Is There a Relationship Between Vancomycin-Resistant Enterococcal Infection and Clostridium difficile Infection?

S206 Is There a Relationship Between Vancomycin-Resistant Enterococcal Infection and Clostridium difficile Infection? Dale N. Gerding From the Medical Service, Chicago Veterans Affairs Healthcare System Lakeside Division, and the Department of Medicine, Northwestern University Medical School, Chicago, Illinois The relationship between vancomycin or metronidazole treatment of Clostridium difficile associated diarrhea (CDAD) and the occurrence of vancomycin-resistant enterococcus (VRE) infection was investigated by review of 18 case-control studies. Fifteen (83%) of 18 studies found vancomycin use, days of use, or grams used to be significantly associated with VRE infection or colonization. Intravenous vancomycin use was a significant risk in nine of 10 studies, and oral vancomycin use was a significant risk in three of four studies that stratified risk by route of administration. Although statistically associated, oral vancomycin use was so infrequent (25% in the study with the most use) that it is likely to have only a minor influence as a risk for VRE infection or colonization when compared with the much more widespread use of intravenous vancomycin. Metronidazole exposure was found to be a significant risk in four of five studies that specifically assessed this variable, but the indication for metronidazole use was not specified. Risk factors for both VRE infection and CDAD include antimicrobial exposure, number of antimicrobials, days of antimicrobial use, specific agents (third-generation cephalosporins, clindamycin, and imipenem), patient age, length of hospitalization, severity of underlying illness, use of electronic rectal thermometers, enteral feedings, environmental contamination, and contamination of the hands of health care workers. of the meetings of the Infectious Disease Society of America, the Interscience Conference on Antimicrobial Agents and Che- motherapy, and the Society for Healthcare Epidemiology of America. No effort was made to include a comprehensive list of published abstracts. All species of VRE were included, and no attempt was made to analyze the reports on the basis of type of vancomycin resistance (van A, B, or C). Methods used to identify VRE were assumed to be appropriate. Studies of VRE infection, colonization, and bacteremia were included. Some studies included patients from single hospital units or wards, whereas other investigations included patients throughout the hospital; still other studies included patients from multi- ple hospitals. As expected, data available in abstracts were more limited than those available in published reports, thus restricting the number of abstracts suitable for inclusion. Selection was restricted to inclu- sion of reports and abstracts that included controls for comparison of risk factors. Criteria for selection of controls were variable but usually included patients exposed to the same environment at the same time who were not infected or colonized with VRE. Controls who had infection or colonization with vancomycin-susceptible enterococci were often selected. The relatively recent descriptions of nosocomial infections due to vancomycin-resistant enterococcus (VRE) have raised questions regarding their relationship to another nosocomial infection, Clostridium difficile associated diarrhea (CDAD). In particular, the question of whether treatment of C. difficile disease with oral vancomycin is a predisposing factor for VRE infection has been proposed. If oral vancomycin use is a risk factor, to what degree does it predispose to VRE infection? More recently, the use of metronidazole, another first-line treatment of C. difficile infection, has been implicated as a risk factor for VRE infection. Still other researchers have noted the similarities of the epidemiology of both infections in hospitalized patients and the possibility that the two infections share common risk factors. The purpose of this study was to review the literature