Development Of A Mathematical Model For Predicting Stone-free Rate After Minimally Invasive Percutaneous Nephrolithotomy

Objective Our study was performed to investigate the prognostic factors related to the success rate of minimally invasive percutaneous nephrolithotomy lithotripsy (MPCNL) and develop a preoperative logistic regression model for predicting the stone-free rate after MPCNL.Methods We retrospectively analyzed the records of 865 patients with upper urinary tract calculi who underwent MPCNL in our department between January 2006 and September 2009. All cases were confirmed by KUB / IVU, ultrasound or CT scan. Stone-free rate was adopted to evaluate the treatment outcome. Stone-free was defined as no residual stones detected on KUB / IVU, ultrasound or CT scan after initial therapy. The statistical analysis was carried out by the SPSS-15 package. The Student t test, Mann-Whitney U test and chi-square test was performed to analyze the relation between preoperative factors and stone-free rate. Finally, binary logistic regression analysis(Forward:LR) was used to determine statistical significant variables and to create predictive mathematical model.Results The stone-free rate after primary MPCNL was 80.1% (693/865). We found that number of stones, stone location, stone size and the degree of hydronephrosis were significant factors which could affect the outcome of MPCNL. Then a logistic regression model was developed using these variables to estimate the stone-free rate after MPCNL. The overall accuracy of the model was 82.0%, the sensitivity 95.5% and the specificity 27.3%. Conclusions Our study demonstrated that number of stones, stone location, stone size and the degree of hydronephrosis were significant prognostic factors of MPCNL outcome. We have developed a logistic regression model with an overall accuracy of 82.0% for predicting the stone-free rate after MPCNL, which is useful for patient counseling and doctor decision and ensuring an optimal combination of maximum efficacy and minimum cost.