Rehospitalization Over 10 Years Among Survivors of TBI: A National Institute on Disability, Independent Living, and Rehabilitation Research Traumatic Brain Injury Model Systems Study.

Rehospitalization data were based on self-report from people with TBI or their proxy, so recall bias is possible. The database includes people treated at specialized TBI rehabilitation centers, so results may not apply to all TBI survivors. Reasons were coded into broad categories, so a hospitalization could reflect planned improvement or decline, and the study cannot say whether it was elective, planned, preventative, or corrective. The analyses looked at rehospitalization during a specific interview year, not cumulative rehospitalization over 10 years. The longitudinal sample was smaller and differed from the cross-sectional sample; it required at least three temporal measures and was not representative of people who died soon after injury or were recently added. The longitudinal models are descriptive, not inferential, and the authors caution that they should not be used to predict prognosis for an individual patient. No treatment was assigned, so the study cannot show that any discharge plan or follow-up service reduces rehospitalization.

The study was funded by NIDILRR within the Agency on Community Living in Health and Human Services. The authors declared no conflicts of interest and no party with a direct interest in the results will confer a benefit.

Do not read the rehospitalization rates or risk trajectories as proof that a particular discharge plan or follow-up service prevents rehospitalization. The study was descriptive and did not test a service or discharge plan. The authors also caution that the models should not be used to predict prognosis for an individual patient.