|Home | About | Journals | Submit | Contact Us | Français|
This retrospective, case-control study aimed to identify variables associated with the incidence of Clostridium difficile–associated diarrhea (CDAD) in acute care facilities and to specifically identify the relationship of fluoroquinolones and acid suppressive agents in the development of CDAD. Seventy-one symptomatic patients positive for C. difficile toxin A or B hospitalized for at least 72 hours were compared with 142 control patients hospitalized for at least 72 hours who were not positive for C. difficile toxin A or B. Two controls were matched to one case patient for age within 5 years, unit of admission, and date of admission. The mean ages for cases and controls were 63.5 and 62.7 years, respectively. After adjusting for two confounding variables—hospital stay within 3 months and Charlson Comorbidity Index—conditional multiple logistic regression identified six risk factors for development of CDAD: gastrointestinal procedures within 60 days (odds ratio [OR] 9.1, P < 0.013), levofloxacin exposure (OR 8.2, P < 0.033), moxifloxacin exposure (OR 4.1, P < 0.026), imipenem exposure (OR 14.9, P < 0.014), laxative use (OR 20.2, P < 0.0001), and immunosuppressive use (OR 20.7, P < 0.034). The risk of CDAD after exposure to levofloxacin or moxifloxacin was not significantly different. Acid suppressive therapy was not a risk factor for CDAD development.
Outbreaks of Clostridium difficile–associated diarrhea (CDAD) have been reported in acute care as well as long-term care facilities (1, 2). Disruption of the normal gut flora allows colonization with C. difficile (3). C. difficile colonization in adult patients may increase by 10% to 30% during hospitalization, but not all patients will develop disease. Known risk factors associated with CDAD development include prior exposure to antimicrobials, gastrointestinal (GI) surgery, feeding tubes, chemotherapy, environmental exposure, advanced age, severity of comorbid conditions, and GI stimulants, stool softeners, and enemas (1, 2, 4, 5). Although the strongest evidence with antimicrobial exposure exists with clindamycin, penicillins, and cephalosporins (4–11), reports have implicated fluoroquinolone use as a potential risk factor in the development of CDAD (2, 11–20). Proton pump inhibitors (PPIs) are another pharmacologic class suggested as a risk factor in the development of CDAD (21, 22).
Hillcrest Medical Center is a licensed 557-bed tertiary care hospital located in northeast Oklahoma. This institution had experienced an apparent increase in CDAD cases over a 2-year period (years 2001–2003) with no defined cause, although a correlation with a formulary change from levofloxacin to moxifloxacin had been postulated. In addition, intravenous PPI therapy became available during this time period. The primary objective of this study was to evaluate the relationship of CDAD with levofloxacin or moxifloxacin use in acutely ill patients. Secondary study objectives included evaluating the relationship between CDAD and PPI use in acutely ill patients and describing the treatment regimens and outcomes of CDAD patients.
This study was conducted in the general medical wards and intensive care units (ICUs) at Hillcrest Medical Center. A retrospective chart review was conducted in patients with CDAD or at risk for CDAD in acute care areas who were admitted between August 1, 2001, and August 31, 2003. Levofloxacin was the formulary fluoroquinolone from August 2001 to June 2002, and moxifloxacin was utilized from July 2002 to the end of the study analysis. Patients were excluded from participation if they were discharged from the hospital in <72 hours, were <18 or >89 years of age, had a previous CDAD diagnosis, showed symptoms within rather than after 72 hours of hospitalization, were treated in nonacute care areas, or did not have a chart available from medical records. This study was approved by the institutional review boards of the University of Oklahoma Health Sciences Center and Hillcrest Medical Center. Medical charts included in the study were identified using microbiological data (identification of C. difficile toxin A/B during the study period; Wompole™ C. Diff QuikChek Complete™, Inverness Medical, Princeton, NJ) and admission data.
A retrospective case-control study design was used. Cases included patients with acute onset of loose bowel movements persisting for ≥2 days who tested positive for C. difficile toxin A and/or toxin B. The date of positive toxin test was defined as the date of diagnosis. For each case patient with CDAD, two control patients were matched for age (within 5 years), unit of admission, and month of admission. Controls had to have a minimum length of stay of 3 days. Data from a single case of CDAD for each patient were included in the study analysis. Data collected included patient age, gender, race, the presence of comorbidities, alcohol abuse or current use of cigarettes as identified in the social history, patient location at time of CDAD diagnosis, and diagnosis at admission. The following variables were documented and assessed as risk factors associated with exposure before the development of CDAD: recent hospital admission within the previous 90 days, admission to the ICU, length of hospital stay prior to CDAD diagnosis, total hospital length of stay, GI procedure or surgery within 60 days of illness or during admission, previous antimicrobial use (e.g., fluoroquinolones, vancomycin, clindamycin, penicillins, cephalosporins, metronidazole), duration of antimicrobial treatment, other medications (chemotherapy, immunosuppressives, PPIs, histamine-2 receptor antagonists [H2RAs]), administration of enteral or parenteral nutrition, and presence of a nasogastric or nasotracheal tube. Data were recorded until the patient was discharged from the hospital, transferred to a long-term care facility, or died.
