-
I used to be very critical of (what I knew of) the STATS19 database. However, since then I've looked at systems in other European countries, and it turns out that STATS19 is pretty much the best of the lot. Yes, it has under-reporting (very, very high for slight injuries, quite considerable for serious injuries, although *very* rare for fatal injuries), but that is much worse in all the other European countries that I've studied. In some countries, there's even a huge shortfall in the registration of deaths.
STATS19 can and should be improved (better training for officers at the scene, different reporting categories (although this would then create a break in data continuity), etc.), but as ever the main problem is the extent to which it is seen as gospel by some and the 'crash-chasing' of using it as evidence for design 'improvements', which often results in worse and counter-productive designs.
It's amazing that there isn't even a reliable database of victims. I know the police figures are disputed, fatalities get missed out.