In the early 1990s, analysts predicted an excess supply of mammography machines in the US.(26
) However, the number of mammography facilities decreased in the decade following implementation of the MQSA.(12
) Our analyses of Medicare beneficiaries and respondents to a large national survey support an inverse relationship between mammography availability and utilization. Despite recent reports that mammography resources are adequate overall,(14
) our results suggest that maldistribution of these resources may impair access to breast cancer screening in some areas.
While our results address the relationship between the availability of mammography machines and their utilization, human resources – mammographic technologists, medical physicists and interpreting radiologists – are important components of mammography capacity. A shortage of breast imagers has been reported in both community-based and academic practices.(18
) Several factors may be contributing to the reported shortage of qualified breast imagers, including fear of malpractice litigation, declining reimbursement, burdensome volume and education requirements stipulated by the MQSA, and greater interest in more sophisticated imaging modalities.(18
) A shortage of mammographic technologists may also be limiting access to breast cancer screening.(30
Our metric of mammography capacity quantifies the availability of machines, but not their spatial accessibility, both of which may independently influence utilization. For example, women in areas with limited capacity may face longer waiting times for an appointment, and women who live far from the nearest mammography facility may face a substantial travel burden. Resource capacity and proximity have been associated with health care utilization in a variety of other settings, including primary care,(31
) prenatal care (32
) and cancer treatment.(33
) Ongoing research regarding travel burden and appointment wait times should further elucidate the role of these factors in access to screening mammography.
Our estimate of the impact of inadequate capacity on mammography use was similar in the two cohorts evaluated, with AORs of 0.89 in the BRFSS cohort and 0.85 in the Medicare cohort. This similarity is remarkable given the differences between the two data sources and the populations they represent. BRFSS data are self-reported and may therefore be influenced by recall problems and social desirability bias. Recent evidence suggests that self-report surveys are likely to overestimate mammography utilization.(35
) Medicare claims, while not subject to these biases, only capture services reimbursed by Medicare. Free or subsidized mammograms will not be identified in claims if providers did not bill Medicare. Claims may therefore underestimate mammography utilization. These differences between the two data sources likely explain the greater estimate of recent mammography use among BRFSS respondents age 70–79, compared with the claims-based estimate among women of the same age in the Medicare analysis. In both cohorts, mammography use was inversely associated with age. This finding is not a surprise, as screening mammography benefits and guidelines are less clear for women age 70 and older than for women 40–69 years of age.(4
While mammography capacity had a similar impact on utilization in the two cohorts, results related to race diverged. In the Medicare cohort black, Hispanic and Asian women were all less likely than white women to receive a mammogram, while black and Hispanic BRSS respondents were more likely than their white peers to report a recent mammogram. Although this may be attributable to true differences between the cohorts, it may also be a function of race-specific variation in the accuracy of each data source for identifying breast cancer screening. Self-report of mammography use may be less sensitive and less specific among black women than white women.(36
Numerous patient, physician and health insurance characteristics have been shown to facilitate or impede timely screening mammography.(21
) In both cohorts we examined, the strongest predictors of a recent mammogram were factors related to interaction with the health care system, specifically having a regular source of care and having a check-up or a primary care visit. However, we also found that mammography capacity in a woman’s county of residence was significantly associated with mammogram use, controlling for these factors and for other important characteristics. We were not able to adjust for psycho-social factors or knowledge, attitudes and beliefs about breast cancer screening.
Additional limitations must be noted. We assumed that with respect to non-spatial factors, mammography facilities were equally accessible by all patients. In fact, some facilities may serve only patients participating in specific health insurance plans or physician practices. We defined mammography capacity at the county level, although women may seek health care services outside their county of residence. Also, if mammography capacity is associated with other county-level or even state-level attributes, our analysis would not be able to isolate the role of those factors. Finally, because we excluded women lacking a county identifier in the BRFSS our findings in that cohort likely underrepresent women in the most sparsely populated US counties, where access to health care resources in general is of particular concern.
Among all US counties, the estimated median female population aged 40 years or older was 6,334 in 2007, and only 36% of counties had at least 10,000 women in this age group. Of the 1,164 counties with mammography capacity <1.2 machines per 10,000 in 2007, the median county population of women in this age group was 3,044, the maximum was 574,665, and only 246 counties (21% of those with inadequate capacity) had at least 10,000 adult women potentially eligible for annual screening mammography. It may be unrealistic to expect every low-capacity county to enhance resource capacity alone if it does not have a population large enough to support the service. Rather, more successful strategies to increase mammography capacity might be regional efforts that serve a larger, multi-county population, perhaps with mobile mammography facilities. Areas where multiple adjacent counties have inadequate capacity would be priority targets for such initiatives.
Prior efforts to increase screening mammography have been multi-faceted and have typically focused on patient and physician behavior. Evaluations of interventions directed at patients (such as behavioral and educational programs) and interventions directed at primary care physicians (such as automated mammography reminder notifications) have found varying degrees of effectiveness.(48
) Our findings suggest a third strategy for improving breast cancer screening rates, namely interventions directed at the local health care infrastructure. Further research on the influence of both technical and human resources will inform the development of specific initiatives, such as subsidies for the purchase of mammography equipment at community health centers or public hospitals in low-capacity areas, increased mammography reimbursement, and incentives for radiologists and technologists to practice in underserved areas. As we have shown here, geographic analysis can identify the communities in greatest need. Targeted enhancement of mammography capacity may boost breast cancer screening rates while optimizing limited health care resources.