Association Between Environmental Factors and Toxigenic Clostridioides difficile Carriage at Hospital Admission

IMPORTANCE Clostridioides difficile infection is the most frequent health care–associated infection intheUnitedStates.However,exposuretothisorganismmightoccuroutsidethehealthcaresetting. OBJECTIVE To examine whether exposure to environmental factors, such as livestock farms, is associated with a higher probability of being colonized with C difficile at hospital admission. DESIGN, SETTING, AND PARTICIPANTS This retrospective cohort study was conducted from May 1, 2017, to June 30, 2018, at a teaching-affiliated hospital in Milwaukee, Wisconsin. All consecutive patients underwent C difficile screening using a nucleic acid amplification test at hospital admission. Data analyses were performed from July 2018 to October 2019. EXPOSURES The distances from patient residence to the nearest livestock farms, meat processing plants, raw materials services, and sewage facilities were measured in addition to risk factors previously evaluated in other studies. MAIN OUTCOMES AND MEASURES The main outcome was a positive result on C difficile screening tests performed within 72 hours of hospital admission. RESULTS total of 3043 patients admitted to the were included in the final analysis. Of those, 1564 (51.4%) were women and 2074 (68.9%) were white, with a mean (SD) age of 62.0 (15.9) years; 978 patients (32.1%) were admitted to hematology-oncology units. At first admission, 318 patients (10.4%) were detected through testing as C difficile carriers. Multivariable logistic regression analyses were performed on a stratified sample of patients based on hematology-oncology admission status. These analyses indicated previously considered risk factors (eg, previous hospitalization), our analysis indicates that residential proximity to certain environmental exposures, such as livestock farms, is associated with a higher risk of C difficile carriage, particularly among the non–hematology-oncology population. This environmental association might be owing not only to higher exposure to C difficile but also to exposure to microbiome disruptors, such as antibiotics and pesticides in water runoff. 31 Future studies should examine the association between C difficile and livestock farms in other geographic areas and, if an association exists, identify the underlying mechanisms for this association.


Introduction
Clostridioides difficile infection is the most common health care-associated infection in the United States. In 2015, the Centers for Disease Control and Prevention reported the incidence of C difficile infection to be approximately 453 000 cases per year, with an associated annual mortality of 29 300 patients. 1 Several studies have suggested that some C difficile transmission occurs outside of the hospital environment. Among registries that include the incidence of C difficile infection, only 65% of cases in the United States and 74% of those in Europe were reported to be associated with the health care environment, 1,2 suggesting that community exposure may be a factor in the remaining cases. In addition, large studies using whole-genome sequencing of strains that cause C difficile infections reported that 45% of those strains were not related to symptomatic cases, 3,4 suggesting that community factors have a role in C difficile acquisition.
Exposure to C difficile in the community setting might occur from various sources, including farms, livestock animals, water, and agricultural produce. [5][6][7][8][9][10] Whole-genome sequencing of C difficile strains in humans and animals indicates a bidirectional spread of strains. 11,12 Previous studies have evaluated factors associated with the presence of C difficile colonization at the time of hospitalization. [13][14][15] However, to our knowledge, environmental exposures to potential sources of transmission have not yet been evaluated. The goal of this study was to identify factors associated with C difficile colonization at the time of hospitalization, with a particular focus on demographic characteristics, comorbidities, and proximity to potential environmental exposures, such as livestock farms and meat processing plants. We hypothesized that in addition to comorbidities and recent hospitalizations, environmental exposure to animals would be associated with a greater likelihood of C difficile colonization.

