Incident Investigation, Near Misses, Root Cause, and Corrective Action

Incident Investigation, Near Misses, Root Cause, and Corrective Action

Investigation Purpose and Reporting

Incident investigation is a prevention process, not a blame exercise. It establishes what happened, why, which controls failed, and what must change. The CHST exam tests whether the investigator stops at a surface cause. Saying a worker fell because they were not tied off may be true, but it does not explain whether anchors were available, the task was planned, supervision was effective, training was understood, or production pressure drove the decision.

Certain events also carry mandatory OSHA reporting timelines under 29 CFR 1904.39: a work-related fatality must be reported within 8 hours, and an inpatient hospitalization, amputation, or loss of an eye within 24 hours. A CHST should know these triggers because a missed report compounds the incident with a citation.

Investigations should cover injuries, illnesses, fatalities, property damage, fires, environmental releases, equipment failures, and near misses — unplanned events that could have caused harm under slightly different conditions, such as dropped objects, trench-wall movement, swinging loads, or near-contact with energized parts.

Immediate Steps and Evidence

First provide medical care, control hazards, stabilize the area, and complete required notifications; investigation never interferes with rescue or mitigation. Once safe, secure the scene and preserve evidence. Capture photographs, measurements, equipment positions, permits, inspection and training records, weather data, SDSs, and witness names promptly. Do not repair, discard, or return to service damaged equipment, failed rigging, PPE, ladders, tools, or electronic records before review — they are the physical evidence.

A basic sequence:

1. Care for the injured and control immediate hazards.
2. Notify required internal and external parties (including OSHA when triggered).
3. Preserve the scene and evidence.
4. Collect facts via photos, documents, measurements, and interviews.
5. Build a timeline of the event.
6. Identify immediate, contributing, and root causes.
7. Select corrective actions using the hierarchy of controls.
8. Assign owners, due dates, interim controls, and verification steps.

Witness interviews should be prompt, private, respectful, and fact-focused. Ask open questions — What did you see? What changed? What task was underway? What instructions were given? Separate direct observation from assumption and rumor, and interview witnesses individually so accounts are not contaminated.

Root Cause and Corrective Action

Root-cause analysis looks beyond the last action before the event. Tools such as the 5 Whys, barrier (fault-tree) analysis, fishbone (Ishikawa) diagrams, and causal-factor charts help, but disciplined thinking matters more than the tool. "Retrain the employee" for every event is usually weak unless the analysis shows why training failed and why training is the right control.

Corrective actions follow the hierarchy of controls — elimination, substitution, engineering, administrative, then PPE — and must be specific, assigned, and verifiable. "Install guardrails on the west mezzanine before work resumes" is far stronger than "be careful." Each action needs an owner, a due date, interim controls until completion, and an effectiveness check. If a fix introduces a new hazard, it must be revised.

Share lessons learned with affected crews while protecting privacy and avoiding unsupported blame. Trend incident and near-miss data for recurring drivers — housekeeping, struck-by exposure, heat illness, subcontractor coordination. The investigation is complete only when corrective actions are implemented and verified, not when the report is filed.

Causal Layers: Immediate, Contributing, Root

The exam expects you to separate three causal layers. Immediate causes are the unsafe acts and conditions present at the moment of the event — an unguarded edge, a bypassed interlock, a missing anchor. Contributing causes are factors that made the immediate cause more likely — a rushed schedule, an incomplete pre-task plan, inadequate supervision, or a damaged tool kept in service. Root causes are the management-system failures underneath — no inspection program, no procedure for the task, training that was assigned but never verified for understanding. A 5 Whys chain moves deliberately from the visible act down through these layers; stopping at the first "why" leaves the system untouched and the event free to recur.

Worked Example

Consider a worker struck by a falling beam during a lift. Immediate cause: the rigging sling failed. Contributing causes: the sling had a cut that an inspection would have caught, and the lift plan did not exclude workers from beneath the load. Root causes: there was no documented sling-inspection program and no enforced exclusion-zone procedure for overhead lifts. Weak corrective action: "counsel the rigger." Strong corrective actions: implement a tagged, periodic sling-inspection program (engineering/administrative), require and barricade exclusion zones under all suspended loads (administrative with physical barrier), and verify both through audits. Each is specific, assigned, and checkable.

Data Quality and Trending

Individual investigations prevent the next identical event; trending prevents categories of events. Aggregating leading indicators (near misses, audit findings, training-verification rates) and lagging indicators (recordable rate, days away) across the project surfaces patterns a single report cannot — for example, a cluster of struck-by near misses tied to one subcontractor's lift practices. The CHST should ensure investigations feed a common dataset with consistent cause coding, so trends are real signals and not artifacts of how each form was filled out.

• Drive every analysis through immediate, contributing, and root layers.
• Reject single-cause, single-person conclusions for serious events.
• Write corrective actions at the system level, not just the individual.
• Trend leading and lagging indicators with consistent cause coding.