Antimicrobial Use and Risk for Recurrent Clostridium difficile Infection

Transcription

1 CLINICAL RESEARCH STUDY Antimicrobial Use and Risk for Recurrent Clostridium difficile Infection Dimitri M. Drekonja, MD, MS, a,b William H. Amundson, BA, c Douglas D. DeCarolis, PharmD, a,d Michael A. Kuskowski, PhD, a,e Frank A. Lederle, MD, a,b James R. Johnson, MD a,b a Minneapolis Veterans Affairs Health Care System, Minneapolis, Minn; b Department of Medicine, University of Minnesota, Minneapolis; c University of Minnesota Medical School, Minneapolis; d College of Pharmacy, University of Minnesota, Minneapolis; e Department of Psychiatry, University of Minnesota, Minneapolis. ABSTRACT BACKGROUND: Although antimicrobial use during and immediately after Clostridium difficile infection (CDI) is discouraged, the frequency and consequences of such use are poorly defined. We sought to determine the frequency of non-cdi antimicrobial therapy during and after treatment for CDI, and the association of such therapy with recurrent disease. METHODS: Retrospective review of all CDI cases at a Veterans Affairs medical center from Outcomes were non-cdi antimicrobial use during and within 30 days after completing CDI treatment, and recurrent CDI. RESULTS: From 2004 to 2006, new-onset CDI occurred in 249 unique patients. No follow-up information was available for 3 patients, leaving 246 as study subjects. Of these, 141 (57%) received non-cdi antimicrobials, including 61 (25%) who received non-cdi antimicrobials during CDI treatment, and 80 (33%) who received non-cdi antimicrobial therapy after CDI treatment. With adjustment for age, disease severity, duration of CDI treatment, and recent hospital or intensive-care unit stay, receipt of non-cdi antimicrobials after CDI treatment was significantly associated with recurrent CDI (odds ratio [OR] 3.02; 95% confidence interval [CI], ), compared with no antimicrobial use. Antimicrobial use during CDI treatment was not associated with recurrent CDI (OR 0.79; 95% CI, ). Neither number of antimicrobial courses nor antimicrobial days was associated with recurrence. CONCLUSIONS: Non-CDI antimicrobial therapy after an episode of CDI is common and is associated with a 3-fold increase in the odds of recurrent disease. The added risk associated with antimicrobial exposure (regardless of duration) should be considered if such therapy is contemplated. Published by Elsevier Inc. The American Journal of Medicine (2011) 124, 1081.e e7 KEYWORDS: Antimicrobial use; Clostridium difficile infection; Recurrent CDI Clostridium difficile infection (CDI) is increasing in frequency and severity, 1-3 and recurrence is both common and difficult to treat. 4 Antimicrobial use is linked to CDI, preceding up to 92% of CDI episodes. 5 The normal fecal microbiota are thought to protect against CDI, 6 and even a Funding: The study was supported by the resources of the Minneapolis Veterans Affairs Health Care System, including the Center for Epidemiological and Clinical Research. Conflict of Interest: All authors have no disclosures to report. Authorship: All authors had access to the data and a role in writing the manuscript. Requests for reprints should be addressed to Dimitri M. Drekonja, MD, MS, Infectious Diseases (111F), Veterans Affairs Medical Center, 1 Veterans Drive, Minneapolis, MN address: drek0002@umn.edu single systemic antimicrobial dose can substantially alter these microbiota, 7 enhancing the ability of C. difficile to cause disease. In mild cases of CDI, withdrawal of the inciting antimicrobial may be adequate treatment, 8,9 whereas more severe cases require administration of a C. difficile-active agent, in addition to cessation of the inciting antimicrobial, if possible. 4,10 Because continuing the inciting antimicrobial predisposes to recurrent disease, avoidance of unnecessary antimicrobials in patients with recent CDI has been recommended for preventing recurrence. 4,10,12-15 The 3 prior studies that assessed non-cdi antimicrobial use during or after CDI treatment as a predictor of recurrent CDI all found an association with recurrent disease /$ -see front matter Published by Elsevier Inc. doi: /j.amjmed

