NCHSR waerarrataces Estimation and Sampling Procedures in the NMCES Insurance Surveys Instruments and Procedures 3 U.S. DEPOSITORY JUL 2 6 1984 U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Public Health Service Office of the Assistant Secretary for Health National Center for Health Services Research ''Abstract This report describes the design and weighting procedures developed for the insurance surveys of the National Medical Care Expenditures Study (NMCES), which produced national estimates of the insurance status and benefits of the civilian, noninstitutionalized U.S. population in 1977. Information on health insurance coverage obtained in the household survey was verified and supplemented in the Health Insurance/ Employer Survey (HIES) and two substudies, the Uninsured Validation Survey (UVS) and the Health Insurance Options Survey (HIOS). Insurance companies, employers, unions, and other organizations identified in the household survey as the source of private insurance coverage were asked to verify reported coverage and to provide information on each subscriber's coverage, its cost, the parties responsible for payment of the premiums, and the availability of alternative or optional plans. Lack of employment-related insurance coverage as reported in the household survey by employed adults was verified through their employers and if coverage was in fact available, premium and benefit information was obtained. Taking household-reported public coverage into account, information similar to that obtained for private policies was obtained for benefits offered by Medicare, CHAMPUS/ CHAMPVA, and the different state Medicaid programs. The HIES data base thus encompasses public as well as private health insurance benefits. The Employer Health Insurance Cost Survey (EHICS) was designed to provide national estimates for 1977 of the cost of health insurance to employers and of some characteristics of employees eligible for employment-related health insur- ance. Particular attention was given to a sampling strategy that allowed for the selection of employers as a conse- quence of their linkage to key participants in the NMCES household sample. A multiplicity framework for deriving national estimates of the health insurance costs of civilian employers and their employees in the U.S. for 1977 is described. In addition, detailed specifications are provided for the computation of the EHICS sampling weights. These procedures permit derivation of national estimates of the number of employees in firms with employment-related health insurance plans and those eligible for such plans, the proportion of payroll expenses going to health insurance, mean insurance premiums, the respective percentages of premiums paid by employers and employees, and the extent to which employers bear the entire premium cost. Glossary of Acronyms and Terms EHICS: HIES: HIOS: HIPF: NMCES: UVS: Employer Health Insurance Cost Survey Health Insurance Employer Survey Health Insurance Options Survey Health Insurance Permission Form(s) National Medical Care Expenditure Survey Uninsured Validation Survey Annual premium: The total payment required in 1977 to keep a health insurance policy in force. Fora family, this includes all premiums for all policies covering family mem- bers eligible for the NMCES survey. Employment-related insurance: For group health insurance, any coverage obtained through employers or labor unions, whether or not they pay all or part of the premium. Nongroup insurance is considered employment related only if an employer or union contributed to the premium. Primary insured: The person in whose name the insurance is issued or carried. Private health insurance: Any insurance for medical or related expenditures, but excluding “extra cash” coverage (small supplemental payments in the event of hospitali- zation); medical benefits linked to diseases such as stroke or cancer (dread disease’); and casualty benefits. Reporting unit: A reporting unit in this study is composed of individuals related by blood, marriage, or adoption. Any persons 17 years or older unrelated to any other person in the housing unit of the individual’s usual residence were considered to be separate reporting units, unless they were college students not living at home but covered through their family’s insurance policy. Size of group: The (largest) number of members, but exclud- ing insured spouses and children, enrolled in a person's group plan(s). Type of contract: This report distinguishes between group and nongroup coverage. Under a group policy specified benefits are provided to group members, and often their spouses and children, under a contract between the insurer and an organization, such as an employer, a labor union, or a voluntary association. A nongroup policy provides cover- age under a contract directly between the insurer and the insured individual or family obtaining the insurance. ¢¢ '' ~ ACHS National Health Care Expenditures Study Estimation and Sampling Procedures in the NMCES Insurance Surveys Instruments and Procedures 3 Steven B. Cohen Pamela J. Farley May 1984 U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Public Health Service Office of the Assistant Secretary tor Health National Center for Health Services Research '' 226/17 39 Pubes Authors Steven B. Cohen received a Ph.D. in biostatistics from the University of North Carolina at Chapel Hill. He is senior biostatistician at NCHSR's Division of Intramural Research and Asso- ciate Professorial Lecturer in the Department of Health Services Administration at George Washington University. Pamela J. Farley holds a Ph.D. in economics from Yale University. She is a Service Fellow in NCHSR's Division of Intramural Research. Acknowledgments The National Medical Care Expenditure Survey (NMCES) was a collaborative effort by many different people in a number of organizations. Much of the overall survey design is based on specification documents prepared by the NMCES data collection contractor, Resarch Triangle Institute, and its subcontractor, the National Opinion Research Center of the Univer- sity of Chicago. The sampling plan for the household survey was developed jointly by William D. Kalsbeek of Research Triangle Institute and Martin Frankel of the National Opinion Research Center. The advice in this respect of E. Earl Bryant and James Massey of the National Center for Health Statistics (NCHS) is gratefully acknowledged. Much of the material is based on survey methodology documents prepared by Ralph E. Folsom and Judith T. Lessler of the Research Triangle Institute. Ralph Folsom was responsible for the sample design of the Health Insurance Employer Survey and preliminary sampling weights specifications. Judith Lessler had major responsibility for the Employer Health Insur- ance Cost Survey sampling plan and weighting Strategy. Social and Scientific Systems, Inc. (SSS) of Bethesda, Maryland provided data processing support for sample weight construction. Leif Karell of SSS provided programming assistance in the construction of sampling weights and implementation of quality control checks. In addition, the authors wish to thank the following for their review of the manuscript and their helpful suggestions: William D. Kalsbeek of the Department of Biostatistics, University of North Carolina; Jai Choi and Robert Casady of the Office of Research and Methodology, NCHS; and Vicki Burt of NCHSR's Division of Intramural Research. We also thank John Carrick for his careful manuscript preparation. '' Contents at 10 14 16 19 20 21 22 25 27. 28 32 33 34 36 38 39 Introduction Survey Procedures Health Insurance/Employer Survey Data Collection Abstracting of Policy Information Assignment of Benefits to Dependents Employer Health Insurance Cost Survey The Health Insurance/Employer Survey HIES Sample Weights Eligibility Insurance Coverage Premiums for Multiple Coverage Benefits From Policy Abstracts Coverage Options in Employment-Related Groups HIOS Eligibility Availability of Optional Coverage Health Insurance Options Summary The Employer Health Insurance Cost Survey Sampling Frame Sample Weights Missing Employment Status Unidentified Employers Missing Multiplicity Missing EHICS Data Summary References Appendix A-1. Health Insurance Permission Form A-2. Unemployed Verification Survey Permission Form A-3. Unemployed Verification Survey Questionnaire A-4. Health Insurance Options Survey Questionnaire B. EHICS Questionnaire HDIV16e2 UY CY IO ey PUBL '' '' Introduction This report is the third in a series on instruments and sampling procedures in the National Medical Care Expenditure Survey (NMCES). It describes the questionnaires, sample design, and weighting procedures developed for two NMCES components, the Health Insurance/Employer Survey (HIES) and the Employer Health Insurance Cost Survey (EHICS). These two insurance surveys supplement and extend the data obtained in the NMCES household component, in which information on sociodemographic characteristics, health status, and use of services during 1977 was obtained from approximately 14,000 randomly selected households in the civilian noninstitu- tionalized population of the United States. (See Bonham and Corder 1981, and Cohen and Kalsbeek, 1981, Instruments and Procedures | and 2 in this series, for detailed information on the household survey.) The design of the National Medical Care Expenditure Survey was guided by the informa- tion requirements and analytic goals of the National Health Care Expenditures Study of the Division of Intramural Research, National Center for Health Services Research (NCHSR). These included an assessment of issues related to the cost and extent of health insurance in the U.S. population, in particular o The breadth and depth of health insurance coverage; o The loss of revenue resulting from current tax treatment of medical and health insurance expenses, particularly with regard to the benefits currently accruing to different categories of individuals and employers, and the potential effects on the federal budget of proposed changes to tax laws; The extent and availability of work-related insurance; The cost of insurance to employers and employees; The extent and depth of private insurance supplementing Medicare benefits; . The cost, utilization, and budgetary implications of changes in federal financing programs for health care and of alternatives to the present structure of private health insurance; o The extent and type of optional health insurance plans offered by employers to their employees. oo°0 0 The NMCES household survey sample was chosen as the basic sampling frame for the insurance surveys. Both insurance surveys and their design and sampling strategy reflect this interrelationship with the household sample and the characteristics of its members. The analy- tical goals of the overall study required comprehensive national estimates of the health insur- ance provided by private and public programs in the United States in 1977. These estimates thus reflect coverage of the population by age, sex, and sociodemographic characteristics; insurance premiums and sources of premium payment; type, breadth and depth of coverage; the availability of private health insurance options and associated premiums and specific benefits; and the extent and nature of coverage for catastrophic illness. The Health Insurance/Employer Survey verified and supplemented information reported by household respondents on private health insurance coverage, particularly as to whether it was '' obtained in relation to the insured's employment, and ascertained the availability of such cover- age to household sample members who reported that they were not covered by employment related health insurance. Also, the existence of optional insurance coverage available from but not obtained from employers was established. Insurance companies, employers, unions, and other organizations identified by respondents in the household survey as the source of their private insurance coverage were surveyed to obtain detailed data on insurance, including verification of reported coverage, information on the provisions of coverage, its cost, and the parties responsible for payment of the premiums. Also, where the employers of persons reporting that they were not covered by employment- related insurance could be identified, these were surveyed to confirm lack of coverage and to collect policy and premium information for both existing and available coverage. Coverage by public insurance mechanisms (Medicare, Medicaid, and CHAMPUS/CHAMPVA) was not verified, but the benefits provided by these public programs were obtained from the administering agen- cies and included in the HIES data. National estimates of the cost of health insurance to employers and of the characteristics of employees eligible for employment-related health insurance were obtained in the Employer Health Insurance Cost Survey. In this survey component, employers provided data on the num- ber of their employees according to wage-rate categories; level of unionization; total annual payroll; total annual health insurance premiums and the share of employer and employee contri- butions; and the number and salary level of employees eligible to participate in health insurance plan(s) sponsored by their employer (see Taylor and Lawson, 1981). The next section of this report briefly describes the data collection methods and instru- ments used in HIES and EHICS. In subsequent sections, the sample design of each survey is outlined, and