Building a predictive model for back re-injury among injured workers in Washington State

Back pain and injuries are the costliest and most prevalent disabling occupational disorder in the United States, resulting in 101,800,000 annual lost work-days. A substantial proportion of workers with back injuries have re-injuries after returning to work, yet there are few studies of occupational back re-injuries; re-injury rates vary from 5-82%. We aim to develop a model for predicting self-reported 1-year back re-injury in occupational settings using early predictors. The Washington Workers' Compensation Disability Risk Identification Study Cohort provided a large, population-based sample to determine baseline and follow-up variables across six domains: pain, health, demographics, work-related factors, injury severity, and medical care. Initially, we identified significant predictors in bivariate logistic regression models. Future stepwise regression models will identify the strongest predictors in each domain, which will then be entered in a multivariable logistic regression predicting occupational back re-injury. In the year following the post-injury baseline interview, 297 of 1,319 workers (22.5%) had an additional occupational back injury after returning to work. Baseline predictors significantly associated in bivariate analyses include demographic factors, high occupational physical demands, a hectic job, whether a provider discussed methods to avoid re-injury, a history of back problems, a previous occupational injury with at least 1 month off work, and previously submitting a workers' compensation claim. Preliminary study findings indicate that variables in multiple domains predict occupational back re-injury. Building a model for accurately predicting occupational back re-injury may allow targeting of high-risk workers for population-based prevention interventions aimed at reducing re-injury risk.