Black Women Receive Less Mammography Even with Similar Use of Primary Care

Methods

Data Sources

Two databases were obtained from the Health Care Financing Administration for 1990. Data were taken from 10 geographically and demographically diverse states: Alabama, Arizona, Connecticut, Georgia, Kansas, New Jersey, Oklahoma, Pennsylvania, Oregon, and Washington. The first database had one record for each Medicare beneficiary and listed sociodemographic information (age, race, state, and ZIP code of residence), dates of Medicare part A (hospital) and part B (physician) coverage, reason for entitlement, and date of death. The second database was a complete physicians' claims file (Medicare part B) that contained one record for each part B bill that was reimbursed by Medicare during 1990. Each claim listed the procedure code for the service provided according to the Physicians' Current Procedural Terminology (CPT) [15] and the specialty of the physician providing the service. The database was edited to remove all duplicate, reprocessed, and denied claims. We also obtained 1990 U.S. Census data by ZIP code; the data included median household income. The two Health Care Financing Administration databases were merged using the beneficiary's identification number, and this file was merged with the census data by ZIP code.

Study Sample

We identified 3 795 097 women who were 65 years of age and older as of 1 January 1990 and who resided in 1 of the 10 states. Approximately 12% of eligible women were excluded because they had had less than 1 year of Medicare part B coverage because of death or enrollment in a health maintenance organization. These women were excluded to ensure that comprehensive claims information was available for all physician services provided in 1990. Because the analysis was restricted to black women and white women, women of other races were excluded (4%). The final study sample consisted of 3 187 116 black or white women 65 years of age and older; the women resided in one of the 10 states and were alive through 31 December 1990.

Study Variables

Physicians' claims data were used to identify women in whom bilateral mammography had been done in 1990 [CPT code 76091]. In 1990, the Health Care Financing Administration did not reimburse providers for screening mammography. The criteria for diagnostic bilateral mammography did, however, essentially allow mammographies done for screening purposes to be included under the diagnostic code. The criteria were as follows: 1) distinct signs and symptoms for which mammography is indicated; 2) a history of breast cancer; or 3) lack of symptoms but, on the basis of the patient's history and other factors, the physician's judgment is that mammography is appropriate [16]. The third criterion was sufficiently vague to allow screening mammography to be done and billed under this code.

Women were considered to have had a primary care visit if they saw a physician who specialized in general practice, family practice, internal medicine, geriatrics, or obstetrics and gynecology. Visits could occur in the office (CPT codes 90010 to 90080), at home (CPT codes 90100 to 90170), in a nursing home (CPT codes 90300 to 90370), or in a rest home (CPT codes 90400 to 90470). Among women who had made at least one visit to a primary care physician, most (95%) had had at least one visit in an office. The number of primary care visits was categorized as 0, 1, 2, and 3 or more.

Because individual socioeconomic information is not available from Medicare, median household income by ZIP code was used as a proxy for socioeconomic status. For each state, the median income by ZIP code was used to divide the population into quintiles. Each woman was assigned to a quintile according to the median income of the ZIP code of the area in which she resided.

Statistical Analysis

Each state was analyzed separately. First, we determined the percentage of women who had had mammography in 1990. We used the chi-square test to identify bivariate associations with mammography use by race, number of primary care visits, income quintile, and state. We analyzed strata defined by the number of primary care visits, income quintile, and state to determine the difference in mammography use between black women and white women. We repeated the analysis within strata defined by income quintile and state to include only the subset of women who had made at least one visit to a primary care physician.

We describe health care use for nearly 3.2 million women who were 65 years of age and older and who were enrolled in the Medicare program in 10 selected states in 1990. Throughout most of this article, we report the observed use of mammography and the difference in use in selected subpopulations (for example, mammography use for black women and white women with one primary care visit). These data are shown separately for Georgia and Alabama, the two states with the most black women (both of these states have more than 50 000 black women each), so that all reported categories pertain to at least several thousand women. Because of the large numbers and the descriptive nature of these full-population findings, no inferential data techniques were used (we do not report P values or CIs for this data). Similar associations were seen in all states (details for the other 8 states are available from Dr. Burns).

We separately fit logistic models to predict mammography use according to age, number of primary care visits (0, 1, 2, and 3 or more), income quintile, and state of residence for the 2.9 million white women and the 250 000 black women in the 10 states. Because the purpose of the modeling is inferential, we report odds ratios and 95% CIs associated with the variables of interest for these models. All statistical analyses were done using SAS statistical software, version 6.08 (SAS Institute, Cary, North Carolina) [17].

