In what I think is a very important BMJ paper, John Ioannidis and Fotini Karassa have shown that systematic reviews, the highest level of evidence, may be highly deceptive. We did, I suppose, know this in a way, but they illustrate the case beautifully. What disappoints me is that nobody has responded to their paper.
I’m reminded of the first time that I ever wrote to a medical journal—to the Lancet in 1974, when I was a medical student. I wrote in response to Ivan Illich publishing a paper saying that “Major medicine is the major threat to health in the world today.” I said that I was amazed that nobody responded. What did that mean, I asked. Did everybody agree? Did nobody have arguments to refute what Illich said? Or did people not think the paper worthy of comment, in which case, I wondered, why did the Lancet publish it?
Similar questions occur to me with the lack of response to the paper by Ioannidis and Karassa.
John Ioannidis has already shown how most research findings are false and how doctors may receive a very distorted view of the evidence by reading the top medical journals. It’s no wonder that he has reached the attention of the New York Times, but I fear that his messages may be a little too uncomfortable for the average reader of medical journals.
Since the article by Ioannidis and Karassa was published, Fiona Godlee and Elizabeth Loder have told us how systematic reviews have misled us on the effectiveness and safety of the antidepressant reboxetine and argued that “the medical evidence base is distorted by missing clinical trial data” and that “urgent action is needed to restore trust in existing evidence.” But, as Ioannidis has shown, there are many sources of distortion and manipulation in addition to failing to publish data.
Sadly, I’m sceptical that the answer lies with journals. High profile journals are very attractive to those who want to push their idea or product–and it’s not only drug companies who want to do so. They guarantee attention but may also amplify the distortion, as Ioannidis has argued.
As Ian Roberts and I suggested years ago, we need another way to make research available. The job of journals should not be to publish original research but rather to assess critically the research that is published in full on databases. Find the comparatively few studies that matter, test them to destruction, and place them in their full context.
The snag, of course, is economic. Subscriptions from institutions and payments for reprints from drug companies are major sources of revenue for journals, and it seems that institutions (the profitable subscribers) are paying primarily so that their users can access research. They are less willing to pay for analysis and comment.
So we may need to find not only new ways of disseminating research but also new business models to support not necessarily journals but rather the process of critically appraising, digesting, and implementing research. Both are difficult problems, but most interesting problems are difficult.
Competing interest. RS is a former editor of the BMJ and chief executive of the BMJ Publishing Group. He will have a pension from the BMA, the owners of the BMJ, and the profits from publishing are important to the BMA. RS is also on the board of the Public Library of Science, an open access publisher that is now profitable largely because of income from PloS One, which might be described as more a database than a journal, RS receives no payment for being on the board, but has his expenses paid twice a year to attend a board meeting in San Francisco. But an economy flight to and from San Francisco over three days is more a punishment than a reward.