Pharmageddon

by David Healy

  The second development in both the United States and Europe was a huge expansion in university training and in the health services industry. This led to a rapidly growing number of medical academics. Where before the pharmaceutical industry had been beholden to a handful of magisterial figures, any of whom could make or break a new drug, now companies could shop around among eager young academics willing to manage a trial in return for the kudos of being a principal investigator and the fees that came with the exercise. These new academics, more vulnerable in their careers and more open to the seduction of being made into opinion leaders, could be expected to accommodate themselves to company interests in a way that their older counterparts might not. Besides, there was little these academics had to do other than be notional investigators—companies could now supply them readymade versions of clinical trial protocols that the previous generation of academics had developed during the 1950s.

  The third development was the increasing dispersion of the locus of clinical trials. Where in the 1950s and early 1960s trials had often been conducted in a single university or hospital, by the late 1960s they were typically multicentered, and by the 1970s they had become multinational. This did not happen because the drugs and the testing were better. Quite the contrary, the more obviously effective the drug the smaller the trial needed to demonstrate that it works, although large trials might still be needed to reveal possible hazards of treatment. The newer drugs were in fact less effective and required larger trials to show benefits that could be considered statistically significant. Increase in the number of participants in typical trials and the geographic proliferation of sites had far-reaching implications; principal investigators could no longer know all the patients, could no longer supervise all the raters, and might no longer be able to speak authoritatively to side effects that may only have been witnessed on sites other than their own. Where before investigators had the data for the whole study, now they typically had a record only of the data from their own site and just a summary of the rest. Companies or their proxies increasingly did the shuttling between sites, and it made sense to lodge the data somewhere central—such as company headquarters. When investigators requested full access to the data—the answer was no, the information was proprietary. Without accessible data, these trials had the appearance of science but were no longer science.

  These three developments changed the interface between medicine and industry. Where clinical trials were once a scientific exercise aimed at weeding out ineffective treatments, they became in industry's hands a means to sell treatments of little benefit. The key to this has been— and still is—industry control of the data from “their” trials, which is then selectively described in articles designed to establish market niches by selling diseases the new drug happens to address or to counteract adverse publicity about treatment hazards.

  THE APPEARANCES OF SCIENCE

  As part of an effort to reengineer their drug development and marketing strategies in the 1970s, the major pharmaceutical companies began to outsource not only their clinical trial division but their medical writing division as well. The job of running clinical trials went to a new set of companies—clinical research organizations (CROs). The early CROs included Quintiles (set up in 1982), Parexel (in business since 1984), and Covance (started in 1987), with a growing number of companies such as Scirex and Target Research coming onstream in the 1990s. By 2010, the clinical trial business was worth $30 billion and CROs were running more than two-thirds of the clinical trials undertaken by industry.2

  In the 1950s and 1960s, controlled trials involving drugs were either funded by independent agencies such as the National Institutes of Health or involved drug companies handing over supplies of new compounds to medical academics who devised the clinical trials to test out the new remedy. These professors and their colleagues personally witnessed the giving of placebos that were outwardly indistinguishable from the trial drugs, along with both older and new drugs to patients. They interviewed the patients and completed the rating scales themselves or personally taught members of their team how to do so. When the data were finally assembled, professors would analyze them and later store the data in their filing cabinet for consultation should questions arise. While some academics availed themselves of the new opportunities to supplement their income, in many instances, the investigators did the work without charge—the new drugs were unquestionably moving medical science forward, which for some was payment enough.

  When a lead investigator wrote a paper for a scientific journal representing what the trial had shown, the resulting article reflected a judgment on the new compound based on familiarity with other compounds in the field. When it came to presenting the findings at an academic meeting, the professor was there to answer audience questions about potential hazards of the new compound or other issues not covered in the article or presentation. At meetings in the 1950s and 1960s, entire symposia were dedicated to the hazards of new treatments, where now it is as difficult to find mention of a treatment's hazards at major academic meetings as it would be to find snow in the Sahara.

  Few clinicians or others outside industry noticed any change during the transition from the 1960s's way of doing things to the 1980s's way. The trials under CRO auspices had all the appearances of previous clinical research. But from the 1980s onward, these trials increasingly diverged from previous clinical research practices.

