Part II | Chapter 8 162 categorical and continuous variables, respectively. Linearity of continuous variables was checked with Martingale residuals. Influential outliers were assessed by calculating dfbeta residuals. Model performance and internal validation The discriminative ability of the model was assessed using Harrell’s C-statistic. Internal validation was performed through bootstrapping with 2000 resamples for each imputation set, in which all modelling steps were repeated. The optimism of each model and shrinkage factors were calculated, and the β-coefficients and C-statistic were adjusted accordingly. The predictive accuracy of the optimism-corrected models was visualised with calibration plots at 12, 24, and 36 months. Nomogramand risk group construction For both models a nomogram and web tool were constructed using the optimism-corrected coefficients. Finally, for each model separately, three risk groups were identified based on the 25th and 75th percentile of the linear predictor. The Kaplan-Meier method was used to display the biochemical disease-free survival curves for each risk group. All statistical analyses were performed using R studio (version 3.6.1, R Foundation for Statistical Computing, Vienna, Austria, https://rstudio.com) and the survival, survminer, rms, pmsampsize, ggplot2, mice, psfmi, DynNom, and regplot packages30. Reporting was according to the TRIPOD statement.29 Results Baseline characteristics and Kaplan-Meier survival analysis Baseline characteristics are displayed in Table 1. Median (IQR) follow-up time was 25.1 months (13.538.1) for all patients and 18.1 months (9.2-29.6) for patients who did not experience biochemical failure (censored). Sixty-one patients (40.7%) experienced biochemical failure after a median (IQR) of 32.9 months (23.5-43.6). Median bDFS was 29.7 months (95% CI: 25.0-38.6) (Supplementary C). Characteristic % Missing Primary treatment Primary treatment, n (%) - EBRT - LDR brachytherapy - HDR brachytherapy 80 (53.3) 67 (44.7) 3 (2) 0 EBRT dose (Gy), median (IQR) 76.0 (71.5-77.0) 12.5 LDR dose (Gy), median (IQR) 145.0 (145-145) 0 HDR dose (Gy), median (IQR) 19.0 (19.0-38.0) 0 PLND at primary treatment, n (%) 30 (20.0) 0 Table 1 – Baseline patient-, tumour-, and treatment-related characteristics. Legend: IQR = Interquartile range. EBRT = External beam radiation therapy. LDR = Low-dose-rate. HDR = High-dose-rate. PLND = Pelvic lymph node dissection.
RkJQdWJsaXNoZXIy MTk4NDMw