In brief
The accident frequency rate is found by dividing the number of accidents causing incapacity by total actual hours worked and multiplying by 1,000,000; the severity rate by dividing the number of days lost by the same hours. For a correct result, hours are taken as actual hours from time-and-attendance records, and days lost are counted as calendar days from the day after the accident to the day before return to work. In comparisons, the multiplier, the accident definition and the counting rule must be kept constant; in small units a single accident can change the rate sharply.
Why do rates say more than the number of accidents?
The sentence "We had 6 accidents this year and 5 last year" means nothing on its own. If headcount rose from 180 to 240, or the plant moved from two shifts to three, more accidents may not mean worse performance. Accident frequency and severity rates solve this problem by dividing the number of accidents by the working time exposed. Accident frequency rate calculation is therefore a standard part of the annual evaluation, management reports and customer audits.
This article explains the formulas, where to collect the data and the common comparison mistakes. Which indicators you should track and how to balance them with leading indicators is the subject of our article on OHS performance indicators.
Accident frequency rate and severity rate formulas
The international basis for the formulas is the ILO resolution concerning statistics of occupational injuries, adopted by the International Conference of Labour Statisticians in 1998. Turkish legislation does not define a separate formula for the rates; businesses use these definitions.
| Indicator | Formula | What does it show? |
|---|---|---|
| Frequency rate (occupational accident frequency rate) | Number of accidents × 1,000,000 / Total actual hours worked | How many accidents occur per million hours worked |
| Severity rate | Days lost × 1,000,000 / Total actual hours worked | How many working days are lost per million hours worked |
| Days lost per accident | Days lost / Number of accidents | The average severity of accidents |
| Incidence rate | Number of accidents × 1,000 / Average number of workers | Accidents per 1,000 workers |
According to the ILO resolution, the severity rate is calculated only for temporary incapacity cases. Days lost may be estimated for fatal and permanent incapacity cases, but these data are compiled separately. Commuting accidents, if included, are also reported separately.
How are hours-worked data collected?
The accuracy of the denominator is the accuracy of the rate. The ILO resolution recommends using hours actually worked where possible, and otherwise using normal working hours corrected for paid absences such as annual leave, sick leave and public holidays.
Collect the data from these sources:
- Time sheets or the time-and-attendance system: Actual hours worked per shift. This is the most reliable source.
- Overtime records: Overtime is also exposure and is added to the denominator.
- Time to be deducted: Annual leave, days on sick leave, unpaid leave and public holidays; if the time sheet gives actual hours, these are already excluded.
- Temporary and subcontractor workers: If you are going to include subcontractor workers on site, take their hours too. Counting their accidents without taking their hours inflates the rate.
If, for a year without hours data, you have to use an estimate such as number of workers × annual working days × daily hours, state in the report that it is an estimate and keep using the same method in subsequent years.
Counting rule for the days lost rate
How days lost are counted is the issue that distorts the severity rate most. The rules of the ILO resolution are:
- Time lost is counted in calendar days; if working days are used, an attempt is made to convert them to calendar days.
- Counting starts the day after the accident and ends the day before return to work. The day of the accident is not counted.
- Absence of a few hours solely for medical treatment is not counted as a day lost.
- For repeated absences related to the same accident, each period is counted separately and added up.
- Temporary incapacity is taken into account for a maximum of one year.
Example: A worker had an accident on 10 March and returned to work on 24 March. The days lost are the 13 calendar days from 11 March to 23 March. Another site counting the same case in working days only would find a smaller number; the severity rates of the two sites therefore cannot be compared directly.
Which events are to be counted as an "accident" must also be defined at the outset. The scope of the ILO resolution is fatal accidents and injuries causing incapacity for at least one day (excluding the day of the accident). Track accidents that do not cause incapacity in a separate indicator; the notification obligation is a separate matter and is covered in our article on notifying minor occupational accidents.
