Randomized controlled trials in surgery.

Abstract

BACKGROUND: The objective was to determine the number of randomized controlled trials (RCT) performed by surgeons, published in surgical journals, or comprising a surgical arm and to assess their characteristics and overall quality. METHODS: RCT in general surgery (including gastrointestinal, breast, surgical oncology, vascular, critical care, and trauma) published in 1990 were retrieved by MEDLINE and analyzed to determine the funding agency, type of therapy, area of surgery, journal published, country of origin, number of centers, and whether a surgeon was the principal author. The completeness of the MEDLINE search was compared to a manual search of the literature. All RCT were assessed with Chalmers' qualitative score. RESULTS: MEDLINE retrieved 202 surgical RCT (46% of those retrieved by a manual search) with a mean score of 0.40 +/- 0.13. However, surgical RCT were performed by surgeons in only one third of trials, compared surgical therapies in only one quarter of trials, and were published in surgical journals in less than one third of trials. Only 22% of surgical RCT were funded by peer reviewed granting agencies. The strongest variables determining the quality of surgical RCT were the type of therapy tested (p = 0.0001), the type of journal published (p = 0.006), and the area of general surgery (p = 0.007). CONCLUSIONS: Although surgical RCT are being performed, there are a relatively low proportion and standard of RCT performed by surgeons as the principal author, published in surgical journals, and comparing surgical therapies. This may reflect a lack of expertise by surgeons in clinical trials, lack of funding for surgical trials, methodologic problems peculiar to surgical trials, or a need for adoption of other research designs to assess surgical therapies.