The joy of stats

I wrote in the post "The poor are stupid..." that I would come back to this when I got a reply. Well, to give them their due a lady at the RCP has replied to my question of where they got the data for the "alcohol related deaths double" statement and it's here.

Now there is of course more than one way of looking at statistics. On the basis of their figures it is legitimate to say that of the proportion of people dying, the number of men doing from alcohol related somethings has, (roughly), doubled.

Of course it is also fair to say that the percentage of men dying from alcohol related somethings has gone from 0.009% to 0.018% and that for women the change is from 0.005% to 0.008%.

I may be doing something wrong here but that doesn't seem like a significant change to me. This is however, not the only issue. There is a question mark over the reporting due to a change from ICD-9 to ICD-10 codes in 1999, (it doesn't seem to have any immediate effect on the stats but I hope to find the raw data at some stage and I mention it for the sake of completeness at the moment).

There are, at least, two other possibilities for the increase. The first and most likely is reporting bias. As people, particularly the medical profession, have become more obsessed with the dangers of alcohol, they are more likely to report a death as alcohol related. What may once have been put down as a heart attack, may in a few instances become part of the alcohol rated statistics. This is not to say that it is not alcohol related but that it would not previously have been considered so. The second is related to immigration, if your genetic background does not include those who drank beer because the water was foul you will be badly affected by alcohol. We have had rather a lot of such people come into the UK and it would be surprising if some of them didn't have a few problems in this area. (No idea if stats are even collected on this but it is an example of how other possibilities can account for small changes in reported incidence).


Any ideas, further data or ability with statistics is welcome.