When Near-Miss Reports Stop: A Process Safety Warning Sign

API Recommended Practice 754 formalizes the logic behind process safety leading indicators, organizing near-misses and precursor events into a tiered structure (Tiers 3 and 4) that functions as a leading indicator of process safety health — distinct from occupational safety indicators like Heinrich's triangle, which were developed from personal-injury data and are not a reliable predictor of major-accident risk. A plant generating near-misses at a stable underlying rate but reporting fewer of them over time has not become safer; it has become less visible to itself.

A sustained decline in reported near-misses is read by plant leadership, absent other evidence, as evidence of improving safety performance — the intuitive but usually wrong interpretation. The more probable explanation, particularly where the decline is not accompanied by any corresponding change in operating conditions, procedures, or equipment, is that the workforce has stopped reporting, not stopped generating near-misses.

James Reason's 'just culture' framework distinguishes between human error, at-risk behavior, and reckless conduct, applying escalating accountability to each — a distinction that only functions if it is applied consistently and is visibly separate from ordinary performance review. Where reporting and disciplinary review share a channel, operators rationally treat every report as a potential liability rather than a safety contribution.

• →Investigate a declining near-miss rate as a governance priority requiring root-cause analysis, not a metric to celebrate
• →Audit whether closure actions on filed reports are visible to the reporting workforce, not only recorded internally
• →Separate the near-miss reporting channel from the performance and disciplinary review process structurally, not just in policy language

A plant that closes the loop visibly on every near-miss report and firewalls reporting from disciplinary consequence sustains a reporting rate that tracks the actual underlying hazard rate, giving leadership an accurate leading indicator. A plant that allows reporting to erode loses that visibility precisely when it matters — and typically discovers the gap only after a serious incident forces a retrospective look at a near-miss history that had gone quiet.

• →Investigate every declining near-miss report rate as a governance priority, not a performance improvement
• →Close the loop visibly on every report, so operators observe directly that reporting changes something
• →Keep the reporting process separate from performance and disciplinary review to protect the trust it depends on