Results

The sociodemographic characteristics of the study sample by state are shown in Table 1. Seven percent of the sample was black; the mean age for the women of both races was 75 years. The percentage of black women varied across the 10 states. Only 1% of the women in Arizona, Oregon, and Washington were black, whereas 20% of the women in Georgia and 21% of the women in Alabama were black. Black women and white women were distributed differently across income quintiles. Black women were concentrated in the quintile with the lowest median income; white women were distributed fairly evenly across quintiles. Forty-four percent of black women and 18% of white women resided in areas with the lowest median income. Moreover, 21% of white women and only 9% of black women resided in areas with the highest median income.

Table 2 shows, by race, the percentage of women who had made 0, 1, 2, or 3 or more primary care visits. Slightly more black women than white women did not visit a primary care physician (34% and 30%, respectively). Black women were also slightly less likely than white women to have made 1 or 2 primary care visits (8% and 10%, respectively); however, one half of both black women and white women had made 3 or more visits to a primary care physician.

Figure 1 shows, by number of primary care visits and race, the percentage of women in Alabama and Georgia who had mammography. In these two states (as in all others), mammography use was positively associated with the number of primary care visits for women of both races. Mammography use was lowest among women who had made no visits to a primary care physician, was greater among women who had made at least one visit, and was greatest for women who had made more than one visit.

At each visit level (0, 1, 2, and 3 or more visits), however, mammography was done less frequently in black women than in white women. Furthermore, a visit to a primary care physician was associated with a smaller increase in mammography use for black women than for white women. In Alabama, the percentages of women who had mammography at each visit level were 3%, 5%, 9%, and 13%, respectively, for blacks and 5%, 14%, 21%, and 24% for whites. In Georgia, the percentages of women who had mammography at each visit level were 4%, 7%, 12%, and 12%, respectively, for blacks and 7%, 15%, 21%, and 20%, respectively, for whites.

Figure 2 presents, by income quintile and race, the percentage of women in Alabama and Georgia who had mammography. In both states and in each income quintile, fewer mammographies were done in black women than in white women. In Alabama, rates of mammography use among black women and white women in each income quintile were, respectively; as follows: lowest, 9% and 16%; low, 8% and 16%; middle, 8% and 16%; high, 11% and 19%; and highest, 9% and 23%. In Georgia, rates of mammography use among black women and white women in each income quintile were, respectively, as follows: lowest, 10% and 16%; low, 8% and 14%; middle, 10% and 14%; high, 8% and 15%; and highest, 10% and 17%.

Figure 3 shows the information from Figure 2 about mammography use by income quintile and race but gives data for primary care users only. Although all of these women had made at least one primary care visit, the deficit in mammography use among black women persisted in each income quintile. In Alabama, the percentages of black women and white women who had had mammography in each income quintile were, respectively, as follows: lowest, 11% and 19%; low, 10% and 20%; middle, 11% and 20%; high, 14% and 23%; and highest, 12% and 28%. In Georgia, the percentages of black women and white women who had had mammography in each income quintile were, respectively, as follows: lowest, 13% and 20%; low, 10% and 18%; middle, 12% and 18%; high, 10% and 19%; and highest, 12% and 22%.

Figure 4 shows mammography use by race and state. Mammography use varied widely by state, from 7% in Oklahoma to 21% in Washington. In each state, fewer mammographies were done in black women than in white women. Separate rates for black women and white women according to state were also lowest in Oklahoma (4% and 7%, respectively) and highest in Washington (16% and 21%, respectively). State differences in mammography use were so great that black women in three high-use states (Connecticut, Pennsylvania, and Washington) had mammography more frequently than did white women in the four lowest-use states (Kansas, New Jersey, Oklahoma, and Oregon).

Figure 5 shows the information from Figure 4 about mammography use by race and state but gives data for primary care users only. Although all of these women had made at least one primary care visit, large differences by state in mammography use and the deficit in mammography use in black women persist in each state. Oklahoma has the lowest rate of mammography use among both black women and white women (5% and 9%, respectively), and Alabama has the highest rate (26% and 35%, respectively).

