Outcome and comparative effectiveness research in traumatic brain injury : a methodological perspective

Traumatic brain injury (TBI) is one of the leading causes of death and disability worldwide. Although research activity in TBI has expanded rapidly, all these endeavors have not yet resulted in major advances in our understanding of TBI.
This thesis addresses two important topics in TBI research. First, outcome following TBI including the prevalence and predictors of psychiatric disorders and post-concussion symptoms. Second, this thesis examines whether comparative effectiveness research (CER) could contribute to evidence generation in TBI, with a focus on current guideline adherence, treatment variation and analytical methods.
We found that prevalence rates of post-concussion symptoms and psychiatric disorders varied widely and were dependent on pre-injury factors, patient population, assessment, analysis strategy and the diagnostic criteria used. Etiology of TBI sequelae is complex and multifactorial and includes biological, psychological and social factors that might be difficult to capture in a prediction model. When studying prevalence and predictors of psychiatric disorders and post-concussion symptoms following TBI, the role of misattribution, research setting, attrition and the overlap in symptomatology between post-concussion symptoms and psychiatric disorders should be taken into account.
Current evidence underpinning TBI guidelines is weak, and consequently, there is large variation in policy and guideline adherence among neurotrauma centers. This variation provides an opportunity for CER. However, the effect estimate obtained in observational CER studies largely depends on the analytical method used. Since all analytical methods have their strengths and limitations and are based on untestable assumptions, it is important that researchers do not consider one particular analytical method as the most appropriate. Different analytical methods that are reasonable for the research question under study should be used as sensitivity analyses and the results of observational CER studies should be confirmed by high-quality RCTs.