Setting the Stage for the Next Phase of Social Determinants of Health Research

Athough social determinants are increasingly recognized as important influences in health outcomes, obtaining accurate data about social determinants of health (SDH) is often a challenge. Publicly available data may be too coarse to be applicable to individual patients, and screening programs to collect individual-level determinants, although expanding, are not yet optimized. In addition, the debate about the usefulness of community-level versus self-reported SDH data remains open, with the former seen as more useful for policymakers or population health and the latter seen as more useful at the individual clinical encounter level. In this issue of AJPH, Udalova et al. (p. 923) demonstrate an approach to fortifying the utility of community-level data by linking Census Bureau data to clinical data at the individual level. The authors accomplished this through a process that began with a complicated data use agreement that required education, negotiation, and specification on both sides and ended with greater than 94% linkage between protected identification keys from the Census Bureau and the patients in the clinical data set. Although 90% ofthe patient matching was based on Social Security numbers, the algorithm performed nearly as well when Social Security number information was removed from the data set. The authors note that microlevel Census Bureau data may be useful for health systems in determining whether they are serving a representative population, correcting for bias in data sources, and evaluating estimates based on larger area statistics (e.g., Census blocks).