OBJECTIVES: To examine the perioperative mortality rates at 90 days (90 dM) after nephroureterectomy (NU) and to devise a model capable of identifying individuals at an elevated 90 dM risk. NU represents the surgical standard of care for patients with invasive, nonmetastatic upper-tract urothelial carcinoma. However, this major abdominal surgery may be associated with a nonnegligible rate of perioperative mortality. METHODS: We identified 6078 upper-tract urothelial carcinoma patients treated with NU from 17 registries of the Surveillance, Epidemiology, and End Results database, between 1988 and 2006. Stratified analyses quantified 90 dM rates according to age, gender, race, year of diagnosis, tumor location, surgery type, T stage, tumor grade, and lymph node status. Subsequently, multivariable logistic regression models identified predictors of 90 dM within the development cohort (n = 3039). The accuracy and calibration of the model were tested in an independent validation cohort (n = 3039). RESULTS: The overall 90 dM rate was 4.4%. Continuously coded age and T and N stages achieved an independent predictor status in multivariable logistic regression models and represented key variables for prediction of individual 90 dM risk after NU, with 73.4% accuracy. Excellent correlation between predicted and observed 90 dM rates after NU was recorded. CONCLUSIONS: In this large-scale population-based analysis of perioperative mortality after NU, age and T and N stages emerged as the most informative predictor of 90 dM. We recommend the use of this tool in individual decision-making and in informed consent considerations.
OBJECTIVES: To examine the perioperative mortality rates at 90 days (90 dM) after nephroureterectomy (NU) and to devise a model capable of identifying individuals at an elevated 90 dM risk. NU represents the surgical standard of care for patients with invasive, nonmetastatic upper-tract urothelial carcinoma. However, this major abdominal surgery may be associated with a nonnegligible rate of perioperative mortality. METHODS: We identified 6078 upper-tract urothelial carcinoma patients treated with NU from 17 registries of the Surveillance, Epidemiology, and End Results database, between 1988 and 2006. Stratified analyses quantified 90 dM rates according to age, gender, race, year of diagnosis, tumor location, surgery type, T stage, tumor grade, and lymph node status. Subsequently, multivariable logistic regression models identified predictors of 90 dM within the development cohort (n = 3039). The accuracy and calibration of the model were tested in an independent validation cohort (n = 3039). RESULTS: The overall 90 dM rate was 4.4%. Continuously coded age and T and N stages achieved an independent predictor status in multivariable logistic regression models and represented key variables for prediction of individual 90 dM risk after NU, with 73.4% accuracy. Excellent correlation between predicted and observed 90 dM rates after NU was recorded. CONCLUSIONS: In this large-scale population-based analysis of perioperative mortality after NU, age and T and N stages emerged as the most informative predictor of 90 dM. We recommend the use of this tool in individual decision-making and in informed consent considerations.