CDC-Commissioned Report Says More Evidence Is Needed to Decide Whether Smoking Bans Are Good

Just kidding. The report, which an Institute of Medicine committee issued yesterday, concludes, per the press release, that "smoking bans reduce the risk of heart attacks associated with secondhand smoke." The committee's chairwoman, Lynn Goldman, a professor of environmental health sciences at Johns Hopkins, sums up the report's findings this way:

It's clear that smoking bans work. Bans reduce the risks of heart attack in nonsmokers as well as smokers. Further research could explain in greater detail how great the effect is for each of these groups and how secondhand smoke produces its toxic effects. However, there is no question that smoking bans have a positive health effect.

As with the 2006 surgeon general's report on secondhand smoke, the press release goes farther than the report itself, which in turn draws conclusions that are not justified by the evidence it presents. The judgment about the effectiveness of smoking bans is based on 11 studies that looked at heart attack rates in eight jurisdictions after smoking bans took effect. "None was designed to test the hypothesis that secondhand-smoke exposure causes cardiovascular disease or acute coronary events," the report concedes. Furthermore, "only two of the studies distinguished between reductions in heart attacks suffered by smokers versus nonsmokers." Even so, the report concludes that smoking bans reduce heart attacks, at least partly by reducing nonsmokers' exposure to secondhand smoke.

To accomplish that impressive feat, the report underplays two major problems with attempts to measure the impact of smoking bans through observational studies. First, in recent decades there has been a general decline in heart attack rates, driven mainly by improvements in preventive medication and treatment. A decrease in heart attacks seen after a smoking ban takes effect could be part of this pre-existing trend. One way to address that possibility is to use comparison jurisdictions that do not have smoking bans but are otherwise similar, a precaution most of these studies did not take. 

Second, random variation means that some jurisdictions with smoking bans are bound to see significant drops in heart attacks purely by chance, while others will see no real change or significant increases. The largest study of the issue, which used nationwide data instead of looking at cherry-picked communities, found that smoking bans "are not associated with statistically significant short-term declines in mortality or hospital admissions for myocardial infarction or other diseases." Furthermore, "An analysis simulating smaller studies using subsamples reveals that large short-term increases in myocardial infarction incidence following a workplace ban are as common as the large decreases reported in the published literature."

That study, published by the National Bureau of Economic Research in March, suggests that publication bias can explain what the IOM panel describes as the "consistent" results of the studies it considered (meaning that they all found drops in heart attacks, although the magnitude of these decreases varied widely, from 6 percent in Italy to an astonishing 47 percent in Pueblo, Colorado). If a researcher runs the numbers for a particular jurisdiction and finds no impact from a smoking ban, he is not likely to write up that result, especially if he supports smoking bans as part of the effort to reduce tobacco-related disease. Even if he does submit an article describing his findings, it is not likely to be published, not just because of an anti-smoking bias but because negative results are perceived as boring. 

The NBER paper was mysteriously excluded from the IOM report, even though the authors say they bent over backward to compensate for publication bias by looking for relevant data that did not appear in medical journals. They also ignored analyses that found no declines in heart attacks following smoking bans in California, Florida, New York, Oregon, England, Wales, and Scotland. The omission of the Scottish data is especially striking because they contradict one of the 11 studies included in the IOM report, showing that a decrease in heart attacks during the first year was exaggerated and in any case disappeared the following year.

The report is slippery in addressing the biological plausibility of attributing immediate, dramatic reductions in heart attacks to smoking bans.  For example, the study of Helena, Montana (the one that started it all) found a 40 percent drop in heart attacks within the first six months of the city's smoking ban. As I've said before, it is simply not possible that the smoking ban caused that big a drop so quickly, unless the CDC and the American Heart Association are wildly off in their estimates of the heart disease caused by smoking and secondhand smoke, respectively. If smoking and secondhand smoke together account for about 25 percent of all heart disease deaths, a smoking ban could not cause a 40 percent reduction in heart attacks even if everyone in the city immediately stopped smoking. 

For some reason (a convenient error, I assume), the IOM report describes the utterly implausible 40 percent reduction in Helena as a 16 percent reduction, as part of an attempt to show that bigger declines were found in the studies with longer follow-up periods. (Most of the studies had follow-up periods of a year or less.) What the comparison of studies actually shows is that the most dramatic results were found in the studies with the tiniest samples (e.g., Helena , Pueblo, and Monroe County, Indiana), where random variation would be expected to have the biggest impact.

