Ascertainment and classification of patients with community-associated Clostridium difficile infection, 2009 through 2011.
Chitnis AS, Holzbauer SM, Belflower RM, Winston LG, Bamberg WM, Lyons C, Farley MM, Dumyati GK, Wilson LE, Beldavs ZG, Dunn JR, Gould LH, MacCannell DR, Gerding DN, McDonald LC, Lessa FC. Epidemiology of Community-Associated Clostridium difficile Infection, 2009 Through 2011. JAMA Intern Med. 2013;173(14):1359-1367. doi:10.1001/jamainternmed.2013.7056
Clostridium difficile infection (CDI) has been increasingly reported among healthy individuals in the community. Recent data suggest that community-associated CDI represents one-third of all C difficile cases. The epidemiology and potential sources of C difficile in the community are not fully understood.
To determine epidemiological and clinical characteristics of community-associated CDI and to explore potential sources of C difficile acquisition in the community.
Design and Setting
Active population-based and laboratory-based CDI surveillance in 8 US states.
Medical records were reviewed and interviews performed to assess outpatient, household, and food exposures among patients with community-associated CDI (ie, toxin or molecular assay positive for C difficile and no overnight stay in a health care facility within 12 weeks). Molecular characterization of C difficile isolates was performed. Outpatient health care exposure in the prior 12 weeks among patients with community-associated CDI was a priori categorized into the following 3 levels: no exposure, low-level exposure (ie, outpatient visit with physician or dentist), or high-level exposure (ie, surgery, dialysis, emergency or urgent care visit, inpatient care with no overnight stay, or health care personnel with direct patient care).
Main Outcomes and Measures
Prevalence of outpatient health care exposure among patients with community-associated CDI and identification of potential sources of C difficile by level of outpatient health care exposure.
Of 984 patients with community-associated CDI, 353 (35.9%) did not receive antibiotics, 177 (18.0%) had no outpatient health care exposure, and 400 (40.7%) had low-level outpatient health care exposure. Thirty-one percent of patients without antibiotic exposure received proton pump inhibitors. Patients having CDI with no or low-level outpatient health care exposure were more likely to be exposed to infants younger than 1 year (P = .04) and to household members with active CDI (P = .05) compared with those having high-level outpatient health care exposure. No association between food exposure or animal exposure and level of outpatient health care exposure was observed. North American pulsed-field gel electrophoresis (NAP) 1 was the most common (21.7%) strain isolated; NAP7 and NAP8 were uncommon (6.7%).
Conclusions and Relevance
Most patients with community-associated CDI had recent outpatient health care exposure, and up to 36% would not be prevented by reduction of antibiotic use only. Our data support evaluation of additional strategies, including further examination of C difficile transmission in outpatient and household settings and reduction of proton pump inhibitor use.
Clostridium difficile is the most common cause of health care–associated infectious diarrhea.1Quiz Ref IDTraditional risk factors for C difficile infection (CDI) include antibiotic use, advanced age, and prior hospitalization.2 Since 2005, CDI has been increasingly reported among young, healthy individuals residing in the community.3- 6 An estimated 20% to 28% of CDI is community associated,5,7 with an incidence of 20 to 50 cases per 100 000 population in the United States,5 Sweden,7 and England.8 Previous studies3- 6 have shown that approximately 40% of patients acquiring community-associated CDI were not exposed to antibiotics, suggesting that additional factors may contribute to infection. Although C difficile has been isolated from soil, food, water, animals, asymptomatic infants, and health care environments, the role of these sources in community C difficile acquisition is not well understood.9 Understanding the importance of novel sources will help guide strategies to prevent community-associated CDI.
We interviewed patients with community-associated CDI identified through a longitudinal, population-based, surveillance program across 32 counties in 8 US states. We describe demographics, clinical characteristics, and outpatient exposures and outcomes and evaluate potential sources of acquisition of CDI in the community.
This project was approved by the institutional review boards at the Centers for Disease Control and Prevention and participating sites. Verbal consent was obtained from all patients interviewed.
