Organizational variation in safety observation practices and their relationship to accidents: A three-year multi-site study

Introduction: Safety observations are widely used as proactive tools in safety management, yet their relationship with accident outcomes remains unclear. Method: This study examines safety observation practices and their association with total recordable injuries (TRIs) across seven production units within the same industrial organization during 2023–2025. Safety observations were normalized as observations per 100 working hours and accident occurrence was measured using TRI rates. Results: The results reveal substantial variation among production units in both observation activity and accident occurrence. Temporal analysis shows a recurring pattern in which observation reporting declines during mid-year and increases toward the end of the year. Correlation analysis indicates no consistent linear relationship between observation activity and TRI levels across units. Conclusions: The findings suggest that safety observation frequency alone does not function as a direct predictor of accident occurrence. Instead, observation activity appears to reflect broader organizational factors such as reporting practices, safety engagement, and local safety management processes. Practical applications : The study highlights the importance of interpreting safety observation metrics within their organizational context when assessing safety performance.