Objective: To assess and compare the quality of nonstructured and structured abstracts of original research articles in three medical journals.
Design: Blind, criterion-based observational study.
Sample: Random sample of 300 abstracts (25 abstracts per journal each year) of articles published in the British Medical Journal (BMJ), the Canadian Medical Association Journal and the Journal of the American Medical Association (JAMA) in 1988 and 1989 (nonstructured abstracts) and in 1991 and 1992 (structured abstracts).
Main outcome measures: The quality of abstracts was measured against 33 objective criteria, which were divided into eight categories (purpose, research design, setting, subjects, intervention, measurement of variables, results and conclusions). The quality score was determined by dividing the number of criteria present by the number applicable; the score varied from 0 to 1.
Results: The overall mean quality scores for nonstructured and structured abstracts were 0.57 and 0.74 respectively (p < 0.001). The frequency in meeting the specific criteria was generally higher for the structured abstracts than for the nonstructured ones. The mean quality score was higher for nonstructured abstracts in JAMA than for those in BMJ (0.60 v. 0.54, p < 0.05). The scores for structured abstracts did not differ significantly between the three journals.
Conclusions: The findings support recommendations that promote the use of structured abstracts. Further studies should be performed to assess the effect of time on the quality of abstracts and the extent to which abstracts reflect the content of the articles.