weighting and adjustment procedures for various types of nonresponse and missing information are described in detail. Proper use of the weights will yield national estimates compatible with household survey data of insurance premiums and sources of premium pay- ments, types of coverage for the civilian noninstitutionalized population in 1977, and of the health insurance premium costs of all U.S. employers. '' Survey Procedures Health Insurance/Employer Survey Data Collection HIES data collection procedures were initiated on the basis of information obtained in each round of the household survey regarding the enrollment of household members in private and public insurance programs. For each health insurance policy reported by members of the household survey, permission was asked from the household to contact the respective insurance company or group (e.g. employer, union, fraternal organization) through which the policy was obtained. In this way coverage reported by the household was verified; additional coverage not reported by the household was discovered; and information concerning benefits and the payment of premiums was obtained. Exceptions were policies covering "dread diseases" only (e.g., stroke or cancer), cash coverage for hospital stays limited to small supplemental payments, and reim- bursement of medical expenses for accidents or injuries under motor vehicle or homeowner insurance, These were not considered health insurance for the purposes of this phase of the study. For all employed but not self-employed persons in the household survey who reported not being insured through their firm of employment, permission was requested to verify lack of coverage with their employers in the Uninsured Validation Survey (UVS), a substudy of the HIES. For these purposes, employment status was determined from a supplementary set of questions on employment asked in either the third or fourth and again in the fifth round of the household survey (see Bonham and Corder, 1981). It should be noted that persons reporting no coverage through their own employment were not necessarily without private insurance accord- ing to the household report. In fact, a significant proportion were covered as a dependent under another family member's policy. The only household survey participants whose private insurance status could not be veri- fied were unemployed or self-employed respondents and their dependents who reported being uninsured during all of 1977. These individuals were assumed to be in fact uninsured. Household reports of public coverage were accepted without verification. It should be noted that al- though verification of insurance status was one reason for undertaking the HIES, classification by private and/or public insurance was in fact changed for less than one percent of the eligible population on the basis of this information (see however, Walden, Horgan, and Cafferata, 1982, for discrepancies in reported and verified premiums and specific benefits). Permission forms for HIES were obtained in two stages. In the second interview of the household survey, all participants identified as the primary insured under policies obtained directly from a health insurer (in the following, referred to as nongroup coverage) were asked to sign a Health Insurance Permission Form (HIPF; see Appendix A.1) authorizing use of a ques- tionnaire to obtain verification of coverage and benefit and premium information from the '' insurer. During the second round, a 25-percent sample of households reporting any group insur- ance coverage was asked to sign a permission form authorizing contact with the group; most often this was an employer, but unions and other sources of group coverage were included. In the fifth household interview, permission forms were obtained from the remaining househalds with members covered by group insurance, from persons who had obtained new or different coverage since the second interview round, and from persons eligible for the Uninsured Valida- tion Survey (UVS). In all, signed permission forms were obtained from approximately 90 percent of household respondents who were the primary insured under group insurance or privately purchased plans, and from approximately 70 percent of household respondents eligible for the UVS. Approximately 16,000 permission forms were mailed to insurance carriers and insurance groups between October 1978 and August 1979. Responses were obtained for 85 percent of household members who had signed either the HIPF or UVS questionnaires or both. These forms served as the HIES data collection instrument for all household-reported insurance plans and for those discovered in the UVS. The forms asked each HIES respondent for verification of cover- age; whether the policy covered a family, a couple, or an individual; whether it was for group or nongroup insurance (employers contacted in the UVS were assumed to provide group coverage); and for data on the total annual premium, the amount paid by the insured, and the payment of premiums by other sources such as the person's employer. A copy of the policy or certificate describing insurance benefits was also requested, Where multiple coverage was held by a person or family, the information obtained was subsequently compiled to represent the person or fami- ly's overall coverage. In addition, a questionnaire was sent to all employment-related group plans (including UVS respondents) in a second substudy, the Health Insurance Option Survey (HIOS), which was de- signed to obtain estimates of optional coverage available to household respondents and their dependents. Here, information was requested on the number of employees enrolled in the specific group plan or plans and on the existence, financing, and benefit provisions of health insurance options available from the group but not elected by household survey participants. (See Appendix A for all HIES survey forms.) Abstracting of Policy Information The policies and certificates describing both insurance plans actually held or available to a household were abstracted by highly trained coders, most of whom were health insurance claims examiners, onto forms initially developed for the Rand Corporation Health Insurance Study (this experiment is described by Newhouse, 1974). Basic and major medical benefits for specific health services were identified, as were deductibles, reimbursement rates, limitations, and other benefit provisions for each covered service. For a 15-percent subsample of NMCES reporting units who had signed at least one Health Insurance Permission Form, a longer form '' was used to code additional information. All data items on the short form were included on the long form, so that estimates for all short-form questions can be derived for household respond- ents with any abstract information. The same abstracting procedures were used to include public insurance benefits in the HIES data set. Separate forms were coded for Medicare, CHAMPUS/CHAMPVA (active), CHAMPUS/CHAMPVA (retired), and each state Medicaid program. Different Medicaid ab- stracts were used for categorically needy and medically needy recipients in states where this was appropriate. Because household respondents reporting Medicaid coverage were not asked to classify themselves as either medically or categorically needy, Medicaid respondents who had received Supplemental Security Income or welfare payments during 1977, or who belonged to a household receiving such benefits, were assigned to the categorically needy benefit group. Otherwise, they were assigned to the medically needy benefit group. Assignment of Benefits to Dependents Since private health insurance policies generally neither enumerate nor identify depend- ents by name, insurance carriers and groups were only asked to confirm the coverage of the primary insured reported by the household. Coverage or lack of coverage for family members was inferred from confirmation that the primary insured was covered by the carrier or group and that the coverage extended to an individual, a couple, or a family. Verification of coverage for dependents and their plan benefits used information concerning the insurers for each pri- mary insured in a family as reported in the household survey. HIES data for a family or couple insured by a particular insurer were then attributed to all appropriate family members for whom coverage from that insurer had been reported in the household survey. An analogous procedure was used to exclude coverage reported in the household survey but not confirmed by an insur- ance carrier or group in the HIES. The exception to this extension of benefits to dependents was insurance not reported in the household survey but ascertained when a HIES respondent, e.g., an insurance carrier or group, reported coverage additional to that shown on the permission form signed by the house- hold respondent, or when a UVS respondent (an employer) confirmed previously unreported coverage for an employee. In these cases, family and couple coverage was assigned to the appropriate dependents by either of the following methods: o If the primary insured under a plan newly ascertained in the survey also subscribed to a family or couple plan that had already been reported in the household survey, house- hold members linked to the reported plans were linked to the discovered plan as well. o Inall other cases, benefits and coverage under discovered plans were assigned on the basis of family relationships, such that family plans held by persons under 65 years of '' age were linked to their children under 21 or who were unmarried college students without their own insurance; and family and couple plans were linked to the spouse of the primary insured. o Discovered UVS plans were linked to dependents through family relationships as described above, and the same rules were used to identify spouses and children who were eligible for assignment of UVS data through the employment of someone else in the family. The identification of these dependents as HIES eligibles was a consideration in the construction of the HIES sample weights. Employer Health Insurance Cost Survey For purposes of EHICS, the Health Insurance Permission Forms signed by privately in- sured, employed household survey participants and the Uninsured Validation Survey Permission Forms signed by employed participants not reporting employment-related insurance were used to identify the employer's name and address. Also, a 10-percent systematic sample of persons in the household survey who provided a permission form not associated with an employment- related policy but who had reported being employed throughout 1977 or for part of the year, as well as those who did not sign any permission forms, were telephoned and asked to provide the name of their 1977 employers. Another 10-percent systematic sample of individuals for whom employment status was missing were likewise telephoned and asked to report their employment status in 1977 and, if applicable, to provide the names of their employers. Firms identified as employers of EHICS sample individuals were contacted by mail or telephone to provide data on their health insurance costs, number of employees, size of payroll, and related information (see Appendix B for the survey forms). EHICS data collection resulted in the completion of questionnaires by 5,845 unique employers (including those with multiplic- ity-only responses; see pages 28 to 36). The total response rate was 85 percent. '' The Health Insurance/Employer Survey The Health Insurance/Employer Survey produces national estimates of the health insur- ance of the civilian noninstitutionalized population of the U.S. in 1977. Based on responses in the household survey, enrollment in various types and combinations of public and private insur- ance plans was determined and private coverage was verified. Health insurance premiums; the amounts paid by employers, employees, and other sources; and variations in these for different types of plans and individuals can be described, and benefits held under both private and public insurance plans can be shown in considerable detail. These data, however, are for a one-year period and do not distinguish between policies held at different times in 1977 or for only part of the year. For example, total health insurance premiums in 1977 are the sum of the annual premium for each policy held during the year and not the sum of premiums in fact paid between January 1, 1977 and December 31, 1977. Data are also provided on the availability, cost, and benefit provisions of health insurance options offered through employer groups. HIES Sample Weights To obtain national estimates from the responses obtained in the HIES and its substudies, the UVS and the HIOS, the development of a series of weights adjusting for various types of nonresponse was required (Folsom, 1981). Adjustments for reporting unit nonresponse and poststratification to represent the civilian noninstitutionalized U.S. population as of 1977 had been incorporated in the sample weights developed for key participants in the household survey (Cohen and Kalsbeek, 1981; there, key participants were defined as persons initially selected into the household sample, babies born into sample households during the study, and initially ineligible household members, i.e., who had entered the military or an institution but who later became eligible by rejoining the household). Given complete response from all private insurance carriers and insurance groups associated