LAUSR

To the Editor: The Laparoscopic Approach to Cervical Cancer (LACC) trial [1] keeps stirring emotions and discussions [2−6]. The LACC trial can be criticized [7] because the surgical proficiency and the adequacy of the laparoscopic radical hysterectomy was not adequately evaluated, because radicality was not appropriately assessed, and because of the poorly defined inclusion of type II and type III surgery. However, our impression is that the main reason for this ongoing debate is because the LACC trial was a randomized controlled trial (RCT) and because many of us do not like or are not ready to accept the results. This discussion highlights the problems of performing and of interpreting the value of an RCT in surgery. It is surprising that only the results of the smaller previous studies and meta-analyses were mentioned, but not the much larger excellent study with very similar results in the same issue of the New England Journal of Medicine [8]. Results are not only similar, but results were robust for laparoscopic or robot-assisted surgery and across histologic types and tumor sizes. Moreover, these results were recently confirmed in 4 other studies [9−12] Randomization is essential to avoid inclusion bias. In addition, randomization is supposed to eliminate the effect of eventual cofactors. Therefore, randomization should be stratified for major factors that affect the outcome and that are not eliminated by randomization, such as the center and the surgeon in the LACC trial. In order to compare 2 techniques of surgery, the surgeon must be equally skilled in both techniques; otherwise, we are evaluating the surgeons’ skills instead of the technique of surgery. Similarly, in cystic ovarian endometriosis surgery, it remains debated whether some results vary with the singer, not the song [13]. It is beyond the scope to discuss in detail the other problems of an RCT, such as blinding, extrapolation of results, and hidden subgroups in a nonhomogeneous population [14] and of the inadequacy for multimorbidity [15]. Other persisting problems for us clinicians, reviewers, and editors are how to estimate the value of a nonperfect RCT, and in surgical trials, how dependent and independent variables are treated. The value of a trial is the design, and the