT DONE® CODE
31. Superficial Temporal
32. Brachial
33. Radial
‘nor DONE (Specify which item number and why not done)
22
QUALITY OF ARTERIAL PULSATIONS
34. A11 Normal ]
RIGHT SIDE NORMAL BOUND ING | DIMINISHED PALPABLE NOT DONE CODE
35. Radial
36. Dorsalis Pedis
37. Post-tibial
LEFT SIDE noRMAL | Bounoing |oiMinisHED | %OT. | wor pone” ODE
38. Radial
39. Dorsalis Pedis
40. Post-tibial
LOWER EXTREMITIES RIGHT | LEFT REMARKS CODE
41. Normal
42. Not Done*
43. varicosities
44, Dependent Edema
45. Ulcers
-
NOT DONE (specify which item number and why not done)
HEART
46. Thrills None O
IF present, specify: Location
Timing
47. Aplcal Impulse Not Felt O
MCL [at or inside [outside |
Interspace [x] [5] [6]
48. Heart Sounds
Normal 0
Accentuated
A, [0]
r [1]
Mm []
Third Heart Sound Od
Diminished
O00
Splitting of second sound abnormal ]
other (Specify)
49. Murmurs If present, specify (in order): location, intensity (grades | through v), pitch,
quality, duration, timing, transmission, and whether significant or non-significant.
Systolic None
Diastolic None Ol
MUSCULOSKELETAL SYSTEM
50. Arthritis and Rheumatism wo positive Findings OJ
If positive findings are present, fill out Summary of Joint Involvement
on next page.
24
SUMMARY OF JOINT INVOLVEMENT
MANIFESTATIONS
Joints
Tender [Swelling|Deformity |Limitation other? Code
51. Shoulder
52. Elbow
53. Wrist
54. Metacarpo-
phalangeal
55. Proximal-
inter—
phalangeal
56. Distal-
inter-
phalangeal
57. Hip
58. Knee .
59. Ankle
60. Feet
Cervical
spine
61
62. Lumbar
spine
63. Other®
Record positive findings as R for right, L for left, RL for both, except for spine (Items 61 and
62) which should be check marked.
Fingers (1tems 54, 55, and 56): Record total number of joints involved on right or left.
lugther” manifestations include Heberden's nodes, subcutaneous nodules, ulnar deviation, pain on mo-
tion, heat, atrophy, and funnel fist.
“nother” joints include temporomandibular, sternoclavicular, sacroiliac, and specific joints of the
feet.
25
ADDITIONAL FINDINGS IN THE PHYSICAL EXAMINATION
none [|
CODE
64. Head
65. Neck
66. Chest
67. Extremities
Neuromuscular System
68. Gait
69. Cdordination
70. Strength
71. Tremor
IMPAIRMENTS
none [|
ETIOLOGY
Birth
Later Illness
or Injury
CODE
72. Cleft palate
73. Club foot
74. Paralysis (Specify site)
75. Missing digits (Specify)
76. Other (Specify)
77. Additional Remarks
26
EXAMINING PHYSICIAN'S IMPRESSION
Cardiovascular Diseases NEGATIVE
Hypertension .......... PRTC E REY ARRAS see nrnre ceri La)
Peripheral arteriosclerosis ...ceoe.... eA AER CRN Kner 0
Organic heart disease..cecriisriiecireannnns wor assenvren rdntrrernttes Tod
Angina pectoris
If positive or suspect,
Etiology
POSITIVE
a
O 00
SUSPECT
a
O 00
Anatomy
Physiology
Functional capacity
Other
Comments
Arthritis and Rheumatism
No arthritis [J
Classical arthritis (give specific diagnosis)
Definite arthritis
Rheumatic complaints
Questionable complaints
Other Diseases and Conditions
Signature
O00O0
APPENDIX III.
ELECTROCARDIOGRAPHIC READINGS
Criteria and Classification the cardiologists who read the ECG's. The draft version
of these criteria was submitted to cardiologists ex-
The following are the criteria and classifications perienced in reading electrocardiograms for survey
used in electrocardiographic (ECG) reading by the purposes, and their criticisms and suggestions were
Health Examination Survey. They were developed by taken into account in this working version.
Category Leads Impressions
1. Q & QS patterns (Q must be 1 mm. or more)
a. Q duration =\0.04 second or more I, II, V;-V, (any) Anterior myocardial
infarction
b. Q duration = 0.04 second or more AVL Anterior or lateral myo-
cardial infarction
c. QS pattern when R wave 1s present Vo-Vg (any) Anterior myocardial AAEEEL,
in adjacent precordial lead to infarction Srash
the right myocardial
infarction
d. QS pattern Vi-v (all) Anteroseptal myocardial
4
infarction
v,-V. (all) Anterior myocardial
1°'5
infarction
v,-V. (all) Anterolateral myocardial
1 °'6
infarction
e. Q duration = 0.05 second or more IIT Posterodiaphragmatic
and a Q wave in AVF myocardial infarction
Posterior
f. Q duration = 0.05 second or more AVF Posterodiaphragmatic myocardial
and R = +3 mm. or more myocardial infarction infarction
gs Q duration = 0.04 second II, III, and Posterodiaphragmatic
AVF (all) myocardial infarction
2. QRS axis deviation
a. QRS axis = -30° or more I, II, and III left axis deviation
b. QRS axis = +120° or more I, II, and III Right axis deviation
3. Ventricular preponderance (hypertrophy)
a. S (+) R= 35 mm. or more "gS" in vy or left ventricular
NOTE: Record associated ST- or V.. and "R" in hypertrophy
T-wave abnormalities 22
separately Vs or Ve
b. QRS duration less than 0.12 second Vy Right ventricular
and R = 5 mm. or more hypertrophy
and R/S = 1.0 or more
and transition zone (decreasing R/S)
left of V
2
4. ST junction and segment
(T-P interval is baseline)
a. ST junction depression 1 mm. or more I, II, AVL, AVF, Subendocardial ischemia
Vv, -Vg (any)
b. ST-J depression 0.5-0.9 mm. and I, II, AVL, AVF,
ST segment horizontal or downward Vv, -Yg (any)
Subendocardial ischemia and/or digitalis effect
¢. No ST-J depression as much as 0.5 mm. I, II, AVL, AVF,
but ST segment sloping down and Vy -Vg (any)
reaching 0.5 mm. or more below
baseline
28
4.
