An assessment of the current diagnostic criteria for infective endocarditis

Infective endocarditis (IE) is a condition in which microbial infection occurs on the endothelial surface of the heart. It is characterized by the formation of valvular vegetations, fenestration, abscess and dehiscence of prosthetic valves. Patients with IE may develop fever, malaise, weight loss, anorexia, and symptoms of immunological and embolical phenomenon. However, because of the protean manifestations of IE and the inaccessibility of the involved organ, the disease is often diagnosed with uncertainty based on clinical manifestations. For example, a new onset of a prolonged PR interval on an electrocardiogram may be indirect evidence of myocardial abscess, but this method is limited by its low sensitivity. The diagnosis of IE is never absolutely certain until explanted valves are assessed pathologically or an autopsy is available. There have been very few published guidelines that assist in the diagnosis of IE. Newly trained physicians often associate the presence of a febrile illness in a patient with a prosthetic heart valve with endocarditis and may subject patients to unnecessary invasive studies and therapy. Should all patients who have prosthetic heart valves and fever undergo an echocardio-

graphic study to evaluate the possibility of IE? Who should be treated with a full course of antibiotics for suspected IE?The purpose of this article is to review the diagnostic criteria used to diagnosis IE.

VON REYN CRITERIA
In 1981, von Reyn et al (1) published clinical pathological criteria for classifying patients suspected of having IE.The criteria classified cases into definite, probable, possible and rejected endocarditis (Table 1).These criteria have been criticized in recent years for a variety of reasons.First, pathological confirmation is necessary to designate a case as definite IE.Because only some patients require surgery, a substantial portion of cases are classified as 'possible IE', and some of them rejected.Second, the criteria do not incorporate information provided by echocardiography, which plays a critical role in the assessment of patients with suspected endocarditis.Third, they do not consider intravenous drug abus-ers as high risk and predisposed to IE due to needle-sharing.Because of these criticisms, there is a need for a more comprehensive set of guidelines for diagnosing IE.

DUKE CRITERIA
Durack et al (2) from Duke University, Durham, North Carolina introduced a new set of criteria that incorporated findings from the two-dimensional echocardiogram and a history of intravenous drug use into the clinical assessment.The new criteria were modelled after the Jones criteria (3) used for the identification of cases of rheumatic fever.
The proposed Duke criteria defined cases of suspected IE into definite, probable and rejected (Table 2).'Definite IE' was intended to identify patients with a very high likelihood of having IE.Cases for which pathological confirmation is available are classified into this category.On the other hand, because only one-third of patients with native valve IE undergo surgery or autopsy during the acute illness, cases that fulfil certain clinical parameters that are considered typical of IE are also included in this category (Table 3).The 'rejected IE' group comprises cases in which the syndrome resolves spontaneously with a short course of antibiotic therapy or when an alternate diagnosis that explains the presentation is made.Cases that belong to the 'possible IE' category are those that do not meet all the criteria for 'definite IE' and yet manifest findings that are sufficiently suggestive of IE to make rejection of the diagnosis clinically unjustified.
The clinical features of IE are separated into two major and five minor criteria under the Duke criteria.Cases that have two major, one major plus three minor, or five minor criteria are considered definite cases of IE.Two features that are crucial with respect to blood cultures are persistent bacteremia and typical microorganisms.rium hominis, Eikenella species and Kingella kingae) are common causes of IE and are infrequently isolated from blood cultures of patients without IE.On the other hand, Staphylococcus aureus and enterococcus may be typical endocarditis pathogens, but they also cause bacteremia in patients with confirmed extracardiac infections without IE, especially those that are acquired nosocomially.The Duke schema weighs persistent enterococcal and staphylococcal bacteremias as major criteria only when they are both community acquired and arise without an apparent primary focus of infection.
Coagulase-negative staphylococcus can cause IE but is often isolated from blood culture usually because of contamination.This explains why persistent bacteremia is a key feature of these criteria.The echocardiographic manifestations of IE include discrete, echogenic oscillating intracardiac masses located at sites of endocardial injury, periannular abscess and new dehiscence of a prosthetic valve.Based on the authors' clinical experience, echocardiography does not appear to be sensitive enough to pick up fenestrations of valvular leaflets; however, identification of fenestrations should be as a major criterion.Nonspecific valve thickening is not evidence of echocardiographic definite IE.False positive echocardiographic features in prosthetic valves could be related to stitches to the valvular ring or noninfective paravalvular leak.New valvular regurgitation by auscultation, however, is considered a main criterion under the Duke classification.
Predisposing factors include patients who have had previous mechanical or tissue prosthetic valves, intracardiac instrumentation or intravenous drug use.The presence of an ischemic leg or focal neurological manifestation consistent with stroke belongs to the vascular phenomenon.These conditions must be confirmed radiologically.Splinter hemorrhages, subconjunctival hemorrhages and clubbing are considered too nonspecific, and, therefore, they are not included in this category.Immunological phenomena include immune complexmediated events such as glomerulonephritis, Osler's nodes and Roth's spots.Echocardiographic features of the Duke minor criteria are those that are suggestive of IE but are not mentioned in the major category; valvular thickening and nodules are examples of this.Microbiological features include bacteremia not due to the typical organisms causing IE or serological evidence of recent infection by a pathogen known to cause valvular lesions (eg, brucella, legionella, chlamydia or Coxiella burnettii), where routine blood cultures may not be able to isolate the pathogen.Bartonella species have been recently recognized as an important cause of apparent 'blood culture-