2 1081.e2 The American Journal of Medicine, Vol 124, No 11, November 2011 However, limitations include analyzing only 7 episodes of recurrence, 11 not specifying the frequency of non-cdi antimicrobial use, 12 and including only a fraction of all CDI cases occurring during the study period. 13 Furthermore, no study differentiated between non-cdi antimicrobial use during versus after CDI treatment. To obtain a more robust assessment of the risk of recurrent CDI associated with non-cdi antimicrobial use, we retrospectively reviewed consecutive cases of CDI at a large Veterans Affairs (VA) medical center over a 3-year period to determine the frequency of non-cdi antimicrobial use during or after CDI treatment and whether such therapy was associated with an increased risk of recurrent CDI. We also examined recurrent CDI risk in relation to the extent of non-cdi antimicrobial exposure. METHODS CLINICAL SIGNIFICANCE Subjects The Minneapolis Veterans Affairs Medical Center (MVAMC) pharmacy service maintained a database of all newly diagnosed CDI cases from January 2004 through December Systematic collection of new CDI cases did not occur before or after this time. Potential cases were identified by comparing pharmacy records for use of drugs active against C. difficile (metronidazole or oral vancomycin) with microbiology laboratory records of positive C. difficile tests. The case definition for CDI was: 1) 4 unformed bowel movements per day or liquid colostomy output of 1300 ml per day over 2 days; or paralytic ileus, unexplained abdominal pain, or leukocytosis; plus 2) a positive C. difficile culture or toxin result or visualization of pseudomembranes on colonoscopy or histopathology. Identification of potential cases (by comparing pharmacy records to laboratory data) and medical record review to assess for presence of clinical inclusion criteria were performed previously by one of the authors (DDD) in conjunction with 2 pharmacy residents. Exclusionary criteria included presence of another explanation for diarrhea and a diagnosis of CDI within the preceding 30 days. Data Extraction and Definitions For the present study, subjects previously identified by the pharmacy service as having new-onset CDI during the study period underwent more extensive retrospective medical record review by one of the authors (WHA or DMD) using the VA computerized patient record system. For subjects with multiple entries in the CDI database, the earliest was used. Abstracted data included age, sex, and primary service Antimicrobial use during and after Clostridium difficile infection is discouraged; how often such use occurs in clinical practice is unknown. Non-C. difficile antimicrobial use during and 30 days after C. difficile infection was common (57%) in this cohort of 246 patients from a large Veterans Affairs hospital. After adjustment for relevant covariates, antimicrobial use after, but not during, C. difficile infection was significantly associated with recurrent disease. (medicine, surgery, or rehabilitation). Because communityassociated CDI has been defined previously as symptom onset occurring more than 12 weeks after the last discharge from a health care facility, 16 we assessed whether subjects had been hospitalized in the preceding 12 weeks, and if so, whether they required treatment in an intensive care unit (ICU). Potential CDI risk factors were recorded, including gastric acid-suppressive therapy, immune-suppressive medications or conditions, and tube feeding. 4,17 Antimicrobial use within 60 days before the index CDI episode was recorded. The severity of the subject s underlying illness was assessed using a modified Horn s index, which is a 4-point scale of disease severity, with 1 mild, 2 moderate, 3 severe, and 4 extremely severe. 18 Details of treatment for the index CDI episode were recorded, including drug, dose, and duration. Details of non-cdi antimicrobial therapy received during or within 30 days following completion of CDI treatment also were recorded. An antimicrobial course was defined by the presence of an order to start a specific antimicrobial, regardless of duration. If multiple antimicrobials were administered over the same time period, each drug counted as a separate course. Non-CDI antimicrobial use was classified in relation to its timing relative to the period of CDI therapy. Because ongoing CDI therapy was anticipated to provide some protection against CDI if co-administered with non-cdi antimicrobials, non- CDI antimicrobials were presumed to be less likely to precipitate recurrence if used during rather than after completion of CDI treatment. Accordingly, non-cdi antimicrobial use was classified as follows: use concurrent with CDI treatment was classified as during, whereas any use after completion of CDI treatment (regardless of use during ) was sufficient to classify the subject as having received antimicrobials after CDI treatment. The extent of non- CDI antimicrobial use was recorded as both the number of courses and the number of antimicrobial days (defined as days during which at least one non-cdi antimicrobial was ordered). 19 Outcomes Assessed outcomes included non-cdi antimicrobial use and recurrent CDI, defined as renewed fulfillment of the case definition after resolution of the initial manifestations of CDI and completion of the initial course of CDI treatment. Recurrence was assessed during the 90 days after completion of CDI treatment.

