A new Human factors incident taxonomy for members of the public (HFIT-MP): An investigation of escalator incidents

Background: Escalators are common across many urban environments, yet, incidents and fatalities can occur during their use. This research takes a Human Factors approach to: 1) produce an Incident Taxonomy for Members of the Public (HFIT-MP); and 2) apply the taxonomy to a unique dataset of serious escalator incidents. Methods: The HFIT-MP was developed using Human Factors literature, Health and Safety Executive (HSE) guidance, and subject matter experts. 403 narrative escalator incident reports were obtained from the HSE and coded according to the HFIT-MP. Cohen's Kappa was used for inter-rater agreement and Multiple Correspondence Analysis (MCA) coupled with cluster analysis was used to identify factors that tended to occur together. Results: The HFIT-MP consisted of four overarching themes and 25 factors. All factors achieved a Kappa greater than 0.60. Analysis of all incident reports indicated that 93% were falls. MCA identified six groups of factor categories that tended to occur together; for example, falls occurred more frequently in people aged 65 or over, who lost their footing, getting onto the escalator, that was not moving unexpectedly, travelling alone, and moving in a descending direction. Conclusion: Results provide health and safety practitioners with target groups (e.g., age), actions (e.g., getting onto the escalator) and locations (e.g., descending escalator) to inform interventions and the need to consider the interaction of multiple factors among incidents. Further application of the HFIT-MP to different sources and modes of data, such as video footage, could further refine the taxonomy for incident investigations.