Emboli in infective endocarditis: the prognostic value of echocardiography

OBJECTIVE: To determine whether vegetations visualized on two-dimensional echocardiography are an independent risk factor for the development of subsequent emboli in patients with infective endocarditis and to assess the timing of emboli relative to the initiation of antimicrobial therapy. DESIGN: Investigator-blinded, retrospective incidence cohort study. SETTING: Tertiary referral center. PATIENTS: Patients with left-sided native valve infective endocarditis who had two-dimensional echocardiography within 72 hours of beginning antimicrobial therapy. MEASUREMENTS AND MAIN RESULTS: The crude incidence rate of first embolic events in patients receiving antimicrobial therapy was 6.2 per 1000 patient-days (95% CI, 4.2 to 9.2). The rates in patients with and without vegetations were 7.1 and 4.9 per 1000 patient-days, respectively (incidence rate ratio, 1.4; 95% CI, 0.6 to 3.3). The relation between vegetations and risk for emboli was microorganism-dependent: Stratified incidence rate ratios were 6.9 (95% CI, 1.1 to 42.5; P less than 0.05) and 1.0 (95% CI, 0.2 to 3.9) for viridans streptococcal and Staphylococcus aureus endocarditis, respectively. The rate of first embolic events diminished over time (P less than 0.001), falling from 13 per 1000 patient-days during the first week of therapy to less than 1.2 per 1000 patient-days after completion of the second week of therapy. CONCLUSIONS: Overall, the presence of vegetations on echocardiography was not associated with a significantly higher risk for embolus in patients with left-sided native valve infective endocarditis. The relative risk for embolic events associated with echocardiographically visualized vegetations may be microorganism-dependent, with a significantly increased risk seen only in patients with viridans streptococcal infection. The rate of embolic events declines with time after initiation of antimicrobial treatment.     

Comparison of two-dimensional and M-mode echocardiography in the evaluation of patients with infective endocarditis.

M-mode and two-dimensional echocardiographic evaluation of infectious endocarditis and its complications was reviewed. In 21 consecutive patients with clinical endocarditis, 22 valves were involved (12 aortic, 5 mitral and 5 tricuspid). M-mode echocardiography detected vegetations in 10 patients (four aortic, two mitral and four tricuspid) and detected complications of endocarditis in 2 patients (one aortic root abscess and one flail aortic cusp). Two-dimensional echocardiography detected vegetations in 9 patients (four aortic, one mitral and four tricuspid) and detected complications in ten patients (five flail aortic cusps, one aortic root abscess, one sinus on Valsalva aneurysm, two flail mitral leaflets and one flail tricuspid valve). Thus, although M-mode and two-dimensional echocardiography had a similar ability to detect actual vegetations, two-dimensional echocardiography was superior to M-mode echocardiography in diagnosing complications of the destructive process.     

The demonstration of vegetations by echocardiography in bacterial endocarditis: An indication for early surgical intervention

The visualization of vegetations by M-mode echocardiography in patients with infective endocarditis has been suggested to imply a poor prognosis regarding the development of major systemic emboli, congestive heart failure and the need for early surgical intervention. The question of using the finding of vegetations by echocardiography as an indication for surgery is controversial. To answer this question, 30 patients with the clinical diagnosis of endocarditis were studied by echocardiography. In 17 of the 30 (57 per cent) vegetations were present (aortic eight, mitral four, both mitral and aortic five), whereas in 13 (43 per cent) no vegetations were visualized. Infecting organisms were similar in each group; Streptococcus viridans being the most common. The patients with echocardiographically demonstrable vegetations had a higher incidence of congestive heart failure compared to the patients without (14 of 17 versus six of 13, p < 0.05), major emboli (eight of 17 versus two of 13, p = NS) and need for valve surgery (17 of 17 versus two of 13, p < 0.001). Mortality was not significantly different in the two groups (six of 17 versus three of 13, p = NS). Urgent or emergency surgery was required in 16 of 17 patients with vegetations. Thus, the demonstration of vegetations by echocardiography i dentified a subset of patients with more severe disease in whom early operative intervention was required.     

