What does the frequency index measure?
The frequency rate, or FR, relates the number of recorded accidents to the working hours of the group being analyzed. Its usual unit is accidents per million hours worked. Therefore, a result of 20 does not mean twenty accidents in the company or twenty percent of people injured: it expresses a normalized relationship between incidents and hours.
This standardization helps interpret situations with varying levels of activity. A company might register more absolute accidents by significantly increasing its working hours and, at the same time, present a lower accident rate. This data does not eliminate the need to investigate the incidents, but it avoids interpreting a count without knowing the corresponding level of exposure during the period.
Formula and variables
The usual expression is: FR = number of accidents × 1,000,000 / hours worked. Before calculating it, it is necessary to determine which accidents are included: for example, accidents resulting in sick leave during working hours. Series can also be constructed using other criteria, provided they are explicitly identified and not compared as if they represented the same set of events.
The hours must correspond to the same group and period as the accidents. If incidents involving contracted personnel are included, a denominator consisting solely of the hours of the company’s own staff cannot be used. Similarly, adding hours from sites whose accidents are excluded from the numerator artificially reduces the result and breaks the indicator’s consistency.
How to prepare the data
For internal calculations, it’s advisable to keep a record of the accident count and the source of hours, in addition to the final result. The record should allow for the identification of duplicates, reclassifications, and the criteria applied to each incident. Scheduled hours, paid hours, and actual hours worked are different concepts; the record should explain which one is used and maintain a consistent methodology.
If the hours are based on an estimate, this should be indicated. When the timekeeping system or the company’s scope changes, it’s advisable to check whether the previous series remains comparable. A decrease that coincides with a new method for calculating the denominator may have an administrative explanation, even if the graph suggests an improvement in working conditions.
Practical example
In a hypothetical example, a company records four lost-time workplace accidents during a period with 200,000 hours worked. The calculation is 4 × 1,000,000 / 200,000 = 20. The result is presented as 20 lost-time workplace accidents per million hours worked, along with the period and the employee group included. The four accidents remain four separate incidents that require analysis.
Another unit records one accident in 50,000 hours: its FR is also 20. Both have the same observed relationship, even though their counts are different. This does not prove that they have identical risks or identical preventive quality. To interpret the comparison, tasks, severity, reporting, and sample size must be reviewed, in addition to the characteristics of the work performed.
How to calculate the cumulative total
The cumulative index is obtained by first adding the accidents and hours for the included periods and then applying the formula. It should not be calculated as a simple arithmetic mean of monthly indices when each month has a different number of hours. The ratio of the sums maintains the actual weighting of the workload for each period.
For example, one accident in 10,000 hours results in an FR of 100, while another in 90,000 hours results in approximately 11.11. The cumulative total is two accidents in 100,000 hours, or 20. Averaging 100 and 11.11 would give approximately 55.56 and would inaccurately describe the overall picture. It is advisable to retain sufficient decimal places during the calculation and round only when presenting the result.
Comparison and interpretation
Comparison requires consistent time periods and criteria. Monthly data should not be interpreted as having the stability of a long-term series. In small groups, a single accident can significantly alter the rate; therefore, it is advisable to also show totals and times. The analysis should consider whether new centers have been added, activities have been modified, or incident reporting has been improved.
The severity index adds another dimension, related to the duration of absences, and the incidence index uses the working population as its denominator. The three indicators address different questions. They should not be combined in the same classification without explaining their units, nor should they be interpreted as a single, comprehensive measure of preventive performance.
Use in OSH management
The frequency rate (FR) can help detect changes that warrant investigation, but it does not, on its own, identify their causes. A variation must be compared with the types of accidents, the operations involved, and the observed conditions. Accident investigation allows us to move from simply analyzing the numbers to identifying the specific mechanisms that can be addressed through technical and organizational measures.
It is advisable to combine this with proactive indicators, such as the verification of important controls or the effective closure of measures. These provide information on preventive activities before damage occurs. Monitoring should promote reliable reporting: pushing to reduce the index without improving conditions can mask incidents and diminish the usefulness of the information system.
Errors and limitations
Common errors include confusing accidents with injured people, mixing incidents that occur during work hours with those that occur while commuting, and using the hours of a different group. It is also incorrect to automatically attribute an improvement to a campaign simply because the rate decreased afterward. Random variations, changes in activity, or differences in record-keeping may be involved and need to be analyzed.
A zero frequency rate (FR) only indicates that no accidents included in that series were recorded during the period. It does not demonstrate an absence of hazards or control of serious scenarios. Prevention must address working conditions and applicable obligations, even if the indicator is low. The statistical value supports decisions, but it does not replace evaluation or the implementation of measures.
