The Washington Post
April 18, 2004, Pg. A01
Antidepressant Use in Children Soars Despite Efficacy Doubts
By Shankar Vedantam
The number of depressed American children being treated with antidepressants has soared over the past decade -- a tectonic shift in the practice of psychiatry -- but new scientific reviews of the research that fueled the trend suggest that the drugs' benefits have been dramatically oversold.
The use of antidepressants among children grew three- to tenfold between 1987 and 1996, data from various studies indicate, and a newer survey found a further 50 percent rise in prescriptions between 1998 and 2002. The explosion in antidepressant use occurred even though the vast majority of clinical trials have failed to prove that the medicines help depressed children.
The spike in prescriptions over the past five years has been especially sharp among children younger than 6, even though there is virtually no clinical trial data on these youngest patients.
Paradoxically, drugs that have never shown benefits for depressed children in clinical trials have some of the largest increases in prescription rates. Pediatric prescriptions for Paxil, for example, doubled between 1998 and 2002, even though the medicine failed to show it was any better than dummy pills in three trials. The drug has not been approved for use in children, and last year the Food and Drug Administration and British health authorities warned physicians not to prescribe Paxil for children, citing safety concerns.
Paxil is not alone. Of 15 trials conducted among depressed children, 10 failed to show antidepressants were better than dummy pills. Two were inconclusive, and three showed positive results. The negative results have mostly been withheld from public scrutiny by the pharmaceutical companies that paid for the trials, which say that the data are proprietary.
Although many psychiatrists swear by the drugs in children and adults, leading specialists agree they have limitations.
"These drugs are by and large efficacious, but they are only moderately efficacious," said Steven Hyman, former chief of the National Institute of Mental Health.
"A lot of clinical trials for antidepressants fail," added Hyman, now provost at Harvard University. "Partly that's the difficulty of trials in a waxing and waning disease, but we also need drugs of greater intrinsic efficacy."
Prozac remains the only antidepressant that the FDA has approved for children's depression, after the agency accepted two studies that demonstrated the drug worked better than dummy pills.
But an FDA internal analysis of the trials found Prozac failed on the statistical measure that researchers had originally chosen as their primary benchmark: "The evidence for efficacy based on the pre-specified endpoint is not convincing."
Senior officials at the agency, however, concluded that the improvement on another measure justified approval. For one of the studies, a senior official, Russell Katz, wrote in July 2001 that "one could argue that this post hoc choice of primary outcome is inappropriate," but in the end he and others said that this was the proper benchmark.
Australian researchers writing this month in the British Medical Journal reviewed the published studies of Prozac and other drugs and concluded they were consistently weak. The review charged that researchers doing the studies had selectively dramatized successes and glossed over problems.
Another analysis this month in the journal Psychiatric Services said the drugs had only modest benefits, with "many treated patients continuing to experience symptoms." The report tracked antidepressant prescription increases among children with private health insurance.
"When a patient takes a medicine or a family physician or a pediatrician prescribes medicine, their understanding when they hear this medicine works is they believe, 'My child will recover from depression,' " said Jane Garland, head of the mood and anxiety disorders clinic at the British Columbia Children's Hospital in Vancouver.
"But the data says they are not going to get any better than on a placebo," she said. "They will have some improvement in symptoms, which is a good thing, but it means there is clearly more than medication needed for treatment."
Concerns over the quality of the data have been heightened by a recent warning by British health authorities that urged caution in using the drugs, citing indications that they may cause suicidal behavior.
"The risk-benefit ratio starts to look dodgy," Jon Jureidini, a child psychiatrist at the Women's and Children's Hospital in Adelaide, said in an interview. "But if you look at the published literature, you can be forgiven for not reaching that conclusion."
The American psychiatric establishment firmly supports the drugs -- even those not specifically approved by the FDA for children. Psychiatrists say children's depression is severely undertreated.
Most psychiatrists say that the fears about suicide risk are overreactions. Patients who suddenly stop taking medicine without consulting their doctors could put themselves at risk, Harvard's Hyman and other physicians say. Many doctors are convinced the drugs save lives.
