1                                                                                                                                    
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                    National Medical Care Expenditure Survey                                                                         
                                                                                                                                     
                     Health Insurance/Employer Survey Data:                                                                          
                                                                                                                                     
                           Person Record EBCDIC Files                                                                                
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                  National Center for Health Services Research                                                                       
                      and Health Care Technology Assessment                                                                          
                          Mail Stop 3-50, Park Building                                                                              
                                5600 Fishers Lane                                                                                    
                              Rockville, MD  20857                                                                                   
                                 (301) 443-4836                                                                                      
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
1                                                                                                                                    
                                                                                                                                     
                          Public Use Tape Documentation                                                                              
                                                                                                                                     
                                       for                                                                                           
                                                                                                                                     
                          Private Insurance Status and                                                                               
                   Premium Data for the U.S. Population, 1977                                                                        
                                                                                                                                     
                                       and                                                                                           
                                                                                                                                     
                     Private Insurance Benefit Data for the                                                                          
                        Population Age 65 and Older, 1977                                                                            
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                   Prepared by                                                                                       
                                                                                                                                     
                    G.L. Cafferata, P.J. Farley, S.B. Cohen,                                                                         
                           R. Wilson, and J.C. Cantor                                                                                
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                   March 1986                                                                                        
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
 The contributions of Cathy Doran and Ase Sewall of Social & Scientific                                                              
 Systems, Inc., and of William Ashby of NCHSR in the preparation of the                                                              
 codebooks and documentation are gratefully acknowledged.                                                                            
                                                                                                                                     
                                                                                                                                     
1     This documentation describes the third set of public use data to be                                                            
 released from the 1977 National Medical Care Expenditure Survey (NMCES), a                                                          
 project of the Division of Intramural Research of the National Center for                                                           
 Health Services Research and Health Care Technology Assessment (NCHSR) that                                                         
 was cosponsored with the National Center for Health Statistics.  The household                                                      
 component of NMCES obtained data from a national sample of the civilian,                                                            
 noninstitutionalized population on their insurance status and on health                                                             
 services use and expenditures for 1977.  These data were complemented by a                                                          
 survey of employers and insurance companies responsible for the insurance                                                           
 coverage of household survey members, the Health Insurance/Employer Survey                                                          
 (HIES).                                                                                                                             
      The private health insurance data obtained in the HIES for persons in the                                                      
 household component are being released as separate tapes containing several                                                         
 Extended Binary-Coded-Decimal Interchange Code (EBCDIC) data sets.  The                                                             
 present tape, of which this documentation is File 1, contains two of these                                                          
 sets:                                                                                                                               
                                                                                                                                     
      Data Set 1 (File 3) provides one record for each of the 40,320                                                                 
      respondents in the NMCES household component on private health insurance                                                       
      status and related information in 1977 as verified in HIES, and including                                                      
      data on premiums and sources of premium payment for those with private                                                         
      insurance.                                                                                                                     
                                                                                                                                     
      Data Set 2 (File 5) provides one record and related information for each                                                       
      of 1,966 household respondents 65 years or older with private insurance                                                        
      in 1977 as verified in HIES, as well as information on the services                                                            
      covered by their private insurance benefits.                                                                                   
                                                                                                                                     
      A third HIES data set, which will be released as a separate public use                                                         
 tape, contains data describing the private insurance benefits of each                                                               
 privately insured person under 65.  The benefits of the elderly and of the                                                          
 population under 65 are described separately because of the substantial                                                             
 differences in their private insurance.  These are due not only to nearly                                                           
 universal Medicare coverage of the elderly U.S. population, much of whose                                                           
 private insurance is specifically designed to cover Medicare gaps, but also to                                                      
 the differences in the availability of employment-related health insurance.                                                         
      Two previous public use data sets have provided (1) person-level data on                                                       
 sociodemographic characteristics, health status, access to care, household-                                                         
 reported health insurance coverage, and use, expenditures, and sources of                                                           
 payment for medical and related services (NTIS No. PB 83-199539; Kasper,                                                            
 Walden and Wilson, 1985) and (2) event-level data on use and related                                                                
 conditions, expenditures, and sources of payment for this population (NTIS                                                          
 Nos. PB 85-246635 and PB 85-246643; Walden and Wilson, 1985).                                                                       
      Like the person file from the household component, the premium data set,                                                       
 in conjunction with the appropriate HIES sampling weights, represents the entire                                                    
 civilian noninstitutionalized U.S. population in 1977.  However, the resulting                                                      
 estimates are based on a smaller sample, i.e., the population for whom                                                              
 responses were obtained in HIES.  Nonetheless, sociodemographic                                                                     
 characteristics, use of health services and expenditures, and related items                                                         
 from the person and event-level public use tapes from the NMCES household                                                           
 component can be used in conjunction with health insurance data for each                                                            
 person on the HIES premium data set, which also identifies persons by their                                                         
 household and permits analyses involving the characteristics and insurance                                                          
 status of all family members.  Analogous linkages can be performed for the                                                          
                                                                                                                                     
                                                                                                                                     
1data set that describes the private insurance benefits of the elderly 65 and                                                        
 older.                                                                                                                              
      This documentation, which is file 1 on the tape, consists of the                                                               
 following:                                                                                                                          
                                                                                                                                     
      a description of NMCES (Section 1)                                                                                             
                                                                                                                                     
      data set description and variable definitions (Section 2)                                                                      
                                                                                                                                     
      information on sample design, construction of population weights, and                                                          
      methods of variance estimation (Section 3)                                                                                     
                                                                                                                                     
      information on imputation of missing data and related procedures                                                               
      (Section 4)                                                                                                                    
                                                                                                                                     
      a description of file structure and contents, including codebooks                                                              
      (Section 5)                                                                                                                    
                                                                                                                                     
      programming information (Section 6)                                                                                            
                                                                                                                                     
                                                                                                                                     
 1.   The National Medical Care Expenditure Survey                                                                                   
                                                                                                                                     
      The NMCES was conducted to collect data for the National Health Care                                                           
 Expenditures Study of the Division of Intramural Research, NCHSR.  The survey                                                       
 provided an information base for studying a wide range of national health                                                           
 policy issues relating to patterns of health services use and expenditures,                                                         
 including the following:                                                                                                            
                                                                                                                                     
      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.                                                                                         
                                                                                                                                     
      The breadth and depth of health insurance coverage of the U.S.                                                                 
      population.                                                                                                                    
                                                                                                                                     
      The proportion of health care costs paid by various insurance mechanisms.                                                      
                                                                                                                                     
      The influence of the Medicare and Medicaid programs on the use and costs                                                       
      of medical care.                                                                                                               
                                                                                                                                     
      How and why Medicaid participation changes over time.                                                                          
                                                                                                                                     
      The cost and effectiveness of different federal, state, and local                                                              
      programs aimed at improving access to care.                                                                                    
                                                                                                                                     
      The loss of national revenue resulting from tax allowances for medical                                                         
      and health insurance expenses, particularly with regard to the benefits                                                        
      accruing to different categories of individuals and employers, and the                                                         
      potential effects on the federal budget of changes to tax laws.                                                                
                                                                                                                                     
      How costs of care vary for different diagnostic categories in various                                                          
      treatment settings.                                                                                                            
                                                                                                                                     
                                                                                                                                     
                                                                                                                                     
1     Funding for NMCES was provided by NCHSR, which cosponsored the survey                                                          
 with the National Center for Health Statistics, both agencies of the Office of                                                      
 the Assistant Secretary for Health, Department of Health and Human Services.                                                        
 Data collection for the survey was done by Research Triangle Institute, North                                                       
 Carolina, and its subcontractors, National Opinion Research Center of the                                                           
 University of Chicago, and Abt Associates, Inc., of Cambridge, Massachusetts,                                                       
 under contract HRA 230-76-0268.  Data processing support was provided by                                                            
 Social & Scientific Systems, Inc., of Bethesda, Maryland, under contracts                                                           
 233-79-3022, 233-80-3012 and 282-83-3001.                                                                                           
                                                                                                                                     
                                                                                                                                     
 1.1  The Household Interview Survey                                                                                                 
                                                                                                                                     
      A panel of about 14,000 households containing a total of 40,320 persons                                                        
 was interviewed in six separate rounds during 1977 and early 1978.  The survey                                                      
 reference period was January 1 to December 31, 1977.  The first interviews                                                          
 began in mid-January 1977; subsequent rounds were conducted at intervals of                                                         
 about three months.  The first, second, and fifth rounds of interviews were                                                         
 conducted in person, as were about 20 percent of the third and fourth rounds                                                        
 and about half of the sixth round; the remainder were conducted by                                                                  
 telephone.  For the entire set of household interview instruments, see NHCES                                                        
+                                                                       _____                                                        
 Instruments and Procedures, 1 (Bonham and Corder, 1981).                                                                            
+_____________________________                                                                                                       
      During each of five rounds of interviews, information was obtained on a                                                        
 set of care indicators of use of health services, charges for services and                                                          
 sources of payment, numbers and types of disability days, and status of health                                                      
 insurance coverage.  At various stages during the second through fifth round,                                                       
 supplementary questionnaires were administered on employment status and type,                                                       
 health insurance, access to care, race/ethnicity, and income and assets.  In                                                        
 the sixth round of interviewing, a final series of questions were asked on                                                          
 chronic limitations of activity or impairments, income tax filing status, and                                                       
 the amount of itemized medical deductions for 1977.                                                                                 
                                                                                                                                     
                                                                                                                                     
 1.2  The Health Insurance Employer Survey                                                                                           
                                                                                                                                     
      The Health Insurance/Employer Survey (HIES) verified and supplemented                                                          
 information reported by household respondents on all private health insurance                                                       
 coverage, excluding only coverage limited to "dread diseases" (e.g. stroke or                                                       
 cancer), "extra" cash coverage for hospital stays, and reimbursement of                                                             
 medical expenses for accidents or injuries under auto or homeowner insurance.                                                       
      For each private health insurance policy reported in the household                                                             
 survey, permission was asked from the primary insured to contact the                                                                
 respective insurance company or group (e.g. employer, union, fraternal                                                              
 organization) through which the policy was obtained.  In addition, permission                                                       
 was requested of all employed but not self-employed persons in the household                                                        
 survey who reported not being insured through their firm of employment, in                                                          
 order to verify lack of coverage with their employers in the Uninsured                                                              
 Validation Survey (UVS), a substudy of the HIES.  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.  For the entire set of health insurance                                                           
 interview instruments and a more detailed description of the HIES design, see                                                       
 NHCES Instruments and Procedures, 3 (Cohen and Farley, 1984, and Attachment 1                                                       
+___________________________________                                                                                                 
 to the paper copy of this document).                                                                                                
                                                                                                                                     
1     Permission forms authorizing contact with employers and insurance                                                              
 carriers in HIES were obtained from approximately 90 percent of household                                                           
 respondents who were the primary insured and from approximately 70 percent of                                                       
 household respondents eligible for the UVS; they were mailed to employers,                                                          
 insurers, and insurance groups between October 1978 and August 1979.                                                                
 Responses were obtained for 85 percent of household members who had signed                                                          
 permission forms.                                                                                                                   
      In order to obtain the specific insurance benefits of NMCES participants,                                                      
 the policies and certificates describing any insurance plan providing coverage                                                      
 to a household sample member 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, 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 respondents with any abstract information.  Copies of the                                                                 
 abstracting forms are available from NCHSR upon request.                                                                            
      The variables constructed from the HIES are derived from all insurance                                                         
 plans covering each eligible person at any time during 1977.  Multiple                                                              
 policies were obtained for about 15 percent of persons in the sample.  Since                                                        
 dates of enrollment were not identified, it was impossible to distinguish                                                           
 coverage that was held simultaneously, and was essentially additive, from                                                           
 coverage that was held at different times during 1977 and was thus not                                                              
 additive.  After reviewing the data, including changes in the employment of                                                         
 primary insured persons with multiple plans, it was determined that most of                                                         
 the coverage was simultaneous; accordingly, premiums were summed and benefits                                                       
 combined based on standards for coordination of benefits.  However, in some                                                         
 instances this overstates both coverage and premiums.                                                                               
                                                                                                                                     
                                                                                                                                     
 2.   Data File Contents                                                                                                             
                                                                                                                                     
 2.1  Insurance status, premiums, and sources of payment (data set 1)                                                                
                                                                                                                                     
      This data set includes all persons ever in the NMCES household survey and                                                      
 identifies those with private insurance as verified in the HIES.  (The private                                                      
 insurance indicators on the person file from the household component are                                                            
 unverified data reported by the household.)  Here, a person is defined as                                                           
 privately insured if coverage at any time in the survey year by at least one                                                        
 plan reported in the household survey was confirmed in the HIES, or an                                                              
 employer contacted in the Uninsured Validation Survey (UVS) reported private                                                        
 insurance not reported in the household survey.  A person with private                                                              
 insurance according to the household survey was defined as having no private                                                        
 insurance only if it was determined from valid responses in the HIES that the                                                       
 person was not in fact covered by any of the plans previously reported.                                                             
 Similarly, a person without private insurance according to the household                                                            
 survey but eligible for the UVS was defined as not having private insurance                                                         
 only if all employers responded and confirmed lack of employment related                                                            
 coverage.                                                                                                                           
      All other persons eligible for the HIES but whose private insurance                                                            
 status could not be established from responses to the follow-up survey were                                                         
                                                                                                                                     
                                                                                                                                     
1labeled as "No HIES data" and assigned coverage weights equal to zero; a                                                            
 record for them is on the file, but there are no data except identifying                                                            
 variables (person and household identification numbers, relationship to                                                             
 household head) and variables related to weights (including the key                                                                 
 participant indicator).  Persons eligible for the NMCES household survey but                                                        
 ineligible for the HIES, i.e, who had no household-reported private insurance                                                       
 and who were either unemployed or self-employed, were classified as having no                                                       
 private insurance but assigned positive HIES weights and suitable values of                                                         
 relevant HIES variables, so as to permit the derivation of national estimates                                                       
 of insured and uninsured persons from the HIES data.                                                                                
                                                                                                                                     
