Annually, >400,000 cases and almost 30,000 deaths from Clostridioides difficile–associated diarrhea occur in the United States.Reference Lessa, Mu and Bamberg1 Efforts to reduce the spread of C. difficile have focused on reducing transmission from patients with symptomatic C. difficile–associated diarrhea.Reference McDonald, Gerding and Johnson2, Reference Carrico, Bryant and Lessa3 However, many patients are C. difficile carriers who do not have diarrhea. Asymptomatic carriers may serve as a reservoir and spread C. difficile to those around them.Reference Curry, Muto and Schlackman4–Reference Clabots, Johnson, Olson, Peterson and Gerding7 However, patients who are carriers are not routinely identified on hospital admission. In addition, though some research suggested that asymptomatic carriage is protective against symptomatic C. difficile,Reference Shim, Johnson, Samore, Bliss and Gerding8 other studies demonstrated that patients can progress from carrier state to symptomatic C. difficile infection (CDI).Reference Riggs, Sethi, Zabarsky, Eckstein, Jump and Donskey9 Data demonstrating how frequently and how quickly this occurs are limited.Reference Furuya-Kanamori, Marquess and Yakob10 Therefore, identification of asymptomatic carriers could reduce the spread of C. difficile through 2 mechanisms: first, isolation of C. difficile carriers could reduce transmission to uninfected patients, and second, interventions targeting C. difficile carriers could potentially prevent progression to symptomatic C. difficile.
Clostridioides difficile is a spore-forming, gram-positive anaerobic bacillus spread by fecal–oral transmission of spores, which remain viable for long periods of time ex vivo.Reference Otter and French11, Reference Czepiel, Drozdz and Pituch12 Although C. difficile carriers do not have diarrhea, they do shed spores that can contaminate environmental surfaces.Reference Curry, Muto and Schlackman4 The proportion of symptomatic infection resulting from transmission from asymptomatic carriers remains unknown, but research indicates that this does occur.Reference Blixt, Gradel and Homann13
Previously, C. difficile carriage was thought to be protective against the future development of symptomatic C. difficile,Reference Shim, Johnson, Samore, Bliss and Gerding8 and progression was understood to be rare. However, recent evidence suggests otherwise.Reference Eyre, Griffiths and Vaughan6, Reference Blixt, Gradel and Homann13–Reference Ponnada, Guerrero and Jury16 Progression to symptomatic C. difficile often follows acquired immunocompromise (eg, steroids or severe illness) or administration of antibiotics which disrupt the gut flora.Reference Donskey, Kundrapu and Deshpande17 Both are frequently encountered in hospitalized patients. Hospitals are responsible for reporting “healthcare-facility onset” C. difficile,18 diagnosed 3 or more days after hospital admission, which does not account for symptoms or carrier status before the admission. Identifying a high rate of progression from C. difficilecarrier to symptomatic C. difficile could change what we consider a “hospital-acquired infection,” especially if the bacteria was not necessarily acquired in the hospital but was present on admission and only the diarrhea began in the hospital.
To estimate the prevalence of asymptomatic C. difficile carriage at the time of hospital admission and to determine the rate and time to progression to symptomatic CDI, we tested and prospectively followed asymptomatic patients being admitted with no diarrhea. The objectives of our 2-part study were (1) to identify asymptomatic C. difficile carriers and (2) to observe carriers and noncarriers for progression to symptomatic C. difficile. We hypothesized (1) that admission from a nursing facility would be positively associated with C. difficile carriage and (2) that asymptomatic C. difficile carriers would be at increased risk for developing symptomatic C. difficile compared to asymptomatic noncarriers.
Study design and setting
We performed a prospective cohort study on a sample of patients being admitted to a large university hospital in the Bronx, New York, between July 2017 and March 2018. The hospital contains >800 beds and receives >45,000 hospital admissions annually. To determine the prevalence of C. difficile carriage, patients being admitted to the hospital without diarrhea were tested for C. difficile. To follow these patients prospectively, we utilized our unified electronic medical record shared among multiple hospitals and outpatient clinics in the health system. All patients were followed within our system for 6 months or until death for the subsequent diagnosis of symptomatic CDI.
