Prepare occupational safety data with matched populations, definitions and hours
Before calculating an injury rate, agree the metric definition, worker population, reporting period and hours-worked source. Keep the underlying counts and classification decisions traceable. A correct division can still produce a misleading rate if the numerator and denominator cover different people or periods.
This guide is for reporting teams working with the responsible health-and-safety and HR owners. It focuses on data preparation for review. It does not classify an individual injury, diagnose a condition, investigate an incident or replace statutory recording and notification duties.
Write the reporting basis before requesting a total
Name the framework, edition and exact measure. Lost-time and recordable-injury rates are different; don’t relabel them. Ask the competent data owner to explain the classification rules, pending cases and changes since the previous period.
GRI 403-9 separates employees from non-employees whose work, workplace or both the organisation controls. Keep each group’s counts and hours distinct; contractor labels don’t determine coverage. GRI 403:2018
Map the covered sites and worker groups to the records used for both injuries and hours. Record exclusions and missing information. If a site's incident data is included but its hours are missing, the resulting rate is not made comparable by adding unrelated hours from another population.
Keep the source schedule reviewable
Metric name, definition, framework and version.
Entity, sites, worker group and exact dates covered.
Aggregate counts supplied by the authorised owner, with classification status.
Hours-worked source, compilation method and any estimates or exclusions.
Rate basis, formula, units, rounding and review owner.
Corrections, unresolved items and the records supporting their treatment.
Keep person-level incident and medical records within appropriate restricted access. A general reporting workbook usually needs an authorised aggregate and a controlled way for a reviewer to trace it, rather than copies of clinical details. Confirm whether source systems count events, affected people or injury records, and resolve differences before combining them.
Apply the selected rate basis consistently
Use a 200,000- or 1,000,000-hour basis. Injury fatalities enter recordable counts but not the separate high-consequence measure; responsible owners confirm classifications. GRI 403-9
Use count ÷ matching hours × stated basis. Document estimates’ method, scope, reason and review; headcount isn’t hours.
A fictional two-population calculation
For one fictional entity and its two sites over 1 April 2025 to 31 March 2026, the authorised schedules provide the following figures. The example uses a rate basis of 1,000,000 hours and assumes the supplied classifications and populations have been agreed for the calculation.
Fictional recordable-injury rates by reporting population
Employees
Recordable injuries: 4.
Hours worked: 800,000.
Rate per 1,000,000 hours: 4 ÷ 800,000 × 1,000,000 = 5.
Other workers whose work and/or workplace is controlled
Recordable injuries: 2.
Hours worked: 200,000.
Rate per 1,000,000 hours: 2 ÷ 200,000 × 1,000,000 = 10.
Two other-worker injuries divided by the combined 1,000,000 hours gives two: a population mismatch. A suitably defined combined measure cannot replace the separate results.
This partial recordable-injury example omits fatality, high-consequence-injury and other required 403-9 information. Missing categories are not zero. These fictional figures cannot establish which workforce is safer.
Interpret changes with the reporting context
Injuries describe harm, not safety. Explain boundary changes, recording/reporting practices and limitations before comparing periods. GRI 403-9
Check whether the prior-period metric used the same population, definition, hours basis and multiplier. Investigate a large movement before attributing it to a programme. Training delivered, actions closed and controls assessed are separate evidence streams; they should not be inferred from one rate.
Close the reporting review without hiding open cases
Reconcile the aggregate schedule to the authorised source extract, confirm the hours total and reproduce the calculation. Have the responsible owner review unresolved classifications and later information that could affect the result. Keep corrections visible rather than quietly replacing the earlier schedule.
Nitium's sustainability and GRI reporting support supports disclosure mapping, evidence organisation and reporting gaps. Discuss your reporting needs with the intended measure, population and available aggregate records, keeping personal and medical details out of the initial enquiry. Medical judgments, technical safety assessments, legal opinions and independent assurance are outside this reporting-preparation guide.
Sources checked 10 October 2026. Check the applicable framework and local duties for the actual business; this is not a complete safety disclosure or compliance assessment.
Comments