                                                                                                                                     
 2.1.1  Identification and family status                                                                                             
                                                                                                                                     
      The first variable in this data set (PID; 10.10) identifies all persons                                                        
 on this file by their unique NMCES person identification number.  Next, to                                                          
 take into account dependent coverage and the purchase of insurance on a family                                                      
 basis, the premium data set uniquely assigns all persons to a family or                                                             
 household through the variables RUIDHIES (10.11) and establishes their                                                              
 position within that household through the variable RELHEADF (10.12).  At the                                                       
 outset of the household survey, persons within a dwelling unit who were                                                             
 related by blood, marriage, or adoption were designated as a household and as                                                       
 a reporting unit (RU).  College students living away from a family in a sample                                                      
 dwelling unit were eligible for the NMCES survey but were designated as                                                             
 separate RUs.  As key participants left their original RU and the composition                                                       
 of the households changed, new RUs were designated and persons were reassigned                                                      
 to households in each round of the household survey.  The variable RUIDHIES is                                                      
 defined as each person's RUID (household identification number) in the last                                                         
 round of participation in the household survey.  Two exceptions take into                                                           
 account coverage obtained by dependents under the family plans of a primary                                                         
 insured who was not in the same RU as the dependent at the end of their                                                             
 participation in the survey.  First, insured college students who were not a                                                        
 primary insured, i.e., had no private health insurance of their own, were                                                           
 assigned to the same RUIDHIES as the primary insured and other family members                                                       
 covered by the same insurance.  Second, privately insured persons in reporting                                                      
 units as defined at the end of the year that did not include a primary insured                                                      
 were reassigned to an earlier reporting unit that did contain a primary                                                             
 insured.  Because RUIDHIES ignores the changing composition of families during                                                      
 the year in order to link persons uniquely to specific families, the primary                                                        
 insured and all dependents associated with a particular policy may not always                                                       
 share the same value of RUIDHIES.  (Note that this is different from the person                                                     
 file from the household survey component, where people are assigned to households                                                   
 on a quarterly basis and an end-of-survey basis using the last known family; see                                                    
 Kasper, Walden and Wilson, 1983.)                                                                                                   
      For persons covered as dependents under a family or couple policy, a                                                           
 final identification variable links all privately insured persons on this file                                                      
 to the primary insured (PRMINPID; 10.13) on the covering policy by means of                                                         
 the latter's NMCES person identification number.  Note that in the records of                                                       
 all primary insured persons, this variable is identical to the person                                                               
 identification number (PID).  For persons privately insured as a dependent on                                                       
 the policies of more than one primary insured, PRMINPID is assigned the                                                             
 identification number of the primary insured with the lowest value of                                                               
 RELHEADF.                                                                                                                           
                                                                                                                                     
                                                                                                                                     
12.1.2  Private insurance status and characteristics                                                                                 
                                                                                                                                     
      A person-level indicator of private insurance status is provided                                                               
 (INSPRVP; 11.10), as well as an indicator of primary insured status (PRIMINSD;                                                      
 12.10).  A separate indicator of private insurance coverage of any family                                                           
 member (INSPRVF; 13.10) is shown on the records of household heads.  These                                                          
 indicators relate to three levels of variables incorporated in this data                                                            
 set.  The first (11.10 to 11.33) provides person-level information on                                                               
 insurance status, plan type(s), and related characteristics; the second (12.10                                                      
 to 12.33) provides analogous information on primary insured persons only; the                                                       
 third (13.10 to 13.33) represents the insurance status of families.  The                                                            
 family-level variables are valid on the records of household heads                                                                  
 (RELHEADF=10) only.  The use of family-level variables from these records in                                                        
 conjunction with the family-level weight (WTINSFAM; 14.11) is equivalent to                                                         
 using a file where the unit of analysis is the family; the family weight                                                            
 contains positive values only on the records of household heads and it sums to                                                      
 the average number of families in the U.S. during 1977.                                                                             
      Variables indicating whether the insurance was employment related and                                                          
 derived from a group policy are provided at the level of each privately                                                             
 insured (11.11-11.121) and at the primary insured level (12.11-12.121).  For                                                        
 each privately insured person, they reflect all plans covering that person,                                                         
 whether as the primary insured or as a dependent under a family member's                                                            
 plan.  For each primary insured, they reflect only those plans obtained in                                                          
 that person's name.                                                                                                                 
      The indicator for employment-related health insurance was constructed                                                          
 from information about the administration of the insurance plan and the                                                             
 payment of all or part of the premiums by employers or unions.  Nongroup                                                            
 insurance was considered work-related only if an employer or union at least                                                         
 contributed to the premium.  Where insurance was obtained through a group, the                                                      
 size of the group is indicated as well, defined as the largest number of                                                            
 employees or union members enrolled in the group plan.  For those covered by                                                        
 more than one plan, the group size of the largest plan is used.  Furthermore,                                                       
 for the primary insured only, indicators are provided of type of coverage                                                           
 (individual, couple, or family) under both group and nongroup plans (12.13-                                                         
 12.14).                                                                                                                             
                                                                                                                                     
                                                                                                                                     
 2.1.3  Premiums                                                                                                                     
                                                                                                                                     
      Premium data are again provided at three levels:                                                                               
                                                                                                                                     
      Person-level premiums, which are suitable for analyses requiring a                                                             
      measure of the implicit cost of each person's insurance, even though the                                                       
      actual premium may have been a single figure covering everyone included                                                        
      in a family plan.  Here, the premium associated with each plan was                                                             
      divided by the number of family members covered by that plan and assigned                                                      
      equally to each person.  For persons covered by more than one plan, the                                                        
      premiums for each plan were summed to arrive at a person-level annual                                                          
      total (11.20-11.33).                                                                                                           
                                                                                                                                     
      Primary insured premiums, which assign the entire premium to the primary                                                       
      insured under each plan, regardless of the number of persons covered,                                                          
      summing across all plans designating that person as the primary                                                                
      insured.  This approach resembles more closely the way health insurance                                                        
                                                                                                                                     
                                                                                                                                     
1     is actually sold and would be more suitable, for example, to analyses of                                                       
      employer contributions to health insurance premiums (12.20-12.33).                                                             
                                                                                                                                     
      Family premiums, which are the sum of person-level premiums across all                                                         
      household members (13.20-13.33).  Use of family premium variables is                                                           
      appropriate where the family is the relevant unit of analysis.                                                                 
                                                                                                                                     
      Within each type, premiums for nongroup insurance and group insurance are                                                      
 identified separately, as are respective sources of payment.  An annual                                                             
 premium is defined as the total payment required in 1977 to keep a health                                                           
 insurance policy in force for a year.  As indicated, information needed to                                                          
 adjust the premium data for policies held only part of the year was not                                                             
 available.                                                                                                                          
      For information on weights to be used with this data set, see Section 3.2                                                      
 on this file.  Information specific to each variable or set of variables is                                                         
 provided in the codebook, which is File 2 on this tape.                                                                             
                                                                                                                                     
 2.2  Private insurance benefits of the population 65 and older (data set 2)                                                         
                                                                                                                                     
      This data set comprises all persons 65 years or older with private                                                             
 insurance coverage according to the HIES (1,966 persons).  For each, the                                                            
 following variables are provided.                                                                                                   
                                                                                                                                     
                                                                                                                                     
 2.2.1  Identification and insurance status                                                                                          
                                                                                                                                     
      The first variable in this file (PID, 20.10) identifies the person to                                                          
 whom the insurance information in the file is linked by the NMCES person                                                            
 identification number.  Ths next variables (COVERAGE, 21.10; EMPRELP, 21.11;                                                        
 and GROUPP and GROUPSZP (21.12 and 21.121)) describe whether the persons in                                                         
 this file hold household-reported public coverage (Medicare and/or Medicaid)                                                        
 in addition to private insurance, and the characteristics of the insurance                                                          
 held by each person on this file in terms of employment-related status, group                                                       
 insurance, and group size (see 2.1 above).                                                                                          
                                                                                                                                     
                                                                                                                                     
 2.2.2  Primary insured characteristics                                                                                              
                                                                                                                                     
      The next set of variables (22.10 to 22.12) identify primary-insured                                                            
 status and whether the policy (both group and nongroup) provides family,                                                            
 couple, or individual coverage.                                                                                                     
                                                                                                                                     
                                                                                                                                     
 2.2.3  Health services coverage                                                                                                     
                                                                                                                                     
      A Form Length indicator (23.00) is provided for each type of benefit to                                                        
 indicate use of the proper weight according to the type of benefit abstract                                                         
 used (see 1.2 and 3.2.2 on this file).  Benefits are then listed for the two                                                        
 major types of private insurance coverage, basic and major medical plans.                                                           
 Basic coverage provides reimbursement for many of the services required during                                                      
 illnesses of limited duration but typically limits coverage in terms of the                                                         
 maximum expense for or utilization of each insured service.  Major medical                                                          
 coverage is designed to provide additional protection; it usually obligates                                                         
                                                                                                                                     
