Lefort A, Lortholary O, Casassus P, Selton-Suty C, Guillevin L, Mainardi J, for the β-Hemolytic Streptococci Infective Endocarditis Study Group. Comparison Between Adult Endocarditis Due to β-Hemolytic Streptococci (Serogroups A, B, C, and G) and Streptococcus milleriA Multicenter Study in France. Arch Intern Med. 2002;162(21):2450-2456. doi:10.1001/archinte.162.21.2450
Limited data exist on infective endocarditis (IE) due to Streptococcus milleri (Sm) or β-hemolytic streptococci (BHS). Because BHS and Sm share some physiologic and pathogenic properties, we wondered whether IE caused by these streptococci might present similarities.
Through a nationwide retrospective study in France, the medical and microbiologic charts of adults with definite Sm or BHS IE observed between January 1, 1991, and December 31, 1996, in university and general hospitals were reviewed.
Fifty-six patients had BHS IE (33 men and 23 women; median age, 59 years; range, 18-91 years) and 29 had Sm IE (18 men and 11 women; median age, 66 years; range, 37-97 years). The BHS group tended to have more underlying medical conditions (46%) and fewer previous cardiopathies (48%) than the Sm group (28% and 71%, respectively). Multivariate analysis showed that a longer median time of IE evolution before diagnosis and identification of the presumed portal of entry were independently associated with Sm IE (P<.01). Extracardiac complications were more frequent in the BHS group (55%) than in the Sm group (39%). Comparison of echocardiographic findings between the 2 groups did not establish any statistically significant difference. Thirty-six patients (64%) in the BHS group and 18 (62%) in the Sm group underwent valve replacement. Overall, mortality was 27% for the BHS group and 14% for the Sm group. For all patients, multivariate analysis retained older age as the only significant risk factor for death (P = .01).
Compared with Sm IE, BHS IE occurs in younger patients with more underlying diseases and fewer underlying cardiopathies and has a more aggressive presentation and evolution.
INDIVIDUALIZATION OF the Streptococcus milleri (Sm) group is difficult in clinical research because of the diversity of hemolytic and Lancefield serologic reactions and the controversial classification of this group of pathogens.1,2 According to the current taxonomy, the Sm group includes 3 species, Streptococcus anginosus, Streptococcus intermedius, and Streptococcus constellatus.1,3 Most Sm isolates are nonhemolytic and cannot be assigned to a group. However, many are β-hemolytic and sometimes carry Lancefield group F or, less frequently, A, C, or G antigens.1,4 Clinically, as for all groups of β-hemolytic streptococci (BHS), Sm are characterized by a propensity for invasive pyogenic infections, often complicated by abscess formation, which differentiates them from viridans streptococci.5,6 Severe invasive BHS infections have been increasingly observed in recent years.7,8 Infective endocarditis (IE) due to BHS or Sm is rare, accounting for 5% to 7% of these infections.9- 11 Because only isolated cases or old small series of BHS IE have been reported in the antibiotic era,7,12- 19 and no previous series focusing on Sm IE can be found in the literature, to our knowledge, little is known about the current epidemiologic and clinical characteristics of IE due to these pathogens in adults. Because BHS and Sm share some physiologic and pathogenic properties, we wondered whether IE caused by these streptococci might have common characteristics. Through a retrospective national survey, we analyzed and compared the epidemiologic, clinical, microbiologic, and echocardiographic features and the outcomes of BHS (groups A, B, C, and G) and Sm IE for a large cohort of adults followed in France between January 1, 1991, and December 31, 1996.
Through the Association pour l'Étude et la Prévention de l'Endocardite Infectieuse, we conducted a nationwide retrospective survey of IE due to BHS of serogroups A (GAS), B (GBS), C (GCS), and G (GGS) or Sm in adults between January 1, 1991, and December 31, 1996. Infectious disease, cardiology, and internal medicine units of 53 university and 220 general hospitals were asked by letter to report all cases they followed during the study period. Data were then cross-checked with those of the corresponding microbiology laboratories. Seventy-nine centers (all 53 university hospitals and 26 general hospitals) responded. Twenty-nine centers (23 university hospitals and 6 general hospitals) reported at least 1 case of BHS or Sm IE, for a total of 89 cases (59 BHS and 30 Sm) during the study period.
