The reporting quality of randomised controlled trials in surgery: a systematic review.

OBJECTIVES

In order to examine the reporting quality of urological RCTs, the initial objectives of this study were to evaluate the degree to which RCTs involving urological surgical techniques (as the intervention) published in the years 2000-2003 complied with the CONSORT statement, and to assess trends and patterns of compliance. Following our initial findings in urology, we extended the methodology to a number of other specialties to assess whether our findings in urology could be generalised to other surgical disciplines.

METHODS

The Royal Society of Medicine (RSM) Library was commissioned to search the Medline and Cochrane databases for RCTs in compliance with the study inclusion criteria below. Additional analyses of five other specialties (non-urological trials: cardiovascular, gastrointestinal, hepatic, orthopaedic and vascular) were also made. For the non-urological trials, 15 English language trials (from each specialty) were randomly picked from a blast search conducted by the RSM for the year 2003. The RCTs were given a score out of 22, reflecting how many of the 22 CONSORT items were complied with (with each item being given equal weighting), this score was termed the "CONSORT score".

RESULTS

Urology trials: In total 122 RCT abstracts were identified which met the inclusion criteria for this study. From these, 32 were excluded as they were follow-up studies, involved a virtual procedure or were a cost analysis, leaving 90 RCTs published across 35 different journals which were analysed (68.2%, 90/122). The average score of 11.1 for urological trials indicates that RCTs in our sample do not comply with the CONSORT statement. No trials reported how they implemented their randomisation process. Only 46% of RCTs stated that they had permission from an ethics review board, 20% had declared their sources of funding, 14% stated whether there were any conflicts of interest and only 1.1% stated their trial registry number. For non-urological studies, an average CONSORT score of 11.2 was obtained.

CONCLUSION

Clinical research teams conducting RCTs in urology and other surgical disciplines demonstrate poor compliance with the CONSORT statement. We would recommend that trials should be registered at their outset and that urological and other surgical journals to consider supporting the CONSORT statement and to have compliance 'hard-wired' into their submission, editorial and peer-review processes. Since it seems the best researchers are unable to produce an RCT results which enable surgical techniques to be critically assessed, there is a need for education about the CONSORT statement and its importance at all levels of surgical training. We believe that an open debate is needed on the possible role of other research designs, such as tracker studies. Whether this study actually raises the question of how appropriate RCTs are to surgical techniques, we leave to the reader.