Monday, September 28, 2009

The myth of the smoking ban “miracle”

Restrictions on smoking around the world are claimed to have had a dramatic effect on heart attack rates. It's not true. See also some prior comments on this blog on 15th.

‘Heart attacks plummet after smoking ban’ declared The Sunday Times earlier this month, as it reported that England’s smoking ban has ‘caused a fall in heart attack rates of about 10 per cent’ (1). A few days later, The Scotsman upped the ante, informing its readers that ‘Smoking ban slashes heart attacks by up to a third across world’ (2).

Tales of heart attacks being ‘slashed’ by smoking bans have appeared with such regularity in recent years that it is easy to forget that there is a conspicuous lack of reliable evidence to support them. It is almost as if the sheer number of column inches is a substitute for proof.

The most recent reports are a case in point. Although The Sunday Times claimed a 10 per cent drop in heart attacks, nowhere in the 500 word article was a source mentioned and no one was quoted giving this figure. The ‘study’ the newspaper referred to does not exist, and the anti-smoking pressure group Action on Smoking and Health (ASH) – not renowned for downplaying the risks of passive smoking – went to the unusual lengths of posting a notice on its website the following day to point out that ‘the figures reported in The Sunday Times yesterday (and now circulating elsewhere) are not based on any research conducted to date’ (3).

Although the story quickly went around the globe, no one seems to know where the figure came from. It’s all rather strange. Basing journalism on anonymous sources is commonplace in the world of politics, but it is surely not necessary in the realms of science.

The second story – reported by a host of news organisations, including the BBC – also had no new data to report. Instead, it took its cue from an article in the journal Circulation which examined previous smoking ban/heart attack studies. If nothing else, the Circulation paper offers an opportunity to reflect on just how feeble the collected evidence is on this issue (4).

The first study to make the claim that smoking bans ‘slash’ heart attacks was met with howls of derision when it was published in the British Medical Journal in 2004 (5). Studying the modest population of Helena, Montana – where the number of monthly heart attacks seldom strayed into double digits – the study’s authors made the astounding claim that the town’s smoking ban had led to the rate of acute myocardial infarction (heart attacks) plummeting by 40 per cent.

Dubbed the ‘Helena miracle’ by a legion of sceptics, the 40 per cent finding was damned by its very enormity. Since the authors were adamant that the drop was due to secondhand smoke (rather than smokers quitting), the finding required the reader to believe that 40 per cent of heart attacks in pre-ban Helena had been solely caused by passive smoking in bars and restaurants. To understand quite how miraculous the Helena miracle was, one must bear in mind that around 10 to 15 per cent of coronary heart disease cases are attributed to active smoking. That passive smoking could be responsible for a further 40 per cent strains all credibility.

Despite the inherent implausibility of the hypothesis, further studies were swiftly commissioned. If smoking bans could be shown to immediately save lives, it would be a compelling reason to implement bans elsewhere and expand those already in place. And since all that was required to ‘prove’ the hypothesis was a rough correlation between a declining heart attack rate and the start of a smoking ban, the prospects were good. Heart attack rates had been falling for years in most countries and there were plenty of smoking bans to choose from. The law of averages dictated that another heart miracle would soon come to light.

Flawed though it may have been, the Helena research was followed by several studies that displayed such a cavalier approach to the scientific process that they bordered on the comical. Researchers in Bowling Green, Ohio, for example, saw a large rise in heart attacks during the first year of the smoking ban. Side-stepping this awkward fact, they simply redefined year two of the ban as the ‘real’ post-ban period and, since that year followed an abnormal peak, there was naturally a decline in the heart attack rate. As a consequence, the researchers could triumphantly declare that the smoking ban had led to a 47 per cent reduction in heart attacks (6).

In the Piedmont region of Italy, there was an inconvenient rise in heart attacks amongst those over the age of 60 after the ban, and so those people were simply ignored. In a study that was trailed by the BBC (‘Smoking ban reduces heart risk’), the researchers focused entirely on those under 60, thereby recording an 11 per cent drop in cases (7).

Studies such as these form the basis for the recent reports of smoking bans slashing heart attacks by ‘up to a third’. The Circulation paper gathers them together and concludes that, on average, smoking bans cause rates of acute myocardial infarction to fall by 17 per cent. It includes the studies from Ohio and Italy, as well as three studies that have never been published and have only been ‘reported at meetings’.

The paper does not, however, include a mammoth (published) study of the entire United States, which concluded: ‘In contrast with smaller regional studies, we find that workplace bans are not associated with statistically significant short-term declines in mortality or hospital admissions for myocardial infarction or other diseases.’ (8)

Nor does it include an (unpublished) paper which found no statistically significant fall in heart attacks amongst the entire populations of California, Florida, New York and Oregon (9).

