Diagnostic Value of Echocardiography in Suspected Endocarditis
An Evaluation Based on the Pretest Probability of Disease
Background We hypothesized that for the diagnosis of endocarditis, (1) transthoracic echocardiography (TTE) would be most valuable in patients with an intermediate clinical probability of the disease and (2) transesophageal echocardiography (TEE) would be most useful in patients with an intermediate probability when TTE either does not yield an adequate study or indicates an intermediate probability of endocarditis. We also sought to investigate the influence of echocardiographic results on antibiotic usage and its duration.
Methods and Results TTE and TEE were performed in 105 consecutive patients with suspected endocarditis. Patients were classified as having either low, intermediate, or high probability of endocarditis on the basis of clinical criteria and separately on the basis of both TTE and TEE findings. TTE and TEE classified the majority (82% and 85%, respectively) of the 67 patients with a low clinical probability of endocarditis as having a low likelihood of the disease. Of the 14 patients with intermediate clinical probability, 12 had technically adequate TTE studies; 10 of these (83%) were classified as either high or low probability. All patients with intermediate clinical probability were classified as high or low probability by TEE. The majority of the 24 patients with high clinical probability were placed in the low-likelihood category by echocardiography (15 by TTE and 12 by TEE). There was concordance between TTE and TEE in 83% of all cases. TEE was useful for the diagnosis of endocarditis in patients with prosthetic valves and in those in whom TTE indicated an intermediate probability; these constituted <20% of patients in our study. The course of antibiotic therapy was influenced only by the clinical profile and not by the echocardiographic results.
Conclusions Echocardiography should not be used to make a diagnosis of endocarditis in those with a low clinical probability of the disease. In those with an intermediate or high clinical probability, TTE should be the diagnostic procedure of choice. TEE for the diagnosis of endocarditis should be reserved only for patients who have prosthetic valves and in whom TTE is either technically inadequate or indicates an intermediate probability of endocarditis.
Two-dimensional echocardiography has the ability to detect vegetations and their associated complications in patients with endocarditis. According to the population studied, transthoracic echocardiography (TTE), which has been used in clinical practice for the past 15 years, has been reported to have a sensitivity of anywhere from 40% to 80% for the detection of vegetations.1 2 3 4 5 6 7 8 With transesophageal echocardiography (TEE), the image quality is better because of no interposing lung tissue and the higher transducer frequencies used. Hence, it has been reported to be superior to TTE for the detection of vegetations and their associated complications.8 9 10 11 12 13 14 15 16
Echocardiography is often requested in patients with fever and/or a murmur who have a low probability of endocarditis. By the same token, it is also requested in patients in whom the diagnosis is virtually certain. Patients with an intermediate probability of endocarditis constitute a minority of those undergoing echocardiography. The practical value of echocardiography in these different populations has not been investigated.
We hypothesized that the value of echocardiography for the diagnosis of endocarditis would be greatest in patients with an intermediate clinical probability of the disease. According to the same logic, we also postulated that TEE would be most useful in patients with an intermediate probability of endocarditis in whom TTE either does not yield a technically adequate study or indicates an intermediate probability. We also sought to determine the influence of echocardiographic findings on the use and duration of antibiotic treatment in patients with suspected endocarditis.
Between October 1992 and June 1994, 105 consecutive patients who were referred to the University of Virginia echocardiography laboratory for evaluation of suspected endocarditis were studied prospectively. All patients gave informed consent to participate in the study, the protocol for which was approved by the Human Investigation Committee at the University of Virginia. There were 59 men and 46 women ranging in age from 16 to 88 years. All patients were categorized as having a low, intermediate, or high probability of endocarditis on the basis of a modification of previously published definitions (Table 1⇓).17 The duration of antibiotic therapy was determined for each patient from the hospital records.
All patients underwent TTE and TEE within 36 hours of each other; in 89%, both procedures were performed in tandem during the same session. TTE was performed with a 2.5- or 3.5-MHz phased-array transducer. Patients fasted for more than 4 hours before TEE, which was performed under local pharyngeal anesthesia; the majority of patients also received intravenous midazolam (0.5 to 4.0 mg). A 5-MHz phased-array transducer (either biplane or multiplane) was used for the transesophageal examination, which consisted principally of two-dimensional imaging and color flow mapping and was performed without any complications within 15 minutes in all patients. The information obtained on echocardiography was made available to the referring physicians.
