Monday, January 11, 2010

More Shoddy Science on Smoking Bans and Heart Attacks: Study from Switzerland Has No Comparison Group and Results Don't Support Conclusion

According to the headline of a Swiss news site -- "Smoking Ban Linked to Drop in Heart Attacks" -- a new study shows that a regional smoking ban in Switzerland led to a 22% decline in heart attacks in the Graubuenden canton where it was adopted. The study was published online in the journal Swiss Medical Weekly (see: Trachsel LD, Kuhn MU, Reinharf WH, Schulzki T, Bonetti PO. Reduced incidence of acute myocardial infarction in the first year after implementation of a public smoking ban in Graubuenden, Switzerland. Swiss Medical Weekly, January 7, 2010).

According to the news article, one of the study authors "says that while causality cannot be proven, there is a strong case for a link between the ban and the number of heart attacks, especially given the lack of other obvious culprits. 'The strongest argument for causality is that more than ten studies worldwide have come up with similar results,' he said. 'If you take all these studies together, you get an average decrease of 17 per cent in the first year after a ban is implemented.'"

The public smoking ban in Graubuenden, which prohibits smoking in all public buildings, cafes, bars, and restaurants, went into effect on March 1, 2008. The study examined the number of heart attack admissions at a large, tertiary care hospital in Graubuenden, which reported is the only hospital in the canton to perform cardiac catheterization and therefore, virtually all patients who suffer a heart attack are reportedly seen at this hospital.

The number of heart attack admissions was recorded for the first year following the smoking ban (March 2008 through February 2009) and compared to the same period during the previous two years (March 2007 through February 2008 and March 2006 through February 2007).

The number of heart attack admissions during these three periods was as follows:
2006-2007: 229
2007-2008: 242
2008-2009: 183

Thus, there was a 20% reduction in heart attacks from 2006/07-2008/09 and a 24% reduction from 2007/08-2008/09, or an average reduction of 22%.

The Rest of the Story

The primary reason why this is shoddy science is that the study lacks something very important: a comparison group.

Simply knowing that the number of heart attacks in Graubuenden declined by 22% in the first year after implementation of the smoking ban went into effect doesn't prove that the smoking ban was the reason for the observed reduction in heart attacks. It could well be that heart attacks for all of Switzerland also declined substantially during the same time period.

The only way to evaluate whether or not the observed decline of 22% was attributable to the smoking ban, rather than just indicative of a general secular trend of declining acute coronary events, would be to compare the trend in heart attacks in Graubuenden with the trend in some comparison area, such as other cantons in Switzerland, which did not have a smoking ban.

In the absence of any such comparison group, it is not appropriate (and not scientifically legitimate) to conclude that the observed decline in heart attacks is attributable to the smoking ban. There are many other factors that could have caused the decline, not the least of which is that heart attack admissions decreased substantially during the same time period in many regions throughout the world, due in part to advances in medical treatment of heart disease, hypertension, and hypercholesterolemia, and to declines in smoking that are occurring even in the absence of smoking bans.

The Study Itself Shows that Its Conclusion is Invalid

What makes this study much more problematic -- going well beyond merely being shoddy science -- is that the article draws a conclusion which is refuted by the very data presented therein.

To understand the rest of this story, you need to first appreciate the nature of the Graubuenden canton. As described in the research article: "because of its unique landscape Graubuenden is the largest holiday destination of the country, offering a multitude of recreational activities in all seasons. Hence, a total of ≈6million visitor overnight stays are recorded in Graubuenden every year. Based on the large number of visitors, the population of the Canton of Graubuenden may almost double during the holiday season."

Now here is the critical point: if the smoking ban were the cause of the decline in heart attacks, then the decrease in heart attacks should be observed only among residents of the canton of Graubuenden. One would not expect to see a substantial decline in heart attacks among non-residents, who just happen to be vacationing there for perhaps a week or even less. In a sense, the non-resident visitors to Graubuenden represent a sort of internal comparison group.

The study did examine the differences in the trend in heart attacks among Graubuenden residents compared to non-residents during the relevant time periods. Here are the results, showing the average observed decline in heart attacks in the first year following the smoking ban for residents versus non-residents:

Residents: 22.8%
Non-Residents: 20.6%

As the reader can easily see, there was absolutely no significant difference between the decline in heart attacks among residents versus non-residents. This pretty much dispels any notion that the observed decline in heart attacks was attributable to the smoking ban. In fact, it provides evidence that heart attacks among residents of other cantons in Switzerland, and among the population in general, were declining during the same time period.

Although in his quote to the press, one of the study authors stated that there are no other obvious "culprits" that could explain a decline in acute coronary events, the research paper suggests otherwise. The paper notes that: "an increase in the use of lipid-lowering drugs during the study period might represent an issue. Indeed, since 2006, sales figures of lipid-lowering drugs increased gradually by about 9% every year in the Canton of Graubuenden."

In my opinion, not only is this new study of the Helena et al. ilk invalid in its conclusions and not only is the study shoddy because of its failure to include a comparison group, but the article seems to be heavily biased and to ignore the very data reported in the paper itself. The interpretation of the study's findings appears to be quite biased towards trying to show that there was a significant effect of the smoking ban.

While I understand the source of this bias (I, too, would love to see an immediate and dramatic effect of smoking bans on heart attacks, especially since I have devoted so much of my career to promoting such policies), it is unfortunately clouding the scientific process.

The bottom line is that this study does not support the conclusion that the smoking ban led to a 22% reduction in heart attacks and in fact, it provides some evidence that any observed reduction in heart attacks was due to a secular trend of declining heart attacks that was observed in residents and non-residents alike.

No comments: