Behind the smokescreen

(Warning: compared to my usual posts this is a bit long and a bit grumpy)

I’m a non-smoking epidemiologist, with an academic focus on non-communicable disease. I should be in favour of reducing the public health threat of smoking, right? You bet! And I should be delighted by the BMA’s recommendations that smoking in cars should be banned? I’m not so sure…

See, it sounds like a good idea in some respects. The harms of second hand smoke are well established and they informed the legislation on banning smoking in public places. But I found myself wondering just how much harm passive smoke exposure in cars actually causes -what is the population attributable disease risk? And whether banning smoking in cars would actually do anything to reduce road accidents and lessen the burden of disease related to second hand smoke?

Media reports were lacking these details so I turned to the BMA’s 22 page publication on the matter.  And I’m still none the wiser. At best, it’s a non-systematic review in which any actual quantitative data on the effects of in-vehicle smoking on health has not been presented and/or is referenced in an oblique fashion (by referencing papers that themselves are referencing the original sources). There are two main lines of argument put forward:

(i)  Passive smoke inhalation is harmful to health, and car travel with a smoker presents a particular hazard in this regard.

(ii) The act of smoking while driving poses a risk of distraction which may cause road accidents.

Let’s unpack these arguments and examine the evidence base presented.

Starting with the section “Smoking in vehicles: health impacts”:

“…studies demonstrate that the concentration of toxins in a smoke-filled vehicle is 23 times greater than that of a smoky bar, even under realistic ventilation conditions.”

One of the sources for this figure makes the comparison between a car and a house, not a car and a bar, but is not itself the original source of that statistic. In fact the “23 times greater” stat seems to have been artfully debunked here but nonetheless persists in the literature.

Next:

 “In England an estimated 30 per cent of smokers smoke in their vehicles, and over half of all journeys made by children aged 16 and under are by private vehicle. It is likely that private vehicles are a significant source of exposure to [second hand smoke] in children.”

So, we have an estimate for prevalence of in-car smoking, and an estimate of frequency of car travel by the under 16s, but no estimate of the proportion of car journeys undertaken by children in which someone is smoking. Some big gaps here. And what does ‘significant’ mean in this context?

Later the report says:

“A study by the British Lung Foundation (BLF) showed that more than half (51 per cent) of eight to 15 year olds have been exposed to cigarette smoke when confined in a vehicle.”

The source for this is footnote in a press release. Unfortunately the methods and detailed results of the survey are only available on request so I can’t check whether those 51% of 1000 children surveyed are exposed on a daily basis or have ever been exposed to in-vehicle smoke, even if just once. So I still say the jury is out on the prevalence and magnitude of this exposure.

They make another attempt here:

“This [passive smoking exposure occurring ‘sometimes’ or ‘often’] is especially true of young people (37 per cent of 18 to 24 year olds), and of adults from lower socio-economic groups (31 per cent of those in socioeconomic groups C2DEb).”

This wording is lifted directly from this RCP report (p. 144). The numbers seem to come from a YouGov 2009 smoking survey but I can’t locate that to check how the question was asked or how many young people were surveyed. So none of the above statements are based on peer-reviewed research in the public domain.

Now for the next section: “Smoking in vehicles: road safety”

“There is some evidence that drivers who smoke are at greater risk of accidents than those who do not, with smokers about 50 per cent more likely to die in a crash.”

This references a paper which was actually an experimental study of 10 people, videoed while driving and subjected to distractions. Presumably the “50% more likely” figure comes from something referred to in the introduction, but I cannot check as the paper is in Italian. It would have helped to cite the original source of this statistic – presently it’s unclear whether the elevated mortality is due to a higher likelihood of crashing, or worse survival once involved in a crash.

“A study by Hutchens et al (2008) showed that smoking was the only unique factor among a host of associated risk factors (alcohol, marijuana, sensation seeking) with greater crash odds, independent of demographic factors and general health risk taking. The study determined that smokers were twice as likely to have had a crash as non-smokers, even after controlling for gender, race, ethnicity, geography, socio-economic status, and the length of licence held.”

This references a paper in which teenagers (not all drivers) were found to be twice as likely to have had a crash if they were smokers. This is not the same as a risk posed by smoking while driving. Indeed the authors of this study are exploring whether risky health behaviours are correlated with risky driving behaviours. As far as I can tell from a quick skim, at no point do the authors suggest that smoking while driving predisposes to accidents, rather that it is something about the risk behaviours of teenage smokers that put them at elevated risk of road accidents.

“Elliot et al (2006) found cigarette use was positively and significantly associated with traffic incidences – including crash incidence – among young drivers.”

Again, this is a study of correlations between risky health behaviours and risky driving behaviours.

“A 1996 study by Lang et al had shown cigarette use to be a key predictor of single vehicle crashes for females during the first two years of driving.”

This is another study of risk behaviours in adolescents.The main message I’m getting so far is that we shouldn’t allow teenage smokers to drive.

“It is important to note that the Highway Code does not make it a specific offence to smoke while driving, any more than it is currently an offence to use audio or navigation systems, or eat while driving. …Alongside the evidence base highlighting that smoking and [second hand smoke] in a vehicle is harmful to health, the BMA policy to ban smoking in vehicles would help to ensure that drivers are not distracted by the act of smoking while driving and this may reduce trauma, health service usage and deaths due to road traffic accidents, although further research is required.”

In all there are a number of gaps between the presented evidence and the postulated effects on health. The authors do concede that “further research is required” with respect to the road safety issue but do not acknowledge the limitation that this places on the strength of their recommendations. There is a lot of supposition about the magnitude of exposure to second hand smoke in vehicles and the likely consequences, without any proposal to quantify this in cohort studies or other research. The evidence that is presented is misleading in places.

The idea that smoking in confined spaces like cars poses a respiratory risk for non-smoking passengers, and may cause an additional distraction hazard is definitely plausible. But in the absence of quantitative estimates of the harms of in-vehicle smoking and the possible benefits of removing this hazard, I think the BMA have overstepped the mark by calling for a change to legislation. A greater benefit to child health could probably be had by legislating against transporting children over 10 by car for distances under one mile: think of the cardiovascular benefit, the reduction in vehicle emissions and reduced number of road vehicle crashes (although probably tempered by the unfortunate hazards of being a pedestrian/cyclist). But as I don’t have the numbers to back that up I’m not going to make a formal recommendation and send out a press release.

Let me be clear – I have no pro-smoking agenda. The only agenda I want to push is one of evidence-based medicine and policy. That politicians misrepresent data to support their arguments is no surprise, but I feel deeply disappointed that an organisation like the BMA would do something as drastic as call for changes in the law when the evidence does not yet support the claimed extent of the hazard nor the likely benefits of such an intervention. Failure to base such recommendations on firm evidence undermines the credibility of public health advice and serves to give more credence to the ‘nanny state’ accusations that are levelled at far more robust recommendations.

I’d love to see smoking rates reduced to nothing and an eradication of all smoking-related illnesses, but I also want to be able to trust that the finite resources and influence that medical professionals have at their disposal are being used in the most effective way. I want to know that lobbying is based upon a solid evidence base supporting a causal relationship, not just ecological studies, surveys and postulation. I want proposals to be honest about what they are, and not pretend to be evidence-based when the evidence is thin and maybe not even the right evidence to be reporting. Is that too much to ask?

———————————————————

25th Nov: Update on this post here.