Sunday, 18 December 2016

Oppressing smokers for fun and profit

According to an article in the New England Journal of Medicine (,
Tripling tobacco tax globally would cut smoking by a third, and prevent  200 million premature deaths this century from lung cancer and other diseases.
This should, of course, be instituted immediately. It is almost the perfect public policy: self-interest dressed up as sanctimony. Not only will we make the lives of non-smokers better at the expense of smokers, but we can do so whilst telling smokers we are doing it for their own good!
The article itself reiterates well known facts about smoking and does so according to the tendentious rules of the anti-smoking lobby that ensure we can feel good about ourselves whilst trampling on freedom. Apparently
Manufacturers' worldwide profits of about $50 billion [£30 billion in real money] in 2012 (approximately $10,000 per tobacco-attributable death) yield enormous political influence that is used, among other things, to try to prevent large tax increases.
Maybe that’s true: I don’t know. What I do know is that laws regulating tobacco impose high costs of entry to the market and thereby increase tobacco company profits by restricting competition. So I would expect them to spend money on lobbying to that effect.

Still, we can all at least enjoy a jolly good hate of the disgusting profiteers who profit from tobacco-attributable death. Let’s also get clear who they are. The profit made by the UK government on UK tobacco sales was about £10 billion in 2012. I’m not sure how to scale that up to the world but it seems thereby likely that the profits of governments from tobacco sales are much greater than the profits to tobacco companies. Much of that profit is entirely wasted by the general wastefulness of government, of course. So all the otiose government employees paid by that waste are profiteers from tobacco-attributable death. Of the profit that is not wasted, it amounts to a transfer of wealth from smokers to non-smokers, all of who are also profiteers from tobacco-attributable death.

Then there is the attribution of deaths to tobacco. On this calculus if you died of lung cancer that you would not have had, had you not smoked, that entire death is attributable to smoking. Well, I see the sense in the idea but I don’t think it is correct. It would be correct if otherwise you would have been immortal, but you aren’t. The author’s say that smoking leads to ‘a reduction in life span by an average of about 10 years’.  So for the sake of argument suppose you died at 60 but would otherwise have died at 70 from cancer. That is to say, other factors that control your ability to forestall cancer would lead you to die at 70 anyway but the smoking brought that forward by ten years. In that case it seems that smoking is only responsible for 1/7th of your death.

Now that case depended on assuming death by cancer anyway and I’m not sure whether it is fair simply to generalise. But I think it does draw out an essential point: that attributing the death to smoking, and thereby attributing the cost of smoking as the loss of an entire life, is misleading. The cost is only the actual loss of life expectancy. If that cost is to be measured in terms of lives, the actual death being caused by smoking does not make the cost of smoking one whole life but rather only the fraction of life lost.

I can see that on first acquaintance this argument may appear stretched so consider this: Suppose our life expectancy was 1,000,000 years. I suggest it is obvious in this case that reducing that expectancy by 10 years would be a relatively trivial cost and that is so because it is a relatively trivial fraction of 1,000,000 years. So the real cost of smoking is only the actual years lost and if we are to measure it in terms of lives, it is the proportion of life lost.

Now if that is right, then it is misleading to present the expected benefit of tripling tobacco tax as the saving of 200 million lives. That is not the expected benefit at all: the expected benefit is saving 2,000 million years of life. And if we want to present that benefit in terms of lives we need to divide through by the life expectancy, which in any case we expect to increase over the century. Suppose, conservatively, that it is 80. Then the expected benefit of tripling tobacco tax is only 25 million lives.

Of course, that is still a lot, which takes us onto the final point: So what? So what if tripling taxes on tobacco saves 25 million lives? For that matter, so what if it is 250 million lives or 2,500 million lives?
Dr Harpal Kumar, Cancer Research UK’s chief executive, said: “there’s an urgent need for Governments to find ways to stop people starting and to help smokers give up”. No there isn’t, and there never has been. There can’t be. It is not and never has been the government’s business whether you,  I or anyone else smokes. The issue is entirely outwith the government’s legitimate concerns.

You might think the government is right to discourage behaviours that reduce life expectancy and that this justifies taxing smokers.  But in that case, why pick on smoking alone? There are plenty of other activities that are dangerous and that reduce people’s life expectancy: horse-riding, ski-ing, climbing, going on holiday and getting drunk on a balcony. Why no call to tax these? These too are no doubt elastic to price and we could ‘save x-million lives in the next century by using taxes to double their price’. But the obvious lunacy of any call to tax these shows that the government is not there to coerce anyone into living safely and hence it is not there to tax risky behaviour.

You might think that the government is there to coerce prudent living and that this justifies taxing smokers. We have yet to establish that smoking is imprudent. Whether it is imprudent depends entirely on whether, for the individual concerned, their life is better or worse if they smoke. The article presents the cost of smoking. But it does nothing whatsoever to consider the benefits of smoking. Only if the costs outweigh the benefits is smoking imprudent. There are plenty of people who show they think their life is better if they smoke. Who are you or I,  Prabhat, Peto or Kumar and, for that matter, the World Health Organisation, to presume we know better?

