- Email: [email protected]
Enterococcal bacteraemia: a prospective study of 125 episodes
Journal
of Hospital
Infection
Enterococcal
27,
179-186
bacteraemia: a prospective 125 episodes
J. Gray, Department
(1994)
P. J. Marsh,
D. Stewart
and
study
S. J. Pedler
of Microbiology, Royal Victoria Infirmary, Queen Victoria Newcastle upon Tyne NE1 4LP, UK Accepted
for publication
12 May
of
Road,
1994
Summary: One hundred and twenty-five episodes of enterococcal bacteraemia occurring over a SO-month period were studied prospectively. Enterococcus faecium was the commonest species, accounting for 76 (59.8%) of the 127 isolates. Overall, 33.1% of isolates were resistant to ampicillin and one isolate (0.8%) to vancomycin; high-level gentamicin resistance was detected in 4.3% of 93 isolates tested. The percentage of nosocomial episodes was 70.4, and 95.2% of the patients had significant underlying illness. Central venous catheters (CVCs) were the commonest source of infection. Eightyfour per cent of episodes were ultimately treated with appropriate antibiotics. The overall mortality rate was 17.6%, and that directly attributable to infection was 8.0%. An increased mortality rate was observed in intensive care and neonatal unit patients, and in patients who had received antimicrobial therapy in the 2 weeks prior to enterococcemia. CVC-related infections were associated with a reduced mortality. No other clinical or microbiological factors were found to influence outcome. Keywords:
Enterococcus;
bacteraemia.
Introduction The importance of enterococci as a cause of 520% of cases of endocarditis is well-established.’ However controversy still surrounds the significance and optimal therapy of enterococcal bacteraemia without endocarditis. A high mortality associated with enterococcal bacteraemia has been longrecognized.2 However patients frequently have serious underlying disease, and enterococci often occur in mixed infections, thus raising doubts as to their pathogenic role.3 There are conflicting reports of the influence of In some series specific antiantimicrobial treatment on prognosis. enterococcal therapy has not been associated with increased surviva1,4-8 and several investigators have suggested that synergistic two-drug regimens are probably unnecessary.7,9-” The frequency of enterococcal bacteraemia is increasing3’” due to factors Correspondence Road, Newcastle 0195-67Ol/94/070179+08
to: Dr upon
J. Gray, Department Tyne, NE1 4LP.
of Microbiology,
$0X.00/0
Royal
Victoria
Infirmary,
0 1994 The Hospital
179
Queen
lnfemon
Victoria
Society
180
J. Gray et al.
such as increased numbers of compromised patients and widespread use of antibiotics, such as cephalosporins, to which enterococci are intrinsically resistant.3’12 Moreover, enterococci that are resistent to ampicillin or to the glycopeptides,‘2”3 or highly-resistant to aminoglycosides,‘2’14 are becoming increasingly common. Infections with such organisms present serious therapeutic difficulties. There have been several studies of enterococcal bacteraemia from North America, 4-8~10~11,15,16 but there are few reports of enterococcal bacteraemia from Europe.‘,17 We have previously reported finding an unusually high frequency of bacteraemia with Enterococcus f~eciurn.‘~ The purpose of this study was to evaluate the epidemiology, clinical significance and treatment of enterococcal bacteraemia in our hospital. Materials
and methods
Setting The Royal Victoria Infirmary is an SSO-bed teaching hospital and tertiary referral centre in North-East England. Specialist services within the hospital include renal dialysis and transplantation, bone marrow transplantation, the regional paediatric oncology unit, a neonatal unit (NNU), and adult paediatric intensive therapy units (ITUs).
Specimens Paired aerobic and anaerobic blood cultures were examined with a BACTEC USA). Cultures were NR730 analyser (Becton Dickinson, Cockeysville, incubated for 7 days before being discarded as negative. Enterococci were identified using standard microbiological methods.” Susceptibility to ampicillin and vancomycin was tested by the break-point method.20 In addition, 93 isolates (E. fuecalis 34; E.faecium 56; E. rafJinosus 2; E. avium 1) were tested for high-level resistance to gentamicin [minimum inhibitory concentration (MIC) >SOOmgl-‘1 by the same method. l3-Lactamase production was tested for by the rapid chromogenic cephalosporin method (nitrocefin; Oxoid Ltd, Basingstoke, UK).
