The topic to be discussed this week is the ‘Marshalling of Science’, or how pharmaceutical companies exert explicit control over every detail of the drugs they create: from the design of the study, to the results they achieve, and even the explicit favouring of positive data, while excluding negative data, in the academic publications they produce. An article from Alastair Matheson, and two from the brilliant Sergio Sismondo of Queen’s University, Kingston, demonstrate how pharmaceutical firms achieve these ends, and what may be done about it.
Corporate Science and the Husbandry of Scientific and Medical Knowledge by the Pharmaceutical Industry – Matheson
In this first article, Alastair Matheson explores how both pharmaceutical and medical device industries operate between commerce, medicine, and science, to produce scientific knowledge with commercial suitability. (355) To understand how this is done, we are provided a brief overview of the pharmaceutical sector and directed to two domains of expertise (marketing and medical) which become increasingly important as basic research and development transitions into clinical development. Marketing experts, as Matheson describes, “are involved with traditional promotional activities, including branding, market testing and advertising, while dedicated medical experts, including qualified doctors and scientists, are involved in ongoing clinical developments…” (358)
Pharmaceutical companies shape medical and scientific knowledge – the direction and volume of particular products (in this case research and developments) are influenced by market considerations. Not only is the efficacy and safety of a drug considered, but also the ability of the pharmaceutical companies to “secure regulatory approval at the least possible risk, and to bolster marketability.” (Matheson, 359) Although these companies strongly influence and develop studies which result in generally favouring the sponsored products, design bias is often difficult to identify as the studies themselves turn out to be statistically and methodologically robust.
To understand new drugs, biological processes, and the areas of medicine in which a drug will be used, pharmaceutical companies have developed an important tool termed the ‘product cannon’. The information derived from this tool is important for pharma because it will be fed back into the current clinical developmental program with which it is affiliated, and new trials will be designed specifically with the goal of confirming it. Resulting trials that confirm information derived from the product cannon will be considered successful – though it is hard to see how the company’s self-confirming strategy would lead to other results. The results of the product cannon are then transformed into ‘key messages’, typically deployed in scientific journals as part of ‘drug narratives’, which are designed to address “why the drug is needed, how it benefits the patient, why it is superior to its competitors and why it is cost-effective.” (Matheson, 361) Additionally, drug narratives may seek to reformulate professionals’ understanding of certain diseases by increased medicalization, often with a reluctance to speak about weaknesses of the drug in question. Further persuasion of potential prescribers and purchasers may stem from endorsement of a drug by key opinion leaders (KOLs), or medical standards councils, and even through incentives to believe a drug narrative (which may include honoraria or rewards). (Matheson, 2008)
In order to release only pertinent information, and successfully market a particular drug, pharmaceutical companies appear to have fairly strict regulations surrounding their discourse structure. As Matheson notes, these companies “modulate the people and networks through which discourse proceeds, and the institutional settings in which discourse is grounded and science is done.” (369) Specifically, key opinion leaders may be “constructed and regulated with respect to what they study, where they go, what they say and write, and with whom they interact.” (Matheson, 369) Despite an abundance of information, uncertainties and non-beneficial results of research are husbanded to preserve the strength of the narrative.
A last subject of particular interest as discussed by Matheson, is the role or goals of pharmaceutical companies within the greater organization, or dispositif, of medical-scientific knowledge production. What these companies will deem successful will be scientific research that delivers a sales-supporting output; their goals are to understand the “operation of science itself to an external end.” (Matheson, 372) As mentioned above, uncertainties and non-beneficial results of research are husbanded, which may increase criticisms of the truth-claims arising from a particular pharma program
“…as an engineered mesh of puzzle/solution narratives. …At every point, from study selection through design and methodology, data analysis and the projection of ‘drug narratives’, the choice exists to select the option likely to yield the ‘best’ data or argument – where ‘best’ relates not to truth, but to the beautification of the product.” (Matheson, 373-4)
Though outright deceptions are likely rare, the development of safeguards including an International Standard of Integrity in Science may be beneficial. To be formulated by leading journals, this safeguard would collect information including: “which company financed the publication; which specific drugs were being promoted, highlighting the lead drug; [and] if the KOL was receiving payments directly and/or in kind…” (Matheson, 375) Since it is not likely that the complications of pharma’s contributions to scientific knowledge will be eliminated, what is needed are carefully constructed policies to help corporate and academic science retain their vibrance, incisiveness, and freedom. (Matheson, 2008)
How Pharmaceutical Industry Funding Affects Trial Outcomes – Sismondo
In this short piece by Sismondo, we focus on the causal routes by which pharmaceutical industry funding affects trial outcomes. Biases created by pharma funding for example, influence the medical literature, and further, this funding has the potential to influence outcomes of clinical trials, increasing the chances of pro-industry results with an odds ratio of between 3.60 and 4.05. (Sismondo, 2008) Pharma funding to publication planning (a marketing strategy in which pharmaceutical companies plan and shape many steps in research, analysis, writing, and publication of articles) is one such creator of this bias.
