By Jerome Burne
There can be few people who think that putting an increasing number of children on SSRI anti-depressants is really a good idea but then reflect that it’s just one of those things. Cash strapped NHS; time-poor GPs; waiting lists of months for therapy; drugs cheap; they may help some.
But we know these drugs can make children suicidal and that they are marginally effective at best. If they did work well, the drugs companies wouldn’t have had to go to such extraordinary lengths in torturing statistics and rigging trials to produce favourable results. Some of the ways this is routinely done is revealed in a feature of mine in the Daily Mail today.
But only one of these drugs – Prozac (fluoxetine) – is licensed for children and adolescents under the age of 18 on the grounds that it is more likely to help and less likely to harm than the others. But is it?
Recently I was sent documents which strongly suggest that this is a fallacy; that Prozac’s supposed greater benefit and safety for children is the result of just as vigorous data rigging as all the rest.
Bringing up the bodies
The Daily Mail feature is based on research, published in the BMJ last month by Professor Peter Gotzche, an expert in drug data manipulation. He describes what he found when he burrowed into the full data collected during trials of half a dozen SSRIs. These reports often run to thousands of pages and contain details of trials never normally seen. It is here the bodies are buried, out of sight of doctors and their patients.
The number of children and adolescents being prescribed an antidepressant has doubled in the last decade – an estimated 100,000 prescriptions in the UK. Presumably the majority of them are for Prozac.
So it is appalling that the data I’ve been sent is not secret or unknown, it has simply been ignored. One part is the evidence used to licence Prozac in the first place in 2001, which now looks remarkably shaky. The other is a journal article published last year that shows how key studies that supposedly support Prozac were very misleading.
The journal article is by a Swedish researcher – Dr Gӧran Hӧgberg of the Child and Adolescent Psychiatric unit at the Karolinska Institute in Stockholm. Five years ago he was shocked at the results of doing something very simple. He compared the abstracts (summaries) of seven trials that have given Prozac the green light, with what was in the full report. There were major differences. The abstracts – all that most medics read – painted a much rosier picture by leaving out key findings.
Suicidal events with Prozac higher than expected
In the detached understated language of science, this is what he found. ‘We concluded that a major, albeit underreported, finding….was the significant increase of suicidal events in in the adolescent on antidepressant medication in comparison with the group on the placebo.’
Notice the understatement – ‘albeit underreported’ Just how much greater was the risk if you were an adolescent on Prozac, the safest SSRI, compared with a placebo? ‘The proportions of suicidal events were 11% and 2.7% respectively.’ So, more than three times greater.
The discrepancies Hӧgberg spotted were subtle but the cumulative effect was to downplay the risks of suicidal events on the drug. For instance in the first study in 2004 the abstract reported that nearly a third of the children had suicidal thoughts at the start of the trial but that they all improved significantly whichever treatment they got. Impressive.
What was missed out was that there were more reports of suicidal events in the Prozac group than among those on a placebo
Leaving out the comparison between the drug and the placebo, which is the clearest way of isolating the effect of the drug, was also a failing of several other abstracts.
Seventeen suicides not mentioned
The most striking omission came in a study published in 2009 entitled: ‘Assessment of safety and long-term outcomes’. The abstract said: ‘No difference between the groups in rates of suicidal events.’ In fact the study had found 17 suicidal events in the in the Prozac group and none in those getting either CBT (cognitive behaviour therapy) or a placebo.
Hӧgberg makes a final point: “None of the seven abstracts …mentioned the fact that there were four times more suicidal events with fluoxetine than with the placebo …this difference was statistically significant.”
In other words the information about Prozac and children that should, at the very least, have set alarm bells ringing in the offices of psychiatrists and regulators, was there in the research they were relying on all along. They just didn’t notice.
It’s worth noting a point about the language used in this research. The ‘suicidal events’ which Hӧgberg refers to, include thinking about suicide, behaviour that involves preparing for suicide and actual suicide.
A confusion that is killing children
There is a professional debate around the relationship between thinking and preparing and doing it. Professor Gotzsche is clear on this: “Not everyone who thinks about it or prepares for it actually commits suicide but far too many do, he says “This confusion is killing children.”
