The steps
- Respond and secure. Treat the injured, make the area safe, preserve the scene and notify the regulator if the incident is notifiable.
- Gather facts. Photographs, measurements, equipment state, documents (procedures, training records, maintenance history) and interviews with those involved and witnesses, separately and soon.
- Build the sequence of events. A timeline of what happened, in order, from the facts.
- Analyse causes. Move from the immediate cause (the unguarded nip point) to the underlying and organisational causes (why it was unguarded, why nobody raised it, why the audit missed it).
- Recommend actions. Controls chosen from the hierarchy of control, with owners and dates.
- Report and share. Findings to the people affected, the leadership team and other sites with the same hazard.
- Verify. Confirm the actions were done and the control holds.
ICAM root cause analysis
ICAM (Incident Cause Analysis Method) is a root cause analysis method for serious and potentially serious incidents. Developed in Australia from James Reason's "Swiss cheese" model of organisational accidents, it is widely used in mining, rail and heavy industry. It works back from the event through four layers:
- Absent or failed defences. The barriers that should have stopped the event: guarding, interlocks, isolation, alarms.
- Individual or team actions. The errors or violations before the event.
- Task and environmental conditions. What made those actions likely: workload, fatigue, time pressure, an unworkable procedure.
- Organisational factors. The system weaknesses behind those conditions, such as training, procedures, maintenance and management of change.
Recommendations are written against the failed defences and the organisational factors, where the lasting fix sits, so the investigation cannot stop at the person.
5 whys and fishbone
For lower-severity incidents and near misses, five whys at the tier board gets to a workable cause in minutes. A fishbone diagram organises causes when several factors combine.
What makes investigations fail
- Blaming the person and stopping.
- Investigating weeks later, when memory and the scene have changed.
- Actions that are all retraining and reminders (administrative controls).
- Reports filed with nobody checking the actions were done.
Incident investigation inside daily management
A near miss raised at the huddle is investigated by the team that day with five whys; a serious event triggers ICAM with a cross-functional team reviewed at Tier 2. In both cases the actions sit on the same board as every other action, with owners and dates, and the verification step is a dated check that the control is in place. Safety management software built into daily management keeps the investigation, its actions and that verification on the one board.