Endocarditis-associated paravalvular abscesses. Do clinical parameters predict the presence of abscess?

STUDY
OBJECTIVE: To determine whether standard clinical and transthoracic echocardiographic criteria considered to be suggestive of the presence of endocarditis-associated paravalvular abscess are predictive of which patients would benefit from reliable but invasive transesophageal echocardiographic investigations for abscess.
DESIGN: Retrospective chart review.
SETTING: A 630-bed university hospital. PATIENTS: Forty-eight patients with 51 episodes of definite endocarditis and 24 paravalvular abscesses. MEASUREMENTS AND
RESULTS: A comparison of abscess and nonabscess populations revealed that clinical parameters (patient demographics, valvular involvement, presence of a prosthesis, infection with a virulent organism, pericarditis, persistent fever, persistent bacteremia, congestive heart failure, history of intravenous drug use, embolization) and transthoracic echocardiographic parameters were insensitive predictors of the presence of abscess. The only statistically significant correlate was the presence of previously undetected atrioventricular or bundle branch block. Paravalvular abscesses were common in our population and were associated with increased mortality. Improved survival correlated with the absence of mitral valve involvement and the absence of moderate-to-severe congestive heart failure.
CONCLUSIONS: Given the accuracy and safety of transesophageal echocardiography and the unreliability of clinical and transthoracic echocardiographic criteria, we recommend that transesophageal echocardiography be considered in all endocarditis patients with previously unrecognized conduction disturbances, aortic or prosthetic valve involvement, or both, or indications for valve replacement, or all of the foregoing.