Posttraumatic Amnesia: A Systematic Review and Meta-Analysis. Proposal for a New Severity Classification

Abstract

Posttraumatic amnesia (PTA) duration is used to predict outcome after traumatic brain injury (TBI): however, no meta-analysis exists. A systematic review was performed following PRISMA reporting guidelines. The databases Scopus-1966, PubMed/MEDLINE, CINAHL, PsycINFO, and Embase were searched for relevant texts. Random effects meta-analysis derived pooled estimates of the odds ratio of outcomes of interest and sensitivity and specificity of PTA at different cutoffs, and subsequently a summary receiver operating curve was derived. PTA prediction of Glasgow Outcome Scale (GOS) and Glasgow Outcome Scale-Extended (GOSE) scores was assessed both qualitatively and quantitatively by pooled odds ratio regarding both a good outcome (GO: GOS-5 or GOSE-7/GOSE-8) and a severe disability outcome (SDO: GOS-3 or a GOSE-3/GOSE-4). Summary receiver operating curve analysis was performed in the prediction of composite of a moderate disability outcome (MDO: GOS-4 or GOSE-5/GOSE-6) and SDO. Twenty-four studies were included in qualitative synthesis, and 9 (12,386 patients; males, 64%-84%) in meta-analysis. The odds of a GO and SDO were significantly different between PTA >56 days and PTA <7 days (P= 0.04 and P= 0.03). PTA <7 days (mild TBI) excluded MDO/SDO and SDO alone with 87% and 90% sensitivity. PTA of 43-86 days (severe TBI) predicted MDO/SDO or SDO with 90%-96% and 80%-90% specificity. However, PTA of 7-42 days (moderate TBI) predicted MDO/SDO or SDO with only 38%-89% and 30%-80% specificity. PTA duration was reliable in predicting outcome when <7 days, and especially when >42 days but was often unreliable between 7 and 42 days duration.