Bias is endemic in research and is not always a bad thing. I am consciously biased by my value system and funder (Cancer Research UK) to choose topics to research that I think will lead to a reduction in tobacco use. I may be unconsciously biased by the fact that I receive research funding from, and undertake consultancy for, the Department of Health and a number of pharmaceutical companies, and by the fact that I personally like and respect many of the people in those organizations with whom I come into contact. I hope we can all agree that it will be impossible, and not always desirable, to remove bias; neither will it be practicable to document with every grant application, review committee membership, editorial role or written output the full panoply of potential sources of bias arising out of conflicts of interest conceived in broad terms. However, we have to take account of potential conflicts of interest to some degree. The fact that different organizations and journals have different ways of doing this is creating a considerable amount of wasted effort, and it appears to be the case that important conflicts of interest are falling through the net.

The ‘common standard’ document [1] has many useful features and the goal of a single common set of principles is desirable. However, there will probably be concerns in some quarters that some of its provisions will add an unnecessary or unacceptable burden on journals and those who write or review for them. In this commentary I would like to address just one particular issue and discuss one possible solution. The issue is that full conflict of interest disclosures will be overly long, contain a great deal of redundancy and still allow important potential sources of bias. One solution might be a ‘traffic light’ system, which alerts users to the potential risk of bias without the need for detail. It should be noted that the interpretation of the signal would not be as in the United Kingdom, where red always means ‘stop’, but as in some other countries (which should remain nameless) where stopping appears to be discretionary! It is also extremely unlikely that one version of the system would suit all needs.

The point about redundancy can be illustrated easily, as follows: Dr X works for a tobacco company. He (who happens to be a man) has written a review on the benefits of nicotine in aiding concentration. Being employed by a tobacco company leads to such a serious risk of bias that there is nothing else he could say about his conflicts of interest that need be said, so mentioning that he has shares in a tobacco company or has received research funding from a pharmaceutical company or is a smoker provides little further useful information.

The ‘traffic light’ solution proposes a tariff of conflicts of interest according to the potential risk of significant bias. Whether a particular disclosure falls into the ‘red’, ‘amber’ or ‘green’ category is determined by the most serious of these, and any less serious ones can be ignored. What particular journals or organizations choose to do in the case of the different traffic signals would be a matter for them. For example, at Addiction a ‘red’ might generally preclude publication of reviews or editorials but might not preclude research reports or letters. Even reviews or editorials might be allowed under exceptional circumstances. A ‘red’ might also preclude taking on any editorial or review role relating to a particular paper or topic. Research reports that are published would carry the ‘red’ warning and cite the reason(s) as listed below. ‘Amber’ would require the most serious conflicts of interest to be declared alongside the label ‘amber’. ‘Green’ would obviously be anything that was not red or amber.

Here is an early prototype of typical conflicts of interest relating to a specific ‘topic of study’ (e.g. an intervention, or a paper written by someone in one's research group) and into which category they might fall:

1 Any receipt of funds (e.g. through employment, shares, fees, research grants) within the past 5 years from an interested party (IP). An IP is a company/industry with any financial interest in the topic of study (TOS), or an organization acting as an agent for such a company/industry in any capacity (e.g. lobbying, ‘watchdog’, media relations, distributing research funds)Red
2 Any strong religious beliefs or values directly relating to the TOSRed
3 Ownership of patents, copyrights or businesses relating to the TOS, the financial gain from which could be influenced by the conclusions drawnRed
4 Grievance against, family tie with, close friendship with or close working relationship with one or more individuals who have a direct personal interest in the TOS (e.g. author of a paper being reviewed)Red
5 Travel funds or hospitality within the past 5 years from an IPAmber
6 Receipt of research funds within the past 5 years from a charity with an interest in the TOS, not linked to an IP, or a governmental agencyAmber
7 Working in a paid or voluntary capacity with a charity or governmental agency with an interest in the TOSAmber

Even at first glance it is clear that this system works only in relation to a specific topic of study. An individual may be red for one TOS and green for another. A different system may well be necessary for more general duties, such as being editor of a journal. There is much to debate about this proposal and, indeed, the area as a whole.

We have come a long way in the past few years, but still have a long way to go to achieve a system that is efficient and transparent. Perhaps a ‘traffic light’ system of some kind (not necessarily the one proposed here) will prove a move in the right direction. Perhaps the concept of traffic lights has merit, but different organizations will need to develop their own systems.

Note: this commentary is written in a personal capacity, not as Editor-in-Chief of Addiction.

Declarations of interest

Red (1, 3 and 4 above).