All patients with an admission order from July 2017 to March 2018 were eligible for inclusion. During the study period a convenience sample of days including weekends, was used for screening and testing. On screening days, all patients admitted within the previous 24 hours from a nursing facility were approached for inclusion. On those same days, a work list was generated, and any patient admitted from the community within the previous 24 hours was assigned a number. Random-number generators were used to determine a random sample of community patients. Because previous studies have suggested that nursing facility residents have a high prevalence of C. difficile carriage, patients from the community were sampled in a 1:4 ratio with patients from nursing facilities. For inclusion in the study, subjects were ≥21 years old and required an admission order from the emergency department within the previous 24 hours. The research team queried any patient, family, or staff and excluded patients with active diarrhea defined as ≥3 episodes of loose stool in the previous 24 hours or ≥2 episodes in the previous 12 hours. Subjects were excluded if there was documentation of comfort care only status, if they had a colostomy, or if they were admitted to the pediatrics, obstetric/gynecologic, or psychiatry services.
Because screening admitted patients for infectious diseases is part of standard infection prevention and control practice, we did not seek consent from each study subject, but we offered the option of dissenting (declining) when approached for participation. Following testing, participants were observed for 6 months or until death. The Montefiore/Einstein Institutional Review Board approved the study, granting a waiver of informed consent.
Data collection methods
Eligible subjects underwent swabbing of their perirectal area with an ESwab collection and transport system (Copan Diagnostics, Murrieta, CA) by a single member of the study team. No invasive rectal swabbing was performed. Rectal swabbing or direct testing of stool specimens are the accepted clinical standards, but previous studies have demonstrated the utility of perirectal swabbing.Reference Kundrapu, Sunkesula, Jury, Sethi and Donskey19 If stool was available, a separate swab was performed directly on stool. Test swab soilage, as defined by any visible material on the swab, was recorded as recommended.Reference Kundrapu, Sunkesula, Jury, Sethi and Donskey19
Specimens were processed by the study team (S.B. and D.D.) on the same day as collection. Two testing methodologies were used for all specimens: (1) C. difficile Quik Chek Complete (Abbott, Chicago, IL) to test for glutamate dehydrogenase (GDH) and toxins A and B and (2) XPert C. difficile/Epi (Cepheid, Sunnyvale, CA) real-time polymerase chain reaction (PCR) assay that detects the toxin B gene.20, Reference Swindells, Brenwald, Reading and Oppenheim21 All specimens were also tested by toxigenic culture using spore-enriched specimens in cultures with selective chopped meat broth incubated for 48–72 hours followed by repeat GDH and toxin A/B testing.Reference Koransky, Allen and Dowell22, Reference Riska23
Demographic and clinical characteristics were extracted from the electronic medical record, which included all inpatient and outpatient visits and lab tests sent from the medical center.
Prevalence analysis. Determination of active diarrhea status was queried directly of patients, family, or staff. All other demographic and clinical characteristics were recorded from the electronic medical record or an extracted replicate of the electronic medical record. C. difficile carrier status was defined as any positive PCR or toxin test or toxigenic culture for C. difficile without diarrhea (diarrhea was an exclusion criterion). If the primary clinical team ordered a subsequent C. difficile test, they were informed of any positive study testing result but otherwise were unaware of subject participation in the study. This allowed for the possibility of direct benefit to subjects in the form of hastened diagnosis and treatment. Independent variables examined included age, gender, nursing facility or community resident, season of enrollment, soilage of the test swab, previous admissions within 28 days,18 previous antibiotics within 90 days,Reference Hensgens, Goorhuis, Dekkers and Kuijper24 and previous CDI within 56 days.18
Outcomes analysis. The primary outcome, symptomatic CDI, was defined as any positive clinical test for C. difficile sent by the primary clinical team as part of usual care. The microbiology laboratory rejects solid stool specimens, so the presence of diarrhea in patients was assumed. The clinical algorithm used in this healthcare system, in accordance with guidelines,18 is a combined GDH and toxin test (Quik Chek Complete) followed by PCR (XPert) if the GDH and toxin results are discrepant. We compared time to C. difficile–positive testing among carriers versus noncarriers, censored at 6 months or death.