Setting and Population
The study was performed at Froedtert Memorial Lutheran Hospital, a 565-bed teaching-affiliated hospital in the Milwaukee, Wisconsin, metropolitan area. In 2016, more than 200 health careassociated laboratory-identified C difficile infection cases from this facility were reported to the National Healthcare Safety Network, resulting in a C difficile standardized infection ratio of 1.45 (45% more than the facility's expected number of infections). The 5 units selected for C difficile screening were those with the highest rates of C difficile infection: 2 hematology-oncology units (66 beds and 18 935 patient-days per year), which included a blood and marrow transplant unit and a solid tumor unit; 1 solid organ transplant unit (27 beds and 7455 patient-days per year); 1 general medical unit (32 beds and 8060 patient-days per year); and 1 intensive care unit (20 beds and 7800 patient-days per year). This study was reviewed and approved by the institutional review board of the Medical College of Wisconsin, and a waiver of informed consent was granted because of the minimal risk associated with the study and because the research could not practicably be carried out without the waiver.
Patients eligible for inclusion were adults who had at least 1 C difficile screening test completed within 72 hours after hospital admission between May 1, 2017, and June 30, 2018. Patients with a history of C difficile infection within the preceding 6 months were excluded from further analysis.
Patients screened for C difficile at more than 1 hospital admission were included only once, using the results of the first screening test from the first admission during which the test was performed.
Demographic variables, clinical information, and laboratory results were obtained from the electronic health record, and patients' home addresses were geocoded using ArcGIS software (Esri). This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for cohort studies.

Clostridioides difficile Testing
All patients underwent C difficile screening using a nucleic acid amplification test at hospital admission. Stool samples were collected by nursing staff and sent to the hospital's clinical laboratory for C difficile testing. Rectal swabs were obtained if patients were not able to produce stool samples.
Patients were allowed to refuse the administration of rectal swabs, which anecdotally occurred in less than 10% of cases. The C difficile screening tests were conducted using the Xpert C difficile assay (Cepheid), which targets a conserved region of the cytotoxin B gene, tcdB, required for virulence. The test was internally and independently validated for off-label use with rectal swab specimens and formed stools. Formed stools were tested by inserting the swab into or rolling the swab on the surface of the stool specimen until the swab was visibly coated. The swab was then eluted into the Xpert C difficile test buffer (provided by the manufacturer) and tested in accordance with the instructions in the product insert.
During the validation, the limit of detection for alternative sources was assessed using swabs coated with stool matrix that had a negative test result for the presence of C difficile. Swabs were subsequently submerged in 10-fold dilutions of a C difficile suspension (American Type Culture Collection BAA-1875) in normal saline. The limit of detection was found to be approximately 5 × 10 3 colony-forming units per swab, or 5 × 10 4 colony-forming units/mL. This limit of detection was 1 log 10 greater than the limit stated by the manufacturer.

Measurement
For the purposes of this study, C difficile colonization was defined as a positive C difficile screening  13 For the analysis, we focused on residential distances to businesses in Wisconsin, Illinois, and Michigan. The geodetic distances (ie, measurements along the earth's surface) between a patient's address and the nearest business within each business group were calculated using ArcGIS software.

Statistical Analysis
Analyses of summary statistics were performed based on the presence of C difficile at hospital admission. Multivariable logistic regression models were first applied to the pooled sample to examine the association between the independent variables and C difficile colonization at admission.
We also evaluated the type of drinking water (ground vs surface) at the zip code level as a potential factor in the likelihood of C difficile colonization. These data were obtained from the drinking water system database maintained by the Wisconsin Department of Natural Resources. 16 After rejecting the post hoc hypothesis of equality of slopes across patients admitted to the hematology-oncology unit compared with those admitted to other units, 14 we reestimated the multivariable models on the 2 subsamples of patients stratified by unit of admission (hematologyoncology vs other units). All tests were 2-tailed and unpaired, with a significance threshold of P = .05.
Data analyses were conducted using Stata software, version 15  exposures, such as number of comorbidities, recent previous hospitalization, or residential distance from livestock farms, did not significantly alter the probability of a positive test result at admission.
To provide an estimate of the relative magnitude of the association between environmental and health care exposures, we calculated adjusted probabilities of C difficile colonization for different scenarios of residential distance to livestock farms (1, 10, and 50 miles) and previous hospitalization (yes or no), holding all other factors constant at their original levels based on the multivariable regression coefficients shown in Table 2. The results of these calculations for patients admitted to units other than hematology-oncology (ie, patients for whom such exposures were significantly associated with colonization) are shown in Table 3. No association was found between the type of drinking water and C difficile colonization.
The independent effect of residential distance to livestock farms was substantial; regardless of health care exposure, the probability of colonization more than doubled for those living 1 mile from a livestock farm compared with those living 50 miles from a livestock farm. Specifically, the probability of colonization increased from 6.5% among those living 50 miles from a livestock farm to 15.7% among those with previous hospitalization and from 4% to 10.6% among those without a recent hospitalization. The increased risk of colonization among patients living 10 miles from livestock farms, a value close to the median distance among our sample, ranged from 48% (risk of colonization increased from 6.5% to 9.6%) among those with previous hospitalization to 52% (risk of colonization increased from 4.0% to 6.1%) among those with no previous hospitalization. A history  (risk of colonization increased from 4.0% to 6.5%) among those with no substantial environmental exposure (ie, those living 50 miles from the nearest livestock farm) and by 48% (risk of colonization increased from 10.6% to 15.7%) among those living 1 mile from such an environmental exposure.    Table 2. Individual-level probabilities were then averaged over the entire sample of patients admitted to non-hematology-oncology units.

