Twitter YouTube LinkedIn

AAPS 2007 Annual Meeting, May 19 - 22, 2007, The Coeur d'Alene Resort, Coeur d'Alene, Idaho.
Back to Scientific Program
Back to Annual Meeting

Twenty Years of Randomized Controlled Trials in Plastic Surgery: An Analysis of Reporting Standards and Trial Impact
Amir H. Taghinia, M.D., Eric C. Liao, M.D., James W. May, Jr., M.D..
Massachusetts General Hospital, Boston, MA, USA.

Purpose: Well-designed randomized controlled trials (RCTs) provide the highest level of evidence in health-care interventions. However, biased results from trials with improper methodology or reporting can mislead medical decision-making and public policy. There is strong evidence to suggest that the quality of reporting of RCTs in the medical literature is suboptimal, and that poor reporting standards reflect poor methodological quality. To remedy this problem, the Consolidated Standards of Reporting Trials (CONSORT) group issued a checklist of items that are to be reported so that readers are clearer about a trial’s design, execution, and analysis. Using these items, we sought to assess the quality of reporting of RCTs in plastic surgery.
Methods: The EMBASE, Medline, and Cochrane Registry were queried for RCTs published from 1986 to 2006 in Plastic and Reconstructive Surgery, Annals of Plastic Surgery, and British Journal of Plastic Surgery. We scored each RCT on a percentage scale based on the CONSORT criteria. We explored the association between the quality of reporting and multiple parameters including: therapeutic category, number of participants, number of times the article had been cited since publication (citation index, a measure of a trial’s overall impact), the country of origin, and the year and journal of publication.
Results: A total of 163 RCTs were evaluated with an average score of 50.4% ± 12.5% (S.D.). Most RCTs were published in Plastic and Reconstructive Surgery, originated from the U.S., and pertained to wound healing. The most under-reported CONSORT items were sample size determination (12% of RCTs), delineation of a primary outcome measure (20%), intention-to-treat analysis (15%, where applicable), and details of randomization including implementation (12%) and allocation concealment (21%). Similar results were seen for reporting of adjustments to statistical analyses for subgroups (15%) and multiple comparisons (13%). In contrast, most reports did provide adequate scientific rationale (98%), clinical objectives (100%), and details of intervention (99%). The overall score was not statistically associated with the number of participants, therapeutic category, the country of origin, or the year or journal of publication. However, the reporting of the following CONSORT criteria was highly predictive of the final score (p<0.001): sample size determination, random sequence generation, allocation concealment, and blinding of outcome assessors. There was no significant linear relationship between citation index and the final score (p>0.05), but trials that were cited more than 20 times scored higher (55.4% vs. 49.5%, p=0.01), thus suggesting a real, albeit weak correlation between these two factors.
Conclusion: In light of ever-increasing reliance on evidence-based practice, the quality of reporting of RCTs in plastic surgery needs improvement. Reporting standards could be improved if the CONSORT criteria are considered when designing and evaluating RCTs, and if better quality control measures for trial reporting and methodology are established. The correlation between the quality of a report and its citation index may suggest that authors are intuitively aware of the methodological quality of a report but that there are other reasons, unrelated to quality, that influence authors to cite a trial.

Back to Scientific Program
Back to Annual Meeting
Quick Links
Annual Meeting
Twitter YouTube LinkedIn