In 2009, the Emerging Infections Program began active laboratory-based and population-based surveillance for CDI in select counties across 10 US states. The surveillance methods have been described elsewhere.10 In brief, surveillance staff at each site identify all positive C difficile toxin or molecular assays in stool specimens from all inpatient and outpatient laboratories serving surveillance catchment area residents. C difficile infection is defined as a positive C difficile toxin or molecular assay on a stool specimen from a surveillance area resident 1 year or older who did not have a positive assay in the previous 8 weeks. Quiz Ref IDFor each patient identified with CDI, medical records are initially reviewed to determine if the infection had a hospital onset (ie, positive stool specimen collected >3 days after admission) or a community onset (ie, positive stool specimen collected as an outpatient or ≤3 days after admission). For all patients with community-onset CDI, an in-depth medical record review is performed, and patients are classified as having putative community-associated CDI if no recent (ie, within 12 weeks before the stool specimen collection date) overnight stay in a hospital or long-term care facility was recorded. Information on disease severity, clinical outcomes, medication exposures, and underlying conditions pertaining to the Charlson comorbidity index,11 as well as those conditions relevant to CDI such as inflammatory bowel disease and diverticular disease, is obtained from the medical records for all patients with putative community-associated CDI.
From January 1, 2009, through May 31, 2011, a sequential sample of patients with putative community-associated CDI was contacted by telephone for an interview in 8 of 10 US surveillance sites (California, Colorado, Connecticut, Georgia, Maryland, Minnesota, New York, and Tennessee). All patients who agreed to be interviewed were initially asked if they had a history of recent overnight stay in a health care facility (ie, nursing homes, acute care hospitals, or long-term acute care hospitals). Those patients reporting an overnight stay in a health care facility were reclassified as having community-onset health care–associated CDI and did not proceed with the interview; patients not reporting an overnight stay were classified as confirmed patients with community-associated CDI and were asked additional questions regarding medical history, clinical symptoms, health care occupation requiring direct patient care, and recent (ie, within 12 weeks before the C difficile–positive specimen) exposures to day care settings, children in diapers, infants younger than 1 year, outpatient health care settings, household members with CDI, and antibiotic and other medication use, as well as food and animal exposures. All 8 participating sites completed at least 50 interviews.
Only confirmed patients having community-associated CDI with diarrhea documented in the medical record or reported in the interview as 3 or more loose stools in a 24-hour period at the time of the C difficile–positive specimen were included in the analyses. Because C difficile is often transmitted in hospital settings,2 where invasive procedures are performed and where the duration and frequency of patient contact with health care providers and the environment are long and high, we a priori categorized outpatient health care exposure in the 12 weeks before the C difficile–positive stool specimen into the following 3 levels: (1) high-level health care exposure, defined as dialysis, a job requiring direct contact with patients, outpatient surgery or an invasive procedure, emergency department or urgent care visit, or inpatient care at a health care facility without an overnight stay; (2) low-level health care exposure, defined as a visit to a dentist, physician, or other outpatient clinic (eg, psychology, warfarin sodium, or pharmacy clinic visit); and (3) no health care exposure, defined as no recent outpatient health care exposure. Patients with community-associated CDI were classified into 1 of the 3 exposure levels based on the highest level of exposure reported during the telephone interview. For example, a patient who reported both low-level and high-level exposures was included in the high-level exposure group; levels of exposure were mutually exclusive.
A convenience sample of C difficile–positive stool specimens (approximately 40%) from interviewed patients was cultured,12 and molecular characterization of recovered C difficile isolates was performed. Pulsed-field gel electrophoresis patterns were analyzed using available software (BioNumerics version 5.10; Applied Maths) and were grouped into pulsed-field types using Dice coefficient and unweighted pair group method with arithmetic mean algorithm clustering, and an 80% similarity threshold was used to assign North American pulsed-field gel electrophoresis (NAP) types.13
Univariate analyses of demographics, clinical characteristics, and potential sources of C difficile acquisition were conducted among patients with community-associated CDI stratified by level of health care exposure. The χ2 test or Fisher exact test was used to compare categorical variables, and the Wilcoxon rank sum test was used to compare continuous variables.
