Bacteriological outcome of combination versus single-agent treatment for staphylococcal endocarditis

Transcription

1 Journal of Antimicrobial Chemotherapy (2003) 52, DOI: /jac/dkg440 Advance Access publication 30 September 2003 Bacteriological outcome of versus single-agent treatment for staphylococcal endocarditis Dragana DrinkoviM 1, Arthur J. Morris 1 *, Sudha Pottumarthy 1, Donald MacCulloch 1 and Teena West 2 Departments of 1 Microbiology and 2 Biostatistics, Green Lane Hospital, Auckland, New Zealand Received 10 June 2003; returned 12 July 2003; revised 23 July 2003; accepted 11 August 2003 Objective: To analyse the bacteriological outcome of versus single-agent antimicrobial treatment in staphylococcal endocarditis. Patients and methods: Retrospective review of 152 episodes: 91 cases of native valve endocarditis (NVE), 74 due to Staphylococcus aureus and 17 due to coagulase-negative staphylococci (CoNS); and 61 cases of prosthetic valve endocarditis (PVE), 29 due to S. aureus and 32 due to CoNS. Results: Valves from patients with S. aureus NVE treated with any kind of antibiotic treatment were no more likely to be culture-negative than those treated with a single agent [19 (45%) of 42 versus 13 (41%) of 32; P = 0.69]. This finding remained unchanged when cases of CoNS NVE were added to the S. aureus group. In PVE, after adjusting for duration of treatment, valves from patients receiving any kind of treatment were 5.9 times (95% confidence interval ) more likely to be culture-negative than those receiving monotherapy (P = 0.024). Patients treated for >14 days were more likely to be culturenegative than those treated for 14 days [49 (83%) of 59 versus 29 (31%) of 93; P < 0.001]. Conclusions: In staphylococcal NVE, treatment is not superior to monotherapy in sterilizing infected valves, but in PVE treatment confers an advantage. Keywords: coagulase-negative staphylococci, culture result, heart valves, Staphylococcus aureus, Staphylococcus epidermidis Introduction Information on the relative efficacy of versus singleagent antimicrobial treatment to achieve valve sterilization in staphylococcal endocarditis is incomplete. 1 Nevertheless, regimens are recommended in both UK and US guidelines. 2,3 The UK guidelines are simpler than those from the US, and make similar recommendations for the choice of agents in both native (NVE) and prosthetic valve (PVE) staphylococcal endocarditis. 2 Gentamicin is recommended for the initial 7 days of treatment, with rifampicin being suggested for difficult cases. 2 The US recommendations make a distinction between NVE and PVE. For NVE gentamicin is suggested for the first 3 5 days of treatment. For PVE initial triple-agent treatment, vancomycin or a β-lactamase-resistant penicillin, rifampicin and gentamicin, is recommended. 3 To test the hypothesis that antimicrobial treatment is more likely to result in valve sterilization than single-agent treatment we have analysed valve culture results of patients undergoing valve replacement surgery for staphylococcal endocarditis with respect to the preoperative antimicrobial treatment. Patients and methods Patient population Green Lane Hospital s cardiothoracic surgical unit serves 1 million people and is the reference centre for complicated cases for all New Zealand ( 4 million people). All patients undergoing valve replacement surgery from September 1963 through December 1999 who had endocarditis recorded as the cause of their valve dysfunction had their charts reviewed. 4 The information recorded for each patient included: identifying information, number of blood cultures taken, number of positive blood cultures sets, organism(s) isolated, duration and type of antibiotic treatment received before surgery, indication(s) for surgery (valve function, heart failure, fever/sepsis, emboli), operation findings and culture result of the resected valve *Correspondence address. Microbiology Laboratory, Auckland City Hospital, PO Box , Auckland 1003, New Zealand. Tel: ; Fax: ; amorris@dml.co.nz JAC vol.52 no.5 The British Society for Antimicrobial Chemotherapy 2003; all rights reserved.

