When raw data contradict the published 'findings' on a new drug, it is time to review clinical trials, argues David Healy.
Readers opening an issue of any major medical journal are likely to be faced with a glossy drug advertisement on one side and a clinical trial for one of the latest treatments in medicine on the other. There is a substantial consensus among lay and professional groups that the glossy adverts are a problem and that the way to counteract company marketing is to ensure that medicine conducts and publishes more high-quality clinical trial reports. Fortunately, to judge by the outcomes of media ambushes of clinicians in expensive resorts, the tactics of pharmaceutical company sales departments, such as the use of adverts and of merchandise, and invitations to meetings in glamorous resorts, have little effect. When asked what influences them, physicians typically state that they are primarily driven by evidence from clinical trials.
For some time, the only hurdle in the way of properly evidence-based medicine has appeared to be the well-known fact that companies fail to publish the results of negative trials. This has led to calls for a register of clinical trials. Given a register, meta-analysts and others would supposedly have the means to combat the pernicious influence of some companies' marketing.
Recent events surrounding the induction of suicidal intent by selective serotonin re-uptake inhibitors (SSRIs), however, have revealed a deeper problem than the failure to publish negative trials. Study 329, looking at the effects of paroxetine in depressed minors, was a negative trial - but one that was published. Having made an assessment in 1998 that 329 did not offer evidence that paroxetine worked in minors, GlaxoSmithKline nevertheless opted to publish selected positive results from one study. The final article, published in the Journal of the American Academy of Child and Adolescent Psychiatry with an authorship including many of the best-known names in American psychopharmacology, was to the effect that paroxetine was well tolerated and effective. Almost every other article representing the results of controlled trials with fluoxetine, paroxetine, sertraline or other antidepressants have made similar statements implying tolerance and efficacy in minors. No academic or clinician was ever likely to be any the wiser, had it not been for the ability of a few journalists to spot deficiencies in the imperial wardrobe. As a result of concerns raised in the media, regulatory actions led to the emergence into the public domain of a fuller set of results than usual from about 20 controlled trials. This has left us facing the greatest known divide in medicine between what the raw data from a set of trials show and what the academic literature that purports to represent those data claims.
A comparable divide exists in the supposedly scientific literature on the use of SSRIs in adults. In the case of paroxetine, raw data point to an up to eightfold increase in suicidal acts among paroxetine users in clinical trials compared with a placebo. But the most widely cited review of these data claims that paroxetine is five times less likely than a placebo to be linked to suicidal acts. This innocent review was published in European Neuropsychopharmacology . The first author was also editor of the journal; the second author apparently has never seen the raw data; and the third author has links to the company making paroxetine. Similar discrepancies between the data and published claims can be found for fluoxetine and sertraline.
This is a crisis for evidence-based medicine rather than merely an issue concerning a side-effect of treatments. It is a crisis that stems from the fact that the marketing - as opposed to the sales - departments of some companies that accept that clinicians are primarily influenced by evidence, have, over the past 20 years, set about providing selective evidence in the best journals and with the best authors. Two structural features aid them in this process. One is the fact that the majority of company trials are probably written up by ghostwriters. Second, the raw data from these trials remain inaccessible to the public. But the real problem is not a matter of our inability to access reports of unpublished trials - unpublished trials are no use to a marketing department. The real problem is whether we can trust the results of any published trials.
When the 1962 amendments to the Food and Drugs Act were put in place, physicians were seen as one check on understandable company excesses in promoting their products. The power of regulators to monitor the claims being made by companies in their advertisements was another. But manufacturers realised some 20 years ago that regulators cannot control what academics say. If, after due consideration of the published evidence, individual academics endorse a product, or groups of experts in consensus panels build that product into guidelines and algorithms that dictate treatment options, the regulators will not intervene. They will not intervene even when the adverts for products contain claims that the regulators would regard as illegal if made by the companies, if these claims are referenced to supposedly independent academic authors.
What clinicians have failed to realise is that certain marketing departments seek to understand them better than they understand themselves - playing even on their sense of imperviousness to the trinkets of sales.
They have failed to understand that it is the goal of certain marketing departments in any industry to capture the consciousness of the consumers rather than selling the product. In the pharmaceutical industry, physicians rather than patients are the consumers.
In the case of antidepressants, this has meant promoting a vision of depression as a lowering of serotonin, despite an almost complete dearth of evidence. This image of the illness marginalises non-drug treatments and judicious non-intervention to see whether the condition will resolve, as a majority of conditions do within weeks of detection, even if untreated. It puts an almost moral onus on people to seek treatment and clinicians to treat this "serotonin deficiency" disorder.
In the course of the 20th century, marketing has been incorporated into all areas of our life. There must, however, be some fundamental opposition between marketing and science, in that marketing operates to build consensus - and what better way to do that than by coming to the obvious conclusions that the "scientific" evidence base mandates - while science traditionally proceeds by fracturing consensus. But there can be no scientific fracturing of the consensus in the absence of data. Only cranks rail against experts when they have no data.
Against this background, maintaining therapeutics or any area in which human beings interface with the new biomedicine as a science would seem impossible at this point, without independent access to the raw data of clinical trials or other studies. Registering unpublished trials will not suffice.
There would seem to be powerful ethical grounds for insisting on access to the raw data from trials, in that these data for the most part stem from vulnerable individuals taking risks on behalf of the community with drugs that often turn out to be too lethal or too ineffective to market. It is also of note that no informed consent forms at present inform those participating in trials that their data will remain unavailable to independent scrutiny. Selecting positive results and marketing these as "science", while ignoring negative results, would seem to be one of the most flagrant violations of the norms of scientific inquiry imaginable, quite aside from any ethical violations.
But this is an issue that goes beyond medicine and science. A great deal of the use of drugs such as the SSRIs in both children and adults appears to have more to do with what can be viewed as either efforts to manage risks or enhance ourselves rather than with traditional efforts to treat disease.
The promise of the new biology lies in just these areas of risk management or enhancement as we seek to better understand who we are. The data that stem from biological and especially genetic sources will shed light on our attitudes and our aptitudes and on many aspects of how we govern ourselves, from our proclivities to particular political or religious orientations through to our physical and mental abilities. But these data will for the most part be owned by pharmaceutical companies. Concealing key data of this sort in the manner that companies have become accustomed to doing in recent years will almost by definition cut across the abilities of a people to govern themselves in the most effective manner possible. We have a profoundly antidemocratic force growing in our midst.
David Healy is director of the North Wales Department of Psychological Medicine at Bangor. He has contributed a chapter to The New Brain Sciences , to be published by Cambridge University Press in October. He was involved in a libel row with the University of Toronto over an employment offer that was rescinded after he gave a lecture linking antidepressants with suicide. Both parties agreed to drop complaints after a settlement, and he was given a one-week visiting professorship in the faculty of medicine.