Date of Award


Document Type

Open Access Thesis

Degree Name

Medical Doctor (MD)



First Advisor

Angie Paik


While the U.S. has the highest incarceration rate in the world, at nearly 1% of the adult population, or over 2 million people, insight regarding health disparities in this population remains limited, particularly for traumatic injuries and surgical care. Our aims with this paper were two-fold. First, we sought to perform the largest national database analysis of criminal-justice system involved (CJI) trauma patients (specifically incarcerated patients). Second, we planned to investigate if incarceration status was associated with differences in a variety of outcomes after trauma for U.S. adults.To accomplish these aims we analyzed data from the National Trauma Data Bank from 2017-18. Using multi-level logistic regression we measured risk-adjusted associations between patient incarceration status (assessed by ICD-10 location codes) and trauma outcomes, namely mortality, any in-hospital complications, aggregate major complications, and failure to rescue – death after complication. We report odds ratios and 95% CIs, adjusting for demographics, transfer status, insurance type, comorbidities, injury mechanism, injury severity, and presenting vitals (systolic blood pressure, pulse, and total GCS). We investigated the potential mediating role that procedural intervention (assessed by ICD-10 procedure codes) plays via logistic regression, both of the association between CJI status and receiving procedural intervention, and of the association between CJI status and mortality and related outcomes, stratified by procedural intervention. A truncated negative binomial model was used to calculate risk-adjusted incident rate ratios of CJI status on length of stay, and hurdle models were used to calculate risk-adjusted incident rate ratios of CJI status on both total ICU days and total ventilator days. In our sample there were 12,888 CJI patients and 1,654,254 non-CJI patients. CJI patients made up about 0.8% of the sample and were, on average, younger (38.7 vs 53.9 years), more likely to be male (94.9% vs 60.55%), Black (27.9% vs 13.9%), and Hispanic (15.7% vs 11.5%), and presented more frequently with minor injuries (ISS


This is an Open Access Thesis.

Open Access

This Article is Open Access