Baseline characteristics of nursing facility residents versus community residents were compared using the χ2 and Fisher exact tests as appropriate. We calculated the prevalence of asymptomatic C. difficile carriage in the overall study population, in nursing facility residents, and community residents. The significance of the difference in prevalences was tested using a χ2 test and univariate logistic regression model. The time of progression to symptomatic CDI or death, censored at 6 months, was plotted for carriers and noncarriers using the Kaplan-Meier method, and a log-rank test was used to test the differences between groups. Finally, a univariate Cox proportional hazards model was constructed to estimate the hazard of progressing to symptomatic CDI in carriers versus noncarriers. Given the small number of outcome events, no multivariate analyses were performed. Statistical analyses were performed using Stata version 14.2. software (StataCorp, College Station, TX). P values < .05 were considered statistically significant.
Of the 351 potential subjects evaluated for inclusion, 220 subjects were enrolled (62.7%). Common reasons for nonenrollment included declined participation (n = 30), discharged before testing (n = 28), and excluded due to diarrhea (n = 14). Also, 28 patients were not eligible for 7 additional reasons based on the exclusion criteria. In addition, 31 patients who were included in random sampling and, thus, in the potential subject pool, but they were never approached for inclusion to maintain the 4:1 sampling strategy. In terms of acceptability to the subjects, only 30 of 351 of all eligible subjects (8.5%) declined participation.
Of the 220 enrolled subjects, most were female (54%), ≥65 years old (67%), enrolled in the summer (58%), did not have a soiled test swab (55%), and were nursing facility residents (76%), in accordance with the 4:1 enrollment strategy. The characteristics of the total study population and the nursing facility and community residents are presented in Table 1.
a Bold type face indicates statistical significance at P < .05.
b Unadjusted statistics on total population.
c Indicates that data is calculated based on 219 subjects total (missing data on 1 subject).
d Using the Fisher exact test.
Of 220 subjects tested, 21 (9.6%) were asymptomatic C. difficile carriers, which included 17 of 168 nursing facility residents tested (10.2%), and 4 of 52 community residents tested (7.7%), a difference that was not significant (P = .60).
The associations between subject demographic and clinical characteristics and the odds of asymptomatic carriage are presented in Table 2. Having a soiled swab was significantly associated with carriage (odds ratio [OR], 2.7; 95% confidence interval [CI], 1.03–6.9; P = .04). In addition, previous antibiotic exposure was nonsignificantly associated with asymptomatic carriage (OR, 2.3; 95% CI, 0.9–5.6; P = .08).
Note. CI, confidence interval.
Among 21 subjects identified as C. difficile carriers, 8 (38.1%) progressed to clinical CDI within 6 months. Among 199 subjects who were not carriers at enrollment, 4 (2.0%) developed symptomatic CDI within 6 months. Most carriers that progressed to symptomatic CDI did so within 2 weeks of enrollment (note that patients were assessed and excluded if diarrhea was present at enrollment). In the time-to-event analysis, C. difficile carriers had significantly increased risk of developing subsequent clinical CDI compared to noncarriers (hazard ratio [HR], 23.9; 95% CI, 7.2–79.6; P < .001) (Fig. 1).
In this prospective cohort study, 9.6% of subjects admitted in a large academic medical center were asymptomatic C. difficilecarriers, including 10.2% of nursing facility residents and 7.7% of community residents. Among C. difficile carriers identified at enrollment, 38.1% were subsequently diagnosed with symptomatic CDI, most progressing within 2 weeks, whereas only 2.0% of noncarriers were subsequently diagnosed with symptomatic CDI. Only 8.5% of potential subjects declined to participate despite the sensitive nature of perirectal swabbing, suggesting that screening of asymptomatic patients is feasible.
Previous studies have reported widely varying prevalence estimates of asymptomatic C. difficile carriage in healthcare facilities. The overall raw prevalence of asymptomatic CDI in our sample (9.6%) is consistent with recent estimates of C. difficile carriage among admitted patients which vary from 0.6%–13%.Reference McFarland, Mulligan, Kwok and Stamm25, Reference Loo, Bourgault and Poirier26 In contrast, we found a lower prevalence of asymptomatic carriage among nursing facility residents (10.2%) than prior studies, which have reported up to 51%.Reference Riggs, Sethi, Zabarsky, Eckstein, Jump and Donskey9, Reference Fulton and Fallon27–Reference Campbell, Beere, Wilcock and Brown29 Our lower prevalence may reflect a different underlying population, geographic variation, and/or the success of antibiotic stewardship programs.