Recording
and analysis of data Clinical and microbiological data on all episodes of clinically-significant enterococcal bacteraemia occurring between 1 August 1989 and 30 September 1993 were recorded prospectively, and subsequently analysed, using the ‘Epi info 5’ database and statistical package (Centers for Disease Control, Atlanta, USA). Statistical analysis was performed by the x2 test, with Yates’ correction where appropriate. Bacteraemia was regarded as hospital-acquired where the initial positive blood culture was obtained more than 48 h after admission. The source of bacteraemia was determined on the basis of microbiological or clinical
Enterococcal
bacteraemia
181
Table I. Species distribution of 127 enterococci isolated from 125 episodes of bacteraemia, and the frequency of resistance to ampicillin [minimum inhibitory concentration (MIC) >8 mg I-‘] and vancomycin (MIC, >4 mg I-‘) No. Species
Enterococcus faecium E. faecalis E. E. raffinosus avium E. durans Total
No. (X) isolates
(“5~) resistant
to
of Amplicillin
76 (59.8) 46 (36.2)
Vancomycin 0 1 (2.2)
2 (1.6) 1 (0.8)
01 (50.0) 0
127
42 (33.1)
P, 0 1 (0.8)
evidence of a focus of infection. Bacteraemia was judged to be related to central venous catheters (CVCs) when it coincided with symptoms related to use of the device, or where microorganisms of the same species, and with the same antibiotic susceptibility pattern, were isolated simultaneously from blood cultures and the device exit-site or cannula tip. Infective endocarditis was defined according to the criteria of Hoge and colleagues.” Deaths that occurred within 7 days of the first positive blood culture were considered to be directly due to infection, unless there was strong evidence to the contrary. Antimicrobial therapy was deemed appropriate if the spectrum of activity of the chosen regimen included activity against the infecting organism. Results
One hundred and twenty-five episodes of clinically-significant enterococcal bacteraemia occurred in 101 patients over the SO-month period. A further seven episodes of uncertain significance were excluded from the analysis. Seventy-four (59.2%) of the episodes were due to E.faecium and 44 (35.2%) to E. fuecalis, whilst E. fuecium and E. fuecalis were isolated simultaneously in two episodes. There were five bacteraemias with other Enterococcus species (Table I). Overall, 33.1% of isolates were resistant to ampicillin (MIC >8 mg l-l), but only one isolate of E. faecalis was resistant to vancomycin (MIC >4mg111), (Table I). Of 93 enterococci tested, four isolates of E. faecalis were highly resistant to gentamicin. P-Lactamase production was not detected in any of the isolates. In 47 (37.6(Y)o cases the bacteraemia was polymicrobial. The additional microorganisms were Gram-positive bacteria in 14 episodes, Gram-negative bacteria in 20, and Candida al&cans in two cases; in the remaining 11 cases there was more than one additional microorganism. Eighty-eight (70.4%) episodes were hospital-acquired. The proportions of community-acquired (595%) and hospital-acquired (59.0%) bacteraemias due to E. fuecium were
J. Gray et al.
182 Table
II.
Occurrence
of 125
episodes
Specialty
Source
bacteraemia
No.
Haematology/Oncology Intensive Therapy Neonatal Unit Paediatric surgery Medical specialties General medicine General surgery Paediatric medicine Obstetrics Surgical specialties
Table
of enterococcal associated mortality
III.
41 17 17 15 10 9 9 4 2 1
Unit
Sources
of infection
Central venous catheters Endocarditis Gastrointestinal tract Respiratory tract Skin and soft tissue Urogenital tract Uncertain
(%)
of 125 episodes
of enterococcal
cases
(32.8) (13.6) (13.6) (12.0) (8.0) (7.2) (7.2) (3.2) (1.6) (0.8)
bactevaemia
No.
(%) 44 2 24 1 6 9 39
cases
(35.2) (1.6) (19.2) (0.8) (4.8) (7.2) (31.2)
by clinical
specialty,
and
No. of deaths (mortality rate %) 4 (9.7) 6 (35.3) 6 (35.3) : 3 1 1 0 0
and associated
(10.0) (33.3) (11.1) (25.0)
mortality
No. of deaths (mortality rate %) 2 (4.5) 2 (100) 6 (25.0) i 2 (22.2) 10 (25.6)
similar. The mean duration of hospitalization prior to development of nosocomial enterococcal bacteraemia was 36.8 days (median 18 days). There was a preponderance of cases at the extremes of life, with 55 (44.0%) episodes occurring in children aged less than 1 year and 28 (22.4%) in patients aged over 60 years. Most infections occurred either in patients undergoing chemotherapy for haematological or other malignancies, or in intensive care settings (Table II). One hundred and nineteen (95.2%) of the patients had significant underlying illness. Eighteen patients (14.4%) had had previous bacteraemia or fungaemia during their hospital stay: E. faecium accounted for 16 of these 18 episodes. An underlying focus of infection was identified in 86 (68.8%) episodes (Table III). CVCs were much the commonest source of infection. The two cases of infective endocarditis occurred in patients with non-Hodgkin’s lymphoma, one of whom had a long-term CVC, and the other rheumatic mitral valve disease. Fifteen of the 39 episodes in which the focus of infection was not identified occurred in patients who were neutropenic: in many of these patients bacteraemia probably originated from the gastrointestinal tract.
Enterococcal
bacteraemia
183
Eleven patients had more than one episode of enterococcal bacteraemia. Four patients had an initial bacteraemia with ampicillin-sensitive enterococci followed within 1 month by infection with ampicillin-resistant E. fuecium, whilst another patient had apparently unrelated bacteraemias with E. fuecalis and E. faecium separated by more than 6 months. The other six patients had recurrent episodes of CVC-related bacteraemia ultimately necessitating removal of their catheters; one of the patients experienced a further three episodes of enterococcemia associated with the replacement catheter. Ninety-four (75.2(Y) o patients were pyrexial (temperature 338°C) at the time of blood culture; none was hypothermic. Two patients had a systolic blood pressure ~90 mmHg, both of whom died within 48 h of onset of bacteraemia. One had E. fuecalis infection, and the other polymicrobial bacteraemia with E. fuecium and Klebsiellu. White blood cell counts were available in 113 episodes. Forty-eight (38.4%) patients had leucocytosis (>lO x lo9 l-l), and 30 (24.0?‘) 0 were neutropenic (cl x 1 O9 1-i). Twenty-two (17.6%) patients died during hospitalization, of whom 10 (8.0%) were considered to have died as a direct result of bacteraemia. There was an increased mortality in ITU (P=O.O4) and NNU (P=O*O4) patients (Table II), whilst CVC-related infections were associated with a significantly reduced risk of death (Table III, P=O.Ol). Clinical factors that did not influence survival included age, presence of neutropenia or leucocytosis, presence of pyrexia, occurrence of polymicrobial bacteraemia, and a history of previous bacteraemia (data not shown). There was no significant difference in the mortalities of patients with bacteraemia with E. faeculis (15.9%) or E. fuecium (17.6%, P=O.S). Seventy-five (60.0%) patients had received antibiotic therapy over the 2 weeks prior to developing enterococcal bacteraemia. The mortality in these patients (25.30/) o was greater than in patients who did not receive prior antibiotic therapy (6.0%, P=O.Ol). 0 ne hundred and ten (88.0%) patients received empirical antibiotic therapy at the time of suspected bacteraemia: this included activity against the infecting enterococcus in 46 (36.8%) cases. There were no significant differences in the mortalities of patients receiving appropriate (15.2%), inappropriate (20.30/) o or no (13.3%) empirical therapy (P=O.7). One hundred and five (84.0%) patients ultimately received appropriate therapy. Of the remainder, three responded to removal of CVCs, therapy was withheld from two patients whose prognosis was too poor to justify active management, Four recovered without treatment, and 11 received inappropriate antimicrobial therapy. The mortality rate of patients who ultimately received appropriate antimicrobial therapy was 17-l%, compared with 20.0% for those who did not (P= 0.8). Five of the 19 (26.3%) patients who received synergistic two-drug regimens died compared with 15.1% of those who were treated with a single appropriate antibiotic (P=O.3). Forty-six (36.8%) patients received appropriate empirical antibiotic
184
J. Gray
et al.
