Nobody has retracted or apologized for a study that was an academic disgrace—but a marketing coup for GSK—which may well have caused untold numbers of deaths, suicide attempts and irreversible anguish to myriad families. Can we stand idly by when we’re told that it “accurately reflects the honestly-held views of the clinical investigator authors who do not agree that the article is false, fraudulent or misleading.”? What is the current market value of the honestly-held views of people who tell lies?
I’d like to reflect on a few major aspects of Study 329 and the recent BMJ restoration study (RS) which raise fundamental ethical issues, and which pose some theoretical problems or raise other important issues which haven’t yet been scrutinized. And, by counterpointing Paxil against Thalidomide, I shall suggest that a seismic cultural shift has made studies like Keller’s almost inevitable.
1. Ethics and Academic Integrity
Study 329 forces us to confront the ethical question once again, before it’s too late, and question the claim that the investigators were chosen only for their expertise, their interest in the subject and their capacity to recruit volunteers. Might they not have been in the pay of GSK, or have been friends, ex-students or colleagues of people like Biederman, Keller, Krystal or Nemeroff? Or chosen for their commitment to drug solutions, regardless of awkward research findings, in line with the FDA which wrote at the time that there may be negative findings even in trials of drugs they know really work? Even at the initial stages, then, there was plenty of room for manoeuvre, allowing GSK and its (paid?) investigators to take advantage of a fuzzy diagnostic category, to cherry-pick the subjects most likely to produce favourable outcomes and to discount the gravity or relative severity of an adverse event if it suited their purposes.
The sociopathic lack of remorse and moral consciousness shown by GSK, Keller, et al., became most shockingly apparent early on, in GSK’s serial re-writing and occlusion of troubling data — designed to stop negative results leaking out to doctors, the public or their sales staff. In fact, a GSK internal memo showed that the company knew that their studies had failed to demonstrate efficacy since at least 1998; and in 2003 the MHRA revealed that GSK’s own studies showed that the drug actually trebles the risk of suicidal thoughts and behaviour in depressed children. Outside the Holocaust, I’ve never come across a more chilling, amoral or sociopathic memo than the one they sent out to senior management, clarifying their decision “to effectively manage the dissemination of these data in order to minimize any potential negative commercial impact…It would be commercially unacceptable to include a statement that efficacy had not been demonstrated, as this would undermine the profile of paroxetine.” In the same cynical vein, an email from a PR executive working for GSK said: “Originally we had planned to do extensive media relations surrounding this study until we actually viewed the results. Essentially the study did not really show it was effective in treating adolescent depression, which is not something we want to publicize.”
An important ethical barometer is the prevalence of authors’ conflicts of interest (COIs), though it is really a wider, cultural, problem, for a large majority of medical and science journals in the world have no COI policy. While we weren’t told about the authors’ COIs here, we now know those of Keller, Ryan, and of GSK employee Mc. Cafferty, whose affiliation was hidden. We know, for example, that Shelley Jofre sent emails to Dr. Ryan in 2002 asking questions about the safety of Paxil, whereupon this “independent” researcher forwarded them to GSK, asking for advice on how to respond to her! We know also, thanks to Peter Doshi, the extent to which Karen Wagner was beholden to BP. I have since followed up her case: it beggars belief. Hunting down the COI hare of the other “authors” could bring to light some very damning evidence indeed.
The prevalence of ghostwriting is another disturbing example of ethical indifference: David Healy, who has monitored this phenomenon rigorously for years, estimates that up to 50% of apparently serious academic psychiatric studies are ghostwritten. Could this be a devious way of getting around the ban on off-label promotion?
The rampant, unethical ubiquity of ghost writing has sullied the reputation of numerous academics and their institutions in the U.S. Hannah Arendt’s reflection is apposite here: “When all are guilty, no one is; confessions of collective guilt are the best possible safeguard against the discovery of culprits, and the very magnitude of the crime the best excuse for doing nothing.” Here, however, we have no confession of collective guilt; not surprising, since guilt implies a minimal moral conscience, which was so rare here and among Chemie-Grünenthal’s (C-G) top brass in the thalidomide tragedy, some of whom had seen active service in the camps, and most of whom had impeccable Nazi credentials. This brings to mind those studies some years ago showing that a large proportion of industry CEOs could be classified as sociopaths: another indicator of a huge cultural shift. Are we now simply facing the inevitable consequences of rampant neo-liberal economics?
