Current issues of ACP Journal Club are published in Annals of Internal Medicine


Two-dimensional echocardiography has high sensitivity and specificity in detecting perivalvular abscesses

ACP J Club. 1994 Jan-Feb;120:17. doi:10.7326/ACPJC-1994-120-1-017

Source Citation

Aguado JM, González-Vílchez F, Martín-Durán R, Arjona R, Vázquez de Prada JA. Perivalvular abscesses associated with endocarditis. Clinical features and diagnostic accuracy of two-dimensional echocardiography. Chest. 1993 Jul;104:88-93.



To assess the diagnostic accuracy of transthoracic echocardiography in the detection of perivalvular abscesses.


Independent, blinded comparison of transthoracic echocardiographic results with diagnosis of valvular ring abscess at surgery or autopsy.


Tertiary referral center in Spain.


36 patients (mean age 48 y) with infective endocarditis and a histologically proven diagnosis of perivalvular abscess and 20 randomly chosen control patients with infective endocarditis in whom myocardial abscesses were not found at surgery. 25 patients had native valve endocarditis and 11 had prosthetic valve involvement. The aortic valve was the most common site of infection (85% of patients).

Description of test and diagnostic standard

2-dimensional and Doppler echocardiographic studies were stored on videotape for later review and analysis. Abscess was defined by any 1 of the following echocardiographic criteria: a definite region of reduced echo density, excessive thickness images, or echolucent cavities within the valvular annulus, or adjacent myocardial structures in the setting of valvular infection; aorto-left ventricular discontinuity; or a mycotic aneurysm of the aortic root. All echocardiograms were evaluated by 2 independent observers before surgery or autopsy. The diagnostic standard was perivalvular abscess determined at surgery or autopsy. A perivalvular abscess was defined pathologically as a region of necrosis containing purulent material and penetrating into the perivalvular annulus or myocardium with cavity formation or an aortic root mycotic aneurysm. Histologic evidence of necrosis in the annulus without formation of an abscess cavity was not considered to indicate a perivalvular abscess.

Main outcome measures

The sensitivity and specificity of 2-dimensional echocardiography in the detection of myocardial abscesses.

Main results

A perivalvular abscess was observed echocardiographically in 29 of 36 patients with histologically proven perivalvular abscesses. 7 of 11 patients had prosthetic valve abscesses. Of the 20 control patients, 3 had an image suggesting an abscess; all 3 patients had prosthetic valves. Sensitivities, specificities, and likelihood ratios are shown in the Table. 42% of all patients died within 30 days of admission.


2-dimensional transthoracic echocardiography was an accurate method for diagnosing perivalvular abscesses associated with endocarditis.

Source of funding: Not stated.

For article reprint: Dr. J.M. Aguado, Department of Internal Medicine, Hospital Nacional "Marqués de Valdecilla," Universidad de Cantabria, Santander, Spain. FAX 34-42-202-678.

Table. Test characteristics of transthoracic echocardiography for detecting perivalvular abscesses*

Outcomes Sensitivity (95% CI) Specificity (CI) +LR -LR
Perivalvular abscess 81% (64 to 92) 85% (62 to 97) 5.4 0.23
Prosthetic valve abscess 64% (31 to 89) 85% (62 to 97) 4.2 0.43

*LRs defined in Glossary; LRs and CI calculated from data in article.


The 81% sensitivity and the 85% specificity for detecting perivalvular abscesses using 2-dimensional echocardiography in this retrospective study are similar to findings in a previous study (1) that used different diagnostic criteria. A recent prospective study (2) comparing transthoracic and transesophageal echocardiography, however, yielded a low sensitivity for transthoracic imaging (28%) and a higher sensitivity for esophageal imaging (87%). In the current study, sensitivity was 64% (7 of 11) for abscess detection in patients with prosthetic aortic valve endocarditis, considerably less than the 88% (22 of 25) sensitivity for abscess detection in native valves.

The retrospective method used in this study yielded a higher rate of detection (72%, 26 of 36) of perivalvular abscess than did the original interpretation of echocardiograms before surgery. The lower value may reflect more closely the results in clinical practice. Because surgical treatment for endocarditis or an autopsy was required for study entry, the results can be expected to apply only to more advanced abscess formation.

This study helps define the limits of detection for perivalvular abscesses that can be expected with transthoracic echocardiography. Because of the high mortality rate with medical treatment, the more sensitive technique of transesophageal echocardiography is the preferred method for assessing patients with endocarditis, particularly when a prosthetic valve is involved. When only the transthoracic method is available, however, most perivalvular abscesses can be detected if multiple diagnostic criteria are used along with a high index of suspicion.

Gary Burggraf, MD
Hotel Dieu HospitalKingston, Ontario, Canada


1. Ellis SG, Goldstein J, Popp RL. Detection of endocarditis-associated perivalvular abscesses by two-dimensional echocardiography. J Am Coll Cardiol. 1985;5:647-53.

2. Daniel WG, Mugge A, Martin RP, et al. Improvement in the diagnosis of abscesses associated with endocarditis by transesophageal echocardiography. N Engl J Med. 1991;324:795-800.