Monday, March 25, 2013

Another Study Concludes that Smoking Ban Reduced Heart Attacks While Data Show No Effect

A study published online in the journal PLoS ONE concludes that a partial smoking ban in Prince Edward Island resulted in a 14%-24% decline in heart attack admissions, while the study data show that the rate of heart attacks actually increased in the five-year period following the smoking ban compared to the five-year period before the ban.

(See: Gaudreau K, Sanford CJ, Cheverie C, McClure C. The Effect of a Smoking Ban on Hospitalization Rates for Cardiovascular and Respiratory Conditions in Prince Edward Island, Canada. PLoS ONE 8(3): e56102. doi:10.1371/journal.pone.0056102.)

A partial smoking ban went into effect in Prince Edward Island on June 1, 2003. Smoking was banned in all public places and workplaces, but was still allowed in designated smoking rooms. The study compares the rate of heart attack admissions before and after the smoking ban.

It finds that: "The mean rate of acute myocardial infarctions was reduced by 5.92 cases per 100,000 person-months (P = 0.04) immediately after the smoking ban."

It concludes that: "A comprehensive smoking ban in PEI reduced the overall mean number of acute myocardial infarction admissions...".

The Rest of the Story

The study conclusion is based entirely on the abnormally low heart attack rates observed during the period 1995-1997. If one compares the trend in heart attacks from 1998-2003 (five years prior to the smoking ban) with the trend from 2003-2008 (five years after the smoking ban), one finds that the smoking ban was actually associated with an increase in the rate of heart attacks.

Here are the approximate trends in the heart attack rates in the five-year periods before and after the smoking ban:

1998-2003 (PRIOR TO BAN): -0.4% per year
2003-2008 (AFTER BAN): +1.1% per year

Thus, prior to the smoking ban, the heart attack rate was quite steady, declining slightly by about 0.4% per year. Following the ban, the heart attack rate increased by 1.1% per year.

These findings are hardly indicative of a significant and substantial (14%-24%) reducing in the heart attack rate that is attributable to the smoking ban. In fact, an examination of the data indicates that the finding reported in the paper is not at all robust and neither the data nor the analysis supports the study's conclusion.

In fact, the authors could just have easily compared the five-year periods before and after the smoking ban and concluded that the ban led to an increase in hospitalizations for acute myocardial infarction.

Incidentally, I would not have expected this policy to reduce heart attacks because it did not eliminate smoking, it restricted it to designated smoking rooms. But whether expected or not, the data simply do not support the study's conclusion that this partial smoking ban resulted in a significant and substantial decline in heart attack admissions.

At this point, it has become clear to me that there exists a strong investigator bias in favor of finding a significant effect of smoking bans on the reduction of heart attacks. So many studies have drawn conclusions that are simply not supported by the data that it is apparent that investigators want to find an effect.

I "want" there to be an effect as well, since I've devoted much of my career to promoting workplace smoking bans. But we have to remain objective and use rigorous scientific methods. In the long run, I don't think it serves us to lower the level of scientific analysis in order to be able to put out "favorable" findings. After all, our scientific integrity and the public's trust of the quality of our science are the foundation upon which pu.blic health is built.

No comments: