REPORTING OF RANDOMIZED CLINICAL-TRIAL DESCRIPTORS AND USE OF STRUCTURED ABSTRACTS

Context.-Structured abstracts, that is, abstracts that describe a stud y using requisite content headings, provide more informative content. Concomitant reporting in the text of the report might improve with str uctured abstract use because of increased awareness by authors or edit ors of important study areas associated with content headings. Objecti ve.-To assess whether structured abstract use is associated with impro ved reporting of randomized clinical trials. Design and Setting.-Surve y of trial reports published the year preceding, of, and following new use of structured abstracts, found by hand searching Archives of Opht halmology (1992-1994) and Ophthalmology (1991-1993), as well as trial reports published concurrently without change in abstract format (Amer ican Journal of Ophthalmology, 1991-1994). Main Outcome Measures.-We m easured the inclusion of 56 criteria derived from Consolidated Standar ds of Reporting Trials (CONSORT) descriptors (JAMA 1996;276:637-639) i n the text of each report and calculated the number of criteria includ ed per report and the proportion of reports including individual crite ria. Reports with structured abstracts were compared with those withou t, and reports published in 1993 and 1994 in the American Journal of O phthalmology were compared with those published in 1991 and 1992. Resu lts.-The mean (SEM) number of criteria included by authors was 15.8 (0 .4) per report in 125 trial reports, We found no difference in the mea n number of criteria included or the proportion of reports that includ ed specific criteria by journal. Following structured abstract use, th ere was no difference in either the mean number of criteria per report or the proportion of reports including a majority of criteria within each CONSORT subheading. Four criteria were included more often and 2 less often following structured abstract use in individual journals. C onclusion.-Using CONSORT descriptor criteria to evaluate reporting qua lity, we found no difference in text reporting associated with structu red abstract use in the journals examined.