I have been informed today by our production editor at the publisher that references appearing in the abstract must be cited in full, not just as a superscript number, as "Abstracts are treated as separate entities as sometimes they can be viewed on their own and the reference numbers will mean nothing to the reader."

I am wondering if this is common practice with other journals and whether others agree it should be done this way.

Toni McCallum Pardey
Interim Editor-in-Chief, AENJ
_______________________________
In our journal, Birth, I get around that issue by deleting reference numbers from the abstract, and making sure they are cited in the introduction of the text.

Diony Young
Editor, Birth
_______________________________
To the best of my knowledge, it is not a common practice to reference in abstract section of an article; at least this is what I've read in books and teach in my workshops on scientific writing. However, if for any reason we accept that we can reference in the abstract, I think your production editor is right, and I believe the references should be cited in such a way that we can retrieve them easily.

Farrokh Habibzadeh
Archives of Iranian Medicine
_______________________________
Your production editor is correct—abstracts must be able to "stand alone". Unfortunately, they are frequently the only part of a manuscript that is read, particularly by those who search articles online and may not have access to the full text. The same premise should be applied to abbreviations, which should be avoided in abstracts, or expanded the first time they are used, as you would normally do in the main text.

It has always been the policy of the MJA to exclude references from the abstract, and I cannot recall ever seeing references in any other journal's abstracts. I have an old copy of Medical style and formatting by Ed Huth (former editor of the Annals of Internal Medicine) in which he states "the text of the abstract must not include citations of references cited in the text of the article or references themselves aside from the reference for the article itself added to the [end of the] abstract", and a not-so-old copy of the American Medical Association's Manual of Style, which simply says, under general guidelines for abstracts, "Do not cite references".

Bronwyn Gaut
Deputy Editor, The Medical Journal of Australia
_______________________________
Why is it so bad to include a reference in an abstract?

I agree that abstracts must be able to stand alone, but I don't see that as the same question as whether they can include references. Certainly this practice is rare, but it is probably not often thought useful by authors and of course most journals don't allow this practice anyway. But is it necessarily a bad thing?

Such citations seem to be used sometimes in methodological papers. Here is a very recent example:

Am J Epidemiol. 2007 Mar 30; [Epub ahead of print]
Estimating Population Size with Two- and Three-Stage Sampling Designs.
Tate JE, Hudgens MG.

Reliable estimates of population size are important for developing and monitoring health programs in at-risk populations. Laska, Meisner, and Siegel (Biometrics 1988;44:461-72) developed an unbiased estimator for the size of a population at a single venue based on a single sample. Because many populations of interest are not contained within a single venue, this article generalizes the Laska, Meisner, and Siegel estimator to incorporate two- and three-stage sampling designs and enable estimation of total population size over multiple venues. Use of the estimator with two- and three-stage sampling designs is illustrated with examples that estimate the size of a population of individuals who socialize over a 4-week period at public venues where transmission of human immunodeficiency virus and other sexually transmitted infections is likely to occur.

Is any harm done by including the citation here? It seems to me to provide useful information. Without obtaining this new article, to which I might not have access, I am able to identify the motivating source publication and indeed conduct a citation search for other related material.

As some people say they have never seen references in abstracts so here are 4 other examples:
PMID: 16365481
PMID: 16732467
PMID: 16279273
PMID: 15231922

We can have a more informed discussion with such examples available. Why not cite a reference if it is helpful to do so? (But I don't recall that I ever did so.) Common sense and good judgment should prevail, not blanket rules.

Doug Altman
_______________________________
I entirely agree with Doug.

Clement O. Adewunmi
Editor-in-Chief, African Journal of Traditional, Complementary and Alternative Medicines
_______________________________
I, for one, do not have any problem with citing in the abstract. But many indexing systems such as PubMed put a 250-word limitation on the size of abstracts. Referencing certainly needs use of extra words so that we may cross this limitation. A solution for this problem may be to only mention the PMID or DOI of the cited article.

Farrokh Habibzadeh
Archives of Iranian Medicine 
_______________________________
Although it is widely believed that PubMed puts a 250 word limit on abstracts this is not correct. Since 2000 the limit has been 10,000 characters, which is clearly far in excess of 1000 words. This information is available at http://www.nlm.nih.gov/bsd/mms/medlineelements.html

We know from a review we carried out last year that some journals' Instructions for Authors include an incorrect statement about PubMed restrictions on word length. 

Doug Altman
_______________________________
Probably not a bad thing, but I suspect the reason people don't use them often is that references contribute to the word count, and many journals are more scrupulous about the abstract word limit than about that of the text. Since superscripts contravene the injunction that abstracts should stand alone, there is no alternative but to add words (only 6 in the case Dr. Altman cites, but they can add up). Omission of authors' names, and some rewording, would require the addition of only two words (eg, Biometrics 1988;44:461-72), however, and it makes sense to use judgment rather than rules.

Rich Rothenberg
_______________________________
I have seen references frequently used in abstracts that (a) comment or respond to another specific piece or (b) are a direct test or refutation of a particular piece of work or theory.

Nancy Darling
_______________________________
Dear Dr. Altman,

Thank you for the information on the new word limit (I really did not know that). But, if this is not the problem with indexing systems, why do many journals still ask authors to restrict the word count in abstract to 250 (in some journals 150)?

Farrokh Habibzadeh
_______________________________
“why do many journals still ask authors to restrict the word count in abstract to 250 (in some journals 150)?”

…for the same reason why a movie trailer is only 2-3 minutes long.

And besides: If an author is not able to tell what and why he/she has done in the study, most probably it is not a good study.

Mark R Graczynski
Medical Science Editor