Electronic Telegraph October 11, 1998
Suspending the rules of science
Robert Matthews, Science Correspondent
SO NOW the facts are out. With the publication last week of the full World Health Organisation report on passive smoking, anyone can check the accuracy of The Telegraph's exclusive story last March, which disclosed that the WHO had failed to find any convincing evidence that passive smoking causes lung cancer.
Yet there has been little of the publicity which would have been expected for so striking a finding from a major study by an official organisation. But this is passive smoking research, where normal rules do not apply, including those of scientific investigation.
Indeed, the most impressive aspect of the WHO study is how the same political correctness has pervaded the organisation's approach to the scientific evidence.
Following our coverage of the study's findings, the WHO immediately issued a press release headlined "Passive smoking does cause cancer", vehemently insisting that it was "untrue" that the study had "failed to scientifically prove that there is an association [with] passive smoking".
Now that the study has been published, it is hard to see what basis the WHO had for making these definitive statements. So unremittingly negative were the study's findings that it is quicker to state the two statistically significant results it did uncover. The first is a hint of increasing risk with a measure of exposure to environmental tobacco smoke at home or the workplace. Given all the other negative risk findings, however, quite what this trend means is far from clear.
But the other positive result is a real headline-grabber: statistically significant evidence that childhood exposure to cigarette smoke cuts the risk of lung cancer by 22 per cent. In other words, exposure to cigarette smoke can be protective. Such a finding, while surprising, is clearly intriguing, and the authors of the WHO report made it their principal study finding. But in just the same way that the WHO can see "proof" of an extra cancer risk in statistically non-significant evidence, so it can fail to see anything at all in significant evidence for a lower cancer risk.
In the WHO report, the statistically significant finding on childhood exposure is transformed into evidence that "Exposure during childhood was not associated with an increased risk of lung cancer". This is Humpty Dumpty science, where words such as "statistical significant" mean what the WHO researchers choose them to mean.