We could of course argue that this entire affair could be reduced to just one issue: is there any justification for ghost-written academic studies, especially when they have serious safety implications? If it is deemed to be professionally acceptable, then the “authors” of Study 329 have committed no sin at all: it’s just par for the course. After all, PLoS Medicine, which has been leading the charge against ghostwriting, found that only 13 of the top 50 medical schools in the U.S. have a policy that prohibits it.
At a 2011 U. Toronto conference on ghostwriting, Trudo Lemmens, a law professor, said that biomedical ghostwriting is a public health issue needing serious attention, since erroneous use of pharmaceuticals is a leading cause of hospitalization. Many studies, like Keller’s, contribute to that by hiding adverse events, negative data, and over-emphasizing benefits, so academics fronting such studies are responsible for any ensuing prescriptions and harms. Yet there seems to be no backing in law for those wanting to hold academics legally and professionally accountable: ghostwriting is perfectly legal, widely practiced yet officially considered unethical, so universities are nicely protected when they routinely protect shady researchers, who bring them prestige and large research grants. McGill, though, showed commendable rigour and oversight in the Sherwin/Wyeth case.
Doshi and Healy have no doubt that Study 329 was ghostwritten, but I’m not yet absolutely sure about this, though the circumstantial evidence is overwhelming. Keller’s recent letter defending his study and criticizing the RS does indicate that the “authors” were centrally involved at all stages of the study. Could they be forced to take an oath on this, as was Sally Laden? I’m still very unclear about which authors knew what, if anything: the much- compromised JAMA tried, unsuccessfully, to find this out when first offered the article. Keller’s slippery filmed statements and his facial language made me very suspicious indeed, as he repeatedly dodged interviewers with statements like: “I don’t remember matters like this…I never read such tables.”
However, let’s now suppose it wasn’t ghost written, then Keller has painted himself into an inescapable corner, since the only excuse he could proffer for producing scholarly work that drew conclusions contradicting the raw data was that it was, in fact, ghost written. Either that or they produced scholarship that would not have been accepted from an undergraduate, but was accepted by the editor of a prestigious journal who brushed aside THE JOURNAL’S OWN REVIEWERS’ DEEP RESERVATIONS about the study. Why?
2. Pseudo-science, Lies, Damn Lies, and Statistics
Another ethical casualty of Study 329 was peer review, a protocol designed to guarantee the integrity, independence and rigour of published scholarship: JAMA followed the protocol strictly, and rejected the study, but JAACAP did not. How did it get away with so few official criticisms or sanctions? The story behind the editor’s ultimate acceptance of it could prove very incriminating, and throw up even more serious questions about academe’s failure to uphold minimal standards by publishing a labyrinth of lies, half-truths and lies by omission. The intellectual dishonesty displayed here by members of a prestigious Ivy League university is a bloody blot on the escutcheon of American academic integrity. Standards like this in very high places should not be tolerated by any academic community worth its salt.
The RS also highlighted an area which is rarely discussed: statistical analysis. We have become hung up on whether the results of a trial had statistical or clinical significance, but this study reminds us of the value of patient narratives, “merely anecdotal” reports and descriptive statistics, as did the thalidomide tragedy. Here, the commitment to descriptive statistics was abandoned, probably because they might have allowed non-specialists to make judgments and informed decisions, whereas more sophisticated techniques offered a smokescreen to hide behind for years, thus allowing billions to accrue. How many of us are sufficiently qualified to evaluate what the statisticians tell us, and how they tell it? Unsurprisingly, GSK told us that they employed the best ones here: this may well have been the only true statement they uttered. It was in their interest to hire the top people, pay them a fortune and make it clear that, without making fools of themselves professionally, they should choose designs, protocols and methods of analysis likely to give the required positive results.
Jureidini’s team forces us to radically critique another untouchable: RCTs, whose status, integrity and even utility must now be thrown into question. Following Breggin, Cohen, Doshi, Harrow, Healy, Kirsch, Moncrieff, Timimi, Whitaker et al., we RCT sceptics now have much more precise ammunition to fight with. In their wake, can we continue to have faith in any previous RCTs and meta-analyses when so many negative trial results are never published and when fundamentals like blinding and adherence to protocols are so flagrantly flouted? Is this just the tip of a very large, murky and treacherous iceberg? After all, Study 329 was published in a top journal, yet deviated from the original protocols, added several new outcome measures, sometimes after un-blinding, and drew conclusions directly at odds with their own findings. In particular, I am deeply suspicious about its level of sustained adherence to blinding protocols.
