Wednesday 2 September 2009

1+1=11: The Helena Miracle (part 3)

The Helena Miracle is perhaps the archetypal junk science study. Now five years old, there is little more that can be said about its shocking methodology and ludicrous conclusion. This piece of nonsense has been thoroughly debunked by many writers, including myself and I would only offer one further observation.

For those unfamiliar with the story, the Helena Miracle is the name given to a study written by Stanton Glantz and two anti-smoking activists which purported to show that a short-lived smoking ban in the small town of Helena, Montana resulted in a 'nearly 60%' fall in heart attack admissions to local hospitals. The unavoidable conclusion, as publicised in a worldwide press release sent out by Americans for Nonsmokers' Rights (ANR), was that reducing exposure to secondhand smoke immediately saved lives.

The study was so flawed as to be comical. No questions had been asked of any of the patients regarding their exposure to secondhand smoke, their lifestyles, diet or age. Two-thirds of them turned out to be either current or former smokers. It transpired that the authors of the paper had merely counted heart attack cases and ascribed their own explanation to it. Furthermore, the numbers involved were ridiculously small; there had been an average of seven cases a month prior to the ban and four cases after it.

Nevertheless, it was immediately picked up by the international media and august news-gathering organisations such as the BBC ("Town slashes heart attacks") and The New York Times ("The Secondhand Smoking Gun") without asking even the most basic questions of the report. It was thoroughly reported in the UK, where the prospect of a national smoking ban was beginning to raise its head, and journalists were quick to extrapolate the findings from Helena onto the population of the UK ("Smoking ban could halve number of heart attacks" - The Independent). New Scientist, to its shame, covered it with the headline "Public smoking ban slashes heart attacks", a phrase so close to the title of the ANR press release that it is most charitable to assume the journal had not seen the (still unpublished) study and instead took the ANR's claim on trust.

But there was a problem with the study that went beyond its atrocious methodology and the obvious bias of its authors. If the smoking ban was responsible for reducing heart attack admissions by 60%, then secondhand smoke must have been responsible for 60% of the heart attacks in Helena before the ban. And since Helena was a typical American town - why else would they do an epidemiological study there? - the only conclusion could be that passive smoking was responsible for the majority of heart attacks in the US.

And, of course, the Helena smoking ban did not eliminate passive smoking. It merely reduced it in bars, restaurants and workplaces. More accurately, then, we should say that passive smoking in some places is responsible for 60% of heart attacks. Perhaps passive smoking in the home causes the other 40% and, therefore, secondhand smoke is the sole cause of heart attacks?

The glaring problem was that active smoking was not responsible for anything close to 60% of heart attacks. No health group has ever claimed that active and passive smoking combined is responsible for any more than 20% of heart disease cases and, of course, the vast majority of these are attributed to active smoking. It was therefore risible to suggest that "nearly 60%" of heart attacks could have been triggered by secondhand smoke exposure in so-called public places. The killer flaw with the Helena Miracle was not in the study itself - awful though it was - but with the conclusion. As with Repace's tornado claim, it is not necessary to question the validity of the passive smoking theory in order to debunk the Helena miracle. Even if we suspend disbelief and accept that passive smoking is lethal to nonsmokers, there is no way on earth that it can possibly be responsible for anything like 60% of all heart attacks.

The mainstream media did not pick up on this obvious point. However, it did not escape everyone's attention and Glantz was called to task on it. It is his response to the criticism that the Helena effect was far too large to be believable that is the focus of this article. His defence was:

"The individual risk of heart attack associated with passive smoking is about 30%, which is within the margin of error of the 60% drop we saw in Helena."(1)

If Glantz believed this then he is, quite simply, mathematically illiterate. He makes the elementary error of confusing relative risk (to the individual) with absolute risk (to the population).

To give an example, driving while drunk might increase the risk of crashing your car by 90% but that does not mean that the number of car crashes in a country will fall by 90% if no one drives while drunk. The reason is simple: not all car crashes are caused by drunks. Let's say - for the sake of argument - that 5% of all car crashes are caused by drunk driving. Eliminating drunk driving entirely will therefore reduce car crashes by 5% but not by 90%. The 90% figure for individual risk has no bearing on the overall incidence of car crashes.