2 Single versus treatment in staphylococcal IE Table 1. Treatment regimens and valve culture results for S. aureus NVE Culture result Group a n Treatment regimen Median (days) positive (%) negative (%) Comment 1 32 single agent (59) 13 (41) three patients were started on gentamicin treatment but were switched early to flucloxacillin alone 2 27 continuous 9 18 (67) 9 (33) most [24 (89%)] received β-lactam and aminoglycoside treatment, usually flucloxacillin and gentamicin 3 9 initial 11 4 (44) 5 (56) initial β-lactam and aminoglycoside, then β-lactam alone 4 6 partial (17) 5 (83) various antibiotics used: β-lactam, gentamicin, vancomycin, rifampicin a Group 1 versus 2, P = 0.56; group 1 versus 2 + 3, P = 0.88; group 1 versus , P = After adjusting for duration of treatment by logistic regression analysis valves from patients receiving any kind of treatment were no more likely to be culture-negative than those receiving monotherapy (odds ratio 0.45; 95% CI ; P = 0.23). Definitions The modified Duke criteria 5 were used to define cases of endocarditis, except that a positive microbiology Gram stain was taken as definitive evidence of endocarditis. 4 PVE was defined as: early when it occurred within 60 days of valve insertion, intermediate between 60 days and 12 months after surgery, and late >12 months after surgery. 6 Combination antibiotic treatment was defined as receipt of two or more antibiotics active against the staphylococcal isolate, and regarded as having the potential to act synergically, 1 simultaneously for at least 1 day. In vitro tests for synergy were not performed. Continuous treatment was defined as receipt of two or more active antibiotics for the entire preoperative period. Initial treatment was defined as receipt of two or more active antibiotics for at least the first 3 days of treatment, but not for entire preoperative course. Partial treatment was defined as receipt of two or more active antibiotics for a part of the preoperative period other than above. Patients received standard doses of the respective antibiotics, e.g. flucloxacillin 12 g daily in six divided intravenous (iv) doses, vancomycin iv 1 g twice a day, gentamicin 1 mg/kg, iv three times a day, with appropriate adjustments for impaired renal function. Laboratory procedures Resected valves were sent for Gram stain and culture as described previously. 4 Briefly, following macroscopic examination, vegetations and material that appeared to be infected were removed, ground in trypticase soy broth, Gram stains made, and solid and four liquid media inoculated. Media were examined regularly and discarded after 7 days. Routine broth subcultures onto solid media were performed after incubation for 7 days and the plates were incubated for 48 h aerobically and anaerobically before being discarded. Adherent clot or other material removed from prosthetic valves was treated as above. Isolates from surgical specimens were compared with available preoperative blood culture isolates. Statistics Data are presented as counts (percentages) and median (inter-quartile range) unless otherwise stated. Groups were compared using the χ 2 -test and logistic regression analysis was used to adjust for indications for surgery and duration of treatment. Statistical analysis was performed using SAS release 8.0 software (Cary, NC, USA). Results One hundred and seventy-four patients had staphylococcal infective endocarditis (IE). Two patients, receiving antibiotic treatment, had negative preoperative blood cultures but positive valve cultures. All other patients had positive preoperative blood cultures. Twenty-two patients were excluded from this study: 12 because they had not received antibiotic treatment before valve replacement and eight because valve cultures were not performed; one patient was excluded because of discrepant susceptibility results between blood culture and valve isolates and another case excluded because there were insufficient susceptibility data to match the isolates. Of the remaining 152 cases, 91 (60%) had NVE and 61 (40%) had PVE; 139 met the criteria for definite endocarditis, 135 (89%) by pathological and four (3%) by clinical criteria, and 13 (9%) met the criteria for possible IE. 4,5 Staphylococcus aureus NVE Seventy-four (81%) of the 91 patients with NVE had Staphylococcus aureus infection (Table 1). Only one