On a more serious note:
From this paper - Bennett, M.H., 2004, Risk, perception and sport--the doctor as policeman? SPUMS, 34(2), pp. 75-80
andA few examples may serve to illustrate these concepts. During 2003, a rogue sniper was infamously active in the Washington DC area. Several individuals were shot whilst in public places such as petrol stations and shopping mall car parks. There was understandably widespread concern for the safety of residents, and many individuals were reported as driving long distances to avoid exposure in such places in the Washington area. One author calculated that during this time there was a 1 in 517,422 chance (p = 0.00002) of being shot by the sniper, while the risk of death through a motor vehicle crash over the average extra miles travelled was likely to be appreciably higher.
Stats are great, but you need to make sure you are comparing apples with apples. This a real problem when the data does not exist, even anecdotally.The consequences of the 1995 report of the UK Committee on Safety in Medicines illustrate the potential dangers. The Committee reported, accurately, that there was a 100% RR [relative risk] increase of significant thromboembolic disease associated with the use of the 3rd generation oral contraceptive, as compared to the 2nd generation.13 Following a flurry of concern, many women abandoned the preparation, resulting in an estimated 8,000 extra abortions in the UK, and an unknown number of unwanted pregnancies. The AR [absolute risk] increase was from 3 to 6 deep vein thromboses for each 1,000,000 users per year, equivalent to a NNH [numbers need to harm] with the new preparation of 333,333 before one additional clot was caused. It is highly unlikely that more harm could be caused by continuing the use of the oral contraceptive pill than was caused by ceasing it. In this case, poor but accurate risk communication probably harmed the community.
Fundamentally we don't know the exposure (numbers diving each year in each 'discipline') nor do we have an accurate or consistent metric (fatalities per year (DAN); fatalities per dive (BSAC, but based on a flawed number of dives per year in my opinion); fatalities per hour in the water (most likely more useful but no-one uses it).
The BDSG are conducting a survey this year to determine the demographic of the diving population with the help of the RNLI. I was involved in looking at the data capture forms and had some concerns about the results they would get back and what they would mean. There is a training session this Saturday for RNLI volunteers who will help in this survey but I am not sure that my concerns have been allayed.