Passive smoking  is not the concern of the article but let us, for completeness, consider it anyway. You might think the government is right to discourage behaviours that harm others and that this justifies taxing smokers. That is sometimes right: preventing harm is at least a necessary condition for government coercion. But passive smoking does not harm others. You may think otherwise because of the dishonesty of the anti-smokers on this point. The burden of proof is on those who claim passive smoking is harmful. The WHO study of 1998 showed, and Enstrom and Kabat (BMJ 2003 have pretty conclusively confirmed, that the claims about harms from passive smoking are unfounded. See also survey: Peres 2013 No Clear Link Between Passive Smoking and Lung Cancer

Originally at

More on passive smoking
 ““Conclusions The results do not support a causal relation between environmental tobacco smoke and tobacco related mortality”, ““Conclusions: Our results indicate no association between childhood exposure to ETS and lung cancer risk.”, “A large prospective cohort study of more than 76,000 women … found no link between the disease and secondhand smoke.”

All the talk about links, weak evidence, etc, is a misleading way of saying that nothing of statistical significance was found but a p value greater than 50% was found. So what this talk really means is we cannot reject the null hypothesis (no effect) because our p-value is greater than 5%. And plainly,  that proves nothing at all. The case for passive smoking was from the beginning based on ‘links’ and ‘weak evidence’ of this kind and it was dishonest for that very reason. So-called links that lack statistical significance are results which could be mere chance.

In fact, the methodology of the positive results on passive smoking was shown some time ago to be riddled with basic errors and they fit Languir’s remarks on pathological science very well: ‘The effect is of a magnitude that remains close to the limit of detectability, or many measurements are necessary because of the very low statistical significance of the results.’

Consider what Ioannadis showed about published research which is statistically significant: ‘Why Most Published Research Findings are False’ (2005 So not even statistically significant results are likely to be true. That certainly means that all these claimed links that do not even have statistical significance should be given no credence.

And that is why the meta-analyses are of little worth: garbage in-garbage out. Of far greater importance are studies with appropriate power based on very large sample numbers, which is what Enstrom and Kabat and the study reported on by Peres are.  

Unfortunately, Peres’s article also reports a lot of opinionating, such as the opinionating of Patel, Wang and Winn. Consider the context we now inhabit about passive smoking:
1.        the general publication bias against publishing negative results,
2.        the social regimentation of opinion that has gone on over this issue,
3.        the social regimentation has among other things supplemented the publication bias and biased all interpretations of negative results
4.        social regimentation explains the willingness of authors of negative results to give sops (‘weak evidence’ instead of ‘cannot reject null hypothesis of no effect’) to those they know will verbally assault them for their negative results
Such opinionating is essentially worthless because it is a mere reflection of this context and the social regimentation of opinion that is prevalent.

Finally, why did I pick on lung-cancer rather than, for example, respiritory infections? Mainly because the article I was responding to was arguing for tax increases on the basis of saving people’s lives. Nevertheless, Jones et al  raises another important area of methodological problems with claims about the effects of passive smoking. Epidemiology is riddled with research publications based on relative risk ratios less than 3 (= 300% or increases in risk less than 200%). This methodology has come under sustained criticism.  
In epidemiologic research, [increases in risk of less than 100 percent] are considered small and are usually difficult to interpret. Such increases may be due to chance, statistical bias, or the effects of confounding factors that are sometimes not evident .[ National Cancer Institute, Press Release, October 26, 1994.]
As a general rule of thumb, we are looking for a relative risk of 3 or more before accepting a paper for publication." - Marcia Angell, editor of the New England Journal of Medicine"
My basic rule is if the relative risk isn't at least 3 or 4, forget it." - Robert Temple, director of drug evaluation at the Food and Drug Administration.
An association is generally considered weak if the odds ratio [relative risk] is under 3.0 and particularly when it is under 2.0, as is the case in the relationship of ETS and lung cancer." - Dr. Kabat, IAQC epidemiologist

The main point is that chance alone can easily produce relative risk ratios greater than 1 and to even approach statistical significance greater than 3 is needed.

The willingness to publish research with low ratios is part of the reason why epidemiology has a low reputation in the biological sciences and why some critics say it should not be regarded as a science at all.
Now all except one of the relative risk ratios reported in Jones et al are less than 2 and one is 2.51, so it falls into the category of epidemiological publications that critics are willing to call not science at all. On top of that, and of great significance, it is itself not a base study but another meta-analysis. So first is liable to the garbage in-garbage out point already made. But what is especially significant is that, despite being a meta-analyis it still hasn’t managed to get relative risk ratios in the statistically significant range. That is really bad. It means that the analyses that it aggregated were so weak that every single one of them might be nothing more than chance. 

No comments:

Post a Comment