regarding VRE infection in an attempt to better define the relationship between VRE infection and C. difficile infection and to compare the risks of both infections, specifically examining the risks of vancomycin and metronidazole use as well as the risk factors common to both infections. It was also the goal of this study to obtain sufficient data to arrive at consensus conclusions with regard to the above questions. Methods The literature on VRE was reviewed by using MEDLINE. Abstracts were identified by manually reviewing proceedings Grant support: United States Department of Veterans Affairs (Merit Review). Reprints or correspondence: Dr. Dale N. Gerding, Medical Service (111), VA Lakeside Medical Center, 333 East Huron Street, Chicago, Illinois 60611. Clinical Infectious Diseases 1997;25(Suppl 2):S206 10 This article is in the public domain. Results Vancomycin exposure. In a Belgian study [1], administration of oral glycopeptide antimicrobials to volunteers was shown to result in the emergence of VRE in the stool. In 1989, teicoplanin or vancomycin was administered orally to volunteers for 21 days; 13 (81%) of 16 volunteers who received teicoplanin were found to have VRE in their stools, and one (17%) of six volunteers who

CID 1997;25 (Suppl 2) Relation Between VRE Infection and CDAD S207 received vancomycin had VRE in stools at the end of the study. Metronidazole exposure. Five of the studies reviewed in All of the volunteers were VRE-negative before receiving glycopeptides. table 1 specifically assessed metronidazole use as a risk factor In 1992, the same investigators sampled stool specimens for VRE infection or colonization [9 11, 15, 18], and four of from 40 healthy volunteers who had not knowingly received glycopeptides; five studies found that the risk was significant when compared low numbers of VRE were found in 11 volunteers with controls. The studies did not differentiate oral from paren- (28%). The researchers indicated that avoparcin, a glycopeptide, teral use and did not indicate what the reason for metronidazole had been used in animal feeds in Europe, including Belgium, for use was (in particular, whether use was for the treatment of ú15 years. CDAD). Use of other antimicrobials active against anaerobic An early study of 100 patients in whom VRE was detected bacteria (imipenem/cilastatin and clindamycin) was also significantly in New York City, although uncontrolled, provided an indication associated with VRE infection or colonization in of the possible risk of vancomycin exposure [2]. Sixty- three studies [9, 11, 15], but two studies [10, 18] failed to three of 100 patients had some exposure to vancomycin prior detect such an association. to detection of VRE, 55 intravenously, 3 orally, and 5 orally Cephalosporin exposure. Six studies (table 1) assessed and intravenously. Only eight of 100 patients had exposure to cephalosporin use (usually third-generation cephalosporins) [6, oral vancomycin. Exposure to other antimicrobials included 8, 11, 12, 15, 18], and four of these studies found a statistically metronidazole (23% of patients) and third-generation cephalosporins significant association when compared with controls. (52% of patients). Exposure to all antimicrobials. Six studies assessed the Data from 18 reports and abstracts that contained case-control risk of VRE infection or colonization with exposure to any comparisons of rates of VRE infection or colonization and antimicrobial (table 1) [3, 5, 6, 8 10], and five studies found vancomycin exposure are shown in table 1 [3 20]. Studies that the risk of VRE infection or colonization was significantly included patients with bacteremia, other infections, and coloni- increased with antimicrobial use of any kind. zation (fecal or skin). Vancomycin use was classified as intravenous, VRE and C. difficile detected in the same patients. In an oral, or any if no distinction was made in the report. uncontrolled study, Jordens et al. [21] found C. difficile toxin Fifteen (83%) of 18 studies showed vancomycin use, days of in stool specimens from 