Anonymous clinical and microbiologic questionnaires were sent to each of the reporting centers. Information was collected by the referring physician or microbiologist of the reporting center or by one of the principal study investigators. The questionnaires requested the following minimum data for each patient: sex, age, previous cardiopathy and type, underlying predisposing condition(s), suspected portal of entry of BHS or Sm, native or prosthetic valve(s) involved, duration of evolution before diagnosis, presence of fever, results of cardiac auscultation, extracardiac symptoms, embolic events (type and date of occurrence), echocardiographic (transthoracic [TTE] and transesophageal [TEE]) findings, doses and durations of antibiotic treatments, cardiac surgery (type, time, and indications), characteristics of the valve (surgeon's observations and histologic report), and outcome. The indication for surgery was considered to be infectious when the infectious syndrome persisted despite antibiotic treatment, cardiac in the case of hemodynamic failure or major valve damage, or embolic for patients who had recurrent emboli or who were at high risk for developing emboli.20 The presumed portal of entry retained was that designated by the patient's treating physician. It was considered to be dental in the case of recent tooth extraction or an infectious buccal focus. During surgery or at autopsy (by direct inspection and histopathologic examination), an intracardiac abscess was defined as a region of necrosis containing purulent material and penetrating into the valvular annulus or myocardium. An abscess was considered to be present on the echocardiogram when a definite region of hypodensity or echolucent cavities within the valvular annulus or adjacent myocardial structures were found.21
All the collected data were reviewed by one microbiologist and one infectious disease specialist (O.L., J.-L.M.). Of the 89 IE cases, only those considered definite according to the Duke criteria22 were retained for analysis, leaving a total of 85 cases (56 BHS and 29 Sm). The day of IE diagnosis was considered to be the day when echocardiographic findings were suggestive of endocarditis.
Streptococcus milleri and BHS were isolated and identified according to standard procedures.1 Classification was based on the appearance of the colonies, hemolysis pattern, Lancefield serologic reactions, growth properties, and biochemical reactions. Lancefield serologic grouping (A, B, C, or G) was carried out using a commercial kit (Streptococcal Grouping Kit; Oxoid Limited, Basingstoke, England). The Sm isolates were also recognized by the caramel-like odor associated with their cultures and by using the Api 20 Strep System (bioMérieux, Marcy-l'Etoile, France). Subspecies within the Sm and BHS groups were identified using the Rapid ID 32 Strep System (bioMérieux). Antibiotic susceptibility was tested using the disk agar diffusion method.23 Macrodilution or microdilution methods or the E-test was used to determine penicillin G or amoxicillin minimum inhibitory concentrations. The critical values for susceptibility and resistance were 0.25 mg/L or less and greater than 16 mg/L for penicillin, and 4 mg/L or less and greater than 16 mg/L for amoxicillin.23
Statistical analyses were performed using a software program (StatView II; Abacus Concepts Inc, Berkeley, Calif). For univariate analysis, tests were conducted using the usual parametric methods. The χ2 test with Yates correction for small numbers, when appropriate, was used to compare qualitative variables, and the t test (2-tailed) was applied to compare quantitative variables, expressed as mean ± SD. The following data were selected for comparison between the BHS and Sm groups and between the GBS and other BHS groups (ie, GAS, GCS, and GGS): age, sex, previous cardiopathy, involvement of a prosthetic valve, chronic alcoholism, diabetes mellitus, underlying malignancy, identification of a portal of entry, valve(s) involved, duration of evolution before diagnosis, congestive heart failure, embolic and septic events, meningitis, valve vegetation(s), valvular insufficiency and TEE-detected intracardiac abscess, surgery and its indication, use of antibiotic monotherapy, combination therapy including an aminoglycoside, treatment including a β-lactam antibiotic, duration of antibiotic treatment, and mortality. For analysis of risk factors associated with dying, the following items were considered: age, microorganism responsible for infection, duration of evolution before diagnosis, previous cardiopathy, valve(s) involved, use of antibiotic monotherapy, and surgery and its indication. The patient's status (alive or dead) 1 month after IE diagnosis was recorded for analysis of prognostic factors. Logistic regression analysis was undertaken to identify the variables independently associated with death or with Sm infection, using those that had been retained after univariate analysis. Statistical significance was defined as P<.05.
The BHS isolates responsible for the 56 IE cases were distributed as follows: 7 GAS, 34 GBS, 3 GCS (including 2 Streptococcus zooepidemicus and 1 Streptococcus equisimilis), and 12 GGS. Of the 29 cases of Sm IE, 12 were due to S anginosus, 2 to S constellatus, and 3 to S intermedius; the species was not determined for the remaining 12 cases.