Multinomial logistic regression analysis was used to identify predictors of no and low-level health care exposure using high-level exposure as the reference group. Predictors of interest included potential sources of C difficile acquisition. Variables eligible for inclusion in models had P < .20 in univariate analysis and were biologically plausible sources of C difficile acquisition. Prior exposure to antibiotics was included as an interaction term with potential sources of C difficile acquisition. A backward logistic regression modeling strategy was used with a stay criterion of P ≤ .10 for all variables. Sensitivity analyses for the final model were conducted by restricting data to patients who received antibiotics. All analyses were conducted using statistical software (SAS version 9.2; SAS Institute, Inc), and 2-sided P ≤ .05 was considered statistically significant.
From January 1, 2009, through May 31, 2011, a total of 1624 patients with putative community-associated CDI were contacted for an interview; 1101 (67.8%) agreed to be interviewed, and 523 (32.2%) could not be contacted or declined participation. Of 1101 patients with putative community-associated CDI for whom interview data were available, 67 (6.1%) were reclassified as having community-onset health care facility–associated CDI because they reported having an overnight stay in a health care facility in the 12 weeks before the positive C difficile stool specimen collection. Of 1034 confirmed patients with community-associated CDI, 1013 (98.0%) completed the entire interview; 984 (97.1%) of these reported diarrhea at the time of collection of the C difficile–positive stool specimen and were included in all analyses (Figure).
Compared with 573 patients (523 putative and 50 confirmed) excluded from analyses, the 984 confirmed patients with community-associated CDI were more likely to be female (66.6% vs 58.4%) and of white race (86.3% vs 59.1%) (P < .01 for both). No difference in the proportion of patients 18 years or younger was detected (13.3% vs 10.8%, P = .15).
Among 984 confirmed patients with community-associated CDI, the median age of patients was 51 years, and the median Charlson comorbidity index was 0; 66.6% were female, and 86.3% were of white race (Table 1). Antibiotics were used within 12 weeks of C difficile–positive stool specimen collection among 631 of 984 patients (64.1%); cephalosporins, β-lactam or β-lactamase inhibitor, penicillins, fluoroquinolones, and clindamycin were most commonly used. Among 631 patients with CDI who used antibiotics, the most commonly reported reasons for receiving antibiotics were ear, sinus, or upper respiratory tract infection (34.7%), followed by dental cleaning or oral surgery (15.1%), urinary tract infection (9.3%), skin infection (7.5%), and bronchitis or pneumonia (7.5%).
Of 984 patients with CDI, 273 (27.7%) reported recent proton pump inhibitor (PPI) use, 91 (9.2%) had used immune-suppressing agents, and 90 (9.1%) had exposure to an H2-receptor antagonist. A higher proportion of patients without prior antibiotic exposure reported PPI use (31.2% vs 25.8%, P = .07) or the use of an immune-suppressing agent such as chemotherapy, oral corticosteroids, and interleukin receptor antagonists (12.2% vs 7.6%, P = .01) compared with patients with prior antibiotic exposure, while the proportion of patients receiving an H2-receptor antagonist (9.6% vs 8.9%, P = .68) did not differ by antibiotic exposure status. Of 91 patients who received immune-suppressing agents, only 17 (18.7%) did not report other medication exposure (ie, PPI, antibiotic, or H2-receptor antagonist).
As given in Table 1, Quiz Ref IDhospitalization within 7 days of a positive C difficile stool specimen collection occurred in 251 patients with CDI (25.5%); for 125 (49.8%) of these, C difficile was listed as the reason for admission. Admission to an intensive care unit (4.8%), toxic megacolon (3.2%), death (1.6%), and colectomy (0.8%) were uncommon among hospitalized patients with CDI.