All echocardiograms were evaluated later during reading sessions by two observers who were blinded to the clinical information. TTE and TEE studies were read in random order on separate occasions, which ensured that the results of the TTE study on a patient were not known to the observers while they were reading the TEE study on the same patient. TTE studies were defined as technically inadequate if both observers deemed the quality of the images to be insufficient to gain diagnostic information regarding the presence or absence of vegetations or their complications. Findings on TTE and TEE were separately categorized as indicating high, intermediate, or low probability for endocarditis as follows: high, any definite vegetation and/or abscess or probable vegetation with evidence of otherwise unexplained valvular dysfunction (greater than mild regurgitation or a paravalvular prosthetic leak); intermediate, a probable vegetation without evidence of unexplained valvular dysfunction; and low, no evidence of vegetation or abscess or a possible vegetation without any evidence of regurgitation. Echocardiographic evidence of vegetation or abscess was determined according to previously defined criteria.18 19 20
Comparisons between groups were made by either a χ2 or unpaired Student’s t test or ANOVA. Comparisons of reliability between various diagnostic methods were based on κ statistics and provided a means for calculating agreement that occurred in excess of chance alone. Cox regression models and Kaplan-Meier analysis were used to evaluate the impact of the clinical criteria and either echocardiographic technique on the use and duration of antibiotic therapy. All tests were two-sided, and a value of P<.05 was considered to be statistically significant.21
According to the criteria presented in Table 1⇑, 67 patients were deemed to have a low, 14 to have an intermediate, and 24 to have a high probability of endocarditis. The historical data were similar among all subgroups, with the exception of a higher incidence of previous valvular disease in the intermediate-probability group (Table 2⇓). The time from onset of illness to performance of echocardiography was also similar between the groups (Table 2⇓). As expected, the presence of many of the clinical abnormalities associated with endocarditis increased with a higher clinical probability of the disease (Table 3⇓). Three of the 5 patients with negative blood cultures and yet a high clinical probability of endocarditis were on antibiotics at the time of sample collection.
There were 12 patients with technically inadequate TTE studies, whereas all TEE studies were adequate for interpretation. Vegetations were identified in 23 patients by each approach, and the prevalence of vegetations by either echocardiographic approach increased with increasing clinical likelihood of endocarditis (Fig 1⇓). In 2 of these patients, an abscess was identified by both approaches, and in 1 additional patient, an abscess was seen on TEE only. In all patients with an abscess, either a definite or probable aortic valve vegetation was seen by both techniques. Eight patients thought to have vegetations on TTE were found to have nonspecific valvular abnormalities (thickening or calcification) on TEE. Six of these were subsequently classified as having a low probability of endocarditis by TEE as opposed to intermediate (n=5) or high (n=1) probability by TTE.
Comparison of Clinical and Echocardiographic Data
Fig 2A⇓ illustrates the echocardiographic results in the 67 patients with a low clinical probability of endocarditis. Both TTE and TEE classified the majority of these patients as low probability (47 of the 57 interpretable TTE and 57 of the 67 TEE studies). Of the remaining 10 patients with low clinical probability who were classified as intermediate or high probability by echocardiography, approximately half had suspected prosthetic valve endocarditis. The remainder were deemed not to have a final diagnosis of endocarditis despite the echocardiographic findings.
Fig 2B⇑ depicts the echocardiographic results in the 14 patients with an intermediate clinical probability of endocarditis. TTE reclassified the majority of the patients (10 of the 12 interpretable studies) into either low or high probability, whereas TEE classified all patients as either high or low probability.
Fig 2C⇑ shows the echocardiographic results in the 24 patients with a high clinical probability of endocarditis. Approximately half of these were classified as low probability by echocardiography (15 by TTE and 12 by TEE). In many of these, however (7 patients), another source of infection was identified later in their hospital course. These patients originally had a high clinical probability of endocarditis because of the presence of nonspecific vascular phenomena (petechiae, embolism) and a murmur. The rest had significant valvular regurgitation due to structural abnormalities of the valve (flail leaflet or rheumatic, sclerotic, or prosthetic valve) without evidence of concurrent endocarditis. No patient with another source of infection was classified as having intermediate or high likelihood of endocarditis by either TTE or TEE.
Evaluation of Prosthetic Valves
Of the 17 patients with prosthetic valves, 12 were classified as low, 2 as intermediate, and 3 as high probability for endocarditis on clinical grounds. TTE classified only 2 of the 11 interpretable studies in patients with low clinical probability into either intermediate or high probability, whereas TEE classified 5 of the 12 patients as such. Both approaches classified the 2 patients with intermediate clinical probability as low probability for endocarditis. The 3 patients with high clinical probability were classified as low probability for endocarditis; another source of infection was later identified in all of them.
