Limitations of mild traumatic brain injury meta-analyses

Abstract

PRIMARY OBJECTIVE: Mild traumatic brain injury (mTBI) meta-analyses are sometimes cited as evidence that by 3 months post-injury the clinical effects are insignificant. In contrast to these conclusions are findings that long-term mTBI outcome can vary depending on (1) the mechanism of the injury, (2) which diagnostic criteria are employed, (3) which assessment tools utilized and (4) whether symptomatic groups are considered separately. RESEARCH DESIGN: The present study was designed to clarify opposing conclusions in the mTBI literature by re-analysing meta-analytic datasets. METHODS AND PROCEDURES: The most frequently cited mTBI meta-analysis is by Binder, Rohling and Larrabee in 1997; this study was updated in 2005 by Frencham, Fox and Maybery. Data combined in these studies were re-categorized according to the four variables above; associated effect sizes and heterogeneity statistics were calculated. MAIN OUTCOMES AND RESULTS: Considerable qualitative and statistical heterogeneity was identified within the meta-analytic data under consideration. Clinically significant lasting effects were identified for a sub-set of neuropsychological measures. CONCLUSIONS: Methodological and statistical heterogeneity in studies combined in mTBI meta-analyses significantly limits the conclusions that can be drawn from small or non-significant overall effect sizes; clinically relevant information can be obscured by meta-analytic procedures. Recommendations for future meta-analytic studies of mTBI are offered.