1the insurer to pay a specified portion of a wider range of medical expenses                                                         
 but provides for deductibles and copayments for most expenses.  It is                                                               
 characterized by few limits for particular services and a high overall limit.                                                       
      In this data set, basic and major medical coverage frequencies and                                                             
 variable descriptions are provided separately but for the same set of                                                               
 services, as follows:                                                                                                               
                                                                                                                                     
                                                                                                                                     
                                              Basic           Major Medical                                                          
                                                                                                                                     
 Inpatient hospital and physician          24.10-24.16         25.10-25.16                                                           
 services and extended care facility                                                                                                 
 benefits                                                                                                                            
                                                                                                                                     
 Outpatient hospital facility             24.20-24.205        25.20-25.205                                                           
 services                                                                                                                            
                                                                                                                                     
 Ambulatory physician and                 24.30-24.37          25.30-25.37                                                           
 nonphysician services                                                                                                               
                                                                                                                                     
 Services for inpatient and               24.40-24.403        25.40-25.403                                                           
 outpatient mental health                                                                                                            
 conditions                                                                                                                          
                                                                                                                                     
 Outpatient diagnostic                        24.50               25.50                                                              
 procedures                                                                                                                          
                                                                                                                                     
 Home health care                             24.60               25.60                                                              
                                                                                                                                     
 Maternity services                           24.70               25.70                                                              
                                                                                                                                     
 Supplemental accident                        24.80               25.80                                                              
 benefit coverage                                                                                                                    
                                                                                                                                     
 Coverage for medical                         24.90               25.90                                                              
 supplies and equipment                                                                                                              
                                                                                                                                     
                                                                                                                                     
      Additionally, indicators of coverage are provided for outpatient                                                               
 prescribed medicines, dental, hearing and vision services, and for hospital                                                         
 indemnity coverage, which pays a specified amount per day, week, or month of                                                        
 hospitalization without reference to the actual expenses incurred for services                                                      
 (26.10-26.50).                                                                                                                      
      Information specific to each variable or set of variables is provided in                                                       
 the codebook, which is File 4 on this tape.                                                                                         
                                                                                                                                     
                                                                                                                                     
 2.2.4  Coverage of selected Medicare gaps                                                                                           
                                                                                                                                     
      In addition to private insurance coverage for particular services,                                                             
 benefits available to the elderly population under policies supplementing                                                           
 Medicare coverage are described.  In the main, this refers to deductibles and                                                       
 copayments and to hospital days not covered by Medicare.  Based upon an                                                             
 analysis of the private health insurance policies of the Medicare enrollees in                                                      
                                                                                                                                     
1the NMCES sample, persons were characterized as having full coverage for each                                                       
 of these Medicare cost sharing provisions if they were highly unlikely to                                                           
 incur any out-of-pocket expense as a result of using services.  In 1977, these                                                      
 cost-sharing provisions under Parts A and B, respectively, included a $124                                                          
 deductible for inpatient hospital stays, a daily $31 and $62 hospital                                                               
 copayment (for days 61-90 and the lifetime reserve, respectively), a 150-day                                                        
 limit on payment for hospital stays, and a $60-deductible and 20-percent                                                            
 coinsurance for physician services.                                                                                                 
      In addition to determining policies that specified as benefits full                                                            
 coverage of Medicare cost-sharing requirements, separate procedures were                                                            
 developed to identify persons with other types of private policies or,                                                              
 occasionally, several relatively limited plans, who would also incur no out-                                                        
 of-pocket expenses under normal circumstances.  Also, because private                                                               
 insurance benefits often differ by site of care (e.g. inpatient versus                                                              
 outpatient services), a distinction not made by Medicare, the potential                                                             
 liability for out-of-pocket expense resulting from the Part B deductible and                                                        
 coinsurance is likely to vary for these types of services and was consequently                                                      
 developed for both sites of care, as described in detail in the following.                                                          
      Part A ($124) deductible.  For policies where full coverage of the                                                             
+     _________________________                                                                                                      
 deductible was not explicitly specified, it was necessary to determine if the                                                       
 insurance would provide at least $124 in benefits for a hospital stay, thus                                                         
 offsetting the Medicare deductible.  As the total insurance benefits for a                                                          
 hospital stay often depend on the length of stay, a seven-day hospital stay as                                                      
 the modal length of stay for persons 65 and older in 1977 was assumed for the                                                       
 purposes of this report (NMCES, unpublished data).  The calculation of whether                                                      
 an insured person would have any liability for the Part A deductible within                                                         
 this period was based on the person's basic or hospital indemnity benefits, if                                                      
 any, and/or their major medical benefits.  If benefits are stated as a percent                                                      
 of the hospital's charges for the stay, the total benefit depends on the                                                            
 assumed charge per day; here, a semiprivate room rate of $90 per day (Health                                                        
 Insurance Association of America, 1977), or a total expense per day of $170                                                         
 (American Hospital Association, 1978) was assumed.  A policy consequently was                                                       
 considered to provide full coverage if the assumed or explicit daily benefit                                                        
 (times 7) exceeded the Medicare deductible (27.11).                                                                                 
      Part B deductible ($60) for physician care.  Full coverage of the                                                              
+     ___________________________________________                                                                                    
 deductible was assumed when the insurance would pay a lump sum of at least $60                                                      
 for the first physician service delivered at either inpatient or outpatient                                                         
 sites, or when the plan had no deductible and the benefit was 100 percent of                                                        
 the usual, customary and reasonable charge.  In view of the difference in                                                           
 insurance provisions regarding site of care, coverage for the Part B                                                                
 deductible is shown separately for inpatient and outpatient sites (27.12 and                                                        
 27.13).                                                                                                                             
      Part A copayments for days 61-90 ($31) and 91-150 ($62).  Where full                                                           
+     ________________________________________________________                                                                       
 coverage of the Part A copayments was not explicitly specified, it was                                                              
 determined if payable benefits would equal or exceed the cost of Medicare                                                           
 copayments.  (The net cost to a Medicare enrollee of a 90-day stay in 1977                                                          
 would have been $1,054, i.e., $124 plus $31/day for days 61-90; the net cost                                                        
 of a 150-day stay would have been $4,774, i.e. the 90-day copayment plus                                                            
 $62/day for days 91-150.)  If payable benefits for room and board and                                                               
 ancillary expenses less any deductible equaled or exceeded a Medicare                                                               
 enrollee's liability, coverage is considered full; if not, coverage is                                                              
 designated partial (27.21 and 27.22).                                                                                               
      Cost of hospital care for stays lasting 365 days.  Where full coverage of                                                      
+     _________________________________________________                                                                              
 a 365-day hospital stay was not explicitly specified, it was determined if                                                          
                                                                                                                                     
