Login to Access Video or Poster Abstract: MP10-03
Sources of Funding: none

Introduction

INTRODUCTION: Radical cystectomy is an extensive operation with complications reported in up to 30.5% of patients. High complication rates contribute to increased costs, patient morbidity and mortality. Accurate prospective predictions of patients’ risk for post?surgical complications have the potential to identify at risk patients. Risk estimators have been developed but often involve an extensive number of factors or produce expansive results that are not clinically useful. _x000D_ OBJECTIVE: Clinically available preoperative risk factors were identified as potential predictors of postoperative complications, at 30 and 90 days, in patients who underwent radical cystectomy for bladder cancer. We developed a postoperative complication risk prediction model using minimal factors obtained in the normal course of preoperative history, physical and staging. not clinically useful.

Methods

METHODS: 330 patients who underwent radical cystectomy for bladder cancer from January 2008 to July 2014 were included in this study. Potential preoperative risk predictors were collected from medical history, TURBT pathology, preoperative labs, proposed procedure type, and prior treatments. Postoperative complications were graded using the Clavien?Dindo scale. Multivariate logistic regression models were used to predict post?operative complications. Accuracy of prediction models was assessed using the area under the receiver operating characteristic curve.

Results

RESULTS: Of the potential preoperative risk factors, 5, 10 and 16 unique predictors along with two way interactions were determined to have strong association with 90 day postoperative complications, yielding an AUC of 0.69, 0.79 and 0.91 respectively. This is illustrated in Figure 1.

Conclusions

CONCLUSIONS: Our findings suggest routinely collected preoperative patient?level clinical variables may be useful for determining patient risk for short?term postoperative complications. The flexibility in our prediction model for the number of predictor inputs allow users to tailor the degree of risk assessment based on a patient’s baseline heath status. A simple and accessible prediction model with selective predictors may help identify at risk patients for patient education, counseling and development of risk reduction strategies. _x000D_

Funding

none