Sensitivity and Specificity of Duke Criteria for Diagnosis of Definite Infective Endocarditis: A Cohort Study

The Duke criteria represented a major advance in the diagnosis of infective endocarditis (IE) by formally incorporating echocardiographic findings into the diagnostic schema for IE. The accuracy of the Duke criteria in making a diagnosis of definite IE has not been re-examined. The purpose of this study was to examine the sensitivity and specificity of the Duke criteria in the diagnosis of definite IE in surgically-treated patients with suspected IE. Surgically-treated illness episodes where a diagnosis of IE was strongly entertained by at least one of the treating services were identified from the Cleveland Clinic Infective Endocarditis Registry and Cardiovascular Infections Registry. IE was defined as presence of acute inflammation or microorganisms, or identification of a pathogen on PCR or culture, on an excised cardiac valve. Not IE was defined as absence of both histopathological and microbiological evidence of IE. Episodes with inadequate evaluation of cardiac valves were excluded. The sensitivity and specificity of a diagnosis of definite IE using the non-histopathological elements of Duke criteria were examined. The person evaluating the clinical, microbiological and echocardiographic elements of the Duke criteria was blinded to the histopathological findings. A total of 581 episodes of illness with adequately examined heart valves were identified in 567 patients. The mean (±SD) patient age was 56 (±16) years, 415 (72%) were males, and 293 (50%) had prosthetic valves or rings. Of these episodes 513 were IE and 68 were not. The sensitivity of a diagnosis of definite IE by Duke criteria was 72% (95% C.I. 66–77%) and the specificity was 74% (95% C.I. 58–87%). These remained in a similar range when examined across subgroups of age, sex, year, and presence of a prosthetic valve or ring (figure). Among patients who undergo surgery, the Duke criteria are unable to provide a diagnosis of definite IE in almost 30% of patients with IE, and about 1 in 4 patients in our institution that meet criteria for definite IE by Duke criteria do not have IE. The Duke criteria are not an acceptable replacement for histopathological examination in making a definite diagnosis of IE in surgically-treated patients. All authors: No reported disclosures. Sensitivity and specificity of Duke criteria across subgroups.