with members of the NMCES household survey eligible for HIES, these household survey weights theoretically allowed for the derivation of national health insurance estimates. In the absence of full HIES participation and response rates, an additional and analogous set of weights was developed to adjust for the following types of nonresponse in the Health Insurance Employer Survey: (1) Permission form nonresponse. Eligible participants in the household survey refused to sign the permission form authorizing contact with their insurance carrier or insurance group (employer in the case of UVS). (2) Health insurer nonresponse. The insurance carrier, insurance group, or employer identi- fied on a signed permission form did not respond at all or did not respond to or provide one or several of the following: '' a) the coverage questionnaire, requesting verification of private insurance coverage as reported by the household and information about premiums and the type of policy; b) acopy of the policy describing applicable benefits; c) the options questionnaire, requesting information about the availability of other health options from the group and their cost; d) acopy of the policy of each optional plan. (3) Partial health insurer response on multiple coverage. Some but not all insurance carriers, insurance groups, and/or employers identified for a particular participant in the household survey responded in the HIES. First, the household survey weights were adjusted for unsigned HIES permission forms and health insurance provider nonresponse. Alternative weights were derived for estimates relating to items (2)a-(2)d above. A fifth factor was developed to adjust premium estimates for partial responses in the case of multiple coverage (item (3)). Eligibility As described previously, key household survey participants were eligible for the Health Insurance/Employer Survey with the exception of the unemployed or self-employed reported as uninsured in the household survey and their uninsured dependents. For the development of national sampling weights, an HIES eligibility indicator, ELG;, was constructed to estimate the eligible HIES population, where ELG; = | if the household survey data implied that key participant i had health insurance coverage (public or private) at any time during 1977; participant / was employed but not self-employed, more than 17 years old, and reported no insurance obtained through his or her employer (UVS eligibility); the participant was the spouse or child of someone meeting the criteria for eligibility above; or if any HIES response was obtained from an employer, group, or insurance carrier for persons not otherwise eligible. = 0 otherwise. ''In addition to this eligibility indicator, mutually exclusive classes of insurance coverage were specified to obtain population estimates according to household-reported health insurance coverage. This indicator took the form if participant i was assigned to insurance coverage category k by the INSCOV,; = 1 household survey data where ea indicates only public coverage in 1977, k indicates any private coverage, and indicates uninsured throughout the year, and ea) otherwise. To allow for nonresponse adjustments within homogeneous weighting classes determined by cross-classification of the age, race, and sex of household survey participants, another indi- cator, C;;, was specified as if key participant i was in age-race-sex class /1 Chi = | and =O) otherwise where /i contains 32 age-race-sex weighting classes resulting from cross-classification of the following characteristics: one Race Sex 0-4 35-44 White Male Daly 45-54 Other Female 18-24 55-64 25-34 65+ The weighted estimate of the population of eligible individuals in insurance coverage class k and age-race-sex class h was then determined as WEnk = ps ELG; Cpj INSCOV,; W; tENMCES ''10 where Ww; is the NMCES household sample weight which reflects the ith individual's selection probability, adjusted for household nonresponse and poststratified, and igNMCES _ indicates key participants in the household survey. Household survey participants ineligible for the HIES, i.e., the unemployed or self-employed reported as uninsured in the household survey, were also considered to be uninsured in the HIES. The population of ineligible individuals in age-race-sex class h was estimated as WIps where WI), = > Cy, Wi 1EINELG and isINELG refers to key participants in the household survey who were ineligible for the HIES and were assumed to be uninsured as reported in the household survey (ELG; =0). Insurance Coverage To compensate for both permission form and health insurer nonresponse in the analysis of coverage and premium data collected in the coverage questionnaire, HIES sample weights were developed for all individuals with a coverage questionnaire response. To this end, their house- hold survey sampling weights, wi, were inflated to represent all individuals eligible for the HIES, including those for whom HIES respondents returned a copy of an insurance policy but no cover- age questionnaire response; these individuals were presumed to have coverage. Nonresponse adjustments were then made within weighting classes defined by cross-classifications of age, race, sex, and household-reported insurance coverage. To account for the possibility of different levels of nonresponse for UVS eligibles who reported no private insurance whatever and those household respondents who reported that they were covered by private plans, a separate indicator, U, was specified for all individuals with ELG; = 1 such that lI U=1 for UVS eligibles without any household-reported private insurance, = 2 for household survey participants who reported only public coverage and were not eligible for the UVS, ''a = 3 all others eligible for the HIES, including UVS eligibles who did not report their own work-related coverage but were reported to be covered by private insur- ance from some other source. For U=1, the weighting classes for nonresponse included an indicator distinguishing be- tween full and part-time employees; here, spouses and children who did not themselves satisfy the employment criteria for eligibility were characterized according to the employment charac- teristics of the primary UVS eligible. This indicator, E,;, was specified for part- and full-time employees such that Ey = 1 if the individual was assigned to employment category t by the household employment data, where ll 1 indicates the individual worked only in jobs limited to less than 35 hours/week (part-time) Hier! indicates full-time and = 10 otherwise To adjust all weighting classes for coverage questionnaire nonresponse, a response indica- tor, Q;, was specified, where O;=1 if one or more of the following for person i: any response to a UVS coverage question if person i had no private insurance according to the household survey; verification of coverage on at least one permission form, or negative responses accounting for all household-reported insurance plans; policy information for a private plan; or the individual reported only public coverage, was not eligible for UVS, and therefore was assigned the coverage reported in the household survey, and = otherwise. Weighted counts of individuals with responses to the coverage questionnaire WR)’;, were derived separately for each value of U. For UVS eligibles without household-reported private insurance(U=1), 48 weighting classes were defined by cross-classification of age (collapsed into ''12 three categories: 0-17, 18-64, 65+), race, sex, the two remaining categories of household re- ported insurance coverage (public only, k= 1, or uninsured, k= 3), and full-time employment, such that WRoy' ky = 2 Q; Cry INSCOV , ;E “W thea 1eU=1 where isU = 1 considers only individuals eligible for UVS without household-reported private insurance and Cyr is an indicator for age-race-sex class h'(h’=1,2...12) where age is collapsed into three categories (0-17, 18-64, 65+). Here, the nonresponse adjustment factor for UVS eligibles without household-reported private insurance is defined as a WE pret kt WR hk with WEy,,, representing the weighted eligible population of UVS individuals with U=1in class h'kt and igU=1 Weighting classes were generally collapsed when these nonresponse adjustment factors exceeded the value of 4, or the weighting class consisted of less than 20 respondents. Household survey participants who reported only public coverage and were not eligible for the UVS(U = 2) were assumed to have only public coverage and were uniformly treated as cover- age questionnaire respondents. Their nonresponse adjustment was consequently equal to one. Among individuals in class U=3 and k= 2, weighted counts (WR,,,) of those with HIES cov- erage questionnaire responses were used for nonresponse adjustment factors. These counts were constructed for weighting classes by age, race, and sex, such that ''13 WRh2 = > Q; Cy INSCOV,; W; isU=3 where considers the individuals eligible for the HIES who had private coverage ac- cording to the household report. isU = 3 The nonresponse adjustment factors for individuals with an HIES questionnaire response in class U=3 took the form WE, 5 Ang = aR WRh» where WEp > = > ELG; Chi INSCOV 5; Wi 1ENMCES This factor adjusts for the nonresponse of HIES eligibles in classU=3,. For each of the classes, U=1,2,3, the nonresponse adjusted HIES coverage questionnaire weight W was con- structed for individuals with an HIES coverage questionnaire response (Q;= 1)as GW. Wi = Ay; where A, = Ah’ kt for U = | =i for U = 2 = Apo for U = 3 and ; wl = 0 otherwise (individuals without an HIES coverage questionnaire response, Q;=0). ''14 The distribution of those eligible for HIES with a coverage questionnaire response, classified by household reported coverage, is shown in Table 1. To correct for potential under-representation of some population subgroups and ensure consistency with Census Bureau population estimates for 1977, the nonresponse adjusted HIES weights were poststratified within each of the 32 initially specified age-race-sex weighting classes (h). While the HIES weight for household survey participants who reported only public coverage (U=2) was not altered, the weights for the remaining eligible individuals(U = 1,3) were adjusted by the factor, P,, where 2 Chi" i (eU=1,3 Ps Cw? fy = and, as before, w; is the poststratified household sampling weight for each household survey participant. Consequently, the poststratified HIES coverage questionnaire weight took the form w2 - a, w2’ for Q, = 1 l where A, = 1 tor U = 2 Az = Py for U =1,3 and ieh and w= 0 otherwise (Q;=0). To provide national population estimates incorporating ineligible uninsured persons (ELG;=0), each individual's poststratified NMCES household weight was assigned to we i.e., we= Wj for all those with LLG. = 0. Premiums for Multiple Coverage In addition to private health insurance coverage, the HIES analysis also required informa- tion about premiums and the respective amounts paid by the insured and others, such as the subscriber's employer. In cases of multiple coverage, failure by one or more insurance carrier or group to provide premium information yielded only partial response on premiums associated ''15 Table 1. HIES Response Distribution for Household Survey Participants, by Household Reported Insurance Coverage Number of Household household survey reported participants Questionnaire coverage eligible for HIES response Public only 6,187 4,591 Eligible for UVS (U =1) 2,285 689 Ineligible for UVS (U = 2) 3,902 3,902 Some private (U = 3) 28,898 19,099 Uninsured (U =1) 2,424 828 Total eligible 37,509 24,518 Total ineligible* 1,306 n/a “Insured neither by private nor public insurance and not employed or self-employed. with an individual's coverage. For those with only partial premium information but who were considered as having coverage questionnaire responses, the available premium data were ad- justed to account for additional unreported premiums. First, valid HIES responses received for each primary insured who signed more than one permission form were matched against permission forms in order to identify partial responses. Subscribers with complete and partial responses were then categorized into three groups ac- cording to the number of permission forms they had signed (2,3, and 4 or more). An inflation factor, /,, was defined for those with partial responses in each group(n=1,2,3)as PREM Pe eee nN preme PREM}, where l, = the inflation factor for subscribers with partial responses in group n, PREM), = average of the sum of premiums per subscriber with complete responses in group n, and PREM? = average of the sum of premiums per subscriber with partial responses in group 7. ''16 The total premium for the coverage of each subscriber with partial premium response in group /” was then computed by multiplying reported premiums by /,. This adjustment produced an estimate of total national premium expenditures in 1977 that is consistent with estimates made by the health insurance industry (Health Insurance Association of America, 1982). Benefits From Policy Abstracts An additional weighting strategy was necessary to account for missing policy abstracts, i.e., insurance benefit information for individuals classified in HIES as having private coverage but for whom insurance carriers or groups, while verifying coverage, had not provided the requested policy description. Therefore, an HIES determined indicator of verified private insurance status was developed and used in conjunction with the nonresponse adjusted, poststra- tified HIES weight, wg to derive the required abstract weights. Also, in order to construct national estimates from the subsample for which long-form abstract information was obtained, a separate long-form abstract weight was required. No separate policy information adjustment was made for individuals classified in the “public only" coverage category. Where only public insurance coverage was reported in the household survey, the benefits offered under these programs were abstracted from available public information and assigned to the appropriate individuals. Complete benefit information was consequently available for all persons so classified, and in their case the appropriate weight for analyzing benefits corresponded to the household-reported coverage weight. The indicator of insurance status as verified by HIES respondents (or, in the case of public insurance, as reported in the household survey) was defined as PC;;- It took the form os if the household survey participant was assigned to insurance coverage category k on the basis of HIPF or UVS responses =0 otherwise where k= 1 indicates only public coverage, k=2 indicates some private coverage, and k=3 indicates uninsured. ''17 National estimates of the civilian noninstitutionalized population with only public or at least some private insurance coverage, falling into the 32 age-race-sex classes, h, were then deter- mined by wrk = a Chi PCy WP i€QR where QR represents individuals with an HIES coverage questionnaire response, Q;= 1. Here, to incorporate all those without any coverage, a national estimate for the uninsured population (k=3), was derived by adding W/,, the estimate of persons in age-race-sex class /h without health insurance and ineligible for HIES, to WP). To describe the benefits of individuals with public and/or private coverage, a short-form abstract indicator, SA;, was constructed as follows: SA; = 1 if short or long-form abstract data for a private plan were ascribed to an individual in category k = 2(some private coverage) or if individual i was assigned to category k=1(only public coverage), i.e, a category for which complete benefit information was available; 0 otherwise. Since the short-form abstract used to code insurance policies was an abbreviated version of the long form, the completion of a long form provided responses to all short form items. Thus, SA;=1 for all individuals with any abstract data on private or public insurance. Two types of weights corresponded to the long and short form. Weighted counts of indi- viduals with short-form abstract responses, WS;,, were derived within the 64 age-race-sex, insurance coverage (public only, some private) weighting classes such that : QO WSik = 2 SAy Cpe PCy, Wi 1EQR The nonresponse adjustment factor for short-form abstracts within age-race-sex insurance coverage class hk (given an HIES coverage response) consequently took the form We hk 1 ''18 When fewer than 20 respondents were observed in any of the hk classes, or the factor RSp, was noticeably greater than 2, the classes were collapsed to reduce variation among sampling weights. The HIES short-form abstract weights, Ww, were therefore defined for all participants in weighting class hk as Ww - RS, WE for SA, = 1 l and w> = 0 otherwise. Long-form abstract weights were constructed only for individuals with some private insurance coverage (k=2), A 15-percent subsample of reporting units with at least one signed Health Insurance Permission Form was designated for long-form coding. As no long-form sub- sample was drawn from the reporting units where only a UVS permission form was signed, survey participants from households or reporting units eligible only for UVS were, therefore, certain to be underrepresented in the long-form abstracts. A nonresponse subsampling adjust- ment was developed to account simultaneously for undercoverage of UVS-only reporting units, long-form nonresponse, and the long-form subsampling rate. To derive this adjustment factor, a long-form abstract indicator, LA;, was specified such that LA; = 1 if private insurance by the HIES insurance indicator, PC,;=1, and long-form but no short-form abstract data for private insurance plans were ascribed to the eligible individual =iK0) otherwise. Individuals with private policies abstracted on short forms as well as long forms were not classi- fied as having a long-form response, because items available only on the long form were missing for some insurance plans. Weighted counts of HIES long-form abstract data response, WL ys were derived within the 32 age-race-sex weighting classes used for insurance class k= 2, where L > ~ yw wae > CR Peg Wi 1EQR The nonresponse subsampling adjustment factor within weighting class /2, RLj,», Was expressed as WP RL), = — h2 WLp> ''19 where, as before, 7 QO WP = > Cig PCy Wi igOR Again, when fewer than 20 respondents were observed in any of the 42 weighting clases, they were collapsed. Long form abstract weights, Wry are expressed as wr = RL, we for all participants in weighting class h2 with LA;=1, we 0 otherwise. ~ Since individuals classified as uninsured in both the household survey and the HIES were not assigned abstract data by definition, they were uniformly attributed short and long-form response indicators with zero values. However, to allow for their inclusion as individuals with- out coverage in the reporting of short-form data, their HIES questionnaire weight, we was assigned to Wie, w> = w? for those with PC3;=1. l l Coverage Options in Employment-Related Groups When coverage from an employment-related group was reported in the household survey, the group was also asked to complete a Health Insurance Options Survey (HIOS) questionnaire requesting information on optional coverage available to but not obtained by the household respondent, as well as policies and premium information for such optional plans. Similar information was collected from employers contacted in the UVS. As most of the UVS eligibles were not in fact uninsured but covered as a dependent under a family member's insurance, the UVS usually confirmed that they had not enrolled in their employer's insurance plan. In these cases, the "optional plan" was therefore not a true alternative plan, but rather the only plan in which they could have enrolled had they elected coverage at their own place of work. The main effect of the UVS options questionnaire was, therefore, to ascertain whether ''20 individuals not enrolled in an employment-related plan had one available to them. Only where the UVS revealed that the employee was in fact enrolled in a plan did optional plan(s) reported by a UVS respondent represent additional choices offered by the group. There were two types of nonresponse in the HIOS. Some employers or groups did not return the optional coverage questionnaire, while others provided this information, but did not forward policy documents describing the benefits offered as options. Both an optional coverage weight and an optional policy abstract weight were consequently required for the analysis of these data. HIOS Eligibility Eligibility for the Health Insurance Options Survey (HIOS) was restricted to participants in the household survey holding employment-related group insurance that was confirmed through a HIES Health Insurance Permission Form and to those UVS eligibles whose employment was confirmed by an employer contacted in the UVS. A first requirement for HIOS eligibility was a HIES coverage questionnaire response for an individual either as a primary insured or as a dependent. The population that was eligible for an optional coverage questionnaire then consisted of three groups. The first group, » = 1, comprised the primary insured named on Health Insurance Permission Forms and the accompanying Health Insurance Options Survey. The second group,y=2, comprised individuals not in g= but eligible for the UVS; these were employed individuals whose lack of coverage through their employer was confirmed. The third group,g=3, comprised dependents of those in the first two groups but who were not themselves in either group. To estimate the population eligible for an optional coverage questionnaire among those with we greater than zero, i.e., with an HIES coverage response, an eligibility indicator, OE gis was defined, where OE,; = 1 if individual i subscribed to an employment-related group plan that was verified on an HIPF or if information describing an optional plan available to such an individual from an employment-related group was received with an HIPF; OE,; = | if OF; + 1 and individual i was an employee eligible for the UVS or infor- mation describing an optional plan available to such an individual was re- ceived with a UVS questionnaire; OE,; = 1 if OF,; # land OF,; + 1 and individual i was the dependent of an individual who was eligible according to any of the criteria above; OE 4; = (0) otherwise. ''21 Employment-related group plans were defined for these purposes as group plans either administered by an employer or union or the premiums for which were at least partially paid by an employer or union. Family members who were ascribed the coverage data of a primary insured from a Health Insurance Permission Form or an Uninsured Validation Survey question- naire were identified as the person's dependents in defining g=3. Availability of Optional Coverage To produce a set of weights for analyzing the availability of optional plans, an HIOS response indicator was established for individuals in each group. Here, OR,; = | if there was a response to at least one optional coverage question in the Health Insurance Options Survey, or benefit or premium data for an optional plan were received with an HIPF for employee i (g= 1); OR,; = 1 if there was a response to at least one optional coverage question in the UVS, or benefit or premium data for an optional plan were received with the UVS questionnaire for employee (g = 2); OR,; = 1 if individual /(g=3)was the dependent of some individual with OR,; Or OR,; = 15 OR 9; = 0. otherwise. Weighted counts, WOR ons of respondents within weighting classes defined by the previously described age-race-sex groups, HIES-determined insurance coverage, and optional coverage eligibility were produced as WOR Gh = >. — ORG Cay PC gy we lENMCES These totals were included in the nonresponse adjustment factor, NOR hk which was defined as WOE opp NOR pny = ——S WOR ohg | v ''22 where WOE ony = 2 OE 9 Oyj PCgy Wj lENMCES is the sum of individuals eligible for optional coverage in weighting class ghk, using the eligibil- ity indicator, OE gis Again, we is the HIES coverage questionnaire weight. Classes were gener- ally collapsed whenever there were fewer than 20 respondents so classified or this factor ex- ceeded the value of 2. The optional coverage weight was then constructed as w?- NOR WE for eligible individuals with an optional coverage response (OR ,; = |) and 0 other wise. To incorporate individuals who were ineligible for the optional coverage questions and were presumed to have no optional employment-related group coverage, poststratified HIES coverage weights were assigned to we, where WO= we, Health Insurance Options Optional plan policies were abstracted in addition to policies describing the coverage actually selected so as to describe all private insurance benefits available to individuals as a result of their employment. First, in constructing a set of weights to describe the extent of optional insurance benefits, an indicator of the availability of optional coverage for persons in group g with HIOS responses (OR gj =1)was established, namely, OC,; = | if at least one optional coverage questionnaire in the HIOS indicated that there was an employment-related group plan available to but not selected by em- ployeei(g=1)s OC,; = 1 if at least one optional coverage questionnaire in the UVS indicated that there was an employment-related group plan available to but not selected by em- ployee i (g= 2); OC,; = 1 if individual i(g=3) was the dependent of an individual with OC; or OC,;=1; OC,; =Q otherwise. ''° lll, 23 A similar indicator was considered for each HIPF or UVS questionnaire in conjunction with the actual coverage reported on the form, in order to determine the policy abstract data neces- sary to represent the various benefits available to the individual from the group. The abstracts required to make this determination for any employee for a particular group fell into four categories (1): Employee Category enrolled Any coverage (r) in group option Abstracts required 1 yes yes held and optional policy 2 no yes optional policy 5 yes no held policy 4 no no none Here r=1 relates to individuals who selected a policy from their group and also had op- tional coverage available; = 2 to individuals whose lack of coverage from their employer was verified but who could have enrolled in a plan had they so desired; ,=3 to individuals who en- rolled in the only plan offered by their group; andr,=4 to those who were offered no employ- ment-related health insurance benefits whatever. No distinction was made between long-form and short-form optional policy abstracts because there were too few long forms available for analysis. A policy abstract indicator for each person was then specified for the construction of a set of weights to analyze the benefit data. This indicator took the form SO,;,;= 1 if the policy abstract data necessary for the respective category, +, were available for at least one HIPF naming