Abstract
BACKGROUND AND OBJECTIVES: Nearly 30% of older adults presenting with isolated spine fractures will die within 1 year. Attempts to ameliorate this alarming statistic are hindered by our inability to identify relevant risk factors. The primary objective of this study was to develop a prediction model that identifies feasible targets to limit 1-year mortality. METHODS: This retrospective cohort study included 703 older adults (65 years or older) admitted to a level I trauma center with isolated spine fractures, without neural deficit, from January 2013 to January 2018. Multivariable analysis was used to select for independently significant patient demographics, frailty variables, injury metrics, and management decisions to incorporate into distinct logistic regression models predicting 1-year mortality. Variables were considered significant, if P < .05. RESULTS: Of the 703 older adults, 199 (28.3%) died after hospital discharge, but within 1 year of index trauma. Risk Analysis Index (RAI; odds ratio [OR]: 1.116; 95% CI: 1.087-1.149; P < .001) and ambulation requiring a cane (OR: 2.601; 95% CI: 1.151-5.799; P = .02) or walker (OR: 4.942; 95% CI: 2.698-9.196; P < .001), ie, frailty variables, were associated with increased odds of 1-year mortality. Spine trauma scales were not associated with 1-year mortality. Longer hospital stays (OR: 1.112; 95% CI: 1.034-1.196; P = .004) and nursing home discharge (OR: 3.881; 95% CI: 2.070-7.378; P < .001) were associated with increased odds, while discharge to rehab (OR: 0.361; 95% CI: 0.155-0.799; P = .014) decreased 1-year mortality odds. A "preinjury" regression model incorporating Risk Analysis Index and ambulation status resulted in an area under receiver operating characteristic curve (AUROCC) of 0.914 (95% CI: 0.863-0.965). A "postinjury" model incorporating Glasgow Coma Scale, hospital stay duration, and discharge disposition resulted in AUROCC of 0.746 (95% CI: 0.642-0.849). Combining elements of the preinjury and postinjury models into an "integrated model" produced an AUROCC of 0.908 (95% CI: 0.852-0.965). CONCLUSION: Preinjury frailty measures are most strongly associated with 1-year mortality outcomes in older adults with isolated spine fractures. Incorporating injury metrics or management decisions did not enhance predictive accuracy. Further work is needed to understand how targeting frailty may reduce mortality.