Advancing Statistical Methods
We have developed and extended Bayesian transmission models used to understand the transmission dynamics of antibiotic-resistance pathogens in healthcare facilities in the absence of complete data. Early work focused on increasing computational efficiency in the algorithms while simultaneously developing continuous-time implementations that reduced bias in parameter estimates. We also relaxed model assumptions that were previously made in order to simplify analysis, allowing for a more realistic model of patient transitions between different infectious states. These Bayesian transmission models have been used to study transmission of methicillin-resistant staphylococcus aureus (MRSA) in hospitals and nursing homes across the Veterans Affairs (VA). Using our models, we have been able to estimate how the burden of colonization changes during hospitalizations and how MRSA transmission has changed across the US over time. We also estimated that use of contact precautions for MRSA carriers across the VA was associated with a two-fold reduction in the transmission rate, and we have also validated our methods in the re-analysis of an infection control trial showing that these methods could be used to evaluate future infection control intervention trials. We are currently working to extend our Bayesian transmission models to situations where test result data is more rare, making adjusting for bias more difficult.
We have also used and developed branching process models to evaluate high-profile outbreaks of pathogens that have been of concern across the globe for public health officials due to the pandemic-potential of some of these pathogens, including MERS and Ebola. We have developed these models to make use of reported cases and our approach is used to estimate the reproductive number of these pathogens while exploring the potential for superspreaders in the population based on the amount of variation in the observed cases. These models have allowed us to investigate the outbreak size probabilities so that we can have a better understanding of the risks in these outbreaks.
