Related Experiment Video
Updated: Oct 14, 2025

Inverse Probability of Treatment Weighting Propensity Score using the Military Health System Data Repository and National Death Index
Published on: January 8, 2020
Social determinants of health and patient-level mortality prediction after trauma
Heather M Phelos1, Nicolas M Kass, Andrew-Paul Deeb
1From the Division of Trauma and General Surgery, Department of Surgery, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania.
Background:
Social determinants of health (SDOH) impact patient outcomes in trauma. Census data are often used to account for SDOH; however, there is no consensus on which variables are most important. Social vulnerability indices offer the advantage of combining multiple constructs into a single variable. Our objective was to determine if incorporation of SDOH in patient-level risk-adjusted outcome modeling improved predictive performance.
Methods:
We evaluated two social vulnerability indices at the zip code level: Distressed Community Index (DCI) and National Risk Index (NRI). Individual variable combinations from Agency for Healthcare Research and Quality's SDOH data set were used for comparison. Patients were obtained from the Pennsylvania Trauma Outcomes Study 2000 to 2020. These measures were added to a validated base mortality prediction model with comparison of area under the curve and Bayesian information criterion. We performed center benchmarking using risk-standardized mortality ratios to evaluate change in rank and outlier status based on SDOH. Geospatial analysis identified geographic variation and autocorrelation.
Results:
There were 449,541 patients included. The DCI and NRI were associated with an increase in mortality (adjusted odds ratio, 1.02; 95% confidence interval, 1.01-1.03 per 10% percentile rank increase; p < 0.01, respectively). The DCI, NRI, and seven Agency for Healthcare Research and Quality variables also improved base model fit but discrimination was similar. Two thirds of centers changed mortality ranking when accounting for SDOH compared with the base model alone. Outlier status changed in 7% of centers, most representing an improvement from worse-than-expected to nonoutlier or nonoutlier to better-than-expected. There was significant geographic variation and autocorrelation of the DCI and NRI (DCI; Moran's I 0.62, p = 0.01; NRI; Moran's I 0.34, p = 0.01).
Conclusion:
Social determinants of health are associated with an individual patient's risk of mortality after injury. Accounting for SDOH may be important in risk adjustment for trauma center benchmarking.
Level Of Evidence:
Prognostic/Epidemiologic, level IV.
Related Concept Videos
Assumptions of Survival Analysis
Actuarial Approach
Consider the example of a high-risk surgical procedure with significant early-stage mortality. A two-year clinical study is conducted,...
Cancer Survival Analysis
Introduction To Survival Analysis
The primary goal of survival analysis is to estimate survival time—the time...
Regression Toward the Mean
Comparing the Survival Analysis of Two or More Groups

