Discussion
In this study, we found that persons exposed to recently hospitalized family members were at substantially increased risk for CDI within 60 days after the family member's hospital discharge. Furthermore, CDI risk among family members increased as total days of within-household hospitalization increased. Because CDI was not diagnosed in recently hospitalized and discharged family members during or after their hospitalization, and because persons in our analysis were not hospitalized themselves, the increased risk could be attributable to asymptomatic C. difficile colonization at the time of hospital discharge in the hospitalized family member.
We also conducted several sensitivity analyses. First, to evaluate whether household confounding because of greater hospitalization in more susceptible family members could explain our findings, we reversed the temporal ordering of hospital exposure and found that incorporating future hospitalizations did not attenuate our primary effect estimates. This finding reinforces our primary hypothesis that the increased risk we observed is attributable to transmission from family members who become asymptomatically colonized during a prior hospital stay. Second, we used a 90-day exposure window and found consistent results but the dose-response effect appeared slightly attenuated. This finding could suggest that household exposures occurring >60 days in the past might convey minimal risk.
Our results have several implications. First, we provide further support for the role of asymptomatic C. difficile carriers in bacterial transmission. Second, we identify a previously underappreciated potential CDI reservoir outside healthcare settings that could support the spread of community-associated C. difficile. Finally, our results suggest that, if patients who are asymptomatically colonized during a hospital stay contribute to transmission in the community, not all CDI cases attributable to hospital exposure can be directly identified based on hospital discharge records.
In hospital settings, patients asymptomatically colonized with C. difficile are increasingly viewed as a major contributor to CDI spread.[12,13] Indeed, asymptomatic C. difficile transmission has been posited as an explanation for the missing epidemiologic links in whole-genome sequencing studies.[14] Asymptomatic C. difficile colonization among hospitalized patients is not uncommon.[12,17–20] For example, a meta-analysis found that ≈10% of hospitalized patients in North America become colonized.[20] In addition, the likelihood of colonization increases with longer hospital stays,[17] as well as the use of chemotherapy,[22] PPIs or H2 blockers,[22] and steroids.[23] Furthermore, colonization likely persists for some time after discharge. For example, prior hospitalization, even 6 months in the past, has been found to be a risk factor for colonization at hospital admission.[18] Because asymptomatically colonized patients can contaminate the environment and C. difficile spores are resistant to many cleaning solutions, household environments could feasibly lead to both symptomatic and asymptomatic CDI in family members.
Despite the increase in community-acquired CDI, relatively little research has focused on the household setting. Instead, most efforts to find the exposure sources for community-associated CDI have focused on healthcare settings outside hospitals, such as outpatient clinics and emergency departments,[6,7] and nonhealthcare sources such as food,[8] household pets,[10] and even exposure to the agricultural industry.[24] A few relatively small studies[10,25] and 1 large study[11] did identify potential secondary C. difficile transmission from symptomatic cases among household members. Thus far, however, few studies, except studies focusing on newborns, have questioned the role of asymptomatic carriers in household settings. Because infants frequently are colonized with C. difficile in their first several months of life, our findings and those from other studies that exposure to infants is potential risk factor for community-associated C. difficile[17,21] are not surprising.
Household transmission has been documented for other gastrointestinal infections, including rotavirus, norovirus, and Giardia.[26–30] In addition, household transmission has been documented for another major healthcare-associated infection, methicillin-resistant Staphylococcus aureus.[31,32] For at least some of these pathogens, asymptomatic or minimally symptomatic cases contribute to disease transmission. Of note, transmission of methicillin-resistant S. aureus, like C. difficile, was first thought to be almost exclusively confined to hospital settings; awareness of spread in community settings emerged later. Close household contact can also contribute to the spread of other fecal–oral pathogens, such as rotavirus and norovirus,[27] via environmental contamination, providing further support for the plausibility of household spread of C. difficile.
In addition to providing support for the contribution of asymptomatic C. difficile colonization to household transmission, our results also might have implications for future C. difficile surveillance and intervention-based investigations. Prior investigations have shown that cases of symptomatic hospital-associated CDI often do not appear until after a patient is discharged[33] and that some of those cases might generate additional symptomatic cases among family members.[11] However, our results raise the policy question of whether secondary symptomatic cases among household members should be considered when measuring the broader costs of healthcare-associated infections, especially those that have a reasonable epidemiologic link (e.g., using genotyping) with discharged patients who are asymptomatically colonized. Our results clearly suggest that hospital-based interventions to control both symptomatic and asymptomatic C. difficile transmission can help reduce spread in the community. Measures based on standard surveillance efforts might also underestimate the full effectiveness of hospital-based infection and antimicrobial stewardship interventions because those measures might not capture potential, positive downstream effects in the community.
One limitation of our study is that we cannot directly identify the exact point of exposure where C. difficile transmission might have occurred. Exposure could have occurred in a household setting after a family member was discharged from the hospital; alternatively, a family member might have become colonized while visiting another family member in the hospital. However, several reasons exist to suspect that family members visiting the hospital are unlikely to fully explain our observed effect. First, healthcare workers often have lower colonization rates than discharged patients.[17] Second, visitors and visiting hours often are limited or restricted and only represent a small portion of a patient's total length of stay. Third, we did not count persons as exposed in our analysis when their corresponding CDI index date occurred before their family members were discharged from the hospital; we only consider exposure to a recently hospitalized family member after discharge occurred. Thus, if visiting the hospital were the primary mechanism driving our results, our analytical method would be greatly biased toward the null.
Another limitation of our study is that we depended on insurance claims data and diagnostic codes to identify CDI events. We did not have access to laboratory test results to confirm CDI diagnoses, nor did we have access to genetic data to confirm whether subsequent CDI cases in family members were genetically related. We also could not observe or confirm that household contact actually occurred in the assumed household setting; family members could be residing in different locations even if they were enrolled in the same insurance plan. Finally, our data might not capture all family members residing in a single location. We only had access to information for family members that are actively enrolled in the same insurance plan, and family members in the same household are often enrolled in different plans. Despite these limitations, our results demonstrate the importance of considering asymptomatic carriers in spread of CDI in household settings.
In conclusion, because patients are frequently colonized with C. difficile during hospitalization and at discharge, and because ≈25 million persons each year have overnight hospital stays in the United States alone,[34] patients recently discharged from hospitals could be spreading C. difficile outside hospital settings. Asymptomatic C. difficile carriers discharged from hospitals could be a major source of community-associated CDI cases and should be considered during surveillance and intervention-based investigations.
Emerging Infectious Diseases. 2022;28(5):932-939. © 2022 Centers for Disease Control and Prevention (CDC)