As described by Sismondo, the pharma companies not only fund clinical trials, “but they also design and shape them. …[C]ompany statisticians usually perform statistical analyses…hired medical writers produce first drafts or edit many papers…And, medical communication companies expertly shepherd manuscripts through the publication process.” (1910) A goal of these companies with the teams they hire, is to develop a relationship in which individual researchers’ dispositions are changed, by seeing the pharma company and their products in a more favourable light, after being provided gratuitous sponsorships (research grants, and the like).
How else might trial outcomes be affected? Sismondo describes in detail five particular causes, which demonstrates that the problem at hand is not only complex, but in need of a comprehensive solution. The first cause is design bias, where funding is provided to designs with an increased likelihood of achieving favourable results, including those involving placebos inappropriate doses, or carefully constructed experimental populations.
Secondly, multiple trials may be created with predictable outcomes after drugs are on the market. This serves to familiarize prescribers with products by designing trials for a population in which researchers already know a drug to be effective.
Thirdly, scientific misconduct may be present in many clinical research trials, but approximately 24% of surveyed medical specialists report questionable behaviour by sponsors in industry funded trials. This may have included “the company prematurely terminating studies, changing protocols while studies were in progress, and writing first drafts of reports.” (Sismondo, 1912)
Industry sponsored trials’ funding can also affect the interpretation of data and the manner in which articles are written. Conclusions stemming from these trials “are shaped as much or more by funding source as by data.” (Sismondo, 1912) Lastly, publication bias in industry funded trials often leads to positive data being over-reported compared to negative data.
Sismondo concludes that post-hoc attempts to correct biases are not likely to be successful. Rather, in order to maintain current levels of funding and innovation, one suggestion may be to nationalize clinical trial planning and funding. This suggestion would dramatically reduce the influence of the pharmaceutical industry however, and there, unfortunately, lies a problem.
Ghosts in the Machine: Publication Planning in the Medical Sciences – Sismondo
In building upon briefly addressed considerations from the previous article, Sergio Sismondo addresses the trend of pharmaceutical firms to employ company-sponsored research with the goal of being developed and deployed in such a manner that they will positively affect the opinions of researchers and prescribers: in other words, the trend of ‘publication planning’.
In order to gain the most ‘commercial value’ from research, industry funded journals promoting a specific product are often ‘authored’ by key opinion leaders. As first authors, these KOL’s are recognized as crucial contributors to articles, but are often portrayed as lazy, only contributing by reviewing penultimate copies of the articles (contrary to International Committee of Medical Journal Editors’ criteria). The major contributors on the other hand, which may include statisticians, researchers, medical writers, and others, are rarely acknowledged in the publications. This cannot wholly be attributed to the ‘laziness’ of KOL’s however, since, as one speaker in Sismondo’s work estimates, 50% of pharmaceutical companies only show KOL’s the penultimate manuscript, critically limiting the possibility for input. (2009)
Although publication planners focus primarily on publications, their influence is not limited – they are also influential in research design. These planners see themselves as needing to ‘create a market’ or an ‘understanding of an unmet need’, but are often criticized as being involved in ‘disease mongering’, and ‘selling sickness’. (Sismondo, 2009) Publication planners must, however, pay attention to marketing. It is simply good business for these plans to address “‘target audiences’, lay out key ‘scientific and clinical communication points’, perform ‘competitor publication and gap analyses’, and outline ‘top-line tactics’ and ‘critical timing’. (Sismondo, 179) The planning is meant to generate revenue by producing information that will increase sales, and it does so almost entirely through publications, that is, without direct contact with physicians.
The work of publication planners may in some respects seem trivial, however, this work can involve important and extensive decision-making. As noted by Sismondo, “[i]n addition to making their own contributions, planners facilitate their teams’ work, keeping in contact with medical writers, making sure that all documents produced are consistent with the plan, managing information, and reconciling divergent demands and suggestions.” (190) The work of the planner, in other words, is creative mediation between a multitude of people, that aims to develop a complex manuscript that will fare well in peer review. We have a commercialized science whose published work is valuable as a marketing tool, not by itself, but with the help of KOLs to endorse it. That being said, we must now hope that “a combination of authorship guidelines, standardized procedures for the performance and analysis of clinical trials, and standard formats for journal papers will control for problems of bias,” (Sismondo, 194) something that is not easily identified or addressed in the realm of publication planners and ghost management.