Category
ST junction and segment—Continued
d. ST segment elevation, any of
2 mm. or more
3 mm. or more
e. ST segment elevation and ST
contour upward (convex),
with elevation
2 mm. or more
3 mm. or more
f. ST segment elevation and
concave, with elevation
2 mm. or more
3 mm. or more
T wave
a. T = .5 mm. or more and QRS
mainly upright
b. T wave flat or small diphasic (+ 1 mm.)
and when QRS mainly upright
and R = + 5 mm. or more
ce. T =-1 to 5 mm.
when R = (+) 5 mm. or more
when QRS mainly upright
A-V conduction
a. Complete A-V block (permanent or
intermittent)
b. Partial (varying) A-V block
¢. P-R interval over 0.21 second
(any heart rate)
d. Accelerated conduction
Ventricular conduction
a. QRS duration 0.12 second or more
and R peak duration 0.06 second
or more (in absence of infarct
criteria, category 1, above)
b. R prime greater than R and QRS
duration over 0.12 second
c. R prime greater than R and QRS
duration not over 0.12 second
and not less than 0.10 second
d. QRS of 0.10 second or more, but
without LBBB or RBBB
Leads
1,11,III, V,,
Vg {any} °
v,-V, (any)
I, II, III, V,
52
Vg (any)
v,-V, (any)
I, IT, 111, Vg,
Vg (any)
LA (any)
I, II, III, AVL, AVF,
Vy-Vg (any)
I, II, V,-V, (eny)
AVL, AVF (either)
I, 1%, Avi, Vo-Vg (any)
AVL
AVF
Any
Any
I, II, III (any)
Any
I, II, III (any)
I, AVL, Vg, Vg (any)
I, II, III (any)
Impressions
Current of injury
Nonspecific T-wave abnormality
left ventricular ischemia
Complete heart block
Partial A-V block
First degree heart block
Wolff-Parkinson-White syndrome
Left bundle branch block
Right bundle branch block
Incomplete right bundle branch block
Intraventricular block
29
Category
8. Arrhythmias
a. 3 or more premature ventricular Any
contractions in sequence
be. Atrial fibrillation or flutter Any
c. Atrial (over 120/minute), nodal or Any
supraventricular (over 100/minute)
tachycardia
d. Nodal rhythm (up to 100/minute) Any
PR interval less than 0.11 second with
either a positive or negative P wave
or absent P or P following QRS
9. Low QRS, high T
a. Total R or S amplitude in
leads I plus II plus III
equals less than 15 mm.
b. T wave over 12 mm. Any
10. Premature beats and miscellaneous
a. Premature atrial, nodal, or Any
ventricular systoles
Rare (up to 3 in 40 complexes)
Frequent (4 or more in 40 complexes)
b. Miscellaneous items not mentioned elsewhere
1. QT interval >0.42, at any rate Any
2. P waves notched, or peaked (3 mm.), Any
or prolonged (>0.12 second)
3. Q duration of 0.03-0.04 second
(but not diagnostic of posterior
myocardial infarction)
X, IZ, IIT (all)
leads Impression
Ventricular tachycardia
Atrial fibrillation or flutter
Atrial, nodal, or supraventricular tachycardia
Nodal rhythm
Low QRS voltage
High T voltage
Premature atrial, nodal, or ventricular
systoles
Prolonged QT
P-wave abnormality
III and AVF (both) Other Q-wave abnormality
NOTE: In each category the ECG readers were allowed to designate abnormalities outside of criteria. For some
categories such findings were fairly common.
The general ECG reading procedure is described
in the main body of this report.
Three exceptions to this procedure were accepted.
(1) When a case was reviewed the full documentation was
considered. If the ECG was found to have an abnormality
which had been overlooked in the routine reading, this
abnormality was taken into account in the diagnosis;
similarly ECG readings that were found not to meet
the criteria were discounted on review. This led to
very few changes. (2) All cases of MI outside criteria
were reviewed by Dr. Abraham Kagan of the Framing-
ham Heart Study. One was found to meet the criteria
and the diagnosis was changed accordingly. A number
of other cases were found to nearly meet the criteria.
In ordinary usage they would be considered diagnostic
of MI but it was decided not to alter the criteria to in-
clude them. (3) The voltage criteria used in the finding
of LVH (S in Vv, or v, plus R in A or Ver whichever
is greater) made it possible to obtain this finding by
having clerks measure the ECG's. S in vi and R in
Vg were measured on all ECG's. It was found on the
30
basis of a sample of electrocardiograms that the S
wave was almost always greater in lead vy than lead
Vv, and the R wave was almost always greater in lead
Vs than lead Ver so measurements were confined to
leads V, and V_. If their sum was 35 mm. or more
and the person was 35 years or older, this was considered
evidence of LVH for purposes of diagnosing hypertensive
heart disease. A review of a sample of these cases in-
dicated that the measurement was sometimes in error
but it was assumed that other ECG's were under-
measured and hence that there was a counterbalancing
error. The measurement added a fairly large number
of cases. Of persons 35 years or older having definite
or borderline hypertension, 111 had LVH by meas-
urement but not by the readings of the cardiologists.
