Patient-specific risk of undetected malignant disease after investigation for haematuria, based on a 4-year follow-up.

Objectives: • To estimate the diagnostic accuracy of a guidelines-based haematuria clinic protocol by measuring the incidence of undetected malignancy during a follow-up period. • To estimate an individual's post-test risk of having undetected malignancy using the protocol likelihood ratio and the population prevalence of disease.

Methods: • Data were collected prospectively on a cohort of 4020 consecutive patients who were referred to a 'one-stop' haematuria clinic between 1998 and 2003. • All patients had a plain radiograph taken and underwent ultrasonography and flexible cystoscopy as a part of 'first-line' investigation. • Intravenous urography was performed where indicated after abnormal first-line tests or in patients with persistent haematuria where no abnormality had been detected. • Records of the initial 687 participants from the first year of the study were reviewed 4 years after the original consultation. Missed diagnoses of urinary tract malignancy were recorded and sensitivities, likelihood ratios and the post-test probability of missing all disease and upper tract malignancy were calculated.

Results: • As previously reported, the overall prevalence of malignant disease was 12.1% (18.9% for macroscopic haematuria compared with 4.8% for microscopic haematuria). • The records of the first year's cohort of patients (N = 687) were analysed 4 years after their original consultation and 10 potentially 'missed' tumours were identified. • The sensitivity of the protocol was 90.9% for the detection of all urinary tract malignancy (95% CI, 82.4 to 95.5) and 71% for upper tract tumours alone (95% CI, 45.4-88.3). The latter improves to 78.6% (95% CI, 52.4-92.4) with the addition of further upper tract testing. • The probability of missing malignant disease overall was 1.7% (95% CI, 0.95-3.04) but this rose sharply to >4% for males over 60 with macroscopic haematuria. • For those with non-visible haematuria, the percentage probability of missed malignant disease was less than 1%.

Conclusions: • The haematuria clinic protocol described is robust but it is not infallible. • The risk of missing malignant disease in the higher risk groups identified in the study is much greater than previous studies would suggest. • If additional upper tract testing or interval follow-up were to be recommended, it could be rationally targeted at these groups, given the measurable risk shown here.