Abstract reporting in randomized clinical trials of acute lung injury: An audit and assessment of a quality of reporting score*
- Additional Document Info
- View All
OBJECTIVES: To evaluate the quality of reporting among abstracts of randomized controlled trials (RCTs) in acute lung injury and to highlight important trial information for abstract inclusion. DESIGN: Audit of published RCT abstracts. SAMPLE: A total of 56 RCTs, identified in MEDLINE, EMBASE, HEALTHSTAR, CINAHL, and the Cochrane Central Register of Controlled Trials. MEASUREMENTS AND MAIN RESULTS: We used a research focus group and published literature on suggested content for abstracts of original articles to generate a list of 32 recommended items. The focus group reduced this list to a 20-item long form list of highly relevant criteria and a 12-item short form list of essential criteria for inclusion in abstracts of RCTs in acute lung injury. After pilot testing the abstract appraisal form, we evaluated abstracts independently and in duplicate. We scored the quality of reporting of each abstract by dividing the number of criteria fulfilled by the number applicable. Although abstracts described the study objectives and interventions well and the participants, outcomes, and conclusions to an intermediate extent, key deficiencies were noted in reporting the study methods, setting, and results. Mean quality of reporting scores were significantly higher for structured compared with unstructured abstracts using the 32-item, 20-item, and 12-item lists (p = .008, .014, and <.0001, respectively), especially for abstracts published after 1990 (p = .004, .017, and .001, respectively). The 20-item and 12-item lists correlated well with the 32-item list (r = .89 and .62, respectively) and with one another (r = .73). CONCLUSIONS: Key design features and results are frequently under-reported in RCT abstracts, particularly among unstructured abstracts. Checklists may aid authors and editors in prioritizing important criteria for inclusion in RCT abstracts.
has subject area