All of these cases were automatically diagnosed as
having hypertensive heart disease. Actually in 70 cases
the ECG finding simply constituted supplementary
evidence of hypertensive heart disease since there
was also evidence of heart enlargement on the X-ray,
and in only 7 of these cases was the diagnosis changed
from suspect to definite hypertensive heart disease as
a consequence of the ECG measurement. In the re-
maining 41 cases, however, a new diagnosis of hy-
pertensive heart disease resulted—in 23 cases definite,
and in 18 suspect. The net effect of the ECG measure-
ment was to raise the prevalence of hypertensive heart
disease by approximately 9 percent.
The distributions of LVH findings by the readers
for persons 35 years and over against the combined
sum of the S in vy and the R in Vs were as follow:
Number of electrocardiograms
Voltage
sig +R vp I LVH No LVH
finding finding
Under 35 mm=----- 3,903 45 3,858
35 mm---=======-= 62 8 54
36 mm--======-=-=- 53 11 42
37 mm--=======-= 37 11 26
38 mm--========= 35 13 22
39 mm---======--= 33 16 17
40 mm--=--=-===== 22 10 12
41 mm---======== 22 13 9
42 mm---======== 27 22 5
43 mm--======--- 18 15 3
44 mm--===——=--- 14 13 1
45+ mM----====== 80 75 5
This table includes all sample persons, whatever their
blood pressure.
If a person had normal blood pressure, no account
was taken in this report of discrepancies between the
electrocardiographic readings and the measurements
for LVH. Had this been done, the number of persons
considered to have had significant but nondiagnostic
cardiac findings would have been increased by about
10 percent.
The level of agreement between readers in desig-
nating major electrocardiographic findings was gener-
ally very high. Some examples are given below. Need-
less to say, agreement is no assurance of validity, LVH
being a case in point. For most findings, however, it
seems reasonable to assume that relatively few cases
were missed in the ECG reading.
Number of readers
agreeing with final
determination on
Final determination a tien
Total 3 2 3:
Myocardial infarction!- 100 67 | 13 20
Left ventricular
hypertrophy==-======== 397 342 | 29 26
Right ventricular .
hypertrophyl=====--=-- 7 5 2 -
Subendocardial
ischemia li? ===caconoe-- 135 102 23 10
Nonspecific T wave! ---- 207 147| 39 21
Left ventricular
ischemial-===-c-oceu-- 83 67 5 11
Left bundle branch
block====emcceen mean 25 25 - -
Right bundle branch
block======eemem cme 29 26 2 |
I-V block======mmenme=- 50 26 10 14
Atrial fibrillation---- 20 20 - -
Abnormal nodal rhythm-- 14 ii 1 2
!Inside or outside criteria.
With or without digitalis effect.
Some of these categories are fairly broad and if
they were broken into their specific components the
level of agreement would be less than indicated here.
For example, all three readers might agree that the
electrocardiogram showed evidence of a myocardial
infarction but disagree on the location of the infarct
or on whether the finding was inside or outside the
criteria.
In addition, there were instances where one or
more of the readers reported a finding which was not
agreed to in the final review. The number of such cases
of ''false positives'' was as follows:
Myocardial infarction--------- 25
Left ventricular hypertrophy--- 33
Right ventricular hypertrophy-- 1
Subendocardial ischemia ------ 46
Nonspecific T wave----------- 41
Left ventricular ischemia----- 28
Left bundle branch block------ 3
Right bundle branch block----- 6
I-V block---==-ccmcmcccneaao 19
Atrial fibrillation------------- -
Abnormal nodal rhythm------- 1
31
ECG Code Sheet
PH>-3762 ECG CODE SHEET (Clinical)
REV. 12-61 HES-212 Deck 30
CASE NUMBER (1-5) READER (6)
RATE (7-9) PR (10-11) QRS (12-13)
CODE: 1--Abnormal
2--Abnormal--Outside criteria
X--All normal
Y--Unsatisfactory ECG p Column 14
9--No ECG
14 15 16 17 18 19
Ant Post LAD RAD LVH RVH
MI MI
20 21 22
ST
Sor]
Sub. Sub. Isch. / Current of
Isch. digitalis Injury
23 24
T
Wave
Non-Specific LV Isch.
25 26 27 28
AV Cond.
Complete Partial 1st degree WPW
Block Block Block
29 30 31 32
Vent.
Cond.
LBBB RBBB Inc. I-v
RBBB Block
33 34 35 36 37
Arrhyth-
mias
Vent. Aur. Aur., Nod., Vent. Nodal
Tach. Fib. Supra-Vent. Rhythm Rythm
Tach.
38 39 40 41 42
(Circle one)
Low QRS High T Rare Frequent Al V2 N3
Premature Systole
43 REMARKS
Misc.
000
APPENDIX IV.
INTERPRETATION OF CHEST X-RAY
Form Used in Pulmonary
Reading
PHS:3739
4-61
NATIONAL HEALTH SURVEY CHEST
X-RAY INTERPRETATION
X-RAY NUMBER
READER
DATE CHECK HERE IF
FILM IS UNSATIS-
FACTORY Od
NONE
[]
PULMONARY PATHOLOGY
EXISTENCE OF LESION (Check one)
[] Definite [] Indefinite
IF LESION EXISTS, STATE MOST LIKELY ETIOLOGY
CARDIOVASCULAR PAT
QLCGY
NONE
LJ
HEART ENLARGEMENT (Check one)
[] Definite [] Borderline
IF OTHER CVD, PLEASE SPECIFY
OTHER CVD (Check one)
[] Definite [] Borderline
PLEASE SPECIFY BELOW ANY OTHER SIGNIFICANT PATHOLOGY
Instructions for intevpreting cavdiovascular pathology
Heart enlargement: Borderline enlargement is defined as 10 to 20 percent larger
than normal. If enlargement was not generalizeds
pecify the hypertrophied chamber.