patient had methicillin-resistant S. aureus (MRSA) infection. Three patients were injecting drug users, two with tricuspid valve infection. Forty-two patients had aortic valve IE, 23 mitral valve, three both aortic and mitral valve, and six had tricuspid valve IE. Invasive extravalvular infection was present in 38 (51%) patients. Thirty-two (43%) of these 74 patients were treated with a single β-lactam antibiotic, either flucloxacillin or penicillin (Table 1). Nineteen (59%) of the 32 were valve culture positive. Twenty-seven patients had continuous therapy. In two patients rifampicin was added to β-lactam and aminoglycoside treatment and in one patient it was combined with the β-lactam only. Eighteen (67%) of these 27 patients had positive valve cultures. Patients receiving continuous treatment were no more likely to be culture-negative than those receiving single-agent treatment [nine (33%) of 27 versus 13 (41%) of 32, respectively; P = 0.56] (Table 1). Nine patients had initial therapy with β-lactam and aminoglycoside followed by β-lactam alone, and four (44%) had positive valve cultures. When patients receiving continuous 821

3 D. DrinkoviM et al. Table 2. Treatment regimens and valve culture results for staphylococcal PVE Culture result Group a n Treatment regimen Median (days) positive (%) negative (%) Comment 1 14 single agent 17 8 (57) 6 (43) most [12 (86%)] received either β-lactam or vancomycin 2 4 continuous triple (0) 4 (100) β-lactam or vancomycin with rifampicin and gentamicin agent 3 20 continuous dual agent 9 10 (50) 10 (50) either β-lactam with aminoglycoside or vancomycin with rifampicin 4 23 partial 19 4 (17) 19 (83) 18 (78%) patients received β-lactam with aminoglycoside or rifampicin; three (13%) received vancomycin and rifampicin; two (9%) received vancomycin with rifampicin and gentamicin a Group 1 versus 2 + 3, P = 0.36; group 1 versus , P = After adjusting for duration of treatment by logistic regression analysis valves from patients receiving any kind of treatment were 5.9 times more likely to be culture-negative than those receiving monotherapy (95% CI ; P = 0.024). and initial therapy were compared with patients on a single therapy, there was no difference in the likelihood of them having a negative valve culture [14 (39%) of 36 versus 13 (41%) of 32; P = 0.88]. Another six patients were treated with partial treatment, and one (17%) was culture positive. Patients receiving any kind of treatment were no more likely to be culturenegative than patients receiving a single agent [19 (45%) of 42 versus 13 (41%) of 32; P = 0.69] (Table 1). This finding remained unchanged after adjusting for duration of treatment [odds ratio 0.45; 95% confidence interval (CI) ; P = 0.23]. Coagulase-negative staphylococci NVE Seventeen patients had NVE caused by coagulase-negative staphylococci (CoNS). Twelve episodes were caused by Staphylococcus epidermidis, two of which were methicillin resistant. Two episodes were caused by Staphylococcus warneri and Staphylococcus cohnii, respectively, and three episodes by non-speciated CoNS. Eleven patients had aortic valve endocarditis, five had mitral and one had both aortic and mitral infection. Six (35%) patients had extravalvular extension of infection, with periannular abscess formation in four patients. Vegetations were found at surgery in 16 patients and 10 (59%) patients had ruptured cusps. Four patients were treated with a single β-lactam antibiotic and three (75%) had positive valve culture. Eight patients had continuous therapy: a β-lactam and aminoglycoside in five patients, and rifampicin, combined with either penicillin, vancomycin or fusidic acid, in three patients, respectively. Six (75%) of these eight had positive cultures. Valves from patients receiving treatment were no more likely to be culture-negative than those receiving single therapy [six (46%) of 13 versus one (25%) of four; P = 0.60]. All NVE When cases of CoNS NVE were added to the S. aureus group, the observations regarding culture results and treatment group remained unchanged. Patients receiving single-agent treatment were as likely to be culture-negative [14 (39%) of 36] as those receiving continuous treatment [11 (31%) of 35; P = 0.51], those receiving continuous plus initial treatment [18 (39%) of 46; P = 0.98] or those receiving any kind of treatment [25 (45%) of 55; P = 0.54]. This finding remained unchanged after adjusting for duration of treatment (P = 0.23). PVE Sixty-one patients had PVE, 29 due to S. aureus and 32 due to CoNS (Table 2). Seven episodes of CoNS NVE occurred within 2 months of surgery and six (86%) isolates were methicillin resistant; 10 episodes had onset between 3 and 12 months, five (50%) isolates were methicillin resistant, as were seven (47%) of the 15 isolates causing infection after 12 months. Most [21 (72%) of 29] S. aureus episodes occurred >12 months after surgery, with early and intermediate PVE occurring in four patients each. All S. aureus isolates were methicillin susceptible. Overall, 33 patients had mitral valve IE, 21 patients had aortic IE and seven patients had both valves infected. Extensive extravalvular infection was present in 30 (49%) patients. Current recommendations for staphylococcal PVE do not distinguish between species, 2,3 and therefore culture results versus treatment regimen are analysed for all cases of PVE without reference to species. Fourteen patients had treatment with a single antibiotic: nine were treated with a β-lactam and five, infected with methicillin-resistant isolates, were treated with either vancomycin (three patients), gentamicin (one) or clindamycin (one). Eight (57%) of these 14 had positive valve cultures. Four patients with PVE were treated continuously with three antibiotics, β-lactam or vancomycin plus rifampicin plus gentamicin and all had negative valve cultures. Twenty patients had continuous treatment with two antibiotics before their surgery: 14 with β-lactam and aminoglycoside, and six with vancomycin plus rifampicin for their methicillinresistant isolates. Ten (50%) of these 20 had positive valve cultures. Twenty-three patients had partial treatment. Most of them were treated with two antibiotics, and a β-lactam and an aminoglycoside were used in 15 patients. Rifampicin was combined with either β-lactam or vancomycin in three patients respectively. Two patients had three antibiotics, vancomycin plus gentamicin plus rifampicin, for a part of their preoperative period. Four (17%) of

4 Single versus treatment in staphylococcal IE had positive valve cultures. Although patients treated with continuous treatment were no more likely to be culturenegative than those treated with a single agent [14 (58%) of 24 versus six (43%) of 14; P = 0.36], there was a trend for valves from patients receiving any kind of treatment to be culture-negative compared with valves removed from patients receiving single-agent treatment [33 (70%) of 47 versus six (43%) of 14; P = 0.062] (Table 2). After adjusting for duration of treatment by logistic regression analysis valves from patients receiving any kind of treatment were 5.9 times (95% CI ) more likely to be culture-negative than those receiving monotherapy (P = 0.024). Eighteen patients had rifampicin included in their treatment. Twelve (67%) were valve culture negative. Twenty-seven (63%) of the remaining 43 patients without rifampicin in their treatment regimen had negative valve cultures. Rifampicin was combined with vancomycin in nine patients, flucloxacillin in two, a first generation cephalosporin in one and was a part of triple antibiotic regimen in six. Six (33%) of these 18 patients had positive valve culture at surgery and all had received vancomycin and rifampicin. In two (33%) of six culture-positive patients, rifampicin resistance had developed during treatment. Patients with infection invading beyond the leaflets were no more likely to be culture-positive than patients with infection affecting valves only [38 (51%) of 74 versus 36 (46%) of 78; P = 0.52]. After accounting for duration and type of treatment, prosthetic valves were no more likely to be culture-positive than native valves (P = 0.169). There was no difference in the indications for surgery between the different treatment groups for either S. aureus NVE (P = 0.53), any staphylococcal NVE (P = 0.51) or PVE (P = 0.66). During the study period there were 23 cases before 1980, 58 in the 1980s and 71 in the 1990s. The median (mean) duration of treatment before surgery for the respective time periods decreased: 16 (17), 14 (17.5) and 9 (12) days, respectively (P = 0.006). The proportion of patients receiving monotherapy did not show any trend over time (P = 