The fallacy of impartiality: competing interest bias in academic publications

Abstract

Bias comes in a thousand forms, but even so we think that we have identified a new form: ‘competing interest bias’. This new form is illustrated nicely in an article to which we have contributed that has recently been published in PLoS Medicine.1 In one half of a debate article we have argued that the private sector can make an important contribution to improving healthcare for the world's poor. Four authors from the London School of Hygiene and Tropical Medicine have argued the opposite. In the competing interest statements that accompany the article, the four authors from the London School declare that they have no competing interests, whereas our statements of competing interests run to over 300 words. Two questions arise immediately: is it true that academics, including one of us (RF), do not generally have competing interests, except in those obvious cases of links to the manufacturers of a product that is a subject of the paper in question? And, will readers discount our arguments because we have declared competing interests, when our opponents have not? We should make clear at this point that we are not proposing for an instant that competing interest statements should not be published. We agree that they should be as full as possible.2 But is it true that the authors from the London School, for example, have no competing interests? We are not suggesting that they have been deceptive, but are they subscribing to a common – indeed, near universal – view that those who are linked with for-profit companies are heavily conflicted whereas those employed in public or academic institutions are, generally speaking, not? People who work for public sector institutions regard themselves (and are often regarded) as being neutral, disinterested, and unbiased supporters and defenders of the public interest. There is, however, a large literature by economists and political scientists known as ‘public choice theory’ (that even has its own scholarly journal, Public Choice) that demolishes this pretension.3 Public institutions and the individuals that work for them are found to be self-interested, much like private institutions and their employees. Government bureaucracies seek to maintain and expand their scale and influence, a reality which is captured in arguments against the ideal of impartial civil servants in the Weberian bureaucracy.4–7 United Nations agencies fight over territory and mandates. Individuals working for public institutions with a certain culture (such as the London School of Hygiene and Tropical Medicine, where one of us [RF] was the dean) know that their career prospects may be advantaged by being a part of that culture rather than iconoclasts. As others have noted, being a ‘public servant’, or an ‘international public servant’ or the employee of a university does not make one un-self-interested or un-conflicted.8 Academics, especially in applied fields such as global health and medicine, often have numerous relations with not-for-profit organizations – including governments, foundations, non-governmental organizations and United Nations agencies. These relationships typically include some combination of remuneration for advice or assistance, research funding (which may also include salary support for the principal investigator) and travel support. More generally, these relationships are likely to be career enhancing, as when an academic has multiple relations with the World Health Organization (WHO) and is frequently called upon by WHO for services of various kinds. Many of the organizations with which the academic has relations have stated positions on issues affecting public health and indeed many other topics. Surely there is potential here to influence an academic's expression of views – in other words a potential conflict of interest worthy of declaration. On the second issue concerning whether readers discount the views of those who have declared their potential conflicts of interest, the answer seems to be yes. There is evidence from two studies, in which one of us (RS) participated, that shows that competing interest statements can have a strong impact on how readers perceive articles.9,10 In the first study readers were sent a paper that showed that pain from herpes zoster may have a substantial impact on patients' daily functioning. Readers were randomized to receive either a version of the paper with a declaration that the authors were employees of a fictitious company and potentially held stock options in the company or a version that said the authors were from an ambulatory care centre and had no competing interest. Readers were asked to rate the study in terms of interest, importance, relevance, validity and believability – and they rated the paper where the authors declared a competing interest statement significantly lower on every criterion. Readers presumably thought that employment by a company manufacturing a drug to counter the pain of herpes zoster meant that the authors would exaggerate the impact of the pain. A second study, using a similar design but also including a second paper on the use of problem lists in letters between hospital doctors and general practitioners, showed that the impact of a competing interest statement was less if authors simply had a grant from a company rather than were employed by the company, and if it was less clear how the competing interest could influence the study. Confused thinking over competing interests is common – as the recent article by the editors of PLoS Medicine shows.2 We would argue that this bias against the private sector extends even deeper. For example, one of us (RS), is not able to write for the BMJ, which he used to edit, on health policy issues because he is now employed by a for-profit company. Yet editorials in the journal on general practice are routinely written by general practitioners, who are not just the employees but also the owners of for-profit organizations, their practices, with a significant financial stake in the status quo. Our message is simple: we must recognize that we are all conflicted and declare accordingly.11 A view of the world that sees employees of private for-profit companies as conflicted and doctors, or employees of public or academic bodies, as not, is naive, potentially deceptive and likely to distort reader response to new information.