3 Drekonja et al Antimicrobials During and After CDI 1081.e3 Table 1 Characteristics of 246 Veterans with Clostridium difficile Infection (CDI) at the Minneapolis Veterans Affairs Medical Center from 2004 through 2006 in Relation to non-cdi Antimicrobial Therapy During and After CDI Treatment Category Specific Characteristic Any Non-CDI Antimicrobial Therapy During or Within 30 Days After CDI Treatment Overall (n 246) Yes (n 141) No (n 105) P Value, Yes vs No* Demographics Age, mean ( SD), years 71 ( 13) 70 ( 13) 72 ( 13).44 Male, n (%) 241 (98) 140 (99) 101 (96).17 Site of CDI acquisition, n (%) Community 24 (10) 4 (3) 20 (19).001 Hospital 222 (90) 137 (97) 84 (81).001 Primary service Medicine 204 (83) 113 (80) 91 (87).23 Other 42 (17) 28 (20) 14 (13).23 CDI risk factors, n (%) Hospitalized in prior 12 weeks 222 (90) 137 (97) 85 (81).001 ICU stay in prior 12 weeks 56 (23) 41 (29) 15 (14).01 PPI use 150 (61) 89 (63) 61 (58).43 H2 blocker use 27 (11) 21 (15) 6 (6).024 Other antacid use 29 (12) 14 (10) 15 (14).32 Any antacid use 187 (76) 117 (83) 70 (67).004 Tube feeding 19 (8) 14 (10) 5 (5).15 Immune suppression 30 (12) 21 (15) 9 (9).17 Disease severity Horn index, mean ( SD) 2.65 ( 0.62) 2.79 ( 0.56) 2.48 ( 0.65).001 Treatment duration, index CDI episode Days, mean ( SD) 15.2 (7.7) 16.8 (9.4) 13.3 (3.7).001 CDI Clostridium difficile infection; H2 type-2 histamine receptor; ICU intensive care unit; PPI proton pump inhibitor; SD standard deviation. *P values (by Fisher s exact test for categorical variables, and Student s t test for continuous variables) are for comparisons between subjects receiving antimicrobials during or after CDI vs those receiving none. Other includes all surgical services, neurology, and physical medicine & rehabilitation. Risk factors were assessed at time treatment was initiated for the index CDI episode. Horn index measure of underlying disease severity at completion of CDI therapy, with 1 mild, 2 moderate, 3 severe, and 4 extremely severe. Statistical Methods Continuous variables were compared using a 2-tailed t test. Proportions were compared using Fisher s exact test. Univariable logistic regression was used to evaluate antimicrobial use and other variables for associations with recurrent CDI. Multivariable logistic regression was used to adjust for the effects of other clinical characteristics, including age, severity of underlying disease, recent hospitalization, recent ICU stay, treatment duration for the index CDI episode, and use of acid-suppressive medication, on the potential association of different antimicrobial exposures with recurrent CDI. The association between type of non-cdi antimicrobial and recurrent CDI was assessed by testing the association between recurrent CDI and exposure to each individual antimicrobial class. The study protocol was approved by the MVAMC institutional review board. RESULTS Study Population The pharmacy service s CDI database contained 265 entries for patients treated for CDI at the MVAMC between January 2004 and December Of these, 16 represented repeat entries for patients already listed in the database, so were excluded, leaving 249 unique patients/episodes. Follow-up information was unavailable for 3 of these, leaving 246 unique evaluable subjects as the study population. Overall, subjects were elderly and predominantly male (Table 1). Ninety percent had been hospitalized within the 12 weeks before onset of CDI. Documented antimicrobial use in the 60 days preceding CDI diagnosis was nearly universal (96%). The typical severity of underlying illness was moderate to severe. A large percentage of subjects were receiving gastric acid-suppressive therapy, usually a protonpump inhibitor (PPI), whereas relatively few were receiving immune-suppressive medications or tube feedings. Median treatment duration for the index CDI episode was 14 days (range 1-64). Metronidazole alone was the treatment for the index CDI episode in 202 subjects (82%), whereas vancomycin alone was used for only one episode. In 43 cases (17%), metronidazole and vancomycin were both administered, usually sequentially (in either order), but occasionally together. Metronidazole was administered as 500 mg orally thrice daily for 231 subjects (94%); 14 subjects received the same dose intravenously. Vancomycin was administered orally 4 times daily at 2 dosages: 125 mg in 36 subjects (15%), and 500 mg in 8 subjects (3%). Vancomycin enemas were used for 6 subjects (2%), all of whom also received oral vancomycin.

