Prediction of time in industrial chemical accidents: A survival analysis

Abstract

BACKGROUND: Chemical accidents have imposed casualties and high economic and social consequences to Iranian industries and society. OBJECTIVE: This study investigated the effect of risk factors involved in occurrences of the chemical accidents and predicted the time of occurrences in Iranian chemical factories. METHODS: A cross-sectional study was implemented in 574 chemical facilities with more than 25 employees from 2018 to 2020. Collecting data instruments were 2 checklists with 15 and 25 three-point Likert scale questions, respectively. Chi square and Monte Carlo tests assessed the relationships between independent risk factors and dependent hazardous chemical accidents. Cox semi-parametric and log-normal parametric models were used to predict the upcoming time of chemical accidents based on the impacts of risk factors understudy. Data analyses were performed using Stata and R software. RESULTS: The results showed that safety data sheets, labeling, fire extinguishing system, safe chemicals storage, separation, loading, transportation and training were statistically significant with occurrences of the chemical accidents (P-value < 0.05). Loading and transportation were mostly related to chemical incidents and reduced significantly the expected time of chemical events (P-value = 0.028). CONCLUSION: Establishing a comprehensive chemical accidents dataset and strict governmental supervision on chemical safety regulations are suggested to decrease the chemical accidents at regional and local levels in chemical plants.