The overall characteristics of patients in the 2 groups did not differ substantially (Table 1). However, more patients in the Sm group (55%) were 65 years and older than in the BHS group (34%). There was a tendency toward a higher rate of underlying medical diseases (46%) and a lower rate of underlying heart diseases (48%) in the BHS group than in the Sm group (28% and 71%, respectively). The percentages of patients in whom no cardiac and general predisposing conditions could be demonstrated were similar in both groups. No patient was an intravenous drug addict or was infected with the human immunodeficiency virus. None of the 34 GBS-infected patients had IE during or after pregnancy. In 11 of 13 patients with prosthetic heart valve endocarditis, the median time since valve replacement was 7 years (range, 6 weeks to 16 years); the interval was not known for the remaining 2 patients.
All infections were community acquired. Most patients (92% of the BHS group and 77% of the Sm group) had acute presentations, with a duration of symptoms before diagnosis of 6 weeks or less. The median time of disease evolution before diagnosis was statistically significantly shorter for BHS IE than for Sm IE (11.0 and 20.5 days, respectively; P<.01) (Table 2). The complication rate was slightly higher for the BHS group than for the Sm group (55% and 39%, respectively; P>.05). In particular, the frequencies of arterial emboli, meningitis, mycotic aneurysms, and other focal events were higher for the BHS group. A presumed portal of entry was found more frequently in the Sm group (n = 22; 76%) than in the BHS group (n = 22; 39%) (P<.01). When the Sm portal of entry was suspected, it was dental for most patients (n = 17; 77%); ear, nose, and throat for 2 (9%); gastrointestinal for 2 (9%); and genitourinary for 1 (5%). In the BHS group, suspected portals of entry were cutaneous for 11 patients (50%; 2 GAS, 5 GBS, 1 GCS, and 3 GGS); ear, nose, and throat for 5 (23%; 3 GAS, 1 GCS, and 1 GGS); gastrointestinal for 3 (14%; all GBS), genitourinary for 2 (9%; both GBS); and pulmonary for 1 (5%; GBS). Multivariate analysis showed that a longer median time of IE evolution before diagnosis and identification of the presumed portal of entry were independently associated with the occurrence of Sm IE (P<.01).
All patients underwent echocardiography, including TEE in 41 patients (73%) in the BHS group and 21 (72%) in the Sm group. According to the Duke criteria, echocardiographic findings suggesting IE were obtained for all patients. No statistically significant differences were found between the echocardiographic findings of the BHS and Sm groups (Table 3). The mitral valve was predominantly involved in both groups. Valvular insufficiency was detected in more than 83% of the patients in both groups. When the size of vegetations was specified, it was 1 cm or larger in 12 of 18 patients in the BHS group and in 7 of 9 in the Sm group. Twenty-nine patients (52%) with BHS IE and 14 (48%) with Sm IE underwent TTE and TEE. The latter was better able to detect vegetations than was TTE, since valvular vegetations were detected only by TEE for 15 patients (27%) in the BHS group and 3 (10%) in the Sm group. In 2 patients with BHS IE, multiple vegetations were detected by TEE, whereas TTE visualized only 1 vegetation. The size and mobility of vegetations and valvular abnormalities were more precisely described in TEE than in TTE reports. Intracardiac abscesses were also better detected by TEE, since it visualized an intracardiac abscess in 5 patients, 4 from the BHS group and 1 from the Sm group.
The isolate responsible for IE was isolated from the blood cultures of all but 2 patients, 1 with GBS isolated from cerebrospinal fluid and 1 with Sm isolated from a bone biopsy specimen. All isolates were susceptible to amoxicillin. Of the 36 BHS isolates for which susceptibility to macrolides was determined, 30 (83%) were susceptible to erythromycin and 6 (17%) were intermediate or resistant; 2 of these nonsusceptible isolates were also resistant to lincomycin. Of the 13 Sm isolates for which susceptibility to macrolides was determined, 3 (23%) were resistant to erythromycin, including 1 also resistant to lincomycin.
