Populist instead of professional

The evidence on the eYcacy of mammographic breast cancer screening is as firm as any can be in medical science. Randomised trials show clearly that breast cancer mortality is about a third lower in screened women. Many expert groups have reviewed the evidence and concluded that breast cancer screening is worthwhile. In spite of this a recent paper by Gøtzeche and Olsen re-examined the same evidence and concluded that breast cancer screening is not justified. The paper was published in The Lancet, and the sensation of the claim, seemingly supported by so prestigious a journal, attracted considerable publicity in the media. Experts interviewed at the time reinforced the opinion that women should be screened but did not give clear reasons why the report was judged to be flawed, and so people remained confused. Medical journals and the media, who take their cue from papers published in the medical journals, sometimes tend to encourage controversy rather than scholarship and sound interpretation. Sensational headlines and increased circulation become more important than accuracy and good judgment. Gøtzesche and Olsen’s paper lacks scientific merit. The accompanying editorial by de Koning stated the problems in the paper and one might ask why his comments were not referred back to the authors with a view to correcting it instead of publishing both in the same issue together. The flaws in the paper published in The Lancet were covered in detail in the subsequently published correspondence. Gøtzesche and Olsen considered seven randomised trials of breast cancer screening and inappropriately rejected five valid trials because they failed to acknowledge that minor chance diVerences in age and social class between screened and control groups will arise with the cluster randomisation used in most of these trials. Such diVerences, being random, will tend to cancel out when trials are combined. Their analysis rested on only two trials, one of which did not compare screening with no screening (it was a comparison of mammography plus breast palpation with breast palpation alone). The other trial was valid, but preliminary results were used rather than more recent results, and data on women under 50 and over 50 were combined: they should have been analysed separately because it remains uncertain whether mammography significantly reduces mortality in women under 50. To what extent the paper by Gøtzeche and Olsen has undermined the confidence of women invited to attend breast cancer screening remains to be seen. More generally, however, reports like this will tend to make the public sceptical of the quality and reliability of all medical findings reported by the media—to regard medical reports as no more than the “opinions” of one group that are probably going to be countered by the “opinions” of others. Medical science becomes medical opinion and the disagreement becomes entertainment. The Lancet should not have published this paper. Editors of medical journals have a duty to try to publish valid work and to correct known errors of fact or interpretation before papers are published. This is particularly the case where the issue is an important public health measure that will aVect millions of people. To publish a paper which the accompanying editorial rightly criticises as being unsound is being more populist than professional.