Although the authors work hard to make patently ridiculous claims seem plausible, they never settle on an explanation of how, exactly, these reductions in heart attacks are accomplished. When people stop smoking, their heart attack risk declines gradually over several years, and it's hard to see why the heart attack risk in people exposed to secondhand smoke would fall any faster. Given a lack of information about individual exposure patterns in studies of secondhand smoke and heart attacks, the report says, "the committee could not determine whether acute exposures were triggering acute coronary events, chronic exposures were causing chronic damage that eventually resulted in acute coronary events, or a combination of chronic damage and an acute-exposure trigger led to the increased risk of acute coronary events." But if the impact of a smoking ban is seen within six months (or two, as in one of the Italian studies), the only possible explanation, since heart disease takes years to develop, is that the ban prevents heart attacks in people who would otherwise keel over from acute exposure.

Who are these people? The report acknowledges that people with severe heart disease are most vulnerable, but it also intimates that perfectly healthy people might drop dead from a heart attack after spending a half-hour in a smoky bar. "The risk of acute coronary events is likely to be increased if a person has preexisting heart disease," it says, implying that even people without pre-existing heart disease take their lives in their hands by grabbing a drink in a bar where smoking is allowed. At the same time, the report concedes "there is no direct evidence [as opposed to suggestive laboratory results] that a relatively brief exposure to secondhand smoke can precipitate an acute coronary event." No kidding. Not to put too fine a point on it, but where are the bodies? The sort of immediate effect they are talking about should be readily apparent. Yet the committee cites no reports of people, whether sick or healthy, having heart attacks after brief exposures to secondhand smoke.

Assuming that smoking bans do reduce heart attacks, that result could be due to declines in smoking, declines in secondhand smoke exposure, or some combination of the two. The report settles on that last explanation, even though only two of the 11 studies bothered to distinguish between smokers and nonsmokers. The authors do concede that, given the paucity of the data, "it is not possible to determine the magnitude of the effect that is attributable to changes in nonsmokers compared with smokers."

Likewise, because of the limitations of the studies and the wide variations between them, the report says, "the committee has little confidence in the magnitude of the effects and, therefore, thought it inappropriate to attempt to estimate an effect size." Michael Siegel, a Boston University public health professor who was one of the report's reviewers, notes that "if you can't even estimate the magnitude of an effect—if you have no confidence in even providing an estimate—you are hardly in a position to conclude that there is a significant effect of smoking bans on heart attacks, an effect which exceeds random variation combined with the known secular decline in heart attack rates." Siegel faults the authors for a "sensationalistic" approach, especially as reflected in their attempt to "scare people into thinking that they could drop dead from a heart attack from a brief tobacco smoke exposure."

The main goal of this project, which was commissioned by the CDC, seems  to have been producing a document that could be waved around at city council meetings and state legislative hearings. If so, the authors have succeeded. "The evidence is clear," says CDC Director Thomas Frieden. "Smoke-free laws…prevent heart attacks in nonsmokers." Although the IOM panel said it would be reckless to estimate the impact of smoking bans on heart attacks, Frieden (an avid smoking ban advocate when he was New York City's health commissioner) is willing to go there. "These findings suggest that tens of thousands of heart attacks could be prevented each year," he claims. "States and communities that do not have comprehensive smoke-free laws could have significant cardiovascular health benefits by doing so." Neal Benowitz, a member of the IOM panel, insists that "smoking bans need to be put in place as quickly as possible," because "the longer we wait, the more disease we are accepting."

My own view is that the scientific findings are not relevant to the policy question, which is a matter of property rights. Regardless of the hazards posed by secondhand smoke, business owners should be free to set the smoking rules on their own property, which no one is forced to enter. At the same time, I recognize that reports like this provide ammunition to advocates of smoking bans, so I am more inclined to be skeptical of claims about the impact of such laws than someone who has no problem with them would be. (I do agree that smoking bans, to the extent that they encourage smokers to quit, can be expected to gradually reduce heart disease over the long term, as opposed to dramatically reducing it immediately.) Yet the people conducting these reviews are not neutral observers either; as reflected by Benowitz's comments, they are committed partisans in the push to extend strict smoking bans across the country.