Unpublished data reverses risk-benefit of drugs
NewScientist.com news service
16:20 23 April 04
Unpublished studies on the effects of anti-depressant drugs on children suggest some are both ineffective and potentially harmful, according to a new review of research. The unpublished data contradict published results, fuelling the debate on how pharmaceutical companies reveal trial data.
The new study was conducted by Tim Kendall, deputy director of the Royal College of Psychiatrists' Research Unit in London, UK, and colleagues. One of Kendall's roles is to help analyse medical research to draw up the clinical guidelines on which the UK government bases its drug regulations.
After recent revelations that drug companies have suppressed unfavourable data on their drugs, Kendall and his colleagues contacted pharmaceutical companies requesting unpublished studies that might bear on the guidelines. None of the companies complied, so Kendall contacted a government agency which provided six unpublished studies on three anti-depressants known as selective serotonin reuptake inhibitors (SSRIs).
Kendall and his team added those results to a review of five published studies on the effects of various SSRIs on children.
"When we got the unpublished data and put it in with the published data, something happened. Instead of being safe and effective, the risk-benefit reversed," Kendall told New Scientist.
Of the five SSRIs reviewed - fluoxetine, paroxetine, sertraline, citalopram, and venlafaxine, only fluoxetine (Prozac) offers more benefits than risks in children. Unpublished studies of venlafaxine, for example, suggested the drug increased suicide-related events such as suicidal thoughts or attempts by 14 times compared with placebo.
"This data confirms what we found in adults with mild to moderate depression: SSRIs are no better than placebo, and there is no point in using something that increases the risk of suicide," says Kendall. "The key point is, can we trust the published evidence now?"
An editorial in The Lancet, which published Kendall's study, suggests the answer is no. Research on SSRIs in children is marked by "confusion, manipulation, and institutional failure", it states.
Analyses of published data - which governments rely on to set regulations - are "made entirely redundant if [the] results are so easily manipulated by those with potentially massive financial gains".
The editorial reports that GlaxoSmithKline sold almost $5 billion worth of its SSRI paroxetine (known as Seroxat or Paxil) in 2003. In June 2003, paroxetene was shown to increase suicidal thoughts and behaviour in children by as much as three-fold over placebo.
And in March, a GlaxoSmithKline memo on the drug's effect on children surfaced that read: "It would be unacceptable to include a statement that efficacy had not been demonstrated, as this would undermine the profile of paroxetine."
In 2003, the UK government's Committee on Safety of Medicines banned the use of all SSRIs except fluoxetine in children. But the government estimates that half of the 40,000 children and adolescents in the UK taking anti-depressants are using other SSRIs "off licence".
"Clearly, if so many children are being prescribed SSRIs, then all data must be made available to properly address the balance of risks and benefits," says Kendall's co-author Craig Whittington of University College London.
Richard Ley, a spokesman for the Association for the British Pharmaceutical Industry, says it has set up a new website for drug companies to voluntarily report the results of clinical trials done in the UK.
And he says a UK law should take effect in May that forces drug companies to provide some information on drug trials to government regulatory agencies - though not to the public at large.
"We're very keen to extend the information that's available - within reason," Ley told New Scientist. "It costs about £500 million pounds to develop a single new medicine, and it takes 10 to 12 years. You cannot afford to give all the information you have away free."
Journal reference: The Lancet (vol 363, p 1341)