Sample Characteristics by Race and BMI
Of the 5,738 women between ages 50 and 75 years, 5,277 (92%) had complete data on BMI, race, and mammography use. Of these women, 72% reported mammography use in the preceding 2 years. The rate was 74% for white women and 70% for black women (P = .08). The prevalence of obesity (BMI ≥ 30 kg/m2) in this age group was 23% among white women and 38% among black women (P < .001). We noted significant differences in marital status, education, and the insurance status of black and white women (). Furthermore, compared to white women, black women were less likely to see a health provider in the preceding year but were more likely to be hospitalized and to have certain medical conditions and physical limitations ().
Characteristics of Women in Study Sample Overall and by Race
In both black and white women, we found similar associations between obesity and many clinical and sociodemographic factors. In black and white women, obesity was significantly associated with higher age, and lower education and income; obesity was also associated with poorer health status, higher numbers of medical conditions, and greater mobility difficulties. However, there were some differences by race. The relationship between BMI and source of health care varied slightly by race. Compared to normal weight white women, obese white women were more likely to have a usual source of care (97% vs 95%) and to have a specialist as their usual provider (4% vs 3%). Conversely, obese black women were less likely than normal weight black women to have a usual source of care (92% vs 96%) or to have a specialist as their usual provider (1% vs 4%). Furthermore, while obese white women reported a significantly higher hospitalization rate than normal weight white women (15% vs 10%), obese black women reported a significantly lower hospitalization rate compared to their normal weight counterpart (16% vs 19%). Both black and white women with obesity were significantly more likely to feel sad or hopeless in the last 30 days compared to normal weight women. However, being obese was significantly associated with feeling worthless only among white women; while 14% of white women with obesity reported feeling worthless, 8% of normal weight women had similar feelings.
BMI and Mammography Use: Unadjusted Results
shows the unadjusted association between BMI and mammography use among all women in our sample and among white and black women separately. Among white women, rates were highest among nonobese women. Among black women, rates were highest among women who were moderately obese.
Mammography Use (%) Among Women Age 50 to 75 years by Weight Category Before Adjustment
BMI and Mammography Use Among Eligible Women Overall
When we examined for a relationship between BMI and mammography use after adjusting serially for sociodemographic factors and access, illness burden, psychological functioning, and health habits, women with moderate to extreme obesity overall were least likely to report breast cancer screening, although the trend was not statistically significant (P = .08). Results did not differ substantially among our adjusted models. shows results from our primary analysis, which adjusted for sociodemographic factors, health care access, general health status, and illness burden (model 2).
Relative Risk for Mammogram Use According to Weight Category for Women Age 50 to 75 Years After Adjusting for Demographic Factors, Health Care Access, Comorbid Conditions, and Mobility Status
BMI and Mammography Use Among Eligible White Women
also shows the results for mammography use and BMI among white women from our primary model (model 2). The likelihood of undergoing mammography decreased with increasing BMI category among white women (P value for trend = .02). Compared to results from model 1, which was adjusted for sociodemographic factors, health care access and general health status, further adjustment for illness burden (model 2), psychological functioning (model 3), and health habits (model 4) did not change the relationship between mammography use and BMI substantially ().
FIGURE 1 Influence of different factors on the relationship between BMI and mammography use among white women. ** Normal weight women served as the reference group with a relative risk of 1.00. † Model 1 was adjusted for sociodemographic factors, health (more ...)
BMI and Mammography Use Among Eligible Black Women
Among black women, mammography use increased with higher BMI except for women with extreme obesity, who had similar rates to normal weight black women (; ); however, the trend was not statistically significant (P = .30). Adjusting for different groups of factors did not change this relationship appreciably ().
FIGURE 2 Influence of different factors on the relationship between BMI and mammography use among black women.** Normal weight black women served as the reference group with a relative risk of 1.00. † Model 1 was adjusted for sociodemographic factors, (more ...)
BMI and Mammography Use: Comparing Black and White Women
In order to directly compare the relationship between mammography and BMI in black and white women, we excluded underweight women from our primary analysis (model 3) and added an interaction term for race and BMI. We found the interaction to be statistically significant at P = .001. We also repeated our adjusted analyses combining both black and white women and using normal weight white women as the reference group (). Compared to the reference group, black women regardless of weight were generally as or more likely to report screening with mammograms, with moderately obese black women being 20% more likely to report screening after adjustment (). Prior to adjustment, normal weight black women were significantly less likely to report mammography use (RR, 0.82; 95% CI, 0.70 to 0.94) than normal weight white women; however, this was substantially attenuated by adjustment for sociodemographic factors, health care access, and general health status (RR, 0.92; 95% CI, 0.79 to 1.04; model 2) and unchanged with further adjustment (models 3 and 4).
FIGURE 3 Mammogram use among black and white women by weight category compared to normal weight white women.* Normal weight white women served as the reference group. Analyses adjusted for sociodemographic factors, health care access, general health status, medical (more ...)
Our study suggests that among white women aged 50 to 75 years, higher BMI was modestly associated with lower breast cancer screening in a dose–response manner. White women with moderate obesity were least likely to undergo mammography. Among black women, however, higher BMI was associated with similar or greater use of mammograms. Differences in mammography use by race and BMI could not be explained by sociodemographic factors, health habits, or indicators for health care access, illness burden, or psychological functioning. We also noted that normal weight black women had significantly lower unadjusted rates of mammography use compared to their white counterparts; however, these differences were attenuated after accounting for differences in sociodemographic factors and health care access.