Table 3 shows the age-adjusted logistic models that separately predict mammography use for white women and black women in all 10 states. Primary care use has a powerful effect on use of mammography. Among white women, odds ratios for persons with 1, 2, and 3 or more visits to a primary care physician were, respectively, as follows: 2.73 (CI, 2.70 to 2.77); 3.98 (CI, 3.93 to 4.03); and 4.62 (CI, 4.58 to 4.67). Analogous odds ratios for black women also show a powerful, but systematically weaker, effect of primary care visits on mammography use (odds ratios of 1.77 [CI, 1.67 to 1.87], 2.49 [CI, 2.36 to 2.63], and 3.15 [CI, 3.04 to 3.25], respectively).

Mammography use varies greatly from state to state, and the state of residence affects women of both races similarly (state odds ratios are as high as 4.92 for blacks and 3.76 for whites). Income quintile, however, has little effect on mammography use for white women (odds ratio, 1.41 [CI, 1.39 to 1.42] in the highest income quintile) and no effect on mammography use for black women (odds ratio, 1.03 [CI, 0.98 to 1.09] for the highest income quintile).

Discussion

Our study included more than 3 million women 65 years of age and older residing in 10 geographically and demographically diverse states. We found a deficit in mammography use among older black women. For both black and white women, greater mammography use was associated with more visits to a primary care physician. The observed deficit for black women persisted at each income level and in each state, even after use of primary care was considered. The number of primary care visits did less to “boost” mammography use for black women than for white women. Thus, although the number of visits made to a primary care physician does influence mammography use, it is not sufficient to overcome socioeconomic and regional differences in mammography use. These results suggest that the nature of primary care may vary within and among states and between black women and white women.

We found that 15% of women 65 years of age and older had had mammography during 1990 (11% of blacks and 16% of whites). Several other studies [11, 18, 19] have used Medicare data to examine mammography use among women 65 years of age and older. Each of these studies has also shown a difference in mammography use between black and white women.

For both races, an increased number of visits to a primary care physician was associated with greater mammography use. Even after the number of primary care visits was considered, however, the deficit in mammography use among black women persisted at each income level and in each state.

We found that, in each income quintile, fewer mammographies were done in black women than in white women. Previous studies [11, 12] have also shown that socioeconomic status influences mammography use. In our study, receiving primary care did not correct the difference in mammography use between black women and white women. Black women who had made at least one visit to a primary care physician had fewer mammographies than white women in each income quintile.

We found large differences in mammography use among the 10 states studied. In each state, however, mammography use was lower for black women than for white women. Previous studies have also shown geographic differences in mammography use [12, 20]. In our study, the difference in mammography use among the 10 states was so great that black women in certain states had mammography more frequently than did white women in other states. Regional differences in mammography use remained, even among women who had received primary care. Large differences among states and the deficit in mammography use among black women persisted in each state among women who had made at least one visit to a primary care physician.

We used logistic modeling to further explore the relation between mammography use and number of primary care visits. We found that the effect of primary care visits in “boosting” mammography use is systematically lower for black women than for white women. This may be because we could only account for the number of primary care visits and not for the reasons for the visit, the nature of the visit, or the site of the visit.

Data from the 1985 National Ambulatory Medical Care Survey showed that the primary reason women 65 years of age and older visited a physician was for a general medical examination and that the most common principal diagnosis was essential hypertension [21]. The reasons black women visit their primary care physicians may, however, differ from that of white women. Black women may be more likely to present for an urgent visit than for a routine examination, and they may present with more, or possibly more complex, medical problems.

We do not know the nature of the primary care visits made by the women in this study; in particular, we do not know the details of the physician–patient encounter. Research has shown that physicians are more likely to encourage older white women to have mammography than older black women [22]. Alternatively, because black women have been shown to be less knowledgeable about mammography than white women [23], physicians may be unwilling or unable to spend the additional time necessary to sufficiently educate black women about the importance of the procedure. Research has shown that patient education is worthwhile, because the difference in mammography use among black women and white women is eliminated after adjustment for knowledge-related variables [24].

Finally, we do not know the site of the primary care visits made. Women who received care in hospital-based clinics may have been more likely to face barriers to obtaining mammography than women who received care in private, community-based offices. In addition, although all women in the study had Medicare, several recent studies [11, 25, 26] have suggested that Medicare coverage alone is not sufficient to eliminate cost as a barrier to mammography use. We could not determine whether the women had supplemental insurance, a factor that may partially explain the deficit in mammography use among black women.

