Editorial Note: This is the first of four Trick or Treat posts. They make most sense when read in conjunction with the RxISK Prize posts especially the series of 3 posts starting tomorrow.
In 1962 RCTs were added to the regulatory requirement for marketing medicines. This looked like a definitive stake through the heart of hucksterism. No longer would we have to carry crucifixes and garlic around to ward off blood-suckers hawking ineffective remedies.
In 1962 doctors were viewed as the key element of the regulatory apparatus. A great part of their input centred on descriptions of the harms medicines could cause. In addition to news of the latest advances, medical meetings were filled with symposia and medical journals with articles on the harms of treatment and how to manage them.
Company funded RCTs, which played a minor part in clinical practice, were all done in hospitals and universities and academics had the data.
Around 1980, as new drugs got weaker and weaker, company trials became multi-centered, coordinated by CROs, written up by ghost writers, with academic names attached afterwards. Everyone seemed as happy to enable this increasingly Hollywood like remake of medicine as they were to enable Harvey Weinstein. In the process access to trial data was lost.
The key event that marked a transition to the modern world came in 1990, when Teicher et al published 6 cases of people becoming suicidal on Prozac in the American Journal of Psychiatry. According to all canons of causality, this paper offered undeniable evidence SSRIs can cause suicide. Hearings were convened to discuss the need for warnings.
But coincident with these hearings, BMJ published a meta-analysis of Prozac trials which Lilly claimed showed Prozac was not linked to suicidal events. This paper had been rejected elsewhere, The BMJ reviewer was lukewarm. But Richard Smith, BMJ’s editor, perhaps spotting an opportunity to push a new kid in town, Evidence Based Medicine, and its shiny new meta-analytic machine, published.
The published data showed an increased risk on Prozac, which Lilly and BMJ ignored, claiming nothing was statistically significant. Beyond this, Lilly played some of the tricks other companies later played – the small print shows the only placebo event hadn’t happened on placebo, so that technically there was a statistically significant infinitely greater risk on Prozac.
The follow-up letters told RS he had been naïve and wrong. But no one accused him of conflict of interest. What could be wrong with letting good quality evidence (even though he got that wrong) triumph over anecdote?
The way this played in public was that the stories of suicides on Prozac were tragic but anecdotal. The scientific evidence demonstrated that patients and doctors just can’t believe the evidence of their own eyes and ears. They have to be told what’s what by experts.
This dangerous and misguided message triumphed with regulators, and later in Courts.
This message killed any interest journals like AJP and BMJ had in taking Case Reports. Besides companies didn’t buy reprints of these, whereas they handed over huge amounts of money for reprints of ghost-written fraudulent RCTs with zero access to data, and even more for the best science money could buy – meta-analyses of these trials. Evidence Based Marketing was here.
A medicine is a chemical plus information. The information about the effects of chemicals should come primarily from practice on the ground, as in the early days of antibiotic or psychotropic drug use, or street drug use now, or from use of anti-retroviral combinations for AIDS in the 1990s.
When the benefit (not the harms) of a drug is equivocal, RCTs offer a means of examining efficacy ambiguities by focusing a magnifying lens on the ambiguity. They do this at a cost – ignoring safety. They are an act of hypnosis that gets investigators to focus on one thing while ignoring everything else going on around them.
If we ignore the ignoring of safety, as we do, we compromise rather than enhance safety. The harms vanish. The drugs that come out of an RCT have no shadow – no harms. They are vampire medicines. Put another way, all clinical trials (RCTs) cause harm but some can also be helpful. They pose the same problem AI and viruses do – if you create one, can you control the consequences?
Co-incidentally around the time vampires began to flourish, companies had poison symbols removed from medicines and from discourse. Any mention of the celebrated medical wisdom that all medicines are poisons in an expert report will now be met by company motions to have the word poison struck as prejudicial.
If there was a conflict of interest involved in linking trials to regulation, it was born from hostility to pharmaceutical companies.
But it’s now clear that even if done by angels, RCTs are the gold standard way to hide adverse events. Their intense focus on a primary outcome rather than the overall picture means that in even the most independent of studies they necessarily neglect adverse events. Companies overlay a lot of creative hiding on this neglect but this is an extra source of difficulties not the primary problem.
It was doctors who used to go around with stakes. Regulatory bureaucrats never did. RCTs have made life easier for the bureaucrats. According to the head of Britain’s MHRA, Ian Hudson, formerly of GSK, if events are not statistically significant – and strictly speaking as no adverse event is ever a primary outcome measure, they can’t be – then they aren’t happening. BMJ helped consign stakes to the twentieth century.
If there was a conflict of interest involved in the move from RCT to Evidence Based Medicine (EBM), it was born from hostility to companies.
EBM was not unreasonable in the early 1990s but it has created a bandwagon that is now out of control contributing hugely to a conversion of poisons into fertilizers to be sprinkled as extensively as possible, from as young an age as possible.
If there was a conflict of interest involved in using trials to create Guidelines in the 1980s, it was born from hostility to pharmaceutical companies.
But now, if NICE and other Guidelines, based entirely as they are in the case of on-patent medicines, on ghostwritten articles whose data is inaccessible, hadn’t been created, Pharma would have to invent them.
Drug harms that took a year or two to come to light in the 1960s take a decade or two now. In 2003 I predicted we were on our way to having lethal and common drug harms being contested years after the drug had gone off patent. In April 2017, the Dolin case in Chicago centred on just this issue – ten years after paroxetine had gone off patent.
