NOTE: Comparing Beneficiaries of the Medicare Savings Programs with Eligible Nonparticipants

Social Security Bulletin, Vol. 64, No. 3, 2001/2002

This note was prepared by James Sears, Division of Policy Evaluation, Office of Research, Evaluation, and Statistics, Office of Policy, Social Security Administration. An earlier version of this paper was presented at the November 2001 meetings of the Association for Public Policy Analysis and Management (APPAM). The author would like to thank participants in the APPAM session "Medicare Part B Buy-in: Addressing the Lack of Participation" for their questions and helpful suggestions.

The findings and conclusions presented in the Bulletin are those of the authors and do not necessarily represent the views of the Social Security Administration.


This note focuses on participation in two entitlement programs that help reduce out-of-pocket expenses for low-income Medicare beneficiaries: the Qualified Medicare Beneficiary (QMB) program and the Specified Low-Income Medicare Beneficiary (SLMB) program. As of 1999, about 2.75 million eligible, noninstitutionalized individuals were not enrolled in these Medicare savings programs. The eligible nonparticipants differed substantially from the QMB and SLMB participants in that they were less likely to be Supplemental Security Income beneficiaries and more likely to be elderly, nonblack, and in relatively good health. These findings, which could help target future outreach efforts, are based on Survey of Income and Program Participation data matched with administrative records from the Social Security Administration.


Medicare savings programs, also known as buy-in programs, use Medicaid funds to help reduce out-of-pocket expenses for Medicare beneficiaries who have limited income and resources. These programs initially had very low enrollment rates (see, for example, Families USA Foundation 1993 and U.S. General Accounting Office 1994). Although more recent outreach efforts and simplified enrollment processes have led to increased participation, many eligible individuals are still not enrolled. Future outreach efforts may be most effective if they are better targeted toward that group. To help improve the targeting of such efforts, this note describes the population of eligible nonparticipants and compares it with the beneficiary population.


The Qualified Medicare Beneficiary (QMB) and Specified Low-Income Medicare Beneficiary (SLMB) programs are both entitlement programs that cover the Medicare Part B premium. Beneficiaries can save over $600 per year by enrolling in either of these Medicare savings programs, since the Part B premium is typically $54 per month in 2002. The QMB program pays Medicare deductibles and coinsurance as well as premiums for individuals with income at or below the poverty guideline (that is, $759 per month for one person in 2002). The SLMB program serves people with income between 100 percent and 120 percent of the poverty guideline but covers only the Medicare Part B premium. For a more complete description of the Medicare savings programs, see Nadel and others (2000).

Rupp and Sears (2000) estimate that 63 percent of the 7.8 million eligible, noninstitutionalized individuals were enrolled in the QMB and SLMB programs as of 1999. Although this percentage is higher than most previous estimates, it still leaves about 2.5 million eligible individuals who are not enrolled. Rupp and Sears describe the entire eligible population, but they cannot distinguish which members of that group are beneficiaries of the Medicare savings programs. Indeed, no single data source is suitable for characterizing the eligible individuals who are not yet enrolled. Survey data are useful in describing the eligible population, but they generally do not reveal which individuals are already enrolled. Program records do identify the enrollees, but they provide little information about them. Only when linked with outside information are the program records useful for comparing participants with eligible nonparticipants.

Only one previous study of Medicare cost-sharing programs has used survey data matched with administrative indicators of enrollment. Under contract to the former Health Care Financing Administration (HCFA), the Barents Group (1999) analyzed Medicare Current Beneficiary Survey (MCBS) data matched with HCFA administrative records. The matched data combined reliable indicators of enrollment in the QMB and SLMB programs with self-reported health and sociodemographic information. However, the limited financial data provided by the MCBS hindered the identification of program eligibles. The Barents Group found that Supplemental Security Income (SSI) beneficiaries were much more likely to enroll in the QMB program--an expected finding since QMB enrollment is automatic for SSI beneficiaries in many states. The Barents Group also found an increased probability of QMB and SLMB enrollment associated with being young (under age 45), being black, or having an outpatient hospital visit during the year. Lower enrollment rates were associated with being married, reporting good or better health, or being at least age 80.


