The number of nonsmokers admitted to hospital after heart attacks fell by 20 per cent in the ten months after the ban came into force in March 2006, compared with the same ten months in the year before, Jill Pell, of Glasgow University, said.... In the ten months of the year leading up to the ban, there were 3,235 admissions, while in the matching period after the ban, the figure was 2,684.The general angle of the entire piece is that there the two fact when juxtaposed are somehow evidence of causality. Indeed, Jon Ayres, head of the University of Aberdeen environmental and occupational medicine department, is quoted as saying "This also suggests that the important thing was the smoking ban." Actually it doesn't at all.
All it suggests is less people had heart attacks. To draw conclusions of causality based on a single, and deliberately chosen catalyst based on nothing more than correlation with one previous year is entriely foolish. In fact, let's flip this on its head for a second.
If the number of heart attacks amongst non-smokers had increased during the same time what would the reaction of these same people be to someone standing up and saying something like "Non-smokers suffer more heart attacks after smoking ban therefore the ban should be abolished"? You can bet you're bottom dollar that they would deride such conclusions as dangerous statisical correlation masquerading as causality.
I imagine, if one had a dig around here, one could easily find that since the smoking ban the amount of effluent in Scottish rivers had changed too, but would they stand up and try to claim the two were linked on the basis of numbers alone?
9 comments:
*Fewer
As Devil's Kitchen pointed out, there were also many other health "drivers" going on in Scotland at the same time.
It really is pathetic for a supposedly "scientific" person such as the Aberdeen person making such a mistake. The first thing we learnt in statistics classes at school was that "correlation does not imply causation".
Of course you're right about statistics and causality but I think this is another media misrepresenting science story. John Ayres says "it suggests" which is good, balanced comments from a scientist. He doesn't say it proves because that would require proper research of the biochemical changes in many test subjects. I assume the researchers looked at a host of other changes to see if their is another suggested correlation but we'll have to wait for the full article when it is published. Professor Pell does make some rather conclusive statements that I would wince at.
I understand that the incidence of heart disease has been declining for decades. My mother-in-law thinks the original rise may have had to do with the import of Spam and hard margarine in WW2. We now know that transfats stick to your arteries in a way that butter doesn't.
Have these people never heard of "confounding factors"?
Wow! That was quick! The smoking ban only came into effect a couple of months ago, and already people aren't suffering from heart disease any more! Neat!
Dangerous though it is to draw inferences, if the figures suggest anything it is that smoke inhalation triggers heart attacks in people who already have heart problems. I don't recall this being an issue when the ban was imposed.
The statistic I hate the most is the '50% increase' in risk if you do this or that. But 50% of what?
It initially provides a scare story, but often, when analysed, the actual increased risk works out at a small fraction of 1%.
Verity - our Caledonian neighbours have laboured under a smoking ban longer than the English.
If the number of heart attacks amongst non-smokers had increased during the same time what would the reaction of these same people be to someone standing up and saying something like "Non-smokers suffer more heart attacks after smoking ban therefore the ban should be abolished"? You can bet you're bottom dollar that they would deride such conclusions as dangerous statisical correlation masquerading as causality.
Just because the statistics agree with your viewpoint doesn't necessarily mean that they're wrong. I haven't looked at the study in any more detail than what you cover here, but there is at least some other logical reasons to back up their assertion - even if we do not have anything other than correlation to pack it up.
Post a Comment