The graph shows hospital admissions for acute coronary syndrome in Kanawha County, West Virginia. It's pretty unexciting, I think you'll agree. As in many parts of the world, including England, heart disease is declining at a steady but constant rate.
Can you guess when the smoking ban started? Well, actually there were two. A full smoking ban, including bars, was enacted in January 2008. That doesn't seem to have had any effect, although it is right at the end of the timeline. Prior to that, there was what the authors of this paper call a 'comprehensive' ban that included restaurants and all public places, but not bars. That started in January 2004.
Can you see how the 2004 ban affected the heart attack rate? That's right. It didn't. And the authors* admit as much...
We did not find additional significant change between, before, and after the removal of smoking areas in restaurants (the key change in the [Clean Indoor Air Regulation] revision that took effect January 1, 2004) after accounting for the sustainable decline of ACS hospitalizations since the 2000 regulation revision.
So how is this evidence of smoking bans reducing hospital admissions for acute coronary syndrome?
To answer that, we need to go all the way back to 1995, when Kanawha County introduced what the authors admit was a 'modest smoking regulation'. Modest indeed, it covered public buildings, but not bars or restaurants, or—I dare say—quite a few other places. Restaurants were, however, obliged to have a designated nonsmoking section of 50% of the floor space.
This is not what anyone today—let alone people in tobacco control—would describe as a smoking ban, but then in 1995, even California didn't have a full smoking ban.
But what—I hear you cry—has the 'modest smoking regulation' of 1995 got to do with the heart attack rate of 2000-08? If you're a true student of voodoo science you'll pay attention, because this is a belter...
In conclusion, our results demonstrate that from 2000 through 2008, the rate of hospital admissions for ACS has consistently declined in Kanawha County in the presence of an existing [Clean Indoor Air Regulation].
Yes, you read that right. Heart attacks declined "in the presence of" a very mild smoking regulation that came into force five years before the timeline of the graph begins. Therefore, post hoc ergo propter hoc, smoking bans reduce heart attacks.
Was the rate rising before 1995? We don't know, because they haven't bothered to get the data. Did the appearance of a real smoking ban lead to an acceleration of the decline? No it didn't. Is it true to say that the heart attack rate declined "in the presence of" absolutely anything that began prior to the year 2000? Yes it is.
This is beyond sad. This 'research' was published by the Centers for Disease Control, America's most prominent public health organisation. Don't they feel just a twinge of shame at being associated with this cow-pat of a study?
Probably not, because, as ever, godawful pseudo-science is fine so long as its for a good cause.
* One of whom is Juhua Luo, a perennial grant recipient from tobacco control who has an uncanny ability to find epidemiological associations where no one else can. In the past she has authored several studies that tried to show a link between snus and pancreatic cancer, as well as a recent attempt to link smoking to breast cancer.
Thanks to Michael McFadden for the tip.
