ISBN:  Unavailable

Category: Unavailable

Page count: Unavailable

In the literature addressing the important problem of work injuries, there are wide variations in the analytic statistical methods applied, as well as in the reported effect-estimates. In this study effect-estimates mainly from the health and the economics literature, for some traditionally reported predictors of return to work (such as age), were transformed into one metric (relative risk) in order to allow comparison. Moreover, all of the frequently used analytic methods were applied to one Ontario WCB dataset (of 1845 injured workers), in order to determine the differences/similarities in the effect-estimates which result from the choice of analytic method only. From this study it was evident that ignoring inter-subject variation in Follow-Up and censoring, by failing to utilize survival analytic approaches, may result in a less precise effect-estimates. Also, a misleading effect-estimate may result when no attention is given to a possibly changing relative risk over time i.e. the time-dependency nature of certain predictors. Moreover, in the future, researchers have to be careful in the way they report their effect-estimates, and make explicitly justified choices between one single simple effect-estimate and a more accurate but complicated changing estimate over time.