Stool specimens from 388 patients with CDI (39.4%) were collected and submitted for toxigenic C difficile culture. Clostridium difficile was recovered from 313 of 388 toxin-positive specimens (80.7%); NAP1 (21.7%) was the most common strain type detected, followed by NAP4 (11.5%) and NAP11 (10.9%); NAP7 and NAP 8 were uncommon (6.7%).
Of 984 patients with community-associated CDI, 177 (18.0%) had no recent outpatient health care exposure, 400 (40.7%) had a low-level health care exposure, and 407 (41.4%) had a high-level health care exposure (Table 2). Patients having CDI with no, low-level, and high-level outpatient health care exposure differed in age, PPI use, medical conditions, antibiotic exposure in the 12 weeks before CDI, and exposure to household members who had active CDI, who were infants younger than 1 year, or who were diapered children younger than 4 years (Table 3). Patients having CDI with no outpatient health care exposure were less likely to have received antibiotics in the prior 12 weeks compared with patients having low-level or high-level health care exposure (P < .001). Exposure to PPIs, H2-receptor antagonists, animals and different types of food, and household members who had active CDI or who were infants younger than 1 year was similar between patients having CDI with no and low-level health care exposure (P > .05). Of 177 patients having CDI with no outpatient health care exposure, 108 (61.0%) reported prior medication exposure; 42 (38.9%) of those were exposed only to antibiotics, 16 (14.8%) only to PPIs, 8 (7.4%) only to H2-receptor antagonists, and 6 (5.6%) only to immune-suppressing agents.
In multivariable analysis, patients having CDI with low-level health care exposure were more likely to have no medical conditions (odds ratio, 1.7; P < .01) and have household members who were infants younger than 1 year (odds ratio, 2.1; P = .05) compared with patients with high-level health care exposure (Table 4). Although the association between having household members with CDI and low-level outpatient health care exposure was strong, it was not statistically significant (odds ratio, 6.9; P = .07). For patients having CDI without outpatient health care exposure, no statistically significant association was found with having household members younger than 1 year or household members with active CDI; however, the point estimates were high and were similar to point estimates found for patients having CDI with low-level health care exposure, suggesting that an association may exist. No differences in the final model were detected when analyses were restricted to 631 patients who used antibiotics.
Although community-associated CDI is defined based on the interim surveillance recommendations14 as the absence of inpatient overnight stay in a health care facility, we found that 82.0% of patients acquiring C difficile in the community had either a recent outpatient health care exposure or an inpatient health care exposure without an overnight stay. Outpatient settings such as physicians’ offices, emergency departments, and dialysis facilities can be the source of C difficile acquisition by exposure to contaminated environmental surfaces, as well as the prescription of antibiotics that disrupt the lower intestinal microbiota. In our study, 64.1% of patients with CDI received outpatient antibiotics within 12 weeks before infection, and the most common indications for antibiotic therapy were ear, sinus, or upper respiratory tract infection or a dental procedure. Multiple studies15- 20 have noted that ear, sinus, or upper respiratory tract infections are common reasons for inappropriate antibiotic use in outpatient settings. The many patients receiving antibiotics for dental procedures was notable because the current American Heart Association guideline for prevention of infective endocarditis restricts prophylactic antibiotic use for dental procedures to patients with underlying cardiac conditions associated with the highest risk of adverse outcome from infective endocarditis.21 Therefore, it is likely that a substantial proportion of patients in our study received antibiotics inappropriately, emphasizing that antibiotics should be prescribed more judiciously by outpatient health care providers and that the overuse of outpatient antibiotics may have an adverse effect on community-associated CDI rates. Antimicrobial stewardship programs in acute care facilities have been associated with decreases in CDI rates up to 60%22,23; aspects of these strategies may need to be considered for use in outpatient health care settings as well.
