Tuesday, 16 October 2012

Smoke and submarines

A nice example of how to mislead with statistics at the quasi-journal Tobacco Control. It looks at smoking in US naval submarines which was banned on 31 December 2010 largely thanks to this study. Its conclusion is:

This study provides evidence that non-tobacco using submariners were exposed to SHS. Exposure was seen in all submarine classes and was not limited to personnel working in proximity to the smoking area. The existing policy was inadequate to protect non-smokers from exposure to SHS and required revision. As a result of a policy review, informed by this study, smoking below decks was banned aboard all US Navy submarines effective 31 December 2010.

Although policy-makers felt no need for the study to be published before they acted on it, it's worth seeing how robust the evidence was for a ban that had a pretty profound effect on submarine crew—nipping outside for a smoke not being an option. Prior to the ban, smoking was confined to ventilated rooms (the whole sub is ventilated, obviously) and a previous study had found that "passive smoke exposure appears to be minimum [sic]".

This new study looked at cotinine levels in urine. Cotinine is a (harmless) biomarker for nicotine and, although cotinine can be found elsewhere (such as in nightshade vegetables), it is a reasonable proxy for tobacco exposure. The study's headline figure is that cotinine levels doubled once the crew were in the submarine compared to their levels when in port.

Overall, deployed cotinine levels were 2.1 times the in port levels in non-tobacco using submariners

To be precise—for the devil is in the detail here—the average background level of urinary cotinine was 0.33 ng/ml and the average level found once in the submarine was 0.69 ng/ml. That is a doubling. No doubt about it. But is it of any clinical significance? Let's look at what typical urinary cotinine levels are in smokers and nonsmokers.

Wall (1988) found a level of 6.0 ng/ml in unexposed nonsmokers, rising to 9.2 ng/ml for 'exposed' nonsmokers. Thompson (1990) found a level of 4.4 ng/ml in nonsmokers who lived with nonsmokers and a level of 11.4 ng/ml for those who lived with smokers. Repace (2005) found levels of around 3 ng/ml for nonsmokers, rising to around 8 ng/ml after 6 hours in a smoky bar.

All these figure are considerably higher than both the 'unexposed' and 'exposed' members of submarine crews today, presumably because the background level of smoke exposure has fallen sharply in the last 25 years due to smoking bans and the decline in the number of smokers. Levels of urinary cotinine in 'exposed' submarine crew are a fraction of those found in 'unexposed'—and therefore, by implication, 'safe'—nonsmokers a few years ago.

But it is only when we compare these levels with those found in smokers that we really see the figures in context. Wall (1988) found levels of 646.8 ng/ml in smokers of 10 or fewer cigarettes per day. For heavier smokers, the level was 1,100.7 ng/ml. Thompson (1990) found an average level of 1,623 ng/ml in smokers. Holl found an average level of 2,671 ng/ml.

Put in this context, a "doubling" in "exposure" from 0.33 ng/ml to 0.69 ng/ml looks like the meaningless finding it is. I have attempted before to show cotinine levels in smokers and passive smokers in the form of a graph. Alas, you need very good eyesight to see the bar for passive smokers, such is it dwarfed by the bar for smokers (take the optical challenge here). If we take Thompson's figure as a median estimate, when in port, the submariners have cotinine levels that are 0.02% of a smoker's. When in the submarine, this rises to—wait for it—0.04%.

If 0.33 ng/ml is a safe level, why isn't 0.69 ng/ml? Both indicate 'smoke exposure' that is less than 1/2000th of that associated with being a smoker. The authors of the study feel no urge to explain why this should be a health concern. Instead they defer to a scientifically insupportable comment made by the last Surgeon General in a press conference.

This study did not attempt to quantify risk; as the US Surgeon General advises, there is no safe level of exposure to SHS.

There you have it. No need for biological mechanisms. No need for epidemiology. A ridiculous comment from a glorified GP is enough for us to assume that trace levels of a harmless biomarker represent risk. This is the natural, and quite predictable, outcome of the "no safe level" woo. Once again, anti-smoking fanatics are doing the hokey-cokey on the grave of science.

There is, however, a much more reasonable conclusion that could be drawn from this study. Submarines are the very definition of an enclosed space and yet they have ventilation systems that reduce secondhand smoke exposure to a level that is, for all practical purposes, zero.

So, if a submarine can do it, why can't a pub?