Prediction of Incontinence after Robot-Assisted Radical Prostatectomy: Development and Validation of a 24-Month Incontinence Nomogram.

Abstract

Simple SummaryMany men fear urinary leakage after radical surgery for prostate cancer and may even choose against operation for unrealistic fears of leakage. Many urologists are unaware of their own results, and some urologists who collect their results do so in different ways. We collected urinary leakage data from 680 men in a uniform and simple way at 6, 12, and 24 months after operation: no pads, 1–2 pads, or ≥3 pads required daily. We used many patient characteristics to identify the key factors that predict recovery of urinary control after operation: age, race, height and weight, and preoperative erectile function. Easy-to-use nomograms were constructed that should be tested by other urologists to make sure they perform equally well in their patients. Nomograms like these allow men and the urologists counseling them to share patient-specific information about the timeline for, and the chance of, recovery of urinary control after operation.Incontinence after robot-assisted radical prostatectomy (RARP) is feared by most patients with prostate cancer. Many risk factors for incontinence after RARP are known, but a paucity of data integrates them. Prospectively acquired data from 680 men who underwent RARP January 2008–December 2015 and met inclusion/exclusion criteria were queried retrospectively and then divided into model development (80%) and validation (20%) cohorts. The UCLA-PCI-Short Form-v2 Urinary Function questionnaire was used to categorize perfect continence (0 pads), social continence (1–2 pads), or incontinence (≥3 pads). The observed incontinence rates were 26% at 6 months, 7% at 12 months, and 3% at 24 months. Logistic regression was used for model development, with variables identified using a backward selection process. Variables found predictive included age, race, body mass index, and preoperative erectile function. Internal validation and calibration were performed using standard bootstrap methodology. Calibration plots and receiver operating curves were used to evaluate model performance. The initial model had 6-, 12-, and 24-month areas under the curves (AUCs) of 0.64, 0.66, and 0.80, respectively. The recalibrated model had 6-, 12-, and 24-month AUCs of 0.52, 0.52, and 0.76, respectively. The final model was superior to any single clinical variable for predicting the risk of incontinence after RARP.