Skip to main content
American Journal of Public Health logoLink to American Journal of Public Health
. 2004 Oct;94(10):1795–1799. doi: 10.2105/ajph.94.10.1795

Changes in Racial Differences in Use of Medical Procedures and Diagnostic Tests Among Elderly Persons: 1986–1997

José J Escarce 1, Thomas G McGuire 1
PMCID: PMC1448536  PMID: 15451752

Abstract

Objectives. We used 1997 Medicare data to replicate an earlier study that used data from 1986 to examine racial differences in usage of specific medical procedures or tests among elderly persons.

Methods. We used 1997 physician claims data to obtain a random sample of 5% of Medicare beneficiaries aged 65 years and older. We used this sample to study 30 procedures and tests that were analyzed in the 1986 study, as well as several new procedures that became more widely used in the early 1990s.

Results. Racial differences remain in the rates of use of these procedures; in general, Blacks have lower rates of use than do Whites. Between 1986 and 1997, the ratio of White to Black use moved in favor of Blacks for all but 4 of the established procedures studied.

Conclusions. The White–Black gap in health care use under Medicare is narrowing.


Racial and ethnic disparities in health care use in the United States are pervasive and widely acknowledged.1,2 The Institute of Medicine’s recently released Unequal Treatment2 contains an appendix with 600 articles documenting disparities. Although empirical research on disparities has burgeoned in the past decade, important gaps remain in the literature. The Institute’s report concludes that little is known about the mechanisms that generate disparities from within the clinical encounter; much less is known about disparities for non-Black racial or ethnic minorities. Also, because most research on disparities features data at a single point in time, we know little about whether disparities between Blacks and Whites are growing or shrinking. Researchers have speculated on how changes in the organization and financing of health care, particularly the growth of managed care, may influence the magnitude of racial and ethnic disparities in treatment, but data documenting trends in disparities are scarce. In a recent review, Gornick3 concludes that racial disparities in Medicare “tend to persist over time” but the most recent data cited on surgical procedures were for 1992.

Data from the Medicare program offer an opportunity to monitor trends over time. Medicare records census-like information about race and ethnicity in the enrollment process, and this information is reliable for Blacks and Whites.4 Medicare provides health insurance to virtually every US adult older than 65 years, about 40 million people in total, and is the nation’s largest single payer of hospital care. In 1997, 87.0% of Medicare beneficiaries were White, 8.4% were Black, and 2.3% were Hispanic, with the balance distributed among persons of Asian/Pacific Islander, American Indian/Alaska Native, Other, and Unknown race/ethnicity.4 Enrollment and health care use data in the program have been stable for many years. Some of the more notable studies documenting health care disparities have used Medicare data.5–7 In the report by Escarce et al. that served as a model for our study,8 Medicare data for 1986 were used to study differences in health care use between Blacks and Whites for 32 selected procedures and tests. For most services, and particularly for newer or high-technology services, Escarce et al. found that Whites had age- and gender-adjusted rates of use that exceeded those of Blacks.

Our study replicates the Medicare 1986 analysis with data from 1997. We used the same data files and examined the same procedures, defined in the same way, as did the 1986 study; we also examined several newer procedures that became widely used in the early 1990s. For purposes of comparison, we adjusted the data by means of the same methods used in the earlier article, thereby permitting direct comparison of rates over time.

METHODS

Data and Study Population

Two data files were merged to combine information on health care use with information about beneficiaries. The 1997 Part B Beneficiary File we obtained from the Centers for Medicare and Medicaid Services contains a detailed record of physician and other services paid within the fee-for-service system of Medicare on behalf of a 5% sample of beneficiaries. Information about these beneficiaries is contained in the enrollment database also maintained by the Centers. Available information includes age, gender, race or ethnicity, date of Medicare eligibility, Medicaid eligibility status, health maintenance organization (HMO) enrollment status, and date of death (if applicable). To replicate the Escarce et al. study, we examined only Whites and Blacks. Data on race and ethnicity were derived from the Social Security Administration’s Master Beneficiary Record. Owing to expansions in the racial and ethnic categories, the definition of White and Black is not completely comparable between 1986 and 1997. For 1997, when the enrollment data were compared with the more detailed information collected face-to-face for the Health Care Financing Administration Medicare Current Beneficiary Survey, both the sensitivity (97% and 95%, respectively) and the specificity (89% and 99%, respectively) of the White and Black categories were very high.4 Certain non-Black minorities were probably more frequently misclassified in 1986. However, because these racial/ethnic minorities’ numbers were small in 1986, any bias resulting from this measurement error can be ignored.

