See the Editorial Commentary by van der Werf on pages 835-6.) Background. The diagnosis of the neglected tropical skin and soft tissue disease Buruli ulcer (BU) is made on clinical and epidemiological grounds, after which treatment with BU-specific antibiotics is initiated empirically. Given the current decline in BU incidence, clinical expertise in the recognition of BU is likely to wane and laboratory confirmation of BU becomes increasingly important. We therefore aimed to determine the

more »

... stic accuracy of clinical signs and microbiological tests in patients presenting with lesions clinically compatible with BU. Methods. A total of 227 consecutive patients were recruited in southern Benin and evaluated by clinical diagnosis, direct smear examination (DSE), polymerase chain reaction (PCR), culture, and histopathology. In the absence of a gold standard, the final diagnosis in each patient was made using an expert panel approach. We estimated the accuracy of each test in comparison to the final diagnosis and evaluated the performance of 3 diagnostic algorithms. Results. Among the 205 patients with complete data, the attending clinicians recognized BU with a sensitivity of 92% (95% confidence interval [CI], 85%-96%), which was higher than the sensitivity of any of the laboratory tests. However, 14% (95% CI, 7%-24%) of patients not suspected to have BU at diagnosis were classified as BU by the expert panel. The specificities of all diagnostics were high (≥91%). All diagnostic algorithms had similar performances. Conclusions. A broader clinical suspicion should be recommended to reduce missed BU diagnoses. Taking into consideration diagnostic accuracy, time to results, cost-effectiveness, and clinical generalizability, a stepwise diagnostic approach reserving PCR to DSE-negative patients performed best.