Echocardiographic and clinical correlates in drug addicts with infective endocarditis. Implications of vegetation size

Thirty-four drug addicts with endocarditis were studied to evaluate the prognostic significance of vegetation size and its short-term changes, as determined by two-dimensional echocardiography. Among 43 episodes of endocarditis, vegetations were detected in 27 (63%), confined to the tricuspid valve in 20 patients, mitral valve in one, aortic valve in two, and both tricuspid and mitral valves in four. All vegetations were large (greater than or equal to 1 cm) (mean maximal dimension, 1.7 +/- 0.5 cm). Medical cure was achieved in all 16 patients without vegetations and in 18 (90%) of 20 patients with tricuspid valve vegetations. One patient with tricuspid vegetation and polymicrobial infection died of respiratory failure. Surgery was required for one patient with tricuspid vegetation, all three patients with isolated left-sided endocarditis, and two of four patients with multivalve involvement. Short-term changes of tricuspid valve vegetations during therapy (one to eight weeks) did not correlate with clinical outcome. Although large tricuspid vegetations may occasionally identify a subset at risk for complications, most patients with isolated tricuspid valve endocarditis have a benign prognosis.     

Serious infections caused by Streptococcus milleri

Viridans streptococci continue to be the most common etiologic agents in bacterial endocarditis. The spectrum of diseases other than endocarditis caused by these organisms, however, has received sparse attention. Moreover, little clinical information is available concerning the individual viridans streptococcal species. During the 16 months of this study, one species of viridans streptococci, Streptococcus milleri, caused a surprising number of clinically significant suppurative infections. Clinical syndromes included the following: abscesses, 10 cases; peritonitis, four cases; endocarditis, three cases; cholangitis, empyema and cellulitis, one case each. A gastrointestinal source was apparent in eight of these 20 cases. Review of 58 cases of viridans streptococcal endocarditis demonstrated that although Strep. milleri was an infrequent cause of endocarditis, this species was associated with an unusual frequency of suppurative complications. Our experience suggests that Strep. milleri may be more commonly associated with serious suppurative infections than other viridans streptococci and that species identification may be a clinically useful procedure.     

Enhanced morphological diagnosis in infective endocarditis by transoesophageal echocardiography.

Thirty three consecutive patients with clinically suspected endocarditis were studied by both precordial cross sectional echocardiography and transoesophageal echocardiography. The diagnostic value of both techniques was assessed. The data were compared with findings at operation in 25 patients. In 21 patients with native valve endocarditis precordial echocardiography showed evidence of vegetations in six patients and suggested their presence in nine. Transoesophageal echocardiography identified vegetations in 18 patients. Complications were seen in four patients at precordial echocardiography and in nine patients at transoesophageal echocardiography. Precordial echocardiography did not show vegetations in any of the 12 patients with prosthetic valve endocarditis whereas transoesophageal echocardiography showed vegetations in four. Complications were seen in four patients at precordial echocardiography and in 10 at transoesophageal echocardiography. Echocardiographic findings were confirmed at operation in all 25 operated patients. In two patients both echocardiographic techniques had missed the perforation of the cusps of the aortic valve that was seen at operation, but this had no effect on patient management. Transoesophageal echocardiography is the best diagnostic approach when infective endocarditis is suspected in patients with either native or prosthetic valves.     

Value of echocardiography in detection of vegetation and their clinical significance

The aim of the study was to assess usefulness of echocardiography in bacterial vegetations detection as well as their clinical value as the indicator for surgical treatment. 44 patients aged 16-65 (mean 37.6) with infective mitral and aortic valve endocarditis underwent the study. Authors assessed clinical state taking into consideration blood culture tests as well as M-mode and parasternal and apical projections two-dimensional echocardiographic examinations. Data were compared with intraoperative or pathomorphological findings to estimate specificity and sensitivity of echocardiography in bacterial vegetation detection. Vegetations were pathomorphologically or intraoperatively stated in 21 patients (48%). M-mode echocardiography revealed changes in 16 patients, and two-dimensional one in the next 3. In the group of 23 patients without vegetations, concordance between intraoperative findings and echocardiographic results was stated in 19 subjects. Therefore, sensitivity and specificity of two-dimensional echocardiography were respectively 91% and 83%. Vegetations stated in two-dimensional echocardiographic examination had an unfavourable prognosis. Embolic complications were observed in 14, and myocardial infarction in 7 of 21 patients with bacterial vegetations. Authors thought it advisable to early operate on such patients. Whereas in patients without vegetations embolic complications were stated only in 3, and myocardial infarction in 1 patient. Therefore two-dimensional echocardiography making bacterial vegetations detection possible in patients with infective valve endocarditis allows to identify patients with higher risk of thromboembolic complication or death.     