                                                                                                                                     
1payable benefits would have covered the cost of such a stay beyond Medicare                                                         
 benefits, which provide for a 150-day life-time maximum.  Liability for                                                             
 hospital charges during a stay lasting 365 days in 1977 can be estimated at                                                         
 $41,324 net of Medicare, i.e., the Part A deductible, all Part A copayments,                                                        
 and all charges for days 151-365 (assumed at $170 a day; American Hospital                                                          
 Association, 1978), but excluding expenses such as inpatient physician care.                                                        
      Thus, a policy was considered to provide full coverage if indemnity                                                            
 benefits for a 365-day stay equaled or exceeded $41,324 or if payable benefits                                                      
 from major medical coverage equaled or exceeded $41,324 and the policy paid                                                         
 100 percent of all costs exceeding $25,500 (the total cost of days 1-150).                                                          
 For plans with more than one type of benefit (e.g. basic and major medical                                                          
 coverage), benefits were combined to estimate coverage.  Plans with less than                                                       
 full coverage were examined to determine if they provided any benefits beyond                                                       
 the Medicare lifetime reserve (150 days).  A plan was considered to provide                                                         
 partial coverage of days 151-365 if the benefits for hospital care were not                                                         
 full but exceeded the costs of the Part A deductible and copayments through                                                         
 day 150 (27.23).                                                                                                                    
      Part B coinsurance for physician care.  Coverage of the 20-percent                                                             
+     ______________________________________                                                                                         
 Medicare coinsurance for Part B services was defined in relation to the first                                                       
 service provided after satisfaction of the policy deductible or the first                                                           
 visit or service if there was no deductible.  Again, separate variables were                                                        
 constructed for benefits at inpatient or outpatient sites.  Because a small                                                         
 fraction of policies (20 percent with coverage of inpatient physician care and                                                      
 less than 10 percent with benefits for physician office visits) provided for a                                                      
 flat amount per service or visit rather than a percentage, certain costs per                                                        
 visit were assumed to establish the percent paid ($20 for a physician office                                                        
 visit, $15 for an inpatient physician visit or other services).  Policies with                                                      
 coverage of physician services which were obligated to pay 20 percent or more                                                       
 of total charges were assumed to provide full coverage; additionally, for both                                                      
 sites of care, the variables indicate if the policy stipulated payment only up                                                      
 to the Medicare allowable (UCR) limit (27.31 and 27.32).                                                                            
                                                                                                                                     
                                                                                                                                     
 3.  Sampling Information, Weights, and Variance Estimation                                                                          
                                                                                                                                     
 3.1  Sampling                                                                                                                       
                                                                                                                                     
      The NMCES household sample was a national area probability sample of                                                           
 about 14,000 households designed to represent the civilian,                                                                         
 noninstitutionalized population of the 50 states and the District of                                                                
 Columbia.  To maximize geographic dispersion of the sample and the statistical                                                      
 precision of resulting estimates, these households were selected through a                                                          
 four-stage process rather than by simple random sampling.  Two national area                                                        
 samples were drawn independently from the national general purpose area                                                             
 samples of Research Triangle Institute (RTI) and National Opinion Research                                                          
 Center (NORC).  Within each of these samples, the four stages of the design                                                         
 were primary sampling units (PSUs) that are counties, parts of counties, or                                                         
 groups of contiguous counties; secondary sampling units (SSUs) that are Bureau                                                      
 of the Census Block Groups or Enumeration Districts; smaller area segments                                                          
 generally consisting of at least 60 housing units; and residential housing                                                          
 units which were the final sampling units.                                                                                          
      The household residing within each sample housing unit was designated as                                                       
 a reporting unit (RU) if all residents were related by blood, marriage or                                                           
 adoption.  Separate reporting units were established within housing units that                                                      
                                                                                                                                     
                                                                                                                                     
1contained unrelated persons.  The NMCES response rate, measured as the percent                                                      
 of eligible reporting units providing data for all rounds of household data                                                         
 collection, was 82.2 percent for the RTI sample and 81.7 percent for the NORC                                                       
 sample.                                                                                                                             
      All household survey participants, with the exception of unemployed or                                                         
 exclusively self-employed individuals and their dependents who were reported                                                        
 as uninsured in the household survey, were eligible for the Health Insurance                                                        
 Employer Survey.  For a more detailed discussion of the HIES sample design,                                                         
 see NHCES Instruments and Procedures, 3 (Cohen and Farley, 1984).                                                                   
+    ___________________________________                                                                                             
                                                                                                                                     
                                                                                                                                     
 3.2  Sample weights                                                                                                                 
                                                                                                                                     
      The sample weights developed for all HIES data reflect the differential                                                        
 probability of selection into the sample resulting from the complex four-stage                                                      
 sample design.  Furthermore, to obtain national estimates from the responses                                                        
 obtained in the HIES and its substudy, the UVS, the development of series of                                                        
 weights adjusting for various types of HIES/UVS nonresponse was required, in                                                        
 addition to adjustments for reporting unit nonresponse and poststratification                                                       
 to represent the civilian noninstitutionalized U.S. population as of 1977.                                                          
 (The latter adjustments are incorporated in the sample weights developed for                                                        
 key participants in the household survey; Cohen and Kalsbeek, 1981.)  Weights                                                       
 were 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, insurance group, or employer;                                                                               
                                                                                                                                     
 (2)  Health insurer nonresponse.  The insurance carrier, insurance group, or                                                        
      employer identified on a signed permission form did not respond at all or                                                      
      did not provide one 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)  a copy of the policy describing applicable benefits.                                                                      
                                                                                                                                     
                                                                                                                                     
 3.2.1  Data set 1                                                                                                                   
                                                                                                                                     
      Two types of weights are provided for each record in the file.  WTINSP is                                                      
 to be used when the unit of analysis is a person, WTINSFAM is to be used when                                                       
 the unit of analysis is a household.                                                                                                
      The household survey weights were initially adjusted for unsigned HIES                                                         
 permission forms (item (1) above) and health insurance provider nonresponse                                                         
 (item (2)(a) above).  The resulting person weight, WTINSP (14.10), must be                                                          
 used for items on insurance status, type of coverage, and amounts and sources                                                       
 of premium payment at the person level, including estimates of the primary                                                          
 insured population only.  The positive values of WTINSP, summed across the                                                          
 person records on this file, yield the total civilian, noninstitutionalized                                                         
 population of the U.S. in 1977.  A corresponding family weight, WTINSFAM                                                            
 (14.11), is supplied for use when the family as defined by RUIDHIES is the                                                          
 unit of analysis.  The positive values of WTINSFAM, summed across household                                                         
                                                                                                                                     
                                                                                                                                     
1heads (RELHEAD=10), yield the average number of families that existed at any                                                        
 given time during 1977.  (See Section 2.1.1 with regard to the person file                                                          
 from the household survey component.)                                                                                               
      Each record also contains an indicator of whether a person was a key                                                           
 participant in the household survey (WTKEY, 14.20).  Key participants were                                                          
 defined as persons in the sample households at the start of the survey along                                                        
 with babies born in to sample households, and initially ineligible household                                                        
 members (i.e. who had entered the military or an institution but who later                                                          
 became eligible by rejoining the household).  Nonkey participants are persons                                                       
 who joined the households of key participants later in the survey year, as in                                                       
 the case of marriage.  Because of the possibility that nonkey participants                                                          
 held insurance covering key participants, they were included in the HIES data                                                       
 collection.                                                                                                                         
      For detailed information about the construction of HIES weights, see                                                           
 NHCES Instruments and Procedures, 3 (Cohen and Farley, 1984).                                                                       
+___________________________________                                                                                                 
                                                                                                                                     
                                                                                                                                     
 3.2.2  Data set 2                                                                                                                   
                                                                                                                                     
      For the variables on the file for persons 65 or older, WTINSP (28.10) is                                                       
 the appropriate weight.  While this file is not by itself suitable to family-                                                       
 level analyses, data for family members, including those less than 65 years,                                                        
 and the suitable family-level weight (WTINSFAM 14.11) can be obtained by                                                            
 linking to the premium file.                                                                                                        
      For individual benefit variables derived from the health insurance                                                             
 policies obtained in the HIES, two additional weights were developed, which                                                         
 further adjust WTINSP for estimates of the benefits under a person's                                                                
 coverage.  Variables that were abstracted on the short form (see Section 1.2                                                        
 of this documentation) are indicated by an "S" as the last character in the                                                         
 variable name.  Variables abstracted onto the long form are indicated by an                                                         
 "L" in the variable name and are available only for the long-form subsample.                                                        
 The variable FORMLENG (23.00) identifies each person as having no abstract                                                          
 data (FORMLENG=9), short-form data only (FORMLENG=2), or having both short-                                                         
 and long-form data (FORMLENG=1).  (Because items on the short form were a                                                           
 subset of items on the long form, both short-form and long-form variables are                                                       
 available for the long-form subsample.)  The short-form weight, WTSFPID                                                             
 (28.21) is appropriate for estimates for short-form variables, and the long-                                                        
 form weight, WTLFPID (28.20) is suitable for estimates using any abstract                                                           
 variable.                                                                                                                           
      Values of WTINSP, WTSFPID, or WTLFPID (28.20) summed separately across                                                         
 the person-level records in this file yield the number of privately insured                                                         
 persons 65 years or older in 1977.  The estimates from these weights differ                                                         
 slightly because of rounding.                                                                                                       
      Again, each record indicates key participant status in the household                                                           
 survey for each person on the data file (WGKEY, 28.30).                                                                             
                                                                                                                                     