therapy at the time of suspected bacteraemia, whilst 105 (84.0%) patients ultimately received appropriate therapy. Mortality was not influenced by either empirical or definitive antibiotic therapy. There were no significant differences in the mortality rates of patients receiving appropriate, inappropriate or no anti-enterococcal therapy either empirically or definitively. Discussion
Previous studies have used varying criteria for identifying cases of significant enterococcal bacteraemia. In our prospective study we used a less stringent definition than some others10’“,‘7 and included all episodes where there was strong clinical evidence for enterococcal infection, irrespective of the number of positive blood cultures or whether a focus of infection was confirmed microbiologically. We believe our approach accurately reflects clinical practice, and do not think it significantly over-estimates the incidence of true bacteraemia. Indeed, the sub-group of 58 episodes meeting stricter criteria for clinical significance differed from the study overall only in having a significantly lower frequency of polymicrobial bacteraemia (data not shown). Most studies of enterococcal bacteraemia have found E. faecalis to be the predominant species, accounting for over 85% of isolates.7~‘0~11~15~16 To some extent the preponderance of E. faecium in our study is due to the emergence of this organism as a frequent pathogen in paediatric oncology patients in our hospital:21 E. faecium accounted for 24 of the 3 1 enterococcus isolates from this unit. However, E. faecium also predominated in the remainder of the hospital, accounting for 52 (54.2%) of 96 isolates. Our findings accord with a small study by Ruoff and colleagues22 who identified five out of 10 consecutive blood culture isolates of enterococci as E. faecium. There was no common antibiotic susceptibility pattern amongst the isolates to suggest that widespread dissemination of a single strain of E. faecium was occurring. We have previously reported that faecal carriage of ampicillinresistant E. faecium occurs more frequently in patients with recent exposure to antibiotics,23 and likewise 63.2% of patients with bacteraemia with E. faecium had received previous antibiotics (PcO.05). In previous series fewer than 5% of blood culture isolates of enterococci have usually been found to be ampicillin-resistant.7*11~13~24The much higher prevalence of ampicillin resistance in our study (33.1%) reflects the unusually high proportion of bacteraemias due to E. faecium, which is more resistant to ampicillin than E. faecaZis.12’18The reported frequency of high-level gentamicin resistance in blood culture isolates of enterococci has ranged from O-62.7%.‘4z2”26 I n our study only 4.3% of 94 enterococci tested were highly gentamicin-resistant. As in previous studies,7,11,13,24 resistance to vancomycin was rarely encountered. The percentage of cases of bacteraemia which were polymicrobial was
Enterococcal
bacteraemia
185
37.6, which is consistent with previous reports of 18~2-47%.~11~‘5~‘6The proportion of bacteraemias that were hospital-acquired is also in agreement with previous reports of 52~7-77%.s~7~8~10~1*‘15,16 However, many of the patients in our series were frequent hospital inpatients, so that the true incidence of bacteraemia with enterococci of nosocomial origin may be higher. As in other studies, almost all patients had serious underlying disease. As reviewed by Graninger and Ragette’ the urinary and gastrointestinal tracts, and cutaneous wounds have been the principal sources of enterococcal bacteraemia in previous series, although CVC-related infections have been reported to account for up to 20% of enterococcemias.24 In our study the number of bacteraemias originating from CVCs was unusually high, despite the proportion of patients with these devices (68.8%) being comparable to previous reports of 69 and 86%.4,24 Ho wever, the 44 episodes of CVCrelated infection occurred in only 30 patients, with six patients experiencing between two and five episodes of CVC-related enterococcemia. All six had long-term surgically-implanted CVCs, leading to an understandable reluctance to remove the catheters promptly. The overall mortality in our study was 17.6%, compared with 29-57% in other recent studies.““,15-“,25’26 The mortality attributed directly to infection was 8%, again rather lower than previous reports of 12-43%.7,‘o,24,26 The low mortality in our study is partly explained by the high frequency of CVC-related infections: only two of the 45 patients with CVC-related infections died, neither as a direct result of sepsis (P=O.O2). Excluding episodes due to CVC-related infections, the mortality rates were 25% overall, and 12.5% due to infection. Previous studies have revealed no clinical characteristics that are consistently associated with a higher mortality rate.’ We observed an increased mortality in ITU and NNU patients and in patients whose enterococcemia was preceded by antibiotic therapy, while CVC-related infections were associated with a reduced mortality. Sixty per cent of patients had received previous antimicrobial therapy, compared with previous reports of 42-98%.4’7-11 DeCelis and colleagues17 reported that overall mortality was significantly lower in patients who received appropriate antimicrobial therapy at the onset of bacteraemia. However we found no association between outcome and antibiotic prescribing. Eighty-four per cent of the patients in our series ultimately received appropriate antimicrobial therapy, which compares favourably with previous reports of 69~3-89~1%.5~7-11~‘7 Whilst some workers have observed that patients who receive appropriate antimicrobial therapy have an increased survival,5,9-11 we, like others,“-’ found no such effect. As in previous studies7Y9-11we observed no benefit from synergistic two-drug regimens in patients without endocarditis. In conclusion, our study confirms the importance of enterococcal bacteraemia in patients with serious underlying disease. The predominance of E. fuecium in our series is in contrast to the experience of most others.
