Investigating patterns and causes of struck-by accidents in roadway construction projects.

Abstract

Struck-by accidents involving vehicle intrusions and heavy equipment in construction work zones, particularly utility systems, highways, streets and bridge projects, pose significant safety risks. These incidents often resulting from interactions between workers, machinery and vehicles, frequently lead to serious or fatal injuries. This study utilized a comprehensive dataset of 3,268 OSHA accident reports from 2000 to 2022 to examine patterns, contributing factors, and the severity of struck-by accidents. Through a multi-method approach combining descriptive statistics, chi-square analysis, and predictive modeling (logistic regression, Random Forest, and Gradient Boosting), the study addressed four core research questions focused on frequency, risk factor association, and injury severity prediction. The results revealed that struck-by incidents comprised over 50% of all reported construction work zone accidents and accounted for nearly 70% of related fatalities. Risk factors significantly associated with fatality included injury nature (e.g., amputation, asphyxia), worker occupation (e.g., construction laborers, highway maintenance workers), project type, cost, time of day, and specific activities such as excavation and trenching. While logistic regression offered interpretability (AUC-ROC = 0.74487), ensemble models provided greater predictive accuracy (AUC-ROC = 0.78-0.79). It also underscores the need for standardized data reporting to enhance future modeling efforts.