Marketing trials, marketing tricks — how to spot them and how to stop them

To understand the interaction between marketing and clinical trials, it is necessary to shift perspective from “marketing trials”, which define only one point along a spectrum of commercial influence, and map the problem more systematically. In particular, it is helpful to separate the marketing functions industry trials perform from their marketing-related features, that is, the details of framing, design and reporting which have been commercially configured the better to serve marketing goals.

As Table 1 illustrates, various strategic considerations determine the choice of trials manufacturers undertake. Once the selection is made, however, trials serve numerous marketing functions. Most obviously, they generate reservoirs of data, as a basis for evidence-based medicine, and equally, evidence-based marketing. This is the case for off-label as well as on-label research, and a seemingly innocuous trial exploring a new use of a drug may constitute off-label marketing — indeed, this is sometimes a specific strategy [7].

Trials are also a tool for recruiting and leveraging key opinion leaders (KOLs) [8, 9]. A large trial keeps in place a substantial group of KOLs, maintains high-level interest and helps the company forge relationships with and among KOLs, creating a network of “influencers” to serve the company’s interests. Large long-term trials are also used directly to promote drugs: for example, most angiotensin receptor blockers had long-term outcomes trials [10‐15] which addressed clinically relevant issues but were also used during their courses to generate interest and product differentiation, for instance, at congresses or through journal publications. Finally, trials can be used for product seeding, by spreading the study population thinly across numerous investigator sites, each of which enrolls relatively few patients. Such “seeding” features do not only function, however, to familiarize prescribers with the product. They promote regular contact between the company’s representatives and prescribers, building familiarity and trust and assisting sales not only of the study drug, but of other products. They also facilitate “internal marketing”, that is, generating enthusiasm for the product within the company; they help local teams acquire greater knowledge of the product; and when recruitment is a challenge, they may enable trials to be completed more quickly. There may also be regulatory considerations; for instance, some countries require their own nationals to be involved in pivotal trials for a license to be granted, and the need for an ethnically diverse population may also increase the number of study sites and countries [16].

None of these marketing functions, with the exception of seeding, necessarily compromises the quality of the science in industry trials. The best industry trials of promising new drugs, including many pivotal regulatory trials where the company stands to lose billions if the study fails, have robust designs and are objectively reported. In other cases, however, marketing encroaches into the trial’s features, influencing the framing of research questions, steering the design choices, driving the accumulation of biases and shaping the reporting of results (Table 1) [17‐25]. The foremost difficulty with industry research is not the fatuous marketing trial, but the coexistence within the same clinical studies of defensible science and subtle marketing spin, rendering the results harder to evaluate, but easier to sell.

Who is responsible for commercial encroachment into this science? To some degree, the academic authors are culpable, whether through lack of awareness or poor judgement. Many such individuals are trusted partners who enjoy financial relationships with the manufacturers [26, 27] and know that if they disappoint their patrons, they may not be invited back. The primary culprits, however, are the companies themselves, who in most cases are proprietors and corporate authors of this research, and its one essential component: the academic authors and institutions recruited into these projects are replaceable by others [28]. To refer to companies as mere “sponsors” of these trials or providers of “funding” or “support” conceals their true role, and such terms should be discarded.

To demonstrate the extent of corporate involvement, it is first necessary to distinguish between true industry projects funded and instigated by manufacturers, and studies that merely received a degree of industry funding. Barbour et al. kindly shared their database to enable me to undertake this analysis (see Additional file 1). A total of 68 trials in their cohort of 194 studies received exclusive manufacturer funding, and in the large majority of these trials, the company was involved directly in design (82%), analysis (83%) and also reporting (76%), deploying company employees as coauthors (88%) and using trade writers (69%). Lundh and colleagues have previously reported similar findings in this journal for a cohort of articles published in the Lancet [29]. Furthermore, designs suggestive of seeding were not limited to a few “seeding trials” but were commonplace among the manufacturer-funded trials in the cohort. The median number of investigator centres in trials funded exclusively by the drug manufacturer was 111, but remarkably, the median number of patients randomized per centre was only 9. These findings suggest that seeding is a widespread feature of manufacturer-financed clinical research.

Yet while these trials are corporate projects, they are not presented to readers as such. The most visible form of bias in this literature is attributional bias, wherein the role of the academic participants in the study is highlighted and industry’s downplayed [28, 30, 31]. This ensures industry work is presented to readers under the lead authorship of credible academics, and by reducing the impression of commercial influence, may also increase the prospect of journal publication [32]. Barbour et al. did not investigate attributional spin, but I identified all the industry-financed articles in their cohort with both academics and industry employees as coauthors, which numbered 70 in total (see Additional file 1). Among these, all 70 had academic lead authors: there was not a single article where an employee of the company fronted the work. This is the first time the systematic fronting of industry projects by academics has been reported for a defined cohort of articles.

Readers of journal articles that report clinical trials should always check for industry funding, industry involvement in research and reporting, ties between academic authors and the manufacturer and academic lead authorship. When present, these features should alert readers to the possibility that marketing is at work, and mandate careful scrutiny of the article. But what of the details of framing, methodology and reporting? Can readers be provided with a simple checklist to spot bias at this level?

