New Study From Ireland Update
The Next Helena: New Study Concludes that Smoking Ban in Ireland Caused 15% Decline in Heart Attack Admissions
By Michael Siegel
A new study presented this week at the annual scientific symposium of the European Society of Cardiology concludes that the smoking ban in Ireland resulted in a 14.5% decline in heart attack hospital admissions during the first year after its enactment.
A Reuters article from yesterday boasts: “Heart attacks tumble after Irish smoking ban” and quotes the study authors as arguing that their research demonstrates the need for further smoking bans.
According to the article: “Ireland’s rate of heart attacks fell by around a tenth in the year following the introduction of the world’s first nationwide ban on workplace smoking, boosting the case for more similar bans, doctors said on Tuesday. Edmond Cronin and colleagues at Cork University Hospital said an analysis of people admitted with heart attacks to public hospitals in southwest Ireland showed an 11 percent fall in the year after the ban came into effect in March 2004.”
What these news articles and quotes do not tell you is that while the study authors compared the number of heart attack admissions in southwest Ireland in the year following the smoking ban with the number of heart attack admissions in the year preceding the smoking ban, they only looked back an additional nine months in order to assess the baseline trend and year-to-year variability in the number of heart attack admissions in this region.
What this means is that there is simply no way to assess whether the observed 14.5% decline in heart attack admissions from 2003 to 2004 was due to the smoking ban or if it was instead due to a pre-existing secular trend in heart attack admissions, or if the change merely reflects underlying variability in the data.
In order to establish the baseline trend in heart attack admissions over time and to assess the degree of year-to-year variability in heart attack admissions, one would need to go back in time much more than simply nine months. Otherwise, these data are virtually meaningless.
For example, the study reports that there were 1277 admissions in 2003 and 1092 admissions in 2004. Suppose that in 2001 there were also about 1280 admissions. This would be consistent with the conclusion that the smoking ban resulted in a significant decline in heart attacks.
However, suppose that in 2001 there had been 1500 heart attack admissions. Then, it would be clear that a 15% decline in heart attack admissions from one year to the next is a common occurrence and that it could not be inferred that the 15% decline from 2003 to 2004 was attributable to the smoking ban.
Complicating matters is the fact that the authors found no decline in heart attack admissions from 2004 to 2005, despite the continued presence of the smoking ban.
In fact, it turns out that there is a secular trend of sharply declining heart attack mortality in Ireland. While this doesn’t necessarily mean that heart attack incidence is declining, it does suggest that there may have been a trend of declining heart attacks in Ireland even before the smoking ban went into effect.
Cardiovascular disease mortality in Ireland has in fact been falling rapidly. Over the past 15 years, it has declined by about half. Some of this decline is due to decreased heart attack incidence. And some of that reduced incidence is due to a drop in smoking rates. All of this happened, of course, before the smoking ban went into effect.
Ischemic heart disease mortality in Ireland decreased by 8.6% in 2003, the year immediately preceding the smoking ban. In 2004, ischemic heart disease mortality declined by only 1.8%.
Does this mean that the smoking ban resulted in an increased rate of death from heart attacks than would have been expected? Of course not. You can’t simply look at a change from one year to the next and attribute it to the smoking ban. You have to carefully examine long-term secular trends, variability in the data, and other factors that affect cardiovascular disease rates.
These problems, however, did not seem to stop the authors of this study from drawing a sweeping conclusion that is completely unjustified by the data which they report. This seems to be par for the course in the tobacco control movement right now. Junk science is passing for perfectly valid science in tobacco control these days.
It will be interesting to see how long it takes for the anti-smoking groups to catch wind of these data and to start spreading these junk science claims widely to the public and policy makers. My guess: it will take only as long as it takes for secondhand smoke exposure to cause hardening of the arteries.
The Helena heart attack study examined a base population of 25,000 and reported a 40% decline, although the authors refused to divulge any numbers indicating the percent of reported decline in nonsmokers.
?
The Pueblo study, with a base population of 100,000 only reported a 27% decline, with the authors of this study evidently not even collecting data on nonsmokers despite extensive previous criticism of the Helena’s authors’ failure to provide such an analysis.
