The authors aimed to address two questions: (1) how do race and income affect mortality and use of services covered by Medicare and (2) how much do racial differences diminish when the rates are standardized according to income? Data for this study included the 1993 file of Medicare enrollment and median household income by zip code from the 1990 United States Census.
This analysis of Medicare data on hospital discharges among people 65 years of age and older revealed that black beneficiaries underwent 17 common procedures less often than white beneficiaries. The data also indicated that certain other less common procedures were performed more frequently among blacks.

The study found substantial race and gender differences in the 1993 mortality rates among Medicare beneficiaries. The mortality rate was 8.0 per 100 for black men, 6.7 per 100 for white men, 5.2 per 100 for black women, and 4.5 per 100 for white women. In each subgroup except black women, the highest income group had the lowest mortality rates.

In 1993, black Medicare beneficiaries made 7.2 visits per person to physicians for ambulatory care, as compared with 8.1 visits per person among whites, resulting in a statistically significant black:white ratio of 0.89. However, black beneficiaries had 376 hospital discharges per 1000 persons, 14% higher than the rate of 329 among whites. The least affluent white beneficiaries (as measured by household income by zip code) visited physicians for ambulatory care 18% less often than the most affluent whites (7.3 versus 9.0 visits), and were discharged from the hospital 24% more often than the most affluent group. Among black beneficiaries, the lowest income group also had fewer visits (7.1 versus 8.0 visits), but there were no income-related patterns in hospitalization rates.

With regard to racial patterns in cardiac procedures, the black:white ratio was 0.46 for percutaneous transluminal coronary angioplasty (PCTA) and 0.40 for coronary-artery bypass surgery (CABG). There were no income-related patterns for cardiac procedures for whites, but among blacks, the least affluent had a 24% lower angioplasty rate and a 16% lower CABG rate than the most affluent group. These patterns were not consistent with the racial/income-group patterns in the rate of hospitalization for ischemic heart disease.

There were race and income patterns in the use of other procedures also. The black:white ratio for mammogram was 0.66. Among whites, the rate for the least affluent women was 33% less than that for the most affluent women; among blacks, it was 22% lower. The rate of reduction of fracture of the femur was lower among black than white women (black:white ratio=0.42), but hip repair rates differed only slightly by income. The rate of amputation of all or part of the lower limb and the rate of orchiectomy were higher among blacks compared with whites (black:white ratio for amputation = 3.64; black:white ratio for orchiectomy = 2.45). Income group had a larger effect on the use of these procedures for whites than for blacks. (The authors also add that there was a race effect in the use of immunizations, with a black:white ratio=0.61. Income was measured individually using the Medicare Current Beneficiary Survey rather than zip code income data, and it showed that the least affluent blacks and whites had lower immunization rates, 26% lower for whites and 39% lower for blacks.)

Adjusting the rates of mortality and use of services among blacks and whites for difference in income affected the black:white ratios relatively little, although generally such adjustment reduced the differences between the races. Similarly, when using the individual income data from the Medicare Current Beneficiary Survey for the 845 blacks and 7911 whites, the analyses of the effects of individual income generally validated the analyses of the effects of income by zip code (presented above). The one exception for whites was visits to physicians for ambulatory care. For that variable, use of individual income (Medicare Current Beneficiary Survey) showed no income effect, whereas the analysis based on income data by zip code area (Census data) showed a moderate effect. For blacks, the race difference in mammography was more pronounced when using the Medicare Current Beneficiary Survey than when using the zip code income data.

Based on their analyses using the sub-sample of patients with individual-level income data, the authors note that their evaluation in the main analysis, which used zip-code income estimates, is likely to reflect the correct direction of income effect but is also likely to underestimate the magnitude of the effect of income. This finding is important to interpreting other studies that use similar approaches.

In sum, the patterns found indicated that blacks and lower income beneficiaries received less preventive services and were at higher risk for procedures associated with less than optimal management of chronic disease. The authors suggest that these racial and income patterns "may reflect a multitude of factors, including education, cultural and behavioral variables; individual preferences; differences in supplementary insurance; and the availability of services." The small and inconsistent effects of income on rates of hip fracture repair suggests that race and SES may not play an important part in access to non-elective services for elderly people enrolled in Medicare. The findings related to racial and SES patterns in coronary procedures and other common surgical procedures suggest that there may be barriers to elective surgical procedures for some groups of beneficiaries. The authors conclude that the differential patterns in the use of many specific services according to race and income indicate that the provision of health insurance alone does not suffice to promote effective patterns of use by all beneficiaries.