Data collection and statistical analysis
Clinical, surgical, histopathological and follow-up data were collected from medical records. Baseline characteristics were compared between the different learning phases. The institutional review board approved this study and waived individual consent requirements for the use of these pseudonymised retrospective data. Participants were not involved in the development of this analysis and no core outcome sets were used in this study.
Primary outcome of interest was surgical proficiency when introducing a novice, based on cervical recurrence rate. To evaluate surgical proficiency (i.e. learning curve) in this new phase, a risk-adjusted cumulative sum analysis (RA-CUSUM) was performed (see previous publication for a detailed description of this analysis8). The probability of recurrence for each individual patient was modelled by a multivariate logistic regression analysis. To prevent overfitting, we limited this model to three degrees of freedom. As previously described, we used two RA-CUSUM functions: the RA-CUSUM+ chart, designed to detect a doubling of the odds of recurrence, and the RA-CUSUM- chart, designed to detect a halving of the odds of recurrence. Details of these RA-CUSUM functions are provided in the supplementary material (Appendix S1).
Secondary outcomes were disease-free survival (DFS), disease-specific survival (DSS) and overall survival (OS) in this new phase compared to the previous phases. Survival was defined as the time interval between date of diagnosis or first visit and date of disease recurrence diagnosis (DFS), diagnosed clinically, by imaging, or histopathological biopsy, or death due to the sequelae of cervical cancer (DSS) or death from any cause (OS). Survival curves were estimated using Kaplan-Meier method and the newly added group (group 3) was compared with the previous groups (group 1 and group 2) using log-rank test: group 1 versus group 3 and group 2 versus group 3.
In addition to the previous analysis, we also assessed differences in operation time and short-term surgical complications between the three learning phases. Operation time was defined as the time (in minutes) between first incision and end of closure by the surgeon. If the procedure was performed in two tempi the operation time of the uterine procedure was included in the analysis. All intraoperative complications, defined as complications diagnosed intraoperatively and postoperative complications within 42 days were registered in adherence to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0 (published November 27, 2017).27 In the current analysis main focus was on severe surgical complications, i.e. grade 3 to 5 adverse events.
Continuous data were compared between the groups using one-way analysis of variance (ANOVA) or its non-parametric counterpart, as appropriate. In the event of significance, post-hoc tests were conducted to determine differences between each pair of groups. Categorical data were reported as proportions and compared between groups using chi-square test or Fisher’s exact test, as appropriate. Statistical tests were two-sided with significance set at p<0.05, with confidence intervals (CI) at the 95% level.
Analyses were performed using the Statistical Package for the Social Sciences version 27.0.0 (SPSS; International Business Machines, Armonk, NY, USA) and Microsoft Excel 2016 (Microsoft, Redmond, WA, USA).

Results