The clinical role of transoesophageal echocardiography

The role of transoesophageal echocardiography (TEE) was evaluated in a consecutive series of 100 procedures performed in 86 patients (age 17–81, mean 56 years). All patients had prior transthoracic echocardiography (TTE). TEE was performed with a 5 MHz phased array transoesophageal transducer with pulsed wave Doppler and colour flow mapping capability. Forty-four per cent of patients received intravenous sedation and 36% received antibiotic prophylaxis. There were no complications of TEE. The TTE and TEE findings were compared. In patients referred for possible cardiac source of embolism, left atrial thrombi were detected in 8/27 TEE studies but in none of 27 TTE studies. In 12 patients with prosthetic valve dysfunction TEE distinguished prosthetic from periprosthetic regurgitation in 9/12 studies compared to 3/12 with TTE. In 11 patients with suspected aortic dissection TEE correctly detected dissection in all seven cases in which the diagnosis was subsequently confirmed, whereas TTE showed only equivocal findings in two cases. Vegetations were detected by TEE in 4/5 studies in patients with proven native valve endocarditis and by TTEin 2/5. No vegetations were detected by TTE or TEE in five studies in patients with proven prosthetic valve endocarditis. Compared with other investigations there were no false positive TEE studies and one possible false negative study. We conclude that TEE is a safeprocedure which often provides additional clinical information to transthoracic echocardiography.     

Infective endocarditis in adults in Glasgow, 1976-81

Seventy-one cases of endocarditis in adults, defined on strict clinical criteria over a period of 5 years were studied. The demographic profile was similar to other published studies except that no bimodal peak in age distribution was detected. The major presenting features were infection, cardiac failure and embolism. Rheumatic valvular disease (44%) was the most common predisposing condition. Twenty percent of our patients developed endocarditis on an apparently normal valve and 28% presented with prosthetic valve infection. M-mode echocardiography was carried out in only 44% of cases and with a low rate of detection of vegetations. Streptococcus viridans was the most common causal organism. The mortality for the total series (45%) was considerably greater than that in other recent reports. Our study confirmed that failure to show early response to appropriate antimicrobial therapy was an indication to consider immediate valve replacement.     

Spectrum of echocardiographic findings in bacterial endocarditis.

Forty-seven echocardiograms were obtained in 32 patients with bacterial endocarditis. Preexistent abnormalities were found in 14 patients. In five of them thought to have bacterial endocarditis on normal valves, echocardiography showed mitral stenosis (one), bicuspid aortic valve (two), and prolapse of mitral valve (two). Definite vegetations were seen in 22 patients--on the aortic valve in seven, the mitral valve in 12, and both valves in three. Ten patients had milder changes suggestive but not diagnostic of vegetations. In 12 patients, surgery confirmed the echocardiographic findings. Fourteen had systemic embolic episodes and all had echocardiographic evidence of vegetations. Abnormalities secondary to bacterial endocarditis, other than vegetations, were common. Twenty-one patients had left ventricular volume overload. Ten had a flail posterior leaflet of the mitral valve, three of which were confirmed surgically. Eight had abnormal coarsely fluttering echoes in the left ventricular outflow tract consistent with a prolapsing aortic valve or underlying aortic vegetations; four were confirmed by surgery. Five had signs of severe aortic regurgitation of recent onset (premature mitral valve closure) and all had confirmation by surgery. Echocardiographic abnormalities persisted after successful medical treatment. We conclude that echocardiography is helpful in patients with bacterial endocarditis. It permits recognition of unsuspected preexistent lesions and the characteristic vegetations, as well as the extent and nature of valvular damage secondary to bacterial endocarditis. However, echocardiography does not differentiate between active and healed lesions.     

Echocardiographic assessment of patients with infectious endocarditis: prediction of risk for complications

To enhance the echocardiographic identification of high risk lesions in patients with infectious endocarditis, the medical records and two-dimensional echocardiograms of 204 patients with this condition were analyzed. The occurrence of specific clinical complications was recorded and vegetations were assessed with respect to predetermined morphologic characteristics. The overall complication rates were roughly equivalent for patients with mitral (53%), aortic (62%), tricuspid (77%) and prosthetic valve (61%) vegetations, as well as for those with nonspecific valvular changes but no discrete vegetations (57%), although the distribution of specific complications varied considerably among these groups. There were significantly fewer complications in patients without discernible valvular abnormalities (27%). In native left-sided valve endocarditis, vegetation size, extent, mobility and consistency were all found to be significant univariate predictors of complications. In multivariate analysis, vegetation size, extent and mobility emerged as optimal predictors and an echocardiographic score based on these factors predicted the occurrence of complications with 70% sensitivity and 92% specificity in mitral valve endocarditis and with 76% sensitivity and 62% specificity in aortic valve endocarditis.     

