- Background: Equity in organ transplantation cannot be fully achieved without an understanding of how social determinants of health (SDOH) mediate patient outcomes. Unfortunately, SDOH variables are currently inconsistently collected across national transplant data registries and public health research databases. Health system data may be a superior source of SDOH data, and Epic has recently implemented a dedicated form for SDOH collection. The goal of this study was to characterize the extent to which these data sources captured SDOH data, using transplant patients as a model of a complex surgical population.
Methods: Social determinants of health data. Our data collection was guided by 28 SDOH variables listed in the NIMHD PhenX Toolkit. National data sources. We selected active national data sources based in the United States that were previously identified in a systematic review of the health equity literature in transplant. For each source, data dictionaries, data collection forms, registration worksheets, and other publicly available data element lists were examined for inclusion of SDOH variables. Health system data. We examined electronic health record data from adults referred to the Duke Transplant Center for liver or kidney transplant between January 1, 2017 and December 31, 2020. SDOH data collection forms in Epic were examined for inclusion of SDOH variables and patient medical records were reviewed for capture of discrete SDOH data to determine the percentage of missingness of these variables.
Results: National data sources. In total, we reviewed 10 national data sources. Of the 28 SDOH variables evaluated, eight basic demographic variables were included in ≥80% of the data sources and seven variables ranged between 30-60% inclusion. The remaining 13 variables addressing personal identity, healthcare access, and social need were included in ≤20% of the data sources with five variables included in none of the sources. NHANES and ACS were the most comprehensive data sources, including 78.6% and 71.4% of the reviewed SDOH variables, respectively (Figure 1). Health system data. A total of 18 variables (64.3%) were included within SDOH forms in Epic, including variables that were poorly represented in the national data source review (access to health services, gender identity, sexual orientation, food insecurity, spirituality, and wealth). Seven of 18 variables exhibited 100% missingness in the transplant cohort containing 7421 patients. The other 11 variables ranged in missingness within the cohort from 0 – 10.9% after data quality review.
Conclusions: Current national data sources include limited SDOH data beyond core demographic information. While health system data appears a potential solution to these data deficiencies, data collection is inconsistent, illustrating the need for a more robust data infrastructure to support efficient collection, organization, and access.