Comparison of TTE and TEE
In those with interpretable studies, there was 83% agreement between TTE and TEE for determining the likelihood of endocarditis. Concordance between the two techniques was high in all clinical categories and was significantly greater than could have occurred by chance alone (Table 4⇓). The probability of chance agreement varied among the three clinical probability groups because of the variable prevalence of positive echocardiographic findings. The greatest discordance was noted in the 10 patients with an intermediate probability of endocarditis on TTE, 7 of whom were reclassified as low or high probability by TEE. As depicted in Fig 3⇓, the distribution of valves showing possible, probable, or definite vegetations was similar for both techniques. Concordance between TTE and TEE for either presence or absence of vegetations was high, with an overall concordance of 96% for all four valves.
Implications for Management
The probability of endocarditis based on clinical criteria was the only variable that predicted the use and duration of antibiotics by univariate analysis (Fig 4⇓). On multivariate analysis, the relative likelihood of being free of antibiotic use was approximately twofold in the low compared with the high clinical probability group (P<.05). Neither TTE nor TEE significantly influenced antibiotic use in this patient cohort, although there was a trend for increased antibiotic use in those who had a high probability of endocarditis on the basis of TEE. Four patients underwent valve replacement (3 of native and 1 of prosthetic valve) because of unremitting congestive heart failure. Three of these patients had a high probability of endocarditis both clinically and by either echocardiographic approach.
Our results indicate that echocardiography, both TTE and TEE, has a low diagnostic yield in patients with a low clinical probability of endocarditis, who constitute the majority of patients in whom it is requested to “rule out” the condition. We found a concordance between TTE and TEE findings in the majority of our patients. It is only in those in whom TTE suggested an intermediate probability that TEE offered incremental value. The probability of endocarditis on the basis of clinical criteria was the only variable that predicted the use and duration of antibiotic treatment. Neither TTE nor TEE significantly influenced antibiotic use in this patient cohort.
Diagnosis of Endocarditis
Despite advances in technology, endocarditis remains a clinical diagnosis. As exemplified by Libman and Celler in 1910,22 the diagnosis rests largely on isolation of the blood-borne pathogen, a notion still held today.23 In certain subsets of patients clinically suspected of endocarditis, however, confirmatory evidence may be lacking, such as in those with “culture-negative” endocarditis. Such patients may have already received antibiotics before blood samples are obtained for microbiological analysis or may have pathogens that cannot be readily isolated. These patients are in the minority and could benefit from echocardiography, in which the presence of a definite vegetation may help confirm the diagnosis.23
In the majority of other patients with endocarditis, however, blood cultures are usually positive. In these patients, it is unlikely that echocardiography can offer any additional information regarding the presence or absence of endocarditis. Certain findings on physical examination, such as splinter hemorrhages, Janeway lesions, or Osler’s nodes, increase the likelihood of endocarditis. Similarly, other sources of fever or other sites of infection that are not usually associated with endocarditis make it unlikely that a patient has endocarditis. The clinical picture rarely results in a certain diagnosis, but neither does echocardiography.
Echocardiography in Native Valve Endocarditis
As is the case with most diagnostic tests, the earlier studies showing the value of TTE in endocarditis were performed in highly selected patients in whom the presence or absence of endocarditis was virtually certain on clinical grounds.6 8 10 The high positive and negative predictive values of the test were influenced by the high pretest likelihood of having or not having the disease. As the clinical experience increased and as the test was used more indiscriminately, the predictive value decreased.14 16 24
More recently, TEE has been tested in selected patients with a high likelihood of endocarditis. Because of its superior imaging capabilities, the level of confidence for determining the presence of a vegetation is higher than that with TTE.8 9 10 11 12 13 14 15 16 This may be particularly relevant for less experienced observers. As can be noted in our study, despite our extensive experience with echocardiography, we also tended to place more patients with abnormal native valves in the intermediate probability category for endocarditis on the basis of TTE. In some of these patients, we were not sure whether a vegetation coexisted with the underlying abnormality, and it was in these patients that TEE provided the greatest advantage. It is important to note, however, that in the majority of patients with abnormal native valves, we were able to determine the likelihood of endocarditis from TTE alone.
Echocardiography in Prosthetic Valve Endocarditis
We concur with previous suggestions12 15 25 that patients with prosthetic valves suspected of having endocarditis do best by having TEE directly. The clinical picture in these patients can be atypical, and it seems prudent not to place a patient with a prosthetic valve and fever in a low probability category under the criteria used for native valves. Because of reverberations and attenuation produced by the prosthetic material, it may not be possible to discern a small vegetation in these patients by TTE, with the result that these patients tend to be placed in the intermediate probability category anyway if this approach is used. TEE offers the greatest value in these patients by more clearly defining the presence or absence of vegetations.
