Predicting post-radiation genitourinary hospital admissions in patients with localised prostate cancer

Rowan David, Mrunal Hiwase, Arman A. Kahokehr, Jason Lee, David I. Watson, John Leung, Michael E. O‘Callaghan

Research output: Contribution to journalArticlepeer-review

1 Downloads (Pure)


Purpose: The risk of treatment-related toxicity is important for patients with localised prostate cancer to consider when deciding between treatment options. We developed a model to predict hospitalisation for radiation-induced genitourinary toxicity based on patient characteristics. Methods: The prospective South Australian Prostate Cancer Clinical Outcomes registry was used to identify men with localised prostate cancer who underwent curative intent external beam radiotherapy (EBRT) between 1998 and 2019. Multivariable Cox proportional regression was performed. Model discrimination, calibration, internal validation and utility were assessed using C-statistics and area under ROC, calibration plots, bootstrapping, and decision curve analysis, respectively. Results: There were 3,243 patients treated with EBRT included, of which 644 (20%) patients had a treated-related admission. In multivariable analysis, diabetes (HR 1.35, 95% CI 1.13–1.60, p < 0.001), smoking (HR 1.78, 95% CI 1.40–2.12, p < 0.001), and bladder outlet obstruction (BOO) without transurethral resection of prostate (TURP) (HR 7.49, 95% CI 6.18–9.08 p < 0.001) followed by BOO with TURP (HR 4.96, 95% CI 4.10–5.99 p < 0.001) were strong independent predictors of hospitalisation (censor-adjusted c-statistic = 0.80). The model was well-calibrated (AUC = 0.76). The global proportional hazards were met. In internal validation through bootstrapping, the model was reasonably discriminate at five (AUC 0.75) years after radiotherapy. Conclusions: This is the first study to develop a predictive model for genitourinary toxicity requiring hospitalisation amongst men with prostate cancer treated with EBRT. Patients with localised prostate cancer and concurrent BOO may benefit from TURP before EBRT.

Original languageEnglish
Pages (from-to)2911-2918
Number of pages8
Issue number12
Publication statusPublished - Dec 2022


  • Decision curve analysis
  • External beam radiotherapy
  • Genitourinary complications
  • Genitourinary toxicity
  • Hospital admission
  • Hospitalisation
  • Machine learning
  • Prostate cancer
  • Radiation therapy
  • Radiotherapy


Dive into the research topics of 'Predicting post-radiation genitourinary hospital admissions in patients with localised prostate cancer'. Together they form a unique fingerprint.

Cite this