Quiz Ref IDThirty-six percent of patients in our study did not report antibiotic exposure in the 12 weeks before infection. Since discovery of the causal role for C difficile in pseudomembranous colitis was made in the late 1970s,24 there have been occasional reports of CDI occurring without precedent antibiotic exposure.25- 27 However, the overall importance of community-associated CDI and its frequent occurrence in the absence of antibiotic exposure were not appreciated until approximately 8 years ago.3 Our study is the largest assessment of antibiotic exposures among patients with community-associated CDI in the United States to date, and the proportion we identified without such exposure is consistent with other recent estimates.3- 6 Although it is unknown from these or other data whether CDI in the absence of antibiotic exposure is increasing, other emerging factors may have a role similar to that of antibiotics in weakening the important host defense afforded by intact lower intestinal microbiota.
We found that patients having community-associated CDI without antibiotic exposure had a trend toward having received PPIs more frequently than patients with antibiotic exposure. In some studies,28,29 PPIs have been shown to increase the risk of community-associated CDI, and the US Food and Drug Administration30 issued a recent warning advising physicians of the increased CDI risk in patients receiving PPIs. However, no data indicating the effect of restricting PPI use on CDI incidence are available to date. In addition, the mechanism by which PPIs may increase the risk of CDI is not fully understood, and it has been suggested that PPIs may have a more important role in patients with minimal antibiotic exposure.31 Based on our data, if the effect of reducing unnecessary PPI use on community-associated CDI is limited to those patients who have not received recent antibiotics, such an intervention would prevent only 11.2% of community-associated CDI.
Clostridium difficile spores can survive for prolonged periods in the environment,2 and the health care environment where patients with C difficile are treated can serve as a source of transmission.32 To identify sources of C difficile in the community other than the outpatient health care environment and the transiently contaminated hands of health care personnel, we compared C difficile patients by level of health care exposure. In these exploratory analyses adjusted for antibiotic use, a plausible association existed between low-level health care exposure and exposure to household members younger than 1 year. Infants younger than 1 year are known to be frequent asymptomatic carriers of C difficile, with the results of some studies33,34 suggesting up to a 70% colonization rate. Our findings are consistent with a study by Wilcox et al,8 which found that contact with children younger than 2 years was associated with an increased risk of community-associated CDI. Although C difficile–colonized infants and children can shed the organism into the environment35 and a study36 has reported a C difficile outbreak in a day care center, additional studies in day care, home day care, and household settings are needed before setting-specific environmental recommendations can be made. We also found higher odds of having a household member with CDI among the no and low-level health care exposure groups. However, due to the low prevalence of household members with CDI, this association was not statistically significant. This finding is consistent with a recent Canadian study,37 which demonstrated that household contacts with patients having active CDI are at increased risk of infection. Our data provided no evidence to support a role for food or animal exposure as a source of C difficile acquisition beyond health care exposure. Only 6.7% of culture-positive isolates were NAP7 or NAP8, strains primarily detected in food and animals.38 In recent studies39- 42 in North America, C difficile detection in retail meat samples has ranged from 0% to 10%. This low prevalence of C difficile among retail meat in conjunction with our findings suggests that food and animal exposures could account for only a small proportion of community-associated CDI. Furthermore, antibiotics may be present in consumed foods,43 and it is unclear at this point whether food can be a source of C difficile or another potential factor that can disturb the gut microbiota and predispose patients to CDI.
Despite that a large sample of patients across multiple geographic locations was included in our analyses, the study is subject to several limitations. Quiz Ref IDFirst, only a sample of patients having community-associated CDI was interviewed, and these patients were more likely to be female and white compared with patients having CDI who refused to be interviewed. In addition, only a convenience sample of the patients interviewed had stool specimens sent for further testing. Therefore, patients and C difficile isolates included in this analysis may not be representative of all US patients with community-associated CDI, and the data should be interpreted cautiously because women of perimenopausal age, for example, may be submitted to more medical maneuvers or may be receiving other medications to counteract menopause symptoms. Second, because interviews were conducted up to 12 weeks after detection of C difficile and because exposures to medications and sources of C difficile acquisition were self-reported, it is possible that these exposures were misclassified. Nevertheless, this study assesses exposures for C difficile using medical records and health interviews and may provide a more accurate description of exposures compared with studies that solely relied on data collected from medical records. Third, the lack of a comparison group without CDI precluded us from confirming risk factors for community-associated CDI that we observed in this study. Fourth, because few patients had CDI without outpatient health care exposure, we were likely limited in our ability to detect any statistically significant association among this group. Nonetheless, our findings raise important hypotheses to be tested in future studies.
