T. Somorin1, D. Barkan1, E. Faridmoayer1, M. R. Nakeshbandi2, S. E. Sharath1, P. Kougias1, D. H. Berger1 1SUNY Downstate Health Sciences University, Department Of Surgery, Brooklyn, NY, USA 2SUNY Downstate Health Sciences University, Department Of Infectious Diseases, Brooklyn, NY, USA
Introduction: The area deprivation index (ADI) is increasingly used as a measure of social risk adjustment in surgical outcomes and hospital quality rankings. It is possible that the ADI incompletely represents hospital-level patient characteristics. Enrollment in Medicaid is consistently associated with patient social disadvantage. The objective of this study was to test the hypothesis that the ADI predicts hospital patient characteristics, as represented by Medicaid enrollment.
Methods: 2021 inpatient discharges from the Statewide Planning and Research Cooperative System (SPARCS) data set were summarized into hospital-level observations by Medicaid proportions in New York state. By facility, proportion of patients with Medicaid primary payer status was calculated. Hospital-level Medicaid proportions were classified into quintiles to define five Medicaid burden levels, including: low, low-moderate, moderate, moderate-high, high Medicaid burden facilities. The area deprivation index associated with the 9-digit zip code of the hospital was the primary dependent variable. Linear regression was used to examine the primary hypothesis with hospital-level Medicaid proportions and ADI ranking as the independent and dependent variables, respectively. Cubic spline fitted with cross medians as knots was used to visually represent the primary predictive relationship.
Results: We identified 204 hospitals in New York state. Of these, 75 (37%) hospitals were classified as high Medicaid burden facilities serving up to 20,000 (per facility) Medicaid-covered patients in 2021 (moderate-high Medicaid burden mean = 4668, standard deviation [SD] = 3939; high Medicaid burden mean = 5460, SD = 4751). Increasing ADI scores represent increasing disadvantage. As hospital Medicaid burden increased, ADI scores decreased. At moderate-high and high Medicaid burden hospitals, median ADI scores were 24 (interquartile range [IQR]: 15-79) and 17 (IQR: 8-45), respectively. In the linear regression model, a statistically significant inverse association was observed – where decreasing disadvantage was associated with facilities with increasing Medicaid patient burden (ADI score = -4.61, p<0.001; Figure).
Conclusion: In New York state, the ADI does not correlate with hospital patient characteristics. This is likely due to the fact that in urban areas, socioeconomic diffusion resulting from high population density drives decision-making related to hospital choice, where patients of increased means seek care outside their locality. Measuring hospital quality and performance based on the ADI should be done with extreme caution, as it may penalize facilities located in high ADI areas with a mainly socioeconomically disadvantaged population.
