Background: Social determinants of health (SDOH) play a critical role in predicting hospital readmission rates across various patient populations. Key SDOH factors, such as neighborhood crime index, income, and household size, are important predictors of unplanned readmissionsi. Importantly, including SDOH in readmission prediction models has been shown to improve accuracyii, highlighting the need for addressing these factors in both risk assessments and when developing intervention strategies to reduce readmissions and improve patient outcomes. While the effect of SDOH on individual patients is well-studied, how SDOH in a community affects hospital readmissions remains unexplored. Therefore, we examined the relationship between neighborhood SDOH on hospital 30-day readmission rates for three major conditions using national data.
Methods: We obtained data from four sources: 1) hospital characteristics and risk standardized 30-day readmission rate data on pneumonia, heart failure (HF), and acute myocardial infarction (AMI) from CMS from 2005-19, 2) regional demographic and economic data from the US Census Bureau, and 3) Zip code mapping to hospital catchment area data from the Dartmouth Atlas of Healthcareiii, 4) Social Deprivation Index (SDI) National percentile rank, calculated from 2011-2015 American Community Survey files at Zip Code leveliv. The four datasets were merged so that each hospital had an SDI percentile rank based on the average ranks of Zip Codes in its catchment area. Data were summarized using mean (SD) or percentages, and differences were analyzed with the Chi-square test or Wilcoxon rank-sum test. Univariable and multivariable linear models were used to examine the relationship between SDI and 30-day hospital readmissions while accounting for correlation due to multiple observations from the same hospital. Adjusted models included confounding variables such as hospital ownership, teaching status, number of beds and nurses, rural/urban location, and catchment area demographics (total population, median income, poverty percentage, African American percentage, and insurance percentage).
Results: In the 4,657 unique hospitals, mean (SD) SDI percentile rank of the hospital catchment area was 43.5 (17.5) and readmissions for HF, AMI, and pneumonia during the study period were 24.4 (2.01), 19.9 (1.34), and 18.1 (1.65), respectively. As compared to hospitals in the lowest SDI quartile, hospitals in the highest quartile were in areas with lower insured population (79% vs. 90%), lower median income ($43,000 vs. $60,000), higher poverty rate (22% vs. 9.4%), and more African Americans (19% vs. 2.9%); all P< 0.001 (TABLE). As compared to the lowest quartile, hospitals in the highest quartile area had higher readmissions for HF (1.12%, 95%CI=0.97, 1.27), AMI (0.70%, 95%CI=0.51, 0.90), and pneumonia (0.65%, 95%CI=0.53, 0.78) in unadjusted models. Results were unchanged after adjusting for potential confounders (FIGURE); lowest vs. highest quartile for HF = 0.88% (95%CI=0.74, 1.03), AMI = 0.40% (95%CI=0.27, 0.52), and pneumonia = 0.55% (95%CI = 0.42, 0.67).
Conclusions: We found that SDOH, as measured by SDI, are associated with higher 30-day readmission rates for in this nationwide study. Studies examining interventions aimed to ameliorate social disparities across populations are needed.
