Back to the future: What do accident causation models tell us about accident prediction?

• Published in: Safety science Vol. 104; pp. 99 - 109
• Main Authors: Grant, Eryn, Salmon, Paul M., Stevens, Nicholas J., Goode, Natassia, Read, Gemma J.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier Ltd 01.04.2018; Elsevier BV
• Subjects: Accident analysis; Accident causation; Accident prediction; Accidents; Causality; Causation; Complex systems; Drift and FRAM; Forensic engineering; Mathematical models; Normal Accident Theory; Risk management; Risk management framework; STAMP; Systems analysis; Accident prediction; Normal Accident Theory; Drift and FRAM; STAMP; Risk management framework; Accident causation
• ISSN: 0925-7535; 1879-1042
• DOI: 10.1016/j.ssci.2017.12.018

•The research is a synthesis of multiple accident causation models representing a comprehensive review of contemporary thinking on accident causation.•Despite the diversity in models there is considerable agreement regarding the core tenets of system safety and accident causation.•To support accident prediction, both safe and unsafe properties are required to capture the influences underpinning systematic weaknesses. The prediction of accidents, or systems failure, should be driven by an appropriate accident causation model. Whilst various models exist, none is yet universally accepted, but elements of different models are. The paper presents the findings from a review of the most frequently cited systems based accident causation models to extract a common set of systems thinking tenets that could support the prediction of accidents. The review uses the term “systems thinking tenet” to describe a set of principle beliefs about accidents causation found in models proposed by Jens Rasmussen, Erik Hollnagel, Charles Perrow, Nancy Leveson and Sidney Dekker. Twenty-seven common systems thinking tenets were identified. To evaluate and synthesise the tenets, a workshop was conducted with subject matter experts in accident analysis, accident causation, and systems thinking. The evaluation revealed that, to support accident prediction, the tenets required both safe and unsafe properties to capture the influences underpinning systematic weaknesses. The review also shows that, despite the diversity in the models there is considerable agreement regarding the core tenets of system safety and accident causation. It is recommended that future research involves applying and testing the tenets for the extent to which they can predict accidents in complex systems.