Thursday, March 17, 2005

A Rank secondhand analysis of secondhand smoke

Posted by Craig Westover | 12:05 PM |  

I will be the first to admit it’s a fine line easily blurred where science plausibility becomes fact. I have been wrestling with that issue as I explore the growing body of scientific evidence supporting a connection between the mercury content in childhood vaccines and the epidemic increase in autism and childhood neurological diseases.

Despite knee-jerk criticism from physicians that for the sake of their patients ought to be more responsible in their criticism, as does David Kirby in his book Evidence of Harm,” I make a good faith effort not to push the growing body of scientific evidence of harm beyond what it is -- evidence-- to a false certainty for the sake of propaganda. That is as unconscionable as the knee-jerk criticism that resorts to ad hominem speculation about motivations instead of examining and refuting the facts.

That being said, one can possibly understand why I find columns such as that by Brian Rank in today’s Pioneer Press (Statewide smoking ban is a public health priority) so galling. It is not his premise; while I disagree with his solution of the necessity of government intervention, I am willing to debate the scientific plausibility of some of his contentions.

No, what galls me about this Rank column is the way he distorts his science and extends it beyond the bounds of truth. He trots out the criticism-worn Helena, Mont., heart attack study -- calling it a “landmark” study -- without any context, without any discussion, using it solely as a scare factor.

Were a politician to do this, one with legislation to promote, an agenda to serve, an election to win, it’s sloppy but understandable as an act of ignorance. But Mr. Rank is not a politician. He is medical director of the HealthPartners Medical Group and chair of the Cancer Plan Minnesota Steering Committee. He is in a position of authority and credibility. Given his position of power, Mr. Rank’s abuse of science is as inexcusable as it is inaccurate. To wit: he writes --

In April, the Centers for Disease Control advised anyone with heart disease to avoid indoor settings where smoking is allowed. The CDC issued the advisory after a landmark study revealed evidence that even short-term exposure to secondhand smoke can trigger heart attacks. The study, published in the British Medical Journal, found the number of heart attacks reported in Helena, Mont., fell by 40 percent during a six-month period in 2002 when the city's comprehensive smoke-free law was in effect.
The paper referenced is “Reduced incidence of admissions for myocardial infarction associated with public smoking ban: before and after study,” by Richard P. Sargent and Robert M. Shepard, attending physicians at St. Peter’s Community Hospital in Helena, Montana. A third author is Stanton A. Glantz, professor of medicine at the University of California in San Francisco.

Without delving deeply into scientific methodology, which can be found here for those interested), the rationale for this study was that St. Peter’s is the only hospital in the Helena area and therefore conducting a study of patients admitted for specific medically defined set symptoms that we layman lump under the heading “heart attack,” would yield data abut the affect of the smoking ban instituted in Helena in June of 2002 (ruled unconstitutional and discontinued Nov 2002). As Mr. Rank states, among the results was a 40 percent decrease in the number of “heart attacks.”

Again, let’s be clear. What is galling about Mr. Rank’s designation of this study as “landmark” and the conclusiveness with which he asserts those findings far outstrip the extent and validity of the study, which again is inexcusable for a man in Mr. Rank’s position.

Consider these comments, also appearing in the British Medical Journal , from Terry F. Pechacek, associate director for science and Stephen Babb, coordinator, secondhand smoke work group of the Office on Smoking and Health at the Centers for Disease Control and Prevention in Atlanta.
Although the results of the study by Sargent and colleagues are consistent with the literature on the risks of acute myocardial infarction associated with secondhand smoke, the study has some important limitations. Among those limitations, Pechacek and Babb not that the study contains no data on actual exposures to secondhand smoke among residents or cases. Despite reduced exposure created by the law --

Some proportion of non-smokers would still havebeen exposed in their homes, cars, or other enclosed places not covered by the ordinance. Thus, without more data, the proportion of non-smokers in Helena among whom exposures were significantly reduced during the six months that the ordinance was in effect cannot be known.

A second concern is that the geographical isolation of the city, while making this type of study feasible, also resulted in a small number of admissions for acute myocardial infarction. As reported elsewhere, the typical number of acute myocardial infarction events per month before the ordinance was only about six or seven and was highly variable, with the actual number per month ranging from none to about 10-12.

Although conservative statistical analyses were applied to these data, due to the small number of events and the lack of data on changes in active smoking, random variation and factors other than secondhand smoke exposure may have contributed to the findings.

