How organisations learn from safety incidents: a multifaceted problem

Abstract

Purpose This paper seeks to review current approaches to learning from health and safety incidents in the workplace. The aim of the paper is to identify the diversity of approaches and analyse them in terms of learning aspects. Design/methodology/approach A literature review was conducted searching for terms incident/accident/near misses/disaster/crisis modified with learning/training and safety. Shortlisted articles were analysed by questioning who is learning, what kind of learning process is undertaken, what type of knowledge is employed and the type of problem that these incidents addressed. Current approaches to learning from incidents were critically analysed and gaps identified. Findings Very few papers addressed all the envisaged aspects when developing their learning from incidents approaches. With support from literature, it was concluded that all the four perspectives, namely participants of learning (participation and inclusion), learning process (single loop, double learning), type of incident and its relation to learning (Cynefin complexity framework) and types of knowledge (conceptual, procedural, dispositional and locative) are important when deciding on an appropriate learning from incidents approach. Research limitations/implications The literature review focused on journal articles and identified keywords, which might have narrowed the scope. Further research is needed in identifying ways to embed the learning from incidents aspects in the organisation. Practical implications The framework developed could be useful by safety planners, safety managers, human resource managers and researchers in the area of organisational learning and safety. Originality/value The paper concludes by outlining key questions and proposing a framework that could be useful in systematically analysing and indentifying effective approaches to learning from incidents.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 78 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.