diabetes, passive smoking, and fuzzy math

Shame, NewScientist, for publishing results of a passive smoking study without even a whiff of skepticism.

US researchers examined more than 4500 men and women in Birmingham, Alabama, Chicago and Minneapolis, Minnesota, and Oakland, California.

The volunteers' health and smoking habits were first assessed in 1985, and again 15 years later. The investigators found that during this time, 22% of smokers developed glucose intolerance. This is the precursor to diabetes, when the body can no longer produce enough of the hormone insulin, which regulates blood sugar.

Those who had never smoked, but who were frequently exposed to second-hand smoke came next, with 17% developing glucose intolerance. This was higher than the 14% risk rate in the group of people who had previously smoked and then given up the habit.

Those least at risk were non-smokers who were not exposed to other people's smoke. In this group, 11.5% of people developed glucose intolerance.

Compare that to the unmentioned qualification in the paper [pdf]:

The hazard ratio for passive smoking exposure was significant only for white men, although the decreases in sample size that accompany these stratified analyses need to be considered in the interpretation of these results.

Angst is premature for other reasons. In a "rapid response" to the article, Alastair G. Browne notes, among other problems,

Moving on the results of the experiment, and specifically table four which details glucose intolerance by category. Looking particularly at the incidence precentages, in one case the incidence percentage works out higher for "never" smokers with passive smoke exposure than it does for current smokers. It would be reasonable to expect that current smokers are subject to greater levels of tobacco smoke than non-smokers of whatever category. The results do not reflect this and there is no attempt at explaining this within the text. Additionally, the majority of the figures do not show a deviation of greater than 10% between smoker categories. Such a deviation is considered the minimum acceptable in order to give meaning to experimental results. Accordingly, this paper appears to be inconclusive.

Table 4 is reproduced to the right.

Policymakers and concerned citizens are only too eager to dispense with caution and care when interpreting initial and inconclusive results. How long before this study is used to usher in further antismoking legislation?