Analysis of Infectious Disease Incidence and Complex Survey Data in Space and Time

Abstract

Immunization is one of the most cost-effective health interventions for preventing infectious diseases. Despite the tremendous advances in global immunization coverage in the past few decades, challenges remain in many low- and middle-income countries (LMICs). Timely and comprehensive evaluation of immunization program performance, especially at the sub-national level, is essential for monitoring progress and planning future interventions. Since the administrative data in many LMICs is unreliable due to inadequate health information systems, other data sources, such as disease surveillance systems and household surveys are turned to, to obtain valid estimates for policy guidance. This dissertation develops statistical methods for analyzing infectious disease incidence and complex survey data in space and time to facilitate planning and assessing immunization programs in LMICs. In particular, we develop space-time smoothing models for estimating sub-national measles routine immunization coverage using survey data, and build time-series models for estimating efficacy of measles supplementary immunization activities using incidence data. We also consider vaccination coverage mapping via Bayesian geostatistical modeling of household survey data, demonstrate appropriate ways of accounting for the survey design in modeling, and develop a novel approach to coverage map presentation to allow comparison and control of the overall map uncertainty level.