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