Predicting duration of posttraumatic amnesia (PTA) from early PTA measurements

To determine a set of variables that would reliably predict duration of posttraumatic amnesia (PTA) in patients with traumatic brain injury and to test the efficacy of the model. Simultaneous standard multiple regression analyses. Two independent samples of patients with traumatic brain injury who were in the early stages of PTA: a test sample (n = 61) and a cross-validation sample (n = 25). The Modified Oxford PTA Scale (MOPTAS) is a 12-item test measuring orientation (8 items) and anterograde memory (4 items). The Galveston Orientation and Amnesia Test (GOAT) was also used on a subset of the test sample. Patients were examined daily until they emerged from PTA. A statistically significant model, using three predictor variables, was derived that reliably predicted duration of PTA, accounting for 89% of the variance. A second model, using two predictor variables readily available to the clinician (day posttrauma on which PTA testing began and aggregate PTA scores over the first 5 days of testing) had comparable predictive accuracy. A third model, using GOAT data, was also statistically significant and successfully accounted for 72% of the variance. The MOPTAS model showed excellent application to an independent (validation) sample, with an intraclass correlation coefficient between observed and predicted durations of PTA of 0.95. Regression equations for all three models are provided to enable calculation of the predicted duration of PTA. These models can be readily applied in clinical practice and will provide clinically useful estimates of the duration of PTA within the first week of testing after admission to rehabilitation. This information will be important in terms of family counseling and planning of rehabilitation programs.