Data concerning the antibiotic and surgical treatments are given in Table 4. In both groups, all patients were treated within 24 hours of diagnosis. The shortest durations of antibiotic therapy corresponded to patients who died early. Of 54 patients who underwent surgery, detailed reports of surgical findings were available for 47 (31 in the BHS group and 16 in the Sm group) and described IE lesions in all cases. Surgery found vegetations in 22 BHS IE cases (71%) and 14 Sm IE cases (88%) and a valvular perforation in 6 BHS IE cases (19%) and 5 Sm IE cases (31%). An intracardiac abscess was detected in 14 patients (45%) with BHS IE and 4 (25%) with Sm IE. The TEE images of vegetations and valvular insufficiency corresponded closely to surgical findings, but not for the diagnosis of intracardiac abscesses and perforations. Indeed, in 4 patients in the BHS group and 1 in the Sm group, the abscess was not detected by TEE. A perforation was evidenced only during surgery in 5 patients with Sm IE. For 2 patients with BHS IE and 1 with Sm IE who had TEE-detected vegetations and underwent surgery more than 1 month after IE diagnosis, no vegetations were seen during the intervention. Results of histologic examination or direct gram staining of the vegetations were available for 15 patients with BHS IE and 6 with Sm IE; gram-positive cocci were identified in 5 and 2 patients, respectively. Bacteriologic cultures of the valves were positive for 2 of 26 patients with BHS IE and 1 of 12 with Sm IE for whom culture results were available.
The outcome was known for all patients in the Sm group and 55 of 56 with BHS IE (Table 4). For all patients, univariate and multivariate analyses retained age as the only risk factor for death. Indeed, patients who died were significantly older than those who survived (66 ± 10 vs 56 ± 17 years; P = .01). The median duration of follow-up for patients alive 1 month after IE diagnosis was 5 months (range, 30 days to 4 years) for BHS IE and 4 months (range, 30 days to 7 years) for Sm infections. Three patients from each group relapsed, and 2 other patients with BHS IE developed another episode of IE due to another pathogen. The mortality rate tended to be higher for patients with GBS IE (n = 12; 36%) than for those with other BHS serotypes (n = 3; 14%) or Sm IE (n = 4; 14%) (P = .05).
Streptococci organisms are most frequently responsible for IE, accounting for 50% to 70% of cases.9,10 Although viridans streptococci, enterococci, and Streptococcus bovis are most often encountered, other streptococci represented 7% of all IE episodes in a 1-year French survey.10 We report herein on a series of 85 adults with BHS or Sm IE (all fulfilled the Duke criteria).22 Because these recent observations were made during a period that saw no major change in diagnostic and therapeutic procedures, some conclusions can be drawn.
β-Hemolytic streptococci IE has typically been described as an acute infection that often occurs on normal heart valves, producing large valvular vegetations and responsible for frequent cardiac complications or major systemic emboli.13,15- 17,19,24- 26 In our series, nearly half of the patients with BHS IE had severe underlying conditions, including diabetes mellitus, malignancy, chronic alcoholism, or cirrhosis, in accordance with the literature.13,15,18,27- 29 Serogroup B was the most frequent organism recovered from patients with BHS IE, in agreement with Baddour,27 and it was associated with the highest mortality rate. Although GBS infections cause substantial pregnancy-related morbidity, none of our patients with GBS IE were pregnant. Of the 21 patients with GBS IE described by Baddour,27 only 2 had a pregnancy-related risk factor, that is, elective or spontaneous abortion 6 and 3 weeks before the onset of endocarditis symptoms. In accordance with reported studies,8,27,30 our observations confirm that GBS infections represent a major cause of severe invasive disease in nonpregnant adults, especially the elderly and those with chronic diseases. Although other researchers8 found human immunodeficiency virus infection to be a predisposing underlying disease for invasive GBS infections, none of our patients were infected with human immunodeficiency virus. None of our patients were intravenous drug addicts, which contradicts the literature, where BHS IE, and usually GAS and GGS IE, have been isolated from intravenous drug users.7,14,28,31 In 28 of our patients (52%) and 19 (61%) of a recent 31-case series of GBS IE,27 no predisposing cardiac lesion could be identified. A high rate of patients with BHS infections (18% in our series and 20% in the series by Baddour27) had prosthetic valve IE, but we did not find a prosthesis to be a significant risk factor for mortality.
Strains of Sm have long been characterized by their propensity for invasive pyogenic infections, which differentiates them from other viridans streptococci.6,32- 34 It has been shown35 that Sm strains possess a variety of pathogenic properties that may be involved in the formation of abscesses and endocarditis. In this study, distant suppurative complications were detected in 7 patients (25%) and intracardiac abscesses in 4 (14%). In a previous study analyzing the characteristics of Streptococcus viridans IE,34 no differences between Sm and S viridans IE were found concerning underlying medical or cardiac conditions, age, type and duration of presenting signs and symptoms, TTE characteristics, frequency of complications, treatment, and outcome. However, in that study, only 4 patients with Sm IE, all due to S intermedius, were analyzed. Our observation differs from that of other researchers,6,36- 39 who found Sm infection to be frequently complicated by the formation of myocardial abscesses or distant septic metastases.
