Tuesday, January 03, 2006

Anti-Smoking Advocates Told How to Respond to Dissent about Pueblo Findings

In a message sent to hundreds of anti-smoking advocates, a prominent tobacco control researcher has apparently provided instruction for these advocates in how to respond to my questioning of the validity of the unpublished Pueblo study, which concluded that an observed 27% decline in heart attack admissions during the 18 months following implementation of a smoking ban (compared to the 18 months before the ban) was due to the smoking ban.

Advocates are apparently being told not to engage in discussion on the topic, and are referred to an electronic response comment on the published Helena study which essentially reiterates the findings of the two papers without directly addressing or even mentioning the specific criticisms rendered.

The message states: "We have received several enquiries from people on how to respond to criticisms on various smokers' rights websites and other blogs of ... study reporting that AMI's fell in Helena. We believe that the proper forum for such a scientific discussion is through the journal that published the paper... ." Readers are then referred to (link here).

Those comments essentially reiterate the findings of the studies, but do not address the specific alternative explanations I and others have offered:

"In thinking about possible alternative explanations for the findings in our and the Pueblo study, it is important to consider all the observations:
  1. In both cities there was a substantial drop in AMI admissions when the smokefree laws went into effect.
  2. There was no such drop in AMI admissions from people from the surrounding area (and, in the case of the Pueblo study, a nearby city) who were not covered by the ordinance.
  3. There was a rebound in AMI admissions in Helena when enforcement of the law was suspended.
  4. There was no change in the underlying pattern of AMI admissions in the surrounding area when enforcement of the Helena ordinance was suspended.
None of the individuals who have questioned either our or the Pueblo study have provided a plausible alternative explanation for these facts."

The Rest of the Story

This post is actually not about the scientific findings of the Pueblo study, but since I brought it up, I'll address the above comment first.

My suggested alternative explanation for the findings in the Pueblo study is that the results are due simply to chance. In other words, that there is a random variation in the annual (in this case, 18-month) trends in heart attack admissions in Pueblo and that with simply 2 data points, it is virtually impossible to establish what the underlying random variation in heart attack admission rates in Pueblo is; thus, it is impossible to attribute an observed 27% decline from one time period to the next to the smoking ban, rather than simply to random variation in the underlying data.

For the purposes of assessing the published argument, let's assume that this alternative hypothesis is correct. What would be the consequences for each of the 4 points made by the authors?

1. In both cities there was a substantial drop in AMI admissions when the smokefree laws went into effect.

If the observed drop in heart attack admissions in Pueblo was due to random variation in the overall trend, rather than due to the smoking ban, then one would expect to see a substantial drop in heart attack admissions when the smokefree law went into effect (this point is tautological, but I want to address each point directly).

2. There was no such drop in AMI admissions from people from the surrounding area (and, in the case of the Pueblo study, a nearby city) who were not covered by the ordinance.


If the observed drop in heart attack admissions in Pueblo was due to random variation in the overall trend, rather than due to the smoking ban, then one would expect to see no such drop in heart attack admissions in a nearby city because there is, by definition, no reason why the random variation pattern in Pueblo would also be followed exactly by the comparison city. If one looks at secular trends in heart attack admissions in Pueblo vs. Colorado City, one will not see that the patterns exactly mirror each other.

3. There was a rebound in AMI admissions in Helena when enforcement of the law was suspended.


If the observed drop in heart attack admissions in Pueblo was due to random variation in the overall trend, rather than due to the smoking ban, then one would expect to see a rebound in heart attack admissions in Helena in the precise time period when enforcement of the law was suspended. By definition, if the observed decline in hospital admissions for heart attacks were just a random variation, then one would expect to see those admissions rebound after the decline. Otherwise, it would not be a random variation but a true, sustained decline in heart attack admissions.

4. There was no change in the underlying pattern of AMI admissions in the surrounding area when enforcement of the Helena ordinance was suspended.