4 (11%) of 35 patients colonized with use, or grams used to be significantly associated with VRE VRE. Shay et al. [9] found prior C. difficile infections in 10 infection or colonization when compared with controls. Intravenous of 46 patients with VRE bacteremia vs. one of 46 controls vancomycin use was a significant risk factor in nine of (OR, 12.5; 95% CI, 1.6 550). Seville et al. [22] screened stool 10 studies, and oral vancomycin use was a significant risk specimens submitted for C. difficile testing for VRE by using factor in three of four studies that stratified risk by administra- selective vancomycin-containing media and found that 16 of tion route. Any vancomycin use was a risk factor in six of 129 patients had VRE in their stools, but they did not comment eight studies that did not distinguish oral from intravenous use. on the frequency of C. difficile in the specimens. Shah and The studies did not indicate the reason for intravenous or oral colleagues [23] retrospectively analyzed 103 cases of CDAD vancomycin use. and 47 cases of VRE infection that occurred over the same 12- In the three studies that noted that the risk for VRE infection month period. They found 17 patients who had both infections or colonization was significantly associated with oral vancomy- (36% of patients with VRE infection had CDAD, and 17% of cin use and that provided the number of patients [6, 9, 10], few patients with CDAD had VRE infection). The incidence of patients were exposed to vancomycin by this route. For example, both infections was highly correlated (r Å.80; P õ.001), and Handwerger et al. [6] found that four of 16 patients vs. zero of both infections were associated with antimicrobial use and long 16 controls (P õ.05) received oral vancomycin compared with length of hospitalization. seven of eight patients vs. five of 16 controls (P õ.05) who Common risks for VRE and C. difficile infections. Factors received intravenous vancomycin. Similarly, Morris et al. [10] that have been found to be associated with VRE infection and found that four of 28 patients with VRE in urine vs. zero of 56 CDAD include antimicrobial exposure, number of antimicro- controls (P Å.01) had received oral vancomycin; however, 18 bial exposures, days of antimicrobial treatment, and specific of 20 patients vs. 24 of 40 controls (OR, 6.0; 95% CI, 1.1 59) antimicrobial agents (third-generation cephalosporins, clinda- had received intravenous vancomycin. In the largest of these mycin, and imipenem) [3, 5, 6, 8 11, 24]. Other factors observed studies, Shay et al. [9] found that 11 of 46 patients with VRE to be common to both infections include patient age, bacteremia vs. four of 46 controls had received oral vancomycin length of hospitalization, severity of underlying illness, use of (OR, 3.4; P Å.05), and the mean number of days of oral electronic rectal thermometers, enteral feedings, gastrointestivancomycin administration was 3.1 for patients vs. 0.5 for con- nal colonization, environmental contamination, and contamination trols (P Å.036). However, exposure to intravenous vancomycin of the hands of health care workers [7, 9, 10, 24 27]. occurred in 34 of 46 patients vs. 11 of 46 controls (OR, 12.5; 95% CI, 3.3 28), and the mean duration of intravenous vancomycin Discussion administration was significantly longer for patients than Review of the literature suggests that prior exposure to van- for controls (P õ.001). comycin is common in patients infected or colonized with VRE.

