Development and External Validation of a Highly Accurate Nomogram for the Prediction of Perioperative Mortality After Transurethral Resection of the Prostate for Benign Prostatic Hyperplasia

Abstract

Benign prostatic hyperplasia affects 60% of men at the age of 60 years. Transurethral resection of the prostate is the gold standard of therapy. We assessed the 30-day mortality rate after transurethral resection of the prostate for benign prostatic hyperplasia, identified risk factors related to 30-day mortality and developed a model that discriminates among individual 30-day mortality risk levels. We performed development (7,362) and external validation (7,362) of a multivariable logistic regression model predicting the individual probability of 30-day mortality after transurethral resection of the prostate based on an administrative data set (Quebec Health Plan) of 14,724 patients 43 to 99 years old treated between January 1, 1989 and December 31, 2000. Overall 30-day mortality occurred in 58 patients (0.4%) undergoing transurethral resection of the prostate. On univariable analyses increasing age (p <0.001) and increasing Charlson comorbidity index (p <0.001) were statistically significant predictors of 30-day mortality after transurethral resection of the prostate. Conversely annual surgical volume was not. On multivariable analyses age (p <0.001) and Charlson comorbidity index (p <0.001) reached independent predictor status. The accuracy of the age and Charlson comorbidity index based nomogram that predicts the individual probability of 30-day mortality after transurethral resection of the prostate was 83% in the external validation cohort. Age and Charlson comorbidity index are important determinants of 30-day mortality after transurethral resection of the prostate. The combination of these parameters allows an 83% accurate prediction of individual 30-day mortality risk after transurethral resection of the prostate. Despite limitations such as the need for additional external validations and possibly the need for inclusion of clinical parameters, the use of the current model is warranted for the purpose of informed consent before transurethral resection of the prostate and/or for patient counseling.