Feature request: conflict of interest statements

[TomHardwicke]

Super helpful app - thanks! Not sure if this is in your scope, but as you already collect information for funding statements, I wonder if you see an opportunity to capture information from authors for conflict of interest statements? Even if authors have no conflicts of interest, it would be useful to confirm that explicitly with each author.

[alexholcombe]

Thanks Tom, @marton-balazs-kovacs and I discussed this and have an idea for how to implement the aspect to indicate no conflict of interest, which we agree would be quite useful. However we don't have enough experience with how COIs are included in manuscripts to have good judgment on how to format the text together with the author list. That is, how should tenzing spit out the COI text with all the author information? Let us know if you have ideas about this.

[TomHardwicke]

Great! I'm not sure what you mean by together with the author list, but I just imagined this working like the funding statement, so it's a separate output. I guess the simplest case would be all authors say they have no conflicts, then the statement can probably just say "All authors declare no conflicts of interest." In the event that one or more authors do declare a conflict, that changes to "All other authors declare no conflicts of interest" and goes at the end of the statement. Then the individual conflicts are included in whatever manner they were entered as a free text response in the spreadsheet e.g, "T.E.H. declares he is emotionally invested in the success of Tenzing. All other authors declare no conflicts of interest."

If you want some real-world examples, we have many extracted from our studies assessing the prevalence of transparent research practices:

• For psychology, see this data file, column AI
• For the social sciences, see this data file, column AK

If you're thinking of a more structured approach then perhaps check out the ICJME guidelines, there's something on that here in BMJ's policy.

[alexholcombe]

That's super-helpful, thanks Tom - we should go ahead and implement most of this!
My impression is that the term "competing interest" is becoming increasingly popular, over "conflict of interest", is that your impression as well?

For the contributor information spreadsheet that users enter their information into (https://docs.google.com/spreadsheets/d/1Gl0cwqN_nTsdFH9yhSvi9NypBfDCEhViGq4A3MnBrG8/edit#gid=0), to avoid unmanageable clutter we try to avoid adding multiple columns for new features. So one idea is to instruct users to enter "N" in the COI column if they have nothing to declare, otherwise to enter the full text of their COI.

[jcolomb]

"N" is not very good and not necessary
One can think of a system independent of what people write ("author A declares no conflicts of interest, authors B, C, and D declares to be emotionally invested in the success of Tenzing, all other authors declares to need a paper for their CV" ,

ordering via the number of identical declaration.

danger is to get: author A declares no conflict of interest, author B declares no competing interest, but I think this may be a feature more than a bug ?

[TomHardwicke]

That's super-helpful, thanks Tom - we should go ahead and implement most of this! My impression is that the term "competing interest" is becoming increasingly popular, over "conflict of interest", is that your impression as well?

I don't know, I've seen plenty of both. I also don't know of any principled reason to use one or the other. Perhaps reasonable to follow ICJME and use competing.

I may have misunderstood @jcolomb comment, but I do think you need an active declaration even when there's no COI — because there's a difference between an active declaration and someone not paying attention / completing the form.

[jcolomb]

I was not clear. I mean that having a shortcode (Alex proposed "N") is not necessary, and probably not a very good practice. We can keep it as a free text entry and later on in the code check for identical entries to group the authors.

something in the sense of:

# create list of non-identical entries, order by number of 
COI-list = unique (data$COI) %>% 
    order... #code to order via number of authors with that COI

# create output

COItext= ""
for (i in COI-list){
 Author_coi = list() ... #get authors with that COI

COItext= pasteo(COItext, paste=(Author_coi, sep=", "), "declares", COI-list[i]), sep="; ")
}

But indeed, the cell should be tested as being non-empty during validation.

[alexholcombe]

This sounds sensible!
We could have "No competing interest" in the Google Sheet template for at least two authors, but not all, to model good practice of having multiple authors with exactly the same text and showcase that tenzing will group those together.
If the cell is empty, I'm thinking that at validation a warning should be given but not an error, because competing interest declarations are not used in some fields/journals.

[marton-balazs-kovacs]

This sounds great! Besides the contributors' information spreadsheet template, I thought we could also provide a contributorship terms template text document. This document could be modified by the corresponding author at the beginning of the project and outline the types of contributions that are expected from the contributors and the different contribution levels. The corresponding author could also use this document to give project-specific instructions to others regarding the contributors' information template.

For example, please write "No competing interest" in the COI variable in the spreadsheet in case you do not have any competing interests.

This would help the corresponding author to get consistent responses (kind of), but we would not restrict anyone on the app's side.

[marton-balazs-kovacs]

210ebb7

[TomHardwicke]

People may also want to say “is” e.g., “John Smith is the editor of…” and there are probably other possibilities — so perhaps its better to be maximally flexible not to add anything as a default after the name?