As in the earlier study, Medicare beneficiaries younger than 65 years and those with end-stage renal disease were excluded, as were individuals with partial-year enrollment or HMO enrollment. HMO enrollment increased from about 6% in 1986 to 16.8% in 1997, a finding that may indicate that the fee-for-service risk pool had on average become sicker by 1997. (Values for 1997 are derived from our own data analysis.) However, because the 1997 rates of HMO enrollment for Blacks (16.2%) and Whites (16.5%) were very similar, differential selection outside the fee-for-service sector was unlikely to affect Black–White comparisons. Our study sample consisted of 1 547000 persons older than 65 years, about 25% more than in the earlier study.

Study Services

We examined 30 of the 32 medical procedures studied in the 1986 analysis. Two groups of cerebrovascular procedures were dropped because they contained a heterogeneous mix of services for which a comparison over time would have little value. We also examined 11 new procedures that entered common use in the early 1990s. As in the original study, the selection of these 11 procedures was based on the frequency of the procedure’s use or on its importance in treating conditions with high morbidity or mortality. Although rates for the new procedures cannot be compared with corresponding rates from 1986, we thought it would be useful to include them in the analysis.

Study services were defined by CPT-4 codes used as a basis for procedure information in Medicare. Procedures and tests are often defined by more than 1 code. For purposes of rate calculations, we counted only 1 of each procedure or test for each person per year. Rates therefore should be interpreted as rates of persons with at least 1 procedure or test in each category per year. The algorithms used to define the procedures are available upon request.

Data Analysis

For each service studied, we calculated age-and gender-adjusted rates of use among elderly Blacks and Whites. Adjustment for age and gender was by direct standardization with 1997 population weights. We assessed differences in rates of use by comparing Blacks to Whites with relative risks (RRs) adjusted for age and gender by the Mantel–Haenszel method.9 Test-based 95% confidence intervals (CIs) were used to check the association of race with differences in rates. When we compared relative risks for the same service in 1986 and 1997, we assumed that the relative risks were statistically independent.

RESULTS

Elderly Whites comprised 92.22% of the study population; elderly Blacks made up 7.78%, a slight increase from 1986 (Table 1). Elderly Blacks were more likely than elderly Whites to be female, to be younger, and to live in an urban county. In spite of being younger and more often female, Blacks died at a higher rate than did Whites. The largest difference between Black and White beneficiaries was in the rate of Medicaid coverage: 26.7 % of elderly Blacks in Medicare also had Medicaid coverage, compared with only 6.9% of elderly Whites.

TABLE 1—

Characteristics of 1997 Study Population, by Race/Ethnicitya

Whites, No. (%) (n = 1 380 570) Blacks, No. (%) (n = 116 406)
Fraction of total study population 92.22% 7.78%
Gender
    Male 556 611 (40.32) 43 226 (37.13)
    Female 823 959 (59.68) 73 180 (62.87)
Age, y
    65–69 368 271 (26.68) 33 845 (29.07)
    70–74 359 932 (26.07) 31 031 (26.66)
    75–79 290 712 (21.06) 22 938 (19.71)
    80–84 197 725 (14.32) 15 236 (13.09)
    ≥85 163 930 (11.87) 13 356 (11.47)
Residenceb
    Rural 342 088 (25.11) 20 684 (18.2)
    Urban 1 020 291 (74.89) 92 989 (81.8)
Died during 1997 66 813 (4.84) 6 476 (5.56)
Medicaid coverage in December 1997c 94 516 (6.85) 31 096 (26.71)

a Full sample of 5% denominator file used.

b “Urban” refers to residence in a metropolitan county; “rural” refers to residence in a nonmetropolitan county.

c Elderly persons who had both Medicaid and Medicare coverage.

Table 2 is comparable to Table 2 in the 1986 Escarce et al article. Columns 1, 3, and 5 show new data from 1997, whereas columns 2,4, and 6 for 1986 are taken directly from the earlier report. A comparison of 1997 rates for the procedures analyzed in the previous article with the rates from 1986 shows substantial changes in rates, although the rates for Blacks and Whites moved in the same direction. The only exception was the cardiac exercise stress test, for which the White rate fell slightly and the Black rate rose slightly. The changes in rates in the cardiac area were particularly large, increasing by more than a factor of 10 between 1986 and 1997 for radionuclide stress tests and Doppler echocardiograms. Rates at least doubled for coronary angiograms, coronary angioplasty, bypass surgery, and Swan-Ganz catheterization. Rates for 2D and M-mode echocardiograms, by contrast, fell to a third of their previous levels.