The study discussed here relies on Survey of Income and Program Participation (SIPP) data matched with Social Security Administration (SSA) administrative records. The SIPP provides detailed information about assets and income that allows precise estimation of eligibility in the QMB and SLMB programs. SSA records also improve the calculation of financial eligibility by providing reliable SSI and Social Security benefit amounts. Accurate identification of eligibles is critical to the study; if eligibility estimates are not reliable, then neither are descriptions of the eligible nonparticipants. One drawback of the matched SIPP vis-à-vis the Barents Group's matched MCBS is that it does not allow for separate identification of QMB and SLMB enrollees. Since any outreach effort would almost surely target persons eligible for either program, studying the QMB and SLMB programs together does not seem particularly problematic.

Virtually all of the SIPP data analyzed here come from wave 3 of the 1996 SIPP panel, and they were gathered between December 1996 and March 1997. This wave was chosen because it includes a topical module on assets, which is used in calculating QMB and SLMB eligibility. Some health-related variables are merged with the analysis file from wave 5 of the 1996 SIPP panel to provide additional descriptive statistics.

SIPP data have been matched to the Numident, which is SSA's master file of all Social Security number (SSN) holders and applications. The ability to match data depends on the respondents' providing their SSN. About 84 percent of adult respondents to wave 3 of the 1996 SIPP panel have Numident file matches. The match rate is slightly higher for people aged 65 or older (85.4 percent) than for those aged 18 to 64 (83 percent). SIPP weights are divided by these match rates so that the matched sample can represent the entire U.S. noninstitutionalized adult population.

The Numident match links SIPP data with administrative records from two other files: the Master Beneficiary Record (MBR), which contains data on Social Security receipt, and the Supplemental Security Record (SSR), which contains data on Supplemental Security Income receipt. The MBR also provides a history of state buy-in of the Medicare Part B premium, which indicates enrollment in a Medicare savings program.


Rupp and Sears (2000) use data from the 1993 SIPP panel along with SSA administrative records to estimate a participation rate for the combined QMB and SLMB programs as of 1999. Repeating this exercise but substituting the 1996 SIPP panel yields a 1999 participation rate of 61 percent. This rate is marginally lower than the one estimated by Rupp and Sears, and it suggests that about 2.75 million eligible people remain to be enrolled in the QMB and SLMB programs.

The one piece of administrative data not available at the time of the Rupp and Sears study was the history of enrollment in Medicare savings programs. That history is now available through March 2000, when the data were extracted. It divides the QMB- and SLMB-eligible population into three groups:

Among eligibles who were not enrolled at the time of the interview, 17.5 percent had state buy-in at some point before March 2000. Those late enrollees include individuals who had responded to some past outreach efforts; the nonenrollees are those who would be the focus of future outreach.


Various characteristics of the individuals in each of the three eligible groups are shown in Table 1 (a weighted tabulation based on 2,215 SIPP respondents who are eligible for the QMB or SLMB program). The data reveal several differences between participants in Medicare savings programs and eligible nonenrollees. The most dramatic difference concerns receipt of SSI. Over 64 percent of initial enrollees (those who were enrolled at the time of the interview) are SSI beneficiaries, as are just 20 percent of those who never enrolled. In this respect, the later enrollees are quite similar to those who never enrolled, with only 16 percent receiving SSI. However, for most other variables, the characteristics of later enrollees are somewhere between those of the initial enrollees and those of individuals who never enrolled. Enrollees are much more likely to be disabled (that is, under age 65) than are eligible nonparticipants. About 37 percent of initial enrollees and about 31 percent of later enrollees are younger than age 65, compared with only about 12 percent of those who never enrolled. The enrollees also appear disproportionately likely to be black, to be in fair or poor health, to be divorced or never married, to visit a doctor many times during the year, and to report difficulty with activities of daily living. Eligible widow(er)s appear less likely to enroll.