Because BHS and Sm share some physiologic and pathogenic properties, we compared the characteristics of the severe infections they may provoke. Patients infected with BHS tended to be younger, to have more underlying medical conditions but fewer underlying cardiopathies, to have a more acute presentation, and to have more extracardiac complications. In addition, more patients underwent surgery during the first month and more died, suggesting a more aggressive course of BHS IE. Inversely, multivariate analysis showed that a longer duration of evolution before diagnosis was independently associated with the occurrence of Sm IE. Our results suggest that BHS are more likely to cause acute and aggressive infections that often affect patients with underlying medical diseases and are responsible for frequent embolic complications and high mortality rates, as we40 recently described for Streptococcus pneumoniae IE. In contrast, Sm IE has a more subacute presentation, affects older patients with underlying cardiopathy, and provokes embolic complications much less frequently, more closely resembling S viridans IE.34 The presumed portal of entry could more often be identified in Sm-infected patients, perhaps because it was dental for most of them (n = 17; 59%) and thus easier to discern than the usual BHS portals of entry, which are more often deep and detected only by invasive procedures.
All patients described in this series underwent echocardiography, particularly TEE (>70% of patients in both groups). Although BHS IE had a more acute presentation than did Sm IE in our series and was reported to induce major intracardiac damages,41 the rates of echocardiographic abnormalities were similar for the 2 groups. In addition, intracardiac abscesses in the BHS group were less frequent than would have been expected.41 We noted a strong similarity between TEE-detected cardiac damage and perioperative observations of vegetations and valvular insufficiency, but not for the diagnosis of intracardiac abscesses and valvular perforations. However, all patients with TEE abnormalities had a surgical diagnosis of IE, suggesting that, among cases fulfilling the Duke criteria for definite IE, there will be few false-positive diagnoses.
Given the small numbers of patients in the previously published series, no precise recommendations concerning the optimal antibiotic therapy for Sm or BHS IE have been established. Penicillin was considered the antibiotic of choice,15,26,42 and although the benefit of combined therapy with an aminoglycoside has not been clearly demonstrated,13,19 some authors16- 18 recommend initial combination therapy with penicillin and an aminoglycoside, especially since tolerance may be encountered in some isolates. All isolates in our series were susceptible to amoxicillin, which is in accordance with a recent in vitro study in which none of the 44 Sm isolates were resistant to β-lactam antibiotics.43 More than 92% of our patients received combination therapy, which included an aminoglycoside for all but 2 patients. The fact that 17% to 23% of the isolates were resistant to macrolides, including some lincomycin-resistant isolates, is of particular importance because clindamycin or pristinamycin is currently recommended against IE for patients who are allergic to β-lactams and are undergoing dental procedures44,45 and thus may not always protect against the risk of Sm IE.
Nearly two thirds of our patients underwent surgery, which is in contrast to the few patients who received combined medical-surgical treatment in previous studies.12,17,25,46 Indeed, 24 (67%) of patients with BHS and 8 (44%) with Sm IE who were operated on required surgery during the first month, suggesting an aggressive presentation, especially for BHS infections. The most frequent indication for surgery was acute valve dysfunction. Comparing the mortality rates of surgical and nonsurgical patients, the benefit of surgery could not be clearly demonstrated. However, the overall mortality rate of BHS IE was lower in our series (14% for groups A, C, and G and 36% for group B) than in previous studies (14%-70% for GAS, GCS, and GGS IE and 41%-52% for GBS IE),13,15,17,24,28 in which only a few patients had surgery. We think that more aggressive surgical intervention most likely contributed to the lower mortality rate shown in our series and as recently demonstrated for S pneumoniae IE40 and Staphylococcus aureus prosthetic valve IE.47 That surgery might play a protective role in BHS IE had already been suggested by some authors,27 who attributed the low mortality rate (12.9%) observed to the high rate (48.4%) of patients who had surgery. In the absence of previous large series, the mortality rate due to Sm IE could never be clearly determined. In the study by Casariego et al,36 1 of 7 patients with Sm IE died. We had 14% mortality, which is lower than the overall mortality for IE (19%-21%10,48) but slightly higher than that for S viridans IE (6%-8%11,34).
