Mortality from infective endocarditis: clinical predictors of outcome.

OBJECTIVE To identify clinical markers available within the first 48 hours of admission that are associated with poor outcome in infective endocarditis. DESIGNS Retrospective cohort study. SETTING Teaching hospital. PATIENTS 208 of 220 patients with infective endocarditis. METHODS Consecutive patients with infective endocarditis presenting between 1981 and 1999 to a tertiary centre were studied. Clinical, echocardiographic, and haematological data recorded within 48 hours of admission were obtained. Data were analysed using logistic regression models. MAIN OUTCOMES MEASURES Mortality at discharge and at six months. RESULTS Data were obtained for 93% of patients who were eligible for inclusion. 194 (93%) were positive for Duke criteria. Mean age was 52 (1.2) years, and 138 (66%) were men. 82 (39%) were transferred from other hospitals. 181 (87%) were blood culture positive, and 47 (23%) infections were Staphylococcus aureus. The infection was located on aortic (n = 85, 41%), mitral (n = 77, 37%), tricuspid (n = 18, 9%), and multiple valves (n = 20, 10%). 67 (32%) had prosthetic valve endocarditis. 48% of the cohort were managed with antibiotics alone. Mortality at discharge was 18% and at six months 27%. Duration of illness before admission, age, sex, valve infected, infecting organism, and left ventricular function were not predictors of adverse mortality. However, abnormal white cell count, serum albumin concentration, serum creatinine concentration, or cardiac rhythm, the presence of two major Duke criteria, or visible vegetation conferred a poor prognosis. CONCLUSIONS Conventional prognostic factors in this study did not appear to predict outcome early during hospital admission. However, simple clinical indices, which are readily available, are reliable, cheap, and potentially powerful predictors of poor outcome.