Objectives: To establish and to evaluate discriminant models to predict the outcomes of transurethral prostatectomy.
Methods: Clinical data of patients treated with transurethral prostatectomy between January and December 2013 were collected, including medical history, symptoms, biochemical tests, ultrasonography and urodynamics. Surgical efficacy was evaluated at 6-month follow up. Predictive models were constructed by logistic regression. Receiver operating characteristic curve and diagnostic tests were used to test the accuracy of models before the predictive value between models was compared.
Results: A total of 182 patients were included, with 73.6% having an effective outcome. History of recurrent urinary tract infection (OR 1.33), score of storage phase (OR 2.58), maximum flow rate (OR 2.11) and detrusor overactivity (OR 3.13) were found to be risk factors. International Prostate Symptom Score (OR 0.13), transitional zone index (OR 0.58), resistive index of prostatic artery (OR 0.46), bladder wall thickness (OR 0.78), ultrasonic estimation of bladder weight (OR 0.28), bladder outlet obstruction index (OR 0.20) and bladder contractility index (OR 0.83) were found to be protective factors. The areas under the curve of models using factors from ultrasonography and urodynamics were 0.792 and 0.829 respectively, with no significant difference being found between them (P = 0.348).
Conclusions: Surgical efficacy of transurethral prostatectomy is positively correlated to severe voiding phase symptoms, outlet obstruction and better detrusor contractility, and negative correlated with urinary infection, severe storage phase symptoms and excessive detrusor contractibility. Ultrasonography might replace urodynamics in selecting patients for whom transurethral prostatectomy is more likely to be beneficial.
Keywords: predictive model; transurethral resection of prostate; treatment outcome; ultrasonography; urodynamics.
© 2015 The Japanese Urological Association.