individuali(g=1)as the primary insured; SO,; = | if the necessary policy abstract data were available for at least one UVS questionnaire for individual /(¢ = 2); SO,, = 1 if individual i(g=3) was the dependent of an individual with SO,; or SO,; equal tO 13 SO,; = 0 otherwise. ''24 Each individual with OR; =| was also assigned to a weighting class defined by the type of ab- stracts required for SO i= 1 (r=1,2,3,4). Individuals with complete abstract data for more than one insurance group were assigned to the lowest value of + which was applicable to any one HIPF or UVS questionnaire. An adjustment factor to account for abstract nonresponse was computed within weighting classes defined by the previously defined demographic and coverage groups, the three eligibility groups g, and the values of r. It took the form rso, = 1 OR hker MRT WOS i boy where Lite panic O WOShkor = 2 50 gi Oni PC Wj ier and WOR pk gr 2 OR gi Chi PC Ki we For individuals with r=4, for whom no abstract information was required, RSOnxer = 14 Whenever the minimum weighting class size constraint of 20 respondents was not satisfied, cells were collapsed. The final optional policy abstract weights were computed as s w? 7 RSO ngy WE for SO,, = 1 l and wPS = 0 otherwise. l As before, to incorporate individuals ineligible for the options coverage questions and presumed to have no employment-related group coverage options, poststratified HIES short- form weiglits were assigned to ws (wos = we Je = ''25 Summary Information on health insurance coverage obtained from the households participating in the National Medical Care Expenditure Survey was verified and supplemented in the Health Insurance/Employer Survey (HIES). Insurance companies, employers, unions, and other organiza- tions identified by respondents in the household survey as the source of their private insurance coverage were asked both to verify the coverage reported in the household survey and to pro- vide information on the provisions of each subscriber's coverage, its cost, and the parties re- sponsible for payment of the premiums. In a substudy, the Uninsured Validation Survey (UVS), lack of employment-related insurance coverage reported in the household survey by employed adults was verified through their employers and if coverage was in fact available, premium and benefit information was obtained. The only household survey participants whose private insur- ance status could not be verified were unemployed or self-employed respondents and their dependents who reported being uninsured during all of 1977. These individuals were assumed to be in fact uninsured. Although no attempt was made to verify enrollment in public insurance programs reported during the household survey, information similar to that obtained for private policies was ob- tained for benefits offered by Medicare, CHAMPUS/CHAMPVA, and the different state Medicaid programs. By taking household-reported public coverage into account, the HIES data base thus encompasses public as well as private health insurance benefits. For household respondents whose employment-related insurance was verified, the Health Insurance Options Survey (HIOS) obtained information about the availability of alternative or optional plans from HIES respondents, as well as data on enrollment and the payment of premi- ums for optional plans similar to those collected for coverage actually held. Particular attention has been given to the sampling strategy adopted to survey the insur- ance carriers, insurance groups, and employers identified by NMCES household survey partici- pants. Detailed specifications are presented for the weighting strategies employed in the HIES and its substudies. To illustrate the procedures which allow for the derivation of national estimates of the health insurance coverage of the U.S. civilian population in 1977, the discussion includes a description of weighting adjustments employed to account for the different probabil- ities of selection that characterize sampling units and for different levels of nonresponse. ''''ror 27 The Employer Health Insurance Cost Survey In addition to the Health Insurance Employer Survey (HIES), the National Medical Care Expenditure Survey incorporated an Employer Health Insurance Cost Survey (EHICS). Designed ‘to produce national estimates of the availability of employment related health insurance and its costs in the United States for 1977, it obtained information from employers identified in the household survey on firm-level characteristics not otherwise available, such as size of firm, unionization, distribution of wages within their workforce, and access to employment-related health insurance. Instead of constructing a separate sample frame defining employers as sampling units, the strategy adopted in this survey to obtain a national sample of employers was based on data already collected in the household survey. Specifically, the employer sample for EHICS was defined as the employers associated with persons in the household sample, which was designed to represent the civilian noninstitutionalized population of the U.S. (Cohen and Kalsbeek, 1981). The rationale for this strategy was that since each employer in the nation is linked to one or more employed persons, the NMCES sample of employed individuals yields an associated sample of employers (Lessler, McNeill, and Waddell, 1980) with the probability of selection for an employer contingent upon the number of persons employed by each firm and the respective probability of an employee being drawn into the NMCES household sample. The EHICS design thus involved the definition of a frame of employed individuals at a particular point in time or period of time; the selection of a sample of employed individuals; the identification of firms linked to the sample of employees; and a survey of these firms concerning health insurance costs. Sampling Frame The design of the EHICS sample allows for the selection of employers as a consequence of their linkage to employed (but not self-employed) persons in the NMCES household sample. Since knowledge of the number of employees linked to a sampled employer was required in this estimation strategy, the questionnaire was designed to obtain from each employer a count of civilian employees as of December 31, 1977. While the optimal strategy would have been the construction of a frame of persons who were employed as of December 31, 1977, the Employ- ment Supplement in the household survey (Bonham and Corder, 1981) did not reflect employ- ment status for a single point in time. Consequently, the frame of employees was constructed to permit membership in the sample for any person who was employed at any time during 1977. A second measure of employment status of NMCES participants was obtained during Rounds 3,4, or 5 of household data collection (see Bonham and Corder, 1981). Briefly, all pri- vately insured household survey participants 14 years old or older were asked to sign a Health Insurance Permission Form; employed participants not reporting coverage by employment- related insurance were asked to sign an Uninsured Validation Survey Permission Form. For purposes of EHICS, these permission forms were used only to identify the employer's name and ''28 address. At the completion of Round 5, household survey participants were then divided into five groups for the development of the EHICS sampling frame. The first group, which was drawn into the EHICS sample with certainty, consisted of NMCES participants who reported being employed and holding employment-related insurance in Rounds 3, 4, or 5 of the household survey, and who signed the Health Insurance Permission Form. Individuals in the second and third group, from which 10-percent systematic samples were drawn, were those reporting employment in either Rounds 3,4, or 5 in the Employment Supplement Questionnaires but providing a permission form not associated with an employment- related policy, and, respectively, those who did not sign permission forms. Members of these two noncertainty sample groups were telephoned and asked to provide the names of their 1977 employers. Another 10-percent systematic sample was drawn from a fourth group consisting of individuals for whom employment status was missing. Members of this group were likewise telephoned and asked to report their employment status in 1977 and, if employed, to provide the names of their employers. A fifth group consisted of all NMCES participants with survey nonre- spondent status in rounds 3, 4, or 5 or who were not employed. This group was excluded from the EHICS sample frame. The initial counts for the groups constituting the EHICS sample frame (Table 2) included NMCES ineligibles, i.e., military personnel, institutionalized persons, or other than key partici- pants. This was due to limitations in the availability of data necessary to identify ineligibles at the time of frame construction. Once the employers of the sample individuals were identified, they were contacted by mail or telephone to provide the required data. EHICS data collection resulted in the comple- tion of questionnaires (including those with multiplicity-only responses) by 5,845 unique employ- ers, a response rate of 85 percent. (See Appendix B for the survey forms.) Sample Weights To derive national estimates of employer characteristics and health insurance costs from the sample described, it was necessary to define a set of weights which accurately reflected the probability of inclusion in the sample for each employer who completed the EHICS question- naire. Since this employer sample was based upon a subsample of household survey participants, knowledge of their selection probabilities was required in the weighting process. For all eligible employees in the EHICS certainty sample (Group 1; see Table 2), these were equivalent to their NMCES household selection probabilities, 7;. Given the 10-percent subsampling rate for the NMCES household participants selected in each of EHICS sampling groups 2-4, their probability of inclusion in EHICS was 7;/10. The EHICS employee level weight was then defined as the reciprocal of its selection probability. ie ''29 Table 2. Sampling Frame for the Employer Health Insurance Cost Survey Employment Total Total Group status number eligible* 1 Employed, with HIES employer permission form 8,634 8/523 2 Employed, nonemployer HIES permission form 2,692 267° 3 Employed, no permission forms 3,952 384? 4 Employment status uncertain 5,845 so 5 Not employed or inactive 17,9211 0 “Eligibility was restricted to key participants in the household survey. >Based ona 10-percent sample. “Of whom 120 were shown to be employed. SOURCE: Lessler, McNeill and Waddell, 1980. Further, the possibility had to be taken into account that several employees in the house- hold sample were linked to the same employer. Thus, the transformation from a sample of employed household survey participants to a sample of their employers required an adjustment to the EHICS sample employee weights to account for the possibility of selecting more than one NMCES household survey respondent for a given employer. The extent of such multiple linkages can be understood by considering a person-based survey in which all eligible employees in the U.S. are included, and then enumerating all unique employee-employer linkages for each em- ployer. The resulting count would equal the number of eligible employees working for the employer in 1977 and is referred to as the employer's multiplicity. The sum of all such linkages is equal to the total number of employee-employer linkages, which is the total number of jobs or employment positions in the nation in 1977. Thus, in order to produce estimates from the EHICS sample that are related to the number of employers rather than to the number of employee-employer linkages (or employment posi- tions), it was necessary to account for the number of employees in each firm or the employer multiplicity. This adjustment was incorporated intc the EHICS employer sample weight specifi- cations, the weight being derived by dividing the sum of sample weights for NMCES household survey participants associated with a specific employer by the employer's multiplicity. Barring missing data and nonresponse, application of these weights will yield statistically unbiased ''30 national estimates of relevant employer population parameters for 1977 (Sirken, 1970). Specifi- cally, if one considers the sum, Y, of a relevant employer characteristic determined from an enumeration of all employers, that sum can be expressed as where Y; is the value of the relevant characteristic for the jth employer, and j= 1,2,....M, with M denoting the total number of employers in the employer universe, Within the multiplicity framework of the EHICS sample, the enumerated total, Y, can be esti- mated by m r= 2 wij m4 where = is the sum over the distinct m(j=1,2,...,m) sample employers linked to mem- jel bers of the EHICS sample of employees. Here, Wig the sample weight for the jth sample employer, is determined by ''31 where % is the number of persons who are 14 years old or older and who are employed by the jth employer (referred to as the multiplicity), 1; is the probability of selection for the ith sample NMCES participant for EHICS, and fj > is the sum over all (Lj) participants in the sample reporting the jth employer. [=] Thus, the EHICS sample person weight was initially denoted as fe -1 we = (t.) i i As in most surveys, however, the assumption of complete response and