Development and Validation of a Clinical Prediction Rule of the Return-to-Work Status of Injured Employees in Minnesota.

Abstract

Vocational rehabilitation services can be a valuable resource to injured employees at risk for sustaining permanent disability. The aim of this study was to develop and validate a predictive model of return-to-work (RTW) status at workers' compensation claim closure that may assist rehabilitation counselors tasked with determining how to allocate such services. A cross-sectional, retrospective study was conducted using data obtained from 15,372 workers' compensation claims in Minnesota's administrative claims database. The association between a set of 15 predictor variables representing medical and contextual factors and the RTW status as of claim closure of the accessible population was assessed using backward stepwise logistic regression. The most parsimonious set of variables that reliably predicted the outcome was selected as the optimal RTW model. This model was then internally validated via a split-dataset approach. Risk factors for failure to RTW by claim closure include the following: (1) attorney involvement; (2) higher level of permanent impairment (PI); (3) shorter job tenure; (4) lower pre-injury average weekly wage (AWW); (5) injury affecting the head and neck or the back; and (6) lower level of educational attainment. The optimal RTW model included four main effects (attorney involvement; severity of PI; age; job tenure) and three first-order interaction effects (pre-injury AWW × pre-injury industry; attorney involvement × severity of PI; attorney involvement × job tenure). When applied to the full dataset, the overall classification rate was 74.7%. This study's optimal RTW model offers further support for evaluating disability from a biopsychosocial perspective. Given the model's performance, it may be of value to those assessing rehabilitation potential within Minnesota's, and possibly other, workers' compensation system(s).