Posted By Ian Waldram A tricky area, it's easy to get confused! Traditionally, accident data was compiled 'per hours worked' - this started in the days when most workers clocked on, so organisations knew the hours without much difficulty. If they didn't, the usual assumption was '2000 hours per worker per year' (= 40/week x 50weeks). In UK, the data is typically per 1000 employees (= 200,000 hours), but in US they usually use 100,000 hours. More recently HSE has not bothered to assume the 2000 hours per employee, so just quoted rates per 1000 employees. Of course, the only injuries HSE knows about are those reported to it, i.e. lost time and major injuries. As safety improves, the actual number of reportable injuries per 1000 empoyees per year is often less than 1, so the number get a bit small. Thus HSE's most recent figures (see their website for 1998/99 data just released per occupational type) are for 100,000 workers. The average for all occupations is 1,490, or 1.49 per 1000 employees. The 'worst' occupations have rates 3.2 times this average. For internal purposes, it is more statistically reliable to count 'all injuries', or at least all that require some sort of medical treatment, i.e. not just a plaster from a First-Aider. This is what are called 'recordable injuries' in US legislation (they have very detailed definitions to explain what this means in practice, which can lead to very creative post-injury arguments and efforts as to why something shouldn't actually BE recorded - thus completely losing sight of the objective of learning how to prevent it happening again!). Thus many US-based organisations use 'all-injury' data as a performance indicator, and this usually means 'recordable injuries', whereas HSE 'all-injury' data means everything notified to them - lost-time+ major+ fatalities. If comparing with US-based recrdable injury data, remember this wil be per 100,000 hours, i.e. 500 employees. 'Severity' data is usually a measure of the total time lost per lost-time injury. I think the best way to measure it is in days. It becomes VERY confusing when it is quoted a 'hours lost per 100,000 hours worked', as it often is in US. As we all know, the actual time lost has a quite poor correlation with the actual severity of the injury, so I've never found severity data helps much. There is also the issue of how many hours to 'count' if you have a fatality or severe disablement and the employee NEVER comes back to work!