                                                                                                                                     
 3.3  Variance estimation                                                                                                            
                                                                                                                                     
      Variance estimates of sample statistics for the construction of                                                                
 confidence intervals and hypothesis testing require that the complex nature of                                                      
 the HIES sample survey design be taken into account.  Here, the assumptions of                                                      
 independence and equal probability of selection inherent in simple random                                                           
 sampling are not satisfied.  Several methods are available that approximate                                                         
 the sampling variance for statistics derived from HIES data.  Three generally                                                       
                                                                                                                                     
                                                                                                                                     
1accepted and frequently used techniques for obtaining variance estimates for                                                        
 various types of statistics have been incorporated in several widely used                                                           
 statistical packages.  They are:                                                                                                    
                                                                                                                                     
      Balanced Repeated Replication (Bean, 1975; McCarthy, 1966),                                                                    
                                                                                                                                     
      the jack-knife method (Kish and Frankel, 1974; Brillinger, 1964), and                                                          
                                                                                                                                     
      the Taylor Series linearization method (Woodruff, 1971; Tepping, 1968).                                                        
                                                                                                                                     
 See Note 1 for information on the programs available for specific types of                                                          
 estimates.                                                                                                                          
      Each variance estimation method is compatible with a multistage,                                                               
 stratified sampling design with at least two primary sampling units drawn from                                                      
 each stratum.  Due to the sampling design, the requirement of paired sampling                                                       
 units was not initially satisfied, as primary sampling units were selected                                                          
 from two independently drawn national samples; the primary sampling units                                                           
 were, therefore, paired as members of pseudo strata to allow for variance                                                           
 computations.  To obtain variance estimates of sample statistics by means of                                                        
 standard statistical packages, two additional variables must be used for all                                                        
 insurance status, premium, and benefit information in both data sets:                                                               
                                                                                                                                     
      WGPSEUDO (14.30; Data Set 1), (28.40; Data Set 2), which indicates the                                                         
      respective pseudo stratum, and                                                                                                 
                                                                                                                                     
      WGMEMBER (14.31), (28.41) which denotes the pseudo primary sampling unit                                                       
      to which an individual belongs within a given pseudo stratum.                                                                  
                                                                                                                                     
      For a more complete discussion of the NMCES sample design and estimation                                                       
 and variance procedures, see NHCES Instruments and Procedures, 2 (Cohen and                                                         
+                             ___________________________________                                                                    
 Kalsbeek, 1981).                                                                                                                    
      In some instances, relative variance curves are a cost-effective                                                               
 alternative to direct methods of variance estimation.  This technique was used                                                      
 for a range of parameter estimates derived from NMCES data on different types                                                       
 of health care variables (see Cohen, 1979, and the NHCES Data Preview series).                                                      
                                                                                                                                     
                                                                                                                                     
 4.   Missing Data:  Imputation and Adjustment for Partial Response                                                                  
                                                                                                                                     
      The HIES was a separate survey component of NMCES in which eligible                                                            
 members of the household survey were asked for permission to contact their                                                          
 insurers or employers (the respondents in HIES) to provide insurance and                                                            
 premium information.  Consequently, missing data on these files are due to one                                                      
 or several of the following:                                                                                                        
                                                                                                                                     
      (1)  refusal of household members to sign the permission form;                                                                 
                                                                                                                                     
      (2)  health insurer nonresponses to                                                                                            
                                                                                                                                     
           the coverage questionnaire requesting verification of private                                                             
           insurance coverage as reported by the household and information                                                           
           about premiums and the type of policy; or                                                                                 
                                                                                                                                     
           a copy of the insurance policy describing applicable benefits;                                                            
                                                                                                                                     
                                                                                                                                     
1     (3)  partial response on multiple coverage, where some but not all HIES                                                        
      respondents identified for a particular participant in the household                                                           
      survey responded in the HIES.  In cases of multiple coverage, failure by                                                       
      one or more insurance carriers or groups to provide premium information                                                        
      yielded only partial response on the total premiums associated with an                                                         
      individual's coverage.                                                                                                         
                                                                                                                                     
      In addition to record editing and consistiency checks, three techniques                                                        
 were used to adjust for these types of nonresponse, sometimes in conjunction                                                        
 with one another:                                                                                                                   
                                                                                                                                     
      Imputation of missing items and missing records                                                                                
                                                                                                                                     
      Weight adjustments (see also Section 3.2)                                                                                      
                                                                                                                                     
      Adjustment for partial response (see Attachment 1, NHCES Instruments and                                                       
      Procedures 3, p. 15)                                                                                                           
                                                                                                                                     
      The following discusses the type of missing data and describes the                                                             
 procedures used.                                                                                                                    
                                                                                                                                     
                                                                                                                                     
 4.1  Private health insurance status                                                                                                
                                                                                                                                     
      Where the private insurance status of any person eligible for the HIES                                                         
 could not be established from responses to the HIES, that case was treated as                                                       
 a nonresponse and a weight adjustment was made.  A similar weight adjustment                                                        
 was used to correct for missing data with respect to a family's health                                                              
 insurance status, but the weight adjusts only for complete nonresponse (i.e.,                                                       
 individual nonresponse for all family members eligible for the HIES).                                                               
 Families with partial responses regarding individual members were treated like                                                      
 families with complete responses in the construction of family weights, i.e.,                                                       
 their insurance status was defined from the insurance status of the individual                                                      
 member(s) for whom responses were received.                                                                                         
                                                                                                                                     
                                                                                                                                     
 4.2  Plan type                                                                                                                      
                                                                                                                                     
      Group plans were defined as employment-related if a union or employer                                                          
 administered the plan or a union or employer contributed to the premium.                                                            
 Thus, a group plan with missing data regarding the administration of the plan                                                       
 was classified as employment related provided a union or employer contributed                                                       
 to the premium.  If the employer or insurer contacted in the HIES did not                                                           
 identify a plan as being for an individual, couple, or family, the type of                                                          
 coverage was defined based on the number of family members reporting the                                                            
 applicable insurance company as a source of coverage in the household survey.                                                       
                                                                                                                                     
                                                                                                                                     
 4.3  Premiums                                                                                                                       
                                                                                                                                     
      For approximately 3.6 percent of primary insured persons in the HIES, the                                                      
 response received from at least one employer or insurance company had missing                                                       
 or incomplete data on annual premiums and sources of premium payments.  These                                                       
 data were imputed by the so-called "hot-deck" procedure developed at the U.S.                                                       
                                                                                                                                     
                                                                                                                                     
1Bureau of the Census for their Current Population Reports series.  This                                                             
 procedure randomly assigns data from individuals with complete data to                                                              
 individuals with missing data but similar characteristics; here the primary                                                         
 insured's age, sex, marital status, employment, and industry of employment as                                                       
 well as the type of insurance policy (group or nongroup contract), the number                                                       
 of people covered under the policy (an individual, couple, or family), and                                                          
 types of benefits.  In addition, not all employers or insurance companies who                                                       
 provided a primary insured with coverage during the survey year responded.  A                                                       
 partial response inflation factor was used to adjust the available premium                                                          
 data for such missing plans.                                                                                                        
                                                                                                                                     
