The Food and Drug Administration plans to update warning labels on antidepressants after a new review of study data showed a possible link between the drugs and suicidal tendencies among children.

The debate over whether the drugs are at fault, however, is unlikely to be settled by the FDA review. It continued to rage yesterday as doctors parsed the statistics in the 130-page report posted on the agency's Web site.

"The best estimate you can get out of it is that there is a significant increased risk," said Dr. Martin Teicher, a Harvard University associate professor who raised the question of a link in the early 1990s.

But, he said, "there's not enough statistical power to be absolutely certain, so people are going to equivocate and they're going to do nothing."

FDA officials, who disagree among themselves about what the data mean, will take the issue back to an expert committee on Sept. 13 for advice. Among other things, the FDA will ask the panel to advise it on what the new warning labels should say and what additional research to conduct.

Agency officials declined to be interviewed yesterday, preferring to let documents posted yesterday on the agency's Web site do the talking.

But the FDA warned that although there appears to be an increased risk of suicidal thoughts or actions with some of the drugs, unexplained inconsistencies remain - even among different clinical trials involving the same drug.

"It is important to note that the data are not 'black and white' in providing a clear and definitive answer to the question of a link between the drugs and pediatric suicidality," Dr. Thomas P. Laughren of the FDA said in a Monday memo.

Early complaints

The issue has been vexing for the agency since the late 1980s, when patients began to complain that Prozac - the first of a new class of antidepressants - prompted some adults who reported no previous suicidal thoughts to take violent action against themselves or others. Some have theorized that people are more likely to act on suicidal thoughts right after they begin taking a drug because they feel energized enough to act on their impulses.

The agency held hearings in 1991, after a paper by Teicher suggested a link between suicidal thoughts and Prozac. It found no proven connection.

Issue resurfaces

The FDA raised the question again last year, this time with respect to children, after one of its reviewers noticed a troubling "signal" in data submitted by GlaxoSmithKline on clinical trials of Paxil.

The FDA asked GlaxoSmithKline for more information, then requested similar data from all pharmaceutical companies that had conducted trials of drugs with children. Dr. Andrew D. Mosholder, an FDA reviewer, concluded then that children on the nine antidepressants he reviewed were nearly twice as likely to be suicidal than those who weren't taking one. The conclusions were based on 20 clinical trials involving more than 4,000 children.

He suggested to his colleagues that the FDA warn doctors not to write "off-label" prescriptions of the drugs for children, a widely used practice in which doctors prescribe drugs approved for one illness or one population for another.

At the time, Britain had already issued a similar warning.

Only Prozac, generically known as fluoxetine, has been approved as an effective treatment for major depression in children.

But an FDA advisory committee considering the issue in February never heard Mosholder's conclusions, which were suppressed by agency officials who disagreed with them and who didn't want to unnecessarily

worry or confuse patients. On the advisory committee's advice, the FDA did ask manufacturers of 10 drugs - Prozac, Zoloft, Paxil, Luvox, Celexa, Wellbutrin, Effexor, Serzone, Remeron and Lexapro - to update labels warning doctors to carefully monitor patients after starting them on antidepressants.

A furor erupted over the suppression of Mosholder's conclusions. The FDA responded yesterday by posting his findings and comments from FDA officials who disagreed with his conclusions.

'Statistically significant'

The recent FDA review was written by Dr. Tarek Hammad of the agency's neuropharmacological drug products division with statistical help from an expert panel at Columbia University.

Hammad concluded that no individual trial showed a statistically significant indicator for suicidality, a term that encompasses suicidal thoughts and actions. But he found that when trials were grouped together, "some of the overall estimates, across various groupings, were statistically significant."

Overall, the review said, patients taking the drugs in the trials were 1.78 times more likely to have serious suicidal thoughts or actions than those on placebo. Mosholder's earlier conclusion was that children on the drugs were 1.89 times as likely to be suicidal.

No patient committed suicide during the trials.

The new review "also indicates a statistically significant association of suicidal events with antidepressant drug treatment in short-term pediatric clinical trials," Mosholder concluded in a memo written Monday and posted yesterday.

Trials questioned

Disagreements over what the data mean are partly due to statistical confidence in the results, which are based on relatively small clinical trials, and questions about whether trial designs may have been at fault.

Dr. John Walkup, a Johns Hopkins University psychiatrist, said one possibility is that patients in the drug group of a study might be sicker than the patients in the placebo group, skewing results.

Teicher said there is no evidence of that. But all seem to agree that more study is needed.

"The next step is to look at clinical characteristics of those cases" to see whether other factors could explain the risk numbers, Walkup said. "So the story is not over."