186
J. Gray
et al.
However, there were few noteworthy differences between characteristics of our patients and those in previous studies.
the clinical
References
:: 3. 4. 5. 6. 7. 8. 9 10. 11. 12. 13. 14.
15. 16.
17. 18.
19. 20. 21. 22. 23. 24. 25.
26.
Megran DW. Enterococcal endocarditis. Clin infect Dis 1992; 15: 63-71. Wheeler SM, Foley GE. A note on non-group-A streptococci associated with human infections. J Bacterial 1943; 46: 391-392. Murray BE. The life and times of the enterococcus. Clin Microbial Rev 1990; 3: 46-65. Barrall DT, Kenney PR, Slotman GJ, Burchard KW. Enterococcal bacteremia in surgical patients. Arch Surg 1985; 120: 57-63. Bryan CS, Reynolds KL, Brown J J. Mortality associated with enterococcal bacteremia. Surg Gynecol Obstet 1985; 160: 557-561. Garrison RN, Fry DE, Berberich S, Polk HC. Enterococcal bacteremia. Clinical implications and determinants of death. Ann Surg 1982; 196: 43-47. Gullberg RM, Homann SR, Phair JP. Enterococcal bacteremia: analysis of 75 episodes. Rev Infect Dis 1989; 11: 74-85. Malone DA, Wagner RA, Myers JP, Watanakunakorn C. Enterococcal bacteremia in two large community teaching hospitals. Am J Med 1986; 81: 601-606. Graninger W. Ragette R. Nosocomial bacteremia due to Enterococcus fuecalis without endocarditis.‘Clii Infect Disease 1992; 15: 49-57. Hoge CW, Adams J, Buchanan B, Sears SD. Enterococcal bacteremia: to treat or not to treat, a reappraisal. Rev Infect Dis 1991; 13: 600-605. Maki DG, Agger WA. Enterococcal bacteremia: clinical features, the risk of endocarditis, and management. Medicine 1988; 67: 248-269. Gray JW, Pedler SJ. Antibiotic-resistant enterococci. J Ho@ Infect 1992; 21: l-14. Watanakunakorn C. Enterococci from blood cultures during 1980-I 989: susceptibility to ampicillin, penicillin and vancomycin. J Antimicrob Chemother 1990; 26: 602-604. Watanakunakorn C. Rapid increase in the prevalence of high-level aminoglycoside resistance among enterococci isolated from blood cultures during 1989-I 991. r Antimicrob Chemother 1992; 30: 289-293. Shlaes DM, Levy J, Wolinsky E. Enterococcal bacteremia without endocarditis. Arch Intern Med 1981; 141: 578-581. Watanakunakorn C, Pate1 R. Comparison of patients with enterococcal bacteraemia due to strains with and without high-level resistance to gentamicin. Clin Infect Dis 1993; 17: 74-78. DeCelis G, Pallares R, Gudiol F, Ariza J, Nogues FF. Role of appropriate antimicrobial therapy in outcome of enterococcal bacteremia. Am J Med 1987; 82: 1283. Gray JE, Stewart D, Pedler SJ. Species identification and antibiotic susceptibility testing of enterococci isolated from hospitalized patients. Antimicrob Agents Chemother 1991; 35: 1943-194s. Facklam RR. Collins MD. Identification of Enterococcus suecies isolated from human infections by’s conventional test scheme. J Clin Microbial i989; 27: 731-734. Working Party of the BSAC. Breakpoints in in-vitro antibiotic sensitivity testing. J Antimicrob Chemother 1988; 21: 701-710. Gray JW, Pedler SJ, Craft AW, Kernahan J, Windebank KP, Pearson ADJ. Changing causes of septicaemia in paediatric oncology patients. Eur J Pediutr 1994; 153: 84-89. Ruoff KL, de la Maza L, Murtagh MJ, Spargo JD, Ferraro MJ. Species identities of enterococci isolated from clinical specimens. J Clin Microbial 1990; 28: 435-437. Gray JW, Stewart D, Pedler SJ. Faecal carriage of antibiotic-resistant enterococci in hospitalised and non-hospitalised individuals. JAntimicrob Chemother 1992; 30: 1 IO-1 12. Rimailho A, Lamp1 E, Riou B, Richard C, Rottman E, Auzepy P. Enterococcal bacteremia in a medical intensive care unit. Crit Cure Med 1988; 16: 126-129. Huycke MM, Spiegel CA, Gilmore MS. Bacteremia caused by hemolytic high-level gentamicin-resistant Enterococcus faecalis. Antimicrob Agents Chemother 1991; 35: 16261634. Noskin GA, Till M, Patterson BK, Clarke JT, Warren JR. High-level gentamicin resistance in Enterococcus faecalis bacteremia. J Infect Dis 1991; 164: 1212-I 215.