4 1081.e4 The American Journal of Medicine, Vol 124, No 11, November 2011 Non-CDI Antimicrobial Use Of the 246 subjects, 141 (57%; 95% confidence interval [CI], 51%-64%) received non-cdi antimicrobial therapy during or after completion of CDI treatment; 87% of this use was administered while the subjects were hospitalized. Non-CDI antimicrobial use occurred during CDI treatment in 61 subjects (25%; 95% CI, 20%-31%), and after completion of CDI therapy in 80 subjects (33%; 95% CI, 27%- 39%). Within the latter group, 43 subjects (17% of total; 95% CI, 13%-23%) received non-cdi antimicrobials both during and after CDI treatment. Subjects who received non-cdi antimicrobials during or after CDI treatment differed significantly from non-antimicrobial-using subjects according to multiple characteristics (Table 1). Specifically, they were more likely to have been hospitalized or to have had an ICU stay in the prior 12 weeks and to be taking a type-2 histamine receptor blocker or using any antacid. They also received longer-duration treatment for the index CDI episode, and had more severe underlying disease. In contrast, among non-cdi antimicrobial-receiving subjects, no significant differences in patient characteristics were evident according to the timing of the non-cdi antimicrobial use, that is, during versus after CDI treatment (data not shown). The 141 subjects with non-cdi antimicrobial use during or after CDI treatment collectively received 445 antimicrobial courses, totaling 2147 days of therapy. Extent of use varied considerably with respect to both the number of treatment courses (median 2; range 1-17) and total number of antimicrobial days (median 12 days; range 1-49). Recurrent CDI Recurrent CDI within 90 days after completing (and responding favorably to) the initial CDI treatment course was documented for 74 (30%; 95% CI, 24%-36%) of the 246 subjects. The median time to recurrence was 10 days (range 3-74), with 92% of recurrence episodes occurring within 8 weeks of initial symptom onset. Recurrence was significantly associated with non-cdi antimicrobial use, occurring in 40% (95% CI, 32%- 49%) of subjects who received antimicrobials during or after CDI treatment, compared with only 16% (95% CI, 10%-25%) of those who did not, for an odds ratio (OR) for recurrence of 3.51 (95% CI, ). Among the 38 subjects who developed recurrent CDI following exposure to non-cdi antimicrobials after CDI treatment, recurrence onset was during or within 1 week of exposure to non-cdi antimicrobials in 30 (79%), and 7 days after the non-cdi antimicrobial exposure in the remaining 8 (21%). Assessment of the risk of recurrent CDI in relation to the timing of non-cdi antimicrobial use (relative to the timing of CDI treatment) identified recurrence in 31% (95% CI, 20%-44%) of subjects who received non-cdi antimicrobials only during CDI treatment (vs no antimicrobials: OR 2.32; 95% CI, ), and in 48% (95% CI, 36%-59%) of those who received antimicrobials after CDI treatment (vs no antimicrobials: OR 4.63; 95% CI, ) (Table 2). In contrast, among non-cdi antimicrobial-receiving subjects, the risk of recurrence was unrelated to the extent of such antimicrobial therapy, whether assessed as number of courses (OR 0.96; 95% CI, ) or days of antimicrobial therapy (OR 0.99; 95% CI, ). The relationship between type of non-cdi antimicrobial and recurrent CDI risk was assessed by evaluating exposure to different antimicrobial classes. Because many subjects (106/ 141; 75%) were exposed to at least 2 antimicrobials, the multiple possible antimicrobial combinations limited our ability to model the risk of recurrent CDI associated with each antimicrobial combination. Accordingly, the association between recurrent CDI and exposure to each individual antimicrobial class was assessed in an exploratory analysis, which showed a significant association of recurrent CDI only with beta-lactams and fluoroquinolones (Table 3). The only nonantimicrobial study variable significantly associated with recurrence was