I recently contacted Medline to draw their attention to an incorrect authorship listing for a set of guidelines published in the European Journal of Pain (which I had worked on). I was very surprised to receive the following response:

"Our policy requires a corporate author for guidelines, since guidelines are issued by a corporate body. From reading this guideline, the best match for the corporate name appears to be "European Federation of Chapters of the International Association for the Study of Pain." If there is a better attribution for the responsible corporate body, please have an erratum published identifying the body that is responsible for the guideline."

It seems that Medline has simply decided that there MUST be a corporate author for these guidelines and therefore included the one they thought fitted best, although this federation had nothing to do with the guidelines. (Their choice seems to be based on the fact that the guidelines had been presented as an abstract at an EFIC scientific meeting. In fact, the guidelines were prepared by a group of independent experts with some funding from a commercial company—this support was clearly acknowledged in the publication.)

I am particularly concerned by this extraordinary action by Medline, since their citations are held in high esteem, and I know several journals recommend authors should check the accuracy of their references against the Medline entry which would, in this case, result in incorrect authorship attribution.

Have any journal editors or authors encountered this problem before? Can we, at WAME, request that Medline (NIH/NLM) rethink this strange policy to ensure that the Medline citation accurately reflects the author list as published in the journal?

Liz Wager
Sideview
Princes Risborough, Bucks

___________________________

The same guideline's author list is indexed in Medline differently:
Moher D, Schulz KF, Altman DG; CONSORT. (BMC Med Res Methodol).
Moher D, Schulz KF, Altman DG. (Lancet).
Moher D, Schulz KF, Altman D; CONSORT Group (Consolidated Standards of Reporting Trials) (JAMA).

And will please anybody tell me how can I search for "American Diabetes Association Clinical Practice Recommendations" in Medline? Is American Diabetes Association an author? Part of the title word? Or is "Expert Committee on the Diagnosis and Classification of Diabetes Mellitus" the author as for one in the series? See it for yourselves and don't get surprised if you find piece of news about ADA such as this one: http://www.nursingcenter.com/pdf.asp indexed in Medline as if "authored" by American Diabetes Association!
Kennedy MS; American Diabetes Association. Good news and bad news from San Diego. Am J Nurs. 2005 Aug;105(8):29.

Arash Etemadi
Scientific Writing and Publishing Advisor, Vice-Chancellor for Research, Tehran University of Medical Sciences

___________________________

I agree with everything that has been said by Doug and others—IF there is indeed a Medline policy that stipulates that guideline papers MUST HAVE corporate authorship, then this is complete nonsense.

However, I have trouble believing that such a policy actually exists. In fact, there are examples for papers which are indexed with PT=guideline that do not have a corporate author. Thus, before I believe this and before WAME approaches NLM with this I really want to see such a policy in writing.

What is more likely is that there is a policy which asks Medline indexers to be on the lookout for corporate authors, especially if the article is a guideline, which absolutely makes sense.

Let me also stress something which few people here seem to be aware of. Corporate authors are NOT indexed in the author field in Medline (even though in the default PubMed output "corporate authors" appear behind the individual author names and make them look like being an "author"). Rather, Medline has a special database field which is called CN (Corporate Author), which identifies the corporate or collective authorship of an article. If you want to see what is in the Medline db fields select MEDLINE in the "Display" drop-down box (which by default says "Abstract") when looking at an abstract in Pubmed. Then you will be able to see that corporate authors are in the field CN, whereas personal authors are in the field AU.

The Pubmed help file says that "corporate names display exactly as they appear in the journal." This is further evidence for what I suspect—that there is no such policy at NLM to "invent" corporate authorship.

The CN field is relatively new—citations indexed pre-2000 and some citations indexed in 2000-2001 retain corporate authors at the end of the title field.

To answer the question below: you can search for corporate authors by entering a string like

"American Diabetes Association"[cn]

into the query field of Pubmed.
For pre-2000 articles, it must be something like "American Diabetes Association"[ti] to search in the title field.

Note that "American Diabetes Association"[au] does not lead to any hits, for the reasons described above.

For all this see for example http://www.ncbi.nlm.nih.gov/books/bv.fcgi?rid=helppubmed.box.pubmedhelp.Box_1_Search_Field_D#pubmedhelp.Corporate_Author_CN

It always amazes me how unfamiliar many researchers—even the most senior ones—are with the intricacies of bibliographic databases, even though searching the literature is one of the pillars of good research.