Table 1. Characteristics of persons eligible for Medicare savings programs, by enrollment status, 1996-1997
Characteristic Enrolled at interview Enrolled later Never enrolled All eligibles
Percentage distribution
39 or younger 11.9 10.4 2.7 7.3
40 to 49 9.1 10.2 3.6 6.6
50 to 59 10.8 7.1 3.4 6.8
60 to 64 5.0 3.1 2.1 3.4
65 to 69 17.2 19.5 20.5 19.0
70 to 79 29.4 29.4 42.3 35.6
80 or older 16.6 20.4 25.3 21.2
Total 100.0 100.0 100.0 100.0
Female 63.5 62.3 65.4 64.3
Male 36.5 37.7 34.6 35.7
Total 100.0 100.0 100.0 100.0
Race and ethnicity
White 55.3 62.8 60.9 58.8
Black 27.0 27.6 18.4 22.9
Asian or other 3.7 2.7 8.6 6.0
Hispanic 14.0 6.9 12.0 12.3
Total 100.0 100.0 100.0 100.0
Marital status
Spouse present 16.8 18.4 20.7 18.8
Separated or spouse absent 6.3 6.9 8.0 7.2
Divorced 18.0 14.1 12.1 14.8
Widowed 35.5 40.0 45.5 40.8
Never married 23.5 20.6 13.7 18.5
Total 100.0 100.0 100.0 100.0
Living arrangement
Alone 43.0 44.4 43.3 43.2
With others 57.0 55.6 56.7 56.8
Total 100.0 100.0 100.0 100.0
High school graduate 34.1 35.2 39.9 37.0
Less than high school 65.9 64.8 60.1 63.0
Total 100.0 100.0 100.0 100.0
Health status
Excellent 2.8 3.0 5.8 4.2
Very good 6.8 6.4 11.4 9.0
Good 19.2 23.8 30.3 25.0
Fair 39.1 41.2 30.5 35.2
Poor 32.1 25.6 22.1 26.6
Total 100.0 100.0 100.0 100.0
Doctor visits in past year
None 9.4 16.9 14.4 12.6
One 6.7 11.6 7.5 7.6
Two to four 26.8 27.5 32.1 29.4
Five or more 57.1 44.0 46.0 50.4
Total 100.0 100.0 100.0 100.0
Hospitalized in past 12 months
Yes 29.5 22.4 20.6 24.5
No 70.5 77.6 79.4 75.5
Total 100.0 100.0 100.0 100.0
Percentage reporting physical limitations
Has difficulty--
Seeing 22.7 25.6 17.8 20.6
Hearing 16.4 14.5 16.3 16.2
Lifting 10 pounds 45.0 44.0 36.5 40.8
Climbing stairs 57.0 54.0 43.8 50.4
Walking quarter mile 52.4 51.3 44.1 48.3
Using phone 11.3 13.0 8.2 10.0
Getting around home 16.3 14.8 12.6 14.4
Getting around outside 36.1 32.9 27.9 31.9
With chair or bed 19.9 20.0 14.5 17.3
Bathing 19.1 20.9 14.4 17.0
Dressing 12.9 12.9 10.0 11.5
Eating 6.4 6.7 4.2 5.4
Using toilet 10.1 11.6 6.1 8.3
Preparing meals 21.1 20.0 14.4 17.8
Doing light housework 26.2 24.2 17.7 22.0
Unable or requires help--
Lifting 10 pounds 17.2 19.7 14.2 16.0
Climbing stairs 28.9 29.1 25.7 27.4
Walking quarter mile 18.9 15.6 15.6 17.0
Using phone 5.9 7.7 4.5 5.4
Getting around home 8.6 7.1 6.3 7.4
Getting around outside 31.0 30.0 23.0 27.1
With chair or bed 9.2 8.1 6.2 7.6
Bathing 13.3 12.6 9.8 11.6
Dressing 8.8 8.1 6.6 7.7
Using toilet 7.0 7.3 4.4 5.8
Preparing meals 17.9 16.5 12.0 15.0
Doing light housework 21.1 18.5 12.7 16.9
Receiving SSI 64.3 16.1 20.1 38.3
SOURCE: Estimates are based on weighted data from the 1996 panel of the Survey of Income and Program Participation (SIPP) file matched to SSA administrative records. The SIPP data were gathered between December 1996 and November 1997.
NOTE: Percentages may not add to 100 because of rounding.