The answer, unfortunately, is no. As Barbour et al. correctly conclude, “individual trials have a unique combination of features reported in the journal publications. The pattern of features makes marketing-influenced studies difficult to identify by the average reader.” Certain features, such as uninformative comparisons, noninferiority designs, surrogate or composite endpoints and speculative conclusions, should raise concerns, but none points with certainty to commercial bias, and in many cases they are reasonable or agreed with regulators. Positive spin in the conclusions affects noncommercial as well as commercial research, and is in any case encouraged by journals [33, 34]. Expert reading of articles by clinicians in the field and authorities in trials design can identify many flaws, and readers should seek independent commentaries and online forums for further insight. However, perhaps the greatest difficulty with commercial bias is that it can prove undetectable even by experts on the basis of the information reported in the article, and may only become fully apparent on rare occasions when the manufacturer’s study reports and database trial data are independently scrutinized [35]. Ultimately therefore, the question of how to spot marketing tricks has a troubling answer. In many cases, they cannot be spotted, and certainly not by the everyday prescribers who are this literature’s primary targets.

Barbour et al. rightly argue that only independent research can fully remedy these problems, but in its absence many actions are possible. The concept of research integrity should be recalibrated such that commercial secrecy, spin and bias are viewed with the same gravity as individual falsification and fraud. There have recently been encouraging steps towards data sharing [36‐38], but far more must be done to ensure academics have ready access to all commercial trials databases, study reports and protocols, including data from past trials for licensed products, and without undue red tape.

Among the various other possible interventions, I briefly consider two. Firstly, two important resources, the CONSORT trial reporting standards and the Cochrane Risk of Bias Tool [39, 40], require further development. Good CONSORT compliance does not necessarily point to high trial quality [41], and both CONSORT and the commonly used trial assessment and bias tools have been criticized for lack of scope and detail [42‐44]. Commercial biases may escape detection not only because they are subtle or hidden in the published article, but because they are poorly captured by the available bias categories. This is particularly the case with what might be termed “design bias” or “hard-wired bias” involving details such as the choice of comparator, drug doses, dose escalation schedules, efficacy criteria and adverse event coding [21, 44, 45]. The Cochrane Handbook for Systematic Reviews addresses this difficulty in a number of ways and the Risk of Bias Tool includes an “other biases” category [44], but the issue remains poorly articulated. In some cases, CONSORT compliance and favourable bias scores may have the unintended effect of conferring credibility on research that does not deserve it, and it is notable that advocates for the commercial writing trade draw attention to the CONSORT compliance of the articles they develop as evidence of their validity [46, 47]. These considerations challenge the trials community to improve the guidance for reporting trials and assessing their quality, particularly in the setting of industry research. The sheer diversity of potential biases in commercial trials and their close relationship with the clinical particularities of the trial setting may make this a thorny task.

Secondly, what of Barbour’s primary concern, the journal articles themselves? Many journals and their publishers have substantial conflicts of interest in respect of industry, whose trials fill their pages and whose reprint purchases swell their revenues [48‐50]. Journals dependent on industry patronage have a financial incentive to publish commercial research even when its importance is low, and to allow trial reports to be framed and communicated in a fashion which assists their marketing functions. Yet journals are the custodians of the academic medical canon and have both responsibility and power to demand open data, rigorous scientific and reporting standards and appropriate attribution. Not all journals are fulfilling these responsibilities with adequate scientific zeal, and because this problem is enduring, systematic and potentially dangerous to patients, it should be considered as important a challenge to the integrity of academic medical literature as research misconduct.

There have been calls for journals to cease publishing industry trials [51, 52], but a more realistic goal is to raise journal standards. The International Committee of Medical Journal Editors (ICMJE) guidelines recommend a number of steps which improve the transparency of commercial literature, but they remain permissive of pharmaceutical marketing [53]. No article should be deemed acceptable for journal publication without the accompanying publication of the trial protocol, completed to SPIRIT standards [54], an unsparing account of the work’s scientific and clinical limitations, and prompt access to patient-level data and clinical study reports. Furthermore, since the flaws and biases in this literature are often difficult for readers to identify, a stronger onus should be placed on the trials’ own stakeholders to report any features which might favour the product, as a matter of research integrity. Altman and Moher have called for journals to require lead authors to make a signed declaration of the work’s honesty, accuracy and transparency [55], and building upon this approach, journals should require authors and product manufacturers to supply detailed information on the commercial provenance and design features of the work, a task that could be accomplished using a checklist (Table 2).

The uncomfortable truth, however, is that journals and publishing corporations inured to industry patronage are unlikely to seek reform of a culture in which they prosper. One final possibility might therefore be the development of a new editorial standard, supplementary to that of the ICMJE and more exacting, for journals committed to rigorous science and reporting. If readers came to recognize the standard and its logo as a mark of scientific stringency, this might stimulate wider uptake by journals and raise the quality of clinical trial reporting.