of Hospital
Infection
Enterococcal
27,
179-186
bacteraemia: a prospective 125 episodes
J. Gray, Department
(1994)
P. J. Marsh,
D. Stewart
and
study
S. J. Pedler
of Microbiology, Royal Victoria Infirmary, Queen Victoria Newcastle upon Tyne NE1 4LP, UK Accepted
for publication
12 May
of
Road,
1994
Summary: One hundred and twenty-five episodes of enterococcal bacteraemia occurring over a SO-month period were studied prospectively. Enterococcus faecium was the commonest species, accounting for 76 (59.8%) of the 127 isolates. Overall, 33.1% of isolates were resistant to ampicillin and one isolate (0.8%) to vancomycin; high-level gentamicin resistance was detected in 4.3% of 93 isolates tested. The percentage of nosocomial episodes was 70.4, and 95.2% of the patients had significant underlying illness. Central venous catheters (CVCs) were the commonest source of infection. Eightyfour per cent of episodes were ultimately treated with appropriate antibiotics. The overall mortality rate was 17.6%, and that directly attributable to infection was 8.0%. An increased mortality rate was observed in intensive care and neonatal unit patients, and in patients who had received antimicrobial therapy in the 2 weeks prior to enterococcemia. CVC-related infections were associated with a reduced mortality. No other clinical or microbiological factors were found to influence outcome. Keywords:
Enterococcus;
bacteraemia.
Introduction The importance of enterococci as a cause of 520% of cases of endocarditis is well-established.’ However controversy still surrounds the significance and optimal therapy of enterococcal bacteraemia without endocarditis. A high mortality associated with enterococcal bacteraemia has been longrecognized.2 However patients frequently have serious underlying disease, and enterococci often occur in mixed infections, thus raising doubts as to their pathogenic role.3 There are conflicting reports of the influence of In some series specific antiantimicrobial treatment on prognosis. enterococcal therapy has not been associated with increased surviva1,4-8 and several investigators have suggested that synergistic two-drug regimens are probably unnecessary.7,9-” The frequency of enterococcal bacteraemia is increasing3’” due to factors Correspondence Road, Newcastle 0195-67Ol/94/070179+08
to: Dr upon
J. Gray, Department Tyne, NE1 4LP.
of Microbiology,
$0X.00/0
Royal
Victoria
Infirmary,
0 1994 The Hospital
179
Queen
lnfemon
Victoria
Society
180
J. Gray et al.
such as increased numbers of compromised patients and widespread use of antibiotics, such as cephalosporins, to which enterococci are intrinsically resistant.3’12 Moreover, enterococci that are resistent to ampicillin or to the glycopeptides,‘2”3 or highly-resistant to aminoglycosides,‘2’14 are becoming increasingly common. Infections with such organisms present serious therapeutic difficulties. There have been several studies of enterococcal bacteraemia from North America, 4-8~10~11,15,16 but there are few reports of enterococcal bacteraemia from Europe.‘,17 We have previously reported finding an unusually high frequency of bacteraemia with Enterococcus f~eciurn.‘~ The purpose of this study was to evaluate the epidemiology, clinical significance and treatment of enterococcal bacteraemia in our hospital. Materials
and methods
Setting The Royal Victoria Infirmary is an SSO-bed teaching hospital and tertiary referral centre in North-East England. Specialist services within the hospital include renal dialysis and transplantation, bone marrow transplantation, the regional paediatric oncology unit, a neonatal unit (NNU), and adult paediatric intensive therapy units (ITUs).
Specimens Paired aerobic and anaerobic blood cultures were examined with a BACTEC USA). Cultures were NR730 analyser (Becton Dickinson, Cockeysville, incubated for 7 days before being discarded as negative. Enterococci were identified using standard microbiological methods.” Susceptibility to ampicillin and vancomycin was tested by the break-point method.20 In addition, 93 isolates (E. fuecalis 34; E.faecium 56; E. rafJinosus 2; E. avium 1) were tested for high-level resistance to gentamicin [minimum inhibitory concentration (MIC) >SOOmgl-‘1 by the same method. l3-Lactamase production was tested for by the rapid chromogenic cephalosporin method (nitrocefin; Oxoid Ltd, Basingstoke, UK).
