Ranking the occupational incident contributory factors: A Bayesian network model for the petroleum industry

Abstract

IntroductionA vast amount of research has been conducted to identify human and organizational factors that contribute to the occurrence of occupational incidents. Considering the identified factors, the question is how much the occupational incident probability will decrease in the absence of one or more recognized contributory factors. MethodsTwenty-one fatal accident reports were selected for Root Cause Analysis (RCA). The contributory factors were identified by content analysis of the accident scenarios. A 5-point Likert questionnaire was developed to measure the probability of identified factors. Using the identified contributory factors and their corresponding probabilities, a Bayesian network model was constructed for estimating the probability of the occupational incident in the absence of each contributory factor. ResultsProcedure violation, poor risk perception, and poor management commitment were three top-ranking contributory factors. The Bayesian network estimated that preventing procedures violation could cause a reduction of 44 % in the occupational incident probability. ConclusionUsing Bayesian network’s advantages is an effective technique for quantifying occupational safety risks. Ranking the contributory factors enables us to choose the most effective prevention strategies. Procedure violation (a type of unsafe act) was the most influencing factor in occupational incident probability.