Title
A Prediction Model Of Rotavirus Vaccine Performance Based On Meta-Analysis And Variable Selection
Date of Award
1-1-2022
Document Type
Open Access Thesis
Degree Name
Master of Public Health (MPH)
Department
School of Public Health
First Advisor
Virginia V. Pitzer
Second Advisor
Daniel D. Weinberger
Abstract
Background: Rotavirus is the leading cause of diarrheal diseases worldwide and is responsible for substantial morbidity and mortality among children younger than 5 years of age. The introduction of national immunization programs against rotavirus has greatly reducedhospital admissions and deaths from rotavirus gastroenteritis, but vaccine efficacy and effectiveness vary substantially between high- and low-income settings. Given the disparity of vaccine performance across countries, we built a model to predict how well the vaccines would work in settings without vaccine introduction. Methods: Data on vaccine performance and relevant predictor variables were extracted from the literature and repositories. We performed a meta-analysis and subgroup analysis to evaluate vaccine efficacy and effectiveness and to examine potential sources of heterogeneity. Finally, we implemented a meta-regression, testing several variable selection methods, and cross-validation models to identify the best set of predictors of vaccine efficacy and effectiveness. Results: The visualization of vaccine performance disparity among countries was consistent with our prior understanding. Using subgroup analysis, we found that heterogeneity in vaccine efficacy and effectiveness could be explained by income level. Based on our meta-regression analysis, the best predictor combination at the study level was under-five mortality, Global Burden of Disease region (GBD), and their interaction. At the country level, diarrhea prevalence, GDP per capita, and study indicator were the best predictor combination. Conclusion: Variability in published estimates of rotavirus vaccine efficacy and effectiveness can be explained by country income level, diarrhea prevalence, and under-five mortality. Our prediction model is a critical tool to understand how a future rotavirus vaccine program would help to decrease mortality and morbidity in countries with high diseases burdens and no vaccination implemented.
Recommended Citation
Li, Yueqi, "A Prediction Model Of Rotavirus Vaccine Performance Based On Meta-Analysis And Variable Selection" (2022). Public Health Theses. 2174.
https://elischolar.library.yale.edu/ysphtdl/2174