Recording
and analysis of data Clinical and microbiological data on all episodes of clinically-significant enterococcal bacteraemia occurring between 1 August 1989 and 30 September 1993 were recorded prospectively, and subsequently analysed, using the ‘Epi info 5’ database and statistical package (Centers for Disease Control, Atlanta, USA). Statistical analysis was performed by the x2 test, with Yates’ correction where appropriate. Bacteraemia was regarded as hospital-acquired where the initial positive blood culture was obtained more than 48 h after admission. The source of bacteraemia was determined on the basis of microbiological or clinical
Enterococcal
bacteraemia
181
Table I. Species distribution of 127 enterococci isolated from 125 episodes of bacteraemia, and the frequency of resistance to ampicillin [minimum inhibitory concentration (MIC) >8 mg I-‘] and vancomycin (MIC, >4 mg I-‘) No. Species
Enterococcus faecium E. faecalis E. E. raffinosus avium E. durans Total
No. (X) isolates
(“5~) resistant
to
of Amplicillin
76 (59.8) 46 (36.2)
Vancomycin 0 1 (2.2)
2 (1.6) 1 (0.8)
01 (50.0) 0
127
42 (33.1)
P, 0 1 (0.8)
evidence of a focus of infection. Bacteraemia was judged to be related to central venous catheters (CVCs) when it coincided with symptoms related to use of the device, or where microorganisms of the same species, and with the same antibiotic susceptibility pattern, were isolated simultaneously from blood cultures and the device exit-site or cannula tip. Infective endocarditis was defined according to the criteria of Hoge and colleagues.” Deaths that occurred within 7 days of the first positive blood culture were considered to be directly due to infection, unless there was strong evidence to the contrary. Antimicrobial therapy was deemed appropriate if the spectrum of activity of the chosen regimen included activity against the infecting organism. Results
One hundred and twenty-five episodes of clinically-significant enterococcal bacteraemia occurred in 101 patients over the SO-month period. A further seven episodes of uncertain significance were excluded from the analysis. Seventy-four (59.2%) of the episodes were due to E.faecium and 44 (35.2%) to E. fuecalis, whilst E. fuecium and E. fuecalis were isolated simultaneously in two episodes. There were five bacteraemias with other Enterococcus species (Table I). Overall, 33.1% of isolates were resistant to ampicillin (MIC >8 mg l-l), but only one isolate of E. faecalis was resistant to vancomycin (MIC >4mg111), (Table I). Of 93 enterococci tested, four isolates of E. faecalis were highly resistant to gentamicin. P-Lactamase production was not detected in any of the isolates. In 47 (37.6(Y)o cases the bacteraemia was polymicrobial. The additional microorganisms were Gram-positive bacteria in 14 episodes, Gram-negative bacteria in 20, and Candida al&cans in two cases; in the remaining 11 cases there was more than one additional microorganism. Eighty-eight (70.4%) episodes were hospital-acquired. The proportions of community-acquired (595%) and hospital-acquired (59.0%) bacteraemias due to E. fuecium were
J. Gray et al.
182 Table
II.
Occurrence
of 125
episodes
Specialty
Source
bacteraemia
No.
Haematology/Oncology Intensive Therapy Neonatal Unit Paediatric surgery Medical specialties General medicine General surgery Paediatric medicine Obstetrics Surgical specialties
Table
of enterococcal associated mortality
III.
41 17 17 15 10 9 9 4 2 1
Unit
Sources
of infection
Central venous catheters Endocarditis Gastrointestinal tract Respiratory tract Skin and soft tissue Urogenital tract Uncertain
(%)
of 125 episodes
of enterococcal
cases
(32.8) (13.6) (13.6) (12.0) (8.0) (7.2) (7.2) (3.2) (1.6) (0.8)
bactevaemia
No.
(%) 44 2 24 1 6 9 39
cases
(35.2) (1.6) (19.2) (0.8) (4.8) (7.2) (31.2)
by clinical
specialty,
and
No. of deaths (mortality rate %) 4 (9.7) 6 (35.3) 6 (35.3) : 3 1 1 0 0
and associated
(10.0) (33.3) (11.1) (25.0)
mortality
No. of deaths (mortality rate %) 2 (4.5) 2 (100) 6 (25.0) i 2 (22.2) 10 (25.6)
similar. The mean duration of hospitalization prior to development of nosocomial enterococcal bacteraemia was 36.8 days (median 18 days). There was a preponderance of cases at the extremes of life, with 55 (44.0%) episodes occurring in children aged less than 1 year and 28 (22.4%) in patients aged over 60 years. Most infections occurred either in patients undergoing chemotherapy for haematological or other malignancies, or in intensive care settings (Table II). One hundred and nineteen (95.2%) of the patients had significant underlying illness. Eighteen patients (14.4%) had had previous bacteraemia or fungaemia during their hospital stay: E. faecium accounted for 16 of these 18 episodes. An underlying focus of infection was identified in 86 (68.8%) episodes (Table III). CVCs were much the commonest source of infection. The two cases of infective endocarditis occurred in patients with non-Hodgkin’s lymphoma, one of whom had a long-term CVC, and the other rheumatic mitral valve disease. Fifteen of the 39 episodes in which the focus of infection was not identified occurred in patients who were neutropenic: in many of these patients bacteraemia probably originated from the gastrointestinal tract.
Enterococcal
bacteraemia
183
Eleven patients had more than one episode of enterococcal bacteraemia. Four patients had an initial bacteraemia with ampicillin-sensitive enterococci followed within 1 month by infection with ampicillin-resistant E. fuecium, whilst another patient had apparently unrelated bacteraemias with E. fuecalis and E. faecium separated by more than 6 months. The other six patients had recurrent episodes of CVC-related bacteraemia ultimately necessitating removal of their catheters; one of the patients experienced a further three episodes of enterococcemia associated with the replacement catheter. Ninety-four (75.2(Y) o patients were pyrexial (temperature 338°C) at the time of blood culture; none was hypothermic. Two patients had a systolic blood pressure ~90 mmHg, both of whom died within 48 h of onset of bacteraemia. One had E. fuecalis infection, and the other polymicrobial bacteraemia with E. fuecium and Klebsiellu. White blood cell counts were available in 113 episodes. Forty-eight (38.4%) patients had leucocytosis (>lO x lo9 l-l), and 30 (24.0?‘) 0 were neutropenic (cl x 1 O9 1-i). Twenty-two (17.6%) patients died during hospitalization, of whom 10 (8.0%) were considered to have died as a direct result of bacteraemia. There was an increased mortality in ITU (P=O.O4) and NNU (P=O*O4) patients (Table II), whilst CVC-related infections were associated with a significantly reduced risk of death (Table III, P=O.Ol). Clinical factors that did not influence survival included age, presence of neutropenia or leucocytosis, presence of pyrexia, occurrence of polymicrobial bacteraemia, and a history of previous bacteraemia (data not shown). There was no significant difference in the mortalities of patients with bacteraemia with E. faeculis (15.9%) or E. fuecium (17.6%, P=O.S). Seventy-five (60.0%) patients had received antibiotic therapy over the 2 weeks prior to developing enterococcal bacteraemia. The mortality in these patients (25.30/) o was greater than in patients who did not receive prior antibiotic therapy (6.0%, P=O.Ol). 0 ne hundred and ten (88.0%) patients received empirical antibiotic therapy at the time of suspected bacteraemia: this included activity against the infecting enterococcus in 46 (36.8%) cases. There were no significant differences in the mortalities of patients receiving appropriate (15.2%), inappropriate (20.30/) o or no (13.3%) empirical therapy (P=O.7). One hundred and five (84.0%) patients ultimately received appropriate therapy. Of the remainder, three responded to removal of CVCs, therapy was withheld from two patients whose prognosis was too poor to justify active management, Four recovered without treatment, and 11 received inappropriate antimicrobial therapy. The mortality rate of patients who ultimately received appropriate antimicrobial therapy was 17-l%, compared with 20.0% for those who did not (P= 0.8). Five of the 19 (26.3%) patients who received synergistic two-drug regimens died compared with 15.1% of those who were treated with a single appropriate antibiotic (P=O.3). Forty-six (36.8%) patients received appropriate empirical antibiotic
184
J. Gray
et al.
