Head injury, a multivariate study: predicting long-term productivity and independent living outcome

As the second of a two-part research project, this study attempted to build three clinically useful prediction equations; these equations would be used to predict three long-term functional outcomes: productivity status, living arrangement status, number of hours of assistance required per week. There were ten predictor variables used in each analysis. These ten variables belonged to four theoretical classes: pre-injury demographics, pre-injury social characteristics, severity of injury, and post-injury environmental variables. The analytical method used was the General Linear Model of multiple regression, rather than the most often used Stepwise regression procedure. The results indicate that in none of the three prediction equations was more than 28% of the criterion variance accounted for, and only a minority of the predictor variables were significant predictors. It is concluded that, while the equations are of only very limited utility, the theoretical model and the analytical technique will be useful for future investigators.