Date of Degree
Doctor of Philosophy (Ph.D.)
Environmental, Occupational, and Geospatial Health Sciences
C. Mary Schooling
Harm van Bakel
Bacterial Infections and Mycoses | Clinical Epidemiology | Epidemiology | Genomics | Infectious Disease | Public Health
Clostridium difficile, nosocomial outbreaks, whole genome sequencing, asymptomatic carriage, outbreak tracking
Genomic epidemiology of Clostridium difficile colonization and transmission in an intensive care unit cohort
by Brianne Ciferri, MPH
Advisor: C. Mary Schooling, PhD
Introduction: Clostridiodes difficile (C. difficile) is a leading cause of healthcare associated infections (HAI) in the United States and responsible for an estimated incidence of 223,900 cases and 12,800 deaths per year1,2. C. difficile can cause gastrointestinal illness with symptoms ranging from mild diarrheal illness to a life-threatening condition. C. difficile is an opportunistic pathogen in which spores can live in an undisturbed dormant state within the intestinal tract and become active in the presence of favorable conditions. Conditions suitable to C. difficile proliferation occur when the gut microbiome is in a state of imbalance, usually as a result of treatment with oral antibiotics, chemotherapy, or illness1,7,8.
The major predictors for developing a C.difficle infection (CDI) include exposure to a contaminated environment or infectious person and disruption of microbiota. The individuals most likely to experience illness caused by C. difficile are those with pre-existing health complications, with infection resulting in a further health decline among individuals already ill. Given that patients admitted to intensive care units (ICU) are often in a declined state of health and have many of the predictors associated with CDI, this population is at a greater risk for developing infection. In this study I sought to explore the relationship between ICU patients and CDI by examining demographic, clinical, and genomic data.
Methods: Stool samples were collected from two sources over a four-year period; ICU patients and positive CDI case patients. The specimens collected from ICU patients included both positive CDI patients and non-infected patients. The specimens collected from patients who tested positive for CDI and classified as cases, were collected from the clinical laboratory. Patients who did not test positive for CDI in the 4-year study window were either matched to case patients or were selected randomly for screening to determine the rate of asymptomatic carriers within the ICU. We sequenced the isolates from the patients identified as carriers and the cases using PacBio long-read sequencing technology. Following genome assembly, open-source Pathogen Sequencing Phylogenomic Outbreak Toolkit (PathoSPOT) was used to compare the nucleotide differences to determine the presence of transmissions based on clonal relation which is defined as beingclusters, and the possible contribution of asymptomatic carriers, I examined hospital admission history to determine patient overlap and explore possible routes of transmission or common unit reservoirs. I also examined the proportion of recurrent CDI to determine if multiple CDIs were the result of an incomplete decolonization or the result of a new infection.
Results: I identified an 8.0-8.1% carrier rate within the matched and random controls when comparing the case, non-infected, and carrier patients. I identified that case patients were significantly more likely to have had more prolonged contact with the hospital environment compared to non-infected patients.
There were 40 transmission clusters identified, 12 of which included asymptomatic carriers, that ranged in samples collected days to years apart. I next examined hospital admission history to identify common unit overlaps and observed 1 unit which was significantly associated with transmission cluster patients visiting that unit prior to CDI. To further examine the possibility of reservoir units, with the collaboration of the research team we sought to determine if there were clusters of units associated with transmission rather than a single unit given that patients often visit more than 1 unit during a hospital admission. We performed a permutation analysis to identify groups of units that patients commonly transfer between. We identified a total of 10 significant unit clusters, and observed a significant association between CDI patients who were admitted to the surgical unit cluster and a higher likelihood of being in a transmission cluster compared to CDI patients not in a transmission cluster. We compared these units with the overall rate of reported CDI cases within the hospital and observed that the units significantly associated with a transmission cluster were not the units with the highest rates of CDI. Lastly, I examined CDI patients who had more than 1 infection and observed that 85.5% of the recurrent infections were not related.
Discussion: The data observed in each step of the analysis suggests environmental persistence is the main driver of CDI within the examined cohort. I first identified that case patients had significantly higher levels of hospital exposure compared to non-infected patients. This suggests that prior to any genomic analysis, there were already indications that increased exposure to the hospital environment was more associated with testing positive for CDI compared to non-infected patients. After we performed genomic sequencing and identified the 40 transmission cluster, we joined the genomic data with hospital admission data and observed that the majority of patients in transmission clusters (77.5%) did not overlap in the same units at the same time(?). This suggests that patients likely encountered the bacteria while visiting the same unit at different times, rather than direct transmission from patient encounters. We identified that a cluster of units associated with surgical patients were more associated with patients involved in a transmission compared to patients not involved in a transmission. These findings suggest that the pre- and post- surgical units may have higher levels of environmental contamination given that clonal isolates were observed in patients who were admitted to these units months, and in some cases years, apart. Additionally, given that the units in the surgical cluster were not the units with the highest reported overall incidence of CDI, suggests that these units are specifically associated with transmissions, and not artificially inflated due to a heightened overall CDI rate. Lastly, the observation that recurrent cases are more often associated with new infection rather than a persistent infection, further supports my hypothesis that environmental contamination is the driving factor in many transmission cases of CDI. This study has yielded findings to further our clinical and scientific understanding of how C. difficile spreads within the healthcare environment and the drivers of transmission. I observed transmission clusters containing patients with clonally related strains with unit overlap spanning as little as 1 week, to as much as 4-years apart. Additionally, the data suggests that the unit overlap was not in the units in which the CDI was diagnosed. These results strongly suggest that the source of transmission is environmental exposure to units visited earlier in the hospital admission prior to having a positive CDI test result.
The standard protocol following a CDI diagnosis is cleaning and containment, which is seemingly effective given that we did not observe patient overlap in the units where patients tested positive. Consequently, this suggests that direct patient-to-patient transmission does not play a major role in transmission, and that environmental persistence caused by ineffective cleaning of units visited prior to diagnosis is driving in transmission. This study has provided new data to inform evidence-based infection control guidelines to reduce the incidence and prevalence of CDI in the hospital environment.
Conclusion: Genomic sequencing technology was the critical method used in this study, and is a valuable tool in not only outbreak tracking but understanding how diseases move through an environment and population. Genomic technology provides a level of precision surveillance without which we would be oblivious to what is a coincidental case of same strain infections versus a true outbreak. This technology allowed us to identify environmental persistence as the driver of transmission within the patient population. Using genomics also enabled observation of transmission clusters that spanned multiple units, patients, and years, which would otherwise have remained undetected. We were then able to merge this transmission cluster data with patient admission history and identify the presence of reservoir units. The methods utilized in this study should be applied to other healthcare associated infections to increase our understanding of persistence in the presence of effective containment and cleaning protocols. The future of infection control must focus on expanding strategies beyond defensive measures in response to infection, and must apply proactive efforts to mitigate the root of transmission. With the implementation of evidence-based strategies we can work towards reducing the burden of C. difficile and ultimately eliminate the associated morbidity and mortality caused by HAI.
Ciferri, Brianne, "Genomic Epidemiology of Clostridium Difficile Colonization and Transmission in an Intensive Care Unit Cohort" (2021). CUNY Academic Works.
Available for download on Wednesday, December 14, 2022