Who's Not "Statistically Significant Then"
During March 1988 great controversy raged over the study conducted by the World Health Organization. Below we re-print in full the text that appeared on pages 76 and 77 of its report. There then follows an extract from the article "Behind the Smoke Screen by Victoria Macdonald and Robert Matthews that appeared in The Sunday Telegraphon 15th March 1998 explaining what the risks really mean.
WHO Report, page 76-77. 3.7.2 Lung cancer and exposure to environmental tobacco smoke (there then follows a list of those involved in the study - available from FOREST upon request - Ed)
"Environmental tobacco smoke (ETS) is a likely cause of lung cancer [27,37], while evidence of an association with other neoplasms is inconclusive. However, the quantitative aspects of the association between ETS exposure and lung cancer risk are not well established, nor is the interaction between exposure to ETS and exposure to other carcinogens. (Ed -- my emphasis).
An IARC-coordinated International collaborative case-control study was aimed at investigating the relationship between exposure to ETS and to other environmental and occupational risk factors and the risk of lung cancer in subjects who have never smoked tobacco. A total of 650 cases and 1542 controls have been enrolled in 12 centres in seven European countries. Information on exposure to occupational carcinogens, urban air pollution, background radiation and dietary habits, as well as lifelong exposure to ETS, has been collected by personal interview of cases and controls. Self-reported (non-)smoking status was confirmed by interviews of relatives. The relative risk (RR) of lung cancer risk was 1.16 (95% CI 0.93-1.44) for exposure to ETS from the spouse, 1.17 (95% CI 0.94-1.45) for workplace ETS exposure and 1.14 (0.88-1.47) for combined spousal and workplace exposure. Several quantitative indicators of ETS showed a dose-response relationship with lung cancer risk; RRs were higher for squamous cell carcinoma than for adenocarcinoma. There was no association between lung cancer risk and ETS exposure during childhood. Additional analyses are continuing on risk factors other than ETS."
Extract from `Behind the smokescreen' by Victoria Macdonald and Robert Matthews, which appeared in The Sunday Telegraph, 15th March 1998.
"In its [WHO's] report, the study team states its findings in the recherche language of statistical analysis. Those exposed to environmental tobacco smoke from their spouse, it claimed, have a `relative risk' (RR) of lung cancer of 1.16, with a so-called 95 per cent confidence interval (more of this later) ranging from 0.93 to 1.44.
On the face of it, the conclusion is clear: the study had found a 16 per cent greater risk of cancer among those living with a smoking spouse.
That, at least, was the message that the WHO and its supporters wanted emphasised. But as ever with statistics, the devil is in the details ...
The WHO study was based on results obtained by recruiting hundreds of people and taking into account all the obvious sources of lung cancer to which they were exposed, such as urban pollution.
Then, by sifting out the effect of all these other potential causes, the research team was able to give an estimate of the risk of lung cancer due to inhaling someone else's smoke.
Yet no matter how careful the researchers are to rule out other explanations from such statistical studies, one simple one remains: fluke.
Statisticians have developed ways of stating their confidence that they haven't been fooled by fluke, and the WHO team chose one known as the confidence interval method. This gives the range of results which the study might reasonably have got if nothing more than chance were at work.
And the bad news for those hailing the WHO results as proof-positive of a cancer link with passive smoking is that the study's own confidence interval of 0.93 to 1.44 includes a relative risk of 1 - that is, no extra risk.
In other words, the results are - as The Sunday Telegraphstated - consistent with there being no real cancer risk for non-smokers, and even with there being a reduced risk.
The WHO last week accused The Sunday Telegraphof concentrating on the study's confidence intervals while ignoring the study's 'headline' relative risk figure of 1.16. But what the WHO declined to point out was that this headline figure alone is meaningless - as, unlike a confidence interval, it does not show how likely fluke is to give such a result.
The chances of getting this headline figure by fluke alone can be calculated. And once again, the bad news for the WHO and its followers is that those chances are so high that the headline figure is not even close to being statistically significant."
Owing to media interest in the WHO study the government appointed Scientific Committee on Tobacco and Health (SCOTH) rushed out a report on ETS on 11th March, No Smoking Day.
Unlike the WHO study, this report contains no new or original research, and merely continues the anti-smokers' practice of reviewing reviews of studies on ETS. The report does not consider the WHO study referred to above, and is issued prior to five further WHO studies on ETS due to report during 1998 or later.
While the SCOTH report was hailed as being a report on the effects of ETS on non-smokers, the bulk of the report actually concentrates on primary smoking and factors that affect smokers and not non-smokers.
In keeping with earlier studies it reminds readers that middle age is defined as being anyone between the age of 35 - 69 years old. Given that average life expectancy in the UK is 74 years for men and 79 years for women, the definition of middle age leaves little time for old age.
Also ignored by SCOTH is the fact that nearly 40% of the 120,000 deaths in the UK allegedly due to smoking occur in men and women above the average age of life expectancy (Hansard, 24th July 1997).
In the section on ETS, at point 2.7, reference is made to a November 1997 report by the Australian National Health & Medical Research Council (NHMRC). This seems odd because at the beginning of 1997 an Australian court prevented the publishing of its report because the NHMRC had failed in discharging its statutory duty of public consultation. In April 1997 a member of the working party on the report, Simon Chapman, revealed to the Australian Press Association that the calculations of risk to non-smokers who were exposed to ETS were so low that journalists "will be hard pressed to write anything other than `Official: passive smoking cleared-no lung cancer'." (For further details see Free Choice, May/June 1997).
At paragraph 2.20 SCOTH puts the risk of lung cancer in non-exposed non-smokers at 10 people per 100,000. It then goes on to note that an increased risk of 20-30% of cancer in non-smokers exposed to ETS would mean 12-13 people per 100,000. However, as pointed out above, no account was taken of the WHO study, and it is likely that if the method of the confidence interval were used to take fluke into account, it would still include a zero risk.
Of greatest concern must be the claims in relation to cot deaths. The only material considered were two reviews, one from the UK in 1994 and one from the US 1995, and a report by Professor Peter J Fleming published in the British Medical Journalin July 1996.
No account was taken of original research conducted by Jim Sprott and Barry Richardson during the 1980's, or a series of other studies which failed to establish a causal link between cot death and exposure to ETS (for full details see Free Choice, September/October 1996).
Of the 37 recommendations, 46% were statements as to why the Government or the public should believe everything SCOTH says, including three recommendations on what the Government should force tobacco companies to do. 38% of the recommendations could be described as job related.
Insofar as they called for further research, training or monitoring which, of course, require people to do the job, and money to pay them. A further 11% of recommendations carried the implication that not only should the Government or public believe them but that jobs should be created to enable these beliefs to be put into action. The remaining recommendations were policy proposals concerning bans on smoking and advertising, and increases in tobacco taxation.
The greatest surprise was recommendation number 30 which acknowledged that there were some beneficial effects of active smoking and called for more research.
The report, however, does not appear to be as welcomed by the Government as one might expect. In a report by the Press Associationon 11th March the following appeared:
"But the Department of Health will stress that today's report is no more or less than the opinions of the scientists and not a blueprint for the White Paper." (my Emphasis --- Ed)
Given that the majority of those named in the report are scientist with names well known in the anti-smoking movement, while none of those who have been critical of the original research on ETS or found no correlation between ETS and ill health in non-smokers sat on the Committee, it is hardly surprising that inconvenient research was not considered, the report was written and issued before the WHO research has been published in full, or that their opinion should be what it is.