Statistical analysis was performed using SAS software, version 9.1 (SAS Institute, Cary, NC) (23). Univariate analysis was performed separately for each of the variables. Variables with a P value of ≤0.2 in the univariate analysis were included in a conditional multiple logistic regression model, as were the indicator variables for prior exposure to moxifloxacin and levofloxacin since these were the primary variables of interest. A manual backward selection process was used. Risk factors were checked for confounding and colinearity. Confounders were included in multivariate models if covariate inclusion changed the coefficient of any statistically significant variable in the logistic regression model by 10% or greater. The odds ratios for prior exposure to moxifloxacin and levofloxacin were evaluated using a Z test comparing parameter estimates. All tests were two tailed, and a P value of ≤0.05 was considered significant in the multivariate model.
Based on microbiological data, 302 patients were identified as positive for C. difficile toxin A or B during the study period. Of these patients, 46 were excluded for prior C. difficile infection, 26 patients were symptomatic within 72 hours of hospitalization, 102 patients were admitted to the rehabilitation or long-term care units, 27 patients did not meet the age requirements, and 30 patients were unable to be matched to the two required control patients. Data were collected and analyzed for 71 case and 142 control patients.
Only three patient characteristics were significantly different between the case and control groups: recent hospitalization, GI procedure in the last 60 days, and GI procedure in the hospital (Table (Table11). The most commonly performed GI procedures for both groups included colonoscopy, exploratory laparotomy, esophagogastroduodenoscopy, and percutaneous endoscopic gastrostomy tube placement. The use of several types of medication was associated with CDAD, but PPIs and H2RAs were not among them (Table (Table22).
In the logistic regression model controlling for hospital stay within the previous 3 months and Charlson Comorbidity Index, six variables were found to increase the risk of CDAD (Table (Table33). The risk of developing CDAD after moxifloxacin or levofloxacin exposure was not significantly different. The case patients who received levofloxacin therapy received it primarily on an outpatient basis where a majority of patients who received moxifloxacin were treated on an inpatient basis. Treatment of the CDAD episode for case patients included metronidazole therapy in 89% and withdrawal of the suspected causative agent in 11%. As shown in Table Table44, patients who developed CDAD had longer lengths of stay for both the ICU and total hospitalization. Mortality was similar in the case and control patients.
Our study suggests that there is an increased risk of CDAD after exposure to moxifloxacin or levofloxacin in patients who develop symptoms and test positive for C. difficile toxin A or B after 72 hours of admission to the hospital. No statistically significant difference in risk of CDAD development was identified between moxifloxacin exposure and levofloxacin exposure. The hypothesis that moxifloxacin could increase CDAD rates as compared to levofloxacin is due to its greater propensity to affect anaerobic organisms (24). Several case reports and reviews have identified this potential risk (25). Our findings are similar to a recent case-control study of outpatients that found comparable risks of hospitalization for CDAD with gatifloxacin, moxifloxacin, and levofloxacin (26). Two studies have also postulated an increase in CDAD related to a change in formulary fluoroquinolone, although only one of these studies was able to control the outbreak by switching back to the original fluoroquinolone (2, 27). A large retrospective case-control study completed within the Veterans Administration (VA) documented that the increased CDAD rates observed were not associated with fluoroquinolone formulary change to gatifloxacin but with seasonal variation in CDAD rates (28). Another case-control study in the VA system confirmed that an increase in CDAD rates was not associated with the addition of gatifloxacin to the formulary (29). A prospective study of nosocomial CDAD in several Canadian institutions found an increased risk with ciprofloxacin, levofloxacin, and gatifloxacin in a multivariate model (30). Continuing research in this area suggests that fluoroquinolones in general are risk factors for CDAD development (11, 31–33).
