Trauma Risk Score Matching for Observational Studies in Orthopedic Trauma

Abstract

ObjectivesTo determine if matching by trauma risk score is non-inferior to matching by chronic comorbidities and/or a combination of demographic and patient characteristics in observational studies of acute trauma in a hip fracture model. DesignRetrospective cohort study SettingLevel-1 Trauma Center Patients1,590 hip fracture [AO/OTA 31A and 31B] patients age 55 and over treated between October 2014 and February 2020 at 4 hospitals within a single academic medical center. InterventionRepeatedly matching randomized subsets of patients by (1) Score for Trauma Triage in Geriatric and Middle-Aged (STTGMA), (2) Charlson Comorbidity Index (CCI), or (3) a combination of sex, age, CCI and body mass index (BMI). Main Outcome Measurements“Matching failures” where rate of significant differences in variables of matched cohorts exceeds the 5% expected by chance. ResultsSTTGMA and combination matching resulted in no “matching failures”. Matching by CCI alone resulted in “matching failures” of BMI, ASA class, STTGMA, major complications, sepsis, pneumonia, acute respiratory failure, and 90-day readmission. ConclusionsSTTGMA matching in observational cohort studies is less likely to yield significant differences of demographics and outcomes than CCI matching. STTGMA matching is noninferior to matching a combination of demographic variables optimized for each treatment cohort. STTGMA matching is apt to reflect equipoise of health at admission and outcome likelihood in observational cohort studies of orthopedic trauma, while maintaining consistent weighting of demographic and injury characteristic variables that may expand the generalizability of these studies. Level of EvidenceLevel III