This is the first of 3 posts laying out the philosophical basis for RxISK.org which will be live in the next few weeks. The others are Once is Never & the Unbearable lightness of being.
“[I suggest] a meeting with yourself and your reviewers. I have spoken in public on these issues and offered to speak on any platform. I’ve visited the MHRA [British equivalent of FDA]. Part of my motivation has stemmed from a willingness, indeed almost a desire to be proven wrong, as the consequences of not being proven wrong are in many respects horrific. If you have a reviewer who can point out the error of my ways I will take any steps that might be called for in the light of any conversion”.
The consequences of not being proven wrong are horrific
This cri de coeur (cry from the heart) is from a letter I sent to Richard Horton, the editor of the Lancet, in 2000, when his journal was reviewing an article I had submitted (see Healyprozac.com). The Lancet didn’t hear my desperation or if they did didn’t take up my offer. I had been to and went again to the MHRA who listened to what I had to say, but never offered any evidence to make me change my view that SSRIs can cause suicide.
The point at the heart of the Lancet article that led to my offer to visit and be persuaded of the error of my ways remains unanswered to this day. It was not about whether antidepressants can cause suicide but about how we assess adverse events. Suicide on antidepressants provides the perfect natural experiment to bring out the problem.
Suicide on antidepressants the perfect natural experiment
In 1990 Martin Teicher, Carol Glod and Jonathan Cole triggered concerns about a suicide risk on Prozac. They outlined 6 cases in which patients became suicidal after treatment started, the problem cleared when treatment stopped and re-emerged when treatment was restarted. In a series of subsequent reports other groups found the emergence of suicidality could be blocked by an antidote.
These reports met all the standard judicial, clinical and scientific criteria for determining cause and effect (see Let Them Eat Prozac). Prozac caused suicide.
No-one was prepared to say RCTs were simply not the way to answer this problem
But Lilly’s response in 1990/1991 was that their clinical trial data showed no risk, and that reports of patients becoming suicidal were just anecdotes. Lilly took their trial data to FDA and published it in an article in the BMJ. In an Evidence Based Medicine era where randomized controlled trials (RCTs) are described as offering gold standard evidence and clinical experience is dismissed as anecdotal this argument worked extraordinarily well. No-one was prepared to say RCTs were simply not the way to answer this problem.
But RCTs are not the way to answer a problem like antidepressant induced suicidality. The repetition that RCTs are a gold standard for everything and good case reports are anecdotes is deliberate propaganda. It’s the effect of repetition that causes us to agree with something that in fact stands common sense on its head.
No-one stepped up to the plate
Evidence based medicine and RCTs are supposed to help us control the pharmaceutical industry. One might have expected some of the advocates of RCTs and EBM to step up to the plate and say that Lilly were misusing the clinical trial data – these trials were not designed to test whether SSRIs caused suicide.
No-one stepped up to the plate then. Will they flinch now?
We live in a world where EBM and RCTs have been effectively deployed as a means to silence people injured by treatment. Too many adherents of EBM acknowledge and bemoan this but say it is down to pharmaceutical company “fraud” – lack of publication, miscoding of data, lack of access to the data etc. If the fraud were eliminated RCTs would deliver the goods.
Trickery plays a part but… RCTs are not the answer
While such trickery undoubtedly plays a part, when it comes to adverse events, RCTs are simply not the answer to determining cause and effect.
In the case of SSRIs and suicide, we can now see that the anecdotes so called were from the start completely consistent with the trial data which in the case of Prozac showed a clear doubling (2.0) in the relative risk of suicidal acts on Prozac compared to placebo. The only company defence as of the 1990s was that the published data was not statistically significant (see Psychotic doubt).
Some 14 years later when a sufficiently large cohort of trials were assembled the relative risk of a suicidal act on SSRI antidepressants became statistically significant. At this point, FDA stated that this statistically significant doubling of the relative risk (2.0) of a suicidal act demonstrated a causal effect (where a close to statistically significant doubling of risk had not).
FDA being plain bizarre
The idea that adding a few more patients to the mix should transform our views from this drug doesn’t cause suicidal acts to this drug causes suicidal acts should strike anyone who thinks about it as plain bizarre.
But here is Russell Katz, the Director of the Central Nervous System division of FDA saying essentially this when discussing the doubling of the relative risk of a suicidal act on anticonvulsants compared to placebo:
“We are unequivocally comfortable with using the word, the “c” word, with saying that this establishes causality. Again, we have talked about this a fair amount. This is how we determine causality, this is how we base our findings of effectiveness for drugs.
We do randomized trials, we analyze them prospectively, we have an outcome measure, and if it’s statistically significantly different from placebo, we say the drug caused it, you know, once you rule out chance and fraud and bias and that sort of thing, which we think we have done here”
There is a far deeper problem here than challenging FDA claims that they have ruled out chance, fraud and bias in these company trials in which it’s openly known some patients didn’t exist.
1958 – where the story really starts
To see the bigger problem we have to step back in time to a meeting in Cambridge, England in 1959. This was one of the earliest meetings at which psychiatrists had a chance to talk about their clinical experience with the first tricyclic antidepressant, imipramine. Imipramine had been introduced in 1958. Several of those there stated on the basis of their clinical observations that imipramine could cause suicide.
ECT had previously also been linked to suicide. ECT it was said and is still said could mobilize stuperose and suicidal patients so that they were able to carry out their plans before those plans began to fade. The doctors in Cambridge were aware of this “rollback” theory about ECT indirectly leading to suicides. But they said something else was happening on imipramine. It could directly cause suicide by increasing agitation. No one disagreed.
Unlike the later SSRIs, the tricyclic antidepressants could also bring about dramatic improvements in severe depression (melancholia). When the SSRIs were first produced in the 1970s, they very nearly didn’t become antidepressants because they were of no use in severe depression. Several companies almost binned their SSRIs after a series of studies showed them to be less effective than older non-selective serotonin reuptake inhibitors like imipramine, clomipramine and amitriptyline.
The tricyclic antidepressants are therefore effective in a patient group at a substantially higher risk of suicide than the patients entered into SSRI trials. Because SSRIs were ineffective in severe depression, the SSRI studies were conducted in mild to moderately depressed patients where there was so little risk of suicide from the illness that the risks from the drugs stood out.
How to hide a rabbit in a hat
So let’s mount a thought experiment. To pull a rabbit out of a hat, the first thing you have to do is put the rabbit in the hat. If we were to undertake a large placebo controlled trial of imipramine in severe depression we would expect the rate of suicidal acts on imipramine to be lower than the rate of suicidal acts on placebo because imipramine reduces the suicide risk from the illness more than any risk that stems from the drug. The relative risk of a suicidal act on imipramine in such trials would be less than 1.0, perhaps as low as 0.5.
Where would this leave FDA? Given the corner they have painted themselves into with the SSRIs, they could not agree imipramine causes suicide on the basis of trial data like this.
They cannot concede that imipramine causes suicide on the basis of the challenge-dechallenge–rechallenge reports from the 1950s or 1960s even though these criteria are still embodied in standard adverse event causality metrics – because they didn’t concede that SSRIs cause suicide on the basis of even better reports.
FDA and the politics of suicide
But FDA has in fact labelled imipramine and other tricyclic antidepressants as causing suicide. They did so for political reasons. In 2004, they put a class-wide labelling for suicide risk on all antidepressants so that the makers of the latest drugs would not be disadvantaged. They had in fact offered to put a class-wide warning on all antidepressants as early as 1990.
It must be clear from this that we are about to descend into a morass. This sequence of events poses far more problems than might be immediately apparent.
To be continued.