Multi-institutional validation of vasectomy reversal predictor

J Urol. 2006 Jan;175(1):247-9. doi: 10.1016/S0022-5347(05)00027-3.

Abstract

Purpose: Some urologists who perform vasectomy reversals are not experienced with performing VE. A model to preoperatively identify patients who may require referral to an experienced VE surgeon was created (). We tested the model at multiple institutions.

Materials and methods: The model had previously been designed in 483 patients who underwent vasectomy reversal at 1 institution (100% sensitive and 59% specific for predicting the need for VE). It was based on time since vasectomy and patient age. We tested it prospectively in 33 patients and retrospectively in a total of 312 at 6 other institutions. The predictive accuracy of the model was compared to using a simple duration from vasectomy cutoff alone, as is used in clinical practice.

Results: The model had 84% sensitivity and 58% specificity for detecting the need for VE in a total of 345 patients at 7 institutions. If using only a duration from vasectomy cutoff of 10 years to predict the need for VE, sensitivity was only 69%. At a cutoff of 4 years sensitivity was 99% but specificity was only 23%. Thus, the model performed better than any specific duration cutoff alone.

Conclusions: The predictive model provides 84% sensitivity for detecting patients who may require VE during vasectomy reversal across 7 institutions (58% specificity). The model more accurately predicts the need for VE than using a specific duration from vasectomy cutoff alone.

Publication types

  • Multicenter Study
  • Validation Study

MeSH terms

  • Adult
  • Humans
  • Male
  • Middle Aged
  • Neural Networks, Computer*
  • Prospective Studies
  • Reproducibility of Results
  • Retrospective Studies
  • Sensitivity and Specificity
  • Vasovasostomy / statistics & numerical data*