Over the years, we have had many scandalous instances of unethical manipulation which allowed BP an easy ride. Keller et al. gave us an egregious example of this when they disbanded reliance on Hamilton, but never registered the change, blithely declaring that since the profession had moved beyond Ham-D between 1993 and 1998, they needed to look at other more appropriate measures! As we say in Dublin: ”Pull the other one, it has bells on it!” I know of no evidence that he was right, and their unseemly scratching around for scales, criteria and about 20 new secondary outcomes that might show Paxil in a more favourable light was sad, desperate, even comical.
In short, we’ve had many reasons to be extremely sceptical of RCTs and critical of the uses to which they have been put: Study 329, for example, was merely a super-lucrative advertisement for a dangerous, addictive drug with doubtful benefits and numerous toxic side-effects. (In 2002 over two million paediatric Paxil prescriptions were written in the USA alone, just in time to beat the patent deadline.) Now whereas the law, the FDA and blind faith in RCTs bought an awful lot of time for GSK to make many billions unimpeded in subsequent years, a modest number of academic articles and individual “anecdotal” reports from doctors between 1958 and 1961 stopped thalidomide in its tracks by 1962. They didn’t hang about too long in those days: one critic was scandalized that C-G knew for six whole weeks of the dangers posed by thalidomide!
The RS, then, may well deal a mortal blow to our naïve faith in academic integrity, our bedazzlement by RCTs, and to our easy acceptance of them as gold-standard benchmarks of truth and rigour. It should, at the very least, put them in their place as one potential, yet fallible, tool in the arsenal of psychiatry, but one that should never be allowed to usurp the centrality of the empathic, dialogic relationship between a singular patient and healer, nor the full story of individual suffering, which is obfuscated by algorithms and statistical averages.
On the other hand, the RS might also, paradoxically, provoke a more positive, nuanced view of RCTs, suggesting that some modicum of scientific respectability could be maintained, but NOT IN THE PRESENT VENAL CULTURE: Jureidini et al. have shown us what they might yield if carried out rigorously by top independent researchers putting in Herculean work for no financial gain: had their study appeared in 2003 many wasted billions and many, many lives might have been saved. But how often are we going to get such a highly qualified, deeply committed, hard-working group together in the present climate, with its routine financial inducements? Should we not, perhaps, use the thalidomide case to reconsider the value of strong, repeated “anecdotal” pointers, thus emphasizing individual stories of psychotropic drug use and the withdrawal therefrom? David Healy, Luke Montagu and James Davies have given us a very strong lead in this domain.
Another problem hovering over, but not addressed, by both studies is that of causality: in psychiatric and neuroscientific discourse there’s a lot of slippery semantic skating going on – “linked to,” “associated with,” ”implicated in,” “correlated with,” and the like, but rarely “caused by.” This is not always helpful, but it does bespeak an understandable and judicious reluctance to assign one cause to any outcome. The Study 329 authors cleverly exploited this difficulty when they tried to minimize serious adverse effects related to treatment by declaring that “causality cannot be determined conclusively.” A single cause for any illness can rarely be proven conclusively – even in cancer, as a number of commentators have pointed out. Thalidomide, however, does appear to have been the sole cause of all those awful deformities; but even in a case like PKU, which is habitually deemed to have a purely genetic aetiology, the gene will not express itself if rigorous environmental controls are put in place – in this case, diet.
Now while it seems clear that judges, biomedical academics and psychiatrists have grossly underestimated antidepressants’ (ADs) contribution to suicidal behaviour, it is also possible to overstate the case by insisting that a given SSRI caused violent, murderous or suicidal behaviour, thus negating the potential impact of other confounding variables. Yet some, but not many judges, taking the advice of people like Peter Breggin and David Healy, have declared that an AD like Prozac had indeed induced a murder. But, pace both of these great fighters, I’d suggest that legal sanctions would be far easier to obtain if we focussed not on a conclusive single cause for a violent act, but on the very high risk factors associated with a given SSRI and on the full personal and contextual story.
3. The Matter of Suicide
In the opening of The Myth of Sisyphus, Camus argued that suicide is the only serious philosophical question, but for many, suicidal ideation (SI) is a dangerous by-product of SSRI consumption; a serious adverse effect. And further, SI means different things to different researchers. For some, it is a particular preoccupation with suicide, ranging from regular random thoughts, to more sustained and frequent thoughts of suicide. For others, however, SI includes suicidal gestures or behaviours such as the planning or rehearsing of the act, incomplete attempts, or more serious attempts designed to fail. Even the MedDRA confuses the issue by coding suicidal ideation OR self-harm OR attempted suicide as suicide events. The RS rightly criticizes Keller for using the term “emotional lability” to fudge the severity and differences in suicidal behaviours, but have they not also engaged in some fudging by lumping suicidal thinking and events under one heading, “suicide-related adverse events”?
