BPH nomogram using IPSS, prostate volume, peak flow rate, PSA and median lobe protrusion for predicting the need for intervention: development and internal validation.

ABSTRACT

OBJECTIVE: The aim of the present study was to develop a nomogram to accurately predict the need for intervention in patients suffering from LUTS due to benign prostatic hyperplasia (BPH) and internally validate it. MATERIAL AND METHODS: The data was collected from the community subjects from the state of Gujarat in western India. All the demographic data, physical examination, PSA, uroflowmetry and prostatic ultrasound was collected in 92 subjects and were followed up after 2 years. The data was analyzed and logistic regression model was used to build a predictive model. A nomogram was build using R software. Nomogram was internally validated using 50 subjects. RESULTS: 92 subjects were analyzed for developing the nomogram. Out of these, 17 patients needed intervention. 8 patients were started on medical therapy and 9 patients were taken up for surgical intervention. Of all the statistically significant predictors, peak flow rate was the most significant and was followed by median lobe enlargement, PSA, prostate volume and IPSS. These variables were used to develop a prediction model for the intervention required using reduced logistic regression model. The predictive accuracy of the model was 95.65% with a sensitivity of 88.28%, a specificity of 97.33%, a positive predictive value (PPV) of 88.24%, and a negative predictive value (NPV) of 97.33%. The AUC of the model was 0.799. Internal validation was done on 50 subjects which had sensitivity, specificity and AUC of the model at 89.66%, 90.48% and 0.968 respectively. CONCLUSION: The study demonstrates the clinical application of nomogram which uses IPSS, PSA, peak flow rate, prostate volume and median lobe enlargement (intravesical prostatic volume). It has a sensitivity of 88.24%, specificity of 97.33%. It predicts the need for intervention in BPH patients with accuracy of 95.65% which was internally validated with an accuracy of 90%.