therapy at the time of suspected bacteraemia, whilst 105 (84.0%) patients ultimately received appropriate therapy. Mortality was not influenced by either empirical or definitive antibiotic therapy. There were no significant differences in the mortality rates of patients receiving appropriate, inappropriate or no anti-enterococcal therapy either empirically or definitively. Discussion
Previous studies have used varying criteria for identifying cases of significant enterococcal bacteraemia. In our prospective study we used a less stringent definition than some others10’“,‘7 and included all episodes where there was strong clinical evidence for enterococcal infection, irrespective of the number of positive blood cultures or whether a focus of infection was confirmed microbiologically. We believe our approach accurately reflects clinical practice, and do not think it significantly over-estimates the incidence of true bacteraemia. Indeed, the sub-group of 58 episodes meeting stricter criteria for clinical significance differed from the study overall only in having a significantly lower frequency of polymicrobial bacteraemia (data not shown). Most studies of enterococcal bacteraemia have found E. faecalis to be the predominant species, accounting for over 85% of isolates.7~‘0~11~15~16 To some extent the preponderance of E. faecium in our study is due to the emergence of this organism as a frequent pathogen in paediatric oncology patients in our hospital:21 E. faecium accounted for 24 of the 3 1 enterococcus isolates from this unit. However, E. faecium also predominated in the remainder of the hospital, accounting for 52 (54.2%) of 96 isolates. Our findings accord with a small study by Ruoff and colleagues22 who identified five out of 10 consecutive blood culture isolates of enterococci as E. faecium. There was no common antibiotic susceptibility pattern amongst the isolates to suggest that widespread dissemination of a single strain of E. faecium was occurring. We have previously reported that faecal carriage of ampicillinresistant E. faecium occurs more frequently in patients with recent exposure to antibiotics,23 and likewise 63.2% of patients with bacteraemia with E. faecium had received previous antibiotics (PcO.05). In previous series fewer than 5% of blood culture isolates of enterococci have usually been found to be ampicillin-resistant.7*11~13~24The much higher prevalence of ampicillin resistance in our study (33.1%) reflects the unusually high proportion of bacteraemias due to E. faecium, which is more resistant to ampicillin than E. faecaZis.12’18The reported frequency of high-level gentamicin resistance in blood culture isolates of enterococci has ranged from O-62.7%.‘4z2”26 I n our study only 4.3% of 94 enterococci tested were highly gentamicin-resistant. As in previous studies,7,11,13,24 resistance to vancomycin was rarely encountered. The percentage of cases of bacteraemia which were polymicrobial was
Enterococcal
bacteraemia
185
37.6, which is consistent with previous reports of 18~2-47%.~11~‘5~‘6The proportion of bacteraemias that were hospital-acquired is also in agreement with previous reports of 52~7-77%.s~7~8~10~1*‘15,16 However, many of the patients in our series were frequent hospital inpatients, so that the true incidence of bacteraemia with enterococci of nosocomial origin may be higher. As in other studies, almost all patients had serious underlying disease. As reviewed by Graninger and Ragette’ the urinary and gastrointestinal tracts, and cutaneous wounds have been the principal sources of enterococcal bacteraemia in previous series, although CVC-related infections have been reported to account for up to 20% of enterococcemias.24 In our study the number of bacteraemias originating from CVCs was unusually high, despite the proportion of patients with these devices (68.8%) being comparable to previous reports of 69 and 86%.4,24 Ho wever, the 44 episodes of CVCrelated infection occurred in only 30 patients, with six patients experiencing between two and five episodes of CVC-related enterococcemia. All six had long-term surgically-implanted CVCs, leading to an understandable reluctance to remove the catheters promptly. The overall mortality in our study was 17.6%, compared with 29-57% in other recent studies.““,15-“,25’26 The mortality attributed directly to infection was 8%, again rather lower than previous reports of 12-43%.7,‘o,24,26 The low mortality in our study is partly explained by the high frequency of CVC-related infections: only two of the 45 patients with CVC-related infections died, neither as a direct result of sepsis (P=O.O2). Excluding episodes due to CVC-related infections, the mortality rates were 25% overall, and 12.5% due to infection. Previous studies have revealed no clinical characteristics that are consistently associated with a higher mortality rate.’ We observed an increased mortality in ITU and NNU patients and in patients whose enterococcemia was preceded by antibiotic therapy, while CVC-related infections were associated with a reduced mortality. Sixty per cent of patients had received previous antimicrobial therapy, compared with previous reports of 42-98%.4’7-11 DeCelis and colleagues17 reported that overall mortality was significantly lower in patients who received appropriate antimicrobial therapy at the onset of bacteraemia. However we found no association between outcome and antibiotic prescribing. Eighty-four per cent of the patients in our series ultimately received appropriate antimicrobial therapy, which compares favourably with previous reports of 69~3-89~1%.5~7-11~‘7 Whilst some workers have observed that patients who receive appropriate antimicrobial therapy have an increased survival,5,9-11 we, like others,“-’ found no such effect. As in previous studies7Y9-11we observed no benefit from synergistic two-drug regimens in patients without endocarditis. In conclusion, our study confirms the importance of enterococcal bacteraemia in patients with serious underlying disease. The predominance of E. fuecium in our series is in contrast to the experience of most others.