There is too much fuzzy thinking here: surely logic and scientific rigour oblige us to clarify our terms and avoid assigning the same weight to SI and clear suicide attempts, since most people who have suicidal ideation do not go on to end their lives, even though it must be considered a (low?) risk factor for suicide? To put this in context, it was estimated that in 2008-9 over 8.3 million adults aged 18+ in the U.S. reported suicidal thoughts in the previous year, but of these, only one adult in 140, 60,000 people, actually took their own lives. SI does not usually mean that someone wants to die, but that s/he does have a desperate need to express her/his hopelessness, impotence, overwhelming pain or collapse of meaning. It may often be a natural strategy employed in a hopeless situation: an imaginative rehearsal or symbolic expression of entrapment in the reptilian freeze response; of the impossibility of fight, flight or social engagement, to put it in terms of Polyvagal Theory.
I’m suggesting, then, that we must beware of falling into the trap of dramatizing or pathologizing something that may be absolutely normal, even healthy, in certain circumstances: for example, if one is homeless, disadvantaged or marginalized, living in unbearable social conditions, or trying to deal with overwhelming loss or failure the lack of SI would surely be a highly dissociative move.
Pain, danger and thoughts of suicide often attend the birth-pangs of any major life transition or change in status, as it did for Tolstoy and J.S. Mill. (See James Davies’ fascinating book, The Importance of Suffering.) Camus’s Meursault normalizes suicide by reminding us that everyone at some time has wanted to kill the one s/he loves: for adolescents needing to separate and carve out their own identity such thoughts are routine, healthy, and not necessarily caused by SSRIs. And for them SI can be one way of managing the maelstrom of hormonal turmoil, passionate intensity, emotional lability and high-risk behaviour attending the scary transition from childhood to adulthood. It may simply well be a way of saying that s/he sees no meaning in life and cannot go on this way any more, especially at a time of collapse in traditional values.
We need to remember, too, that modern adolescence, bereft of mentors, is a particularly fraught, unstable time, in which risky behaviour is far more common than in childhood or adulthood; one in which accidents are the main cause of death, followed by suicide, regardless of any SSRI consumption. (See Sarah-Jayne Blakemore’s research on risk-taking behaviour in adolescents.) For example, in one recent case it became clear that a vicious racism had fueled multiple murders in a local school. In another, a recent school killer’s notebooks showed that he was in a state of absolute despair because doctors had told him he was condemned to live with “a broken brain” for life.
However, well-documented adolescent suicide attempts do have to be taken very seriously, even though they may well be just last-ditch screams for help or understanding: after all, David Healy and others point out that for every eleven suicide attempts there’s one actual suicide.
I believe, then, that it is important to de-emphasize and downgrade SI since the term is used too loosely and can a have a positive valency. Why not confine the term solely to thoughts, and prioritize clear suicidal behaviours? And, taking into account the wider cultural framework, would it not make more sense for judges and researchers to focus on the individual contexts and narratives subtending acts of serious self-harm, clear suicide attempts and very uncharacteristic acts of violence? Drug are usually part of a much bigger story, which we neglect at our peril. Why not, then, let the law get on with assessing the strength of the circumstantial evidence and decide whether a given drug is a major contributor to a suicide beyond any reasonable doubt?
4. Neo-liberal Culture
Martin Keller recently made the point that since nobody has been able to pin anything on the authors, there’s obviously nothing to apologize for. The terrible thing is that he’s right: they have merely acted in a way that is condoned by academic journals, the FDA and the universities; sanctioned by the psychiatric establishment and the wider culture which makes it so easy for negative findings to run for cover, since there is no firm regulatory obligation to report all the results from clinical trials. In this case, Brown is in the dock, but numerous U.S. academic establishment have been regularly shamed, even though a few of them, like Harvard, have had the courage and integrity to come clean in similar cases, so why not Brown, whose officials conveniently lost some vital incriminating data? (Precisely the same thing happened with a vital incriminating registered letter at the C-G thalidomide trial.) Harvard did take some nebulous action against Biederman et al., but only when the scale of the deceit had become too blatant to hide. It is quite shameful that most U.S. universities routinely offer no comment when the media or people like Senator Grassley ask the tough questions about questionable research or COIs. Indeed, Grassley has said that they seem incapable of monitoring the COIs of faculty. Like the all-powerful gun lobby that even Obama can do little to tame, Marcia Angell’s 800-lb. gorilla has run amok, and thumbs its nose with impunity at the helpless Sorcerer’s Apprentice. And at all of us who protest so loudly and so often. But is this enough? Do the times not demand more concrete, radical action?
