A bivariate zero-inflated Poisson regression model to analyze occupational injuries

Area of Science:

• Occupational Health and Safety
• Biostatistics
• Ergonomics

Background:

• Occupational safety interventions aim to decrease injury rates.
• Injury count data often exhibit excess zeros, complicating effectiveness measurement.
• Zero-inflated models are suitable for data with a point mass at zero and a parametric component.

Purpose of the Study:

• To propose and evaluate a bivariate zero-inflated Poisson (BZIP) regression model.
• To assess the effectiveness of a participatory ergonomics team intervention.
• To analyze injury incidence in a hospital cleaning services department.

Main Methods:

• Development of a bivariate zero-inflated Poisson (BZIP) regression model.
• Application of the BZIP model to analyze injury data from a hospital cleaning staff intervention.
• Evaluation of intervention impact on musculoskeletal (MLTI) and non-musculoskeletal (NMLTI) injuries.

Main Results:

• The BZIP distribution demonstrated a good fit for the observed injury data.
• The participatory ergonomics intervention led to a significant reduction in overall injury incidence.
• A significant decrease in mean musculoskeletal (MLTI) injuries was observed, with a marginal reduction in non-musculoskeletal (NMLTI) injuries.

Conclusions:

• The BZIP regression model is a valuable tool for analyzing occupational injury data with excess zeros.
• The evaluated ergonomics intervention was effective in reducing overall and MLTI injuries.
• This methodology can be applied to assess interventions in other high-risk occupational groups.