Most patients identified with community-associated CDI had received antibiotics and had outpatient health care exposure. Prevention of community-associated CDI should primarily focus on reducing inappropriate antibiotic use and better infection control practices in outpatient settings. Our data support evaluation of additional strategies, including further examination of C difficile transmission in outpatient and household settings and reduction of PPI use.
Corresponding Author: Fernanda C. Lessa, MD, MPH, Division of Healthcare Quality Promotion, Centers for Disease Control and Prevention, 1600 Clifton Rd, Mail Stop A-24, Atlanta, GA 30333 (firstname.lastname@example.org).
Accepted for Publication: March 22, 2013.
Published Online: June 17, 2013. doi:10.1001/jamainternmed.2013.7056.
Author Contributions: Drs Chitnis and Lessa had full access to the data and take responsibility for the integrity of the data and accuracy of the data analysis.
Study concept and design: Chitnis, Holzbauer, Bamberg, Farley, Dumyati, Beldavs, Dunn, Gould, Gerding, McDonald, Lessa.
Acquisition of data: Holzbauer, Winston, Bamberg, Lyons, Farley, Dumyati, Wilson, Beldavs, Beldavs, Dunn, MacCannell, Gerding, McDonald.
Analysis and interpretation of data: Chitnis, Holzbauer, Belflower, Winston, Farley, Wilson, Beldavs, Gould, MacCannell, Gerding, Lessa.
Drafting of the manuscript: Chitnis, Holzbauer, Belflower, Dunn, MacCannell, McDonald, Lessa.
Critical revision of the manuscript for important intellectual content: Chitnis, Holzbauer, Belflower, Winston, Bamberg, Lyons, Farley, Dumyati, Wilson, Beldavs, Dunn, Gould, MacCannell, Gerding, McDonald.
Statistical analysis: Chitnis, Holzbauer, Belflower, MacCannell, Lessa.
Administrative, technical, and material support: Holzbauer, Belflower, Bamberg, Wilson, Beldavs, Dunn, Gould, MacCannell, Gerding.
Study supervision: Winston, Farley, Dumyati, Beldavs, Dunn, Gould, Gerding, McDonald, Lessa.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was funded by the Emerging Infections Program Cooperative Agreement between study sites and the Centers for Disease Control and Prevention under the following grants: U50CK000201 (California), U50CK000194 (Colorado), U50CK000195 (Connecticut), U50CK000196 (Georgia), U50CK000203 (Maryland), U50CK000204 (Minnesota), U50CK000199 (New York), and U50CK000198 (Tennessee).
Disclaimer: The findings and conclusions in this study are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
Previous Presentation: Some of the study data were presented in preliminary form as an abstract at Anaerobe 2012: the 11th Biennial Congress of the Anaerobe Society of the Americas; June 27 through July 1, 2012; San Francisco, California.
Additional Contributions: The following contributed to data collection and patient interviews: Joelle Nadle, MPH, Erin Garcia, MPH, Erin Parker, MPH, Kelly Kast, MSPH, Helen Johnston, MPH, Jim Meek, MPH, Leigh Ann Clark, MPH, Andrew Revis, MPH, Rebecca Perlmutter, MPH, Malorie Givan, MPH, Ruth Lynfield, MD, Rebecca Tsay, MPH, Deborah Nelson, RN, Samir Hannah, MD, L. Amanda Ingram, MPH, and Brenda Rue, RN. Susan Sambol, BS, MT, Laurica Petrella, BS, Lydia Anderson, BS, and Brandi Limbago, PhD, processed the Clostridium difficile specimens included in the study. Jessica Cohen, MPH, and Scott Fridkin, MD, critically reviewed the manuscript.