Other cardiovascular pathology is to be specified as follows: Calcification of the
ascending aorta, calcification eof the aortic knob, calcification of other portions of
the aorta, abnormality of shape ofaorta (specify),
increased pulmonary vascularity.
33
Form Used in Cardiovascular Reading
1-5
Record Number
CHEST X-RAY (CV)
HEART OTHER
6 7 8 9 10 11% 12% 13%
GCE LVH AH RVH Other Pulmonary Position Calcifica-
Contour artery tion other
than aortic
. CODES
- 1
SORTA 1-Abnorma
14 15 16 A7 18 2-Doubt ful
X-Normal for entire row
Asc Arch Desc Calcified Other
Tortuous
F*Description
34
Pulmonary Readers
Initially, the X-ray films were interpreted by
three radiologists with a special interest in pulmonary
disease. While their primary concern was withevidence
of pulmonary pathology, abnormalities of the heart or
vessels were also noted. So far as the diagnosis of
heart disease was concerned, the two findings of special
concern were those of generalized cardiac enlargement
(GCE) and those of chamber enlargement, especially
left ventricular hypertrophy (LVH). Borderline GCE
was defined as present if the heartwas 10 to 20 percent
larger than normal; larger hearts were considered to
have definite GCE. No criteria were given for LVH.
In the following discussion a reading is considered
positive if a finding of GCE or LVH, definite or border-
line, was made. In order to determine how frequently
a film with evidence of GCE was missed by the
pulmonary readers, a series of 190 films were measured
by the method of Hilbish and Morgan, ? and the heart
size as measured was compared with the findings of
the pulmonary readers.
H t
hee Suber Reader | Reader | Reader
of normal) films 1 2 3
Number of positive
readings
Total=-=-=-=--- 190 56 42 22
Under 105------ 134 17 10 1
105-109-------- 20 10 9 3
110-114=-=====- 9 6 3 1
115-119-======- 6 5 5 2
1204-=====-=-- 14 13 13 11
Could not
measure=------- 7 5 2 4
Hawes, radiologist for the Framingham Heart Program,
was chosen. In other words, Framingham practice in
X-ray reading was the standard chosen.
Dr. Hawes was given a set of 192 Survey films which
had been selected to include a high proportion of posi-
tives. He found 96 of these ''positive'’; the number of
positive readings by the three pulmonary readers were
56, 42, and 22, respectively. Thus, even the two highest
counts were substantially below the level of readings
by Dr. Hawes.
Cardiovascular Readers
It was evident that to make the cardiovascular
findings of the Survey comparable with those of the
Framingham Heart Program another group of readers
would have to be used to read the X-ray films for
cardiovascular abnormalities. It was felt that training
radiologists to conform to standards was beyond the
resources of the Survey; it was decided, instead, to
choose radiologists who conformed naturally and with-
out instruction to Dr. Hawes' standards.
A series of radiologists were asked to read the
standard set of films. The four who conformed most
closely to Dr. Hawes' readings compared with him
as follows:
Reading by Reader
Hawes Other reader A B Cc D
There were 29 films found to be 10 percent or more
enlarged on measurement. Reader 1 read 24 of these
as positive, reader 2 read 21 as positive, and reader 3
read 14 as positive. The findings of readers 1 and 2
were consistent with the criteria for GCE. Reader 3
seemed to be following a different rule, generally
recording enlargement when it was 20 percent or greater
but seldom if it was 10-19 percent. The positive
findings reported for the smaller hearts are not incon-
sistent with the rules, since the films may have exhibited
abnormalities of shape indicative of cardiac hyper-
trophy.
Next, it was determined in what way, if any, the
cardiovascular readings of the pulmonary readers
differed from readings by radiologists who specialize
in cardiovascular reading. To answer this it was
necessary to obtain a set of cardiovascular reading
standards, or, in more concrete terms, to have a set
of films read by a standard radiologist. Dr. Lloyd E.
Number of
films
Total-=====ceeceanaanx 185] 192] 192 | 183
Agreement
positive positive 68| 84| 73 54
negative negative 64| 56 | 80 86
Disagreement
positive negative 24 12 23 37
negative positive 29| 40 16 6
In terms of reading levels the four readers read
the following percentage of films as positive.
Dr. Hawes ---==mccmmmcemuaa 50.0
Reader A ----c-mecmcmmeenae 52.4
Reader B-----memmmcmcceaen 64.6
Reader C----mmmmmmcmmaeao 46.4
Reader D---memmcmmcmmeee oe 32.8
(Although it later turned out that reader D could not
participate in the cardiovascular readings, his readings
on the standard films are included in some of the sub-
sequent analysis.)
35
The procedure used in the cardiovascular readings
allowed for a distinction between generalized heart
enlargement and left ventricular hypertrophy and for a
designation of findings as abnormal or doubtful. These
distinctions were ignored in the final determinations
because the readers clearly had no common standards
for such details. This is shown in the following tables.
