Objectives To identify factors aiding the selection of patients with gynaecological cancer with malignant urinary obstruction who are least likely to benefit from palliative urinary diversion (UD), and to create a risk-stratification model for decision-making.

Methods This historic cohort study comprised 74 consecutive patients with urinary obstruction resulting from gynaecological malignancies. All underwent palliative UD by percutaneous nephrostomy (PCN). Using the Cox proportional hazards regression model and Kaplan-Meier curves with the log-rank test, we developed a prognostic score identifying candidates least likely to benefit from the intervention.

Results The median follow-up was 4.72 (range 0–5.71) years. Hydronephrosis was diagnosed in most patients on recurrent or persistent disease (81%). It was bilateral in 37.8%. Intervention-related complications included urinary sepsis (8%), catheter dislodgment requiring replacement (17%) and gross haematuria necessitating blood transfusions (13%). After PCN, conversion to an internal ureteral stent was feasible in 46%. The median survival was 11.13 (range 0–67) months. Two patients died within a month of UD. Multivariate analysis identified diabetes mellitus (DM), poor Eastern Cooperative Oncology Group (ECOG) performance status >1 and ascites as significant negative survival factors. A prognostic index based on those factors identified the short-term and long-term survivors. Risk factor-based mortality HRs were 11.37 (95% CI 4.12 to 31.37) with one factor, 26.57 (95% CI 9.14 to 77.26) with two factors and 67.25 (95% CI 15.6 to 289.63) with three factors (all with p<0.0001).

Conclusions Our proposed prognostic index, based on ascites, ECOG performance status and DM, might help select patients with gynaecological cancer least likely to benefit from palliative UD.