The purpose of this study was to analyze the predictive value of different anatomic echocardiographic findings for diagnosis of prosthetic endocarditis.

Prognosis in endocarditis has improved in recent years after the wide acceptance of new clinical diagnostic criteria. One of the most important issues in clinical diagnosis is to use echocardiography for identification of endocardial involvement, but prosthetic material impairs echo quality.

In all, 49 patients with 58 episodes of suggested prosthetic endocarditis were prospectively studied using transthoracic and transesophageal echocardiography. The patients were divided into two groups: group A, patients with 34 episodes of definite endocarditis according to Durack's criteria; and group B, patients with 24 episodes who were eventually classified as not having endocarditis, either by surgical exploration or by a mean of 32.6 months (range: 8-38 months) of follow-up.

In group A, valve dehiscence was observed in 4 episodes of suggested endocarditis, pseudoaneurysms in 3, fistulae in 2, and moderate to severe perivalvular regurgitation in 15. No patient in group B had these abnormalities (P < .001). Vegetations were present in 17 episodes in group A (50%) versus 1 in group B (9%; P < .001); perivalvular abscesses were seen in 19 episodes in group A (56%) versus 1 in group B (P < .001). Mild perivalvular regurgitation was observed in only 1 episode for group A (3%) and in 14 episodes for group B (58%; P < .01). The presence of any of the above echocardiographic finding, when used in combination with the exclusion of mild perivalvular regurgitation, had a positive and negative predictive value for diagnosing endocarditis of 94% and 96%, respectively. Isolated mild perivalvular regurgitation had a poor positive predictive value (6%).

Isolated mild perivalvular regurgitation should not be used as diagnostic criteria in patients with suggested prosthetic valve endocarditis.