once weekly 12 times) instillation; D, early and high-dose. The influence of the instillation protocols and tumour characteristics on the probability of recurrence-free survival was examined using Kaplan-Meier analysis and a Cox regression hazard model.
RESULTSThe early-instillation and high-dose groups had relatively lower recurrence rates after 6 months (A, 30%; B, 25%; C, one of 12; and D, none) and 1 year (50%, 35%, four of nine and one of eight, respectively). Patients who received 360 mg epirubicin (C and D) had a significantly better recurrence-free survival than those receiving 180 mg (A and B; P = 0.012). Preoperative urine cytology and tumour multiplicity were significantly associated with recurrence. However, multivariate analysis of the risk of recurrence using a Cox proportional hazard model showed that urine cytology (hazard ratio 3.11, 95% confidence interval 1.08-8.94, P = 0.04) and total dose (0.32, 0.11-0.92, P = 0.03) were independent prognostic factors for recurrence.