                                                                                                                                     
 4.4  Private health insurance benefits                                                                                              
                                                                                                                                     
      For 111 persons in the sample of privately insured persons 65 years and                                                        
 older, the employer or insurance company contacted in the HIES confirmed                                                            
 coverage and provided the premium amount but did not enclose the policy.                                                            
 Since the weights, WTSFPID (28.21) and WTLFPID (28.20), adjust for this type                                                        
 of total benefit nonresponse, persons are shown as having missing or unknown                                                        
 benefit data only when the information in the policy or policies received for                                                       
 abstracting was incomplete with regard to that variable.  With regard to                                                            
 coverage for particular health services, missing values were sometimes                                                              
 eliminated on individual abstracts by edits and consistency checks.  The                                                            
 remaining missing data, i.e., where a policy or policies did not clearly                                                            
 specify coverage or lack of coverage for a service, were not imputed, and are                                                       
 denoted as unknown.  Likewise, where the abstracted policy or insurance                                                             
 brochure did not provide a clear or complete description of benefits and it                                                         
 was consequently impossible to determine the extent of supplementary coverage                                                       
 of selected Medicare cost-sharing provisions, no adjustments were made and                                                          
 variables are left as unknown.                                                                                                      
                                                                                                                                     
                                                                                                                                     
 5.   Codebook Information                                                                                                           
                                                                                                                                     
      Two codebooks are included as FILE 2 and FILE 4 on this tape.  File 2                                                          
 contains variable information and frequency distributions for a total of                                                            
 40,320 person records representing private health insurance status, premiums                                                        
 and sources of premium payment, as described in Section 2.1 of this                                                                 
 documentation.  File 4 contains variable information and frequency                                                                  
 distributions for a total of 1,966 person records representing health service                                                       
 coverage and related characteristics for the privately insured Medicare                                                             
 population age 65 and above, as described in Section 2.2 of this                                                                    
 documentation.  Each codebook describes a EBCDIC data set providing the                                                             
 following programming identifiers for each variable:                                                                                
                                                                                                                                     
                                                                                                                                     
          Identifier     Description                                                                                                 
                                                                                                                                     
          NAME           Variable name (8 characters or less)                                                                        
          DESCRIPTION    Variable description                                                                                        
          LENGTH         Length of variable in bytes                                                                                 
          START          Beginning column position of variable within the                                                            
                         record                                                                                                      
          END            Ending column position of variable within the record                                                        
                                                                                                                                     
                                                                                                                                     
1     The codebooks describe each variable in some detail, including                                                                 
 instructions for use, sources of data by NMCES component, and cross-references                                                      
 to other variables.  Unweighted counts of records and weighted frequencies,                                                         
 where appropriate, are presented for each category of the discrete                                                                  
 variables.  For continuous variables, unweighted counts and weighted                                                                
 frequencies for various ranges are presented to indicate the distribution of                                                        
 cases.                                                                                                                              
      Weighted frequencies for both types of variables were tabulated using the                                                      
 SUMMARY procedure from the SAS (Statistical Analysis System) computer software                                                      
 system with the length indicated in the specification (LENGTH).  Other means                                                        
 of tabulating weighted frequencies may yield a slightly different estimate.                                                         
 Also, as a result of continued cleaning and editing of the data, the weighted                                                       
 estimates in this codebook may differ from previously published estimates of                                                        
 NMCES data.  The use of weights to produce weighted estimates is discussed in                                                       
 Section 3.2.  For a discussion of variance calculation, see Section 3.3.                                                            
 Several computer packages which produce variance estimates appropriate for                                                          
 complex survey data are listed in Note 1.                                                                                           
      The sequence of variables in each codebook reflects major components of                                                        
 information provided.  A multi-digit numerical identifier (TAG) was developed                                                       
 for this purpose, where the first digit indicates the universe to which the                                                         
 data file was assigned.  The second digit before the decimal point indicates                                                        
 the type of data subsumed by the variables in that sequence, and the digits                                                         
 following the decimal point designate either a subsection within each category                                                      
 and are simple sequence indicators.  Note that the TAG numbers are for                                                              
 information only.  They are not part of each record.                                                                                
                                                                                                                                     
                                                                                                                                     
 6.  Programming Information                                                                                                         
                                                                                                                                     
      This documentation is the first file on the tape.  Each codebook file is                                                       
 followed by its accompanying data file.  The specifications for each file on                                                        
 the tape are as follows:                                                                                                            
                                                                                                                                     
     VOLUME:  HIES02                                                                                                                 
          FILE 1:                                                                                                                    
               Description -  HIES Public Use Tape Documentation                                                                     
               Dataset Name - HIES.PERSON.EBCDIC.DOCUMENT                                                                            
                                                                                                                                     
          FILE 2:                                                                                                                    
               Description -  HIES Public Use Data -- Private Insurance Status                                                       
                              and Premiums:  Codebook                                                                                
               Dataset Name - HIES.PERSON.EBCDIC.PREMCODE                                                                            
                                                                                                                                     
          FILE 3:                                                                                                                    
               Description -  HIES Public Use Data -- Private Insurance Status                                                       
                              and Premiums:  Data File                                                                               
               Dataset Name - HIES.PERSON.EBCDIC.PREMIUMS                                                                            
               Number of Observations -    40,320                                                                                    
               Number of Variables -           44                                                                                    
                                                                                                                                     
          FILE 4:                                                                                                                    
               Description -  HIES Public Use Data -- Private Insurance                                                              
                              Benefits for Persons 65 and Older:  Codebook                                                           
               Dataset Name - HIES.PERSON.EBCDIC.MEDICODE                                                                            
                                                                                                                                     
                                                                                                                                     
1         FILE 5:                                                                                                                    
               Description -  HIES Public Use Data -- Private Insurance                                                              
                              Benefits for Persons 65 and Older:  Data File                                                          
               Dataset Name - HIES.PERSON.EBCDIC.MEDIGAP                                                                             
               Number of Observations -     1,966                                                                                    
               Number of Variables -           91                                                                                    
                                                                                                                                     
     The codebook files contain information and frequency distributions for                                                          
 each variable in each of the data files.  The documentation and the codebooks                                                       
 are stored as Operating System (OS) EBCDIC files containing American Standard                                                       
 Association (ASA) carriage control characters in the first position in each                                                         
 record which will direct the line printer during output to skip lines, begin a                                                      
 new page, etc.                                                                                                                      
     Using a text editor computer program such as WYLBUR, the documentation or                                                       
 a codebook that has been copied to disk can be retrieved on-line to view                                                            
 and/or modify.  The ASA control characters will direct the line printer if a                                                        
 hard copy listing is desired.                                                                                                       
     The data sets contain identifiers of private insurance status, premiums                                                         
 and sources of payment and, respectively, characteristics of coverage of the                                                        
 population 65 and older with private insurance.  Both data files were created                                                       
 in SAS (Statistical Analysis System) and converted to EBCDIC format using the                                                       
 SAS computer software system, and are stored as EBCDIC data sets (see SAS                                                           
+                                                                      ___                                                           
 User's Guide, SAS Institute, 1982).                                                                                                 
+____________                                                                                                                        
                                                                                                                                     
                                                                                                                                     
1Note 1                                                                                                                              
      To obtain variance estimates for the data in this codebook, the variables                                                      
 WGPSEUDO, 14.30 (Data Set 2), 28.40 (Data set 2), and WGMEMBER, 14.31 (Data                                                         
 Set 1), 28.41 (Data Set 2) must be used.  The following indicates several                                                           
 well-known variance calculation programs using different procedures for this                                                        
 purpose (see Section 3.3 for a discussion of variance estimation in complex                                                         
 survey designs):                                                                                                                    
                                                                                                                                     