?
And then the Piedmont Region in Italy, population 3.5 million was studied and a decline of 11% was found, again without any breakdown of smokers and nonsmokers.
?
Given that if I go to the hospital with a broken tooth one of the first questions I am asked is whether I am a smoker, it always seemed odd to me that these studies of smoking bans and heart attacks were always so oddly deficient in this regard, particularly when some of their authors became so outspoken about the meaning of their results regarding secondary smoke exposure by nonsmokers.? (See:
?
http://bmj.bmjjournals.com/cgi/eletters/328/7446/977#67440
?
and
?
http://bmj.bmjjournals.com/cgi/eletters/328/7446/977#123038
?
Now finally we have a study that claims to have broken the two populations down. and has examined a total population of 625,525.?? While the Telegraph article’s lead-off sentence claims the study found a 14% drop “in Ireland”, of course this 625,000 is only a 15 % sampling of Ireland’s 4.23 million.? The authors of the study offer no rationale for why they didn’t compute the figures for the other 85% of the population, at least they didn’t in the abstract.? Perhaps they will here?
?
On the plus side, this study finally seems to have taken the brave leap of both recording AND reporting on the relative drops in smokers and nonsmokers, obviously a very important improvement for a study that will be widely used to promote the perception that secondhand smoke represents a significant danger.
?
The study abstract reports a rough reduction of 23% among smokers and 10% among nonsmokers.? As Dr. Siegel has accurately pointed out in his analysis a drop of 10% or even a drop of 23% could very well be completely meaningless in terms of any causality: if the authors had examined longer term data for that area of Ireland they might very well have found that drops and increases from year to year in heart attacks might very well normally make such jumps.? Again, it seems odd that they did not take such a simple step, but again, maybe they’ll explain their reticence here.
?
In terms of the size of decrease and the size of the population studies, this Irish study fits in perfectly with the extension of the trend findings of the study I conducted with Dave Kuneman almost two years ago, the essentials of which are available at:
?
http://kuneman.smokersclub.com/hospitaladmissions.html
?
In the expanded study submitted to the British Medical Journal and rejected primarily on the grounds that the editors did not think it “added enough, for general readers, to what is already known about smoking and health” (See the reproduction of the ACSH article at:
?
http://www.smokersclubinc.com/modules.php?name=News&file;=article&sid;=4503???
?
This seemed a bit odd, since our results were diametrically opposed to the three main previous studies that had been done and was also based on a far larger population base: 70,000,000.? However it is clearly within the province of a medical journal’s editor to make such a decision.
?
One of the points we noted in that rejected submission was that? large heart attack variations seem to be larger in smaller selected population groups and smaller in larger more generalized ones.? Helena, with the smallest population base, reported the largest drop: 40%; Pueblo, with 100,000 reported 27%; Piedmont, with 3,500,000 reported only an 11% drop.? Our own study, roughly 20 times the size of the other three studies combined, found no drop.
?
The Irish study falls perfectly in line with this trend: with a population base of 625,000 its reported drop of 14% falls quite nicely between Pueblo’s 27% and Piedmont’s 11%.? It reinforces the conclusion that in large populations it is likely that there is no drop experienced at all… that the drops are statistical artifacts unique to selected populations.
?
Without seeing the full body of the study it is difficult to say much more.? Dr. Siegel has already done an excellent job of analysis with the available information and I am sure that we are all eagerly waiting to see the rest of the details (Well, maybe not all of us: I’m quite sure that the media will jump in head over heels spurred on by press releases from antismoking organizations and authorities seeking to reinforce the threatened ban in England against the brewing revolt there.)
?
Hopefully as well we’ll see the authors respond to the two questions I raised above.
?
?
Michael J. McFadden
Author of “Dissecting Antismokers’ Brains”
Mid-Atlantic Director, Citizens Freedom Alliance, Inc.
Director, Pennsylvania Smokers’ Action Network (PASAN)
http://pasan.thetruthisalie.com/
Midwest Regional Director
Citizens Freedom Alliance, Inc.
The Smoker’s Club
http://kuneman.smokersclub.com
