PURPOSE: We devised a model to predict, preoperatively, the need for a vasoepididymostomy (VE) when performing a vasectomy reversal. Urologists could use it to identify those patients who need a referral to an experienced VE surgeon. MATERIALS AND METHODS: We performed a retrospective review of 483 patients who underwent vasectomy reversal by a single surgeon (AJT) including 393 vasovasostomies and 90 vasoepididymostomies. Selection was based on chart availability. Established criteria were used in deciding the type of reversal (eg gross appearance and microscopic examination of vasal fluid). Type of reversal, patient age and time since vasectomy were recorded. Univariate analysis revealed that patient age (p <0.001) and time since vasectomy (p <0.001) were significant predictors of reversal type. On multivariate logistic regression analysis, time since vasectomy (p <0.001) was the only significant independent predictor. We designed a linear regression algorithm based on time since vasectomy and patient age to predict if a VE would be performed. The model was designed using 433 patients and then tested on a separate randomly selected 50 patient group. The model was designed to be 100% sensitive in detecting patients requiring VE. RESULTS: In the test group the model was 100% sensitive in predicting VE with a specificity of 58.8%. The area under the ROC curves for the design and test groups was 0.8. Palm (PalmSource Inc., Sunnyvale, California) and Windows (Microsoft Corporation, Redmond, Washington) versions are available as free shareware from www.uroengineering.com. CONCLUSIONS: The model is 100% sensitivity in detecting those patients who may require a VE during vasectomy reversal (specificity of 58.8%). It may allow urologists to preoperatively identify these patients.
PURPOSE: We devised a model to predict, preoperatively, the need for a vasoepididymostomy (VE) when performing a vasectomy reversal. Urologists could use it to identify those patients who need a referral to an experienced VE surgeon. MATERIALS AND METHODS: We performed a retrospective review of 483 patients who underwent vasectomy reversal by a single surgeon (AJT) including 393 vasovasostomies and 90 vasoepididymostomies. Selection was based on chart availability. Established criteria were used in deciding the type of reversal (eg gross appearance and microscopic examination of vasal fluid). Type of reversal, patient age and time since vasectomy were recorded. Univariate analysis revealed that patient age (p <0.001) and time since vasectomy (p <0.001) were significant predictors of reversal type. On multivariate logistic regression analysis, time since vasectomy (p <0.001) was the only significant independent predictor. We designed a linear regression algorithm based on time since vasectomy and patient age to predict if a VE would be performed. The model was designed using 433 patients and then tested on a separate randomly selected 50 patient group. The model was designed to be 100% sensitive in detecting patients requiring VE. RESULTS: In the test group the model was 100% sensitive in predicting VE with a specificity of 58.8%. The area under the ROC curves for the design and test groups was 0.8. Palm (PalmSource Inc., Sunnyvale, California) and Windows (Microsoft Corporation, Redmond, Washington) versions are available as free shareware from www.uroengineering.com. CONCLUSIONS: The model is 100% sensitivity in detecting those patients who may require a VE during vasectomy reversal (specificity of 58.8%). It may allow urologists to preoperatively identify these patients.