S208 Gerding CID 1997;25 (Suppl 2) Table 1. Summary of data on risk of exposure to antimicrobials for patients with VRE infection or colonization and controls. No. of Any VRE-positive VRE antimicrobial [Reference] patients condition Vancomycin use* Cephalosporin use* Metronidazole use* use* [3] 8 C 8/8 vs. 2/13 (P õ.002) (any)...... 8/8 vs. 5/13 (P õ.007) [4] 6 I, C 6/6 vs. 3/12 (P Å.01) (any)......... [5] 32 I 12/32 vs. 1/35 (OR, 20.2) (any)...... 30/32 vs. 16/35 (OR, 16.3) [6] 8; 16 I; I, C 7/8 vs. 5/16 (P õ.05) (iv); 4/16 vs....; 16/16 vs. 11/16...;......; 16/16 vs. 0/16 (P õ.05) (oral) (P õ.05) 13/16 (P Å NS) [7] 12 I, C?/12 vs.?/12 (OR, 6.0; P Å NS)......... (any) [8] 10 I, C?/10 vs.?/20 (P Å.002) (iv)?/10 vs.?/20...?/10 vs.?/20 (P Å.03) (P Å.04) [9] 46 B 34/46 vs. 11/46 (OR, 12.5; 95% CI,... 17/46 vs. 5/46 5.8 d vs. 2.7 d 3.3 28) (iv); 11/46 vs. 4/46 (OR, (OR, 4.8; (P õ.001) 3.4; 95% CI, 0.9 16) (oral); 3.1 d 95% CI, 1.5 18) vs. 0.5 d (P Å.036) (oral) [10] 20 I, C 18/20 vs. 24/40 (OR, 6.0; 95% CI,... 14/20 vs. 11/40 36.3 d vs. 1.1 59) (iv); 14.8 d vs. 6.2 d (OR, 6.2; 95% 11.5 d (P õ.0003) (iv); 4/28 vs. 0/56 CI, 1.6 24) (P õ.0002) (P Å.01) (oral) [11] 11 B 2/11 vs. 0/22 (P Å.11) (any) (P Å NS) (P Å.07; OR, ;... 95% CI, 2.5 ) [12] 11 C Significant (any) Significant...... [13] 22 C 13/22 vs. 1/12 (P õ.01) (iv)......... [14] 33 I 6/33 vs. 0/33 (P õ.01) (any)......... [15] 32 I Significant (iv) Significant Significant... [16] 27 I, C (OR, 13; 95% CI, 2 79) (any for ú2......... w) [17] õ86 C 56% vs. 11% (P õ.001) (iv)......... [18] 54 C 25/54 vs. 35/92 (P Å NS) (iv) 36/54 vs. 47/92 23/54 vs. 19/92... (P Å NS) (P õ.01) [19] 10 C 4.3 g vs. 1.4 g (P õ.003) (iv);......... insignificant (oral) [20] 20 B 16/20 vs. 10/29 (P õ.001) (iv)......... NOTE. B Å bacteremia; C Å colonization; I Å infection; NS Å not significant; VRE Å vancomycin-resistant enterococcus;? Å unknown;... Å data not evaluated. * No. of patients/total no. of patients vs. no. of controls/total no. of controls, mean duration of use for patients vs. mean duration of use for controls, percentage of patients vs. percentage of controls, or dose for patients vs. dose for controls (P value, OR, and/or 95% CI) (route of administration). increase the risk of VRE infection and an alternative agent (metronidazole) for the treatment of CDAD is available, recom- mendations to defer first-line use of vancomycin for the treatment of CDAD seem appropriate given the lack of any effective therapy for systemic VRE infection [28]. It is possible that some reduction in oral vancomycin use following the 1995 recommendations of the Hospital Infection Practices Advisory Committee [28] may already be reflected in data from some of the studies in table 1, particularly abstracts published in 1996; however, lag times of 1 year generally precede publica- tion of data, even in abstract form. Unfortunately, epidemiological studies of VRE have also identified metronidazole use as a risk factor for VRE infection or colonization, although the indication for use of metronida- zole was not given (i.e., use of metronidazole for the treatment The major contribution to risk of VRE infection or colonization is associated with intravenous (not oral) use of vancomycin. Although some studies did not distinguish the method of administration, those that did identified oral vancomycin exposure in only a few patients (25% of patients in the study with the most use), albeit at a significantly higher rate than among controls. Thus, it appears that whereas oral vancomycin is observed to be a significant risk in some studies (and may be of higher risk than intravenous vancomycin), the total exposure to oral vancomycin (because so few patients receive it compared with intravenous vancomycin) accounts for a small part of the total vancomycin exposure; this finding suggests that treatment of CDAD (assuming that oral vancomycin is used for the treatment of CDAD) is not a major contributor to the overall risk of VRE infection. Since any vancomycin exposure appears to

CID 1997;25 (Suppl 2) Relation Between VRE Infection and CDAD S209 of CDAD was not mentioned). In studies that compared both 8. Moreno F, Grota P, Crisp C, et al. Clinical and molecular epidemiology of vancomycin-resistant Enterococcus faecium during its emergence in vancomycin use and metronidazole use, metronidazole use was a city in southern Texas. Clin Infect Dis 1995;21:1234 7. more frequent than oral vancomycin use [9] and was approxi- 9. Shay DK, Maloney SA, Montecalvo M, et al. Epidemiology and mortality mately as frequent as intravenous vancomycin use [10, 18]. risk of vancomycin-resistant enterococcal