recent hospitalization, with a recurrence rate of 73/222 (33%), versus 1/24 (4%) for no Table 2 Unadjusted and Adjusted* Odds Ratios for Recurrent Clostridium difficile Infection (CDI) According to Post-CDI Antimicrobial Exposure among 246 Subjects with CDI in the Preceding 90 Days Type of Non-CDI Antimicrobial Exposure n Recurrent CDI n (%) Odds Ratio for Recurrent CDI (95% CI) Unadjusted Adjusted* None (16%) 1 (referent) 1 (referent) During (31%) 2.32 ( ) 0.79 ( ) After (48%) 4.63 ( ) 3.02 ( ) CDI Clostridium difficile infection; CI confidence interval. *Variables adjusted for were: age, Horn index, recent hospitalization, recent intensive care unit stay, duration of CDI treatment, and any antacid use. Defined as any non-cdi antimicrobial use occurring only concurrent with CDI treatment. Defined as any non-cdi antimicrobial use occurring within 30 days after completing treatment for CDI, regardless of any during use.

5 Drekonja et al Antimicrobials During and After CDI 1081.e5 Table 3 Recurrent Clostridium difficile Infection (CDI) among 246 Subjects with Current or Recent CDI, in Relation to Exposure to Specific non-cdi Antimicrobial Classes During or After the Index Episode of CDI* Antimicrobial No. of Subjects Exposed (% of 246) Recurrent CDI Proportion (%) of Subjects Exposed Unexposed P Value Beta-lactams 93 (38) 38/93 (41) 36/153 (24).006 Fluoroquinolones 75 (30) 34/75 (45) 40/171 (23).001 Vancomycin 55 (22) 18 (33) 56/191 (29) TMP/SMZ 20 (8) 9/20 (45) 65/226 (29) Antifungals 14 (6) 7/14 (50) 67/232 (29) Tetracyclines 13 (5) 6/13 (46) 98/233 (29) Aminoglycosides 10 (4) 0/10 (0) 74/236 (31) Macrolides 10 (4) 2/10 (20) 72/236 (31) Linezolid 5 (2) 3/5 (60) 71/241 (30) Metronidazole 5 (2) 0/5 (0) 74/241 (31) Clindamycin 4 (2) 2/4 (50) 72/242 (30) Rifampin 3 (1) 0/3 (0) 74/243 (31) Nitrofurantoin 2 (1) 1/2 (50) 73/244 (30) CDI Clostridium difficile infection; TMP/SMZ trimethoprim/sulfamethoxazole. Non-CDI antimicrobial exposure was assessed during and 30 days after CDI treatment. The total number of exposures is greater than the number of subjects who received any antimicrobials (n 141) because of subjects being exposed to multiple antimicrobials. Beta-lactams includes penicillins, beta-lactamase inhibitor combination agents, cephalosporins, and carbapenems. P values (by Fisher s exact test) are shown where P.01 (to account for multiple comparisons). Metronidazole use for conditions other than CDI was classified as non-cdi antimicrobial use. recent hospitalization (OR 11.27; 95% CI, ). In contrast, age, Horn index, any antacid use, duration of CDI treatment, and recent ICU stay were not significantly associated with recurrence (data not shown). Multivariable Analysis Next the association of non-cdi antimicrobial use during or after CDI treatment with recurrent CDI was assessed with use of multivariable logistic regression to adjust for recent hospitalization, age, recent ICU stay, any antacid use, duration of CDI treatment, and Horn index. After adjustment, non-cdi antimicrobial use during CDI treatment was not associated with increased episodes of recurrent CDI, compared with no antimicrobials (Table 2). In contrast, non-cdi antimicrobial use after CDI treatment remained significantly associated with recurrent CDI (OR 3.02; 95% CI, ). In a separate model, receipt of any non-cdi antimicrobials during or after CDI treatment also was significantly associated with recurrent disease (OR 2.80; 95% CI, ). The only non-antimicrobial variable that was independently associated with recurrence together with non-cdi antimicrobial use was recent hospitalization; this association was limited to the model involving non-cdi antimicrobials administered after CDI treatment (OR 9.28; 95% CI, ). Temporal Differences Neither the annual number of index CDI cases nor the by-year risk of recurrence showed a significant temporal trend over the 3-year study period, with 71 CDI cases occurring in 2004, 88 in 2005, and 87 in 2006, and corresponding annual recurrence rates of 38%, 27%, and 