On Thursday, the ICMJE (International Committee of Medical Journal Editors) is expected to announce that its member journals will only publish results for clinical studies entered into a public registry at the beginning of the trial, before a trial sponsor could know if results would be favorable or disappointing.
Reacting to lawsuits and public pressure, pharmaceutical companies are promising to post results of all clinical trials for drugs that have been approved by the FDA, and the PhRMA (Pharmaceutical Researchers and Manufacturers of America) has offered to host a Web site to contain the database. The Web site proposed by the PhRMA may ease some of the logistics in making more data more available, but because submission would be voluntary, it will not address the issue of sharing negative information.
Instead of half-hearted efforts that seem motivated only at improving public relations, industry groups should champion the idea of controlled data-sharing as a tool to make research more efficient.
Within the industry, the notion of "publication bias" is well-understood. No company wants to publish results showing that a competitors product works better. For example, in a trial of anti-cholesterol drugs, Bristol-Myers Squibb sponsored a study that showed that patients taking Pfizers Lipitor were less likely to suffer heart attack and other cardiovascular diseases than patients taking its own Pravachol. (The devil is in the details, though: The difference in efficacy is modest, and, overall, patients taking Lipitor had slightly more side effects.)
True, BMS paid for a study that looks good for Pfizer, but it will also encourage the use of this drug class, and it offers clues to designing improved drugs.
To address publication bias, though, the drug industry needs a shift in culture. Before the study began, BMS and the researchers at Harvard Medical School agreed that both favorable and unfavorable results could be published. This condition is often not included in contracts between the company that sponsors a drug trial and the researchers who conduct it, but it should be standard.
The counterargument to disclosure is that companies wont conduct studies if they are worried about damaging results, so crucial information about disease and treatment will remain unknown. This is a real concern and could be addressed if the FDA offered incentives to companies for adding valuable information to the public domain.
But reluctance to share results could also be addressed if companies saw disclosure as helping the industry as a whole. If companies can learn from each others mistakes and good ideas, the ballooning cost of drug development might go down, and with it, the cost of drugs and political pressure for price regulation. Yes, that view is a bit simplistic, but its got a nugget of truth.
The drug industry has collaborated before to its advantage: for example, in setting data standards for submitting data to the FDA.
A database of clinical trials results should be something the drug companies welcome, not something they just pretend to accept.
M. L. Baker writes about biotechnology and health IT. Write to her at firstname.lastname@example.org.