fully answered questionnaires was not met, restricting the EHICS data base to responding employers. Corres- ponding adjustments in the weighting strategy had to take into account the following types of nonresponse affecting the EHICS sample: o Housing units drawn for the NMCES household sample refused to participate or contact could not be made; o Information on employment status was missing for active household survey participants; o The name of the employer was not obtained; Data on the number of employees employed by a particular employer (essential for the multiplicity factor) were missing on the returned employer question- naire; o All other data items were missing on the returned employer questionnaire. A set of poststratified NMCES household participant sampling weights, w,, that adjusted for housing unit nonresponse were available (Cohen and Kalsheek, 1981). For the EHICS cer- tainty sample (group 1), the initial EHICS employee weight, we was set equal to the initial NMCES household participant weight (WE w,), since all eligible members of this frame were included in the EHICS sample of employees. For all NMCES household respondents selected in EHICS sampling groups 2-4, household sampling weights were multiplied by a factor of ten to ''32 account for the 10-percent subsampling rate. Consequently, the initial EHICS employee weights took the form we = 10 w. l I Additional adjustments to the EHICS sample weights to control for the remaining types of nonresponse are discussed in the following. Missing Employment Status The original sample size for those with unknown employment status (group 4; see Table 2) was 584, leaving a sample of 559 after excluding nonkey household survey participants. For 409 members of this group who were telephoned and asked to report their employment status in 1977, employment status was determined; 14 individuals were found to be less than 14 years of age and ineligible; and 136 did not respond. There were 120 employed individuals in this re- spondent group. The EHICS sample weights of responding employed NMCES participants in this group were then adjusted to account for employed but nonresponding individuals. On the as- sumption of an equivalent employment status distribution for respondents and nonrespondents, the total employed in this group was estimated by inflating the weight for each employed re- sponding participant by the ratio of the sum of sampling weights for the 545 eligible partici- pants to the sum of weights of the 409 respondents. Consequently, the EHICS weights of group 4 members known to be employed were adjusted by the factor Sw ieR' ieR ''33 h 7 ; hE considers eligible group 4 sample respondents and nonrespondents and ieR considers eligible group 4 sample respondents only, such that wr AWE for employed group 4respondents (ieR) and we'- otherwise (i¢R), Unidentified Employers Employer identification was missing for members of sample groups 2 to 4 if contact was not established or household survey participants refused to identify their employer. Also, for group l, the certainty sample, some of the organizations named on the permission forms were not in fact employers. All employees with missing employer identifiers were defined as not linked to an employer, and all employees with identified employers were defined as linked to an employer. A nonresponse adjustment was used to allow the sampling weight of those partici- pants linked to employers to characterize the total population of persons 14 years old or older who were employed at some time during 1977. On the assumption that linked and nonlinked EHICS employees were similarly distributed by type of employer, the weights of linked employ- ees were inflated by the ratio of the weighted sum of those employed to the weighted sum of those employed and linked. This was done separately for each sample group (1-4). The number of linkages and nonlinkages and ne corresponding adjustment Pee are shown in Table 3. Here, the EHICS sample weights, wr ,of the linked employees (wr ‘for members of sample group 4, adjusted for missing employment status) were adjusted by the factor A, within sample groups. The new weights , wis for the linked employees took the form w4 ae AW for employees linked to an employer (ieL) and = -0 otherwise (i¢L). ''34 Table 3. Employer Linkages, Nonlinkages, and Adjustment Factors Number of Number of Sample employees employees size of linked to not linked to Adjustment Sample employees" EHICS employers EHICS employers factor (Aj) Group 1 8,523 8,361 162 1.109 Group 2 267 198 69 15352 Group 3 384 216 168 1.781 Group 4 120 95 25 1.245 “As derived from employed key household survey participants. SOURCE: Lessler, McNeill and Waddell, 1980. where ° Sw ieL' > we teL A, = ieL' considers all (L') EHICS employed individuals and ieL considers only the employed EHICS participants linked (L) to an identified em- ployer. The sum of these adjusted sampling weights provides national estimates of the number of em- ployment positions held in 1977. Missing Multiplicity On the assumption that employees of both responding and nonresponding employers were similarly distributed, the employee sampling weights adjusted for missing employment status and employer linkage were further adjusted for missing employer multiplicity, i.e., the number ''35 of sample employees in a nonresponding firm. Adjustment factors were derived separately within each primary sampling unit (or within each pseudo-PSU; see Cohen and Kalsbeek, 1981) as defined in the household survey. Specifically, the EHICS person weights wit of employees linked to employers providing multiplicity information were adjusted by factor A», within each PSU (\ ). This resulted in the new employee weight wit such that wi - A, wi for linked employees with employers providing multiplicity information and = 0 otherwise where > fai jeM'» Bay = hai JeMnr is computed within each PSU/(A), and Lin) by ian, i=1 where Lin) Ss is the sum over all Litx) EHICS employees linked to the jth employer in i=1 PSU(A)s jeM'» considers all identified employers linked to sample employees in PSU(A), and jeMnr considers all identified employers linked to sample employees in PSU/(\) and responding to the multiplicity question. There were 6,956 employer records with multiplicity information and 254 records with missing multiplicity response. An employer with sample employees in S PSUs is represented by S em- ''36 ployer weights. The adjustment factors range from 1.000 to 1.755 and the data characterize 5,845 unique employers. EHICS employer weights were then constructed by determining the respective EHICS employees that were linked to a distinct employer within a particular PSU, summing the weights of all linked employees, and dividing this weight by the multiplicity of the employer. This weight, Wr, is expressed as hin) wM "ey = i=1 J where wit is the EHICS person weight adjusted for missing multiplicity, Lien) >» is the sum over all EHICS employees reporting the jth employer within the Ath: PSU, i=1 and vj is the jth employer's multiplicity. Missing EHICS Data Of 6,956 records for responding employers, 723 contained information only on multiplic- ity. Consequently, the final EHICS employer weights had to be further adjusted for question- naire nonresponse attributable to these 723 employer records. Assuming the EHICS employers who answered only the multiplicity question to be distri- buted similarly to fully responding employers of similar size, the EHICS weights of responding employers were adjusted to account for questionnaire nonresponse. These adjustments were derived within weighting classes (p) determined by 20 equal intervals on the multiplicity per- centile distribution, specified by increments of 5 percent, to control for number of employees. Consequently, the final employer EHICS sample weight, wi , took the form ''37 we =A ‘p we for employers responding to the EHICS questionnaire in addition to providing multiplicity information and = 0 otherwise where We is the EHICS employer weight adjusted for missing multiplicity and > vf J 2 jeQ'.p 3p . 2», we J JeQ.p jeQ’ considers all responding employers, jeQ considers responding employers excluding those reporting multiplicity only, and p considers all employers within the pth class(p=1,2,...,20)of the multiplicity distri- bution. It should be noted that the differential adjustments for multiplicity and nonresponse imparted noticeable variation to the final sampling weights. The application of these differen- tial sampling weights in the derivation of population estimates will generally increase variance. After adjusting for this type of questionnaire nonresponse, the EHICS analysis file con- tained 6,233 records which characterized 5,176 unique employers. An employer associated with EHICS sample employees in more than one pseudo-PSU would have a positive EHICS sampling weight for each PSU represented. The sum of these EHICS employer sampling weights yielded an estimate of 5,068,000 employers in the nation for 1977 and, when used in conjunction with the multiplicity data, yielded a national estimate of 97,131,000 employee positions. ''38 Summary The design of the Employer Health Insurance Cost Survey has been described in detail. Particular attention has been given to the sampling strategy that allowed for the selection of employers as a consequence of their linkage to key participants in the NMCES household sam- ple. The multiplicity framework for deriving national estimates of the health insurance costs of civilian employers and their employees in the U.S. for 1977 is also presented. In addition, detailed specifications for the computation of the EHICS sampling weights have been provided to ensure a better understanding of the estimation procedures. They include a description of the selection probabilities characterizing sampled EHICS employees for multiplicity estimation, and adjustments for nonresponse at both the employee and employer level. The final EHICS employer sample weights, Wi, allow for the examination of employment- related health expenditures for 1977 through the derivation of national estimates of the number of employees in firms with employment-related health insurance plans and those eligible for such plans; the proportion of payroll expenses going to health insurance; mean insurance premi- ums; the respective percentages of premiums paid by employers and employees; and the extent to which employers bear the entire premium cost. ''39 References Bonham, G.S. and Corder, L.S. 1981. National Health Care Expenditures Study, Instruments and Procedures 1: NMCES Household Interview Instruments. Department of Health and Human Services Publication No. (PHS) 81-3280. National Center for Health Services Research. Cohen, S.B. and Kalsbeek, W.D. 1981. National Health Care Expenditures Study, Instruments and Procedures 2: Estimation and Sampling Variances in the Household Survey. Depart- ment of Health and Human Services Publication No. (PHS) 81-3281. National Center for Health Services Research. Folsom, R.E. 1981. Health Insurance Employer Survey and Uninsured Validation Survey - Weight Adjustment Methodology. Research Triangle Park, N.C.: Research Triangle Institute (processed). Health Insurance Association of America. 1982. Source Book of Health Insurance Data, 1981- 1982. Washington, D.C.: HIAA. Lessler, J.T., McNeill, J.J., and Waddell, R.D. 1980. Employer Health Insurance Cost Survey Methodology Report. Research Triangle Park, N.C.: Research Triangle Institute (pro- cessed), Newhouse, J.P. 1974. A design for a health insurance experiment. Inquiry 11(1):5-27. Sirken, M.G. 1970. Household surveys with multiplicity. J Am Stat Assoc 65: 257-266, Taylor, A.K. and Lawson, W.R. 1981. National Health Care Expenditures Study, Data Preview 7, Employer and Employee Expenditures for Private Health Insurance. DHHS Publication No. (PHS) 81-3297. National Center for Health Services Research, Walden, D.C., Horgan, C.M., and Cafferata, G.L. 1982. Consumer Knowledge of Health Insur- ance Coverage. Paper presented at the Fourth Biennial Conference on Health Survey Research Methods, Washington, D.C. ''''A-1-1 National Medical Care Expenditure Survey Dear Street | am participating in a study being conducted for the United States Public Health Service to gather information about medical care expenditures, utilization of health services, and health insurance coverage. This information will benefit all Americans by enabling our public and private agencies to better understand the health of our people and our medical care system. My family is one of twelve thousand American households participating in this study. To supplement the information provided by families like my own, health insurance companies and the groups, employers, or unions that administer health insurance plans, are being asked to provide information about health insurance benefits, coverage, and pre- miums of sampled families. By this statement or by a facsimile of it, | hereby authorize and request that you furnish to the National Medical Care Expenditure Survey a copy of each health insurance policy | hold with your company or group and information about each policy’s premiums and coverage. Please cooperate with these researchers by providing them with the information requested on the following pages and a copy of each health insurance policy | hold with your company or group. If you have questions about this study that are not covered by the instructions in the enclosed booklet, or need assistance to complete