26%, respectively. DISCUSSION In this 3-year study of consecutive cases of new-onset CDI at a large referral VA medical center, we documented that most subjects (57%) received non-cdi antimicrobials during CDI therapy or within 30 days after completion, and that such use was significantly associated with recurrent CDI. Specifically, non-cdi antimicrobial use after CDI treatment was significantly associated with recurrence even after adjustment for relevant covariates, including recent hospitalization, duration of CDI treatment, and disease severity. These findings support and extend previous data implicating non-cdi antimicrobial use as a risk factor for recurrent CDI, demonstrate that such use is worrisomely frequent in a population at high risk for recurrent CDI, and suggest that post-cdi antimicrobial exposure of any duration carries a significant risk of recurrent CDI. Although avoidance of non-cdi antimicrobials during or after CDI is widely advocated, 10-13,15 the extent and timing of such use is poorly documented. In particular, few data

6 1081.e6 The American Journal of Medicine, Vol 124, No 11, November 2011 describe post-cdi antimicrobial use patterns in routine clinical practice, as opposed to that occurring within the context of prospective studies. 11,13,20 In such studies, non-cdi antimicrobial use could be artificially low due to a Hawthorne effect, whereby clinicians are especially vigilant about factors that could influence the study, even if they are not study team members themselves. 21 Accordingly, our retrospective observational study design may more accurately capture the frequency of non-cdi antimicrobial use during or after a CDI episode. Linsky et al recently demonstrated that non-cdi antimicrobial use modified the association between PPI use and recurrent CDI, with subjects who received both PPIs and non-cdi antimicrobials having increased rates of recurrent CDI. 22 However, the direct effect of non-cdi antimicrobials on recurrence was not assessed. Thus, our study assessing the association between non-cdi antimicrobials and recurrent CDI is the largest of which we are aware, and the first to stratify risk based on timing of antimicrobial exposure. Our findings demonstrate that a majority of patients with CDI in our institution receive non-cdi antimicrobial therapy during or after their CDI episode, and that such therapy sometimes entails multiple or lengthy treatment courses. We expected that reports of increasing incidence and severity of CDI 1,2,23 might raise awareness about the risk of recurrent CDI, and inspire more sparing use of antimicrobials in patients with recent CDI. On the contrary, the high frequency of non-cdi antimicrobial use is greater than that noted in prior studies (37%-47%), 11,13,20,22 even though we assessed such use only during and for 30 days after CDI treatment, whereas previous studies assessed use for 60 to 90 days. 11,13,20,22 The frequency of non-cdi antimicrobial use in this highrisk population raises the question of whether all such use was necessary. That antimicrobials are prescribed unnecessarily is well documented, including in our institution, with reported rates of inappropriate prescribing ranging from 20%-60% Thus, a portion of the antimicrobial use observed likely was avoidable. However, the size of the avoidable fraction remains undefined. Non-CDI antimicrobials given after CDI treatment remained significantly associated with recurrent CDI even after adjustment for other relevant risk factors, whereas non-cdi antimicrobials given during CDI treatment did not. The latter finding could be because 1) our study was underpowered to detect the association of recurrence with non- CDI antimicrobials given during CDI treatment, 2) concurrent CDI therapy may mitigate a portion of the risk for recurrent disease, or 3) there may be no association. Although the timing of antimicrobial exposure significantly influenced the recurrence risk, the intensity or duration of such exposure did not. That is, any exposure