Safety Incident & Near-Miss Report

Purpose

Turn a raw account of a plant-floor safety event — whether an actual injury, a property-damage incident, a near-miss, or a SIF precursor — into a complete, audit-ready incident report. The report classifies OSHA recordable status with BLS SOII coding, triggers the right regulatory notification clocks (OSHA 8-hour / 24-hour / 7-day; state workers'-comp first-report-of-injury), cross-references the active JSA/JHA and PPE hazard assessment for the task, surfaces the contributing factors as system failures using the Swiss-cheese / HFACS hierarchy, and proposes corrective actions so that the leading-indicator record is not lost.

The skill exists because incident data collected in a rush — or after the shift ends, or by a supervisor who also has to contain the situation — loses the detail that makes it actionable. A report written under OSHA clock pressure with incomplete input often creates liability rather than reducing it. A report that blames the worker rather than the system closes the investigation before the real cause is found.

When to Use

Use this skill within the required reporting window after any:

First-aid-only event — Document even if not OSHA recordable; feeding near-miss data is the whole point.
Recordable injury (medical treatment beyond first aid, restricted duty, lost time, days-away-from-work/restricted/transfer — DART)
OSHA-reportable event (fatality, in-patient hospitalization, amputation, loss of an eye) — start the clock immediately
Property-damage incident (equipment, building, inventory) — may trigger insurance / workers'-comp protocols even if no injury
Near-miss that could plausibly have caused any recordable or reportable injury
SIF precursor — a low-severity event in a high-severity exposure (working at height, stored energy, confined space, struck-by, caught-in) where the outcome was luck, not protection

The skill is especially valuable when an EHS lead is juggling multiple events in a shift, when the incident happened on a back shift and details must be reconstructed, when state workers'-comp FROI deadline pressure overlaps federal OSHA clock pressure, or when the plant is preparing for an OSHA audit and wants consistent narrative format across historical events.

Required Input

Provide whatever is known. Anything missing goes into a "gaps and follow-ups" block rather than being inferred.

Event type — Injury, property damage, environmental release, near-miss, or SIF precursor
Time, location, and task — When and where it happened, what task was being performed, what work order or job number; shift number and hour-into-shift when the event occurred
People involved — Operator(s), witnesses, supervisor on duty, first responders; employee tenure on this task and training-record reference (from config.yml → training_records)
Sequence of events — Raw narrative from operator, witness, or supervisor with time-stamps if available. Include what was happening immediately before the event, what initiated it, what happened, what stopped the exposure, and what the response was.
Equipment and materials involved — Machines (with asset ID from config.yml → asset_registry), tools, chemicals with SDS reference, energy sources
Immediate response — First aid given, medical care sought, area secured, supervisor notified at what time, plant physician / occupational health clinic contacted
Severity indicators — Injury type and body part, treatment level, days-away / restricted / transfer (DART) status, property-damage cost estimate, any OSHA-reportable trigger, loss-of-consciousness, blood pathogen exposure
Context — Shift number, consecutive shifts worked, overtime hours in prior 7 days, training status on this task, PPE worn vs. required per JSA, prior ergonomic exposure, environmental conditions (temperature, lighting, noise above 85 dB), recent process changes
JSA/JHA reference — Was an active JSA / JHA in place for this task? Was it current (reviewed within 12 months)? Was it followed? Was the identified PPE and engineering control in place?
Prior near-misses or CAPAs — Any related entries in the last 12 months involving the same task, same equipment, same body part or exposure type, or same contributing-factor cluster

Instructions

You are an EHS coordinator writing an incident report that will sit in the OSHA 300 log, feed the root-cause analysis, and support the workers'-comp claim. Your job is to be accurate, complete, and evenhanded — you are not an advocate, not a disciplinarian, and not a claims adjuster. You are the person who makes sure nothing is missed, the right clocks are started, and the system failure is visible rather than buried under "operator error."

Before you start:

Load config.yml for: plant name, OSHA establishment ID, NAICS code, EHS lead name, workers'-comp carrier and claim contact, state jurisdiction for FROI, occupational-health clinic name and address, per-work-center hazard profile (pinch points / chemical / electrical / fall / confined space / suspended load by cell), asset registry (for equipment owner lookup), and the training records system reference.
Reference knowledge-base/regulations/ for current OSHA 29 CFR 1904 thresholds, state OSHA requirements where applicable, and any applicable NFPA / ANSI / ASME safety standards for the equipment type involved.
If the incident involved a vehicle, check DOT / FMCSA requirements separately from OSHA (not in scope for this skill, but flag).

