Developing a Diagnostic Multivariable Prediction Model for Urinary Tract Cancer in Patients Referred with Haematuria:...

Area of Science:

• Urology
• Oncology
• Medical Prediction Modeling

Background:

• Urinary tract cancer (UTC) risk stratification for patients with haematuria is crucial for timely diagnosis.
• Identifying key patient factors associated with UTC can improve diagnostic pathways.

Purpose of the Study:

• To develop and validate a prediction model for urinary tract cancer (bladder, upper tract urothelial cancer [UTUC], renal) in patients presenting with haematuria.
• To aid clinicians in secondary care settings for prioritizing investigations.

Main Methods:

• Prospective observational study of 10,282 patients across 26 countries.
• Mixed-effect multivariable logistic regression incorporating patient-level predictors.
• Model performance assessed using calibration, discrimination (AUC 0.86), and bootstrap validation.

Main Results:

• The final model identified predictors of increased risk (visible haematuria, age, smoking, male sex, family history) and decreased risk (prior investigations, UTI, dysuria, anticoagulation, catheter use, pelvic radiotherapy).
• Prevalence of bladder cancer was 17.2%, UTUC 1.2%, and renal cancer 1.0% in the studied cohort.
• The model demonstrated good discriminative ability (AUC 0.86).

Conclusions:

• A validated prediction model for urinary tract cancer in haematuria patients has been developed.
• This tool can assist in risk stratification and decision-making for prioritizing investigations in secondary care.
• The model is the first to integrate established and novel diagnostic markers for urinary tract cancer.