Be said to be influenced by personal prejudice (six letters)

In August last year the Lancet published a narrative review of melatonin-based therapies for depression by Professor Ian Hickie and Associate Professor Naomi Rogers (Hickie and Rogers, 2011). The review highlighted the potential worth of the drug agomelatine, which is manufactured by Servier. In a disclosure that ran to a couple of column inches, both authors revealed potential conflicts of interest arising from benefits received from numerous external funders. Specifically, Professor Hickie disclosed that he had ‘led projects for health professionals and the community supported by . . . drug industry partners [including] Servier’ and had participated in ‘a multicentre clinical trial of agomelatine’ and ‘family-practice-based audit of sleep disturbance and major depression, supported by Servier’. Associate Professor Rogers disclosed that she had received ‘grant support’ and ‘honoraria for lectures’ from Servier.

In January this year, the Lancet ran six letters critical of the Hickie and Rogers review. All of the letters declared that the original papers’ claims of clinically meaningful or relative efficacy of agomelatine could not be supported by available data and/or that the authors had overstated the tolerability of the drug (Barbui and Cipriani, 2012; Howland, 2012; Lloret-Linares et al., 2012; Carroll, 2012; Jureidini and Raven, 2012; Serfaty and Raven, 2012). One of the letters explicitly noted that the authors’ acceptance of benefits from Servier may have biased their interpretation of the evidence (Lloret-Linares et al., 2012). Two others, laced with rhetorical flourish, clearly implied that the review may have been penned for ‘promotional effect’ (Carroll, 2012) or that Elsevier, the publisher of the Lancet, might gain improper financial benefit from its publication (Jureidini and Raven, 2012). Hickie and Rogers responded to the criticisms (Hickie and Rogers, 2012) and the dispute splashed into the wider media (Dunlevy, 2012; Griffiths, 2012).

The controversy raises two distinct issues that are often poorly understood and frequently confused with one another. The first issue is the possibility of unconscious bias. To be biased is to have one’s opinions influenced by personal prejudice (Macquarie Dictionary, 2009). There are myriad potential sources of personal prejudice in academic life. The most obvious are the receipt of financial benefit from a third party, such as the research support and honoraria Hickie and Rogers disclosed. Others, less obvious and rarely disclosed include: the known possibility of receipt of future benefit; the personal investment of time and energy into supporting a belief, claim or cause (intellectual passion); and the existence of strong personal alliances or enmities.

All reporting of data involves some degree of interpretation and all interpretation is open to unconscious bias. It is important therefore that steps be taken to minimise that bias by manoeuvres such as blinded peer review, and since unconscious bias cannot be eliminated, potential sources of personal prejudice should be disclosed so that readers may note them and include them in their calculation of a manuscript’s worth. It is also important that other academics who disagree with any interpretation of data offered be given an opportunity to present their objections and state an alternative case. This is how our system works and how it is seen working in the adverse commentary on the review’s conclusions.

The second issue is the possibility that authors might deliberately seek to mislead or deceive readers by the conscious manipulation of data so as to achieve some personal benefit. While unconscious bias is an ever present and unavoidable possibility to be countered primarily by disclosure, the deliberate fraudulent misrepresentation of data is a matter of professional misconduct that can only be countered by a regulatory regime that will impose harsh penalties for proven occurrences.

The data set out in the Hickie and Rogers review is complicated and extensive. It seems likely that any conclusions that might be drawn from the data are matters upon which reasonable minds might differ. It seems quite reasonable (and indeed non-controversial) to claim that Hickie and Rogers may have been unconsciously biased, or at least that there was room for a reasonable apprehension of unconscious bias. However, there was apparently no evidence at all that the authors had set out to mislead or deceive.

The Lancet was wrong to publish correspondence that made thinly veiled and entirely speculative claims of professional misconduct. Allegations of that sort should not be made without firm evidence and should be directed to regulatory bodies, not correspondence columns. A proven breach of professional standards should result in disciplinary action and be followed by a public retraction of any implicated publication.