The principal limitations of our study arise from use of a claims database. The three major issues are our ability to capture all mammographies done, our ability to distinguish diagnostic from screening mammographies, and our ability to evaluate factors that influence mammography use. We discuss each of these below.

We found that 15% of women of both races had had mammography in 1990. This percentage is similar to the percentage of women 65 years of age and older from the 1987 National Health Interview Survey who reported having had mammography in the past year, but it is less than that found in the 1990 survey (14% for black women and 28% for white women) [13]. It is also less than the 22% of women 65 years of age and older from the 1990 Mammography Attitudes and Utilization Survey who reported having yearly mammographies [27]. These rates suggest that our study results are representative of most, but not all, mammographies done in 1990.

Although the Health Care Financing Administration did not formally reimburse providers for screening mammography until 1991, our data suggest that screening mammographies were being done and billed under the diagnostic code. The criterion that allowed this was “a patient was (is) asymptomatic but, on the basis of the patient's history and other factors, the physician's judgment was (is) that a mammogram was (is) appropriate” [16].

Two other studies have used Medicare data to examine mammography use. Both were done after the Health Care Financing Administration began paying for biennial screening mammography in 1991. Both studies found that Medicare claims could not reliably distinguish screening mammography from diagnostic mammography and therefore included both screening and diagnostic mammography in their analyses [11, 19].

In addition, although Medicare claims data cannot distinguish diagnostic from screening mammography, diagnostic mammography represents only a small proportion of all mammographies done. Data from the National Health Interview Survey show that only 13% of all mammographies done in the past year among women 65 years of age and older were for so-called health problems and that black women and white women older than 65 years of age had similar numbers of diagnostic mammographies done (10% and 13%, respectively) (Breen N. Personal communication).

It could also be possible that the observed difference in mammography use between black women and white women could partially be due to the status of Medicare reimbursement during 1990. Physicians caring for black patients may have been less willing to designate screening mammography as diagnostic, and black patients may have been less willing to risk refusal of Medicare reimbursement for the service. Blustein [11] and a report from the Centers for Disease Control and Prevention [19], however, used Medicare data subsequent to 1991 when screening mammograms were reimbursed; both researchers found a difference in mammography use between black women and white women.

We examined potential explanatory factors for the racial difference seen in mammography use during 1990. Mammography use has been increasing over time, and greater increases have been seen for black women than for white women [12]. Results for 65 years of age and older are, however, less encouraging. The report from the Centers for Disease Control and Prevention Medicare used claims data from 1991 through 1993 and found minimal changes; 24% of women in 1991 and 25% of women in 1993 had mammography. Trontell found persistent differences between black women and white women; 18% of black women and 26% of white women had mammography in 1993 [19]. Given the few older women having mammography and the persistent difference in mammography use between the two races, examining factors that explain this difference remains important.

We were able to explore the association between various factors and mammography use. Potential problems are associated with some of these factors. Race was classified as black, white, or other on the basis of self-report to the Health Care Financing Administration. We excluded women who reported their race as “other” from analyses considering race. Although this is a coarse classification, it provides insight into the relation between race and mammography use.

We did not know the individual incomes of the study participants. Instead, women were assigned to an income quintile on the basis of the median income of the ZIP code of the area in which they reside. This method is imperfect but has been used previously [28-31]. Misclassification errors caused by this method are most likely to bias findings toward no association [32]. The presence of a consistent association between income and mammography use across states lends confidence to the patterns identified.

This administrative database provides useful insights into factors associated with mammography use in a large, diverse population of older women. We found lower levels of mammography use for black women than for white women in the same age group. For both, greater mammography use was associated with more visits to a primary care physician. However, deficits for black women persisted at each income level and in each state, even after use of primary care services was considered. In fact, primary care contacts did less to “boost” mammography use for black women than for white women. Thus, we have seen that primary care visits influence mammography use, that primary care visits are not sufficient to overcome socioeconomic and regional differences in mammography use, and that not all primary care visits in and among the states have the same effect on mammography use. Future studies of mammography use should examine the number of primary care visits made as well as the reasons for the visits and the nature of the services rendered.

Dr. Burns: Department of Medicine, Beth Israel Hospital, 330 Brookline Avenue, Boston, MA 03215.

Dr. Shwartz: Operations Management, School of Management, Boston University, 621 Commonwealth Avenue, Boston, MA 02215.