Discussion
In this study, we found that 1 in 10 patients admitted to 5 units of a large tertiary suburban academic hospital were C difficile carriers at the time of admission. Although patients admitted to hematology-oncology units were more likely to be colonized with C difficile at admission, with the exception of race, no significant association was observed between patients' sociodemographic and  15 and exposure to penicillins and cephalosporins was more common in the group with toxigenic C difficile colonization. To our knowledge, this is the only study to evaluate race as a factor, finding no association between a patient's race and C difficile colonization. 15 For decades, C difficile was thought to be primarily associated with hospital-related exposures; however, data from studies conducted from 2006 to 2018 suggest that farm animals, 12,21-24 pets, 7,25 fresh produce, 8,10 retail meat, 26 and even potable water 9 are contaminated with C difficile. Exposure to such environmental sources of C difficile may result in its transmission to humans. Farmers with animals colonized with C difficile are at higher risk of being C difficile carriers, with identical C difficile ribotypes to those of animals. 12 In 2018, a worldwide collection of C difficile strains from both humans and farm animals reported a substantial overlap between C difficile strains, indicating bidirectional transmission of C difficile from animals to humans and vice versa. 11 In 2017, Anderson et al 27 reported that the incidence of community-onset C difficile infection (ie, patients who experienced active infection in the community setting) was higher in participants whose residences were nearer to livestock farms and other environmental exposures. Wisconsin has a high number of dairy and pig farms, and these animals have been previously associated with C difficile carriage. 11 One source of C difficile colonization in humans is thought to occur by water runoff from farms contaminating the water supply. 28 Although we found an inverse association between C difficile colonization and the distance from a patient's residence to livestock farms, we did not find similar associations with water treatment plants, meat processing plants, or farm raw material plants. It is important to note that for each type of environmental exposure (ie, livestock plant or water treatment plant), we used the distance to the nearest exposure rather than the number of these exposures in the immediate proximity of the

JAMA Network Open | Infectious Diseases
Association Between Environmental Factors and Toxigenic Clostridioides difficile Carriage residence. We did not find an association between type of drinking water and C difficile colonization, which could in part be owing to the lack of individual-level data for sources of drinking water.

Limitations
This study has several limitations. Because it was a single-center study, it was limited to the hospital's geographic catchment area. In addition, we did not explore the variables of outpatient antibiotic exposure, chemotherapy, and medical procedures before hospital admission, as these data points are unreliably documented in the medical records. However, antibiotic and immunosuppressant exposures have been studied as independent variables by other researchers. 15,17,18 Also, we did not explore the presence of antibodies for C difficile toxin B, which was previously reported to be associated with carriage. 17

Conclusions
Factors associated with C difficile colonization at hospitalization differed based on admission to hematology-oncology units. In addition to previously considered risk factors (eg, previous hospitalization), our analysis indicates that residential proximity to certain environmental exposures, such as livestock farms, is associated with a higher risk of C difficile carriage, particularly among the non-hematology-oncology population. This environmental association might be owing not only to higher exposure to C difficile but also to exposure to microbiome disruptors, such as antibiotics and pesticides in water runoff. 31 Future studies should examine the association between C difficile and livestock farms in other geographic areas and, if an association exists, identify the underlying mechanisms for this association.