Blood culture negative endocarditis: analysis of 63 cases presenting over 25 years.

objectiveTo analyse cases of blood culture negative endocarditis (BCNE) seen at St Thomas' Hospital, London, between 1975 and 2000.

methodsData on all episodes of endocarditis with negative blood cultures seen at St Thomas' Hospital between 1975 and 2000 were collected prospectively and analysed retrospectively.

resultsSixty three patients with BCNE were seen during the study period: 48 (76%) with native and 15 (24%) prosthetic valve infection. BCNE accounted for 12.2% of the 516 cases of endocarditis seen at St Thomas' Hospital. The diagnosis of endocarditis was clinically definite by the Duke criteria in only 21% (7 of 34) of cases of pathologically proven native valve endocarditis but in 62% (21 of 34) of cases by the St Thomas' modifications of the criteria. Comparable figures for the 11 cases of pathologically proven prosthetic valve endocarditis were 45% and 73%. Despite negative blood cultures a causative organism was identified in 31 (49%) of the 63 cases: in 15 by serology (8 Coxiella burnetii, 6 Bartonella species, and 1 Chlamydia psittaci); in 9 cases by culture of the excised valve; in 3 by microscopy of the excised valve, on which large numbers of Gram positive cocci were seen although the culture was sterile; and in the other 4 by isolation from a site other than the excised valve (2 respiratory specimens, 1 from the pacemaker tip, and 1 from an excised embolus). In addition 5 of the 6 cases of Bartonella infection were confirmed by polymerase chain reaction study of the excised valve. Two thirds of the 32 patients for whom no pathogen was identified had received antibiotics before blood was cultured. Thus truly "negative" endocarditis was very uncommon (6% of the cases).

conclusionIf blood cultures are negative in definite or suspected endocarditis, serum should be analysed for Bartonella, Coxiella, and Chlamydia species antibodies, and the excised valve or (rarely) embolus should be analysed by microscopy, culture, histology, and relevant polymerase chain reaction. Other specimens may be relevant. The Duke criteria performed poorly in BCNE; St Thomas' additional minor criteria gave more definite diagnoses.