Reports that the drug giant GlaxoSmithKline buried evidence that its top-selling antidepressant, Paxil, is not effective in children have stimulated calls for a public registry of company-sponsored research. It’s about time. You might imagine that this was an isolated case of corporate misbehavior, but in fact suppressing unfavorable research results is fairly standard practice in the pharmaceutical industry.
Before drug companies are allowed to bring a new drug to market, they are required by the Food and Drug Administration to conduct clinical trials to show that the drug is safe and effective. They are supposed to provide the agency with the results of all the trials, not just some. But they don’t have to publish or otherwise make the results available, and neither does the FDA. Moreover, the FDA may approve a drug on the basis of a few trials that make it look good, even though other trials may indicate it is not effective. The companies selectively publicize the most positive results, and legions of sales representatives then visit doctors to tout the positive findings.
Just as troubling as the outright suppression of unfavorable research is the bias in the research that is published. Clinical trials can be set up to make drugs look good, and that, too, has become common. Companies can, for example, enroll only young people as human subjects, even if the drug is likely to be used by older patients thereby lessening the chance that side effects will show up. Or they can compare a new drug with an older drug administered at too low a dose, which means the new drug is bound to look better.
Sometimes they sift through the results of a trial and select just the positive findings for publication a practice known as “data dredging.” For instance, a trial of the effect of a drug on heart attacks may be converted to a trial of the drug’s effect on blood pressure, if it begins to look as if the drug isn’t much good for heart attacks. Or a one-year trial may be converted to a six-month trial if the drug looks good only in the first half of the study.
Researchers, even in our best universities, often go along with this sort of thing, because sponsors insist on it. A public registry of clinical trials would help eliminate many of these abuses, but only if it met the following three conditions:
First, it should contain all trials, not just those that the companies want to include, and it should be administered by a publicly accountable agency, such as the National Institutes of Health. Registration should be a condition of enrolling human subjects in research, since volunteers expect their participation to contribute to scientific knowledge and the public has a right to accountability. The argument that listing trials might reveal important proprietary secrets is not persuasive, because by the time drugs enter clinical trials, they are already patented and largely developed.
Second, it is crucial that trials be registered at inception, not after they are completed. The initial information should include a brief description of the design of the trial, including the kinds of patients to be enrolled, the drugs and doses to be tested, what the new drug will be compared with, the duration of the trial, the endpoints that will be measured, and the methods of measurement. That way any bias in the design of the study would be immediately apparent, and sponsors could not shift the endpoints as the trial progressed. At completion, the salient results should be reported. Not all trials are worth publishing in medical journals, but companies will still want their best studies to gain the added prestige and attention afforded by publication in a good peer-reviewed journal. That is fair only if information about all of the trials is available in the registry.
Third, the registry should identify the main researchers and disclose any financial ties they have to the companies whose drugs they are studying. Since drug companies don’t have direct access to patients, they depend on academic researchers and private doctors to supply human subjects and collect the data. Identifying them in the registry would hold them responsible for the quality of the research they perform.
The evidence of bias in medical reports is undeniable. It’s time to break the industry’s hold over the testing of its own products so that we will know whether new drugs are as effective as their makers claim. Talk of “miracles” and “innovation” is no substitute for evidence. The establishment of a government registry of clinical trials is a good way to begin, but it must be done right.