If the observed drop in heart attack admissions in Helena was due to random variation in the overall trend, rather than due to the smoking ban, then one would expect to see no change in the underlying pattern of heart attack admissions in the surrounding area. If the observed decline in Helena were due to random variation in those trends, then there is no reason to expect to see a particular change in heart attack admissions in the surrounding area. A corresponding decline in admissions in the surrounding area would signal a true secular decline in heart attack incidence, but that is not what is being posited in my argument. Instead, I'm simply positing that there is random variation in the Helena trends; one would not expect to see a similar "random" variation in the trends in the surrounding area.

In other words, each of the 4 pieces of evidence purported by the authors of this comment to be incompatible with my alternative explanation for the study findings is not only compatible with the study findings, but is, in fact, the precise finding that would be expected under my alternative hypothesis.

So my alternative hypothesis may well be wrong, but it is certainly not negated by the argument provided by this particular comment.

The Rest of the Rest of the Story

Now to the main point of this post.

There are actually two.

First, I think it is interesting to observe how this story is framed in terms of "how to respond" to my commentary, rather than on a discussion of the actual scientific issues involved. In other words, it appears that what the inquiring advocates wanted to know or what they were told was not the scientific validity of the comments in question, but instead, how to respond to these comments in order to dismiss them.

If someone from within the tobacco control movement wrote a critique of a study I had conducted, I doubt that I would send a message out to advocates telling them how to respond to that dissent from within the movement. I might well defend my study and my conclusions, but I am not in the business, nor do I think we should be in the business, of instructing people on how to respond.

Isn't there some room for advocates and organizations to analyze the scientific issues on their own and come up with an informed opinion on the issue? Do we not think for ourselves anymore? Are we simply automatons that merely follow directions rendered from above? Must we always assume that any conclusions suggested by a researcher that tend to support the anti-smoking agenda must automatically be correct and that any dissent from those opinions must be counteracted, rather than acknowledged as potentially credible?

Second, I think it is interesting to observe how this message sent to hundreds of anti-smoking advocates aims to stifle discussion on the issue, rather than promote it, by suggesting that any and all discussion of the issue must be done only in one form: through the British Medical Journal.

I didn't realize that the British Medical Journal was the great gatekeeper for all scientific discussion of tobacco control issues. Nor did I realize that it is simply inappropriate to discuss the rationale for tobacco control policies or the science behind these policies in any forum other than a medical journal. I didn't realize that scientific discussion of scientific findings relevant to policy issues in a public forum was simply inappropriate and not to be tolerated.

That's interesting, because in all the years I spent responding to criticisms of my own research showing the hazards of secondhand smoke for restaurant workers (and this is the science that actually, I believe, provides support for smoke-free policies, not the alleged effects on heart attack rates), I was never once told that it was inappropriate to discuss the issues outside of JAMA, the journal in which I published my findings. The only thing that seems to have changed is that I'm now on "the other side" of the issue, if one accepts that there is such thing as an established anti-smoking agenda that one can be on "the other side" of.

Actually, I think it is valuable to be able to engage a wide audience of people in a meaningful discussion of this scientific issue, and I don't see any reason why that discussion must or should be confined to the electronic response system of the British Medical Journal.

Moreover, if the premise of this message sent to advocates is correct, then the Pueblo study should never have been reported publicly, because it has not been published in any peer-reviewed, scientific journal. If scientific journals are the only appropriate forum for discussion of science, then there is no excuse for the press releases that were issued publicizing the results of the Pueblo study.

And furthermore, the Pueblo study was not published in any journal. So if it is true that the only appropriate forum to discuss the scientific issues raised by Pueblo is in the journal in which the study was published, then I guess I'm out of luck. There would simply be no way to express my views until (and if) the study finally appears in a journal (and there's no guarantee that it will even make it through the peer review process given its serious flaws).

Well perhaps that's the point. Perhaps the idea is that I shouldn't be expressing my opinion, since it apparently represents dissent from the "established" views of the anti-smoking movement.

OK - my bad.

No comments: