Most journals do not post conflict-of-interest statements on PubMed
Key takeaways:
- The use of the voluntary conflict-of-interest field — added in 2017 — has increased since its introduction.
- The authors called for more uniformity in how journals present conflict of interest in articles.
Use of PubMed’s conflict-of-interest field has increased over time, but utilization remains limited, according to research published in PLOS One.
Utilization of the field even remains low among high-impact journals results showed.
The National Library of Medicine added the voluntary conflict-of-interest (COI) field to PubMed’s abstract page in 2017.
Peter Lurie, MD, MPH, president and executive director of Center for Science in the Public Interest (CSPI) and colleagues examined use of the field for journals and articles indexed on PubMed between 2016 and 2021.
Researchers calculated the percentage of all journals with at least one article that included a posted COI. They also analyzed 100 randomly chosen articles published between June 2021 and May 2022 from each of the 40 highest-impact journals.
Approximately 7,000 journals published articles each year. The percentage of journals with at least one article that included a posted COI statement increased from 25.9% in 2016 to 33.2% in 2021.
Of the nearly 400,000 articles published annually, the percentage that included a posted COI increased, from 9% in 2016 to 43% in 2021.
Researchers randomly selected up to 100 articles from between 2021 and 2022 for each of the 40 highest-impact journals. Less than one-third (30.2%; 95% CI, 28.7-31.6) had published COIs, 63.3% (95% CI, 60.4-66) of which also included a posted COI in the voluntary field.
“I was somewhat encouraged by these findings,” Lurie told Healio. “This is voluntary, and it’s been 8 years or so. Clearly, there is a certain amount of work involved to implement a process using the COI field, and the numbers are still rising. I’m optimistic, and I’m hoping our article will create an additional impetus to use the field.”
Healio spoke with Lurie about the rationale for the study, the key findings and their implications.
Healio: What motivated you to perform this study?
Lurie: CSPI is the group that pushed for this new field. Before I was part of this organization, they had an advocacy campaign in which they wrote a letter to the National Library of Medicine. They got several people to sign on and ask for the creation of this field. They granted our request to create this field, and we thought because it was voluntary, it would be interesting to see what the degree of uptake had been years after the field was announced.
Healio: How did you evaluate uptake?
Lurie: Broadly, there were two parts to what we did. The first part was a kind of computer analysis, and the second was more of a close-read analysis. For the computer analysis, we asked two questions.
One was about the fraction of journals in any given year that use the field at least once. We worked from the assumption that any journal in the course of a year should have one article for which a COI disclosure would be appropriate. If a journal didn’t use it in a given year, it probably means they missed some opportunities to use it and are not using it fully or at all.
The second, simpler question looked at the fraction of articles that have the disclosure and use the field. We took the top 40 journals ranked by impact factor, and for each of them, we pulled and read about 100 articles. Some of those articles will have COIs that are disclosed somewhere within the published article. A good chunk of them will not have a disclosure, because there is, in fact, no COI to disclose. Among those that do have a COI to disclose, then the question is whether they actually use the PubMed field to disclose it.
Healio: What did you find?
Lurie: We found the fraction of articles and the fraction of journals that used the field both are increasing. The analysis of the articles in high-impact journals showed roughly one-third of them have some kind of conflict; of those, two-thirds of them disclose it — so one-third have a conflict and do not disclose.
Healio: What are the potential explanations for these patterns?
Lurie: The way things get into PubMed, in general, is that when journals that are indexed there publish an issue, they transfer certain elements of the article over to PubMed. To do that, they have to have corresponding fields. Let’s say a journal has a field in its computer that says “title” and PubMed has a field that says “title.” You match those up and do the same for “author” and everything else. When you send it, all the matching fields get transferred over to PubMed. It’s really all done by the journals using coding.
There are a few reasons why the field is not being used. The first is that the journal has not decided to use the field or has not yet written the necessary code. The second is that they use it sometimes but, for reasons that are unclear to us, they don’t use it consistently.
The third problem is that when we read those 4,000 or so articles, we found COI statements in various places. Sometimes the field is called the “conflict of interest” field, sometimes it is in the “funding” section, and sometimes it’s in “acknowledgements.” That funding and acknowledgements information never makes it over to the PubMed field. It’s a bit of a complicated coding problem because there’s not a one-on-one correspondence between the data fields.
Healio: What do you think needs to be done to address this problem?
Lurie: We’ve written to a few dozen journals asking them to look into this, and we’ve gotten a few respectful responses from large journals who say they will.
Other potential solutions might be greater uniformity in the COI sections of journals, or possibly a coding solution. At the end of our article, we cite a group that has evidently written some code that a journal could use to facilitate the transfer. My guess is that to solve the problem of COI lurking in the funding or acknowledgement fields, we will need some more elaborate code than what has been written. And there’s always the possibility of an AI solution.
Reference:
For more information:
Peter Lurie, MD, MPH, can be reached at plurie@cspinet.org.
Collapse
Healio Interviews