Predictive risk factors for periannular extension of native valve endocarditis. Clinical and echocardiographic analyses

The study objective is to identify clinical, microbiologic, and/or echocardiographic risk factors present early in the course of native valve endocarditis that predict subsequent development of periannular extension of infection. A multivariate computer-generated analysis of 21 clinical-microbiologic parameters and 11 two-dimensional echocardiographic parameters in patients with native valve endocarditis was designed. These parameters were statistically compared in operated-on patients with native valve endocarditis with and without periannular extension of infection. The study took place in a 600-bed acute-care, nonreferral, municipal hospital primarily servicing an indigent patient population. Seventy-three documented episodes of native valve endocarditis occurred between the years of 1973 and 1987, including 29 operated-on patients with surgically confirmed periannular extension of infection and 44 operated-on patients without periannular extension of infection. Multivariate logistic-regression analyses of multiple clinical, microbiologic, and echocardiographic parameters which are potentially predictive of eventual periannular extension of native valve endocarditis were carried out. The only two independent parameters that significantly predicted periannular infection among patients with native valve endocarditis were (1) aortic valve involvement and (2) abuse of intravenous (IV) drugs (p less than 0.01; p less than 0.01, respectively, multivariate analysis). The relative risk of developing periannular extension of endocarditis among patients with aortic valve involvement and/or IV drug abuse was increased by approximately 2.5-fold compared with patients without these characteristics. Factors not significantly associated with increased risk of periannular extension of native valve endocarditis included the following: prolonged febrile morbidity; Staphylococcus aureus etiology; or two-dimensional echocardiographic demonstration of vegetations, large vegetations (greater than or equal to 1 cm), multiple vegetations, or enlargement of aortic root or annulus. These data suggest that patients with native aortic valve endocarditis, particularly in the setting of IV drug abuse, should be considered for routine, serial noninvasive evaluation for the early detection of periannular extension of their infection.     

Echocardiographic identification of an aortic valve ring abscess

A valve ring abscess was diagnosed in four patients with a prosthetic aortic valve by identifying an echo-free space on two-dimensional echocardiography. Three of the patients presented with severe aortic regurgitation and congestive heart failure after an episode of endocarditis, but two of them did not have evidence of active endocarditis. The fourth patient had endocarditis, but no evidence of aortic regurgitation or heart failure. All four patients required valve replacement. Similar findings in all 11 previously reported cases suggest that a valve ring abscess can be diagnosed by two-dimensional echocardiography. It may be found without clinical evidence of endocarditis, in the absence of aortic regurgitation, without echocardiographically identifiable vegetations or during resolution of endocarditis.     

Clinical outcome and echocardiographic findings of native and prosthetic valve endocarditis in the 1990s

