I Suppose You Think It Was Just a Coincidence!

For the last few years, anti-smoking activist Stanton Glantz has been been struggling to come up with a credible response to criticism of his claim that smoking bans cause an immediate, dramatic reduction in heart attacks. His main rejoinder so far has been to assert that the critics, while arguing that his claim is utterly implausible, have not offered an alternative explanation for the decreases in heart attack admissions that followed smoking bans in Helena, Montana, and Pueblo, Colorado. But that isn't true. The alternative explanation is that the decreases occurred by chance and had nothing to do with the smoking bans.

Glantz makes much of the fact that Helena's drop in heart attacks was statistically significant, which he says means it is "unlikely to be due to underlying random fluctuations." But if there's a 5 percent chance that the decrease was "due to underlying random fluctuations" (corresponding to a confidence level of 95 percent), that means you'd expect to see such a drop in one out of 20 cities with smoking bans, even if the bans had no impact on heart attacks. Since hundreds of municipalities in the U.S. are covered by smoking bans, there are apt to be more than a few where heart attack admissions just happened to fall after the bans were implemented.

In a recent BMJ "rapid response," University of Alabama at Birmingham epidemiologist Philip Cole and Brad Rodu, chairman of tobacco harm reduction research at the University of Louisville, present data that reinforce this alternative explanation. Instead of looking at heart attack admissions at one hospital in Helena, as Glantz and his co-authors did, Rodu and Cole examined heart attack fatalities in Montana's Lewis and Clark County, which includes Helena. They found there was no reduction in such fatalities after the smoking ban took effect, which means either there was no reduction in Helena or it was washed out by a simultaneous increase elsewhere in the county. "We conclude that the smoking ban certainly had no effect on the number of people in Lewis and Clark county who died from AMI in 2002," they write. Rodu and Cole also found that the heart attack fatality rates in Helena involved small, highly variable numbers, and that the same was true of Pueblo. Have a look at their graphs, and you can see it's not at all implausible that the reductions in heart attacks cited by Glantz occurred purely by chance--especially when no such effect is visible in much larger populations covered by smoking bans.

Meanwhile, Michael McFadden, one of the researchers who looked at heart attack admissions outside of Helena and Pueblo (the data weirdly neglected by Glantz), reports in the letter just below Rodu and Cole's that most of the "rebound" in heart attacks that supposedly occurred after Helena's smoking ban was repealed actually happened while the ban was still in effect. "While the graph indicating this was made available during the initial press release parties in 2003 and was displayed on the Internet," he writes, "the incriminating data was eventually removed from both the net and the final BMJ publication."