The Medicare Current Beneficiary Survey (MCBS) is conducted by the Office of Strategic Planning of the Centers for Medicare & Medicaid Services (CMS). It is a continuous, multipurpose survey of a nationally representative sample of the Medicare population, providing information on aged and disabled Medicare beneficiaries living in communities and long-term care facilities. The sample is selected from Medicare enrollment files, and sample persons are interviewed three times per year over a 4-year period. Sample data are collected through computer-assisted personal interviews of the beneficiary or a proxy respondent if the sample person is not available for the interview. Survey questionnaires are tailored to the residence of the sample person—community or long-term care facility. If the sample person moves from one setting to the other during the period covered by an interview, a separate interview is administered for the time spent in each setting. As a consequence, beneficiaries are followed continuously over time regardless of where they live.
The primary goal of the survey is to collect comprehensive information on all medical services used by Medicare beneficiaries, sources of payment, health insurance coverage, and socioeconomic and demographic characteristics of the population. These data are used by the Office of Strategic Planning and the Office of the Actuary to support a host of activities related to the administration of the Medicare program (health insurance for the aged and disabled) and the Medicaid program (Federal grants to States for medical assistance programs). CMS also uses the data to evaluate the Medicare and Medicaid programs and to assess the financial impact of programmatic change on beneficiaries and the Medicare trust fund.
Two public use files are created for each calendar year of data collected in the MCBS: Access to Care and Cost and Use. Access to Care files are made available to the public about 1 year after the close of field work for the fall interview period (September-December). These public use files contain the “always enrolled” segment of the Medicare population; i.e., individuals who were enrolled in Medicare Part A or Part B as of January 1 of the year of data collection and alive at the time of the fall interview. They exclude beneficiaries who enrolled in Medicare during the year and beneficiaries who died before they could be interviewed in the fall round.
Access to Care files contain a mix of Medicare program data from administrative files and survey data from the fall round of interviews. The program data include summaries of Medicare-covered health care use and expenditures for the year. The survey-reported data include information from the fall round of interviews on health insurance coverage, health status and functioning, and access to and satisfaction with care. These public use files are most appropriate for users interested in analyzing the continuously enrolled population and its use of Medicare-covered services. The files also contain longitudinal weights that can be used to track cohorts of Medicare beneficiaries over time (2, 3, or 4 years).
Cost and Use files are released about 2 years after the close of field work for the calendar year. They contain a nationally representative sample of “ever enrolled” Medicare beneficiaries, i.e., all persons who were in the program at any time during the calendar year. The data include survey-reported information on health care use, Medicare claims data for unreported services, and administrative data such as Medicaid buy-in status and Medicare managed care plan enrollment.
MCBS data for the 2006 National Healthcare Disparities Report (NHDR) are primarily drawn from the Cost and Use file for the data year 2002. The Cost and Use file was chosen over the Access to Care file as the primary source of data on Medicare beneficiaries for three reasons: (1) it contains a nationally representative sample of all Medicare beneficiaries rather than just the always enrolled population; (2) it has information on health care expenditures not included in the Access to Care files (e.g., non-covered services such as dental and vision care and prescribed medicines); and (3) information on beneficiary income is complete and continuous for each sample person in the Cost and Use file, as opposed to incomplete and categorical in the Access to Care file.
However, the Cost and Use file could not be used independently of the Access to Care file to produce all of the statistics requested for the NHDR. It lacks information collected in the fall round on satisfaction with care, access to care, and usual source of care. This deficiency was corrected by appending the needed data from the Access to Care file for 2003 to sample persons in the Cost and Use file. Then the augmented Cost and Use file was used to generate statistics for both elderly beneficiaries 65 and older who lived in community settings (H_AGE >=65 and TYPE = ”C”) and disabled beneficiaries under 65 (H_AGE <65 and TYPE=”C”).