bloodstream infections. J Infect Since other agents with activity against anaerobes (imipenem, Dis 1995;172:993 1000. clindamycin, and ampicillin/sulbactam) have also been associantimicrobial 10. Morris JG, Shay DK, Hebden JN, et al. Enterococci resistant to multiple agents, including vancomycin. Ann Intern Med 1995;123: ated with increased risk of VRE infection or colonization, it is 250 9. not clear that altering metronidazole use can impact rates of 11. Edmond MB, Ober JF, Weinbaum DL, et al. Vancomycin-resistant Entero- VRE infection or colonization if alternative agents for the treat- coccus faecium bacteremia: risk factors for infection. Clin Infect Dis ment of infections due to anaerobic bacteria are used. No alter- 1995;20:1126 33. native agents that are as effective as vancomycin and metronistudy 12. Weinstein JW, Roe M, Towns M, et al. Resistant enterococci: a prospective of prevalence, incidence, and factors associated with colonization dazole are available for the treatment of CDAD [24]. in a university hospital. Infect Control Hosp Epidemiol 1996;17: Although it is not likely that the specific use of vancomycin 36 41. or metronidazole for the treatment of CDAD is a major contrib- 13. Beezhold D, Slaughter S, Hayden MK, et al. Prevalence of skin colonization utor to the risk of VRE infection (because relatively few patients with vancomycin-resistant enterococci (VRE) in hospitalized pautor tients receive treatment for CDAD), there is a remarkably high with bacteremia [abstract no 108]. Clin Infect Dis 1995;21:738. 14. Meyers BR, Meyers J, Mendelson MH, Sheiner P, Miller C. Nosocomial number of risk factors that have been identified as common to infections (inf) with vancomycin resistant Enterococcus faecium both of these nosocomial infections. Factors that have long (VREF) in orthotopic liver transplant (OLT) patients (pts): risk factors been known to predispose patients to CDAD (such as older for inf and mortality in inf versus non-inf pts [abstract no J73]. In: age, long hospitalization, antimicrobial exposure, environmencrobial Program and abstracts of the 35th Interscience Conference on Antimi- tal contamination, contamination of the hands of personnel, Agents and Chemotherapy (San Francisco). Washington, DC: American Society for Microbiology, 1995. electronic rectal thermometers, and enteral feedings) are now 15. Currie JB, McNicholl IR, Hershow RC. Risk factors for infection with appearing in the list of risk factors for VRE infection. Infection vancomycin-resistant Enterococcus faecium at a VA Medical Center control and prevention strategies to reduce the incidence of [abstract no J77]. In: Program and abstracts of the 35th Interscience VRE infection and CDAD have also been remarkably similar. Conference on Antimicrobial Agents and Chemotherapy (San Fran- Both barrier precautions (gloves, gowns, and hand washing) cisco). Washington, DC: American Society for Microbiology, 1995. 16. Hauer K, Axelrod P, Fekete T, Woodwell J, Kostman J. Outbreak investiand reduced or restricted use of antimicrobials have been the gation of vancomycin-resistant enterococci: a case-control study [abmainstay of infection control efforts; however, each effort restract no 1170]. In: Program and abstracts of the 32nd Interscience quires high levels of compliance to be effective, and the chal- Conference on Antimicrobial Agents and Chemotherapy (Anaheim). lenge to all health care providers is to follow these recommen- Washington, DC: American Society for Microbiology, 1992. dations if we are to reduce the rates of both these increasingly 17. Byers KE, Titus MG, Alexander CH, et al. A hospital outbreak of vanco- mycin-resistant Enterococcus faecium (VRE) controlled with intensive common nosocomial infections [7, 24]. surveillance culturing and gown and glove isolation [abstract no J41]. In: Program and abstracts of the 35th Interscience Conference on Antimicrobial Agents and Chemotherapy (San Francisco). Washington, DC: References American Society for Microbiology, 1995. 