Developing a tool for identifying high-risk employers for inspection.

Background: Workers' Compensation Board (WCB) data and other information are sometimes used to calculate an 'Occupational Health and Safety (OHS) index' as a way of identifying businesses considered 'high risk' to be inspected as part of enforcement work. However, no evidence on the validity of this index exists. Aims: To evaluate the performance of the Alberta OHS index, a 'score' based largely on WCB claims data, and to see if an index calculated using different information could perform better. Methods: Data from the Alberta Compliance Management Information System database, 2011-2015, and WCB claim database, 2007-2014, were retrieved. Issuing 'stop work' or 'stop use' orders in inspections was defined as a proxy of high-risk outcome. The performance of the current and a modified OHS index were assessed using receiver operating characteristics (ROC) and regression analyses. Results: In large employers, neither the current nor the modified OHS index was particularly effective in identifying 'high risk' employers with the area under the ROC curve (AROC) of 0.55 (95% confidence interval [CI] 0.52-0.57; P < 0.001) and 0.59 (95% CI 0.57-0.62; P < 0.001), respectively. In small employers, neither index seemed very effective with an AROC of 0.54 (95% CI 0.53-0.56; P < 0.001) and 0.55 (95% CI 0.53-0.56; P < 0.001), respectively. These results were consistent in subgroup analyses of assignments without specific initiatives, both in large and small employers. Conclusions: Neither the current nor a modified OHS index seemed to effectively identify high-risk employers. Heterogeneous results in large and small employers suggest that approaches to different-sized employers are appropriate. [ABSTRACT FROM AUTHOR]