  The CROs competed among themselves for clinical trial business on the basis not only of price but also of ensuring rapid access to patients and timely completion of competent study reports. Clinicians who enrolled suitable research subjects were paid per patient, and paid more for patients who completed a course of treatment. There was no one to keep an eye on whether those recruited and deemed to have a particular disorder actually had the disorder, however. In addition, patients were increasingly recruited by advertisement rather than from clinical care. And an increasing number of patients reported in these trials didn't exist. In 1996, as one indicator of this trend, Richard Borison and Bruce Diamond from the Medical College of Georgia were jailed for the conduct of their clinical trial business, which recruited nonexistent patients to trials of many of the antidepressants and antipsychotics now on the market.3

  Where once clinical trial protocols had to be approved by a hospital or university ethics committee—an institutional review board—now they may be subject only to the CRO's privatized review system for company studies. And where university or hospital review boards typically commented on the science in addition to the ethics of a study, often forcing researchers to improve their designs, privatized committees may simply nod company protocols through. As the clinical trial business grew and competition for patients increased, CROs initially moved trials out of university and hospital settings and contacted physicians in general practice to get study subjects. As getting patients in the United States and Europe became more difficult, even from primary care physicians, CROs began moving trials first to the former Eastern European bloc during the late 1990s and subsequently to Asia and Africa. Regardless of location, though, it's still likely to be a Western academic's name that appears as the notional principal investigator on the trial protocol or subsequent articles.

  But the key difference in the shift toward privatization of trial research lies in what happens with the data. Where once a professor might have analyzed the data from a trial, now personnel from the CRO collect the records from each center and assemble them back at base. The events in a patient's medical records or the results of investigations are then coded under agreed-upon headings from dictionaries listing side effects. These items are then cumulated in tabular form. Such tables are the closest that any of the academic investigators are likely to get to the raw data, except what they themselves collected. If later asked at an academic meeting what happened to patients on the drug, this is what they will cite. In this way the blind lead the blind—giving a whole new meaning to the idea of a double-blind study.

  For
instance, academics presenting the results of clinical trials of Paxil for depression or anxiety in children wrote about or showed slides containing rates of emotional lability on Paxil and placebo. The presentations were relatively glib, and there is no record of any member of any audience being concerned by higher rates of emotional lability on Paxil, probably because none of those involved realized that emotional lability is a coding term that covers suicidality; in fact, had they had the raw records in front of them clinicians would have realized there was a statistically significant increase in rates of suicidal acts following commencement of treatment with Paxil.

  Other decisions about how the data is presented can also drastically affect how a drug's effects are perceived. For instance, if a patient experiences a sequence of side effects starting with nausea, then several other side effects, and those are followed by suicidality, and the person then drops out of treatment, if nausea is listed first rather than the suicidal act that actually triggered the drop out, this patient will likely count as a drop out for nausea rather than suicide. Unless there is someone at the heart of the system who understands the issues, the default in these new trial procedures conceals rather than reveals problems.

  But there is no independent academic at the helm any more. The first semijudicial setting in which these issues were tested involved Richard Eastell and Aubrey Blumsohn from Sheffield University in England in a hearing in 2009. Eastell and Blumsohn had been senior investigators on a study of the effects of Proctor and Gamble's Actonel for osteoporosis. Behind the veneer of professional ethics lies another world, as Aubrey Blumsohn found out when Proctor and Gamble set about ghostwriting articles for him. A senior company figure wrote to him “to introduce you to one of The Alliance's external medical writers, Mary Royer. I've had the great privilege of working closely with Mary on a number of manuscripts, which The Alliance has recently published. Mary is based in New York and is very familiar with both the risedronate (Actonel) data and our key messages, in addition to being well clued up on competitor and general osteoporosis publications.”4 Blumsohn found himself faced with articles written up under his name containing tables of “data.” He asked to see the raw data, but access was refused. Using subterfuge, he eventually managed to get hold of the data and found that the tables had been cropped to leave out a great deal of data; when this information was included the claims being made for the drug did not hold up.5 He withdrew from authorship, a move that ultimately cost him his job at Sheffield.

  Eastell, who was the dean of medical research at Sheffield, meanwhile had authored several papers on the same research, one of which required him to declare he had access to the data. He said he had had access. In September 2009, Britain's General Medical Council (GMC), were asked to rule on whether Eastell's behavior was appropriate or whether he should be debarred from medical practice. At the heart of the issue was the question of what are data. After two days of legal argument, the GMC came up with a formulation that most readers of this book have likely already come to—that access to the data means access to the raw medical records. Summary tables of events or results are not data. (Eastell, as one might expect, pleaded that this had not been his understanding of what the term “data” meant at the time of his declaration.6)

  If the key data are the raw medical records, who owns these data? In point of fact, no one knows. When it comes to legal cases, companies regularly keep any paperwork they have stemming from “their” trials and other data off the US mainland just in case a court should request access. If in responding to an article presenting a company's drug trial results, I were to request access to the raw data I would be told to get lost. But this is an exercise of force majeure rather than a position that has scientific, legal, ethical, or moral support. Companies may be practicing their version of evidence-based medicine, but they are not practicing data-based medicine, which is the soul of scientific medicine, and they are actively stopping anyone else from practicing data-based medicine.