18. Anglim AM, Byers KE, Anneski CJ, et al. Risk factors for VRE 1. Van der Auwera P, Pensart N, Korten V, Murray BE, Leclercq R. Influence colonization during an epidemic [abstract no 15]. In: Abstracts of of oral glycopeptides on the fecal flora of human volunteers: selection the 6th annual meeting of the Society for Healthcare Epidemiology of highly glycopeptide-resistant enterococci. J Infect Dis 1996;173: ofamerica.washington,dc:infectioncontrolandhospitalepide- 1129 36. miology, 1996;17(suppl):18. 2. Frieden TR, Munsiff SS, Low DE, et al. Emergence of vancomycin- 19. Shea KW, Fishbane S, Ruggian J, Domenico P, Maesaka JK, Cunha resistant enterococci in New York City. Lancet 1993;342:76 9. BA. Vancomycin resistant enterococci (VRE) colonization in an 3. Rubin LG, Tucci V, Cercenado E, Eliopoulos G, Isenberg HD. Vancomy- outpatient hemodialysis (HD) unit [abstract no 27]. In: Abstracts cin-resistant Enterococcus faecium in hospitalized children. Infect Con- of the 6th annual meeting of the Society for Healthcare Epidemioltrol Hosp Epidemiol 1992;13:700 5. ogy of America. Washington, DC: Infection Control and Hospital 4. Karanfil LV, Murphy M, Josephson A. A cluster of vancomycin-resistant Epidemiology, 1996;17(suppl):20. Enterococcus faecium in an intensive care unit. Infect Control Hosp 20. Stosor V, Postelnick M, Peterson LR, Noskin GA. Enterococcus faecium Epidemiol 1992;13:195 200. bacteremia: does vancomycin resistance make a difference? [abstract 5. Boyle JF, Soumakis SA, Rendo A, Herrington JA, Gianarkis DG. Epidemi- no J34]. In: Program and abstracts of the 36th Interscience Conference ologic analysis and genotypic characterization of a nosocomial outbreak on Antimicrobial Agents and Chemotherapy (New Orleans). Washingof vancomycin-resistant enterococci. J Clin Microbiol 1993;31: ton, DC: American Society for Microbiology, 1996. 1280 5. 21. Jordens JZ, Bates J, Griffiths DT. Faecal carriage and nosocomial spread of 6. Handwerger S, Raucher B, Altarac D, et al. Nosocomial outbreak due to vancomycin-resistant Enterococcus faecium. J Antimicrob Chemother Enterococcus faecium highly resistant to vancomycin, penicillin, and 1994;34:515 28. gentamicin. Clin Infect Dis 1993;16:750 5. 22. Seville MT, Carey RB, O Keefe JP. Vancomycin-resistant enterococcus 7. Boyce JM, Opal SM, Chow JW, et al. Outbreak of multidrug-resistant (VRE) infections and the prevalence of VRE in patients suspected to Enterococcus faecium with transferable vanb class vancomycin resis- have Clostridium difficile diarrhea (CDD) [abstract no J75]. In: Program tance. J Clin Microbiol 1994;32:1148 53. and abstracts of the 35th Interscience Conference on Antimicrobial

S210 Gerding CID 1997;25 (Suppl 2) Agents and Chemotherapy (San Francisco). Washington, DC: American 25. Livornese LL, Dias S, Samel C, et al. Hospital-acquired infection with Society for Microbiology, 1995. vancomycin-resistant Enterococcus faecium transmitted by electronic 23. Shah L, Johanson WG, Smith SM, Eng RHK, Dever LL. Epidemio- thermometers. Ann Intern Med 1992;117:112 6. logic correlation of the incidences of vancomycin-resistant Enterococcus 26. Noskin GA, Stosor V, Cooper I, Peterson LR. Recovery of vancomycin- faecium (VREF) infection and Clostridium difficile colitis resistant enterococci on fingertips and environmental surfaces. Infect in a VA Medical Center [abstract no 16]. In: Abstracts of the 6th Control Hosp Epidemiol 1995;16:577 81. annual meeting of the Society for Healthcare Epidemiology of 27. McFarland LV, Mulligan M, Kwok RYY, Stamm WE. Nosocomial acquisition America. Washington, DC: Infection Control and Hospital Epidemiology, of Clostridium difficile infection. N Engl J Med 1989;320: 1996;17(suppl):18. 204 10. 24. Gerding DN, Johnson S, Peterson LR, Mulligan ME, Silva J Jr. SHEA 28. Hospital Infection Control Practices Advisory Committee (HICPAC). Speposition paper. Clostridium difficile-associated diarrhea and colitis. In- cial report: recommendation for preventing the spread of vancomycin fect Control Hosp Epidemiol 1995;16:459 77. resistance. Infect Control Hosp Epidemiol 1995;16:105 13.