conferred the same risk of recurrent CDI as did receipt of multiple or prolonged antimicrobial courses. This finding is relevant for both perioperative prophylaxis and the use of empiric antimicrobial coverage to treat a possible, but undiagnosed, infection, both of which usually result in administration of relatively short courses of antimicrobials. The high recurrence rate after any antimicrobial exposure among patients with recent CDI is a compelling reason to categorically avoid antimicrobial therapy in such patients, unless a strong suspicion of serious infection necessitates rapid treatment, or prophylaxis is indicated for an urgently needed procedure. The non-cdi antimicrobials administered to subjects were predominantly beta-lactams, fluoroquinolones, and vancomycin. Although exposure to beta-lactams or fluoroquinolones was associated with recurrent CDI, these results should be interpreted with caution. Most subjects were exposed to more than one antimicrobial, many drug classes were used in only a small number of subjects, and overall sample size precluded an analysis to account for the multiple possible combinations of antimicrobial exposure. Study limitations include its single-center nature; other facilities, particularly non-va hospitals, may differ. Likewise, the period during which we assessed for recurrent CDI, 90 days, differs from the definition of recurrent CDI (occurring within 8 weeks of initiating therapy) recommended by McDonald et al, 16 but is similar to that used in other studies of recurrent CDI. 11,22 Also, we did not assess for medications prescribed outside the VA system, although such use has been reported to be minimal (0-2%). 30 Additionally, subjects receiving non-cdi antimicrobials could have differed from other subjects in ways other than those for which we could adjust. Finally, the absence of an antimicrobial stewardship program (such as exists at some other institutions) could have contributed to the seemingly high rate of observed antimicrobial use. However, per-patient antimicrobial usage is lower at the MVAMC than at similar VA facilities (written personal communication, Matthew Goetz, MD; September 2010), suggesting that non-cdi antimicrobial use may be as or more common elsewhere. In summary, in this 3-year retrospective cohort study of patients with CDI at the Minneapolis Veterans Affairs Medical Center from , we found that most subjects received non-cdi antimicrobial treatment during or soon after completing treatment for an index CDI episode. Non- CDI antimicrobial use after CDI treatment was associated with significantly increased odds of recurrent CDI, even with adjustment for other relevant risk factors. In this highrisk population, the added risk of recurrence must be weighed against the potential benefits of non-cdi antimicrobial therapy. ACKNOWLEDGMENTS The authors thank Lindsey Bervig, PharmD, and Jennifer Bolduc, PharmD, for their assistance in compiling the database of patients with Clostridium difficile infection. References 1. Pepin J, Valiquette L, Cossette B. Mortality attributable to nosocomial Clostridium difficile-associated disease during an epidemic caused by a hypervirulent strain in Quebec. CMAJ. 2005;173(9):

7 Drekonja et al Antimicrobials During and After CDI 1081.e7 2. McDonald LC, Killgore GE, Thompson A, et al. An epidemic, toxin gene-variant strain of Clostridium difficile. N Engl J Med. 2005; 353(23): Barbut F, Mastrantonio P, Delmee M, et al. Prospective study of Clostridium difficile infections in Europe with phenotypic and genotypic characterisation of the isolates. Clin Microbiol Infect. 2007; 13(11): Gerding DN, Muto CA, Owens RC Jr. Treatment of Clostridium difficile infection. Clin Infect Dis. 2008;46(Suppl 1):S32-S Gerding DN, Olson MM, Peterson LR, et al. Clostridium difficileassociated diarrhea and colitis in adults. A prospective case-controlled epidemiologic study. Arch Intern Med. 1986;146(1): Fekety R, Shah AB. Diagnosis and treatment of Clostridium difficile colitis. JAMA. 1993;269(1): Sullivan A, Edlund C, Nord CE. Effect of antimicrobial agents on the ecological balance of human microflora. Lancet Infect Dis. 2001;1(2): Olson MM, Shanholtzer CJ, Lee JT Jr, Gerding DN. Ten years of prospective Clostridium difficile-associated disease surveillance and treatment at the Minneapolis VA Medical Center, Infect Control Hosp Epidemiol. 