Prosthetic valve endocarditis is considered to be associatedwith a more severe prognosis than native valve endocarditis.Among other factors, inappropriate visualization of vegetationsin prosthetic valve endocarditis by transthoracic echocardiographyis responsible for this observation. Since the introductionof transoesophageal echocardiography into clinical practicethe diagnostic sensitivity and specificity of the detectionof vegetations located on prosthetic valves have been enhanced.Therefore we aimed to determine and compare the prognosis ofprosthetic valve endocarditis and native valve endocarditisin the era of this improved diagnostic approach. One hundred and six episodes of infective endocarditis in 104patients were seen at our institution between 1989 and 1993.Eighty patients (77%) had native valve endocarditis and 24 (23%)had late prosthetic valve endocarditis. In the latter grouptwo patients had recurrent infective endocarditis. Patientswith prosthetic valve endocarditis were older (mean age 64 vs54 years in native valve endocarditis; P<0.00l) and the majoritywas female (62% vs 38% in native valve endocarditis; P<0.001In prosthetic valve endocarditis, infection of a valve in themitral position predominated (65% vs 30% in native valve endocarditis;P<0.0l), whereas in native valve endocarditis more than halfthe cases had isolated aortic valve endocarditis (51% vs 27%in prosthetic valve endocarditis; P<0.01). In prostheticvalve endocarditis more cases were caused by Staphylococcusaureus (31% vs 14% in native valve endocarditis; P<0.08),whereas in native valve endocarditis the most frequent organismswere streptococci (29% vs l9% in prosthetic valve endocarditis;P<0.12). Differences in the clinical features of native valveendocarditis and prosthetic valve endocarditis could not befound except for a higher rate of embolism in native valve endocarditis(40% vs l9% in prosthetic valve endocarditis; P<0.05). Vegetationscould be detected by transthoracic echocardiography more frequentlyin native valve endocarditis (71% vs 15% in prosthetic valveendocarditis; P<0.0001). Transoesophageal echocardiographyvisualized vegetations in 95% of the episodes of native valveendocarditis and in 80% of the episodes of prosthetic valveendocarditis (P<0.09). Thus, the diagnostic gain by transoesophagealechocardiography was greatest in prosthetic valve endocarditis.Patients with native valve endocarditis had significantly largervegetations than patients with prosthetic valve endocarditis(P<0.05 for length, P<0.00l for width). The median timeto diagnosis was similar in native valve endocarditis and prostheticvalve endocarditis (31 vs 28 days). Surgery was performed in 74% of patients with native valve endocarditisand in 58% of those with prosthetic valve endocarditis; themedian time delay between the diagnosis of infective endocarditisand surgery tended to be shorter in prosthetic valve endocarditisthan in native valve endocarditis (45 vs 60 days). The in-hospitalmortality and the mortality during a follow-up of 22±10 months did not significantly differ between native valveendocarditis and prosthetic valve endocarditis (21% vs 17% 28%vs 25%). In summary in the era of transoesophageal echocardiography,late prosthetic valve endocarditis does not seem to carry aworse prognosis than native valve endocarditis. This can beattributed in part to the improved diagnostic accuracy achievedby transoesophageal echocardiography leading to comparable diagnosticlatency periods in both patient groups. Finally, better characterizationof vegetations on prosthetic valves by transoesophageal echocardiographyallows early lifesaving surgery in patients with prostheticvalve endocarditis.     

Native valve endocarditis due to Streptococcus vestibularis and Streptococcus oralis

Viridans streptococci are the commonest cause of native valve infective endocarditis (IE). The taxonomy of this group is evolving allowing new disease associations to be made. Streptococcus vestibularis is a recently described member of the viridans group, first isolated from the vestibular mucosa of the human oral cavity. It has rarely been associated with human disease. Streptococcus oralis, another member of the viridans group resident in the human oral cavity is a well known cause of IE and bacteraemia in neutropenic patients. We report the first case of native mitral valve endocarditis due to S. vestibularis in a patient with co-existing S. oralis endocarditis.     

Prognostic value of echocardiographic finding of valvular vegetations in patients with infectious endocarditis

42 consecutive patients with infective endocarditis on native valves, according to Pelletier and Petersdorf's criteria of definite (13 pts), probable (12 pts.) and possible (17 pts) endocarditis, were identified and prospectively followed-up with M-mode and two-dimensional echocardiography, since 1980. We compared: 1) these three groups; 2) survivors not referred for surgery versus surgical patients plus nonsurvivors; 3) patients who suffered embolic events versus those who did not; 4) patients with severe-moderate heart failure versus those with no failure or mild failure; 5) patients with aortic valve echocardiographic vegetations versus those with mitral valve vegetations. Furthermore 11 of these patients who did not undergo surgery (9 with mitral and 2 with mitro-aortic vegetations on echo) were serially followed-up with echocardiography for 6-42 months (average: 32 months). The presence of ultrasound detectable vegetations itself and their size, without considering their site, did not identify a major risk of embolization, heart failure, death or need of surgery. The site of vegetations was the only significant feature in our series. It identified a high-risk group and a relatively low-risk group. Aortic valve involvement, with echocardiographic vegetations, was related to severe or moderate heart failure (P less than 0.01), death or need of surgery (P less than 0.05). Mitral valve involvement carried on a relatively low risk. The 9 patients with mitral valve vegetations only, not referred for surgery and followed-up, did well on medical treatment and returned to work. They did not have relapses or embolization. On serial echocardiographic examinations, mitral vegetations become smaller in the long run. Two years after the acute episode, usually echocardiography did not allow identification of vegetations.     

