Estimating the avoidable burden and population attributable fraction of human risk factors of road traffic injuries in iran: application of penalization, bias reduction and sparse data analysis

Abstract

The aim of this study was to prioritize human risk factors for preventive interventions by estimating the avoidable burden and population attributable fraction (PAF) of each risk factor using penalization and data augmentation method. To avoid the sparse data bias, Bayesian logistic regression via data augmentation methods, were used for multivariable analysis. Informative normal priors adopted from the studies were used for the studied human risk factors. Weakly informative log-f was used for the covariates. The population attributable fraction was calculated based on direct method. The comparative risk assessment methodology of the WHO was used to estimate the potential impact fraction for each risk factor. The most important human factors influencing the traffic-related deaths were overspeeding (OR = 9.6, 95% CI: 2.45-37.7), reckless overtaking (OR = 8.6, 95% CI: 1.82-40.7), and fatigue and drowsiness (OR = 6.7, 95% CI: 1.79-25). The total PAF for the all studied risk factors was about 56% (PAF = 0.567, 95% CI: 0.37-0.7). The greatest avoidable burden was related to fatigue and drowsiness, overspeeding, and not fastening seatbelt. By considering the high contribution of human risk factors in occurrence of fatal traffic injuries appropriate legislation and prevention programs for these risk factors would decrease half of such deaths.

Keywords:

• Bayesian logistic
• data augmentation
• population attributable fraction
• potential impact fraction
• fatal traffic injuries
• Iran

Disclosure statement

No potential conflict of interest was reported by the authors.