This really is basic stuff. Let us, for a moment, take Glantz seriously when he says that passive smoking increases an individual's risk of a heart attack by 30%. Let us also assume that obesity, diabetes, physical inactivity, stress, excessive alcohol, too little alcohol, high cholesterol, high blood pressure and a family history of heart disease also increase risk by the same amount (2). Add in passive smoking and we have ten factors which increase heart attack risk by 30%. Now let us say that each of these factors is responsible for 100 heart attacks a year in a community = 1,000 heart attacks in total. If we eliminate all secondhand smoke exposure from that community then the most we can possibly hope for is a fall in the heart attack rate from 1,000 to 900 ie. 10%. Therefore, even if a smoking ban eliminates secondhand smoke exposure entirely - which it won't - it cannot possibly bring about more than a 10% fall.

And that is just when we take 10 risk factors into account. In fact - according to the World Health Organisation (3) - there are over 300 risk factors for heart disease. So let us go back to our theoretical community and say that there are 300 factors that cause 100 deaths each. This equals 30,000 deaths and one factor - passive smoking - causes 100 of them. Passive smoking is therefore responsible for 0.3% of all the heart attacks in the community. It cannot be responsible for any more. If secondhand smoke is removed completely, the heart attack rate cannot fall by more than 0.3%.

Lest I be accused of peddling junk science myself, I should quickly add that I am in no way conjuring up estimates of how many people are killed by passive smoking, active smoking, obesity, stress or anything else. This is nothing more than an illustration of the difference between relative and absolute risk but I hope it suffices. I am not claiming that each of the WHO's 300 risk factors raise individual risk by exactly 30% or are responsible for exactly 0.3% of all heart attacks. The major risk factors - genetics, stress, smoking and so on - have far more effect than others, including passive smoking. But this only underlines how insignificant secondhand smoke is in the overall picture.

Even if you believe that secondhand smoke exposure does raise individual risk by 30% you must accept that its contribution to total heart attack incidence cannot possibly be anything close to 30%, let alone the 60% initially claimed by Glantz (when the study was eventually published in the British Medical Journal (4), the percentage had fallen to 40%, a discrepancy that was never explained*.) And once you accept that, you must accept that the conclusion of the Helena study is not just flawed but mathematically impossible.

At this point you might say: "Well, smoking bans may not save as many lives as some of the more fanatical anti-smoking types claim, but they must still save some lives". And, in a way, this is the message that all the studies covered in this series of articles want you to be left with. Any intelligent person can see that "tornado-level ventilation" is not required to disperse secondhand smoke. Anyone with any common sense can see that secondhand smoke cannot be "more dangerous" than firsthand smoke. And it doesn't take much to work out that passive smoking cannot be responsible for 70%, 60%, 40%, 27%, 17% or 11% of all heart attacks. Even if everything the anti-smoking groups say about passive smoking is true, it would be extraordinary if a smoking ban could result in even a 1% drop in heart attacks.

No one wants to see themselves as gullible by believing big lies but, equally, no one wants to be seen as a crank by dismissing the whole thing. Big lies are useful to the anti-smoking movement because they give intelligent people the scope to be sceptical about what they see as 'exaggerations' while inviting them to conclude that there must be some truth at the heart of the exaggeration. The big lie does not have to be swallowed whole. It only has to spark the thought that there is no (secondhand) smoke without fire.

The anti-smoking movement's descent into hyperbole almost invites us to be cynical about the big lies while insisting that we believe the little lies. But why should we believe either? After all, the same people are responsible for both. Stanton Glantz's botched defence of his Helena study rested on his claim that passive smoking increases heart disease risk by 30% but he does not mention that this figure comes from a paper written by himself in 1991; a report that was rejected by the Environmental Protection Agency when he tried to have it tied to their notorious 1992 report. James Repace gave us the tornado-level winds but he was also an early proponent of the passive smoking theory in the late 1970s and today makes a living as a professional 'secondhand smoke consultant'.

We should be able to expect reliable and rational information from the world's foremost medical journals and, frankly, shame on the British Medical Journal for publishing the Helena study. And shame on Stanton Glantz - a man who has written a textbook on biostatistics - for submitting such nonsense for publication.
* All mention of the 60% figure has been removed from the Americans for Nonsmokers' Rights website. However the original press release can still be viewed here.

(1) http://listsrv.ucsf.edu/OldArchives/hypermail_archives/stanglantz-l/0070.html

(2) For a list of the major causes of heart disease see: http://www.americanheart.org/presenter.jhtml%3Fidentifier%3D4726+risk+factors+heart+disease&hl=en&ct=clnk&cd=1&gl=uk

(3) http://www.who.int/cardiovascular_diseases/en/cvd_atlas_03_risk_factors.pdf (pdf) To name just a few other risk factors: depression, excess homocysteine, blood coagulation, inflammation, lipoprotein, hypertrophy

(4) http://bmj.bmjjournals.com/cgi/content/full/328/7446/977

No comments: