Predicting 1-year disability and mortality of injured older adults

The growing incidence of elderly patients injured from falls, combined with a growing understanding of the contribution of cognition and frailty to mortality, prompted this work. Our objective was to develop a clinical risk prediction model for prognosticating disability and mortality among injured older adults 1 year after hospitalization.

Secondary analysis of prospective longitudinal data from an urban Level 1 trauma center. A proportional odds regression model was used to model mortality and functional status as ordinal outcomes. Death was treated as the lowest functional status, and 3 ordered groups of the Barthel Index were treated as higher functional status. 188 patients aged 65 and older who were admitted through the emergency department from 2013 to 2014 with a primary injury diagnosis comprised the prospective cohort. Follow-up assessments were performed at 30-days, 90-days, 6-months, and 1-year. Predictors in the model included: baseline physical function, baseline cognition, two physical frailty measures, age, injury severity, a comorbidity index, gender, living location, mechanism of injury, and hospital admitting service.

The full model yielded an R² of 0.45, and Life Space Assessment, Vulnerable Elders Survey, and Injury Severity were the most influential predictors. Approximated models (to encourage clinical use) yielded an R² of 0.86. Calibration assessment (i.e., accuracy) demonstrated a mean squared error <0.003 at all 3 intercepts.

A moderate statistical signal was discovered that contributed to a highly accurate clinical prediction model. Approximated models and nomograms could be used by clinicians, patients, and families in shared decision making during hospitalization.