Percent of positive findings designated doubtful:
Dr. Hawes --=-=-ccmmecmmunn 32.3
Reader A--------mmeceemua- 18.6
Reader B----=ccccmccommnaan 1.6
Reader C---=--mmmmemceenm- 44.7
Reader D------cemcmecmana- 15.0
Percent of positive findings designated as gener-
alized enlargement:
Dr. Hawes ---=----ommemmum- 41.7
Reader A -----comemmemmanmn 14.4
Reader B-----=emcemmceena- 38.7
Reader C---cmcmmmmccmcmna-n 55.9
Reader D------cmemmcmcmeae 6.7
Since the cardiovascular reading was to proceed
without training the readers or reconciling their dif-
ferences, it was felt advisable to assimilate all positive
findings to one class. In the case of one reader (reader
C), possible findings were actually assimilated to
negative, since the threshold between possible and
definite in his case seemed to correspond to the
threshold between negative and possible for the other
cardiovascular readers.
Final Evaluation
The procedure adopted for using both the pulmonary
and the cardiovascular readings to arrive at a final
evaluation of heart abnormalities on the X-ray was
essentially ad hoc but can be justified by both the
standardization experience and the Survey findings.
The readings made during the standardization process
were used only as an aid in selecting readers. The
films were re-read routinely for their final evaluation.
The evaluation technique adopted has been de-
scribed in the text. The combination of possible findings
by the pulmonary and cardiovascular readers is sum-
marized:
Final evaluation code
Cardiovascular
Tn
Pulmonary reader ender
——————— Negative | Negative | Negative
—————— Positive | Positive | Positive . ve
— Positive | Positive | Negative —
—————— Positive | Negative | Positive ve
or or
negative | positive
——————— Positive | Negative | Negative | Positive
or or
negative | positive
————— Negative | Negative | Positive | Positive
——————— Positive | Negative | Negative | Negative
or or
negative | positive
m—————— Negative | Negative | Positive | Negative
NOTE: Codes 1-5 are considered positive, all others negative.
36
There were 183 films which were interpreted by
Dr. Hawes and readers A, B, and D. The distribution
of films according to the findings of these four readers
and the final evaluation code is shown:
Number of films according
to the number of positive
Final evaluation initial readings
code
Total || 0 1 2 |3 4
Total======c-=u- 183 ||46 | 31 | 30| 31 | 45
mmmmmmmmmm mm —————— 63 13415 7| 5 2
i i 32) -| -| 2| 5 25
mmm if -| -| -| - 1
mmmmm mmm ——————— 23 1| - 3] 10 9
meme mm——————— 5] - 1 1 - 3
mm 31 2 71 9 9 4
OB 1 - Ty = wo
mmemmmemmm—————————— 271 9 7] 8 2 1
(Of the four only A and B subsequently engaged in
routine reading for the Survey.) There were 92 films
with positive codes 1-5. The average number of positive
readings by the four readers (A, B, C, and D) was 91.
Dr. Hawes found 91 films positive. All three counts were
practically the same.
The preceding table can be summarized in terms of
the percentage of the initial readings positive for each
code.
Final evaluation Percent
code positive
Total-===mmcemaaaan 49.7
[tt 20.6
J EE EEE 93.0
BJ i= esos mm *
dem mm mmm 78.3
Fh io To i *
Sm mmm eee 54.8
6 -— J
7 mmm mmm meme eee 30.6
One final piece of evidence may be considered. It
is well recognized that heart enlargement—whether
generalized or confined to the left ventricle—is highly
correlated with blood pressure. The following table
shows the percentage of films coded to each of the
specified codes which came from persons having
hypertension.
Nias Percent with
Final evaluation ber hypertension
code of
films Border-
Definite line
Om 4,461 9.1 11.4
A 272 54.0 18.8
2emmmmmm meme 17 * *
Brennen nnn 343 33.8 22.2
fommmmm mmm meen 61 16.4 24.6
Ee a i 506 35.8 20.0
Br ees mm oo 73 11.0 19.2
mmm ————————————— 661 19.4 22,1
Missing -——————————— 278 6.1 5.4
Since both heart enlargement and hypertension be-
come more common with age these percentages exagger-
ate the correlation between the two findings. Nonethe-
less, they do generally tend to support the evaluation
procedure used.
A comment is in order with respect to the "missing"
films. Some 278 examinees had no X-ray or, in a few
instances, had a film taken which was too poor to be
interpreted. The large majority of these persons were
women of childbearing age. It was the Survey policy not
to X-ray a woman where there was evidence suggesting
pregnancy. Persons with missing films were distributed
by age and sex as follows:
Number
Men ==-cmm mmm eee 34
Women ==== ===emeemmcec mc c—————— 244
18-24 years =====-=cemmmmmaeaa- 102
25-34 years -=======cmcmcmeaaa- 80
35-44 years ~==--cmcmmmccemeaa- 41
45-79 years ====--mmcmmmmceaae 21
37
The missing films were treated as negative in this re-
port. Judging from the small number of persons with
missing X-rays who had hypertension (15 definite,
17 borderline) this decision seems reasonable. It is
unlikely that treating the missing X-rays as negative
resulted in an appreciable understatement of heart
disease prevalence. Some of these persons were diag-
nosed as having heart disease even without the evidence
of the X-ray, but even if this were not the case there
would seem to be no alternative to the procedure chosen.
Finally, some note should be made of the unusual
nature of the X-ray evaluation procedure. The use of
a screening procedure which picks up all suspicious
findings initially and then, at a second stage of evaluation,
applies more stringent rules to the cases selected is
not uncommon. The Survey procedure was the reverse.
The initial (pulmonary) screening was the more con-
servative, the final (cardiovascular) reading the less
conservative. Actually the contrast between the two
readings is greater than appears from the standard
38
000
films. Since these films included an unusually large
proportion of very large hearts, there would be more
agreement on them than on a purely random sample of
the population. The contrast for the Survey films as a
whole was much greater, the cardiovascular readers
finding 27.9 percent positive on their initial reading, the
pulmonary 8.2 percent.
