Mapping Quality in Health Care Systems – A Novel Tool for Population Management

Background: As primary care networks, including community health centers, develop population-based systems of care, geographic information systems (GIS) may be a useful tool for identifying communities or neighborhoods with poor quality outcomes. We used GIS to examine breast cancer (BRC) screening rates among patients within a large academic primary care network by the communities in which they reside. Methods: We obtained GIS coordinates from the addresses of 142,690 primary care patients within our network cared for by 174 PCPs working in 13 primary care practices, including 4 community health centers. Using data from an electronic record repository, we identified all patients aged 42-69 years old eligible for BRC screening and calculated the proportion overdue for screening by town or city neighborhood and by census block group to identify geographic areas with low BRC screening rates. We then categorized census block groups by higher and lower BRC screening rates and compared differences in patient socio-demographic characteristics and median household income. These analyses were limited to census block groups with greater than 30 qualifying patients. Results: Overall, 9033 (21.3%) of eligible patients within our network were overdue for BRC screening; overdue BRC screening rates by census block group ranged from 0.0% to 53.3%. Maps generated from these data showed considerable geographic variation in the rate of overdue BRC screening among all network patients in Boston neighborhoods and surrounding towns. When we limited the analysis to patients receiving care at one network-affiliated community health center, maps identified several specific census block groups within the community with particularly high rates of overdue BRC screening. Across the Massachusetts area served by our network, census block groups with lower BRC screening rates had higher proportions of minorities (30.5% vs 17.9%, p<0.001), non-English speaking patients (17.3% vs 9.0%, p<0.001), and those not completing high school (13.4% vs 6.7%, p<0.001), and a lower median household income ($54594 vs $75739, p<0.001). Conclusion: Mapping quality indicators using GIS coordinates derived from administrative data may be a useful tool for geographically targeting resources and for tailoring interventions to the needs of specific communities. This detailed geographic approach may enable health systems to identify communities that are at high risk for lower quality outcomes and thereby reduce disparities in health and healthcare.