Conflict of Interest and Academic Papers - can we resolve it?

[C21Beancounter]

[//]: # "======================= Instructions for posting issues: (1) Write your post (below this text blob) using Markdown (as per https://guides.github.com/features/mastering-markdown/ ) or just plain text. (2) Add the appropriate labels to your post, e.g. 'idea' or 'lightning talk' (or both for idea pitches), 'interests', 'skills' or 'experience'. Use 'help wanted' as needed. (3) Don't worry about these introductory lines - you can leave or delete them, as they won't display anyway (you can check this via Preview). (4) Hit the 'Submit new issue' button. ============================" Background

• At NHS Hackday 2012 a team tried to resolve the issue of Conflicts of Interest (COI) assoociated with Academic Papers (https://www.slideshare.net/ideogram/london-nhs-hack-day-presentation-may-2012 , page 15, 'Coincidence Gate').
• The problem was that COI declarations were submitted in free text and were therefore, not machine readable.
• Our team's solution was to build a device that allowed whistleblowers to submit information about questionable research (e.g. a whistleblower might want to inform the public that authors of a paper research about the efficacy of a drug were, in fact, funded by the company that owned the IP related to the drug itself).
• Unfortunately, our team's "solution" simply shifted the doubt over the integrity of journal authors elsewhere - namely to the integrity the whistleblowers!
• It would have been preferable if we had used a sample of existing COI declarations to create a finite but comprehensive set of COI categories. The categories could be assigned to papers by authors (along with a limited amount of free text, where absolutely necessary). Open Repositories could then publish COI data in a more analysable form.

Where we are today

• Apparently this issue is still unresolved as COI declarations are still published in free text form.

If you want to do something...

• I have a spreadsheet of c100 COI declarations (from 2012!), taken from the (then) c2000 journals related to one particular pharmaceutical product and a (crusty) flow diagram derived from discussions between the team during the hackday (see files, attached, below).
• Feel free to create a structure journal authors could use (plus possibly a digital prototype/proof of concept/tool which used those structures and demonstrated their utility in COI research).

Additional issues to bear in mind a. Identify links between (declared) donors to entities who are NOT declared (e.g. PharmaCo X might be 100% parent/funder to Research Charity X, but only Charity X is declared);

b. At the time of the hackday, there were appparently possible COI issues inherent in the process of pharmaceutical licensing (i.e. companies conducting research on behalf a licensing body may have been connected to the medicine being tested). Im not sure if that is the case today (or whether it was back then). In the EU medicines are licenced by http://www.ema.europa.eu/ema/ and in the UK it is https://www.gov.uk/government/organisations/medicines-and-healthcare-products-regulatory-agency), but I am not aware of the screening processes involved;

c. Academic researchers might work in departments funded by significant 'general' donors with an interest in influencing research (i.e. coercive powers might not be in the form of funds for a 'specific' research paper, they may be indirect). An initiative called HEIDI used to provide info about university funding. I think donor (and other) income might be published here these days: http://www.adls.ac.uk/hesa/hesa-finance/?detail .

d. Conflicts are linked to academic papers, not authors per se. So people need to be careful not to suggest conflicts might exist (e.g. for someone who worked somewhere 20 years ago), when in reality they might not.

Why bother? I assume COI is not restricted to healthcare research, the applications of a solution in this domain could be vast.

Files Conflict of Interest in Medical Research Papers - Part 1.xlsx Conflict of Interest in Medical Research Papers - Part 2.xlsx

[fionabradley]

Sorry I can't attend tomorrow but am very interested in following this!

[HKLondon]

Interesting idea. Do you have a vision of what a machine readable COI would look like?

If I take this paper as an example: http://www.bmj.com/content/356/bmj.j510 (This is a paper published by the company I work for). The competing interests statement is pasted below. If we wanted to make it more machine readable how might we go about doing this?

Competing interests: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: support from the Obel Family Foundation for the submitted work; TBL has served as an investigator for Janssen Scientific Affairs, LLC, and Boehringer Ingelheim and has served as a speaker for Bayer, BMS/Pfizer, and Boehringer Ingelheim; PBN has served as a speaker for Boehringer Ingelheim and consultant for Bayer and received unrestricted research grant from BMS/Pfizer; GYHL has served as a consultant for Bayer, Astellas, Merck, Sanofi, BMS/Pfizer, Daiichi-Sankyo, Biotronik, Portola, and Boehringer Ingelheim and as a speaker for Bayer, BMS/Pfizer, Boehringer Ingelheim, Daiichi-Sankyo, and Sanofi Aventis; FS has served as a consultant for Bayer.

[Daniel-Mietchen]

Thanks for the example. The basic idea in making this machine actionable (as opposed to machine readable) is to use machine actionable identifiers (MAI) when possible, and to limit the choices to unambiguous options. Ideally, the MAI would also be globally unique.

Let me walk through your example a bit:

Competing interests

This section could itself not just be tagged in terms of being a document section but also as being about the concept "competing interests" (which could have some MAI) or containing "Competing interest statements" (which might have a MAI as a concept and also as the specific instances of that concept, which uniquely identify the statements by the specific authors of that paper).

The phrase

All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf 

could actually be more or less hard coded, giving authors the response option "yes/ no", and in the case of the latter, the phrase would not be shown in the manuscript (and the manuscript would probably not be published).

Some form of the

and declare

could be inserted automatically.

support from the Obel Family Foundation for the submitted work

• "support" should have a set of predefined options, e.g. financial support, material support, legal support, in-kind support, logistic support etc., possibly with some additional level of granularity.
• "the Obel Family Foundation" should have an MAI that should be stated somewhere in the machine readable version of that statement.
• "for the submitted work": the kind of support indicated under "support" should have some limited set of predefined options to express whether and how the support is limited in scope, e.g. "for this manuscript", "for research project X" (with X having an MAI), "for this person" (with the person having an MAI such as ORCID) etc.

TBL has served as an investigator for Janssen Scientific Affairs, LLC, and Boehringer Ingelheim 

• "TBL" should uniquely identify the person in a straightforward manner. the current way of using initials is not straghtforward even within the context of this manuscript; a machine readable version should just repeat the MAI for that person in this statement, e.g. their ORCID.
• "has served" should be tied to some controlled vocabulary about roles that people may have had in such contexts, which is aware of things like "investigator", "speaker", "consultant" etc., again possibly with additional levels of granularity.
• all of the "Janssen Scientific Affairs, LLC", "Bayer", "BMS/Pfizer" and "Boehringer Ingelheim" should be referenced by their respective MAI as described for "Obel Family Foundation" above, and these statements should not be lumped together but tied to the specific roles mentioned in the previous bullet point.

The rest of the statement would then apply these principles to the remaining authors.

[C21Beancounter]

Hi Finona/HL London.

FB: I should have put this under Daniel's thread about things that others could take forward thread (because I am working on something entirely separate and cant work on this!).

HKL: I never actually got round to sketching out a vision. Indeed, I have however, just read Daniel's response and I couldnt have put it better myself. Such a pity I didnt raise an issue for this earlier!