A Comparison of Scoring Systems for Predicting Short- and Long-term Survival After Trauma in Older Adults.

Abstract

Early identification of geriatric patients at high risk for mortality is important to guide clinical care, medical decision making, palliative discussions, quality assurance, and research. We sought to identify injured older adults at highest risk for 30-day mortality using an empirically derived scoring system from available data and to compare it with current prognostic scoring systems. This was a retrospective cohort study of injured adults ≥ 65 years transported by 44 emergency medical services (EMS) agencies to 49 emergency departments in Oregon and Washington from January 1, 2011, through December 31, 2011, with follow-up through December 31, 2012. We matched data from EMS to Medicare, inpatient, trauma registries, and vital statistics. Using a primary outcome of 30-day mortality, we empirically derived a new risk score using binary recursive partitioning and compared it to the Charlson Comorbidity Index (CCI), modified frailty index, geriatric trauma outcome score (GTOS), GTOS II, and Injury Severity Score (ISS). There were 4,849 patients, of whom 234 (4.8%) died within 30 days and 1,040 (21.5%) died within 1 year. The derived score, the geriatric trauma risk indicator (GTRI; emergent airway or CCI ≥ 2), had 87.2% sensitivity (95% confidence interval [CI]= 83.0% to 91.5%) and 30.6% specificity (95% CI= 29.3% to 31.9%) for 30-day mortality (area under the receiving operating characteristic curve [AUROC]= 0.589, 95% CI= 0.566 to 0.611). AUROC values for other scoring systems ranged from 0.592 to 0.678. When the sensitivity for each existing score was held at 90%, specificity values ranged from 7.5% (ISS) to 30.6% (GTRI). Older, injured adults transported by EMS to a large variety of trauma and nontrauma hospitals were more likely to die within 30 days if they required emergent airway management or had a higher comorbidity burden. When compared to other risk measures and holding sensitivity constant near 90%, the GTRI had higher specificity, despite a lower AUROC. Using GTOS II or the GTRI may better identify high-risk older adults than traditional scores, such as ISS, but identification of an ideal prognostic tool remains elusive.