Finally, the observed effect (a decline of an average of 16 admissions for acute myocardial infarction for a six month period)was substantially greater than what might be expected [using statisticalprediction techniques]. Even assuming that the proportion of acute myocardialinfarction cases among smokers was fairly constant across time, that allnon-smokers were frequently exposed to secondhand smoke in public places, that virtually all this exposure was eliminated by the ordinance, and that allcoronary heart disease risk related to this exposure was immediately reversed among non-smokers the maximum impact on admissions for acute myocardial infarction would be predicted to be about 18-19% during the six months that the ordinance was in effect. Taking all of the above assumptions and issues into consideration, a more conservative estimate of the predicted reduction in acute myocardial infarction events might be 10-15%. [Not 40 percent.]

The small number of acute myocardial infarction events in this study produced a wide 95% confidence interval in the analysis that includes the conservative estimate of a 10-15% reduction. The width of the confidence interval underscores the importance of additional, larger studies that could replicate the findings of the Helena study and provide more stable estimates of the effect size.

But nonetheless, Mr. Rank chooses to hail this study as “landmark” and use it as scare tactic, something even those with the same objective as Mr. Rank, but more integrity, cannot abide.

Also from the British Medical Journal is this response to the Sargent et. al. study from a New York epidemiologist that values the integrity of his profession more than the temptation to misuse scientific data.
As a cancer researcher who has published extensively on the harmful effects of smoking, I am in favor of vigorous smoking bans and feel there is no justification for nonsmokers to have to breathe air polluted with tobacco smoke. However, the study by Sargent et al. claiming that the 6 -month smoking ban in Helena, Montana was associated with a drop in heart attacks must be viewed with skepticism.

The authors reported that the number of heart attacks within the city of Helena dropped by 40% immediately following the initiation of the ban. They claim that this is powerful evidence for an effect of exposure to secondhand smoke on heart attacks. But if we look at their data we see just how questionable this claim is.

First, the researchers only had information from hospital records on where a person lived. They did not interview the patients, so they had no information on whether their exposure to secondhand smoke changed as a result of the ban. They also did not present any information on whether smoking habits were affected by the ban. The fact that they had no information on exposure is a major deficiency.

Second, the drop in heart attacks is based on very few cases: 4 per month on average during the ban compared to 7 per month before. Due to these small numbers the reported difference could easily be due to chance or to some uncontrolled factor. The number of heart attacks in the area outside Helena was even smaller. It should not be surprising that, given these small numbers, there are fluctuations of the magnitude seen in this study.

Finally, the “immediate effect” and its magnitude really should make anyone stop and question the connection the authors are asserting. There are few interventions in public health that have such an immediate effect. Even if all active smokers in Helena had quit smoking for at least a year, one would not expect to see such a dramatic effect. No previous epidemiologic study or community smoking cessation program has ever shown that a reduction in smoking or exposure to secondhand smoke causes an immediate decline in heart disease incidence or mortality.

A rigorous study would have involved collecting information on a population of adequate size and interviewing individuals to assess their pre-existing risk for heart disease as well as how their behavior was affected by the ban in order to make a plausible connection between the two phenomena.

The attempt to make claims about the effects of smoking bans based on this very weak ecologic study raises disturbing questions about our ability to distinguish between sound science and wishful thinking.
That last sentence bears repeating lest Mr. Rank miss the point -- there is a distinct difference between sound science and wishful thinking. A city councilman can plead ignorance when citing bad science; people of Mr. Rank’s prestige have no such excuse.

[For a rebuttal of the Sargent et. al. study from a anti-ban activist, click here. For an alternative look at the science of secondhand smoke, click here.]

Mr. Rank then goes on to use other statistical slight-of-hand by trotting out the aggregate statistics that show no overall loss of business when a smoking ban is imposed. Of course, by "loss of business" Mr. Rank means that state tax revenues remain constant. He is not concerned about the small taverns and restaurants that are put out of business by smoking bans and don't show up in aggregate statistics.

And finally, he totally ignores the central argument in the smoking ban controversy -- the fundamental right of people to engage in a legal activity on private property with the permission of the property owner. He offers no criteria for when an acknowledged personal health issue becomes a “public” health issue demanding the intervention of government and damn, I get tired of repeating myself to people that keep bringing up the same issues, ignoring the same issues and refusing to engage in any reasonable debate.

But it’s all here.