this request, please call une of the coordinators for the National Medical Care Expenditure Survey at their toll-free number 1-800-225-1970. Please be assured that this information is subject to the Privacy Act of 1974 (Public Law 93-579) and is therefore strictly confi- dential. Personal information about survey participants will not be disclosed to any parties outside of this survey and will be destroyed upon completion of this study. Sincerely, Policyholder, Certificateholder, or Guardian of Person whose name appears on policy: Date Information contained on this form which would permit identifica- tion of any individual or establishment has been collected with a | | | | | guarantee that it will be held in strict confidence, will be used only for Participant |.D. Purposes stated for this study, and will not be disclosed or released to others without the consent of the individual or establishment in accor- dance with Section 308 of the Public Health Service Act (42 USC 242m). ave LILI : > PLEASE TURN TO PAGE 2 Appendix A-1. Health Insurance Permission Form '' NATIONAL MEDICAL CARE EXPENDITURE SURVEY Sponsored by US Public Health Service NS RESP ARCHCENTER ABT ASSOCIATES INC ORK CAMBRIDGE MASSACHUSETTS (17) 492 7100 Dear Friend® You have been asked to participate in the National Medical Care Expenditures Survey which is being conducted for the National Centers for Health Statistics and Health Services Research of the U.S. Public Health Service. The information collected in this survey will benefit all Americans by enabling our public and private agencies to better understand the health of our people and our medical care system. Because accurate insurance information cannot be collected from any other source, your help in completing this questionnaire is essential. The importance of your cooperation cannot be overstated. Please examine the identifying information for the health insurance policies presented here. For each policy, please answer all questions contained in this folder. Detailed instruc- tions for you to follow in answering these questions appear on pages 3 thru 5 of the attached booklet. When answering question 2 for group insurance policies, please do not enter the premium for the entire health insurance group. After you complete this questionnaire please be sure to enclose a copy of each insurance policy or group insurance certificate in the postage paid envelope that we have supplied. If you have questions about this survey or need assistance to complete this request, please call one of the coordinators for the National Medical Care Expenditure Survey at their toll-free number (1-800-225-1970). Sincerely, SRG Pew Donald N. Muse, Ph.D. Associate Project Director National Medical Care Expenditure Survey If no identifying numbers are available for one or more of the Policies listed on this form, the policyholder’s Social Security Number should PLEASE NOTE: Your Social Security Number is needed only to allow your health insurance policy to be easily and accurately located and identified. In accordance with the Privacy Act of 1974, provision of your Social Security Number is voluntary. The National Medical Care Expenditure Survey is being conducted under Section 306 of the Pub- lic Health Service Act (42 USC 242k). This is the authority for voluntary disclosure. Health Insurance Policy A Name of Insurance Company Policy # or Group # Certificate # Name of Policyholder or Certificateholder (as it appears on policy) Type of Policy O Group a Individual O Other (please specify) Health Insurance Policy B Name of Insurance Company _ Policy # or Group # Certificate # Name of Policyholder or Certificateholder (as it appears on policy) Type of Policy O Group OJ Individual O Other (please specify) Health Insurance Policy C Name of Insurance Company Policy# or Group # Certificate # Name of Policyholder or Certificateholder (as it appears on policy) Type of Policy (1 Group O) individual (aa Other (please specify) ''A-1-3 Al Has the indicated policy or certificate- holder been covered by this policy at any time during 1977? (Please check one box) EJ Yes-®] Is this a family, couple, or individual insurance policy? O Individual 0 Couple Family EE Other (please specify) C1 sno (Please SKIP TO POLICY B) O Other (please specify) coverage, please specify the 1977 annual premium, the payors, and the amount that each payor will contribute toward the total annual premium. 1. Total Annual Premium: Amount $ cee e CT) 2. Insured’s Contribution: Amount $ iq Tt] ° LU] 3. Other Payor’s Name(s) Ve ee nt Amount $ Ci [| ° LL i rn eee ieee a es Amount $ Crh) e LU A2 For this individual, couple, or family A3 Please enclose a copy of this policy or a group insurance certi- ficate in the return envelope with this questionnaire. O Please check this box to indicate policy enclosure. B1 Has the indicated policy or certificate- holder been covered by this policy at any time during 1977? (Please check one box) J Yes] Is this a family, couple, or individual insurance policy? (_] individual LJ Couple L) Family a Other (please specify) C1 no (Please SKIP TO POLICY C) O Other (please specify) __-__-__ coverage, please specify the 1977 annual Premium, the payors, and the amount that each payor will contribute toward the total annual premium. 1. Total Annual Premium: LTA) 2. Insured’s Contribution: LP.) 3. Other Payor’s Name(s) Amount $ Amount $ 1 gale $ [TT ° ae ae FEL) B2 For this individual, couple, or family B3 Please enclose a copy of this policy or a group insurance certi- ficate in the return envelope with this questionnaire. [3 Please check this box to indicate policy enclosure. c1 Has the indicated policy or certificate- ‘holder been covered by this policy at any time during 1977? (Please check one box) C2 For this 2 Yes-m] Is this a family, couple, or individual insurance policy? (J) Individual Couple Family Other (please specify) [1 no (Please TURN TO PAGE 4) (J Other (please specify). == individual, couple, or family coverage, please specify the 1977 annual Premium, the payors, and the amount that each payor will contribute toward the total annual premium. 1. Total Annual Premium: amounts {1 | 1).LLJ 2. Insured’s Contribution: L110). 3. Other Payor’s Name(s) Te —— Amount $ tT LU ° Li 2: — tee Amount $ LIT e ae Amount $ c3 Please enclose a copy of this policy or a group insurance certi- ficate in the return envelope with this questionnaire. EJ Please check this box to indicate policy enclosure. OMB #68-R1528 Expires 12/31/78 —_———> PLEASE TURN TO PAGE 4 ''A-1-4 4. During 1977, has the insured been covered by any other health insurance policy that you administered which was not listed on page 2? (Please check one box) @ yes—__——»j_—sif yes: please specify the insurance company that writes this policy, the policy and certificate numbers, any dependents who are covered by this policy, annual premium for 1977, the payment sources, and OC no amounts that each contributor has or will pay toward this annual premium. O don’t know 5. In case we need to contact you about this questionnaire, please indicate your name, title, address, and phone number. Name Title: Address: — City Sute Zip Coae prones_| | HLL LIL] TTY area code Thank you for completing this Questionnaire. Please return it and a copy of each insurance policy in the enclosed prepaid envelope addressed to: National Medical Care Expenditure Survey 55 Wheeler Street Cambridge, Massachusetts 02138 ''A-2-1 OMB No. 68-R1528 Dear National Medical Care Expenditure Survey P.O. Box 12138, Research Triangle Park, North Carolina 27709 Name Street Address City State Zip study. | am participating in a study being conducted for the United States Public Health Service to gather infor- mation about medical care expenditures, utilization of health services, and health insurance coverage. As part of the survey, any employer for whom | have worked and any union to which | belonged between January 1977 and April 1978, are being asked to supplement the information that | have provided to the April 1978. By this statement or a photocopy of it, | hereby authorize and request you to furnish to the National Medical Care Expenditure Survey any information, including copies of policies, re- quested in regard to health insurance plans or programs available to me from January 1977 to in such programs. Sincerely, Please cooperate with this study by providing the data requested on group health insurance plans or pro- grams that were in effect from January 1977 to April 1978, regardless of whether or not | participated Signature of employee or union member (If 14 or over, please sign above) Parent or Guardian (Please sign for person under 18) Name of employee or union member Date signed to (Dates of employment or membership, if other than from January 1977 to April 1978) Date signed Social Security Number of employee or union member NOTICE — Information contained on this form which would permit identification of any individual or establishment has been collected with a guarantee that it will be held in strict confidence, will be used only for purposes stated for this study, and will not be disclosed or released to others without the consent of the in- fividual or establishment in accordance with Section 308 of the Public Health Service Act (42 USC 242m). No information will be disclosed where prohibited by federal laws and regulations govern- ing the confidentiality of alcohol and drug sbuse patient records, 21 USC 1175 or 42 USC 4582, 42 CFR Part 2. Reporting Unit # (TLL Participant ID # (TTT Permission Form # NOTICE — Your Social Security Number is needed only to allow your employer or union to accurately identify and locate your employment or membership records. In accordance with the Privacy Act of 1974, provision of your Social Security Number is voluntary. The National Medical Care Expenditure Survey is being conducted under Section 306 of the Public Health Service Act (42 USC 382k). This is the authority for voluntary disclosure. Appendix A-2. Unemployed Verification Survey Permission Form ''''A-3-1 NATIONAL MEDICAL CARE EXPENDITURE SURVEY HEALTH INSURANCE QUESTIONNAIRE Questionnaire Conceming: Person: ID: 1. At any time between January 1, 1977 and December 31, 1977, was the above named person an employee of your organization or a member of your union? Yes (oo) at No nt E2 Please skip to question 12 Don’'tknow ( )94 2. During this time period, did your organization administer or provide any health or dental in- surance policies to your employees or union members? Yes ()2 No ( ) 2 Please skip to question 12 Don’tknow ( ) 94 wee 3. Between January 1, 1977 and December 31, 1977, was the above named person e/igib/e to either purchase or enroll in any of these health or dental insurance policies? Yes (ac) No () 2 Please skip t estion 12 Don’tknow ( ) 94 oo Other ( ) «——+® Please specify and skip to question 10 4. At any time between January 1, 1977 and December 31, 1977, was the above person covered by any of your organization's health or dental insurance policies? Yes ee No is Please skip to question 10 Don’'tknow ( )94 lc Other (_ ) «—» Please specify and skip to question 10 Information contained on this form which would permit identification of any individual or establishment has been collected with @ guarantee that it will be held in strict confidence, will be used only for purposes stated for*this study, and will not be disclosed or released to others without the consent of the individual or establishment in accordance with Section 308(d) of the Public Health Service Act (42 USC 242m). OMB No. 68-R1528 Appendix A-3. Unemployed Verification Survey Questionnaire ''A-3-2 5. For the policy which covered the previously mentioned person, please indicate the name of the insurance company or the prepaid practice. 