It should not be a search for someone to blame.

Immediate response boundaries

First priorities may include:

  • emergency response;
  • medical support;
  • making the area safe;
  • preserving evidence;
  • required notification.

The exact response depends on site procedures and jurisdiction.

Evidence collection

Useful evidence can include:

  • scene observations;
  • photographs where permitted;
  • equipment condition;
  • alarms/logs;
  • documents;
  • maintenance history;
  • procedures;
  • training records;
  • witness accounts.

Preserve facts before resetting or altering the scene where practicable and permitted.

Interviews

Good interviews are fact-finding, not interrogation.

Ask:

  • What were you trying to do?
  • What happened before the event?
  • What did you expect?
  • What conditions were unusual?
  • What normally happens?
  • What made the chosen action seem reasonable?

Timeline

Build a sequence of:

  • conditions;
  • actions;
  • system responses;
  • decisions;
  • deviations.

A timeline often reveals interactions missed in narrative summaries.

Causal analysis

Separate:

Immediate event: what physically happened.

Contributing conditions: factors that made it more likely.

System causes: design, planning, supervision, maintenance, workload, training, change, culture or control weaknesses.

Root-cause methods

Methods can include:

  • 5 Why;
  • fishbone;
  • barrier analysis;
  • fault-tree thinking;
  • causal-factor charting.

No method guarantees a correct root cause.

Use evidence.

Corrective actions

Strong actions target the cause.

Prefer, where feasible:

  • elimination;
  • design/engineering change;
  • stronger system controls.

Training alone may be insufficient if the system encourages the same error.

Effectiveness review

Ask:

  • Was the action implemented?
  • Did the hazardous condition change?
  • Has the failure mode repeated?
  • Did the change create new risk?

Closure is not the same as effectiveness.

Reporting

A good report distinguishes:

  • fact;
  • inference;
  • unresolved uncertainty;
  • recommendation.

Portfolio exercise

Use a public incident or fictional case.

Build:

  • timeline;
  • evidence table;
  • causal analysis;
  • actions;
  • effectiveness plan.

A strong investigation changes the system without pretending uncertainty does not exist.

Related content

Sources