Although C. difficile carriers shed fewer spores than symptomatic C. difficile patients,Reference Furuya-Kanamori, Marquess and Yakob10, Reference Donskey, Kundrapu and Deshpande17, Reference Sethi, Al-Nassir, Nerandzic, Bobulsky and Donskey30, Reference Caroff, Yokoe and Klompas31 given their larger numbers, carriers may actually be responsible for a larger C. difficile spore burden and more transmission than symptomatic patients.Reference Eyre, Cule and Wilson5, Reference Guerrero, Becker and Eckstein32 Strategies to reduce transmission from asymptomatic carriers to uninfected individuals have included preemptive modified isolation,Reference Longtin, Paquet-Bolduc and Gilca33 heightened cleaning of units at risk,Reference Orenstein, Aronhalt, McManus and Fedraw34, Reference Barker, Alagoz and Safdar35 intensified antibiotic time outs for carriers,Reference Lee, Frenette, Jayaraman, Green and Pilote36 or even prophylactic treatment for those at highest risk such as oncologic or chronically immunosuppressed patients.Reference Ganetsky, Han and Hughes37 Many of these strategies, however, require routine early identification of carriers.
In this study, the only clinical or demographic feature associated with carriage was swab soilage, which is visible fecal material staining the swab. We hypothesize that swab soilage represents improved testing sensitivity in the presence of frank fecal material or, instead, could mean stool incontinence, poor hygiene, or an inability to care for oneself effectively. Further study could elucidate the cause of this association.
The present study adds to the limited body of literature examining the rate of progression from C. difficile carriage to clinical CDI. An older review of 810 patients in 4 studies admitted to large US hospitalsReference Shim, Johnson, Samore, Bliss and Gerding8 found a lower rate of progression to symptomatic C. difficile among carriers than noncarriers; thus, C. difficile carriage was thought to be protective against symptomatic C. difficile. Newer studies show that carriers are at higher risk for subsequent CDI, though in each study the time frame in which carriers were followed was limited to 14 days,Reference Blixt, Gradel and Homann13 during the admission,Reference Eyre, Griffiths and Vaughan6, Reference Nissle, Kopf and Rosler15, Reference Ponnada, Guerrero and Jury16 or 1 month following discharge.Reference Tschudin-Sutter, Carroll and Tamma14 We found a higher rate of progression from carrier to symptomatic CDI (38.1%) compared with only 2.0% among noncarriers when followed for up to 6 months. Due to the high rate of progression, it is possible that a substantial proportion of “healthcare-facility onset”18C. difficile may actually result from the progression from C. difficile carriage to symptomatic C. difficile, especially within the first 2 weeks of hospitalization.
This study has several limitations. First, given the lower than expected number of C. difficile carriers, the study had limited power to detect a difference in the proportion of nursing facility and community residents who were C. difficile carriers. Second, inquiry about the subsequent development of diarrhea was left up to the primary team, which may have led to a symptomatic CDI going unnoticed and undiagnosed, leading to an underestimation of symptomatic CDI in carriers and noncarriers. Third, to screen for asymptomatic carriage, we used perirectal swabbing rather than rectal swabbing or stool specimens. Although perirectal swabbing may have underestimated the true prevalence of C. difficile carriers, the high frequency of soiled test swabs as well as the likely better acceptability of perirectal swabbing as a screening tool made this the preferred modality. Lastly, retrospective data on antibiotic use and prospective data on C. difficile diagnosis was limited to usage and diagnosis only within our healthcare system as recorded in the electronic medical record.
In conclusion, asymptomatic carriers may represent a significant reservoir for transmission of C. difficile, and progression from asymptomatic carriage to symptomatic CDI may account for a significant proportion of CDI that is classified as “healthcare-facility onset.” Therefore, identification of asymptomatic carriers could reduce the spread of C. difficile. Specific environmental, isolation, and stewardship strategies to prevent spread of C. difficile from carriers to uninfected patients as well as prevent progression to symptomatic CDI warrant further study.
The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health.
The research described was supported by NIH/National Center for Advancing Translational Science (NCATS) Einstein-Montefiore CTSA (grant no. UL1TR001073). Additional support was provided by the Department of Medicine, Montefiore Medical Center, Albert Einstein College of Medicine.
Conflicts of interest
All authors report no conflicts of interest relevant to this article.