Why, then, were the pulmonary readings used?
There were three reasons. First, they were already
largely available at the time the cardiovascular stand-
ards were finally chosen. Second, they were relevant;
clearly heart enlargement found on the pulmonary
readings was meaningful in terms of the cardiovascular
standards. Third, it was possible by using them to
devise a more economical and secure cardiovascular
reading system than would otherwise have been possi-
ble. While it is not suggested that the procedure used
was the optimum one, it seems to have worked quite
satisfactorily.
APPENDIX V.
DIAGNOSTIC REVIEW
The procedure used in case review has been
described in the text. Briefly, every case was first
diagnosed by the computer. The key information was
then printed out and this machine record served as a
convenient summary of the case record, as well as a
place for entering decisions made in a subsequent re-
view, if there were such a review.
There were two important points at which the phy-
sician's judgment was seldom modified by review.
These were the diagnosis of angina pectoris and the
evaluation of a murmur. Findings of angina pectoris
were reviewed, chiefly to verify the coding of the phy-
sician's judgment. The description of a significant
murmur was reviewed to see if it was consistent with
the physician's evaluation of it.
The question arises as to what was done when a
physician's findings at these points appeared todiverge
from the usual. This became a serious question on
two occasions.
At one stand both examining physicians reported an
unusually large number of cases of angina pectoris.
Since both physicians had conducted examinations at
other stands and at these had found an average amount
of angina pectoris, it was felt that their judgments had
to be accepted where they found an unusual amount.
Their descriptions of angina pectoris were reviewed
and where the wording indicated less certainty than
appeared in the coded diagnosis, the coding was altered
to conform. It is likely that the review of cases from
this stand was more critical than usual, butin principle
it was the same as the review of similar cases from
other stands. Having admitted most of the cases from
this stand, it is nonetheless suspected that the chest
pain described for these cases was frequently not
due to coronary heart disease but arose from some
other cause.
The other set of unusual findings was a large
number of murmurs considered diagnostic of rheumatic
heart disease that was reported by one examining phy-
sician. On the average, slightly less than two cases of
rheumatic heart disease were reported by other phy-
sicians for each 160 persons examined. This physician
reported 19. Since he examined persons at two stands
and since the other examining physician at each of
these stands (a different one at each) reported only the
usual number of cases of rheumatic heart disease, the
prevalence of rheumatic heart disease among his ex-
aminees could hardly be attributed to the populations
examined. Three choices seemed open. (1) To select
a subsample of the rheumatic heart disease cases re-
ported by this physician, controlling to the usual
prevalence reported in the Survey. (2) To ignore the
population examined by this physician for the purpose
of computing rheumatic heart disease prevalence. (3)
To accept the cases as reported. Of the three alterna-
tives the first appeared the best and the third the
worst. In effect, a random selection of cases was
made by controlling to the usual prevalence, with a
probability of selection by age and sex proportional to
the distribution of rheumatic heart disease by age and
sex as reported by the other examining physicians.
This obviously was a choice among evils.
The review procedure was altered and became
more efficient as staff experience accumulated. In the
last seven stands reviewed, there were 1,116 cases.
Of these 181 were reviewed. In 23 cases a change was
made in the computer diagnosis as a result of the
review,
One set of review cases warrants special notice.
These are the cases in which the impression of the
examining physician was that heart disease was pres-
ent but the computer did not diagnose heart disease.
In the review of cases from the last seven stands 36 of
the 181 cases reviewed fell in this category. Inspection
of the case records revealed that the physician arrived
at his diagnostic impression in one of two ways. Either
he interpreted the electrocardiogram as abnormal when
the Survey readers did not, or he placed more diagnostic
weight on findings from the physical examination, such
as significant systolic murmurs, than the Survey cri-
teria allowed.
It would be misleading, however, to emphasize
the diagnostic ''misses' by the examining physician
and ignore the "hits." In fact, 80 percent of all cases
39
where the examining physician recorded a diagnostic
impression of definite heart disease and 59 percent of
all cases with a diagnostic impression of suspect heart
disease were ultimately diagnosed as heart disease.
Another 14 percent and 23 percent, respectively, were
found to have either significant electrocardiographic
abnormalities or a significant systolic murmur.
Finally, it ought to be noted that the physician's
diagnostic impression is not the same thing as a
final diagnosis. It was arrived at without having avail-
able the readings of the electrocardiogram and the
chest X-ray by the Survey's specialists and indeed
without an opportunity to inspect the X-ray itself.
Thus, there was a certain class of heart disease cases
which the examining physician could not identify in his
diagnostic impression; consequently the physician's
diagnostic impression led to a substantially smaller
count of heart disease than did the final Survey diag-
noses—458 definite and 545 suspect cases as against
855 and 745 cases, respectively.
CASE NC. 15010
AGE-RACE-SEX 73 MW
CIAGNCSIS FHC /2
CHE /2
MD IMPRESSION
He De. DFFINITF
A. P. OCFFINITF
AVERAGE BLOGD PRESSURE
EKG NORMAL
CHEST X-RAY
ENLARGEMENT YES©
HISTORY He. De. NO
PHYSICAL EXAM
THRILL N°
LAB. STS NORMAL
CIAGNOSTIC REVIEW FOR HEART DISEASE
1R6/109/102
ACRTIC AMEURYSM NO
HYP. YES R. Fa. NO
SIGNIFICANT MURMUR DIASTOLIC /0
HEART SOUND NOPMAL
VENOUS ENGORGEVENT
SYSTOLIC /0
000
40
-r
oy
APPENDIX VI.
STATISTICAL NOTES
The Survey Design
The Health Examination Survey is designed as a
highly stratified multistage sampling of the civilian,
noninstitutional population, aged 18-79 years, of the
conterminous United States. The first stage of the plan
is a sample of the 42 primary sampling units (PSU's)
from 1,900 geographic units into which the United
States has been divided. A PSU is a county, two or
three contiguous counties, or a standard metropolitan
statistical area. Later stages result in the random
selection of clusters of about four persons from a
small neighborhood within the PSU. The total sample
included 7,710 persons in the 42 PSU's in 29 different
States. The detailed structure of the design and the
conduct of the Survey have been described in previous
reports.! 2
Reliability in Probability Surveys
The methodological strength of the Survey derives
especially from its use of scientific probability sampling
techniques and of highly standardized and closely con-
trolled measurement processes. This does not imply
that statistics from the Survey are exact or without
error. Data presented are imperfect for three im-
portant reasons: (1) results are subject to sampling
error, (2) the actual conduct of a survey never agrees
perfectly with the design, and (3) the measurement
process itself is inexact, even when standardized and
controlled. The faithfulness with which the study design
was carried out has been analyzed in a previous report. 2
Of the total of 7,710 sample persons, 86 percent or
6,672 were examined. Analysis indicates that the ex-
amined persons are a highly representative sample of
the adult civilian, noninstitutional population of the
United States. Imputation for the nonrespondents was
accomplished by attributing to nonexamined persons
the characteristics of comparable examined persons.
The specific procedure used > consisted of inflating
the sampling weight for each examined person to com-
pensate for nonexamined sample persons at the same
stand and of the same age-sex group.
While it is impossible to be certain thatthe preva-
lence of heart disease was the same in the examined and
the nonexamined groups, the available evidence indi-
cates that it was. One source of information on this
question is a special inquiry sent to the physicians of
nonexamined persons and to the physicians of a match-
ing set of examined persons. The heart disease preva-
lence reported for the examined and for the nonex-
amined groups was in close agreement. For further
details on this subject see Vital and Health Statistics,
Series 11, No. 1.
Sampling and Measurement Error
In this report and its appendices, several refer-
ences have been made to efforts to evaluate both bias
and variability of the measurement techniques. The
probability design of the Survey makes possible the
calculation of sampling errors. Traditionally the role
of the sampling error has been the determination of
how imprecise the survey results may be because they
come from a sample rather than from measurement of
all elements in the universe.
The task of presenting sampling errors for a study
of the type of the Health Examination Surveyis compli-
cated by at least three factors. (1) Measurement error
and "'pure' sampling error are confounded in the data;
41
Table I. Standard error of estimated preva-
lence of specified heart disease for white and
Negro adults in specified age groups, by sex:
United States, 1960-62
Men Women
Age
White | Negro | White | Negro
Definite
hypertensive heart
disease
Percent of specified
population group
0.6
*
1.0
4.1
18-79 years----==-=--
25-34 years==------
45-54
75-79 years==-=-=----
«
0
©
=
®
1
1
|
1
1
1
1
1
=
oun ¥
or N
nO *w
~w oN
ON ¥N
Suspect
hypertensive heart
disease
Q
* ok AN
18-79 years--------
25-34 years--------
45-54 years--=----=-
75-79 years--------
~P=OOC
+ oF kW
Ea SRUSAV,)
it is not easy to find a procedure which will either
completely include both or treat one or the other sepa-
rately. (2) The survey design and estimation procedure
are complex and accordingly require computationally
involved techniques for calculation of variances. (3)
Thousands of statistics come from the survey, many for
subclasses of the population for which there are small
numbers of sample cases. Estimates of sampling error
are obtained from the sample data and are themselves
subject to sampling error, which may be large when the
number of cases in a cell is small, or even occasionally
when the number of cases is substantial.
In the present report, estimates of approximate
sampling variability for selected statistics are pre-
sented in tables I and II. These estimates have been
prepared by a replication technique which yields over-
all variability through observation of variability among
random subsamples of the total sample. The method
reflects both 'pure' sampling variance and a part of
measurement variance.
In accordance with usual practice, a 68 percentcon-
fidence interval may be considered the range within one
standard error of the tabulated statistic and a 95 per-
cent confidence interval the range within two standard
errors.
An overestimate of the standard error of a dif-
ference d=x-y of two statistics x and y is given
by the formula s4= [vi + vi] , where Vi and
v? are variances respectively of x and vy, orthe
42
squares of the standard errors shown in tables I and
II. For example, the prevalence of definite hypertensive
heart disease (HHD) is x=8.2 percent for white adults
and y=20.8 percent for Negro (table C), while from
table I variances are foundto be V? =0.16 percent and
3 = 5.29 percent. The formula yields the estimate of
the standard error of the difference (d=12.6 percent)
as sy= 2.33 percent. Thus, as the observed difference
is more than three times its sampling error, it can be
concluded that the prevalence of definite HHD is
higher among Negro adults than among white.
Small Numbers
In some tables magnitudes are shown for cells for
which sample size is so small that the sampling error
may be several times as great as the statistic itself.
Obviously in such instances the statistic has no mean-
ing in itself except to indicate that the true quantity is
small. Such numbers, if shown, have been included to
convey an impression of the overall story of the table.
