Safety Metrics: 9 Things About AFR, LTIFR and What They Actually Mean
At some point a tender, an insurer, a customer or a board paper asks for your accident frequency rate. Somebody works it out, a number goes in the box, and nobody asks how it was calculated.
That is a problem, because the same organisation can produce three genuinely different figures from the same underlying incidents, all of them correct, depending on which convention was used. And once a number is in a tender response, it is a representation.
Nine things.
1. Why you are being asked
Four routes, and they have converged in the last few years.
Tenders and supplier questionnaires, which now routinely request injury rates alongside policy documents.
Insurers, at renewal, as part of assessing risk.
Sustainability reporting, where workforce health and safety metrics sit inside social disclosure standards, and where organisations below reporting thresholds get asked by customers who are above them.
Boards, increasingly asking how the organisation compares rather than only how many incidents occurred.
2. Lagging and leading indicators do different jobs
The distinction that should shape any set of measures.
Lagging indicators count what has already gone wrong: injuries, lost days, reportable incidents. They are objective, comparable and entirely retrospective.
Leading indicators measure the activity intended to prevent harm: assessments completed and current, actions closed, training in date, inspections carried out, near misses reported.
An organisation reporting only lagging indicators is steering by the rear window. In a low-hazard office environment this is acute, because the incident numbers are small enough that year-on-year movement is mostly noise.
3. AFR: what it is, and the multiplier problem
Need Expert H&S Guidance?
Our qualified consultants can help you implement the right health & safety measures for your business.
Accident frequency rate expresses incidents against hours worked.
The common British formulation is the number of accidents divided by total hours worked, multiplied by 100,000. Other conventions use 200,000, which approximates 100 employees working a year, or 1,000,000.
The arithmetic is identical; the resulting number differs by a factor of ten or more. A rate of 0.5 against one multiplier and 5.0 against another describe the same performance.
Always state the multiplier. A figure without it is not interpretable, and a tender panel comparing suppliers who used different conventions is comparing nothing.
4. LTIFR: the definition problem
Lost time injury frequency rate counts injuries resulting in time away from work, expressed against hours in the same way.
The difficulty is what counts as lost time. Does a day of restricted duties count? Does the day of the injury? Is the threshold one day, one shift, or more than seven days in line with British reporting requirements?
Different organisations answer differently and each is internally consistent. Across organisations the figures are not comparable unless the definition travels with them.
5. Incidence rate is a different measure again
Frequency rates use hours worked. Incidence rates use headcount, typically expressed per 1,000 or per 100,000 employees.
HSE's national statistics present rates per 100,000 workers, which is the natural comparison point for a British organisation and worth using when a board asks how the organisation compares. The statistics tables allow comparison by industry.
Do not mix the two. A frequency rate and an incidence rate cannot be placed in the same trend line.
6. The denominator is where accuracy fails
An underrated point, and the reason many published rates are wrong.
The numerator, incidents, comes from records people take seriously. The denominator, hours worked, is frequently estimated: contracted hours multiplied by headcount, with no adjustment for part-time patterns, leave, absence or overtime.
Three specific errors. Excluding overtime understates hours and overstates the rate. Including contractors in incidents but not in hours inflates it. And using a year-end headcount for an organisation that grew or shrank during the year distorts it in either direction.
If you report a rate, be able to explain where the denominator came from.
7. Contractors and non-employees
A question every questionnaire eventually asks.
Whether contractors and agency workers appear in your figures is a choice, and it must be the same choice in both numerator and denominator.
There is a reasonable argument for including anybody working under your direction, which reflects where the duties actually sit, as covered elsewhere in this series. Whichever basis you choose, state it, and apply it consistently year to year.
8. Leading indicators worth actually using
For office-based organisations, these tell you far more than the incident count.
Assessments current: the proportion of required assessments reviewed within their trigger period.
Actions closed: the number open, and the age of the oldest. This single measure reveals more about a safety function than any other.
Statutory checks in date: the proportion of required inspections completed on schedule.
Training currency: proportion of required training in date, particularly first aiders and fire wardens, where expiry is silent.
Near miss reporting rate: rising is usually good, and a fall after a serious incident is the clearest warning signal available.
Time to close: how long from a finding being raised to it being resolved.
HSE's framework for managing health and safety places measurement in the check stage, ahead of review and action.
9. Presenting them honestly
The closing discipline, and it protects you.
State the basis with every figure: the multiplier, the lost time definition, whether contractors are included, and the period. Show the underlying count alongside the rate, because in a small organisation one incident can move a rate dramatically and the rate alone misleads in both directions.
Say when the numbers are too small to be meaningful. A thirty-person office reporting a frequency rate to two decimal places is presenting noise as precision, and an informed reader will notice.
And keep the records that support it. Regulation 12 of RIDDOR requires records of reportable incidents and of over-three-day injuries to be kept for at least three years, and a figure you cannot reconstruct is a figure you should not publish.
The nine, in short
| Point | Position |
|---|---|
| Why asked | Tenders, insurers, sustainability reporting, boards |
| Two types | Lagging counts harm; leading measures prevention |
| AFR | State the multiplier, always |
| LTIFR | State the lost time definition |
| Incidence rate | Headcount-based; do not mix with frequency |
| Denominator | Usually the weakest number in the calculation |
| Contractors | Consistent in numerator and denominator |
| Leading indicators | Open actions and their age tell you most |
| Presentation | State the basis; show the count; admit small numbers |
Rows three, four and six are where most published figures go wrong, and all three are about the method rather than the performance.
For international groups
Two considerations, and they make group-level rates harder than they look.
Reportable categories differ by country, sharply. This series has documented absence thresholds ranging from three days to seven, jurisdictions that include commuting accidents, and others that make vehicle accidents on premises reportable in their own right. A group rate built by adding up national reportable incidents is adding different things together.
Working time conventions differ too, which affects the denominator. Contracted hours, statutory leave and public holidays all vary.
The workable approach is a single group definition applied to the underlying incident data, with national reporting handled separately against local rules. That requires consistent incident capture across entities, which is where health and safety consultants and software are worth more together than either alone, and periodic health and safety audits test whether the group definition is the one being applied locally.
Where Arinite fits
Arinite establishes definitions that hold across entities and produces figures an organisation can defend when asked how they were calculated. We support 1,500+ businesses across 50+ countries and protect 100,000+ employees, with 95%+ client retention over 15+ years.
Our health and safety consultants work extensively with legal, insurance, finance and banking and IT and software organisations, all of which are asked for these numbers more often than they used to be.
Where entities sit in several countries, our global health and safety consultants reconcile local reporting categories to one group basis, and our international health and safety consultants keep that current.
If you put a frequency rate in a tender last quarter and could not now say which multiplier was used, a free gap analysis is the right place to start.
Related Articles
Written by
Arinite Health & Safety Consultants
Health & Safety Expert at Arinite