  Compared to the days in which purveyors of proprietary medicine flourished, one of the protections we now have is that someone like me cannot simply stand up at a major medical meeting and claim I have a new treatment for some condition without making the data available. I would be jeered off the podium. Practitioners of alternative or complementary therapies never get asked to medical meetings precisely because their claims are not based on data that can be examined.

  Yet standing up and making claims while refusing to make the raw data available is exactly what pharmaceutical companies, and academics speaking on their behalf, do time after time. At first this wasn't much of an issue as there were only occasional reports of company-sponsored drug trials scattered among other articles in peripheral journals and occasional presentations on the program of academic meetings. Initially many people assumed that the trial data was available just as it was in other studies: the academic presenters of the material must have had access to it. When it became clear that the presenters hadn't seen it, academics took refuge in the thought that at least the regulator would have seen all the data. But now, the majority of treatment trials reported in leading journals, as well as large chunks of academic conference programs, feature company material without data access and even if regulators request all the data, they do not get all the studies and certainly do not get the data in accessible form—although few doctors know this.7 At academic meetings and in scientific journals everyone welcomes the velvet glove of scientific appearances while ignoring the iron fist of refusal to part with data.

  Access to the data of experiments is at the heart of what makes science. Right from the very first scientific meetings in seventeenth- century England in Oxford and London, there was a consensus that what made these meetings scientific and different from all previous meetings was a focus on observable evidence. Whether the subject was physics, chemistry, or biology, participants ran experiments in front of their peers. Whether by virtue of a novel technique or a new method, the experimental art aimed at generating data that revealed a formerly hidden aspect of the world. Where others talked, scientists generated data and made both the data and the method to generate it publicly available. It is inconceivable that Louis Pasteur, Robert Koch, Richard Cabot, or other medical people who came after them would have refused access to the data underpinning any scientific claims they made. Refusal to share key information, however, was precisely what characterized the purveyors of proprietary medicines.

  GHOSTS IN THE MACHINE

  Just as many drug companies began outsourcing clinical trials in the 1970s, they began to do the same with medical writing, giving rise to a huge increase in ghostwriting and, in a further twist, ghost-presenting. This went on quietly for more than twenty years before the issue began to surface in the academic media and later more generally.

  The first companies in the medical ghostwriting business emerged in the 1970s but many more entered the market in the 1980s and 1990s; there are now fifty or so medical writing companies in the United States and Britain.8 These specialized companies compete for drug company contracts for a series of articles on the basis not only of price per article and article quality, but also on the stature of the academic “authors” the outfit is able to recruit and the ability of the ghosts to get an article speedily placed in a prestigious medical journal. Many of these writing agencies have since become subsidiaries of public relations companies, who are in turn branches of global advertising companies such as WPP,9 Omnicon, and Interpublic, some of whom have now taken up the clinical trial business themselves.

  Axis Healthcare Communications, one such company, offers to help pharmaceutical companies “brand the science.” Envision Pharma announces that “data generated from clinical trials are the most powerful marketing tools available to a pharmaceutical company.” GYMR boasts that they “know how to take the language of science and medicine and transform it into the more understandable language of health. We advise clients of the best dissemination strategy for their news and make sure that
the message they deliver is compelling, documented and contributes to other national dialogues in a real and meaningful way.”10 Meanwhile, Current Medical Directions offers to deliver scientifically accurate information that has been strategically developed for specific target audiences. To do this, they “write up studies, review articles, abstracts, journal supplements, product monographs, expert commentaries and textbook chapters, conduct meta-analyses, organize journal supplements, satellite symposia, and consensus conferences as well as advisory boards for clients.”11

  Who are the ghostwriters in these endeavors? They are predominantly women, with PhDs or other science qualifications. The work may offer them an opportunity to work from home, possibly on a different continent than that of either the notional lead investigator or the study site. The job requires no familiarity with what actually happened; indeed it can probably best proceed without firsthand knowledge of the trial— that's not the writer's department, as Tom Lehrer might have sung.

  Most of the medical writing companies and independent writers working on contract have put in place protocols or business plans specifying what they do and do not do, and they see themselves as adhering to these specifications and standards. The protocols may, for instance, mention that they will ensure that any submission to a particular journal takes account of that journal's requirement for conflict-of-interest statements and authorship declarations. And in fact, these requirements are much more likely to be met in articles written by medical writers than in articles written by academics.12

  The ghosts see themselves as occupying a position midway between marketing and the practice of science. They are sure that the marketers, left to their own devices, would not be able to resist overt promotion and the resulting product would be rejected at peer review for not meeting the standards of scientific presentation. Left to the academics, the articles would not be ready in even draft form in time for a drug's launch and inadvertently might put the trial's results in less than the best light. The writers see themselves as producing an acceptable compromise.