1994;15(6): Teasley DG, Gerding DN, Olson MM, et al. Prospective randomised trial of metronidazole versus vancomycin for Clostridium-difficileassociated diarrhoea and colitis. Lancet. 1983;2(8358): Cohen SH, Gerding DN, Johnson S, et al. Clinical practice guidelines for Clostridium difficile infection in adults: 2010 update by the society for healthcare epidemiology of America (SHEA) and the infectious diseases society of America (IDSA). Infect Control Hosp Epidemiol. 2010;31(5): Nair S, Yadav D, Corpuz M, Pitchumoni CS. Clostridium difficile colitis: factors influencing treatment failure and relapse a prospective evaluation. Am J Gastroenterol. 1998;93(10): Fekety R, McFarland LV, Surawicz CM, et al. Recurrent Clostridium difficile diarrhea: characteristics of and risk factors for patients enrolled in a prospective, randomized, double-blinded trial. Clin Infect Dis. 1997;24(3): Hu MY, Katchar K, Kyne L, et al. Prospective derivation and validation of a clinical prediction rule for recurrent Clostridium difficile infection. Gastroenterology. 2009;136(4): Aslam S, Hamill RJ, Musher DM. Treatment of Clostridium difficileassociated disease: old therapies and new strategies. Lancet Infect Dis. 2005;5(9): Fekety R. Guidelines for the diagnosis and management of Clostridium difficile-associated diarrhea and colitis. American College of Gastroenterology, Practice Parameters Committee. Am J Gastroenterol. 1997;92(5): McDonald LC, Coignard B, Dubberke E, et al. Recommendations for surveillance of Clostridium difficile-associated disease. Infect Control Hosp Epidemiol. 2007;28(2): Kelly CP, Pothoulakis C, LaMont JT. Clostridium difficile colitis. N Engl J Med. 1994;330(4): Horn SD, Sharkey PD, Bertram DA. Measuring severity of illness: homogeneous case mix groups. Med Care. 1983;21(1): Berrington A. Antimicrobial prescribing in hospitals: be careful what you measure. J Antimicrob Chemother. 2009;65(1): Lowy I, Molrine DC, Leav BA, et al. Treatment with monoclonal antibodies against Clostridium difficile toxins. N Engl J Med. 2010; 362(3): McCarney R, Warner J, Iliffe S, et al. The Hawthorne Effect: a randomised, controlled trial. BMC Med Res Methodol. 2007;7: Linsky A, Gupta K, Lawler EV, et al. Proton pump inhibitors and risk for recurrent Clostridium difficile infection. Arch Intern Med. 2010; 170(9): Loo VG, Poirier L, Miller MA, et al. A predominantly clonal multiinstitutional outbreak of Clostridium difficile-associated diarrhea with high morbidity and mortality. N Engl J Med. 2005;353(23): Vanderweil SG, Tsai CL, Pelletier AJ, et al. Inappropriate use of antibiotics for acute asthma in United States emergency departments. Acad Emerg Med. 2008;15(8): Polgreen PM, Chen YY, Cavanaugh JE, et al. An outbreak of severe Clostridium difficile-associated disease possibly related to inappropriate antimicrobial therapy for community-acquired pneumonia. Infect Control Hosp Epidemiol. 2007;28(2): Mangione-Smith R, Wong L, Elliott MN, et al. Measuring the quality of antibiotic prescribing for upper respiratory infections and bronchitis in 5 US health plans. Arch Pediatr Adolesc Med. 2005;159(8): Gonzales R, Camargo CA Jr, MacKenzie T, et al. Antibiotic treatment of acute respiratory infections in acute care settings. Acad Emerg Med. 2006;13(3): Drekonja DM, Okoye NC, Kuskowski MA, Johnson JR. Appropriateness of urinary tract infection diagnosis and treatment duration. Arch Intern Med. 2010;170(5): Cope M, Cevallos ME, Cadle RM, et al. Inappropriate treatment of catheter-associated asymptomatic bacteriuria in a tertiary care hospital. Clin Infect Dis. 2009;48(9): Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol. 1997;50(1):