CAN THE ACCURACY OF ABSTRACTS BE IMPROVED BY
SPECIFIC INSTRUCTIONS?
Roy M Pitkin and Mary Ann Branagan
Obstetrics & Gynecology, 1100 Glendon Ave, Suite 1655, Los Angeles, CA 90024-3520, USA
Objective: To test the hypothesis that providing authors with specific instructions for preparing abstracts,
including identifying common errors, will result in improved accuracy.
Design: Randomized controlled trial involving manuscripts reporting original research submitted to a peer-reviewed
medical specialty journal and returned after review, with an invitation to revise. The intervention was an instruction sheet
stressing the importance of an accurate abstract and identifying 3 common types of inaccuracies: (l) data inconsistent in abstract
and body of the paper, tables, and figures; (2) data or information in the abstract that do not appear elsewhere; and (3) conclusions
in the abstract not substantiated in the abstract itself. Revisions were examined specifically for accuracy of the abstract. Preliminary
observations indicated an error rate of about 25%, leading to a projected sample size of 200.
Results: One or more deficiencies were found in 28% (95% CI, 19%-37%) of manuscripts in which the author had been instructed
and in 26% (95% CI, 18%-34%) of the uninstructed group. Of the 55 deficient abstracts, 28 had inconsistencies, 16 contained data not
in text, 8 had both, and 3 included inappropriate conclusions. A spot check of 4 other journals (2 weekly and 2 monthly) found defective
abstracts in 27%-65% of published articles.
Conclusions: Errors and inconsistencies in abstracts are common. Simply providing authors with instructions about abstract preparation
is ineffective, and some other means of improving accuracy needs to be found. In the meantime, journals should pay particular attention to the accuracy of abstracts.
Return to Session Information