In conclusion, results of this recent nation-based study show that BHS IE is responsible for a rapidly progressing aggressive infection with a high rate of extracardiac complications that primarily affects patients with underlying disease and, less frequently, underlying cardiopathy. All our cases were observed during a short, recent period, which permitted comparison of clinical, echocardiographic, and prognostic characteristics between BHS IE and Sm IE. A longer time of evolution of the disease before diagnosis and identification of the presumed portal of entry were independently associated with the occurrence of Sm IE, whereas no differences in the echocardiographic characteristics were found between the 2 groups, despite the more aggressive presentation of BHS IE. The results of this study suggest that patients with BHS IE should be especially carefully monitored because prognosis of BHS IE might be improved by early combined surgical-medical therapy guided by TEE, which might prevent early deaths.
Accepted for publication April 3, 2002.
This work was supported by the Société Nationale Française de Médecine Interne (Lille) and by a grant from SmithKline Beecham Laboratories, Nanterre, France.
This work was presented in part at the 38th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Diego, Calif, September 25, 1998.
We thank the Société Française de Cardiologie (Paris) for its logistical support.
Members of the β-Hemolytic Streptococci Infective Endocarditis Study Group
G. Lepeu and H. Lefrand, Hôpital Henri-Duffaut, Avignon; T. Zenone and M. Chomarat, CHU Lyon-Sud, Lyon; F. Delahaye, F. Vandenesch, and R. Loire, CHU Louis-Pradel, Lyon; O. Launay, J. P. Bédos, V. Le Moing, and R. Ruimy, CHU Bichat-Claude-Bernard, Paris; B. Patri, E. Varon, and C. Marvejouls, CHU Boucicaut, Paris; F. Bricaire and J. Nguyen, CHU Pitié-Salpêtrière, Paris; A. Buu-Hoï, CHU Broussais, Paris; J. L. Meynard and D. Lesage, CHU St-Antoine, Paris; A. Bouvet, CHU Hôtel-Dieu, Paris; N. Galezowski, Hôpital St-Joseph, Paris; V. Leflon, CHU Ambroise-Paré, Boulogne-Billancourt; D. Agnola and D. Aubert, Centre Hospitalier, Nanterre; G. Motte and M. Guibert, CHU Antoine-Béclère, Clamart; D. Magnin and H. Tronel, Hôpital Louis-Pasteur, Dole; J. R. Monties and J. P. Casalta, CHU de la Timone, Marseille; P. Brouqui, CHU Houphouet-Boigny, Marseille; G. Humbert and J. F. Lemeland, CHU Charles-Nicolle, Rouen; I. Beguinot and V. Vernet-Garnier, CHU Robert-Debré, Reims; P. Mornet and A. Boisivon, CHI, St-Germain-en-Laye; R. Roudaut and J. Maugein, Hôpital Haut-Lévêque, Pessac; P. Morlat, CHU St-André, Bordeaux; M. Weber and T. Doco-Lecompte, Hôpital Central and CHU-Brabois, Nancy; C. Alba-Sauviat, Centre Hospitalier, Chaumont; M. Paccalin, C. Burucoa, and P. Babin, CHU La Milétrie, Poitiers; R. Verdon and M. Vergnaud, CHU, Caen; F. Geisler and M. Thouvenin, Hôpital Ste-Catherine, Saverne; F. Fraisse and I. Gros, Hôpital Delafontaine, St-Denis; P. Lesprit and J. Tankovic, CHU Henri-Mondor, Créteil; M. Scavizzi, CHU Avicenne, Bobigny; F. Cartier and J. L. Avril, CHU Pontchaillou, Rennes; A. Millaire, Hôpital Cardiologique, Lille; M. Roussel-Delvallez, CHU Albert-Calmette, Lille; J. Schmidt and J. Sirot, CHU, Clermont-Ferrand; P. Massip and J. Lemozy, CHU Purpan, Toulouse; J. M. Besnier, L. Mereghetti, and M. C. Grangeponte, CHU, Tours; C. Combe, Centre Hospitalier, Villefranche-sur-Saône; S. Dellion and A. Dublanchet, CHI, Villeneuve St-Georges; A. H. Nguyen and J. Caillon, CHU Laënnec, Nantes; and F. Raffi and C. Chamoux, CHU Hôtel-Dieu, Nantes, France.
Corresponding author and reprints: Jean-Luc Mainardi, MD, PhD, Unité Mobile de Microbiologie Clinique, Service de Microbiologie Clinique, Hôpital Européen Georges-Pompidou, 20 rue Leblanc, 75908 Paris CEDEX 15, France (e-mail: firstname.lastname@example.org).