Is Study 329, then, not just one more shocking example of the mercantile values and the druggy doxa that permeate our entire culture? Obama’s choice when appointing the new FDA director is a stark incarnation of just how bad things are. So, before we all get too righteous and complacent we mustn’t forget that JAACAP, GSK, Keller et al. are simply following the ethos of our culture in which ethics and transparency have been all but abandoned, even at the top. Earlier this year, Scott, Rucklidge and Mulder studied the adherence, from 2009-13, by the editors of the five leading psychiatric journals to their own pious protocols of oversight and integrity, concluding that “most trials were either not prospectively registered, changed POMs or the timeframes at some point after registration or changed participant numbers.” When such oversight is so loose at the top, to whom do we turn for safe, rigorous research and unbiased guidelines? Quis custodiet ipsos custodes when they countenance criminal acts?
The Study 329 affair, then, underscores the fact that there is strong, but implicit, cultural permission given to BP and amoral academics for their chicanery since trial periods are so short, and post-marketing surveillance so weak and random. Our modern blind-eye culture has no problem with FDA laxity, blatant academic corruption and an increasingly perfunctory oversight by Congress and the law which permits direct-to-consumer marketing and off-label prescription, which in turn allows BP to profitably dispense with honesty, science or FDA approval.
Counterpointing the thalidomide with the Paxil affair, then, helps us to highlight a monumental shift in culture and values that lies very deep in the neo-liberal psyche, transcending Big Pharma, the universities et al. In the thalidomide case, the culture of the 60s enabled the remarkable integrity, rigour and persistence of the FDA’s Frances Kelsey who seems to have been fully backed by the FDA directorate, though one historian of the FDA wrote that she encountered some opposition there. And where she was supported by the wider culture, which subsequently lavished awards and encomia on her, our culture both encourages and rewards the likes of Keller et al. So in 40 years the 1962 amendment, designed to ensure the efficacy and safety of drugs, was emasculated, abused and disempowered by Big Pharma PR and cynical academics. Had thalidomide arrived in 1994 and been researched by such people, the laxity of the FDA, enabled by an ethics-free drug culture, would likely have ensured its continued presence on the market for years, producing millions of deformed babies facing early death or wrecked lives. Fortunately for mothers and new-borns back then academics had integrity: and the meteoric rise of BP power, allied to the prestige and often spurious truth-value of RCTs, was still some way off. Thus a small number of early academic studies, often in the BMJ, and many “merely anecdotal” reports, succeeded in ousting the drug before it got into its stride.
But, once again, what we’re really talking about here is a seismic cultural shift, over 40 years, which is underwritten by government and the law; one which finds no place for ethics, which favours permissive drug regulation, and a neo-liberal ethos in psychiatric practice, academia and the press.
Mad in America hosts blogs by a diverse group of writers. These posts are designed to serve as a public forum for a discussion—broadly speaking—of psychiatry and its treatments. The opinions expressed are the writers’ own.
Thank you for this insightful article about the pseudo-science of the psychology/psychiatry paradigm. It is staggering what passes for science (including influence peddling, disregard for scientific methodology, lack of replication studies and a lack of transparency) when it supports cultural expectations.
Best wishes, Steve
Thanks for this article Redmon. It is disgusting and despicable what passes for “science” in psychiatry and it “research” on drugs. Negative or equivocal results presented as endorsements of drugs is not limited to study 320 or Paxil of course… it extends to the studies on most of the major psych drugs, with the negative studies being hidden and the (slightly) positive few being blown way out of proportion… this is the case for Seroquel, Prozac, Risperdal, and the others discussed in books by Kirsch, Whitaker, Moncrieff etc.
One truly awful thing about these studies not often discussed is that they only last for a few weeks or months. Most of these studies measure how drugs suppress symptoms over 4-8 weeks. But in the real world, people want positive functional changes in their lives lasting over 1, 2, 3 or more YEARS. So this type of research is completely out of step with the functional changes people really care about – i.e. returning to work, making friends, going to college, and being symptom free for long periods, enough to rebuild your life. There are many reasons for this but surely the covert awareness that drugs do not really help many over the long term is one of the reasons for keeping studies short and focusing on symptom suppression rather than wellbeing and long-term functioning.