Are the Duke criteria really useful for the early bedside diagnosis of infective endocarditis? Results of a prospective multicenter trial.

BACKGROUND: To date, no studies have evaluated the usefulness of the Duke vs the modified Duke criteria for the early diagnosis of infective endocarditis (IE), nor is it known whether a probabilistic approach may be useful in establishing an early clinical diagnosis of IE. The aim of this study was (1) to assess and compare the clinical usefulness of the Duke vs the modified Duke criteria for the early diagnosis of IE, and (2) to evaluate the diagnostic utility of a probabilistic approach based on the echocardiographic criterion.

METHODS: From January 2000 to December 2001, 267 consecutive patients with suspected IE were enrolled in a prospective multicenter trial.

RESULTS: IE was diagnosed in 147 cases (55%) and rejected in 120 cases (45%). The Duke and the modified Duke criteria had a high similar sensitivity, specificity and accuracy. The time to diagnosis was 8.15 +/- 7.4 days for the Duke criteria and 8.18 +/- 7.1 days for the modified Duke criteria. The time to diagnosis based on a probabilistic approach was shorter than that based on the Duke and the modified Duke criteria (4.96 +/- 7.1 days, for all p < 0.001).

CONCLUSIONS: Although the Duke and the modified Duke criteria have a very similar sensitivity, specificity and accuracy, the delay in the time to diagnosis may be significant. A probabilistic approach based on clinical suspicion and echocardiographic evidence may be useful for decision-making, whilst awaiting case definition by means of the Duke criteria.