The trusted source for
healthcare information and
Abstract & Commentary
Synopsis: Use of modified Duke criteria were reasonably accurate in the diagnosis of culture-negative endocarditis—but expert clinical judgment was almost as good.
Source: Kupferwasser LI, et al. Am Heart J. 2001;142: 146-152.
Kupferwasser and colleagues reviewed their database of patients who had undergone echocardiography in order to evaluate the relative accuracy of diagnosis of culture-negative native valve endocarditis by use of the Duke criteria, the von Reyn criteria, and expert clinical judgment.
The records of 49 patients with fever, heart murmur, elevated acute phase reactants, and at least 1 cardiac abnormality considered to be a risk factor for endocarditis, but who had at least 3 negative blood cultures, were examined. In each case, examination of specimens of valve tissue obtained at autopsy (6 patients) or surgery (43 patients) served as the "gold standard" for diagnosis of culture-negative endocarditis (CNE). Clinical judgment was rendered by 2 masked "clinically experienced investigators" who reviewed abstracted patient records (the manner in which investigator disagreement was managed is not indicated).
Thirty-two of the 49 patients proved to have endocarditis; valve culture was positive in only 6 of the 32 (18.8%)—4 with nutritionally variant streptococci and 1 each with Haemophilus spp and Aspergillus spp. Of the 32 with proven endocarditis, 23 (71.9%) were classified as definite and 9 as possible by the Duke criteria. In 5 of the Duke "possibles" who proved to have endocarditis, 0.2-0.3 mm length vegetations had not been detected by transthoracic or transesophageal echocardiography. Each of these 5 had also been classified as possible by clinical judgment. The other 4 patients with proven endocarditis classified as possible by the Duke criteria were classified as definite by clinical judgment. The performance of the von Reyn criteria was inferior.
In 14 of the 17 patients who proved to not have CNE, the diagnosis of endocarditis had been rejected by the Duke criteria as the consequence of the presence of an alternative diagnosis or resolution of the clinical syndrome in fewer than 5 days. The other 3 without endocarditis had been classified as possible by the Duke criteria and as definite by clinical judgment; in each case, echocardiography had been interpreted as revealing a mobile vegetation that proved to be absent at valve examination; all 3, however, were believed to have nonbacterial thrombotic endocarditis.
Overall, the von Reyn criteria were inferior to the other 2 methods of classification. On the other hand, the Duke criteria and clinical judgment yielded identical classifications in 78% of the 49 patients. The sensitivity and negative predictive value of the latter 2 methods for the diagnosis of CNE were each 100% when both possible and definite classifications were considered "CNE-positive" and when only cases in which transesophageal echochardiography (TEE) results were available in application of the Duke criteria. However, the Duke criteria (with TEE data) appeared to be somewhat superior to clinical judgment with regard to both specificity (82% vs 53%) and positive predictive value (91% vs 80%).
|Table 1: Modified Duke Criteria|
|Criteria||Definite Dx.||Possible||No Endocarditis|
|Histologic||Vegetation/abscess; active histologically||Can’t tell||No pathologic evidence w/ < 4 days prior antibiotic Rx.|
|Microbiologic||+ culture or histology of vegetation or cardiac abscess||Can’t tell||No pathologic evidence w/ < 4 days prior antibiotic Rx.|
|Clinical (any one of the following)|
|Doesn't meet criteria resolution of manifestations w/ < 4 days antibiotic Rx or firm alternate diagnosis|
|Major||2||Does not apply|
|Major & Minor||1 major + 3 minor||1 major + 1 minor|
|Table 2: Modified Duke Criteria Continued*|
|a.||Supportive Laboratory Evidence:|
|•||Typical microorganism for infective endocarditis from 2 separate blood cultures (viridans streptococus, S aureus, S bovis, HACEK group, or community-acquired enterococcus in the absence of a primary focus).|
|•||Persistently positive blood cultures, drawn more than 12 hours apart, yielding one of the above organisms.|
|•||Single positive blood culture for Coxiella burnetti or phase I antibody > 1:800.|
|b.||Evidence of Endocardial Involvement:|
|•||Echocardiogram supportive of infective endocarditis: oscillating intracardiac mass on valve or supporting structures, or in the path of regurgitant jets, or on implanted material in the absence of an alternative explanation or myocardial abscess or new partial dehiscence of prosthetic valve.|
|c.||New Valvular Regurgitation (change in pre-existing murmur insufficient)|
|•||Predisposing cardiac condition or intravenous drug use|
|•||Fever > 38.0° C (100.4° F)|
|•||Vascular phenomena: major arterial emboli, septic pulmonary infarcts, mycoctic aneurysm, intracranial hemorrhage, Janeway lesion.|
|•||Immunologic phenomena: glomerulonephritis, Osler’s nodes, Roth spots, rheumatoid factor.|
|•||Positive blood culture not meeting criteria above (excluding coagulase-negative staphylococci and organisms unlikely to cause endocarditis) or serological evidence of active infection with an organism consistent with infective endocarditis.|
|*Modified from Li SJ, et al. Clin Infect Dis. 2000;30:633-638. (www.med.upenn.edu/bugdrug/antibiotic_manual/duke.html)|
Comment by Stan Deresinski, MD, FACP
Previous studies have validated the accuracy of the Duke criteria in the overall endocarditis population, as well as their superiority to the von Reyn criteria, largely as a consequence of the inclusion of echocardiographic data in the former. Studies have also previously demonstrated general agreement between clinical judgment and the Duke criteria in mostly culture-positive patients. In this study of CNE, clinical judgment was more likely to yield false-positive diagnoses of endocarditis than the Duke criteria, but neither method produced false-negatives.
The Duke classification produced somewhat more frequent equivocal results, classifying cases as possible in 24.5% compared to 18.4% when clinical judgment was used. These cases are, as stated by Kupferwasser et al, problematic in that there is no agreement on how such patients should be managed. However, in most instances, such patients will be managed as if they have endocarditis because of the potential consequences of failure to treat if the infection is, in fact, present.
This study, while demonstrating the potential value of the Duke criteria in the diagnosis of CNE, also demonstrates that clinical judgment is almost as effective. Given the small sample size, the performance of the study at a referral center, and possible selected nature of the cohort, it is not possible to say that one method is preferred over the other in the clinical setting, provided that the clinician making the judgment is experienced and sagacious.
Dr. Deresinski, Clinical Professor of Medicine, Stanford; Director, AIDS Community Research Consortium; Associate Chief of Infectious Diseases, Santa Clara Valley Medical Center, is Editor of Infectious Disease Alert.