0.27). Treatment duration was predictive of a negative culture: 49 (83%) of 59 and 29 (31%) of 93 treated for either >14 or <14 days, respectively, were culture-negative (P < 0.001). After adjusting for infecting organism, valve type and type of treatment, each day of treatment increased the likelihood of the valve being culture-negative by 17% (95% CI 11 22; P < 0.001). Discussion S. aureus NVE S. aureus accounts for 25 40% of NVE. 6 It is associated with considerable morbidity and mortality. Rapid sterilization of vegetations could result in less valve destruction and reduce the risk of metastatic complications. In vitro and in vivo observations have shown that a β-lactam antibiotic plus an aminoglycoside are more rapidly bactericidal for staphylococci than a β-lactam alone. 7,8 A multicentre prospective study of S. aureus endocarditis showed that addition of gentamicin for the first 2 weeks of a 6 week course of iv nafcillin failed to improve clinical outcome. 9 Nevertheless, treatment was associated with earlier clearance of bacteraemia in nonaddicts as well as in addicts with right-sided endocarditis. 9 These findings have led to the suggestion that an aminoglycoside be discontinued after 3 7 days of therapy 2,3 in the hope that more rapid control of bacteraemia would be accompanied by a lesser incidence of metastatic infections and accelerated sterilization of heart valves, while avoiding the toxic reactions associated with a more prolonged course of aminoglycosides. 9 Our findings show no significant difference in the rate of valve sterilization in S. aureus NVE between patients treated with single or therapy, the majority (89%) of which was β-lactam and aminoglycoside treatment. CoNS NVE NVE caused by CoNS is relatively uncommon, accounting for 4 8% of all native valve endocarditis. 6 As in previous studies, S. epidermidis was the most commonly isolated species in our patients Although CoNS NVE is usually described as a subacute, sometimes indolent process, 11,13,14 infection can be destructive. 11,12 Ten (59%) of our patients had ruptured valves and six (35%) had extravalvular extension of infection with periannular abscess formation in four of them. Many CoNS that cause NVE are susceptible to β-lactam antibiotics. Although single-agent treatment with a β-lactam has been reported as being effective, treatment may be better. Caputo et al. 11 noted that seven (64%) of 11 patients treated with a single agent were cured, while all 10 patients receiving antibiotic were cured. 11 When their data were combined with those of previous series, 21 (72%) of 29 patients treated with a single agent were cured, compared with 23 (92%) of 25 receiving treatment; however, the difference was not significant (P > 0.05). In our study, treatment did not result in a significant increase of valve sterilization. This observation held when CoNS cases were added to the S. aureus episodes. This finding does not support the notion that the addition of an aminoglycoside in the initial phase of treatment in staphylococcal endocarditis hastens valve sterilization. PVE In developed countries, PVE accounts for 7 25% of all cases of IE. 6 The cumulative risk of PVE is 1% at 12 months and 2 3% at 60 months after surgery. 6 The most common cause of PVE within 12 months after surgery is CoNS, with S. epidermidis predominating. 6,15 CoNS cause 60 70% of cases of PVE in the first 12 months of surgery and 10 12% thereafter. S. aureus causes 30 39% of PVE in the first 12 months and 15 20% of PVE after 12 months. 6 The US guidelines for staphylococcal PVE recommend treatment with either vancomycin or nafcillin (depending on susceptibility results) and rifampicin administered for a minimum of 6 weeks, together with gentamicin for the initial 2 weeks of treatment. 3 In experimental endocarditis in rabbits due to methicillin-resistant S. epidermidis, the of vancomycin plus gentamicin plus rifampicin was more effective in sterilizing vegetations than these agents administered alone, or the s of vancomycin and gentamicin or vancomycin and rifampicin. 16 In our study only four patients received continuous, and two patients a partial, triple antibiotic regimen. All six had negative valve cultures. Although most of our patients with PVE [55 (90%) of 61] did not receive triple treatment for the first 2 weeks, most [33 (60%) of the 55] had negative valve cultures. We observed a trend for valves from patients receiving either continuous or partial treatment for PVE to be culture-negative compared with those treated with a single agent (Table 2). However after adjusting for duration of treatment, valves from patients receiving any kind of treatment were almost six times more likely to be culture-negative than those receiving monotherapy (Table 2). This finding supports current recommendations for treatment for PVE, 2,3 despite the 823