The review in Psychiatric Services said that growing awareness about depression, better diagnosis and incentives by insurance companies in favor of medication rather than talk therapy may have fueled the rise in drug treatment. Although the drugs are also used for anxiety and other conditions, depression accounts for the majority, the review said.
The report also cited doctors' belief that positive data from adult studies can be extrapolated to children. In a 1999 letter to Prozac's manufacturer, however, the FDA expressed "substantial concern about the ability to extrapolate positive antidepressant findings from adult to pediatric patients."
"I'm not anti-drug, but I don't know what to believe," said Wayne Blackmon, a Washington psychiatrist who worries that clinicians have been fed misleading data. "Once you start delving into it, you start going, 'Oh no, no, no -- this is not valid.' "
Across the board in clinical trials of antidepressants, about half of all depressed children improve whether they are on a drug or a placebo (dummy pill). And new evidence suggests that the placebo effect -- the tendency to get better when patients believe a treatment will help -- may be even greater in the real world, because patients deemed susceptible to placebos are screened out of clinical trials.
While supporters and critics of the medicines present the issue in black-and-white terms, the data from clinical trials paint a complicated picture.
In one of the two trials of Prozac used to win FDA approval, for example, the original benchmark was recovery -- how many depressed children recovered on Prozac compared with dummy pills. The difference was not statistically significant.
But Prozac's manufacturer, Eli Lilly & Co., then evaluated how many children improved by 30 percent on a commonly used scale to measure depression. Among children taking Prozac, 58.3 percent had a 30 percent improvement, whereas only 31.9 percent of those on dummy pills improved that much. By this new measure, the difference was statistically significant, and the company claimed success.
The internal FDA statistical analysis, however, found the difference vanished when officials looked at how many children improved by 10 percent. And there was again no difference when they evaluated how many children had a 50 percent improvement: "The largest treatment effects was found when 20 percent or 30 percent cut-off points were chosen," an FDA statistician wrote.
Within the FDA, officials also worried that the group of depressed children who got Prozac included a large number who also suffered from anxiety, raising questions about the validity of the results. FDA officials also said a nurse with access to codes showing who got the drug was involved in evaluating two patients, a potential bias.
In the end, however, senior FDA officials concluded that the drug had succeeded on yet another measure, which they said was the best way to evaluate antidepressants -- the average improvement on the children's depression scale.
FDA officials Thomas Laughren, team leader for psychiatric drug products, and Robert Temple, associate director of medical policy, said in a recent interview that the children with anxiety did not undermine the result. While the study had been conducted as an academic research project, not an industry-sponsored trial, FDA officials said they had confidence in it.
"There is nothing to suggest this was not a rigorously conducted study," Laughren said. Eli Lilly spokeswoman Jennifer Yoder said the company stands behind its drug.
Part of the problem for doctors and parents trying to evaluate the data is that experts starkly disagree on the statistics. Graham Emslie, for example, a professor of psychiatry at the University of Texas Southwestern Medical Center, who conducted the two positive trials for Prozac, said six antidepressant studies in children showed benefits; the FDA counted three.
Emslie co-chaired a panel of the American College of Neuropsychopharmacology that declared in January that the drugs are safe and effective for depressed children.
However, at least one manufacturer, Wyeth, has itself told doctors not to prescribe its drug Effexor for children. Philip Perera, medical director at Glaxo SmithKline, which makes Paxil, said: "Our point of view with respect to pediatrics is that it is still up in the air."
Jureidini's analysis in the British Medical Journal, which examined six published trials for Prozac, Paxil, Zoloft and Effexor, found that of 42 measures used to evaluate patients in these studies, only 14 showed a statistical advantage for the medicine over placebo.
The psychiatrist, who himself prescribes the medicines to children -- but rarely -- said the moderate benefits of the drugs have been oversold. He blamed pharmaceutical industry marketing and the alliances the industry has made with top psychiatrists: Once prominent doctors said they supported the medicines, general practitioners and the public accepted the conclusion, Jureidini said.
The researcher said that doctors had a subtle -- but powerful -- bias: "There's this kind of view that we all know antidepressants work and if the research doesn't support that, there must be something wrong with the research."
§ § §