6. Was this an individual, couple, or family insurance policy? Individual | Couple ( )2 Family ( )3 Other (_ ) « ——+ Please specify and continue to question 7 7. For the policy being discussed, please specify the 1977 annual premium, who pays the premium and the amount that each source contributes toward the total annual premium. A) Total annual premium: Amount $ Don't know ( )o« B) What your employee (union member) paid: Amount $ Don’t know ( )o9« C) Other persons or organizations that help pay the premium: Amount S| | | ft | Don't know ( )94 Amount $ Don’t know ( )o« 8. How many employees (or union members) were covered at the end of the 1977 plan year by this plan? Include employees (or union members) covered in all of your company or union locations. No. Covered: 9. Please enclose a group insurance certificate or a copy of the policy described above and check here. ( ) 1———— Please continue to Question 10. ''A-3-3 10. Through your organization between January 1, 1977 and December 31, 1977, could the previously named person voluntarily have purchased an option (or a different option) of health or dental 1. benefits (such as major medical, high or low option, dental, drug, prepaid group practice) through additional riders, or among alternative plans? Yes () No (1) 2 ( ) Don’t know 94 Please skip to question 12 Please answer questions A - F below for each of the different options or plans this person could have purchased or enrolled in between January 1, 1977 and December 31, 1977, but did not. OPTION OR PLAN # 1 OPTION OR PLAN # 2 Please indicate the name of the insurance company or prepaid practice. Name: What type of option or plan was that (such as major medical, high or low op- tion, dental, drug, prepaid group prac- tice)? Please specify the 1977 total annual premium or enrollment cost for... Individual coverage: $ Farnily coverage: $ e Please specify the 1977 annua/ premium or enrollment cost which the person named on the label would have had to contribute for... Individual coverage: $ Pelco the | 7 [ Family coverage: $ » How many employees (or union mem- bers) were covered at the end of the 1977 plan year by this option or plan? Include employees (or union members) covered in all of your company or union locations. No. covered: Please enclose a group insurance certifi- cate or a copy of the policy described above and check here. C] 1 A. Please indicate the name of the insurance company or prepaid practice. Name: . What type of option or plan was that (such as major medical, high or low op- tion, dental, drug, prepaid group prac- tice)? . Please specify the 1977 total annual premium or enrollment cost for... Individual coverage: $ Family coverage: $ e . Please specify the 1977 annua/ premium or enrollment cost which the person named on the label would have had to contribute for... Individual coverage. $ | u Family coverage: $ : How many employees (or union mem- bers) were covered at the end of the 1977 plan year by this option or plan? Include employees (or union members) covered in all of your company or union locations. No. covered: . Please enclose a-group insurance certifi- cate or a copy of the policy described above and check here. CT 1 ''A-3-4 12. In case we need to contact you, please indicate your name, title, address, and phone number. Name: Title: Address: street city state zip code Phone: bi an area code extension Thank you for completing this questionnaire. Please return it and a copy of each insurance policy you have described in the enclosed prepaid envelope addressed to: National Medica! Care Expenditure Survey 55 Wheeler Street Cambridge, Massachusetts 02138 If you have questions or need additional forms, please cal! our toll free number (1-800-225-1970). ''Health Insurance Options Survey A-4-1 OMB No. 68 - R1528 The person listed below has indicated that he was covered during 1977 by those work-related health or dental plans listed on the label. Please answer questions 1 - 3 below regarding this person's coverage. 1. How many employees (or union members) were covered at the end of the 1977 plan year by each plan listed on the label? Include Person: employees (or union members) covered in all Work-related plans listed in the gray booklet as held dur- of your company or union locations. ing 1977: Plan A oe Pian A Plan B (if any) Pian B Plan C (if any) Pian C . Through your organization in 1977, could the above named person have purchased a different option of health or dental benefits (such as major medical, high or low option, dental, drug, prepaid group practice) either within the plan(s) listed above, through additional riders, or among alternative plans? (Do not consider the choice between in- dividual and family coverage as a different option.) . Yes ( )——® Please answer question 3 below and return this survey with a copy of each certificate/ Policy described below in the prepaid envelope. No Ain} 2 Please return this questionnaire in the enclosed prepaid envelope. Do not complete Don't Know ( Se question 3 below. Please answer questions A - F below for each of the different options or plans this person could have purchased in 1977, but did not. OPTION OR PLAN #1 Do not list option(s) shown on the label. OPTION OR PLAN #2 Do not list option(s) shown on the label. Please indicate the name of the insurance company or prepaid practice. Name: What type of option or plan was that (such as major medical, high or low option, dental, drug, prepaid group practice)? Please specify the 1977 total annual premium or enroliment cost for... Individual coverage: s_ TT TILT) TTT) Please specify the 1977 annual premium or enrollment cost which the person named on the label would have had to contribute for... Individual coverage: s| TT TILT) How many employees (or union members) were covered at the end of the 1977 plan year by this option or plan? Include em, ployees (or union members) covered in all of your company or union locations. No. covered: Family coverage: Please enclose a group insurance certificate or a copy of the policy described above and check here. [] Please indicate the name of the insurance company or prepaid practice. Name: What type of option or plan was that (such as major medical, high or low option, dental, drug, prepaid group practice)? Please specify the 1977 total annual premium or enrollment cost for... Individual coverage: s [TLIC] TTD Please specify the 1977 annual premium or enrollment cost which the person named on the label would have had to contribute for... Individual coverage: s_T TTI) reve ETC How many employees (or union members) were covered at the end of the 1977 plan year by this option or plan? Include em- ployees (or union members) covered in all of your company or union locations. No. covered: Family coverage: Please enclose a group insurance certificate or @ copy of the policy described above and check here. CJ , If you have questions or need additional forms, please call our toll free number (1-800-225-1970). Appendix A-4. Health Insurance Options Survey Questionnaire ''''B-1-1 OMB No. 85-S-78016 Expires 03/31/79 Information contained on this form which would permit identification of any individual or establishment has been collected with ® guarantee that it will be held in strict confidence, will be used only for purposes stated for this study, and will not be disclosed or released to others without the consent of the individual or establishment in accordance with Section 306 of the Public Health Service Act (42 USC 242m). EMPLOYER HEALTH INSURANCE COST SURVEY Please answer the following questions as accurately as possible, as specific answers are essential in our attempt to estimate the cost of health insurance to American business. Please return in the envelope enclosed. Thank you for your time, help, and cooperation. Project Address: Research Triangle Institute, P.O. Box 12194, Research Triangle Park, N.C. 27709 1. How many persons did you employ as of DECEMBER 31, 1977? Please include the total number of employees at all locations and/or branches of your organization. (If these figures are unavailable, please base your figures on those from the end of your most recent fiscal year, or as of last month, and check the appropriate bax below.) TOTAL‘EMPLOYEES ou = FULL-TIME EMPLOVEES, — PART-TIME EMPLOYEES (1) (| Figures as of 12/31/77 ~— (2) [ ] Figures from fiscal year ending (3) [ ] For the month of 2. Please provide a further breakdown of your employees in the chart below. Full-Time Employees Part-Time Employees Salaried Hourly TOTAL Salaried Hourly TOTAL Permanent Temporary TOTAL 3. Are any of these union employees? (1) [ ] Yes (2) [ ] No 4. Equivalent hourly, weekly, monthly, and annual wage rate categories are shown below. Please provide a breakdown of the number of employees (both full-time and part-time) that you listed in Question 1 into these rate categories. HOURLY RATE WEEKLY RATE MONTHLY RATE ANNUAL RATE NUMBER OF EMPLOYEES Under $2.65 Under $106 Under $445 Under $5,329 $2.65 - $2.89 $106 - $115 $445 - $486 $5,329 - $5,843 $290 - $3.24 $116 - $129 $487 - $544 $5,844 - $6,528 $3.25 - $3.74 $130 - $149 $545 - $628 $6,529 - $7,536 Over $3.75 Over $150 Over $628 Over $7,537 5. What was your total annual payroll for your employees for 1977? (If your annual figures are unavailable, please provide figures for your most recent fiscal year or last month’s total payroll, and check the appropriate box below.) TOTALPAYROLL $ (1) [| For 1977 (2) [| For the fiscal yearending _.—==—s——s« (3) [| For the month of Appendix B. EHICS Questionnaire ''B-1-2 6. What was the total annual premium, regardless of who paid, for the health insurance benefits provided by your company in 1977? (If these figures are unavailable, please base your figures on those from the end of your most recent fiscal year, or last month’s total premium, and check the appropriate bax below.) TOTAL ANNUAL PREMIUM $ (1) [|For 1977 (2) [|For the fiscal yearending _.________ (3) [|For the month of 7. How much of the amount in Question 6 was contributed by your company or organization, end how much was contributed by your employees? (A) EMPLOYER'S CONTRIBUTION $ (B) EMPLOYEES'CONTRIBUTION $ 8. Does your premiunt cover health care only or does it also include other benefits, such as life insurance or disability insurance? (1) Zz Health care only — (2) [ | includes other benefits in addition to health care 9. Does the contribution per employee paid by your company or organization for health insurance vary by the employee's salary level? mL] ve aL] wo 10. Are all your employees eligible to participate in your health insurance plan? (1) [| Yes (2) [ | No -— (Please check all categories of employees eligible to participate in your health insurance plan. Indicate whether only salaried or only hourly employees in each category are eligible or whether both hourly and salaried employees are eligible.) (1)]_]Permanent Full-Time Employees —~ (1) [ ] Salaried (2) [ ] Hourly (3) B Both 1) ] Permanent Part-Time Employees — (1) 4 Salaried (2) [ ] Hourly (3) [| Both (0[__} Temporary Full-Time Employees — (1) & Salaried (2) Ld Hourly (3) [ ] Both (0[ | Temporary Part-Time Employees — (1) 4 Salaried (2) [ ] Hourly (3) [] Both tn] Jother (Specify) In case we need to contact you about this questionnaire, please indicate your name, title, address, and phone number. me OD -O00-o00o0 | ; | PLEASE RETURN IN THE ENVELOPE ENCLOSED | #U.S. GOVERNMENT PRINTING OFFICE : 1984 0-421-787/422 '' ''''$9272 101_ REPORT DOCUMENTATION |}. REPORT NO. 2. PAGE NCHSR 84-74 3. Recipient's Accession No. 4. Title and Subtitie NATIONAL HEALTH CARE EXPENDITURES STUDY INSTRUMENTS AND PROCEDURES 3, ESTIMATION AND SAMPLING PROCEDURES IN THE NMCES INSURANCE SURVEYS 5. Report Date 1984 6. 7. Author(s) NCHSR Staff 8. Performing Organization Rept. No. 9. Performing Organization Name and Address Division of Intramural Research National Center for Health Services Research DHHS, PHS, OASH, 3-50 Park Building 5600 Fishers Lane Rockville, MD 20857 10. Project/Task/Work Unit No. 11. Contract(C) or Grant(G) No. (Cc) N.A. (G) 12. Sponsoring Organization Name and Address DHHS, PHS, OASH, National Center for Health Services Research Publications and Information Branch, 1-46 Park Building 5600 Fishers Lane 13. Type of Report & Period Covered Survey report Rockville, MD 20857 Tel.: 301/443-4100 14. 1S. Supplementary Notes DHHS Publication No. (PHS)84-3369. 16. Abstract (Limit: 200 words) Describes the design and weighting procedures developed for the insurance surveys of the National Medical Care Expenditures Study (NMCES), which produced national estimates of the insurance status and benefits of the civilian, noninstitutionalized U.S. population in 1977. Information on health insurance coverage obtained in the household survey was verified and supplemented in the Health Insurance/Employer Survey (HIES) and two sub- studies, the Uninsured Validation Survey (UVS) and the Health Insurance Options Survey (HIOS). Taking household-reported public coverage into account, information similar to that obtained for private policies was obtained for benefits offered by Medicare, CHAMPUS/CHAMPVA, and the different state Medicaid programs. The HIES data base thus encompasses public as well as private health insurance benefits. The Employer Health Insurance Cost Survey (EHICS) was designed 9 provide national estimates for 1977 of the cost of health insurance to employers and of some characteristics of employees eligible for employment related health insurance. cost. Procedures permit derivation of national estimates of the number of employees in firms with employment-related health insurance plans and those eligible for such plans, the proportion of payroll expenses going to health insurance, mean insurance premiums, the respective percentages of premiums paid by employers and employees, and the extent to which employers bear the entire premium 17. Document Analysis ». Descriptors NCHSR publication of research findings does not necessarily represent approval or official endorsement by the National Center for Health Services Research or the Usd. Department of Health and Human Services. b. Identifiers /Open-Ended Terms Health services research, National Medical Care Expenditures Survey, insurance survey, interview instruments and procedures c. COSATI Field/Group 16. Availability Stetemen. 19. Security Cless (This Report) 21. No. of Pages Releasable to the public. Available from National Unclassified Technical Information Service, Springfield, VA 20. Security Clees (This Page) >. Price 22161 Tel.: 703/487-4650 Unclassified (See ANS!-Z39.18) See Instructions on Reverse OPTIONAL FORM 272 (4-77) (Formerly NTIS—35) Department of Commerce '' U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Public Health Service National Center for Health Services Research 1-46 Park Building; 5600 Fishers Lane Rockville, MD 20857 National Center for Health Services Research John E. Marshall, Director Robert A. Fordham, Associate Deputy Director, State and Local User Liaison Donald E. Goldstone, Associate Deputy Director, Medical and Scientific Affairs Division of Intramural Research Donald E. Goldstone, Acting Director Daniel C. Walden, Senior Research Manager Renate Wilson, Editorial Associate DHHS Publication No. (PHS) 84-3369 Pooad ''