Example: calculation at a plant with 240 workers
Let the annual data at a metalworking plant be as follows:
- Actual hours worked from time sheets: 452,600
- Accidents causing at least one day of incapacity: 6
- Total days lost (calendar days, excluding the day of the accident): 87
Calculation:
- Frequency rate = 6 × 1,000,000 / 452,600 = 13.26
- Severity rate = 87 × 1,000,000 / 452,600 = 192.22 (0.19 with a multiplier of 1,000)
- Days lost per accident = 87 / 6 = 14.5
- Frequency on a 200,000-hour basis = 6 × 200,000 / 452,600 = 2.65
When you split the same year by department, the picture changes:
| Department | Actual hours | Accidents | Frequency rate |
|---|---|---|---|
| Assembly | 180,000 | 3 | 16.67 |
| Press | 60,000 | 2 | 33.33 |
| Machining | 212,600 | 1 | 4.70 |
| Total | 452,600 | 6 | 13.26 |
Although assembly has more accidents, the highest frequency rate is in the press department, which is small in terms of hours. The priority for preventive work comes from this row; for measures specific to the press department, see our article on press safety.
Comparing years and units correctly
Rates can only be compared when they are calculated with the same rules. Make these checks:
- Is the definition constant? Is an "accident" one causing at least one day of incapacity, or all recorded accidents? If you have changed the definition, recalculate the previous years.
- Is the multiplier the same? Numbers based on 1,000,000 and 200,000 are not placed side by side.
- Is the days lost rule the same? Apply the calendar day, excluding-the-day-of-the-accident rule every year.
- Are the hours data from the same source? If time sheets were used one year and an estimate the next, note this.
- Small denominator warning: In a 30-person workshop, a single accident raises the frequency rate at a stroke. In small units, a three-year moving average and the raw number of accidents are tracked together.
- Long absences are tracked separately: A single serious case can drive the severity rate on its own. Explain this case separately in the report.
A 12-month rolling rate for monthly tracking
If you present rates to management every month, do not calculate using only that month's accidents and hours. A single month's denominator is too small; one accident doubles the rate and an accident-free month brings it down to zero. Instead, each month, calculate by adding up the number of accidents in the last 12 months and the actual hours of the last 12 months. For example, the March report uses data from April of last year to March of this year. The curve is then less affected by seasonal effects and by a single event, and the real trend becomes visible.
For this tracking, keep at least the following fields in each accident record: date of the accident, department, the injured person's employer (your own worker or a subcontractor's), start date of incapacity and date of return to work, number of days lost, and whether the case is fatal, permanent or temporary incapacity.
A fall in the rates does not always mean improvement. The number of accidents also falls when the reporting culture weakens. Interpret the rates together with the number of near miss reports and site walk findings; for the near miss system, see our article on the near miss reporting system.
For businesses in Bursa and the surrounding provinces, we enter accident records, days lost counts and rate calculations into the annual evaluation report as part of our OHS specialist service. For risks specific to metalworking plants, see our metal and machinery page. You can see the fines for breaches such as not keeping records and not reporting accidents in our OHS administrative fines tool.
Frequently Asked Questions
Should 1,000,000 or 200,000 be used in calculating the accident frequency rate?
Both are used. The ILO definition uses a 1,000,000-hour basis, while OSHA in the USA uses a 200,000-hour basis. What matters is using the same basis across years and sites and stating it in the report.
Is the day of the accident included in days lost?
Not according to the ILO rule. Days lost start the day after the accident, end the day before return to work and are counted as calendar days.
Are subcontractor workers' accidents included in the rate?
This is a definitional choice. If you include them, you must also add the subcontractor workers' actual hours worked to the denominator; otherwise the rate will come out higher than it really is. Apply the rule you choose in the same way every year.
How do fatal accidents enter the severity rate?
According to the ILO resolution, the severity rate is calculated only for temporary incapacity cases. Fatal and permanent incapacity cases are reported separately.
A certified specialist matched to your hazard class: site inspections, approved logbook, training plan, annual plan and report follow-up.
Bursa SİSTEMER OSGB Technical Team — written from the field experience of our OHS specialists and workplace physicians. Legal information is valid as of the publication/update date; contact us for an assessment specific to your business.