…

On 7 Nov 2023, at 19:49, Alex O. Holcombe ***@***.***> wrote: thank you Marton for getting this working. I added an actual conflict of interest to see how the sentence would look. What I added was “has a patent pending for beta-kryptonite.” This resulted in the below: Lex L. Wang declares has a patent pending for beta-kryptonite; John M. Smith and Marton Kovacs declare no competing interest. That's not grammatical and to have the simplest possible default that is least likely to confuse users, I think we should not use “declare” at all but just use the word “has” for an individual, and “have” for groups, so that it would read Lex L. Wang has a patent pending for beta-kryptonite; John M. Smith and Marton Kovacs have no competing interests. which means we should change the default from “no competing interest” to “no competing interests”. What do people think? — Reply to this email directly, view it on GitHub <#69 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABK24KYBFWK4F4SRIUKX4BLYDHKV7AVCNFSM5FZ3OQDKU5DIOJSWCZC7NNSXTN2JONZXKZKDN5WW2ZLOOQ5TCNZZG44TCMJWGM2A>. You are receiving this because you authored the thread.

[alexholcombe]

thank you Marton for getting this working. I added an actual conflict of interest to see how the sentence would look. What I added was “has a patent pending for beta-kryptonite.” This resulted in the below:
Lex L. Wang declares has a patent pending for beta-kryptonite; John M. Smith and Marton Kovacs declare no competing interest.

That's not grammatical. The difficult issue is that a plural verb form is required if there are multiple authors with the same entry, whereas the singular form is required otherwise.

To have a simple default that is least likely to confuse users, maybe we should not use “declare” at all but just use the word “has” for an individual, and “have” for groups, so that it would read
Lex L. Wang has a patent pending for beta-kryptonite; John M. Smith and Marton Kovacs have no competing interests.
which means we should change the default from “no competing interest” to “no competing interests”. What do people think? This will not work in all use cases because the word "has/have" will not be the verb authors will want to use in all cases, for example if their COI is that they are an editor at the journal they will not want to use has/have, instead something like "AOH is an editor at this journal." and for plural, "AOH and TH are editors at this journal." Another case we can't handle is when people write multiple COIs in that field, like "is an editor at this journal, and makes lots of money from his stock options in LexCorp." So I don't think we can entirely solve this, therefore maybe we should just warn users that they may have to change the sentences. What do you all think?

[alexholcombe]

@TomHardwicke, sorry our messages crossed! The problem is the singular/plural handling, which I didn't explain fully. When multiple authors enter exactly the same text in the COI entry, we combine them, e.g. "TH and AH declare no competing interest", which should be declareS if there was only one! Do you see the problem?

[jcolomb]

I think we could keep declare(s) and ask the users to set their message accordingly. the 80-20 percent rule apply here (solve 80% of the cases with 20% of the time, the other 20% takes 80% of the time...) I do think it is not worth investing time to solve these 20% exceptions... at least we should get some feedback on which other cases happen often ?

[alexholcombe]

Marton and I discussed this in a call and have come up with a plan. We agreed that it's ok to not have a full solution, but we want to manage user expectations / don't want users to be surprised when it doesn't turn out fully grammatical, for example.
We are thinking that when a button labelled something like "COI statement" is clicked, a settings box will appear (similar to that for the funding information button). That will have a drop-down for declare/declares, has/have, maybe some other options, which gives the user some insight into how it works and what to expect.

[TomHardwicke]

Great this will be such a helpful feature!

…

________________________________ From: Alex O. Holcombe ***@***.***> Sent: Wednesday, November 15, 2023 12:55:03 PM To: marton-balazs-kovacs/tenzing ***@***.***> Cc: Tom Hardwicke ***@***.***>; Mention ***@***.***> Subject: Re: [marton-balazs-kovacs/tenzing] Feature request: conflict of interest statements (#69) Marton and I discussed this in a call and have come up with a plan. We agreed that it's ok to not have a full solution, but we want to manage user expectations / don't want users to be surprised when it doesn't turn out fully grammatical, for example. We are thinking that when a button labelled something like "COI statement" is clicked, a settings box will appear (similar to that for the funding information button). That will have a drop-down for declare/declares, has/have, maybe some other options, which gives the user some insight into how it works and what to expect. — Reply to this email directly, view it on GitHub<#69 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABK24KYRNAEO4TGLFKUD3ALYEQOHPAVCNFSM5FZ3OQDKU5DIOJSWCZC7NNSXTN2JONZXKZKDN5WW2ZLOOQ5TCOBRGE3DONZQGUZQ>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[alexholcombe]

We noticed that the JATS schematron for XML has a check for correct COI format in JATS-XML, so we should make sure our thing is compatible with that. It looks like it has almost no content, but it would be good if our XML output included the COI statement and validated against the schematron. Same with funding