5 D. DrinkoviM et al. potential toxicity of aminoglycoside-containing regimens. Whether treatment should follow either the UK (initial with gentamicin) or US (initial triple followed by double-agent treatment) guidelines is not settled by our observations, but animal model data suggest that triple-agent therapy may achieve sterilization in a greater proportion of valves. 16 Treatment of experimental foreign body infection is enhanced by regimens, especially ones that include rifampicin. 17,18 It is thought that rifampicin effectiveness is due to its excellent tissue penetration and ability to enter living phagocytes and kill intracellular bacteria, 19 as well as its ability to kill stationary phase microorganisms. 20 In our study the rate of valve sterilization was similar between groups with and without rifampicin [67% versus 63%]. We do not know whether the rate of valve sterilization would be higher in the rifampicin group if it had been used more often in therapy. The possibility of the development of resistance to rifampicin during vancomycin plus rifampicin or triple-agent treatment has been observed in experimental animal models 18,21 and clinical infections In a randomized treatment trial for PVE due to methicillin-resistant CoNS, rifampicin resistance developed in 37% of patients treated with vancomycin plus rifampicin, compared with 0% receiving vancomycin and rifampicin with gentamicin added for the first 2 weeks of treatment. 15 Two (33%) of our six patients with positive valve cultures after treatment with of vancomycin and rifampicin developed resistance to rifampicin. Although all six patients treated with vancomycin plus rifampicin plus gentamicin were culture-negative, the number of cases in each group was too small to make any conclusions regarding the effectiveness of gentamicin addition in prevention of rifampicin resistance. Although our study is limited by its retrospective nature and the limited number of patients in some treatment groups, the data give a unique insight into the efficacy of antibiotic treatments in sterilizing heart valves in staphylococcal endocarditis. We can conclude that valves from patients with S. aureus NVE treated with treatment are no more likely to be culture-negative than those treated with a single agent. This observation does not necessary mean that the recommendation for initial treatment with flucloxacillin, or a similar agent, and gentamicin for S. aureus NVE should be abandoned. It does, however, suggest that if the aminoglycoside is omitted, valve sterilization will not be significantly adversely affected. We observed that patients receiving treatment for staphylococcal PVE were more likely to have culture-negative valves. This finding, together with animal model data, supports the current recommendations for initial treatment. 2,3 Regardless of staphylococcal species, valve type or type of treatment, duration of treatment was predictive of a negative culture. After 14 days treatment valve sterilization is achieved in the majority of patients. This observation is of clinical relevance because if valve replacement is not required immediately, e.g. to restore valve function, most would prefer for surgery to be timed when there is the best likelihood of the new prosthesis being inserted into a sterile surgical field. Acknowledgements Financial support for this study was provided by The Auckland Infectious Disease Education and Research Trust. References 1. Le, T. & Bayer, A. S. (2003). Combination antibiotic therapy for infective endocarditis. Clinical Infectious Diseases 36, Working Party of the British Society for Antimicrobial Chemotherapy. (1998). Antibiotic treatment of streptococcal, enterococcal, and staphylococcal endocarditis. Heart 79, Wilson, W. R., Karchmer, A. W., Dajani, A. S. et al. (1995). Antibiotic treatment of adults