|
Abstract:
|
Many infectious diseases cluster in households, but data from observational studies of infections can be difficult to analyze. Dependencies in family members' disease outcomes may be due to shared environment or transmission within the household; observed cases can arise from within-household transmission or transmission from the community. Semi-parametric epidemic models exist for the spread of an infectious disease within households, and extensions have been developed to account for variation in susceptibility among household members. Regression approaches exist when the time series of household infections is continuously observed, but these fail when observation is discrete. In this presentation, I outline a semi-parametric class of infectious disease regression models that permit estimation of the within-household and community force of infection, allow individual- and household-level covariates, and are tractable under discrete observation. I discuss identification conditions derive a procedure for finding the identification region for interval-identified parameters. I apply these methods to a large study of household tuberculosis in Lima, Peru.
|
