Letter of the week The use and misuse of stats
From Dr John Chalmers, Rosyth, Fife
Dr Malcolm Kendrick is
(under)mining a rich seam, in his recent column on breast cancer screening (Clinical, 24 August). In the efforts to move medicine from its origins in 'arts and crafts' to 'evidence-based', doctors have cast around looking for scientific methodology and measurement tools to nail their convictions to the door. In the main, unfortunately, all they have come up with is the 'science' of statistics.
For my first appraisal, I was looking for something plausible and cynical to engage with over
the following year. My thoughts naturally turned to statistics.
I had always found difficulty in accepting them as anything more than, at best, plausible and their use in medicine is exquisitely cynical. After a year wrestling with the topic, I have an increased respect for the mathematical basis for the subject – almost, but not quite – matching the increase in my suspicion of any statistical 'fact' put under my nose.
Much of the condemnation of Olsen and Gotzsche (whose research Kendrick referred to in
his column) was off-target and
had a whiff of vested interests overreacting in the defence of their empire. The dispute arose not out of the medicine but out of the statistics themselves, as the establishment argued about one thing while Olsen and Gotzsche had said something quite different.
Having caught a brief interview with one of the Danes on the radio while doing my calls for the day,
I was impressed by the low-key, unchallenging and clear repetition of what he and his colleague had said.
Their statistical analysis of all published papers on the subject showed that mammography screening reduced the relative risk of breast cancer mortality, but they argued this was an unreliable outcome and biased in favour of screening. Olsen and Gotzsche concluded that the available reliable evidence did not show a survival benefit of mass screening for breast cancer.
The choice of the 'end point' is left largely to the prejudice of
the researcher or, fortunately,
the informed sceptics in the profession.