The AUA microhaematuria risk tiers performed close to chance

Design

retrospective cohort study with Fine-Gray competing-risk models and comparison of risk classification performance

Population

267,133 adults with de novo microhaematuria in Veterans Health Administration and Medicare-linked data, 2015 to 2024

Primary outcome

new bladder cancer diagnosis within 2 years of the index urinalysis

Effect

2-year incidence 1.5% (95% CI 1.4 to 1.5); model AUC 0.63 vs AUA categories 0.52 (P < 0.001)

A retrospective cohort used Veterans Health Administration data linked to Medicare from 2015 to 2024, covering 267,133 adults with a new diagnosis of microhaematuria, and followed them for a new bladder cancer diagnosis within two years, using Fine-Gray models to account for the competing risk of death.

Two-year cumulative incidence of bladder cancer was 1.5% (95% CI 1.4 to 1.5). Only 11.9% completed cystoscopy. Older age, male sex, smoking history and 25 or more red cells per high-power field were each independently associated with cancer. The important comparison is between risk tools: a model built from these ordinary clinical variables achieved an area under the curve of 0.63, against 0.52 for the current American Urological Association risk categories (P < 0.001). An AUC of 0.52 is barely better than a coin toss, and the model also graded risk within the group the guideline simply labels high risk.

What follows is not that microhaematuria should be investigated less. It is that the current tiers are not doing the job they are used for - they cannot distinguish the 70-year-old smoker with 30 red cells per high-power field from the 45-year-old non-smoker with 5, once both land in the same category. Until a validated individual risk tool is available, weight the four factors this study identifies explicitly when you decide on cystoscopy, and say so in the note. The 1.5% baseline is also worth carrying into the conversation: for most patients this is a low-probability investigation, and telling them so is part of consenting them to it.

• Weight age, sex, smoking history and red cell count explicitly rather than relying on the tier alone
• Treat 25 or more red cells per high-power field as a distinct escalation, not just 'microhaematuria'
• Quote the roughly 1.5% two-year risk when consenting a patient to cystoscopy
• Document which risk factors drove the decision, so a repeat episode is not restaged from scratch
• Do not read a low completion rate as low risk - only 11.9% actually had cystoscopy here

Why it matters

The risk classification urologists use to decide who needs cystoscopy performed barely better than chance in a quarter of a million patients.

Don't overread it

Retrospective, and heavily male and older; the improved model was derived in this dataset and has not been externally validated, so it is not yet a tool to adopt.

The statistics, in plain English

An area under the curve of 0.5 is random guessing and 1.0 is perfect; the guideline categories at 0.52 are close to random, and the improved model at 0.63 is better but still modest - neither is a tool to rule out cancer in an individual. The 1.5% incidence is precise because the cohort is enormous, but it is a cohort of veterans, mostly older men with high smoking rates, so the absolute risk will be lower in a general population and much lower in younger women.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.