EVALUATION OF THE DUKE CRITERIA
In addition to the internal evaluation by the Duke Endocarditis Service, the recently proposed criteria were also evaluated independently by several other centres, namely the University of Southern California in Los Angeles (7), the Uni-versity of Nancy Medical Center in Vandoeuvre cedex, France (8) and the University of Alberta (9).Summarizing the results reported from the four centres, a total of 638 patients with suspected IE were evaluated; 163 had IE proven pathologically and 475 did not (Tables 4,5).Calculation of sensitivity and specificity was difficult because the pathology (the gold standard) is not available on all patients.However, the conclusion drawn from these studies is that, using the new set of criteria, more cases would be classified as definite versus probable, possible or even rejected compared with the old criteria.Dodds et al (10) showed that the negative predictive value of the new criteria was as high as 98% during a threemonth follow-up of patients in whom the diagnosis of IE was rejected.In this study, no new cases were found in these patients except one case where the patient's autopsy revealed evidence that might represent missed IE.Thus, the Duke criteria are more sensitive than the old criteria because they include more patients in the positive test category.Moreover, a patient failing to satisfy the criteria for probable IE is very unlikely to have IE.6 and 7.One hundred and eighty-three and 23 cases were labelled as 'definite IE' and 'probable IE', respectively, under the new criteria (ie, no cases were rejected under the new criteria).There were 90 'definite IE' patients and two in the 'probable IE' group who underwent cardiac surgery.The two patients who belonged to the 'probable' group presented with one major (positive blood culture) and two minor criteria; however, during surgery only one patient was identified as having active infective endocarditis.Of note, 98% of the patients with native valve endocarditis had typical blood cultures, and a large portion of patients also had typical echocardiographic features of IE (Table 6).This implies that in the absence of bacteremia and no preceding antibiotic use, patients who have native valves and present with other features of IE are in fact unlikely to have endocarditis.

THE TORONTO HOSPITAL EXPERIENCE
In addition, more than half of the patients proceeded to transesophageal echocardiography after transthoracic scanning.Moreover, a higher percentage of patients with prosthetic valves had transesophageal echocardiogram as the initial radiological test compared with native valve group, probably because of a high pretest suspicion (presence of positive blood culture and prosthesis) and poor acoustics of prosthesis using transthoracic method.
Among patients with prosthetic valves, the majority had typical microorganisms cultured from their blood, but typical echocardiographic features were seen less frequently than in those with native valve IE.This is likely due to the difficulty of detecting vegetations in the presence of valvular prosthesis, mainly as a result of acoustic shadowing on echocardiograms.Our anecdotal experience suggested that patients whose blood cultures were positive for typical organisms and had a history of previous prosthetic valve surgery should be treated with full antibiotic therapy, as if endocarditis was present.

CONCLUSIONS
Infective endocarditis is difficult to diagnose with certainty.Clinicians often face a diagnostic dilemma when a patient presents with some symptoms suggesting IE, especially in cases when patients with prosthetic heart valves present with febrile illness.However, careful examination using the guidelines from the Duke criteria may assist clinicians in making a diagnosis of IE with greater certainty, and thus the decision regarding antibiotics administration may be made more appropriately.In our centre, the Duke criteria identified most patients with definite endocarditis except for one patient who subsequently was proven to have IE during surgery.The Duke criteria, therefore, provide a more satisfactory set of guidelines than the previous guidelines for diagnosing IE and determining which patients should be treated.

TABLE 1 von Reyn criteria for diagnosis of infective endocarditis
Adapted from reference 2 with permission of the publisher Excerpta Medica Inc

TABLE 2 Duke criteria for diagnosis of infective endocarditis Definite Pathological
criteria: 1. Microorganisms: demonstrated by culture or histology in a vegetation, or in a vegetation that has embolized, or in an intracardiac abscess, or 2. Pathological lesions: vegetation or intracardiac abscess present, confirmed by histology showing active endocarditis Clinical criteria, using specific definitions listed in Table 3: • two major criteria, or • one major and three minor criteria, or • five minor criteria Possible Findings consistent with infective endocarditis that fall short of 'definite' but not 'rejected' Rejected 1. Firm alternate diagnosis for manifestations of endocarditis, or 2. Resolution of manifestations of endocarditis, with antibiotic therapy for four days or less, or 3.No pathological evidence of infective endocarditis at surgery or autopsy, after antibiotic therapy for four days or less Adapted from reference 2 with permission of Excerpta Medica Inc Dis Vol 9 No 4 July/August 1998 237

TABLE 3 Definitions of terminology used in the Duke criteria Major criteria
Positive blood culture for infective endocarditis Typical microorganism for infective endocarditis from two separate blood cultures • Viridans streptococci*, Streptococcus bovis, HACEK group, or • community-acquired Staphylococcus aureus or enterococci,

TABLE 5 Comparison of clinical diagnosis in patients evaluated for di- agnosis of infective endocarditis from four independent centres (2,4-6) using von Reyn criteria and Duke criteria ex- cluding pathologically proven cases Duke criteria von Reyn criteria Definite Possible Rejected Total (%)
Patient selection:The Toronto Hospital is a tertiary referral centre.The authors' experience with IE from 1989 to 1995 matched the data reported from other centres.Two hunderd and six cases were identified based on a discharge diagnosis of IE reported to the hospital registry.The hospital charts were reviewed retrospectively, and each case was classified as 'def- 238Can J Infect Dis Vol 9 No 4 July/August 1998

TABLE 7 Clinical course of patients with infective endocarditis at The Toronto Hospital
The clinical characteristics of the patients, including the causative organisms and the course in the hospital, are presented in Tables