Comparative Effectiveness of Inpatient Rehabilitation Interventions for Traumatic Brain Injury: Introduction

Abstract

The Comparative Effectiveness of Inpatient Rehabilitation Interventions for Traumatic Brain Injury (TBI-CER) project used causal inference methods as an alternative to randomized controlled trials to evaluate rehabilitation practices. The TBI practice-based evidence dataset afforded the opportunity to compare the outcomes of different rehabilitation approaches while controlling for a large set of potential confounders using propensity score methods (PSMs). PSMs rely on 4 assumptions: positivity, exchangeability, consistency, and correct specification of the propensity score model. When these assumptions are met, PSMs provide a transparent means for evaluating potential causal relations between interventions and outcomes using observational data. In combination, the series of studies resulting from the TBI-CER project suggested that the content and approach used in treatment have a stronger effect on outcomes than the amount of time spent in treatment. Further, engagement of the patient and family in treatment is key to optimizing outcomes up to 9 months postdischarge from rehabilitation.