Hint: Sometimes it actually helps to read the help files of pubmed.

Let me close by saying that it is nothing new that there are many inconsistencies within Medline on how are articles indexed, and in particular things like publication type are sometimes quite unreliable (I have seen archaeology papers indexed with PT=historical articles). This makes it even more important for individual researchers to know these things and to be skilful and sophisticated when searching Medline.

In defence of the NLM, one also has to acknowledge that some of these indexing errors could be avoided if the publishers would send the correct metadata to NLM in the first place, so I think publishers and editors share the responsibility for wrongly indexed articles. I wonder how many editors check whether the person at the publisher who compiles the XML files for submission to Pubmed is actually trained to know what a "guideline" is (as opposed to a editorial that talks about a guideline), how many of the technical persons working at publishers who are in charge of XML-tagging have the ability to spot a corporate author, or how many publishers/editors actually check the XML files before they are submitted to Pubmed (I do!!).

In particular for guidelines, I think the editor is the best person to answer the question on whether there was corporate authorship or not—whereas it is a guessing game for the Medline indexer, in particular as there is no standard format across journals on how corporate authorship is indicated, and in particular if the publisher-submitted XML file is already wrongly tagged.

If publishers send garbage to PubMed (in their publisher-submitted bibliographic XML files), they shouldn't be surprised if the Medline-Indexer overlooks such things. At the end of the day, the editor (and author) are actually the most qualified persons to answer these questions on authorship.

So before WAME approaches NLM with this, it might be much more productive for WAME to change what it can change within the editors' community, which is a) clear guidelines on how corporate authors should be indicated in the published article, b) raising the awareness that what is submitted from the publisher to Pubmed (in terms of XML files) is actually an important part of the publishing process. I personally think it should be part of the editors' job description to exercise some oversight over what is submitted to and what ends up in Medline. NLM in principle does make corrections, if approached by the author or editor.

Gunther Eysenbach
Editor, J Med Internet Res (http://www.jmir.org)
Executive Director, WebCite consortium (http://www.webcitation.org)

___________________________

I would like to add a comment to the point raised by Gunther about "changing what it can change within the editors' community" and "to exercise some oversight over what is submitted to and what ends up in Medline" and other databases. Besides corporate authorship, guidelines on how to indicate author affiliation should also be considered by WAME and ICMJE. Databases were created mainly to record data on what is being published in a subject field but now they are also being used for generating institutional or national S&T indicators and policies. MEDLINE input data on author affiliation and countries in the author field (AU) and retrieval of authors by countries or institutions using this field is inadequate since affiliation data is not standardized in journals.

Guidelines on how to input author affiliation (names of institutions and countries) could be very useful for correct identification of author affiliation as well as to improve quality of information. Databases in general do not make corrections and display data exactly as they appear in the journals. So, to reduce inconsistencies and to improve quality in dissemination of information I fully agree that is extremely important that editors agree and comply with guidelines, understand policies of databases and check how entries of their journals appear in database records.

Regina C. Figueiredo Castro
Coordinator, Health Scientific Communication, BIREME/PAHO/WHO

___________________________

I have had a very helpful response from Sheldon Kotzin (Executive Editor of Medline) (showing how useful the WAME list-serv can be).

He has explained that Medline staff "misinterpreted indexing policy in this case" and has agreed to remove both the misleading corporate author and the Practice Guideline publication type from the citation in question.

He also explained that:
"The MeSH definition and NLM Indexing Manual both specify that in order for a Practice Guideline to be appropriate for an article, there must be an organizational body explicitly identified as having developed the guideline. Our Indexing Manual states that "if the practice guidelines appear solely the recommendations of the author(s) and no reference is made to any institutional role in their development, do not use Practice Guideline (Publication Type)."

He has also noted the comments from Doug Altman about inconsistencies in indexing the CONSORT statement and related publications and has agreed to look into this matter further.

While the individual cases now appear to be resolved (thanks to Sheldon's prompt response) listserve members might continue to discuss the issues of group authorship and whether they agree with Medline's criteria that articles must have a corporate author in order to be classified as a Practice Guideline.

In the meantime, thanks to all involved!

Liz Wager
Sideview
Princes Risborough, UK

___________________________