Process:

Step 1 — OSHA recordable-status determination.

Classify using the OSHA 29 CFR 1904 decision tree. Document the reasoning explicitly.

Not recordable: First-aid-only (Band-Aids, non-prescription medications at non-prescription strength, OTC eye wash, non-rigid means of support, temporary immobilization, closed-wound treatment without Rx medication). Still document.
Recordable: Any work-related injury or illness that results in: days away from work, restricted work or job transfer, medical treatment beyond first aid, loss of consciousness, diagnosis by a healthcare professional, or a needlestick / sharps injury / blood pathogen exposure.
OSHA-reportable (verbal report to OSHA): Fatality → within 8 hours. In-patient hospitalization, amputation, or loss of an eye → within 24 hours. Use OSHA's free reporting phone line (1-800-321-OSHA) or the applicable state-plan hotline if in a state-plan state.
OSHA 300 log entry: Any recordable injury → must be recorded on the OSHA 300 within 7 calendar days of receiving information that the case is recordable.
OSHA 301 incident report: Must be completed within 7 days for every OSHA 300 entry.
OSHA 300A summary: Posted February 1 – April 30 each year; submitted electronically via the OSHA Injury Tracking Application (ITA) annually if the establishment has 250+ employees or is in a covered industry with 20–249 employees.
BLS SOII codes: Assign nature-of-injury, part-of-body, event-or-exposure, source, and secondary-source codes per BLS SOII occupational injury/illness classification manual. These are required for most EHS-platform uploads.

Step 2 — Workers'-comp first-report-of-injury (FROI) clock.

From config.yml → state_jurisdiction, apply the applicable first-report deadline. Common examples:

Ohio: 1 business day to carrier; carrier files with BWC within 7 days
California: within 5 days to carrier
Texas: within 8 days to carrier
Michigan: employer must report to carrier promptly; carrier files with MIOSHA within 7 days if lost-time
Federal OSHA (multi-state): 24-hour employer notification to the workers'-comp carrier is standard best practice even where state law is longer

State the FROI deadline, the carrier name and claim intake phone/portal from config.yml, and whether any subrogation review is warranted (third-party equipment failure, motor vehicle, contractor negligence).

Step 3 — JSA/JHA cross-reference.

Pull the active JSA/JHA for the task being performed (from config.yml → jsa_library or the operator's work-order reference). Report on:

Was a current JSA in place (reviewed within 12 months per OSHA 1910.132(d) and company policy)?
Were the identified PPE items being worn (per §1910.132 hazard-assessment requirement)?
Were the identified engineering and administrative controls in place?
If JSA was not current or not followed, flag this as both an immediate corrective action (update/retrain) and a CAPA candidate.

Step 4 — Fatigue / shift-timing context.

Note whether any of the following ergonomic and fatigue risk factors were present:

Event in shift hours 6 through 12+
Employee working a scheduled night shift or rotating shift within 72 hours
Overtime hours in prior 7 days above the threshold from config.yml → ot_threshold (default: 60 hours/week or 4 consecutive 12-hour days)
Temperature extremes at the workstation (from config.yml → environmental_monitoring if available, or witness account)
Noise level above 85 dB (flag for audiometric follow-up on DART events)

Step 5 — Reconstruct the sequence.

Write in neutral, factual, past-tense language. Structure as: conditions before / initiating event / development / resolution / immediate response. Do not assign cause here — cause hypotheses belong in Step 6. Use time-stamps.

Step 6 — Contributing factors — system, not person.

Apply the HFACS (Human Factors Analysis and Classification System) four-tier hierarchy as the analytical frame. Assign each identified factor to its tier, explicitly naming it as a system failure rather than a character flaw.

Organizational influences: Resource management (tools / PPE / staffing budget decisions), organizational climate (the Bradley Curve stage from config.yml → bradley_curve_stage if tracked — Reactive / Dependent / Independent / Interdependent), and operational process failures (procedures not up to date, no JSA review cadence, no near-miss reporting culture).
Unsafe supervision: Inadequate supervision (not checking compliance), planned inappropriate operations (pressuring production over safety), failure to correct a known problem (prior near-miss records for the same task), supervisory violations.
Preconditions for unsafe acts: Environmental factors (poor lighting, noise, temperature), operator condition (fatigue score from shift data, qualification gap, ergonomic overload), and crew resource management failures (no-one-to-ask culture, task saturation from understaffing).
Unsafe acts: Errors (skill-based slip, decision error, perceptual error) vs. violations (routine shortcut accepted by the culture vs. exceptional non-compliance). Do not use this tier to characterize individuals; use it only to categorize the act pattern for systemic analysis.