 Type of estimate           Program for calculating variance                                                                         
                                                                                                                                     
 Regression coefficients    SURREGR:  Standard errors of regression                                                                  
                            coefficients from Sample data accessible through                                                         
                            SAS (Holt, 1977).  Generates standard errors of                                                          
                            regression coefficients using the Taylor series                                                          
                            linearization method for complex survey data.                                                            
                                                                                                                                     
                            REPERR:  Repeated Replication Sampling Error                                                             
                            analysis procedure, accessible through OSIRIS IV                                                         
                            (Van Eck, 1979).  Generates standard errors of                                                           
                            regression coefficients through three repeated                                                           
                            replication procedures:  (a) balanced half-sample,                                                       
                            (b) jack-knife, or (c) user specified.                                                                   
                                                                                                                                     
 Means, proportions,        SESUDAAN:  Standard errors program for computing of                                                      
 population totals          Standardized Rates from Sample Survey Data,                                                              
                            accessible through SAS (Shah, 1979).  Generates                                                          
                            standard errors for these types of estimates using                                                       
                            the Taylor series linearization method for complex                                                       
                            survey data.                                                                                             
                                                                                                                                     
                            Health Examination Survey Variance and Cross                                                             
                            Tabulation Program:  Developed by the National                                                           
                            Center for Health Statistics (Jones, 1977).                                                              
                            Generates standard errors for these types of                                                             
                            estimates based on a balanced half-sample method.                                                        
                                                                                                                                     
 Ratios                     RATIOEST:  Standard Errors Program for Computing                                                         
                            Ratio Estimates from Sample Survey Data accessible                                                       
                            through SAS (Shah, 1981).  Generates standard                                                            
                            errors of ratios using the Taylor series                                                                 
                            linearization method for complex survey data.                                                            
                                                                                                                                     
                                                                                                                                     
1REFERENCES                                                                                                                          
                                                                                                                                     
                                                                                                                                     
 Bean, J.A.                                                                                                                          
          1975      Distribution and Properties of Variance Estimators for                                                           
                    Complex Multistage Probability Samples:  An Empirical                                                            
                    Distribution.  National Center for Health Statistics, Vital                                                      
                    and Health Statistics, Series 2, No. 65, PHS Pub. No. 75-                                                        
                    1339, Washington, D.C., U.S. Government Printing Office.                                                         
                                                                                                                                     
 Bonham, G.S. and Corder, L.S.                                                                                                       
          1981      National Medical Care Expenditure Survey:  Household                                                             
                    Interview Instruments and Procedures Series, 1, DHHS                                                             
                    Publication No. (PHS) 81-3280.                                                                                   
                                                                                                                                     
 Brillinger, D.R.                                                                                                                    
          1964      The asymptotic behavior of Tukey's general method of set-                                                        
                    ting approximate confidence limits (the jackknife) when                                                          
                    applied to maximum likelihood estimates.  Rev Int Statist                                                        
                    Inst 3:202-206.                                                                                                  
                                                                                                                                     
 Cohen, S.B.                                                                                                                         
          1979      An assessment of curve smoothing strategies which yield                                                          
                    variance estimates for complex survey data.  Proceedings,                                                        
                    Survey Research Section.  Washington:  American Statistical                                                      
                    Association, pp. 101-104.                                                                                        
                                                                                                                                     
 Cohen, S.B.                                                                                                                         
          1982      An analysis of alternative imputation strategies for indi-                                                       
                    viduals with partial data in the National Medical Care                                                           
                    Expenditure Survey.  Rev Public Data Use 10:153-165.                                                             
                                                                                                                                     
 Cohen, S.B. and Farley, P.J.                                                                                                        
          1984      National Medical Care Expenditure Survey:  Estimation and                                                        
                    Sampling Procedures in the NMCES Insurance Surveys,                                                              
                    Instruments and Procedures Series, 3, DHHS Publication No.                                                       
                    (PHS) 84-3369.  This publication is Attachment 1 to the                                                          
                    paper copy of this document.                                                                                     
                                                                                                                                     
 Cohen, S.B. and W.D. Kalsbeek                                                                                                       
          1981      National Medical Care Expenditure Survey:  Estimation and                                                        
                    Sampling Variance in the Household Survey.  National Center                                                      
                    for Health Services Research, Instruments and Procedures                                                         
                    Series 2, DHHS Publication No. (PHS) 81-3281.                                                                    
                                                                                                                                     
 Holt, M.M.                                                                                                                          
          1977      SURREGR:  Standard Errors of Regression Coefficients From                                                        
                    Sample Survey Data, Research Triangle Institute, Research                                                        
                    Triangle Park, North Carolina.                                                                                   
                                                                                                                                     
 Jones, G.                                                                                                                           
          1977      Health Examination Survey (HES) Variance and Cross-tabula-                                                       
                    tion Program - Version 2, Internal NCHS Report,                                                                  
                    Hyattsville, Maryland.                                                                                           
                                                                                                                                     
                                                                                                                                     
1Kasper, J.A., Walden, D.C., and Wilson, R.                                                                                          
          1983      National Medical Care Expenditure Survey Household Data:                                                         
                    Person Records SAS File.  National Technical Information                                                         
                    Service, Springfield, Virginia, PB 83-198077.                                                                    
                                                                                                                                     
 Kish, L. and Frankel, M.R.                                                                                                          
          1974      Inferences from complex surveys.  J Royal Stat Soc 36:1-37.                                                      
                                                                                                                                     
 McCarthy, P.J.                                                                                                                      
          1966      Replication:  An Approach to the Analysis of Data From                                                           
                    Complex Surveys.  National Center for Health Statistics,                                                         
                    Vital and Health Statistics, Series 2, No. 14, PHS Pub. No.                                                      
                    9-1269, Washington, Government Printing Office.                                                                  
                                                                                                                                     
 National Health Care Expenditures Study                                                                                             
          1981-85   Data Preview Series, National Center for Health Services                                                         
                    Research, Rockville, Maryland.                                                                                   
                                                                                                                                     
 Shah, B.V.                                                                                                                          
          1979      SESUDAAN:  Standard Errors Program for Computing of Stan-                                                        
                    dardized Rates from Sample Survey Data, Research Triangle                                                        
                    Institute, Research Triangle Park, North Carolina.                                                               
                                                                                                                                     
 SAS Institute                                                                                                                       
          1982      SAS Users Guide:  Basics, 1982 Edition.  SAS Institute                                                           
                    Inc., Cary, North Carolina                                                                                       
                                                                                                                                     
 Shah, B.V.                                                                                                                          
          1981      RATIOEST:  Standard Errors Program for Computing Ratio                                                           
                    Estimates for Sample Survey Data, Research Triangle Insti-                                                       
                    tute, Research Triangle Park, North Carolina.                                                                    
                                                                                                                                     
 Tepping, B.J.                                                                                                                       
          1968      The estimation of variance in complex surveys.                                                                   
                    Proceedings, Social Statistics Section, American Statis-                                                         
                    tical Association.  Washington, pp. 11-18.                                                                       
                                                                                                                                     
 Van Eck, N.A.                                                                                                                       
          1979      OSIRIS IV User's Manual.  5th ed. Survey Research Center,                                                        
                    Institute for Social Research, The University of Michigan,                                                       
                    Ann Arbor, Michigan.                                                                                             
                                                                                                                                     
 Walden, D.C. and Wilson, R.                                                                                                         
          1985      National Medical Care Expenditure Survey Household Data:                                                         
                    Event Records SAS and EBCDIC Files.  National Technical                                                          
                    Information Service, PB 85-246619, PB 85-246627, PB 85-                                                          
                    246635, and PB 85-246643.                                                                                        
                                                                                                                                     
 Woodruff, L.S.                                                                                                                      
          1971      A simple method of approximating the variance of a compli-                                                       
                    cated estimate.  J Am Stat Assoc 88:411-414.                                                                     