186
J. Gray
et al.
However, there were few noteworthy differences between characteristics of our patients and those in previous studies.
the clinical
References
:: 3. 4. 5. 6. 7. 8. 9 10. 11. 12. 13. 14.
15. 16.
17. 18.
19. 20. 21. 22. 23. 24. 25.
26.
Megran DW. Enterococcal endocarditis. Clin infect Dis 1992; 15: 63-71. Wheeler SM, Foley GE. A note on non-group-A streptococci associated with human infections. J Bacterial 1943; 46: 391-392. Murray BE. The life and times of the enterococcus. Clin Microbial Rev 1990; 3: 46-65. Barrall DT, Kenney PR, Slotman GJ, Burchard KW. Enterococcal bacteremia in surgical patients. Arch Surg 1985; 120: 57-63. Bryan CS, Reynolds KL, Brown J J. Mortality associated with enterococcal bacteremia. Surg Gynecol Obstet 1985; 160: 557-561. Garrison RN, Fry DE, Berberich S, Polk HC. Enterococcal bacteremia. Clinical implications and determinants of death. Ann Surg 1982; 196: 43-47. Gullberg RM, Homann SR, Phair JP. Enterococcal bacteremia: analysis of 75 episodes. Rev Infect Dis 1989; 11: 74-85. Malone DA, Wagner RA, Myers JP, Watanakunakorn C. Enterococcal bacteremia in two large community teaching hospitals. Am J Med 1986; 81: 601-606. Graninger W. Ragette R. Nosocomial bacteremia due to Enterococcus fuecalis without endocarditis.‘Clii Infect Disease 1992; 15: 49-57. Hoge CW, Adams J, Buchanan B, Sears SD. Enterococcal bacteremia: to treat or not to treat, a reappraisal. Rev Infect Dis 1991; 13: 600-605. Maki DG, Agger WA. Enterococcal bacteremia: clinical features, the risk of endocarditis, and management. Medicine 1988; 67: 248-269. Gray JW, Pedler SJ. Antibiotic-resistant enterococci. J Ho@ Infect 1992; 21: l-14. Watanakunakorn C. Enterococci from blood cultures during 1980-I 989: susceptibility to ampicillin, penicillin and vancomycin. J Antimicrob Chemother 1990; 26: 602-604. Watanakunakorn C. Rapid increase in the prevalence of high-level aminoglycoside resistance among enterococci isolated from blood cultures during 1989-I 991. r Antimicrob Chemother 1992; 30: 289-293. Shlaes DM, Levy J, Wolinsky E. Enterococcal bacteremia without endocarditis. Arch Intern Med 1981; 141: 578-581. Watanakunakorn C, Pate1 R. Comparison of patients with enterococcal bacteraemia due to strains with and without high-level resistance to gentamicin. Clin Infect Dis 1993; 17: 74-78. DeCelis G, Pallares R, Gudiol F, Ariza J, Nogues FF. Role of appropriate antimicrobial therapy in outcome of enterococcal bacteremia. Am J Med 1987; 82: 1283. Gray JE, Stewart D, Pedler SJ. Species identification and antibiotic susceptibility testing of enterococci isolated from hospitalized patients. Antimicrob Agents Chemother 1991; 35: 1943-194s. Facklam RR. Collins MD. Identification of Enterococcus suecies isolated from human infections by’s conventional test scheme. J Clin Microbial i989; 27: 731-734. Working Party of the BSAC. Breakpoints in in-vitro antibiotic sensitivity testing. J Antimicrob Chemother 1988; 21: 701-710. Gray JW, Pedler SJ, Craft AW, Kernahan J, Windebank KP, Pearson ADJ. Changing causes of septicaemia in paediatric oncology patients. Eur J Pediutr 1994; 153: 84-89. Ruoff KL, de la Maza L, Murtagh MJ, Spargo JD, Ferraro MJ. Species identities of enterococci isolated from clinical specimens. J Clin Microbial 1990; 28: 435-437. Gray JW, Stewart D, Pedler SJ. Faecal carriage of antibiotic-resistant enterococci in hospitalised and non-hospitalised individuals. JAntimicrob Chemother 1992; 30: 1 IO-1 12. Rimailho A, Lamp1 E, Riou B, Richard C, Rottman E, Auzepy P. Enterococcal bacteremia in a medical intensive care unit. Crit Cure Med 1988; 16: 126-129. Huycke MM, Spiegel CA, Gilmore MS. Bacteremia caused by hemolytic high-level gentamicin-resistant Enterococcus faecalis. Antimicrob Agents Chemother 1991; 35: 16261634. Noskin GA, Till M, Patterson BK, Clarke JT, Warren JR. High-level gentamicin resistance in Enterococcus faecalis bacteremia. J Infect Dis 1991; 164: 1212-I 215.