Analysis of VRE Transmission In A Major Hospital Setting Using Hierarchical Clustering and Bayesian Phylodynamic Methods

Healthcare-associated infections (HAIs) can prolong and add substantial costs to hospital stays. One study estimated that 1 out of 25 hospitalized patients were expected to be infected by a HAI on a daily basis. Minimizing HAIs would increase the quality of healthcare within hospitals; thus infection prevention methods must utilize various strategies to identify the cause of HAIs and develop interventions to reduce cases. Currently many healthcare institutions utilize whole-genome sequencing (WGS) in identifying outbreaks and combine them with epidemiological methods in developing protocols to minimize the size of an outbreak. A recent example would be researchers in the University of Pittsburgh – Medical Center Presbyterian Hospital (UPMC), who have developed a machine-learning based method to incorporate WGS data with electronic health records (EHRs) to determine the most likely routes of transmission during an outbreak of vancomycin-resistant enterococci (VRE). Using a ground truth (GT) dataset based on the VRE outbreak, we performed an assessment to compare two methods for categorizing bacterial isolates into transmission routes. We compared hierarchical clustering methods with a Bayesian phylodynamic model to determine which classification had the most similarity to the GT dataset. Our analysis proved inconclusive in identifying a method with superior performance due to computational limitations for the Bayesian phylodynamic model, however the urgency and time constraints of an active outbreak have shown to be better suited for the hierarchical clustering method and we recommend the Bayesian phylodynamic model as part of a retrospective analysis of an outbreak. This analysis which identifies routes of infectious disease transmission within a hospital setting could be utilized in optimizing infection prevention strategies within the hospital setting and lower the rate of HAIs – making a positive public health impact through reducing cost of care and increasing quality of care