Paxil, Glaxo SmithKline and antidepressant's studies fraud. Two Studies, Two Results, and a Debate Over a Drug

"That particular study would have been buried," said that researcher, Dr. Robert Milin of the Royal Ottawa Hospital in Canada. "It would have been buried to the public."

Federal regulators in this country are now scrambling to reassess the effectiveness and safety of antidepressants like Paxil, after British regulators touched off a controversy last year by asking drug companies for unpublished data from antidepressant trials. That data suggested that several antidepressants, including Paxil, might give rise to suicidal thoughts in some young users - a potential problem not revealed in any published studies.

Yesterday, the New York State attorney general, Eliot Spitzer, entered the fray by taking the unusual step of suing Paxil's maker, which is now GlaxoSmithKline. Mr. Spitzer's suit accuses the company of consumer fraud for not disclosing all of its Paxil data.

Officials of GlaxoSmithKline defend their record, saying they provided all the results of their Paxil clinical trials to the Food and Drug Administration, as required by law. But the stories of Study 329 and Study 377 provide a window into a far broader issue - the fact that the results of many human clinical trials of drugs are often never widely publicized and that, in some cases, doctors may never learn that the trials were even conducted.

These days, most drug trials are sponsored by pharmaceutical companies. And for more than a decade, a growing number of medical experts have been urging drug makers to release more trial data and to create uniform means of disclosing results through central registries, so that policy makers and doctors can easily learn the results. Those advocates argue that such central databases are necessary because drug companies, as well as medical journals and researchers, tend to spotlight only trials that show positive results.

Last week, GlaxoSmithKline agreed to make two executives available for this article to discuss its handling of Studies 329 and 377 and how information about them was disseminated. But yesterday, a company spokeswoman, Mary Anne Rhyne, said that in light of Mr. Spitzer's lawsuit the company had decided not to allow those interviews.

Under F.D.A. rules, a company seeking approval of a new drug must submit the results of all the clinical trials it runs. But the agency holds that information in confidence - on the ground that it is proprietary - until the drug is approved for sale. At that point, summaries and descriptions of those trials, but not the complete underlying data, are publicly released.

Somewhat different rules apply to so-called postmarketing tests in which approved drugs are investigated for new purposes - like prescribing antidepressants for pediatric use. Safety information from such trials is supposed to be promptly reported to the F.D.A. But postmarketing data involving the drug's effectiveness need not be reported until the maker seeks agency approval for a new claim or use, a process that could take years.

"We fully understand the desire for access to information and we firmly believe that consumers should be as well informed as possible," said an F.D.A. official, who insisted on anonymity. "However, such a listing would not add the information F.D.A. already receives under current regulations." For their part, drug industry officials say that it is the editors of medical journals, not corporate executives, who decide which trial reports to publish. With a few exceptions, drug makers have resisted the idea of establishing trial registries. And the industry's trade group, the Pharmaceutical Research and Manufacturers of America, has never called for such an initiative. But Dr. Alan Goldhammer, the group's associate vice president for regulatory affairs, said the group might revisit the issue in light of the antidepressant controversy.

Currently, a few trial registries exist, like the National Institutes of Health's database of current trials of drugs for life-threatening diseases. And in two weeks, policy makers at the American Medical Association are expected to vote on a proposal that would urge the government to create a public registry of clinical drug trials. Only a few companies have responded to the calls by some researchers for public registries. In 1996, the British unit of Schering, the German drug maker, agreed to create such a listing. The British company Glaxo Wellcome decided in 1998 to take a similar step. Two years later the company merged with SmithKline Beecham, and officials of GlaxoSmithKline said in a brief statement yesterday that they were now working on an improved version of the registry.

The drive to create databases of clinical trial results soon sputtered for lack of industry support, according to people involved in the effort. But even industry critics acknowledge that drug makers are not the only roadblocks to the wider dissemination of trial results. Journals also favor research reports that show strong findings, and researchers say they have little incentive to push for the publication of a trial without a conclusive finding.

Those who call for trial registries, however, argue that it is unethical not to disseminate trial results widely. The participating patients, they say, typically believe that the findings, whatever the outcome, will benefit medicine. "We are telling people that they are participating in research," said Dr. Kevin A. Schulman, a professor of medicine and business administration at Duke University. "But most of the time what they are participating in is proprietary marketing research by companies who have total discretion over whether to publish the results."

It is unclear how, or even if, a database of trial results would change the practice of medicine. Some researchers say they can often retrieve unpublished data by asking companies for it or by ferreting out abstracts of unpublicized findings presented at meetings.

Still, they point out that without a comprehensive trial registry they cannot tell what they may be missing. Just as important, some experts say, a system of disclosing trial results would help prevent the type of chaos now surrounding the use of pediatric antidepressants because the debate over a drug's efficacy and safety might have already taken place.

The fates of Study 329 and Study 377 appear to underscore that point. Both tests were conducted during the mid-1990's at various hospitals and medical centers - Study 329 at facilities in the United States and Study 377 at test centers outside of this country, including Canada, Mexico, Europe, South Africa and the United Arab Emirates.

Study 329, with its potentially positive findings for Paxil, was completed first. Its results were presented beginning in 1998 at several meetings of medical professionals. Meanwhile, a report of the trial, which was led by Dr. Martin B. Keller, a department chairman at Brown University Medical School, was submitted to a medical journal for publication, a process that subjects a study to peer review by scientists not involved in the trial.

Dr. Keller declined to be interviewed for this article. But Dr. Neil Ryan, a professor at the University of Pittsburgh, who was also involved, said he believed that the study was rejected by some journals before The Journal of the Academy of Child and Adolescent Psychiatry accepted it for publication in 2001.

In the case of Study 377, the one with negative findings, there were no press releases or publications. And without the action of Dr. Milin and a Canadian colleague, Dr. Jovan Simeon, the study's findings might have been seen only by regulators and a few researchers.

Dr. Milin was an unlikely rabble-rouser. In an interview, he said he was a strong believer in the use of antidepressants like Paxil in adolescents. He wanted to report the study's findings, he said, mainly because its negative results might have reflected trial design flaws that he did not want to see repeated in other studies. "I feel you need to present all the data even if it is negative," he said.

While drug trials in adults take place at a few sites, Dr. Milin and other researchers said that one problem with the pediatric antidepressant tests was that they were dispersed across a dozen or more clinical centers because each unit often had only a few young patients who qualified. Dr. Milin said he was spurred to take action after SmithKline officials told him in 1998 that they did not intend to submit the study for publication. An internal 1998 SmithKline memorandum, disclosed this year during the antidepressant controversy, also said the company had "no plans to publish data from Study 377."

Dr. Milin said that when he told SmithKline executives that he planned to present the study's finding at a 1999 meeting of the American Academy of Child and Adolescent Psychiatry, the company did not object. Dr. Milin said he last heard from GlaxoSmithKline officials about six months ago, after the controversy erupted over unpublished data from trials like Study 377. Company officials, he said, wanted to make sure that the copy of the report they had in their file was the same one he had presented five years earlier.

As for Study 377, Dr. Milin said he assumed that SmithKline officials would have publicized the trial, any design problems notwithstanding, had its results been different.

"If they had got a positive outcome," he said, "I would suspect that they would have pushed to get it published."