Isolated native tricuspid valve endocarditis caused by viridans streptococcus

The present report describes a case of native tricuspid valve endocarditis caused by viridans group streptococcus in a 43-year-old man who had recently undergone dental extraction. The patient had no history of intravenous drug use, heart disease or right heart catheterization. Although there have been scattered reports of unusual organisms, to the authors'' knowledge, this is the first case of viridans group streptococcal endocarditis involving only the tricuspid valve after dental manipulation.Key Words: Tricuspid valve endocarditis, Viridans streptococcusIsolated native tricuspid valve endocarditis (TVE) accounts for only 5% to 10% of all cases of infective endocarditis (1,2) and is rarely seen in the absence of intravenous drug use (IDU) or intracardiac catheterization or cardiac anomalies (3,4). The present report describes a case of isolated viridans streptococcal TVE after dental manipulation in a nonintravenous drug user without underlying cardiac disease or central venous catheterization. We are not aware of any cases reported in the English literature of isolated native TVE caused by viridans streptococcal species after dental procedures. The expedient identification of patients with TVE in the absence of predisposing factors can be difficult, and thus delay in diagnosis may delay appropriate therapeutic intervention.     

Reliability and clinical relevance of detection of vegetations by echocardiography in bacterial endocarditis.

The efficacy of M-mode echocardiography in identifying vegetations and the clinical relevance of such a finding were evaluated in 36 patients with bacterial endocarditis, 18 of whom had pre-existing valvar lesions. Of 22 patients with vegetations demonstrated by echocardiography, 21 had major complications of bacterial endocarditis and eight died, whereas only eight of the 14 patients without detectable vegetations had similar complications and only two died. Nine of the 11 patients with pre-existing valvar lesions and echocardiographic evidence of vegetations came to operation or necropsy, and in eight of these the presence of vegetations was confirmed. In the other, none was found at surgery performed after an interval of six months. M-mode echocardiography can reliably detect vegetations in patients with bacterial endocarditis even in the presence of pre-existing valvar lesions, and may permit the identification of a subset of high risk patients who may need early surgery.     

Spondylodiscitis and endocarditis caused by S. vestibularis

Streptococcus vestibularis is a recently described member of the viridans group that was first isolated from the vestibular mucosa of the human oral cavity and described as a new species in 1988. It has been rarely associated with human infections. In few papers, it has been reported as a causal agent of systemic infection in immunosupressed adults and in those with other severe underlying diseases, like coronary valve diseases. A 65-year-old woman was admitted to the hospital with complaints of fever for three months, general malaise, effort dyspnea, weight loss, back pain and myalgia. Both native aortic valve endocarditis and spondylodiscitis due to Streptococcus vestibularis were detected. The patient was successfully treated with intravenous potassium penicillin G and gentamicin for six weeks, followed by oral amoxicillin for three months, in addition to aortic valve replacement. In all patients with spondylodiscitis, infective endocarditis should be considered, particularly in patients with heart valve disease history, since spondylodiscitis may be the presenting sign of an infective endocarditis. Cardiac valve replacement surgery should be performed if the course of fever and inflammatory syndrome is unfavorable after appropriate antibiotic treatment. We report the first case with both native aortic valve endocarditis and spondylodiscitis due to Streptococcus vestibularis.     

Streptococcal meningitis in adult patients: current epidemiology and clinical spectrum.

Streptococci other than Streptococcus pneumoniae are a rare cause of bacterial meningitis in adults. We report 29 cases of streptococcal meningitis (1977-1997). The patients comprised 19 men and 10 women, with a mean age +/- standard deviation of 47 +/- 18 years. Nine cases were secondary to neurosurgical procedures, seven to brain abscess, five to cerebrospinal fluid pericranial fistula, and three to endocarditis. Causative microorganisms included the following: viridans group streptococci, 20 cases; anaerobic streptococci, 3; Streptococcus agalactiae, 3; Streptococcus bovis, 2; and Streptococcus pyogenes, 1. Four Streptococcus mitis strains showed decreased susceptibility to penicillin (MIC, 0.5-2 microg/mL). Five patients (17%) died. The infection is increasing in the hospital setting. Streptococci resistant to penicillin should be considered in the empirical treatment of nosocomial meningitis. In cases of community-acquired infection, anaerobic streptococci or streptococci of the Streptococcus milleri group should alert the clinician to the presence of an undiagnosed brain abscess, whereas oral streptococci of the viridans group suggest the diagnosis of bacterial endocarditis.