This study is a follow-up to an earlier study documenting lower mammography use among women with obesity even after adjustment for sociodemographic factors, health care access, and general health status.4
In that study, we used 1994 data to explore the influence of race on this association and found a suggestive dose–response relationship between lower mammography use and higher BMI among white women. Our results for black women, however, were inconsistent, possibly because of inadequate sample size. Furthermore, our earlier study lacked details on specific medical conditions, functional status, or psychological functioning. The current study uses a larger sample to address some of the limitations of our earlier work and allows us to examine the influence of race, illness burden, and other factors that potentially mediate or confound the relationship between weight and breast cancer screening.
We note several differences and similarities between our earlier and current findings. First, mammography use has improved substantially overall, from a rate of 65% in 1994 to 72% in 1998. Second, while the use of mammograms has increased for most groups we studied, the proportion among normal weight black women has actually decreased from approximately 74% to 62% between 1994 and 1998. Some of this difference was explained by differences in sociodemographic factors and health care access; however, future studies should investigate specific factors responsible for this dramatic reduction. Third, while the relationship between breast cancer screening and BMI among black women was inconsistent in our earlier study, our current findings suggest that black women with higher BMI are as or more likely to undergo screening except for those with extreme obesity. Nevertheless, the disparities we previously noted between BMI and screening among white women appears to persist.
Although the lower proportion of screening among white women with higher BMI is relatively modest, our findings are concerning nonetheless because of the rising and high prevalence of obesity and because breast cancer is the most common cancer among women in the United States.19
Approximately 40,000 women die from breast cancer annually, accounting for 15% of cancer deaths among women.19
Moreover, white women with a BMI above 35 kg/m2
represent 16% of all women aged 50 to 75 years (data available from authors) and are at higher risk for developing and dying from breast cancer.3
The mechanisms for this increased risk are unclear but may be related to physical inactivity, and hormonal and dietary factors.20–22
Fortunately, breast cancer deaths can be reduced substantially through early detection using mammograms.4
Moreover, while there are concerns about the quality of mammogram images in women with high adiposity,23
at least one study suggests that screening mammography may be equally beneficial in detecting early breast cancer for obese women and for normal weight women.24
Our study suggests, however, that this population of obese women, who are at high risk for breast cancer, is paradoxically less likely to undergo screening. Furthermore, screening did not appear to be driven by clinical factors that may influence life expectancy such as comorbidities and other proxies for illness burden in our study.
Unfortunately, we were unable to identify specific reasons for weight-related disparities in breast cancer screening. Despite the higher prevalence of known barriers to health care among women with obesity such as lower socioeconomic status and higher illness burden, adjusting for these factors did not influence our findings appreciably. Particularly striking was the minimal role that comorbid conditions, functional status, and other markers of illness burden seemed to play.
Our disparate findings between black and white women does suggest that patient-centered factors such as patient preferences and cultural beliefs may play a larger role in explaining differences in screening by BMI than either clinical factors or provider-driven factors. The self-perception associated with weight is one factor shown to differ between black and white women.10–12
Relative to their white counterparts, overweight black women are more likely to underestimate their weight category, to be satisfied with their bodies, and to feel attractive.10–12
In our study, obesity was associated with feeling worthless in the preceding 30 days for white but not black women. This qualitative difference in body image perception may interact with health behaviors, such that white women who are obese may be less willing to undergo mammograms. Adjusting for our proxy for self-esteem did not affect our findings related to BMI, screening, and race. However, our proxy likely did not completely capture the concept of low self-esteem or poor self-perception; therefore, future studies should use better measures to test the role of self-image in explaining disparities by BMI. Adjusting for health habits as a proxy for health-seeking behavior also did not explain differences by race and BMI. Thus, while our study does not point to specific mechanisms, it does suggest that disparities by weight and race are highly complex and warrant further study.
Although our findings make provider-driven factors such as bias against referring obese women for screening less likely, these findings do not exclude this possibility completely. One study found that when physicians were presented with similar patients who differed only by weight, physicians reported that they would prescribe more tests for the heavier patients, but spend less time with them.25
Another study found that obese women patients were more likely to refuse pelvic exams than thinner women and that physicians were less likely to pursue these exams if their patients refused.9
Hence, disparities in breast cancer screening may result from an interplay between patient preferences and provider biases. Obese women may have more reservations about undergoing mammograms than their normal weight or black women, and providers may not be spending enough time counseling about the importance of screening in a way that ameliorates patients’ fears. Another possibility is that black women may select providers who have fewer racial biases, and these providers may also harbor fewer biases against patients who are overweight. The validity of these hypotheses should be explored in future research.
We used a rich database that generalizes nationally; however, our study has several limitations. First, all information was based on self-report and may be subject to reporting inaccuracies and recall bias. In particular, overweight women tend to overestimate their height and underestimate their weight to a larger extent than thinner women.26
Hence, disparities noted by BMI are potentially underestimated for obese white women and overestimated for normal weight black women. Second, despite the added detail in the 1998 NHIS, we likely did not control adequately for all confounders or explanatory factors. Specifically, our measures for self-esteem and depression were likely inadequate. Third, we were unable to control for many cancer risk factors, which may have influenced patients’ preferences or providers’ recommendations. However, we chose minimum screening recommendations so that these risk factors would be less likely to influence our outcomes. Finally, we were not able to control for provider characteristics, such as gender or race, or evaluate provider behavior, which are likely to influence screening.27
Nevertheless, our study demonstrates that white women with moderate obesity are less likely to undergo breast cancer screening with mammograms than normal weight white women. On the other hand, overweight and obese black women are as or more likely than both black and white normal weight women to report mammography use. Differences in screening by weight were not readily explained by differences in sociodemographic factors, health care access, indicators of illness burden, functional status, or health habits. More research is needed to determine the reasons for these disparities so that appropriate efforts can be made to improve screening in patients with obesity.