Corrective Actions That Keep Repeating
A 350-person metal fabrication shop holds ISO 45001 certification. Its three-year TRIR has been flat at 4.1 to 4.4, and management describes safety as stable. Over eight months, three operators on the same press brake suffered hand lacerations. Each investigation closed with retrain the operator and add a toolbox talk. An internal audit found that 16 of 22 corrective actions over two years were training-only, and that the risk register had not been updated when a harder high-strength steel replaced mild steel on that line, roughly doubling required tonnage and increasing kickback. Production supervisors sign off on corrective actions. Management review examines only TRIR, lost-time rate, and training completion percentage. A new safety manager wants to change the system, not just the press brake.
1. Evaluate the corrective-action process. What would an effective action look like for this press brake?
Read the worked answer
Diagnosis Three lacerations on one machine in eight months is one common-cause problem, not three operator errors. A process that reliably lands on training has two defects: it stops at the first plausible cause, and it has no hierarchy-of-controls screen. Training-only closure is appropriate only when the verified gap is knowledge or skill; here the hazard is in the point of operation and the material.
Process changes
- Require a causal analysis (events and causal factors or fault tree, not 5-why alone) with a stated root cause and contributing factors.
- Add a mandatory hierarchy screen: elimination, substitution, engineering, administrative, PPE, with justification for skipping higher tiers.
- Define closure criteria at initiation, including how effectiveness will be verified and by whom.
- Screen every injury for fatality or amputation potential, not just actual severity.
Effective action here Light curtain or two-hand control with hold-downs, feed tooling so hands stay out of the die area, sharp-edge removal, and revised setup for the harder steel. Procedure and training follow the engineering change; they do not replace it.
2. Was a management-of-change trigger missed? Describe how MOC should have worked for the steel substitution.
Read the worked answer
Yes, this is a textbook MOC trigger. A material change to high-strength steel altered required tonnage, kickback energy, tooling loads, and possibly the adequacy of existing guarding and procedures. Because the change came through purchasing and engineering rather than safety, no screening occurred.
How it should work
- A trigger list covering material and chemical substitution, new or modified equipment, process, tooling, procedures, organization, and temporary changes.
- A screening step owned by whoever initiates the change, before the first production run.
- For triggered changes: hazard review, updated job hazard analysis and risk register entries, verification of guarding and energy control, revised procedures, training, and a pre-start review with sign-off.
- Communication to affected shifts and to contractors, plus a defined review date after startup.
Why it matters Without MOC the risk register silently becomes fiction, which is exactly what an auditor should test by sampling recent engineering and purchasing changes rather than reading the register itself.
3. Redesign the management review inputs. What should leadership see, and what should be removed or reframed?
Read the worked answer
Reframe
- TRIR and lost-time rate stay, but as outcome measures alongside severity and fatality-potential events, not as the definition of safety. Training completion alone is an activity count, not evidence of competence; pair it with observed behavior checks.
Add
- Risk register currency and percent of changes screened through MOC.
- Corrective-action quality: percent with a verified engineering or higher-tier control, percent verified effective at 30/60/90 days, overdue actions by owner.
- Near-miss reporting rate and quality, audit findings and closure, supervisor safety interactions, and contractor performance.
- Context measures that drive risk: turnover, overtime, production volume, and new-hire ratio.
How it is used Each metric needs an owner, a target, and a decision. A review that receives data but makes no resourcing or priority decisions is a reporting ritual, and it is why a flat TRIR was read as stability when the underlying exposures were deteriorating.