Stanton Glantz's Margin of Error


Stanton Glantz has written a response to my criticism of his study claiming that Helena, Montana's smoking ban cut heart attacks in half. Among other things, he says the study was not really that small, because "there were over 500 cases included in the analysis." He seems to be referring to the total number of heart attacks in Helena and surrounding areas from 1998 through 2002. But the outcome measure was heart attacks per month in Helena, a very small, highly variable number that ranged from 1 to 13 during the study period.

In response to my suggestion that the effect he attributes to the ban is preposterously large, Glantz says, "The individual risk of heart attack associated with passive smoking is about 30%, which is within the 'margin or error' of the 60% drop we saw in Helena." The 30 percent figure is the increase in risk associated with long-term secondhand smoke exposure in some epidemiological studies. It is itself implausibly high–about one-third the increased risk from smoking, which involves much higher levels of exposure–and it tells us nothing about immediate effects like those Glantz is claiming.

In any case, even completely eliminating secondhand smoke could reduce heart attacks by 60 percent only if exposure to secondhand smoke more than doubled the risk–an increase of 150 percent, compared to the 30 percent figure from the studies to which Glantz alludes. That does seem to be a pretty big difference, no matter what Glantz's margin of error was. And it implies that exposure to secondhand smoke is more dangerous than smoking.

Perhaps realizing how ridiculous and inconsistent this claim is, Glantz hedges by arguing that "when you make workplaces smokefree, many smokers quit or cut down, which reduces their risks of heart attack." But as I pointed out in my column, this effect could not possibly account for a drop of the magnitude Glantz is claiming, even if every smoker in Helena quit.