TABLE 2—

Age- and Gender-Adjusted Rates of Use of Study Services per 10 000 Elderly Population, by Race and Year

Service Whites, 1997 Whites, 1986 Blacks, 1997 Blacks, 1986 White–Black RR (95% CI), 1997 White–Black RR, 1986
Cardiac procedures
    Coronary angiogram 228.3 85.6 160.8 43.1 1.40 (1.34, 1.47)** 1.97 (1.79, 2.17)
    Coronary angioplastya 25.7 9.9 15.9 3.2 2.01 (1.71, 2.37)** 3.11 (2.20, 4.38)
    Coronary artery bypass surgery 58.6 30.6 26.4 8.1 2.18 (1.94, 2.44)** 3.73 (3.03, 4.61)
    Radionuclide stress testa,b 325.9 24.5 236.4 14.6 1.72 (1.63, 1.82) 1.67 (1.41, 1.98)
    Exercise stress testa 206.4 215.2 119.6 109.7 1.37 (1.32, 1.42)** 1.95 (1.84, 2.07)
    Swan–Ganz catheterization 84.0 38.8 54.3 31.1 1.52 (1.40, 1.65)** 1.24 (1.10, 1.39)
    Permanent pacemaker 38.7 25.4 29.7 20 1.30 (1.16, 1.45) 1.27 (1.09, 1.47)
    Doppler echocardiograma 922.9 34.7 1006.6 27.7 0.90 (0.88, 0.92)** 1.25 (1.10, 1.42)
    Echocardiograma 112.2 388.8 139.3 443.2 0.79 (0.75, 0.84)* 0.87 (0.85, 0.90)
    Electrocardiogram 3482.3 3535.4 3288.8 3232.5 1.08 (1.07, 1.10) 1.09 (1.08, 1.10)
    Holter monitor 247.4 258.6 0.95 (0.91, 0.98)
    Coronary atherectomya 4.8 2.8 1.68 (1.18, 2.38)
    Coronary stenta 36.9 18.6 1.95 (1.70, 2.23)
    Exercise echocardiograma 84.9 53.6 1.56 (1.44, 1.69)
    Nonsurgical revascularization 61.6 31.0 1.95 (1.76, 2.17)
Cerebrovascular procedures: carotid endarterectomy 28.3 17.3 11.2 5.8 2.50 (2.10, 2.97)** 3.01 (2.32, 3.89)
Gastrointestinal (GI) procedures
    Flexible sigmoidoscopya 206.1 276.6 136.9 183.1 1.55 (1.47, 1.63) 1.51 (1.44, 1.59)
    Rigid sigmoidoscopya 33.2 191.2 18 96.5 1.79 (1.57, 2.04)* 1.98 (1.85, 2.11)
    Colonoscopya 407 172.9 372.9 141.4 1.12 (1.09, 1.16)** 1.22 (1.16, 1.29)
    Barium enemaa 171.4 491.8 190.9 501.4 0.90 (0.86, 0.94)** 0.98 (0.96, 1.01)
    Upper GI endoscopya 336.8 187.9 427.1 201.6 0.80 (0.78, 0.83)** 0.93 (0.89, 0.98)
    Upper GI seriesa 188.2 505.5 209.6 533.3 0.90 (0.87, 0.94) 0.95 (0.92, 0.98)
    ERCP 22.2 21.3 1.05 (0.93, 1.20)
Ophthalmologic procedures
    Cataract extraction with lens 371 278.1 275.3 194.5 1.36 (1.32, 1.41)* 1.43 (1.36, 1.50)
    Laser trabeculoplasty 18.5 16.1 52.4 31.2 0.35 (0.32, 0.39) 0.51 (0.46, 0.58)
    Glaucoma surgeryc 13.9 8 34.6 17.9 0.40 (0.36, 0.44) 0.44 (0.38, 0.52)
    Retinal photocoagulation 17.8 15.2 35.7 27.3 0.47 (0.43, 0.52) 0.55 (0.48, 0.63)
Orthopedic procedures
    Total hip replacement 23 22.5 12.4 9.5 1.99 (1.68, 2.35)** 2.36 (1.92, 2.89)
    Total knee replacement 43.7 18.2 30.1 8.9 1.45 (1.30, 1.61)** 2.02 (1.63, 2.49)
General surgical procedures
    Cholecystectomy 55.38 49.3 35.65 28 1.57 (1.42, 1.72)* 1.76 (1.56, 1.99)
    Inguinal hernia 45.05 3.7 24.73 16.5 1.91 (1.69, 2.16) 2.05 (1.75, 2.40)
Urological procedures: transurethral prostatectomyd 75 208.7 61.6 199.4 0.94 (0.85, 1.05) 1.04 (0.97, 1.12)
Miscellaneous imaging procedures
    Mammograme 2338.2 753.6 1288.2 429.6 1.61 (1.57, 1.64)** 1.76 (1.70, 1.83)
    Magnetic resonance imaging (brain)a 194.7 17.1 173.8 9.6 1.12 (1.07, 1.17)** 1.80 (1.46, 2.21)
    Computed tomography (head)a 576.8 456.8 754 518.5 0.76 (0.74, 0.78)** 0.90 (0.87, 0.92)
    Chest radiograph (two views) 2559.2 2738.7 2369.8 2430.1 1.13 (1.11, 1.15) 1.13 (1.11, 1.14)
    Upper-extremity magnetic resonance imaging 22.6 14.5 1.56 (1.35, 1.81)
    Lower-extremity magnetic resonance imaging 40.7 27.5 1.52 (1.36, 1.69)
    Magnetic resonance angiogram 40.8 38.3 1.06 (0.97, 1.16)
    Spine computed tomographya 79.7 59.2 1.36 (1.26, 1.47)
    Spine magnetic resonance imaging 159.4 113.9 1.41 (1.34, 1.49)

Note. CI = confidence interval; RR = relative risk; ERCP = endoscopic retrograde cholangiopancreatography.

a The 8 sets of services contrasting newer vs more established or higher- vs lower-technology services (the newer or higher-technology service is listed first) are exercise echocardiogram or radionuclide stress test vs exercise stress test, coronary stent or coronary atherectomy vs coronary angioplasty (without stenting), Doppler echocardiogram vs 2D or M-mode echocardiogram, flexible sigmoidoscopy vs rigid proctosigmoidoscopy, colonoscopy vs barium enema, upper GI endoscopy vs upper GI series, magnetic resonance imaging vs computed tomography, and magnetic resonance imaging of the spine vs spine computed tomography.

b Includes radionuclide ventriculography and myocardial perfusion scanning.

c Includes fistulization of sciera, fridectomy, and trabeculectomy ab extremo.

d Men only.

e Women only.

* P < .05 for test of difference of RRs in 1986 and 1997; **P < .01 for test of difference of RRs in 1986 and 1997. All P values are 2-tailed.

Table 2 also shows the usage rates for 5 new cardiac procedures added for the 1997 analysis. The rate of coronary stents had surpassed the rate of angioplasty without stenting by 1997, whereas the rate of coronary atherectomy was 10 times lower. The rate of use of some form of nonsurgical revascularization (coronary angioplasty, stenting, or atherectomy) was 61.6 per 10000 for Whites and 31.0 per 10000 for Blacks in 1997. (New procedures added for this analysis are shown in italics and are grouped at the bottom of each category.)

Rates for 4 of the 6 gastrointestinal procedures fell over the period. Only colonoscopy and upper gastrointestinal endoscopy showed an increase in rates. The rates for cerebrovascular, ophthalmologic, orthopedic, and general surgical procedures all increased. Rates for the single urological procedure, transurethral prostatectomy, fell to a third of the 1986 levels. (This rate was calculated for men only.)

The application of imaging technology changed rapidly between 1986 and 1997. Although rates of computed tomography of the head and radiographs of the chest were stable, the mammogram rate tripled for White and Black women. Rates of brain magnetic resonance imaging increased by a factor of 11 for Whites and a factor of 18 for Blacks. Rates for a number of new imaging procedures are also included in the table. By 1997, magnetic resonance imaging of the spine had become much more common than spine computed tomography.

Our main interest is in the relative risks shown in the fifth and sixth columns of Table 2. Column 5 shows the White–Black relative risks for 1997 and Column 6 shows the 1986 data presented by Escarce et al. The White rate significantly exceeds the Black rate for 27 of the 41 procedures reported for 1997 (the lower bound of the 95% confidence interval for the relative risk for 1997 was greater than 1.0). In this respect, the findings for 1997 are similar to the 1986 results, in which we see in column 6 that the White relative risk significantly exceeds 1.0 for 21 of the 30 procedures. However, for 26 of the 30 procedures that are comparable between 1986 and 1997, the relative risk stayed the same or fell between 1986 and 1997. The only exceptions were 1 gastrointestinal and 3 cardiac procedures. Swan–Ganz catheterization was the single procedure for which the rise in the relative risk from 1986 to 1997 was statistically significant. The general decrease in the relative risk was observed both for procedures in which the 1986 relative risk was greater than 1 (Whites used procedure more than did Blacks) and for procedures in which the relative risk was less than 1 (Blacks used procedure more than did Whites).

