Statistical analysis
There were 37,165 questionnaires completed by the investigators. Patients records without prostate enlargement (< 30 ml in transabdominal sonography examination) and those with missing data were excluded (N = 12,552). Statistical analysis was performed using the STATISTICA 13.0 PL software (Tibco Software Inc, Palo Albo, USA), StataSE 12.0 (StataCorp LP, TX, U.S.). Statistical significance was set at a p -value below 0.05. All tests were two-tailed. No data imputation was performed. Nominal and ordinal data were expressed as percentages, while interval data were expressed as mean value ± standard deviation. Distribution of variables was evaluated by the Shapiro-Wilk test and the Cullen-Frey graph. Homogeneity of variances was assessed by the Fisher-Snedecor test. For comparison of data, the one-way ANOVA analysis was used with RIR Tukey posthoc test. Categorical variables were compared using χ2 tests. Classification and regression trees were built with Gini index as a measure of goodness of fit, equal classification error and 10-times cross validation.