Weak Link

New grounds for skepticism about secondhand smoke claims


New York Mayor Michael Bloomberg says his ban on smoking in bars and restaurants, which took effect last month, will save "literally tens of thousands of lives." Anti-tobacco activist Stanton Glantz claims such bans cut heart attack rates in half.

In a political environment where such extravagant claims are credulously accepted, it's useful to be reminded that the scientific debate about the hazards of secondhand smoke is far from settled. A study in the May 17 issue of the British Medical Journal shows once again how tricky it is to measure the effects of environmental tobacco smoke (ETS).

UCLA epidemiologist James Enstrom and State University of New York epidemiologist Geoffrey Kabat analyzed data from an American Cancer Society study that tracked 118,000 Californians from 1959 until 1998. They focused on 35,561 subjects who had never smoked and whose spouses' smoking habits were determined through a questionnaire. As is typical of research in this area, Enstrom and Kabat used marriage to a smoker as an indicator of ETS exposure, subdividing subjects based on how much their spouses smoked.

"No significant associations were found for current or former exposure to environmental tobacco smoke," they report. "The results do not support a causal relation between environmental tobacco smoke and tobacco related mortality, although they do not rule out a small effect….Given the limitations of the underlying data in this and other studies of environmental tobacco smoke and the small size of the risk, it seems premature to conclude that environmental tobacco smoke causes death from coronary heart disease and lung cancer."

Since anti-smoking activists and public health officials confidently assert annual death tolls from secondhand smoke of 50,000 or more, you may suspect that Enstrom and Kabat's findings are unusual. They are in fact similar to the results of most studies looking for a connection between ETS and lung cancer or heart disease. Such research typically finds small, statistically insignificant associations.

If you pool the data together, you can generate statistically significant results. But such meta-analyses tend to mask the weaknesses of the studies on which they're based.

As Enstrom and Kabat note, these weaknesses include the possible misclassification of current or former smokers as lifelong nonsmokers. Because smokers tend to marry smokers and are at an elevated risk of lung cancer and heart disease regardless of their ETS exposure, that kind of mistake could account for the observed associations.

Spouses also tend to share other habits—related to diet and exercise, for example—that can contribute to illness. Since smokers tend to be less health-conscious and risk-averse than nonsmokers, their spouses might be more prone to disease even if ETS had no effect. Another interesting point raised by Enstrom and Kabat is the effect of being widowed, which happens more often to spouses of smokers and is independently associated with higher death rates.

Finally, there is the problem of publication bias: Studies that find an association are more likely to see print. Hence the link between ETS exposure and fatal disease could be even weaker than it looks in the published research.

The question is not whether tobacco smoke can be dangerous. At high enough levels, it clearly is, as Enstrom and Kabat's study confirms. "As expected," they write, "there was a strong, positive dose-response relation between active cigarette smoking and deaths from coronary heart disease, lung cancer, and chronic obstructive pulmonary disease."

The question is whether there's a threshold below which tobacco smoke has no measurable impact on mortality. Based on the epidemiological data, Enstrom and Kabat calculate that smoking one cigarette a day would be associated with something like a 20 percent increase in lung cancer risk (as opposed to an increase of roughly 1,000 percent for a pack a day) and virtually no increase in heart disease risk.

"It is generally considered that exposure to environmental tobacco smoke is roughly equivalent to smoking one cigarette per day," Enstrom and Kabat note. If so, a small increase in lung cancer risk is possible, but the commonly reported 30 percent increase in heart disease risk—the purported cause of almost all the deaths attributed to secondhand smoke—is highly implausible.

Tobaccophobes such as Michael Bloomberg and Stanton Glantz are determined to eliminate smoking from bars and restaurants regardless of what the evidence shows. But those of us who have more respect for the truth should not let them banish skepticism along with cigarette smoke.