Using routinely collected data to inform infection-prevention policy decisions

Measures to reduce transmission are a vital response to infectious disease epidemics. Collectively such measures are effective in reducing the burden of infectious disease but effectiveness of individual interventions is less certain. Methodologies for causal inference from observational data are well developed, but many methods have requirements that are not met by epidemic data. They may require an individual's outcome to be independent of anyone else's treatment, but the very purpose of infection-prevention measures is to break chains of transmission, benefiting both treated and untreated individuals. I combine causal inference methods, mechanistic models, and observational data to estimate effects of interventions that were used to reduce the spread of severe acute respiratory syndrome coronavirus 2 in the United Kingdom. I combine difference-in-differences methodology with a renewal-equation model. If its assumptions are met, this can detect effects of interventions on transmissibility, but if assumptions are violated, erroneous results can arise with no indication that an error is occurring. I apply the method to mass testing and mandatory use of face masks. Difference-in-differences results suggest that interventions increased incidence of detected infections. I investigate optimal timing of vaccination against respiratory viral infections with models incorporating immune boosting from re-exposure to the virus. Boosting can lead to synchrony in susceptibility and cause periodic outbreaks even without seasonal variation in infectiousness. In scenarios with more immune boosting, vaccinating sooner tends to lead to fewer infections, while in scenarios with less boosting, later vaccination is beneficial. Analyses in this thesis highlight potential problems with causal analyses that disregard mechanisms of disease transmission, and with models that oversimplify immunity. These analyses suggest that greater understanding of changing immunity over time is necessary to determine optimal approaches to reducing transmission of these respiratory viral infections.