Given that SSI receipt plays a large role in the QMB enrollment process, some of the other variables discussed above may merely be capturing differences between SSI beneficiaries and others who are eligible for the QMB and SLMB programs. To permit comparisons in a multivariate framework and thus show the marginal effects of these other variables, the study used an unweighted logistic regression to investigate the probability among eligible individuals of not being enrolled (Table 2). The dependent variable takes on a value of one for a person who is enrolled in a Medicare savings program at the time of the SIPP interview and a value of zero for someone who is not enrolled then or later. Those who enroll in Medicare savings programs after the SIPP interview are excluded from this regression since they are not unambiguously defined as either enrollees or eligible nonparticipants. The explanatory variables are a parsimonious set of the concepts in Table 1 that could plausibly be associated with differential enrollment probabilities. A positive regression coefficient indicates a higher probability of not being enrolled.

Table 2. Probability that persons eligible for Medicare savings programs will never enroll
Characteristic Estimate Standard
Intercept 1.62 0.24
Younger than age 64 -1.35 0.15 0.26
Aged 80 or older 0.24 0.14 1.27
Black, non-Hispanic -0.45 0.12 0.64
Married with spouse present 0.12 0.16 1.12
Widowed -0.17 0.13 0.85
Living alone -0.36 0.12 0.70
High school graduate -0.20 0.12 0.82
Health good or better 0.66 0.11 1.93
Hospitalized in past 12 months -0.32 0.12 0.73
Receiving SSI -1.95 0.11 0.14
SOURCE: Estimates are based on weighted data from the 1996 panel of the Survey of Income and Program Participation (SIPP) file matched to SSA administrative records. The SIPP data were gathered between December 1996 and March 1997.
NOTE: Derived using an unweighted logistic regression.

As expected, SSI receipt has a very substantial effect. All else being equal, an SSI beneficiary appears only 14 percent as likely to not be enrolled as someone who does not receive SSI. However, several other variables also remain relevant. Enrollment is associated with being disabled, being black, having poor or fair health, and living alone. Indeed, marital status (including being a widow(er)) is the only characteristic identified in Table 1 that no longer appears to play a significant role in the multivariate framework, perhaps because of the high correlation between being a widow(er) and living alone.

The patterns that emerge from matched SIPP data are quite consistent with those reported by the Barents Group (1999) from matched MCBS data. Among QMB and SLMB eligibles, those who are still not enrolled are less likely to be SSI beneficiaries and more likely to be elderly, nonblack, and in relatively good health. Furthermore, these qualitative results are observed both individually and in a multivariate framework. The only major MCBS finding that was not fully replicated with the SIPP was related to marital status. Although SSI beneficiaries are unlikely to be married and likely to enroll in Medicare savings programs, this study provides only weak evidence that marriage itself is associated with a lower probability of enrolling.


Barents Group. 1999. A Profile of QMB-Eligible and SLMB-Eligible Medicare Beneficiaries. Report prepared for the Health Care Financing Administration under Contract No. 500-95-0057/Task Order 2, Washington, D.C., April 7.

Families USA Foundation. 1993. The Medicare Buy-in: A Promise Unfulfilled. Washington, D.C.: Families USA Foundation.

Nadel, Mark; Lisa Alecxih; Rene Parent; and James Sears. 2000. "Medicare Premium Buy-in Programs: Results of SSA Demonstration Projects." Social Security Bulletin 63(3): 26-33.

Rupp, Kalman, and James Sears. 2000. "Eligibility for the Medicare Buy-in Programs, Based on a Survey of Income and Program Participation Simulation." Social Security Bulletin 63(3): 13-25.

U.S. General Accounting Office. 1994. Medicare and Medicaid: Many Eligible People Not Enrolled in Qualified Medicare Beneficiary Program. GAO/HEHS-94-52. Washington, D.C.: U.S. General Accounting Office.