Tests of Significance
As shown above, the difference in the prevalence of
definite HHD between Negro and white adults was sub-
mitted to a formal test of significance and found to be
significantly different from zero. This difference could
have been examined in other ways. It might have been
more meaningful, for instance, toask whether the preva-
lence for Negro adults was higher than (rather than "dif-
ferent from') the prevalence for white adults. There
Table II. Standard error of estimated preva-
lence of specified heart disease diagnoses for
white and Negro adults in specified age groups,
by sex: United States, 1960-62
Race Men Women
Percent of specified
Definite coronary
population group
heart disease
Whi be www mswmm 0.5 0.3
Negro -— 1.1 0.7
Suspect coronary
heart disease
White----==-ceccceen- 0.3 0.3
Negro - 0.6 0.3
Rheumatic heart
sease
Total======ecceccaaa- 0.3 0.3
is much evidence indicating this, and the test for a
one-sided hypothesis is more powerful than the test
for a two-sided hypothesis. Alternatively, the question
might have been, "Is the prevalence higher for Negro
adults than for white adults if age is held constant?"
Conceivably, the age-sex-specific means could be
identical for the two groups but a larger proportion of
older people in one group could lead to an overall
higher prevalence for that group.
This last version of the hypothesis can be tested
directly from table 3, with the use of a table for the
binomial variable. The prevalence of definite HHD is
higher for Negro adults in every age-sex group. The
chances of 14 heads out of 14 tosses of a true coin are
0.00006.
Demographic Terms
Age.--The age recorded for each person is the
age at last birthday.
Race.—Race is recorded as "White," "Negro,"
or "Other." "Other" includes American Indian, Chinese,
Japanese, and so forth. Mexican persons are included
with "White" unless definitely known to be Indian or
other nonwhite race.
43
U.S. GOVERNMENT PRINTING OFFICE : 1964 O - 741-678
REPORTS FROM THE NATIONAL CENTER FOR HEALTH STATISTICS
Public Health Service Publication No. 1000
Series 1. Programs and collection procedures
No. 1. Origin, Program, and Operation of the U.S. National Health Survey. 35 cents.
No. 2. Health Survey Procedure: Concepts, Questionnaire Development, and Definitions in the Health Interview Survey. 45 cents.
Series 2. Data evaluation and methods research
No. 1. Comparison of Two-Vision Testing Devices. 30 cents.
No. 2. Measurement of Personal Health Expenditures. 45 cents.
No. 3. The One-Hour Oral Glucose Tolerance Test. 30 cents.
No. 4. Comparison of Two Methods of Constructing Abridged Life Tables. 15 cents.
Series 3. Analytical studies
No. 1. The Change in Mortality Trend in the United States. 35 cents.
No. 2. Recent Mortality Trends in Chile. 30 cents.
Series 4. Documents and committee reports
No reports to date.
Series 10. Data From the Health Interview Survey
No. 1. Acute Conditions, Incidence and Associated Disability, United States, July 1961-June 1962. 40 cents.
No. 2. Family Income in Relation to Selected Health Characteristics, United States. 40 cents.
No. 3. Length of Convalescence After Surgery, United States, July 1960-June 1961. 35 cents.
No. 4. Disability Days, United States, July 1961-June 1962. 40 cents.
No. 5. Current Estimates From the Health Interview Survey, United States, July 1962-June 1963. 35 cents.
No. 6. Impairments Due to Injury, by Class and Type of Accident, United States, July 1959-June 1961. 25 cents.
No. 7. Disability Among Persons in the Labor Force, by Employment Status, United States, July 1961-June 1962. 40 cents.
No. 8. Types of Injuries, Incidence and Associated Disability, United States, July 1957-June 1961. 35 cents.
No. 9. Medical Care, Health Status, and Family Income, United States. 55 cents.
No. 10. Acute Conditions, Incidence and Associated Disability, United States, July 1962-June 1963. 45 cents.
No. 11. Health Insurance Coverage, United States, July 1962-June 1963.
No. 12. Bed Disability Among the Chronically Limited, United States, July 1957-June 1961.
Series 11. Data From the Health Examination Survey
No. 1. Cycle of the Health Examination Survey: Sample and Response, United States, 1960-1962. 30 cents.
No. 2. Glucose Tolerance of Adults, United States, 1960-1962. 25 cents.
No. 3. Binocular Visual Acuity of Adults, United States, 1960-1962. 25 cents.
No. 4. Blood Pressure of Adults, by Age and Sex, United States, 1960-1962. 35 cents.
5.
6.
No. Blood Pressure of Adults, by Race and Region, United States, 1960-1962. 25 cents.
No. Heart Disease in Adults, United States, 1960-1962.
Series 12. Data From the Health Records Survey
No reports to date.
Series 20. Data on mortality
No reports to date.
Series 21. Data on natality
No. 1. Natality Statistics Analysis, United States, 1962.
Series 22. Data on marriage and divorce
No reports to date.
Series 23. Data from the program of sample surveys related to vital records
No reports to date.
Catalog Card
U.S. National Center for Health Statistics.
Heart disease in adults, United States, 1960-1962. A description of the exami-
nation and diagnostic procedures with major findings by age, sex, and race. Washing-
ton, U.S. Department of Health, Education, and Welfare, Public Health Service, 1964.
43 p. diagrs., tables. 27cm. (Its Vital and health statistics, Series 11, no. 6)
U.S. Public Health Service. Publication no. 1000, Series 11, no. 6
1. Heart - Diseases - U.S. 2. U.S. - Statistics, Medical. I. Title. (Series. Series:
U.S. Public Health Service. Publication no. 1000, Series 11, no. 6)
Cataloged by Department of Health, Education, and Welfare Library.
PUBLIC HEALTH SERVICE PUBLICATION NO. 1000- SERIES 11 -NO. 6
U.C. BERKELEY LIBRARIES
(021330358
|