Comparison of the Traumatic Brain Injury (TBI) Model Systems national dataset to a population-based cohort of TBI hospitalizations

Objective: To determine whether severity alone accounts for differences observed between a population-based cohort of acute care hospitalizations for traumatic brain injury (TBI) and the Traumatic Brain Injury Model Systems (TBIMS) national dataset. Design: Prospective cohort. Setting: Acute care hospitals in South Carolina and TBIMS rehabilitation centers. Participants: Subjects enrolled in the TRIMS national dataset and the South Carolina TBI Follow-up Registry (SCTBIFR). Interventions: Not applicable. Main Outcome Measures: Comparable variables in the 2 datasets included demographic characteristics, etiology of injury, initial Glasgow Coma Scale score, Abbreviated Injury Scale score for the head region derived from International Classification of Diseases codes, presence of computed tomography (CT) abnormalities, acute hospital length of stay, and payer source. Results: As hypothesized, TBIMS participants showed greater initial injury severity, frequency of abnormal CT scans, and longer lengths of acute care hospitalization, explaining over 75% of cohort membership. Counter to a priori hypotheses, when all other factors were held constant, there were also differences in racial and ethnic background and insurance payer source. Conclusions: Differences between the TBIMS cohort and patients acutely hospitalized with TBI are primarily due to injury severity; however, an additional difference in payer source may need to be taken into account when generalizing findings. Results showed that TBIMS and SCTBIFR datasets are complementary, each having different strengths for understanding factors that impact long-term recovery after TBI. Recommendations are made for methodologic improvements in both data collection for the TBIMS and future outcome surveillance.