These problems are not being contested because it is difficult to decide if a drug is causing a harm. They are being contested because of the power of companies to shut down debate in journals like BMJ.
BMJ’s news item about the Dolin trial missed the key issues, and it wouldn’t have occurred to them they had a role in the development of the situation.
If it was just one drug harm that might be fine but drug wrecks may well now be the leading source of death and disability on the planet.
BMJ helps this happen in its educational forums by making it clear for instance that giving antidepressants to minors is just fine – no mention of harms.
BMJ helps this happen in its educational forums by making it clear for instance that giving antidepressants in pregnancy – now the most commonly used drugs in pregnancy – is just fine despite a doubling the rate of birth defects and miscarriages and behavioural abnormalities in the children of those taking them.
Richard Smith was succeeded by Fiona Godlee, as free of ties to industry as RS once was.
Starting with RS and then later with FG, I and others had a series of articles on drug harms rejected with an increasingly bizarre set of reasons offered for the rejections, and increasing involvement of BMJ’s legal department.
These rejections have two elements. One is that across all journals anyone interested in a drug’s harms appears to be deemed ipso facto not just biased but perverted. It’s like wanting to publish something about sex around the time Dracula was written.
The second is anyone who has ever been in any way linked to a lawsuit as an expert against a company is deemed irredeemably conflicted.
For those of us working on the Restoration of Study 329, the review process became beyond odd. BMJ editors kept raising queries that had been answered and talking about conflict of interest. To which our response was everyone is biased but the way conflicts are overcome in science is to turn to the data – it’s the data that reveals whether someone’s latent bias is operative or not.
BMJ just didn’t get it. My formulation up till then was that a belief conflict of interest is of supreme importance is incompatible with recognizing the importance of accessing the data. This is still my view in the case of most of those who champion Sunshine.
But then a predicted legal review came into the 329 frame.
BMJ’s lawyers made it clear that if the journal had anything to do with Healy and Jureidini they would be providing GSK or other companies with grounds to claim bias and to sue. This is even though BMJ and GSK are partners in AllTrials and BMJ had a short while before featured Andrew Witty on its front cover as the acceptable face of pharma, helping GSK put a $3 billion fine behind them. Partnership?
BMJ is a business that can no longer support medical practice, a key part of which is describing harms. If a doctor is not prepared to go to Court to stand up for the harms she’s put her name to, she’s writing fiction – a bit like Bram Stoker saying in real life he doesn’t believe in vampires.
That the BMJ editor handling 329 was married to a man who worked for the law firm who defended GSK against a $3 billion Department of Justice fine didn’t seem to be any kind of conflict of interest. But then patients who have committed suicide or homicide on these drugs aren’t going to sue BMJ.
When it comes to on-patent drugs, BMJ provide news stories about medicine, medical entertainment. They are less likely than the NY Times or BBC to check the integrity of their primary sources.
BMJ and all our major journals have been turned.
Water and drugs flow down even the slightest gradient unless there is something in the way.
The BMJ got involved in the Tamiflu saga in 2009 making it the basis of a “transparency” campaign – see Cisparency and Transparency next week – and later AllTrials.
Tamiflu is an efficacy story not a harms story. As Peter Doshi and Tom Jefferson accessed more and more efficacy data the efficacy of Tamiflu shrank to almost nothing. But it only takes 1% efficacy to sell pills provided there are no bumps in the way, and nothing has happened to dislodge Tamiflu from the top of the guideline heap for use in cases of Flu.
The story of antidepressants and teenage girls is even more remarkable. There are now 30 RCTs of SSRIs and related antidepressants for adolescents and children – 29.5 negative. Fluoxetine/Prozac is the most negative of the lot – there are 7 trials in which it has failed to beat placebo on the primary outcome measure including the two that led to its licensing by FDA and MHRA and EMA.
Surely not – you thought Prozac worked for children. Why have you not heard of this? Same reason you never knew Harvey Weinstein was anything other than a nice guy – with the extra reason that whatever about other enablers, bureaucrats never ever admit they’ve made a mistake, and the academic media aren’t any better.
There is an excess of suicidal events on antidepressants in every one of these 30 trials. In the only notionally non-company trial, the investigators in 7 major publications in leading journals from this trial “managed to conceal” the 34 suicidal events on Fluoxetine versus 3 on placebo – this requires an Editorial Nelson to be not just blind in one eye but very short-sighted in the other.
The BMJ has done more than conceal harms. It “enabled” a transformation of these life-ending harms into a “controversy”, when in 2014 it published close to the shonkiest, most ridiculous article any major journal has ever published on anything – the Lu et al article.
There seems no point in submitting an almost entirely data-driven, opinion-free, article on what the trials in teenagers show to BMJ, even though antidepressants are now the most commonly prescribed drugs to teenage girls.
Even though last week, the British Secretary of State for Health declared adolescent mental health services the biggest single weakness in NHS provision. The great concern is that teenagers are self-harming while waiting for access to secondary services – now why would that be?
A few weeks before, the grisliest case report ever appeared in one of the few media still concerned about primary sources (Panorama). It covered the killing of twelve people in Aurora by a 24 year old on an SSRI, Black Boxed up to 25. This case report is totally backed up by the RCT data. BMJ played a stellar role in dissing any possible link to the drug – abetted by a posse of Psychiatric Association Presidents and ex-Presidents.
With the creation of Dracula, Bram Stoker, a Dubliner, globalized the Celtic feast of Halloween. Both Dracula and Stoker left their place of origin and came to England. When last heard of, Stoker was working a few blocks down the road from the BMJ office…
To be continued