Thursday, February 28, 2008

More Science, or Mythology, by Press Release: French Smoking Ban Cuts Heart Attacks by 15%

According to a new, unpublished and apparently unreleased study whose results were issued via press release, the smoking ban in France has resulted in a 15% decline in emergency room heart attack admissions in the first year since the ban was implemented in January 2007. The press release, which also appeared on other news services, reports that there was a "striking" 15% decline in heart attack admissions in France in 2007, which it attributes to the implementation of a smoking ban, which was reported to have reduced secondhand smoke levels in restaurants by more than 35%, but not to have affected active smoking rates. The story is also reported elsewhere (example 1; example 2; example 3; example 4; example 5; example 6; example 7).

A detailed search of all of the articles in which the story was reported, including the press release itself and the websites of the organization which released it (the European Society of Cardiology), failed to find any link to the actual study.

No other details about the study were released, including:
  • how far back the trends in heart attacks were examined;
  • what control population was used to determine whether the observed trends were different from trends in other countries;
  • what other variables related to heart attack incidence were controlled for in the analysis;
  • what type of analysis was performed;
  • how secular trends in heart attack incidence were controlled for; and
  • how the researchers could ensure that the observed changes were not due to other factors, such as the increased use of statin drugs and better management of patients with heart disease.
The Rest of the Story

This is really getting ridiculous, and out of control. The days of tobacco control science appear to be over. Now, it appears to be just a public relations game. This appears to be the strategy:

1. Where a smoking ban has gone into effect, conduct a quick, shoddy analysis of heart attack trends.
2. Since heart attack rates are declining almost universally, you are pretty much assured of finding a decline in heart attacks after the implementation of the smoking ban.
3. Release your finding and your conclusion immediately via press release.
4. Do not bother to wait until your results have been peer reviewed or published before announcing your conclusions to the public.
5. Do not make your methods and results readily available for public scrutiny.
6. If it is later discovered and pointed out that your conclusion was invalid because you failed to account for underlying secular trends in heart attacks or random variation in the underlying trend, or if later analyses fail to bear out your conclusions, it won't matter because the results have already been widely disseminated and have become part of the popular knowledge about the topic.

I would point out that based on my long experience in tobacco control, this was not previously the modus operandi. In the "old" days, we were very hesitant to release findings to the public before the study had been peer reviewed. We went to great lengths to hold off on releasing scientific conclusions to the public until a paper was published. In the rare instances where results were released prior to publication, we made sure to make the entire study available so that the validity of the conclusions could be judged. It was rare, if not unheard of, to release study results only by press release, with the study itself and its methods cloaked in a veil of secrecy.

Now, public relations by press release is what passes for science in tobacco control.

This is truly sad for me to watch because it represents the deterioration of the scientific integrity of the tobacco control movement. It's very difficult to realize that I am a part of this (even though I am distancing myself from these tactics) and to maintain any good conscience about it.

Let me emphasize that I am arguing here not only that this is shoddy science but that the tactics are unethical. I do not believe it is appropriate to conduct science by press release. In general, conclusions like these should not be released publicly until a study is peer reviewed and published. In circumstances where the results are so important that one cannot wait until publication, one is, I believe, under an ethical obligation to make the full study available for public review and scrutiny. Otherwise, one is doing a great disservice to the public.

The real danger of science by press release is that once a conclusion is disseminated, it cannot be retracted or changed if it is subsequently found to be untrue. In addition, it is an inappropriate tactic because it essentially means that an organization can release any conclusion it wants, without having to worry about that conclusion being subject to public scrutiny.

Can you imagine if the tobacco industry practiced this strategy? We would be attacking the companies relentlessly. They actually know better than to sink to such a level. But for some reason, tobacco control organizations can get away with it. Presumably, if your cause is a good one, I guess you no longer are subject to basic standards of scientific ethics and integrity.

In terms of the scientific conclusion itself, it is simply not plausible given the limited data presented. Since smoking rates reportedly did not decline, then the study is essentially concluding that a 35% reduction in secondhand smoke levels in restaurants resulted in a 15% decline in heart attack admissions within one year. That's scientifically implausible.

The fact that secondhand smoke levels declined by only about 35% in restaurants seems to suggest that the smoking ban was quite ineffective. Compliance has apparently been poor. Data from Boston, for example, indicates that there was greater than a 95% reduction in secondhand smoke exposure following the implementation of a bar smoking ban. So a 35% reduction is hardly something to write home about (or issue a press release about).

The scientific truth is that the observed decline in heart attacks in France almost certainly reflects a secular change in acute cardiac events that is not attributable to the smoking ban, but to other factors.

Instead of spending their time taking credit for a decline in heart attacks which has little to do with the smoking ban, I think public health officials in France would be better served to figure out why the smoking ban is not working as well as it should be, and spend some time trying to actually enforce the law, rather than bragging about it.

No comments: