Injury Statistics That Do Not Match the Clinic Log
Golden Harvest Foods employs 320 workers on three shifts. The safety dashboard shows four recordable injuries year-to-date and a TRIR of 1.1, a figure the plant manager presents at client audits. You are reviewing source records and find: the clinic log records 61 visits for lacerations, minor burns, and low-back pain; a worker who received three sutures returned to work the same day and appears in the log as 'first aid'; another wore a wrist splint for three weeks on light duties and was logged as 'first aid, no lost time'; a supervisor's daily report mentions two hand incidents that do not appear in the injury register. The company runs a quarterly pizza-party incentive for departments with zero recordables. A worker tells you his supervisor said, 'Don't file it - you'll cost the whole line its bonus.' Last year's annual exposure data report cannot be located. A client audit team will review three years of injury data next week.
1. Assess the accuracy of the injury data. What principles determine how a case should be classified, and how do the examples fall?
Read the worked answer
Classify by what was done to the worker and what happened to his work, not by what the supervisor prefers to call it. The working test is whether the case involved medical treatment beyond first aid, restricted work or transfer, days away, or loss of consciousness. Apply it: three sutures is medical treatment, so it is a recordable case, not first aid. Three weeks in a splint on light duties is both medical treatment and restricted work, and likely days of restricted activity - recordable, and probably a more serious case than several of the four currently counted. The two hand incidents in the supervisor's daily report need to be traced to clinic entries, witnesses, and payroll, because a missing register entry is either a classification failure or a concealment. The dashboard is not merely rounded; it is unsupported. A reasonable reconciliation exercise: sample 20 clinic entries across the year, classify each against the criteria with the clinic nurse present, then extrapolate carefully and label the estimate as an estimate. Expect the true count to be materially higher than four and the TRIR of 1.1 to be unreliable. Note why this is not a paperwork issue: undercounting hides hazards from the risk assessment, distorts prioritization and budget, and can deprive injured workers of employee compensation entitlements. Say that plainly in the audit response rather than presenting a number you cannot defend.
2. Design a reporting system and a set of leading indicators that resist gaming.
Read the worked answer
Build the system so that the person closest to the event reports, and the person with a bonus at stake does not decide. Elements: one intake route for injuries, near misses, and hazards, available on paper and by QR code with an anonymous option; a written classification guide with worked examples, applied by the safety office and clinic rather than by supervisors; a monthly reconciliation of clinic log, injury register, and payroll light-duty records, signed by two people; a feedback loop that tells the reporter what was fixed; and a no-blame stance for reporting, distinct from the disciplinary process for deliberate unsafe acts. Then shift the management focus to leading indicators: near misses and hazards reported per 100 workers, percentage closed within 14 days, scheduled inspections completed, corrective actions closed on time, training completion with competency verification rather than attendance, exposure monitoring coverage against the plan, OSH committee actions completed, participation in toolbox talks, and housekeeping scores from structured walks. Keep the lagging indicators but stop rewarding them. The design test for any indicator is simple: if a supervisor can improve the number by suppressing a report, it is the wrong indicator. Report the leading set monthly alongside the corrected lagging data so that management sees the relationship between reporting volume and injury outcomes.
3. Analyze the incentive scheme and the supervisor's instruction, and decide how to handle the client audit.
Read the worked answer
The incentive as designed buys silence. A quarterly reward for zero recordables gives every worker and supervisor a financial reason to keep injuries off the register, and the supervisor's instruction to a worker not to file is the predictable result. Under RA 11058 workers are protected when they report, and a scheme that economically punishes reporting has the same practical effect as an explicit prohibition - it should be replaced, not tweaked, with recognition for reporting, hazard identification, and timely closure of corrective actions. The supervisor's instruction requires a separate response: a documented discussion, retraining on reporting duties, and a disciplinary route if it recurs, plus a clear statement from the plant manager that reporting is protected and that no bonus will be affected by a report. Offer the affected workers a re-review of their cases and their compensation entitlements. For the audit, do not present the old dashboard and hope. Present the corrected data, the reconciliation method, the reason the previous figures were wrong, and the corrective action plan with owners and dates; auditors generally accept a discovered-and-corrected problem far more readily than a trend that does not survive scrutiny. Reconstruct or verify the annual exposure data report with the Regional Office so you can evidence the filing. And keep your own written record of what you found and what you escalated - this is the kind of finding that gets a safety officer overruled until it becomes an inspection finding.