Risk Matrix Underestimates a Caustic Drum-Handling Task
A distribution center uses a 5x5 risk matrix with severity 1-5 and likelihood 1-5. For transferring 55-gallon drums of 50% sodium hydroxide from a pallet to a containment curb, the team scored severity 4 for a disabling injury and likelihood 2 for unlikely, giving a score of 8, or "medium." The listed controls were "wear goggles and gloves" and "be careful." Two months later a drum slipped, the bung loosened, and caustic splashed an operator's face; goggles were worn but no face shield. The drum had oil on it from a forklift, and the operator was using a bent drum hook. The team now proposes lowering likelihood to 1 because "everyone has been retrained." The facility has 90 employees and no formal management-of-change procedure.
1. Critique the initial risk assessment. Why was likelihood underestimated, and what does the scoring reveal about the use of the matrix?
Read the worked answer
The initial assessment treated PPE as the primary control and used "be careful" as a control, which is not a control at all. Likelihood 2 was not supported. The task is performed frequently, the drums are heavy, the hook was bent, and oil on the drum created a known slip condition. The team likely anchored on the absence of past incidents and underestimated human variability and tool condition. Exposure frequency alone should have pushed likelihood higher.
The score of 8 also illustrates a matrix limitation: a medium score can hide a severe outcome. Severity 4 was plausible, and a single event could cause blindness or a serious burn. A matrix is only a screening tool; it must be supported by task observation, worker input, and consideration of the worst credible outcome and the reliability of each control.
2. Evaluate the proposal to lower likelihood from 2 to 1 after retraining. Is that justified? Explain.
Read the worked answer
No. Retraining is an administrative control that depends on memory and behavior. It does not remove the oil, replace the bent hook, or eliminate the manual lift. Likelihood should be reduced only when a control is verified to reduce the probability of the initiating event. "Everyone has been retrained" is not verification of control effectiveness.
The team should not lower the score before implementing and validating engineering controls. A drum handler, a non-slip surface, drip pans, proper hook inspection, and a face shield would reduce both likelihood and severity. After those controls are in place, the team can re-score with evidence such as observation, inspection records, and worker feedback. Retraining may be added, but it cannot be the basis for a lower likelihood score on its own.
3. Recommend a revised assessment and control set, including how to verify effectiveness.
Read the worked answer
Revise the assessment with the task broken into steps: staging, lifting, moving, opening, and pouring. For each step, identify the initiating event and consequence. Use severity 4 for the caustic burn potential unless the consequence is reduced by engineering. Likelihood should reflect the oil, bent hook, and manual handling; at least 3 until controls are verified.
Controls: use a drum lifting or rotating device, or move the containment curb to the pallet level to avoid manual lifting; clean oil from drums and provide drip pans; inspect and replace the drum hook; require a face shield over goggles, a chemical apron, and gloves validated for 50% sodium hydroxide; write a procedure and JHA. Verify by observing the task, checking hook inspection records, and interviewing workers. Add a management-of-change review for new drums, tools, or workstations.