By Peter C. Gøtzsche at Brownstone dot org.

The Cochrane Collaboration publishes systematic reviews of healthcare interventions in the Cochrane Library. It was once a highly respected institution, but this has changed, and I shall tell a particularly grotesque story about Cochrane bureaucracy, protection of guild and financial interests, inefficiency, incompetence, censorship, and political expediency, which I consider the beginning of the end for Cochrane.

As the events have lasting historical interest, I have uploaded and referenced the documents we received from Cochrane and our replies when we tried to update our published Cochrane review of mammography screening with additional mortality data.

Background

In 1999, after Sweden had screened for breast cancer for 14 years, a study failed to find an effect on breast cancer mortality. This led the Danish Board of Health to ask me to review the randomised trials. My statistician, Ole Olsen, and I were astonished when we realised that the evidence for the benefit of screening was poor and that screening might do more harm than good because of overdiagnosis and overtreatment of harmless cell changes and cancers, which increase mortality.

We noted in our report that, in contrast to what was claimed, screening had not led to less but increased use of radical treatments including mastectomies because of an overdiagnosis rate of 25-35%. We also noted that screening did not reduce all-cause mortality.

We delivered our report a week before the Danish parliament would vote about introducing screening but the director of the Board of Health, Einar Krag, censured our report and ensured that the minister, who was against screening, did not get it before the voting.

In Krag's view, Danes were not entitled to learn about our findings. My view was that the whole world should know about them, and we published our results in the Lancet, in January 2000. Our paper created a media storm and immense furore among screening advocates.

Lancet's editor Richard Horton noted that the editors of the Cochrane Breast Cancer Group had disowned our work, pointing out that our review was not a Cochrane review and had not been reviewed by them. What's the problem with that? Lancet is not an inferior journal.

Cochrane Steering Group co-chair Jim Neilson complained that our review gave the impression that it was a Cochrane review. It obviously didn't, as it was not published in the Cochrane Library and did not even look like a Cochrane review. What had happened was that some staunch screening advocates went ballistic about our results and complained to Cochrane without providing any scientific arguments.

The 2001 Cochrane Review Scandal

The Board of Health funded us to do a Cochrane review but tried to interfere with our work in the most absurd ways to ensure that we arrived at politically correct results. And when we submitted our review to the Australian-based Cochrane Breast Cancer Group - which had a financial conflict of interest, as it was funded by the centre that offered breast screening in Australia - we ran into a roadblock.

The editors refused flatly to include data on the most important harms of screening, overdiagnosis, and overtreatment of healthy women, even though these outcomes were listed in our protocol the group had accepted and published. We wasted a lot of time negotiating with the group but got nowhere.

It was the biggest scandal in Cochrane's history at the time. As we considered it more important to inform women honestly than to protect Cochrane guild and financial interests, we sent the full review, including the harms, to the Lancet. Horton worked with record speed and ensured that our review came out in the Lancet at the same time as the stymied review appeared in the Cochrane Library.

One of the Cochrane editors, John Simes, said to Horton that we had agreed to the changes they had insisted on, but I provided Horton with internal emails to demonstrate that Simes was lying. Horton ...