with infective endocarditis due to streptococci, enterococci, staphylococci, and HACEK microorganisms. Journal of the American Medical Association 274, Morris, A. J., Drinkovic, D., Pottumarthy, S. et al. (2003) Gram stain, culture and histopathological examination findings for heart valves removed because of infective endocarditis. Clinical Infectious Diseases 36, Li, J. S., Sexton, D. J., Mick, N. et al. (2000). Proposed modifications to the Duke criteria for the diagnosis of infective endocarditis. Clinical Infectious Diseases 30, Mylonakis, E. & Calderwood, S. B. (2001). Infective endocarditis in adults. New England Journal of Medicine 345, Watanakunakorn, C. & Glotzbecker, C. (1974). Enhancement of the effects of anti-staphylococcal antibiotics by aminoglycosides. Antimicrobial Agents and Chemotherapy 6, Sande, M. A. & Johnson, M. L. (1975). Antimicrobial therapy of experimental endocarditis caused by Staphylococcus aureus. Journal of Infectious Diseases 131, Korzeniowski, O., Sande, M. A. & The National Collaborative Endocarditis Study Group. (1982). Combination antimicrobial therapy for Staphylococcus aureus endocarditis in patients addicted to parenteral drugs and in nonaddicts. Annals of Internal Medicine 97, Etienne, J. & Eykyn, S. J. (1990). Increase in native valve endocarditis caused by coagulase-negative staphylococci: an Anglo-French clinical and microbiological study. British Heart Journal 64, Caputo, G. M., Archer, G. L., Calderwood, S. B. et al. (1987). Native valve endocarditis due to coagulase-negative staphylococci: clinical and microbiologic features. American Journal of Medicine 83, Miele, P. S., Kogulan, P. K., Levy, C. S. et al. (2001). Seven cases of surgical native valve endocarditis caused by coagulase-negative staphylococci: an underappreciated disease. American Heart Journal 142, Arber, N., Militianu, A., Ben-Yehuda, A. et al. (1991). Native valve Staphylococcus epidermidis endocarditis: report of seven cases and review of the literature. American Journal of Medicine 90, Baddour, L. M., Phillips, T. N. & Bisno, A. L. (1986). Coagulasenegative staphylococcal endocarditis: occurrence in patients with mitral valve prolapse. Archives of Internal Medicine 146, Karchmer, A. W. & Gibbons, G. W. (1994). Infection of prosthetic heart valves and vascular grafts. In Infections Associated with Indwelling Medical Devices, 2nd edn (Bisno, A. L. & Waldvogel, F. A., Eds), pp American Society for Microbiology, Washington, DC, USA. 16. Kobasa, W. D., Kaye, K. L. & Shapiro, T. (1983). Therapy for experimental endocarditis due to Staphylococcus epidermidis. Reviews of Infectious Diseases 5, S Chuard, C., Herrmann, M., Vaudaux, P. et al. (1991). Successful therapy of experimental chronic foreign-body infection due to methicillinresistant Staphylococcus aureus by antimicrobial s. Antimicrobial Agents and Chemotherapy 35, Lucet, J.-C., Herrmann, M., Rohner, P. et al. (1990). Treatment of experimental foreign body infection caused by methicillin-resistant Staphylococcus aureus. Antimicrobial Agents and Chemotherapy 34, Farr, B. M. (2000). Rifamycins. In Principles and Practice of Infectious Diseases, 5th edn (Mandell, G. L., Bennett, J. E. & Dolin, R., Eds), pp Churchill Livingstone, Edinburgh, UK. 20. Widmer, A. F., Frei, R., Rajacic, Z. et al. (1990). Correlation between in vivo and in vitro efficacy of antimicrobial agents against foreign body infections. Journal of Infectious Diseases 162,

6 Single versus treatment in staphylococcal IE 21. Rouse, M. S., Wilcox, R. M., Henry, N. K. et al. (1990). Ciprofloxacin therapy of experimental endocarditis caused by methicillinresistant Staphylococcus epidermidis. Antimicrobial Agents and Chemotherapy 34, Karchmer, A. W., Archer, G. L. & Dismukes, W. E. (1983). Staphylococcus epidermidis causing prosthetic valve endocarditis: microbiologic and clinical observations as guides to therapy. Annals of Internal Medicine 98, Acar, J. F., Goldstein, F. W. & Duval, J. (1983). Use of rifampin for the treatment of serious staphylococcal and gram-negative bacillary infections. Reviews of Infectious Diseases 5, Suppl. 3, S Simon, G. L., Smith, R. H. & Sande, M. A. (1983). Emergence of rifampin-resistant strains of Staphylococcus aureus during therapy with vancomycin and rifampin: a report of two cases. Reviews of Infectious Diseases 5, Suppl. 3, S