SHOULD WE BE PERFORMING MORE RANDOMIZED CONTROLLED TRIALS EVALUATING SURGICAL OPERATIONS

Background. The objective of this study was, first, to determine what proportion of clinical treatment evaluation questions involving surgic al operations could be answered by a randomized controlled trial (RCT) . Second, for those questions not amenable to a RCT, to determine the problems that potentially preclude the initiation of RCT in an ideal c linical research setting. Methods. A sample of treatment evaluation qu estions involving a surgical procedure was obtained by a computerized search of the surgical literature. Problems precluding a RCT were defi ned. Their face validity and interobserver and intraobserver reliabili ty were assessed. By use of these criteria, the sample questions were evaluated to determine whether a RCT could be performed and, if not, t he predominant reasons precluding RCT of surgical procedures. Results. Only 38.8% of treatment evaluation questions could have been answered by a RCT in an ideal clinical research setting. Patient preference wa s the most common problem encountered (40% of all problems). The princ ipal precluding problem was patient preference in 23.1%, an uncommon c ondition in 24.2%, and lack of community (clinical) equipoise in 10%. Methodologic ic issues (1.2%) and surgical preference (2.3%) were infr equent precluding problems. Questions evaluating therapy for malignant disease, comparing surgical with nonsurgical therapies, and where sur vival was the primary outcome were more likely to have problems preclu ding RCT. Conclusions. In the ideal situation RCT can be performed to evaluate only 40% of treatment questions involving surgical procedures . Patient preferences, uncommon conditions, and lack of surgical commu nity equipoise appear to be the most common reasons precluding the per formance of RCT of surgical operations.