How to better flag retractions? Here’s what PubMed is trying

Hilda Bastian
Hilda Bastian from the National Library of Medicine

If you’ve searched recently for retracted articles in PubMed — the U.S. National Library of Medicine’s database of scientific abstracts — you may have noticed something new.

In fact, you may have had trouble ignoring it, which is sort of the point. “It” is a large salmon banner that looks something like this:

retraction banner



Check one out here.

We were wondering if the aims were similar to those of the retraction database we’re beginning to build — i.e., to make retractions more transparent — so we asked the National Library of Medicine’s Hilda Bastian to explain the changes.

Retraction Watch: PubMed has recently introduced a large banner on abstracts of retracted articles. Why?

Hilda Bastian: Retracted articles have been prominently stamped in PMC (PubMed Central), but retraction was part of the “fine print” on PubMed records. It was more easy to overlook. Now retraction status is prominent in both PubMed and PMC. PubMed alone can’t stop the continuing problem of people being unaware of a retraction when they use or cite an article. [Ed: See this post for some background on the problem of articles continuing to be cited after they’re retracted.] We can reduce the chances of it happening if they first come across an article in PubMed after its retraction, though.

RW: Links to relevant Federal Register notices about Office of Research Integrity findings of fraud and misconduct were also recently introduced. Was this part of the same effort?

HB: No, this is a new effort to review post-publication activity comprehensively, now that PubMed Commons has become a permanent part of PubMed. But we are looking at the visibility and accessibility of these findings across the PubMed system as well.

RW: Are there other plans for making corrections and retractions more visible on PubMed?

HB: We are in the process of evaluating post-publication activity on literature in PubMed in several ways. That includes considering the accessibility of the full range, including comments, letters, corrections, retractions, withdrawn publications, and research misconduct findings. We are looking at several options for changes and functionality. We will be publishing our analyses, and making the data openly available so others can explore post-publication activity on PubMed as well.

RW: For some publishers, there seems to still be some delay between when a paper is retracted and when it appears that way on PubMed. Have you seen progress in that area?

HB: We are looking at the present processes to see if there are any opportunities for reducing time lags. A new PubMed Data Management System will be introduced later this year and we are looking carefully at post-publication activity across the board within that transition, which aims to make publisher submission of data easier and quicker.

For more on what various institutions are doing to highlight  and improve retractions, follow us on Twitter, where we’ll be posting today and tomorrow from “Keeping The Pool Clean,” a conference at the Colorado State University in Fort Collins.

Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our new daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.

12 thoughts on “How to better flag retractions? Here’s what PubMed is trying”

  1. It also needs to be done at the level of text entries. Many people do pubmed searches directly from reference managers. Endnote, as you might imagine, will not be aware of a large salmon banner on a webpage. The word “RETRACTED” needs to be added directly into the title of retracted papers.

    1. I believe that this is already the case. Retracted publications are labelled as such using the Entrez PTYP (Publication Type) field. This is present in XML and Medline format and so should be available to reference managers. The question then is how the reference manager chooses to use or display that information.

    2. To extend Miguel’s suggestion, it is surprising that NLM can’t link comment to Genbank (or like data base) sequences that are related to research found to be falsified. Sequence data that is dependent on potentially falsified research is the purest example of an “information virus,” something that overnight can easily enter/contaminate the results of unsuspecting papers that cite it (by comparing it to their results); Yet a ‘hold’ can be placed innocuously on further use of questioned sequence data until the matter is settled, and it can be done so without loosing priority!

    3. Agree, although a nice addition, it does nothing to the xml, csv and text export. Nor is the red mark visible when the article shows up as a search result. Additionally, it would be good to expand this practice to duplicate publications, especially since only a few duplicates are actually retracted.

  2. I assume that Expressions of Concern (EoCs) will also be similarly flagged? It seems to me that, in some instances at least, EoCs ought to be as noticeable as retractions or major corrections.

  3. “RW: Links to relevant Federal Register notices about Office of Research Integrity findings of fraud and misconduct were also recently introduced.”

    Good plug by RW for an important feature of PubMed Searches (Thanks RW!); but in fact this feature has been around for some time: (see ORI Newsletter Vol 19, (4), Sept 2011, pp 4-8, “Further Correcting the Literature: PubMed “Comments” Link Publications to PHS Research Misconduct Findings” ( (PubMed Result Comment Links to misconduct findings had been occurred before that time, but somehow fell into disuse until ORI pressed the issue, and NIH followed suit.) What is gratifying to discover is that at sometime between 2013 and 2014 NLM became more progressive by including the Comment Link for all PHS findings that have mentioned a publication, irrespective of whether or not retraction was mentioned in the PHS finding or was recommended by the institution. (I found this to be true for all of the 38 papers involved in 11 findings I surveyed between May 16, 2016 and July 10, 2014). What is unfortunate -as far as I know- is that there is no such mechanism for transparency attached to publications involved in misconduct findings for research funded by the NSF, DOE, DOD, etc.

  4. Nice but unfortunately in most cases despite clear evidence of manipulation papers are not retracted,

    1. It is easy to be dismissive, and at one level you are right, but would you rather not have that service? In fact, the PubMed Comment link is one way of shaming journals and coauthors to retract, since with it the community has access to the results of the fact finding process relevant to the underlying science. (True, the weakness in my thinking is that either has to be receptive to that influence.) The alert is there through the results of the PubMed search process, and all one needs is to know how to see it, since the Comment Line is not prominently displayed. But I would assume that scholars know how to look.

  5. Check that the banner works on the mobile version of the site. When I clicked on the example (“Check one out here”) when reading on my mobile phone, there was no banner.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.