The effect of acid suppressive therapy on rates of CDAD has recently been debated (34). In our study, no association was identified between PPIs or H2RAs and increased risk of CDAD. An explanation for this in our study may be the low number of patients using these agents in the outpatient setting and the possibility that increased risk may be related to prolonged use. Several other studies have found no link between acid suppressive therapy and CDAD, including a VA study (28), a recent case-control trial of community-dwelling outpatients (35), a study of hospitalized patients (32), and a prospective study in several institutions in Canada (30). In contrast, a large community-based case-control trial found that the use of acid suppressants, especially PPIs, as well as nonsteroidal anti-inflammatory agents was associated with an increased risk for development of CDAD (36), and a retrospective case-control study in Kansas identified an increased risk of CDAD with exposure to PPI therapy (37). In all, the data remain inconclusive but confirm that PPIs should be screened for inappropriate use and discontinued in those situations to avoid potential adverse sequelae (38).
An interesting finding in our study was the increased risk of CDAD after GI procedures completed within the previous 60 days. The ability to associate the development of CDAD with the procedure itself versus the agents used prior to the procedures (i.e., electrolyte preparation solutions or topical antimicrobial agents) is limited by the lack of outpatient information. A proportion of patients in this study also underwent inpatient GI procedures, raising the question of whether these patients had inflammatory bowel disease that was previously undiagnosed and not captured due to study design. In prior studies, there appears to be a link with preexisting intestinal conditions or procedures and CDAD development (28, 39). The increased risk of CDAD with laxative exposure is not a new finding (4).
The increased risk of CDAD with imipenem/cilastatin found in our study is supported by evidence from other trials in patients with immunosuppression. A significant increase in CDAD occurred with imipenem/cilastatin compared with clinafloxacin for empiric treatment of febrile neutropenia (40). A similar finding of increased CDAD in neutropenic patients was found with imipenem and vancomycin compared with cefoperazone-sulbactam and vancomycin therapy for suspected or documented infections (41). Compared with ceftazidime, use of imipenem in febrile neutropenic patients was associated with greater gastrointestinal toxicity, including CDAD and nausea and vomiting (42). A recent case-control study of CDAD in hospitalized patients also identified that imipenem/cilastatin was associated with increased rates of CDAD (43).
The identification of those patients most at risk for CDAD is paramount in the prevention and treatment of this disease. Scoring models for CDAD have been developed to identify those patients who may be at risk (44, 45). The Waterlow score, used for identifying the risk of developing pressure ulcers, has also been linked to the risk of CDAD (44). Also utilized is the clinical risk scoring model, which has four variables (age, hemodialysis, surgical admission status, and ICU length of stay) and identifies those patients at risk for CDAD (45). These scoring models, when used appropriately, may help guide physicians when prescribing broad-spectrum antimicrobial therapy, ordering GI procedures, and implementing preventative CDAD procedures.
Due to the design of this study, several major limitations must be addressed. This review was conducted at a single institution. Our study focused on patients who may have acquired CDAD after hospitalization in an effort to capture inpatient exposures to medications or other risk factors, which limits the ability to generalize these results to other patient populations. Due to the retrospective nature of the study, the agents selected for treatment for the initial suspected infection or CDAD could not be controlled. In addition, the complete duration of therapy of the various antimicrobial, immunosuppressive, laxative, and acid-reducing agents could not be specified with the current design, as only the inpatient duration could be captured. While patients with CDAD receive isolation control including handwashing, gowning, and gloving as standard procedures in this institution, these variables are not routinely documented in the chart once isolation procedures are ordered. Another limitation to the study is the small number of patients who received moxifloxacin and levofloxacin during the study period, thus limiting the power of the study. This reduction in power may have resulted in the inability to find a significant difference between fluoroquinolone exposures. The number of potential patients was greatly reduced by excluding patients who were symptomatic within the initial 72 hours of hospitalization or who had previous CDAD and the inability to match case patients with controls meeting the specified criteria. The strains of C. difficile were not sent for epidemiologic typing to identify outbreak isolates, nor was antibiotic susceptibility testing completed for any of the isolates. This study is also limited because there could have been some changes in practice or differences in risk of CDAD due to other factors (seasonal or random differences in outbreaks) that could not be controlled.
In conclusion, in patients with CDAD identified after at least 72 hours of hospitalization, an increased risk is present after exposure to levofloxacin, moxifloxacin, imipenem/cilastatin, laxative use, immunosuppressive use, and GI procedures within the previous 60 days. The risk of CDAD exposure was not significantly different between moxifloxacin or levofloxacin exposure. Exposure to PPIs or H2RAs did not increase the risk of CDAD development.