The samples size and weighted sample size used to generate statistics for the 2006 NHDR are given in the following table.
| Residents age 65 and older | Disabled residents under age 65 | ||||
|---|---|---|---|---|---|
| Sample size | Weighted size | Sample size | Weighted size | ||
| Total | 9,597 | 33,499,174 | 1,874 | 5,504,832 | |
| Gender | Male | 4,116 | 14,382,370 | 1,059 | 3,061,651 |
| Female | 5,481 | 19,116,804 | 815 | 2,443,182 | |
| Age | 65 to 74 | 4,301 | 17,668,459 | ||
| 75 to 84 | 3,802 | 12,164,988 | |||
| 85 and older | 1,494 | 3,665,728 | |||
MCBS tables on access, use and quality were created with standardized column headings for race, ethnicity, and poverty status. The MCBS race and ethnicity variables are coded from beneficiary responses to the survey questions, and the categories are recorded as interpreted by the respondent. Race categories include American Indian, Asian or Pacific Islander, Black or African American, White, more than one race, or other race. Hispanic ethnicity categories include persons of Mexican, Puerto Rican, Cuban, Central or South American, or other Spanish culture or origin, regardless of race.
Poverty status categories were created for sample persons in the survey based on their reported or imputed income from all sources. Each person was assigned to a poverty status category by using US Bureau of the Census data on poverty thresholds for 2002, by size of family and number of related children under 18 years. The constructed variables are based on the assumption that unmarried beneficiaries were one person families and married beneficiaries were two person families.
Age – Three age categories were developed for the NHDR tables on elderly community residents: 65 to 74, 75 to 84, and 85 and older. Each sample person was assigned to a category based on his or her age as of July 1, 2002.
Gender – Gender is recorded as reported on the Medicare/Social Security administrative record for the sample person.
Insurance coverage – Insurance coverage in the MCBS is based on a combination of Medicare and Medicaid administrative data and survey-reported information on public and private insurance. The information is used to construct month-by-month timelines of coverage by type of insurance for each sample person in a Cost and Use file. In addition, summary health insurance variables are created by using the month-by-month variables to identify persons who had a particular type of insurance at any time in 2002.
The types of insurance identified in the MCBS include Medicare fee-for-service, Medicare managed care, Medicaid, private employer-sponsored, private individually purchased (Medigap), private health maintenance organization (HMO), and other insurance. Summary versions of these variables were used to create four mutually exclusive groups of Medicare beneficiaries by prioritizing insurance coverage as follows:
Urban/rural – Sample persons are classified as living in a metropolitan or non-metropolitan area based on their addresses in the Medicare administrative files.
Perceived health status – Each sample person in the MCBS is asked to compare his or her general health to other persons of the same age. Persons receiving a community interview answer the question themselves. If a sample person is not able to do so, the proxy for that person answers the question. The possible responses to the question include excellent, very good, good, fair, and poor. Persons reporting fair or poor health are shown in the NHDR tables.
Functional limitations – Sample persons are identified as having functional limitations through their responses to six sets of questions about activities of daily living (ADLs) and instrumental activities of daily living (IADLs). Sample persons are asked whether they have a health or physical problem that makes it difficult to perform any of the specified activities. If the response is yes, they are recorded as having a limitation in that particular activity. If the response is “doesn't do,” they are asked whether it is because of a health or physical problem. Then, if the response to the followup question is “yes,” they are recorded as having a limitation in that particular activity.
The ADLs are activities related to personal care; they include bathing or showering, dressing, eating, getting in or out of bed or a chair, walking, and toileting. The IADLs are activities related to independent living; they include using the telephone, doing light housework, doing heavy housework, preparing meals, shopping for personal items, and managing money. Any limitation reported by a sample person may have been temporary or chronic at the time of the interview.
The 2006 NHDR includes four measures for full-year community residents age 65 or older and one measure for disabled beneficiaries under age 65.
For measures 1-3, numerator includes persons with an expense for at least one of the relevant services, and denominator excludes persons with missing values. Measure 5 is defined the same way, except for the disabled beneficiaries under age 65.
Measure 4 shows the proportion of persons who reported that they delayed seeking care because they were worried about the cost in the year preceding the interview.
The proportions in the tables are weighted, and standard errors are included in the tables in order to assess the impact of sampling variability on the accuracy of the estimates. They were estimated by using a software package (SUDAAN) that accounts for the impact of non-random sampling procedures on the precision of the weighted estimates.
Not all table cells have a value. If a cell had fewer than 30 observations or the relative standard error of the estimate was greater than 30, the value for that cell was suppressed. This rule was implemented to avoid reporting statistically unreliable estimates of the characteristics of elderly community residents.
Return to Appendix B: Detailed Methods
2006 National Healthcare Disparities Report