When we examined the magnitude of the decline in relative risks, we found that 3 of the 30 procedures that are comparable between 1986 and 1997 had relative risks exceeding 3.0 in 1986; relative risks for these 3 fell below 3.0 in 1997. Another 3 of the 30 procedures had relative risks between 2.0 and 3.0 in 1986; relative risks for these 3 fell below 2.0 in 1997. Overall, the relative risk fell significantly for 18 of the 30 procedures reported for both years. For example, in the cardiac group of services, the relative risk for the first procedure, coronary angiograms, fell to 1.40 (95% CI = 1.34, 1.47) in 1997 from 1.97 (95% CI = 1.79, 2.17) in 1986. This reduction is both large (a nearly 50% reduction in the relative risk) and significant.

Our analyses found that White–Black relative risks were higher for the newer or higher-technology services in 4 of the 8 groups of services that contrasted newer with more established services or higher-technology with lower-technology services. However, even here, White–Black differences were less marked in 1997 than in 1986. For example, in 1986 the relative risk for the higher-technology Doppler echocardiogram was 1.25 (95% CI = 1.10, 1.42) and the relative risk for the lower-technology 2D or M-mode echocardiogram was 0.87 (95% CI = 0.85, 0.90); by 1997 the relative risks were 0.90 (95% CI = 0.88, 0.92) and 0.79 (95% CI = 0.75, 0.84), respectively. Similar patterns were observed for colonoscopy compared with barium enema and for magnetic resonance imaging of the brain compared with computed tomography.

DISCUSSION

In its recent report, the Institute of Medicine handed down a flat indictment of the US health care system: “Racial and ethnic disparities in health care exist and, because they are associated with worse outcomes in many cases, are unacceptable.”2(p6) At the same time, the report pointed out that there are very few studies of how changes in the US health care system are affecting the magnitude of disparities.

Medicare is the largest single payer for health care in the United State, and its data provide the ability to compare rates of use for categories of beneficiaries over time. About 85% of Medicare beneficiaries receive care from regular Medicare, in which case procedure data are reported for payment and have been recorded in the same format for many years. Medicare contains large numbers of beneficiaries in all parts of the country, including growing numbers of Latinos, Asians, and other minorities. These groups are growing much faster among the elderly population than are Blacks or Whites. In other research,10 we have shown that small sample sizes do not prevent estimation of disparities in Medicare use for these groups, and that misreporting of race or ethnicity has a relatively small effect on Latino–White comparisons.

The striking finding from our analyses is that the White–Black disparity in procedure use in Medicare is narrowing. Specifically, the White–Black relative risk declined significantly for 18 of the 30 procedures that we could compare between 1986 and 1997. The changes in the relative rates were large, erasing in some cases up to 50% of the original racial difference in rates over the 10-year period. Results of our analyses that contrasted newer procedures with more established ones or lower-technology services with higher-technology services were consistent with this general trend. The earlier finding that Whites’ advantage over Blacks was greater for newer or higher-technology services was less apparent in the 1997 data.

Despite these encouraging findings, clinically important disparities in the rates of specific medical procedures remain, and they are especially prevalent in cardiac care. White beneficiaries were more than twice as likely as Black beneficiaries to receive coronary artery bypass surgery in 1997, and they were about twice as likely to receive nonsurgical revascularization. Furthermore, the relative risks for noninvasive tests for myocardial ischemia—exercise stress test, radionuclide stress test, and exercise echo—and for coronary angiography, although lower in 1997 than in 1986, still reflect sizable disparities. Racial differences in the aggressiveness with which coronary artery disease is diagnosed and treated may result in differences in outcomes. Carotid endarterectomy, total hip and total knee replacement, and inguinal hernia repair were other procedures for which we found substantial racial differences in use in 1997.

Our analyses cannot identify the reason for the narrowing White–Black disparity in procedure use. Nonetheless, 3 possibilities are worth mentioning. First, extension of dual eligibility for Medicaid in the late 1980s to a higher proportion of the low-income elderly population decreased out-of-pocket payments for this group.11 Second, the resource-based Medicare Fee Schedule,12 implemented in the early 1990s, reduced differences in physician payment rates for Medicare beneficiaries with private supplementary (Medigap) insurance versus those with dual Medicaid coverage, removing incentives for physicians to provide more services to high-income elderly persons whose private policies had paid higher fees. Because Black elderly persons had lower incomes than their White peers, they may have been helped more by these policy changes. Third, increased awareness among physicians of racial differences in medical care use also may have contributed to shrinking the White–Black gaps.13

It should be acknowledged that Medicare claims data contain limited information on health status. Studies focused on selected disease areas could confirm our finding that the White–Black disparity is narrowing. Another possibility is that the overall health of the Black population worsened in relation to that of the White population over this period, perhaps owing to differential HMO enrollment. Although national rates of Medicare HMO enrollment are similar for Blacks and Whites, important regional differences could exist in these patterns. If the average health of Blacks in regular Medicare were declining relative to that of Whites, the convergence in rates could be attributable to a decline in health status among Blacks in regular Medicare.

Our analysis raises several important questions. If policy changes or other system changes were associated with the relative reduction of disparities, which policies or system changes were the most beneficial? Alternatively, if increasing physician awareness of disparities were chiefly responsible for the findings, how can this awareness be heightened and fostered to improve clinical decisionmaking?

The findings of our study also suggest that data on racial disparities in health care should be continually updated. Medicare data are well-suited for this purpose, although data on disparities should include the nonelderly population as well.

Acknowledgments

This research was supported by the National Institute of Mental Health (grants P01 MH59876 and R01 MH59254).

We are grateful to Zhun Cao and JiTian Sheu for programming assistance.

Human Participant Protection

No protocol approval was needed for this study; data used were publicly available Medicare data with no identifiers.

Contributors…J. J. Escarce developed the clinical algorithms applied in the study. J. J. Escarce and T.G. McGuire both planned the study, supervised the data analysis, and wrote the article.

Peer Reviewed

References

  • 1.Mayberry R., Mili F, Ofili E. Racial and ethnic differences in access to medical care. Med Care Res Rev. 2000;57:108–145. [DOI] [PubMed] [Google Scholar]
  • 2.Institute of Medicine. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. Washington, DC: National Academy of Sciences; 2002.
  • 3.Gornick ME. A decade of research on disparities in Medicare utilization: lessons for the health and health care of vulnerable men. Am J Public Health. 2003;93:753–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Arday SL, Arday DR, Monroe S, Zhang J. HCFA’s racial and ethnic data: current accuracy and recent improvements. Health Care Financing Rev. 2000;21:107–116. [PMC free article] [PubMed] [Google Scholar]
  • 5.McBean AM, Gornick ME. Differences in the rates of procedures performed in hospitals for Medicare beneficiaries. Health Care Financing Rev. 1994;15:77–90. [PMC free article] [PubMed] [Google Scholar]
  • 6.Gornick ME, Eggers PW, Reilly TW, et al. Effects of race and income on mortality and use of services among Medicare beneficiaries. N Engl J Med. 1996; 335:791–799. [DOI] [PubMed] [Google Scholar]
  • 7.Ayanian JZ, Udvarhelyi IS, Gatsonis CA, Pashos CL, Epstein AM. Racial differences in the use of revascularization procedures after coronary angioplasty. JAMA. 1993;269:2642–2646. [PubMed] [Google Scholar]
  • 8.Escarce JJ, Epstein KR, Colby DC, Schwartz JS. Racial differences in the elderly’s use of medical procedures and diagnostic tests. Am J Public Health. 1993; 83:948–954. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic Research: Principles and Quantitative Methods. New York, NY: Van Nostrand Reinhold Co; 1982.
  • 10.Escarce JJ, McGuire TG. Methods for using Medicare data for comparing rates among Asians, Blacks, Latinos and Whites. Health Serv Res. 2003;38:1303–1318. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Barents Group. A profile of QMB-eligible and SLMB-eligible Medicare beneficiaries. Washington, DC: Barents Group; 1999.
  • 12.Health Care Financing Administration, US Dept of Health and Human Services. Medicare program: fee schedule for physicians’ services—final rule. Federal Register. 25November1999;56:59502–59811. [Google Scholar]
  • 13.American Medical Association, Council on Ethical and Judicial Affairs. Black–White disparities in health care. JAMA. 1990;263:2344–2346. [DOI] [PubMed] [Google Scholar]

Articles from American Journal of Public Health are provided here courtesy of American Public Health Association

RESOURCES