Surgeons' experience and interaction effect in randomized controlled trials regarding new surgical procedures

The most reliable information on any type of medical intervention is provided through the results of randomized controlled trials (RCTs). However, RCTs comparing a new surgical technique with an older one may provide inaccurate conclusions when surgeons participating in the study have disparate experience with the new technique. In this case, the surgeon-to-surgeon variability may confound the outcome and lead to a significant interaction effect between surgeon and surgical technique. Subsequently the RCT design does not ensure the same outcome probability among patients assigned to the group undergoing the new technique. Overlooking the interaction effect may be responsible for inaccurate conclusions, which are usually unfavorable with regard to the new technique. We discuss how this interaction effect could be involved in conclusions provided by several RCTs that compared laparoscopic hysterectomy to vaginal hysterectomy. We demonstrate how this interaction may be revealed using a hypothetical RCT whose data was reasonably presumed on the basis of literature data.

Key words: confounding variables, interaction effect, laparoscopic hysterectomy, randomized controlled trial, vaginal hysterectomy