It’s also worth making it clear how influential these studies had been. They all came from a major, long running study known as TADS (the Treatment of Adolescent with Depression Study), funded by the American National Institutes of Mental Health. It cost 17 million dollars and included 439 children between 12 and 17 who had been diagnosed with major depression. It looked like research you could believe in.
But TADS isn’t important just because it was big and expensive; it also played a major role in creating the notion that Prozac was the go-to drug for depressed kids. When the first report from TADS came out in 2004 the whole idea of giving antidepressants to children was under heavy attack. GlaxoSmithKline had been discovered hiding trials that showed Seroxat wasn’t effective for children and raised their risk of suicide.
That year the FDA had held hearings on increased risk of suicide in children with SSRIs and ordered a black box warning about it. But Prozac had already got a licence even though critics at the time claimed that this was more to do with the favourable way the trials were designed than genuine superior performance.
Concerns about Prozac safety date back ten years
The results of that first TADS study made it look like Prozac did the business. It was a big trial and after twelve weeks the kids getting a combination of Prozac and CBT had a response rate of 71% compared to 35% for the placebo.
Critics at the time pointed out that during the trials in involving CBT the researchers hadn’t been blinded (they’d known who had been getting CBT) and that when Prozac and a placebo had been directly compared there had been no difference. It’s depressing to note that this is the same criticism that Hӧgberg was making ten years later. But back in 2004 TABS turned the tide.
So what about the data that was used to get Prozac licenced for children in 2001? How good was it? It’s still important because the revelations about suicide with SSRIs didn’t become public until 2002/3. So ironically although Prozac is presented as the drug that is least likely to raise children’s suicide risk the trials used to licence it didn’t consider suicide at all.
This is clear from the other document I received recently which was crucial to the licensing of Prozac for kids. It is an assessment by an FDA (Food and Drug Administration) expert of the reliability of the trials sent to them by the makers of Prozac – Eli Lilly – to prove it was safe and effective.
Prozac fails to shine in licensing trials
Its bio-medical statistics are way above my pay grade but certain points from the review stand out. It concludes that Prozac should be approved but it is far from an enthusiastic endorsement.
Not only did the drug fail to meet its primary goal – to be more effective than a comparable drug at improving subjects’ depression scores – but it barely succeeded in its other aim, to increase the time before a relapse. The reviewer wryly commented that promoting the drug for this purpose ‘may not be appropriate’.
What’s striking, given these failures is how far the company went to rig the trials to boost their chances – patients who responded well to a placebo were excluded and so were those who responded poorly to the drug. But even these tricks, which Professor Gotzsche also found in his recent BMJ review, were not enough to provide a convincing win.
And there was something else that makes you wonder why anyone thought Prozac was beneficial for depressed children in general, let alone safer. There was a sub-group of patients with anxiety as well as depression in the trials, who responded particularly well to the drug. Without their positive ratings Prozac’s benefit dropped significantly.
Could Prozac stunt a child’s growth?
There was evidence from the trials that Prozac stunted growth which worried the reviewer that it might have damaging effects on ‘growth development and maturation with longer exposure’.
After all these reservations it is curious that the reviewer baldly concluded: ‘I believe the trials provide evidence that fluoxetine benefits pediatric patients with depression.’
It might be relevant to point out that the scientist who did both of these Prozac licencing trials was Dr Graham Emslie of University of Texas-Southwestern Medical Center. Emslie was also one of the authors on the infamous Study 329 which claimed that another SSRI – Seroxat – was safe and well tolerated for use on children when it found nothing of the sort.
A later analysis found there were ‘clinically significant increases in harms, including suicidal ideation and behaviour’. The company GSK was eventually fined three billion dollars for fraudulently promoting Seroxat. There are more details on Study 329 here.
Time for an independent review
The subsequent history of the FDA reviewer, Dr Andrew Mosholder, is also interesting. In2003, he discovered that antidepressants led some children to become suicidal. His findings were leaked to the press and the FDA responded by preventing Dr Mossholder from speaking to an advisory committee about his analysis – very possibly the one on suicides held in 2004. A year later an independent investigation concluded he had been right.
It’s impossible not to wonder how many lives might have been saved had Dr Mosholder known about the suicide link before he nodded Prozac through as suitable for children.
I’m not the person to make a final judgement on Prozac and children but given the rate at which its use and the death toll is rising, it is surely time for an independent review to decide if the drug’s special status is still justified.