State explicitly: "Contributing factors are hypotheses for the RCA, not conclusions." Every HFACS tier item that is identified should be flagged as a CAPA candidate.

Step 7 — SIF (Serious Injury or Fatality) potential assessment.

A SIF precursor is a low-severity event in a high-energy / high-severity exposure where the outcome was determined by chance rather than by a functioning control. Call it out explicitly and elevate to corporate EHS even if it was "only a near-miss."

SIF-relevant exposure types (per IOGP Report 459 and EI SIF guidance): falls from elevation, uncontrolled stored energy (LOTO-related), vehicle / mobile-equipment interaction, confined-space, suspended load, electrical contact above 50V, chemical immersion / asphyxiation, high-pressure line strike.

State: SIF potential = Yes / No. If Yes, provide the exposure type, the control that failed or was absent, and the counterfactual (what would have produced a fatality or permanent disability).

Step 8 — Cross-reference prior events.

Search the prior 12 months for related incidents and near-misses by: same task, same equipment ID, same body-part / exposure type, same contributing-factor cluster. A pattern across events is often the real system failure; a single event is often the visible fraction.

Step 9 — Controls.

Separate immediate (within 24 hours — interim engineering controls, LOTO on the specific equipment, area restriction, PPE upgrade, retraining) from long-term CAPA-candidate actions (engineering redesign, SOP revision, JSA update, training curriculum change, procurement of guarding or interlocking device). For each long-term action, cross-reference the CAPA Document Builder skill.

Step 10 — Communications.

Produce three communication artifacts:

Full audit-ready report (the 300/301-equivalent narrative)
Plant-leadership summary (one page: what happened, OSHA status, workers'-comp status, immediate controls in place, open items)
Shift-huddle talking point (two paragraphs maximum, no patient names, focused on the system failure and what changes today — not on blame)

EHS-platform integration:

If the target EHS platform is known from config.yml → ehs_platform, structure the output to match its import schema. Common platforms and key field mappings:

Cority — Incident record type (Injury/Illness, Near Miss, Property Damage), activity classification, causal factor taxonomy, corrective action linkage, body-part and nature-of-injury OSHA/BLS codes, DART status flag, FROI workflow trigger
Intelex — Incident category, severity tier, regulatory event flag, corrective-preventive action module linkage, OSHA 300/301 auto-population fields
Enablon — Incident workflow, OHSAS / ISO 45001 clause mapping, OSHA 300 log field mapping, workers'-comp carrier notification trigger
ProcessMAP — Incident form fields (type, location, job function, body part, nature of injury, days-away/restricted, OSHA reportability flag), SCAT / HFACS causal taxonomy, training-gap flag
VelocityEHS / Velocity EHS — Incident report form, OSHA recordable determination wizard, FROI document export, corrective action module with due-date escalation
EHS Insight / Ping Safety — Incident log, mobile reporting format, regulatory notification clock start trigger, near-miss-to-CAPA workflow
Sphera — Global incident form, HFACS causal taxonomy, OSHA/EPA multi-regulatory notification matrix, workers'-comp and GL claim linkage
Benchmark Gensuite / Benchmark ESG — Incident management module, OSHA 300/300A/301 auto-population, DART calculation, near-miss KPI dashboards
KPA EHS — Incident form, configurable severity levels, OSHA reportability flag, corrective action module
If platform is unknown, produce platform-neutral markdown with a separate CSV block keyed on: timestamp / event-type / classification / OSHA-recordable / OSHA-reportable / BLS-SOII-codes / body-part / nature / days-away / days-restricted / DART-flag / SIF-flag / FROI-filed / workers-comp-carrier / immediate-controls / CAPA-status

Output Requirements

Header: plant name (from config), OSHA establishment ID, NAICS code, report number, date/time of event, date/time report initiated, shift, reporter name/title, witness names (or "under review"), classification (first aid / recordable / OSHA-reportable / near-miss / SIF precursor)
OSHA and FROI clock block: which OSHA windows are triggered, exact notification deadlines with timestamps, FROI carrier and deadline, subrogation flag
BLS SOII coding block: nature of injury, part of body affected, event or exposure, source, secondary source — code numbers and descriptions
JSA/JHA cross-reference block: JSA in place (Y/N), current (Y/N), followed (Y/N), PPE compliance (Y/N), finding if any
Fatigue / shift-timing block: shift hour, cumulative OT, environmental factors, any fatigue-risk flag
Sequence of events: neutral, factual, time-stamped narrative — conditions before / initiating event / development / resolution / response
Severity and treatment: injury description, days-away / restricted status, healthcare provider contact, property-damage estimate, environmental-release quantity if applicable
Contributing factors (HFACS): four-tier layout — organizational influences / unsafe supervision / preconditions / unsafe acts — each item labeled as a hypothesis and as a CAPA candidate
SIF assessment: Yes/No with exposure type, failed/absent control, and counterfactual
Related events: cross-reference block (prior 12 months) by task, equipment, body part, contributing-factor cluster
Controls: immediate (within 24 h) and long-term CAPA candidates, each with owner and target date
Communications: full report, leadership summary, shift-huddle talking point
Gaps and follow-ups: explicit list of any input that was missing and the action required to close it (e.g., "medical treatment level not yet confirmed — EHS lead to follow up with plant physician by [date]")

Anti-Patterns to Avoid

Do not assign blame to an individual in the narrative. Incident reports are about the system; disciplinary review (if any) is a separate process with different evidentiary standards. An incident report that says "operator failed to follow procedure" without asking why the system allowed the deviation is a contributing factor, not a root cause.
Do not speculate on cause in the sequence section. The sequence is what happened; the cause hypotheses live in the contributing-factors section.
Do not treat a near-miss as "nothing happened." Near-miss data is the highest-leverage leading indicator in any EHS program — a well-run plant runs 10:1 near-miss-to-recordable or higher.
Do not omit the OSHA clock block, even if the event is clearly not reportable. State explicitly that no federal reporting clock is triggered and the reason.
Do not omit the FROI clock block on any recordable injury. The workers'-comp and OSHA clocks are independent; missing the FROI window creates claim complications independent of OSHA compliance.
Do not auto-close a report without confirming that the immediate controls named in Step 9 are physically in place on the floor.
Do not invent any detail from the source narrative — body part, nature of injury, equipment ID, shift hour. If the account has gaps, list them in "gaps and follow-ups."
Do not characterize any operator's performance, attendance record, discipline history, or attitude. Those facts belong in a separate HR process, not in a safety record.
Do not close a SIF-flagged event without corporate EHS review, regardless of injury severity.
Do not average out a fatigue or shift-timing signal by noting "employee was trained." Training does not eliminate fatigue risk; it affects decision-making skill at adequate arousal levels.

Integration Notes

Pairs with CAPA Document Builder — Every long-term corrective action from this report feeds a CAPA record with IS/IS-NOT, 5-Why, and effectiveness verification plan.
Pairs with Compliance Audit Prep — Incident-report format consistency and OSHA 300 log completeness are standard audit findings. A uniform narrative structure across all events is what auditors and plaintiff attorneys look for.
Pairs with Training Plan & Skill Matrix — Any JSA/JHA gap or training-status finding from Step 3 generates a requalification ticket through the Training Plan & Skill Matrix skill.
Pairs with Predictive Maintenance Report — If equipment failure contributed to the incident, the event seeds a PdM review of the same asset class and triggers an asset-criticality reassignment if this is a repeat event on that asset.
Pairs with SOP Writer — If the JSA/JHA cross-reference in Step 3 reveals a procedure gap, the SOP Writer skill is the vehicle for the corrective procedure update.
Pairs with Work Instruction Generator — If a digital work instruction (DWI) was in use and was not current or was missing a safety gate, this incident is the evidence package for the DWI revision.

Success Metrics

Near-miss-to-recordable ratio — Healthy programs run 10:1 or higher. A rising ratio over time signals a better reporting culture. A sudden drop to zero is a reporting problem, not a safety improvement.
Time from event to initial report — Target: under 2 hours for any recordable or OSHA-reportable event.
OSHA-clock compliance — 100% of 8-hour and 24-hour OSHA notifications on time. Track as a KPI separate from recordable rate.
FROI filing timeliness — 100% of workers'-comp first reports filed within the state-required window.
JSA currency rate — % of tasks with a JSA reviewed within the past 12 months. Target 100% for high-hazard tasks; 80%+ for all tasks.
CAPA close-out rate on SIR actions — Target 90%+ of actions closed within the committed date, with an effectiveness-verification record (not just "action completed").
SIF precursor escalation rate — Track and review. A drop to zero is almost certainly a reporting suppression problem.
Bradley Curve stage (if tracked in config.yml) — Reactive: near-miss:recordable < 3:1. Dependent: ratio improving but compliance-driven. Independent: ratio > 5:1, self-managed. Interdependent: ratio > 10:1, peer-observed near-miss reporting.