FDA Revisits Issue
New Analysis May Pressure
Agency to Set Limit on Use
Because of Suicide Risk
By ANNA WILDE
A new Food and Drug Administration analysis of clinical-trial data shows evidence of a link between antidepressant drugs and suicidal tendencies among young people. The finding is likely to reignite debate over the issue and the agency's handling of it.
The new analysis, which focused on 25 studies of nine drugs, found that children and teenagers who took the medicines were more likely to have behavior or thoughts that appeared suicidal, compared with those who got placebo pills. When the analysis used a far broader category of incidents that might possibly reflect suicidal intent -- some of them classified as "not enough information" -- the young people who took the drugs were twice as likely to engage in the behavior.
What makes the issue so controversial is how the FDA has dealt with an internal debate on the matter. Congressional investigators have been examining the agency's decision to block a staffer from revealing this past February his finding of a tie between antidepressants and suicidal tendencies in young people, which drew on many of the same trials. FDA officials said his conclusion was premature, and based on ambiguous data. The new FDA analysis relies on an in-depth review of the data by outside researchers.
The new findings are likely to intensify pressure on the agency to consider stronger action to limit use of the drugs by young people. The FDA is scheduled to present a final version of the analysis to an advisory committee in September, asking for its recommendations on what safety steps it needs to take.
The FDA's new analysis, which hasn't been made public, was contained in a draft document reviewed by The Wall Street Journal. An FDA spokeswoman said the agency was still reviewing the data, and couldn't comment before the committee meeting.
In March, the agency called on makers of major antidepressant drugs to beef up warnings on their labels urging doctors to watch all patients for indications of increasing depression or suicidal thinking. In Britain, regulators have gone further and specifically discouraged the use of a number of the drugs for depressed patients under the age of 18.
The new analysis by an FDA medical reviewer tabulates the results of an outside re-examination, requested by the agency, of the clinical-trial data. The document, dated July 19, focuses on an array of antidepressant drugs including Pfizer Inc.'s Zoloft, Forest Laboratories Inc.'s Celexa, Wyeth's Effexor, GlaxoSmithKline PLC's Wellbutrin and Paxil, and Prozac, made by Eli Lilly & Co. and available generically.
The results varied sharply from drug to drug, and even multiple trials of the same drugs sometimes found differing levels of possible risk. Also, some of the results weren't strong enough to achieve statistical significance. No participants in any of the trials actually completed a suicide attempt.
Nevertheless, the document's findings appear to be consistent with the controversial earlier analysis done by an FDA epidemiologist, Andrew Mosholder. He wasn't allowed to publicly present his conclusions at an advisory-committee meeting in February. He shared his views with agency officials as early as last December.
The issue has taken on more urgency as antidepressant prescriptions for young people have grown, even though few studies have been published for most of the drugs demonstrating that they work in that population. According to an analysis by Washington State University researchers, the rate of antidepressant prescriptions for children and adolescents more than tripled in the U.S. from the early 1990s to 2001. In 2002, an estimated 10.8 million prescriptions for the most widely used antidepressants were dispensed for patients under 18 years old, according to an FDA analysis. Some psychiatrists credit the drugs for the sharp decline in the suicide rate among young people since 1994.
Drug-industry studies of antidepressants in young people have also become the crux of a broader debate over the disclosure of negative clinical trial results. In June, New York State Attorney General Eliot Spitzer sued GlaxoSmithKline, alleging that the company concealed findings in pediatric tests of Paxil.
The FDA has said it felt Dr. Mosholder's findings were premature and based on ambiguous data. Last February, at the advisory-committee meeting, the agency did reveal the underlying data examined by Dr. Mosholder, but not the staffer's conclusions. Two congressional committees are currently examining the agency's handling of the epidemiologist's report. Dr. Mosholder declined to comment.
To get a more definitive answer about possible ties between the drugs and suicidal behavior, the FDA commissioned a new examination of the pediatric-trial data by an outside team of reviewers led by Columbia University researchers. The point was to sort through the incidents in the trials and figure out which ones truly reflected suicidal tendencies. Dr. Mosholder wrote that this re-evaluation was unlikely to significantly change his conclusion, but a number of FDA scientists said that the ambiguous data he used could produce a misleading result. The new analysis was supposed to be more definitive.
Now, the reclassified data appears to show evidence of a link between the drugs and suicidal tendencies in the pediatric patients who participated in the trials. The July document was prepared by a medical reviewer named Tarek A. Hammad, who works in the division of the FDA's drug center that reviews psychiatric drugs.
The analysis parses the data in different ways, and the results differ depending on which drugs are involved, which conditions were being treated in the trial and how Dr. Hammad selected the incidents he was examining. When he looked at all of the trials lumped together, searching for incidents that were classified as a "suicide attempt," "preparatory actions towards imminent suicidal behavior," or "suicidal ideation," he found that young people who took the drugs were 1.78 times as likely to show those behaviors as those who got placebos.
He also analyzed the same trials with a far broader category of incidents, including those that couldn't be defined and some that involved self-harm but not necessarily suicidal intent. Then, patients who took the drugs were 2.06 times as likely to have such events as placebo recipients.
Other findings weren't so definitive. When Dr. Hammad looked separately at suicidal behavior and suicidal thinking, he found the drug recipients had a higher rate of incidents than the placebo patients. But those results didn't achieve statistical significance, potentially because of the smaller number of events involved.
In another part of the document, Dr. Hammad examined trials for depression, and only those with drugs that were selective serotonin reuptake inhibitors, or five of the nine. Here, the results were ambiguous: The drug recipients were 1.41 times as likely to exhibit the incidents that more clearly reflected suicidal intent as those receiving placebos. That result didn't achieve statistical significance.
More confusing, for regulators and doctors, was the fact that different drugs showed varying results, making it hard to paint a clear picture. Some trials of both Paxil and Effexor appeared to show strong evidence of a link between the drugs and suicidal tendencies. Both drugs have already been the subject of public warnings. But in some cases, other drugs' trials reflected little or no difference between patients who took placebos and patients who got real medicine.
That variability is consistent with earlier examinations of the trial data by FDA officials, including Dr. Mosholder. Dr. Mosholder, in a report dated in February, urged that the agency take steps to discourage pediatric use of the drugs, except in the limited cases mentioned on the labels of some of them. Essentially, this would have meant the agency would recommend that doctors use only Prozac for depressed young patients. This was similar to the recommendation of British regulators.
Write to Anna Wilde Mathews at firstname.lastname@example.org
Contact Us Help E-Mail Setup Customer Service: Online | Print