Prediction of memory performance in long term survivors of severe closed head injury

The Selective Reminding Procedure (SR), the California Verbal Learning Test (CVLT), and the Rivermead Behavioral Memory Test (RBMT) were administered to a group of long term ( greater than one year post-injury) closed head injury survivors. As a memory performance measure, patients obtained therapists' signatures on behavioral contracts after each hourly training module over a two week time period. All three memory tests significantly predicted behavioral contract performance. The RBMT, which was designed to reflect real world memory failures in memory impaired individuals, was a significantly better memory performance predictor than the two verbal list learning tasks. CVLT scores were not significantly better than SR procedure scores at predicting memory performance. Injury severity was significantly correlated with CVLT scores and the memory performance measure but not with RBMT or SR scores. The memory performance measure did not add significantly to the prediction of injury severity over CVLT scores. Predicting outcome, planning interventions, and assessing potential for rehabilitation may be best served by using more functionally oriented ecologically valid tests such as the RBMT.