Not bad science, but awful logic

In today's soar-away Guardian, Ben Goldacre has taken on the subject of plain-packaging:

This week our government committed itself to the removal, albeit slowly, of cigarette displays in shops. But plain packaging on cigarettes has been delayed for further consultation.

The Unite union is unimpressed. It represents 6,000 people in tobacco production and distribution, and put out a statement: "Switching to plain packaging will make it easier to sell illicit and unregulated products, especially to young people." This, the union added, "may increase long-term health problems".

Tory MP Philip Davies said: "Plain packaging for cigarettes would be gesture politics … it would have no basis in evidence."

Everyone is entitled to their own opinions, but not, sadly, their own facts.

Strong words, from which we must assume that Goldacre has found the "facts" to show that plain packaging will not help the illicit trade and does have basis in evidence.

But if he has, there's no trace of them in this article. What we get instead is a mildly interesting précis of surveys showing that some smokers believe that low tar cigarettes are less hazardous. This, says Goldacre rather hyperbolically, is "one of the most important con tricks of all time: people base real decisions on it, even though low-tar cigarettes are just as bad for you as normal cigarettes, as we have known for decades."

The collected data from a million people shows that those who smoke low-tar and "ultra-light" cigarettes get lung cancer at the same rate as people who smoke normal cigarettes. They are also, paradoxically, less likely to give up smoking.

The study he links to is Harris, 2004, which found that smokers of low-tar cigarettes have a similar lung cancer risk to those who smoke what Goldacre calls "normal" cigarettes. But the study also shows that smokers of high tar cigarettes (+22mg tar) have a statistically significant 44% increase in risk compared to those smoking lower tar cigarettes. To be fair, Goldacre's decision not to mention this latter finding could be justified on the basis that cigarettes of this strength are no longer legal in the EU. But when they were available (1930s to 1990s), there is plenty of epidemiological evidence to show that low tar cigarettes reduced risks somewhat.

As late as the 1960s, cigarettes regularly had tar yields of 40mg and over. Reducing tar yields to a quarter of this level does seem to have made them less hazardous in many respects—that is why the EU has made cigarettes progressively weaker. So it is not quite true to say that people believe light cigarettes are less hazardous purely because of the tobacco industry "con trick". For many years, they really were.

The EU's limit is now just 10mg of tar, which means that, by earlier standards, all cigarettes smoked today are ultra-light. The difference between a modern 5mg 'light' cigarette and a 'strong' 10mg cigarette is pretty negligible and the corresponding effect on health is likely to be similarly negligible. For the sake of argument, let's assume that modern lights are indeed no better than modern full-strengths.

What is not true is that smokers of low tar cigarettes are "paradoxically, less likely to give up smoking." The study Goldacre cites does not support that. In fact, it says the exact opposite:

We observed the smoking habits of all participants only at enrolment in 1982. However, based on a 13% subsample of participants who were re-enrolled in the CPS-II nutrition cohort, we found that men and women who smoked very low tar and low tar cigarettes in 1982 were more likely to have quit smoking by 1992.

But what does any of this have to do with plain-packaging? The answer is not very much at all. Goldacre does a pretty good job of debunking an argument that nobody is making, but does a poor job of rebutting what Unite and Philip Davies are saying. He implicitly assumes that by stopping the industry using colours to show which products are low tar, the myth of safer cigarettes will disappear and this will benefit public health. But that's just a hope and a prayer. Low tar cigarettes will still be available under plain packaging and smokers will still know that Marlboro Lights are on sale, regardless of what they are called or what they look like.

And even if low tar cigarettes disappeared altogether, there is no reason to think that people who smoke them will quit rather than simply switch to "normal cigarettes". Some of them might quit, of course, but that is mere supposition and we're supposed to be dealing only in facts here.

The truth is that there is very little evidence on either side of this debate because, if we go through with this scheme, we will be the first country on earth to try it. It is fatuous to pretend that there is any proof that it will work; likewise, there's no solid evidence that it will make things worse. Goldacre finds it plausible that banning colours will remove the illusion of reduced risk and lead people to quit (campaigners said the same when they banned the word 'light', but as Goldacre points out, it made no real difference).

Other people—including Unite—find it plausible that plain packaging will make it easier for the illicit trade to make counterfeit cigarettes and will draw smokers to the black market by turning the white market grey. (And we do know from chemical analyses that counterfeit cigarettes are more hazardous than official brands, if that's our concern.)

There are what-ifs on both sides of the argument, and neither side should claim that they are dealing with "facts" while their opponents deal only in "opinion".

At the end of the day, there can be no evidence for something that has never been tried. It is, therefore, factually accurate for Philip Davies to say that the policy "has no basis in evidence" and wrong for Goldacre to portray his hopes as facts.