Influence of Echocardiographic Results on Antibiotic Course
In the aggregate, it was the clinical suspicion of endocarditis that determined the use and duration of antibiotic therapy. Neither echocardiographic approach significantly influenced antibiotic use. These results should be interpreted with caution, however. Patients with an initially high clinical probability of endocarditis in whom vegetations are not shown by echocardiography may be treated more aggressively with antibiotics simply because they have another source of infection. On the other hand, it can be argued that if the clinical profile and microbiological results are the primary determinants of antibiotic use, echocardiographic results play at best a minor role in this determination.
Limitations of the Study
The incremental value of a test for making a diagnosis is best determined when a “gold standard” is available. Short of a pathology specimen taken either at necropsy, during operation, or from an embolus, there is no way of making a definitive diagnosis of endocarditis. Only a probability can be estimated from clinical and laboratory evidence. Since we were testing the utility of echocardiography, we could not use a vegetation on TEE as a gold standard for the presence of a vegetation.
According to the probability theory, a test should not offer additional information when the pretest likelihood of disease is either low or high.26 27 Unlike the group with low clinical probability, however, echocardiography did offer additional diagnostic information in the group with high clinical probability. Although the prevalence of vegetations on echocardiography was highest in this subgroup, they were not seen in at least half of the patients. One possibility is that vegetations were not large enough to be seen on echocardiography. The other, and we believe more likely, explanation is that there is poor discrimination between intermediate and high probability of endocarditis on the basis of clinical grounds alone. For instance, almost all patients in our study who had a high probability of endocarditis according to clinical criteria but in whom echocardiography showed no vegetations had either petechiae or vascular embolic phenomena and were later found to have another source of infection. Thus, as suggested by others,28 some of the findings that are used to place patients into the high clinical probability category appear to be nonspecific. If it is not possible to separate intermediate- from high-risk patients on the basis of clinical criteria, echocardiography may have a role as an additional diagnostic tool in these patients. As noted in our study, in most patients the additional information can be readily obtained by TTE alone.
A new set of diagnostic criteria has recently been formulated that categorizes findings into major and minor criteria analogous to the Jones criteria for rheumatic fever.28 It has been suggested that these new criteria improve the diagnostic accuracy for endocarditis by placing less weight on nonspecific findings (such as fever, vascular events, and immunologic phenomena). If these criteria were applied to our study population, the number of patients falsely classified as having a high probability by clinical grounds would decrease. This new paradigm, however, requires that echocardiography be performed in all patients with suspected endocarditis regardless of clinical probability. The results of our study indicate, however, that except in the case of prosthetic valves, patients with a low clinical probability of endocarditis do not benefit from echocardiography.
Our study did not address the issue of detecting the complications of endocarditis with echocardiography. Since complications occur in a minority of patients with endocarditis, confirmation with echocardiography can be reserved for those in whom there is a clinical probability of complications. It has also been argued that in patients suspected of having endocarditis, the absence of vegetations may preclude the use of a protracted full course of intravenous antibiotics.29 It is implied that routine use of TEE could therefore be cost saving in patients with suspected endocarditis. The results of our study pertaining to antibiotic use and duration do not support such a contention. Physicians tend to treat the clinical condition rather than the vegetations.
Even though we studied consecutive patients, selection bias is likely in our study for several reasons. First, ours is a tertiary care center, and patients seen at our institution do not necessarily reflect those seen in community hospitals. We serve a rural population, and the type of endocarditis seen at our institution is different from that seen in large metropolitan medical centers. For example, we saw right heart vegetations in only two patients, an incidence much lower than reported from city hospitals in which the incidence of intravenous drug abuse is higher. Nevertheless, unlike left heart vegetations, right heart vegetations are equally well visualized by TTE and TEE.30
Twelve patients had technically inadequate studies on TTE. Excluding these patients from comparative analysis is unlikely to have affected our results, since the demographic, clinical, and TTE findings of these patients were not significantly different from those of the rest of the patient cohort. Finally, the role of echocardiography may change within the same patient if the degree of clinical suspicion of endocarditis changes on the basis of new physical findings or laboratory evidence. Our study was not designed to address this issue.
From the results of our study, we recommend that echocardiography be used with greater prudence in patients with suspected endocarditis. It should not be used to make a diagnosis of endocarditis in those with a low clinical probability of the disease. In those with an intermediate or high clinical probability, TTE should be the diagnostic procedure of choice. The use of TEE for diagnosis of endocarditis should be reserved only for patients who have prosthetic valves and in whom TTE is either technically inadequate or indicates an intermediate probability of endocarditis.
This study was supported in part by a grant (R01-HL-48890) from the National Institutes of Health, Bethesda, Md (Dr Kaul). Dr Lindner is the recipient of a Fellowship Training Grant from the Virginia Affiliate of the American Heart Association, Glen Allen. Dr Kaul is an Established Investigator of the National Center of the American Heart Association, Dallas, Tex.
Presented in part at the 44th Annual Scientific Session of the American College of Cardiology, New Orleans, La, March 19-22, 1995.
- Received July 26, 1995.
- Revision received August 28, 1995.
- Accepted September 25, 1995.
- Copyright © 1996 by American Heart Association
Wann LS, Hallam CC, Dillon JC, Weyman AE, Feigenbaum H. Comparison of M-mode and cross-sectional echocardiography in infective endocarditis. Circulation. 1979;60:728-733.
Stewart JA, Silimperi D, Harris P, Wise NK, Fraker TD. Echocardiographic documentation of vegetative lesions in infective endocarditis: clinical implications. Circulation. 1980;61:374-380.
Stafford WJ, Petch J, Radford DG. Vegetations in infective endocarditis: clinical relevance and diagnosis by cross-sectional echocardiography. Br Heart J. 1985;53:310-313.
Erbel R, Rohmann S, Drexler M, Mohr-Kahaly S, Gerharz CD, Iversen S, Oelert H, Meyer J. Improved diagnostic value of echocardiography in patients with infective endocarditis by transoesophageal approach: a prospective study. Eur Heart J. 1988;9:43-53.
Drexler M, Erbel R, Rohmann S, Mohr-Kahaly S, Meyer J. Diagnostic value of two-dimensional transoesophageal versus transthoracic echocardiography in patients with infective endocarditis. Eur Heart J. 1987;8(suppl J):303-306.
Daniel WG, Schroeder E, Nonnast-Daniel B, Lichtlen PR. Conventional and transesophageal echocardiography in the diagnosis of infective endocarditis. Eur Heart J. 1987;8(suppl J):287-292.
Taams MA, Gussenhoven EJ, Bos E, deJaegere P, Roelandt JRTC, Sutherland GR, Bom N. Enhanced morphological diagnosis in infective endocarditis by transoesophageal echocardiography. Br Heart J. 1990;63:109-113.
Pederson WR, Walker M, Olson JD, Gobel F, Lange HW, Daniel JA, Rogers J, Longe T, Kane M, Mooney MR, Goldenberg IF. Value of transesophageal echocardiography as an adjunct to transthoracic echocardiography in evaluation of native and prosthetic valve endocarditis. Chest. 1991;100:351-356.
Daniel WG, Mugge A, Grote J, Hausmann D, Nikutta P, Laas J, Lichtlen PR, Martin RP. Comparison of transthoracic and transesophageal echocardiography for detection of abnormalities of prosthetic and bioprosthetic valves in the mitral and aortic positions. Am J Cardiol. 1993;71:210-215.
von Reyn CF, Levy BS, Arbeit RD, Freidland G, Crumpacker CS. Infective endocarditis: an analysis based on strict case definitions. Ann Intern Med. 1981;94:505-518.
Karalis DG, Bansal RC, Hauck AJ, Ross JJ, Applegate PM, Jutzy KR, Mintz GS, Chandrasekaran K. Transesophageal echocardiographic recognition of subaortic complications in aortic valve endocarditis. Circulation. 1992;86:353-362.
Khandheria BK, Freeman WK, Sinak LJ. Infective endocarditis: evaluation by transesophageal echocardiography. In: Freeman WK, Seward JB, Khandheria BK, Tajik AJ, eds. Transesophageal Echocardiography. Boston, Mass: Little, Brown & Co; 1994:307-337.
Mood AM, Graybill FA. Introduction to the Theory of Statistics. 2nd ed. New York, NY: McGraw Hill; 1963.
Bayes R. An essay toward solving a problem in the doctrine of chance. Philos Trans R Soc Lond. 1763;53:370-418.
Diamond G, Forrester J, Hirsch M, Staniloff HM, Vas R, Berman DS, Swan HJ. Application of conditional probability analysis to the clinical diagnosis of coronary artery disease. J Clin Invest. 1980;65:1210-1221.