Perhaps the most remarkable aspect of the ongoing heart-miracle farrago is the eagerness to focus on small studies when complete hospital data is so freely available. It is extraordinary that no BBC journalist, for example, has thought of taking a few minutes to see how many people were rushed to hospital with acute myocardial infarction before and after the smoking bans of England, Scotland and Wales. If they did so, they would see that smokefree legislation has had no tangible influence on heart attack rates at all.

The graphs below show the number of emergency admissions for acute myocardial infarction, with the arrow indicating the start of the smoking ban. What is abundantly clear in each case is that the number of heart attack admissions has been falling for some time. Far from causing further dramatic cuts in heart attack rates, the bans had no discernible effect.

Publicly accessible hospital admissions data is like kryptonite to those who are so eager to believe in miracles. In most epidemiological studies pertaining to secondhand smoke, the raw data is not published. Here, it is open to all and shows quite clearly that the long-term downward trend in heart attacks has not been affected in any way by the implementation of smoking bans. It provides such a simple and straightforward rebuttal to the heart attack ‘slashing’ hypothesis that one wonders what level of hubris drives those who still espouse it.

The three graphs above cover a population larger than the sample groups in all the studies reviewed in Circulation combined, but no matter how much empirical evidence exposes the fantasy of the Helena miracle, it may be too late for the anti-smoking lobby to back down on this issue. Too many reputations are at stake.

After five years of covering these stories so uncritically, the same may be true of the media. One can scarcely blame newspapers for covering stories that offer such dramatic conclusions as the heart miracles. The irony is that if they dug just a little deeper, they might find a more interesting, and more believable, tale of human folly.

SOURCE (See the original for graphics)

Smoking status doesn't predict cardiovascular death

This article forms an amusing footnote to the one above. It essentially explains WHY smoking bans don't reduce circulatory disease.

It is also an interesting example of how one must consider intervening variables before drawing causal inferences. Non-smokers live many years longer but that is not due to their non-smoking. Smokers have other health problems and it is the sum of those problems that lead to death. Another example of the poor having worse health, I think. The poor are much more likely to smoke

Could it be good news for smokers? Current and past-smokers with coronary artery disease, cerebrovascular disease, or peripheral artery disease have less than half the cardiovascular mortality than never-smokers, the initial findings from a new study suggest. But don't be so quick to tell your patients to light up: After accounting for potential confounders, the association was not statistically significant.

"The relationship between smoking habit and outcome in patients with established arterial disease remains controversial," Dr. M. Monreal, of Hospital Universitari Germans Trias i Pujol, Barcelona, Spain, and colleagues write in the September issue of the European Journal of Internal Medicine. "Some studies have found that smoking may be associated with a better outcome among patients with acute coronary disease," they note. "As for patients with cerebrovascular disease or peripheral artery disease, there is little information on the influence of smoking on outcome."

The researchers used data from FRENA, an ongoing, observational registry of consecutive outpatients with symptomatic coronary artery disease, cerebrovascular disease, or peripheral artery disease, to compare the incidence of cardiovascular death during follow-up of all enrolled patients according to their smoking status. A total of 2501 patients from 24 participating Spanish hospitals had been enrolled in FRENA as of May 2008. Of these, 439 (18%) were current smokers, 1086 (43%) were past-smokers, and 976 (39%) never smoked.

Compared to never-smokers, current and past-smokers were younger, more often male, and more likely to have chronic lung disease. Diabetes, hypertension, and heart failure were less common in current- and past-smokers. There were a total of 250 major cardiovascular events in 239 patients (9.6% of the original 2501) over a mean follow-up of 14 months. A total of 123 (4.9%) patients died (cardiovascular death, 68) during follow-up.

Significantly lower cardiovascular mortality was observed among current smokers and past-smokers compared to non-smokers (1.1 per 100 patient-years in current smokers, 1.9 in past-smokers, and 3.5 in non-smokers). Similar results were found when patients with coronary artery disease, cerebrovascular disease, or peripheral artery disease were considered separately.

The mean age at cardiovascular death was 82 years for never-smokers, 70 years for past-smokers, and 67 years for current smokers. The mean age for non-cardiovascular death was 79, 74, and 69 years, respectively. "On univariate analysis, age >70 years, body mass index >28, chronic lung disease, heart failure, diabetes, prior history of artery disease, non-smoking status, atrial fibrillation, renal insufficiency, and the use of some drugs were significantly associated with an increased cardiovascular mortality," Dr. Monreal and colleagues write. However, on multivariate analysis, none of the variables, including smoking status, were independent predictors of cardiovascular death.

Eur J Int Med 2009;20:522-526.


No comments: