Dissertation
Bayesian spatial modeling of visceral Leishmaniasis Disease and infection in Brazil at multiple spatial scales
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Summer 2023
DOI: 10.25820/etd.007070
Abstract
This dissertation aimed to enhance the understanding of the spread and control of visceral leishmaniasis (VL) in Brazil using spatial and spatio-temporal statistical models. Our first goal was to estimate the relative risk of VL in the state of Rio Grande do Norte, Brazil, and to evaluate drivers of disease risk. By employing a dimension reduction technique on Census data, this first study identified three latent factors used to explain higher-than-average risk of VL, including socioeconomic status, sanitation levels, and various animal reservoirs. A Bayesian hierarchical model was then applied to estimate municipality specific VL risk longitudinally using data from 2007 to 2020. Based upon our findings, we narrowed the focus of understanding L. infantum exposure among apparently healthy individuals in the community of Nordelandia within the northern Lagoa Azul neighborhood in Natal, Brazil. Second, we explored the spatial patterns of multiple L. infantum exposure outcomes at a granular spatial scale where outcomes for each study participant were derived from multiple independent diagnostic tests. We estimated the spatial distribution of L. infantum exposure and measured the spatial association between the multiple outcomes through a Bayesian multivariate marked point process. A simulation study validated our ability to recover the model parameters. We additionally provided an empirical analysis using data from approximately 500 residents in the Nordelandia community obtained in a 2021 cross-sectional investigation. Next, our final study evaluated the causal effects of a vector-management intervention on host L. infantum infections at a local community level for an endemic region. Here we developed a Bayesian spatial causal model to quantify the effects of an intervention on the vector population in reducing vector-borne infections in host species. Specifically, we built a Bayesian point-referenced model that considered spatial dependence and interference (spillover) between units. Spillover effects between units were modeled using a Gaussian kernel function, where an intervention’s influence on sand fly and infection outcomes gradually diminish through space. Through a simulation study we demonstrated our model’s ability to recover parameters and capture true causal effects of a hypothetical vector-management intervention strategy in an urban endemic region.
Details
- Title: Subtitle
- Bayesian spatial modeling of visceral Leishmaniasis Disease and infection in Brazil at multiple spatial scales
- Creators
- Helin Giselle Hernandez Reyes
- Contributors
- Jacob Oleson (Advisor)Grant Brown (Advisor)Christine Petersen (Committee Member)Brian Smith (Committee Member)Emily Roberts (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Biostatistics
- Date degree season
- Summer 2023
- DOI
- 10.25820/etd.007070
- Publisher
- University of Iowa
- Number of pages
- xiv, 149 pages
- Copyright
- Copyright 2023 Helin Giselle Hernandez Reyes
- Language
- English
- Date submitted
- 07/25/2023
- Description illustrations
- Illustrations, tables, graphs, charts
- Description bibliographic
- Includes bibliographical references (pages 136-149).
- Public Abstract (ETD)
This dissertation aims to enhance our understanding of visceral leishmaniasis (VL) disease spread in Brazil and characteristics related to increased risk of VL through the application and development of statistical models. VL is a globally distributed neglected tropical disease known for its potential to cause severe illness, which disproportionately affects people in the lowest socioeconomic bracket. While VL is endemic in 13 countries, Brazil accounted for 97% of reported cases in 2019. Despite the public health significance of VL, this disease continues to spread in urban areas within Brazil, especially in the northeast region. The primary objective of this dissertation is to apply and develop statistical models to shed light on the risk of VL in Brazil at various spatial scales.
First, a spatio-temporal model is applied to estimate how risk of VL compares in different municipalities of a state, revealing spatially heterogeneous patterns and identifying opportunities for tailored public health interventions. Building upon these findings, the focus shifts to a high-risk community in northern Natal, Brazil, where the spatial patterns of multiple individual-level L. infantum exposure outcomes are examined, risk factors associated with those patterns are investigated, and the association between multiple outcomes is measured. Finally, the dissertation presents a model to assess the causal effects of a vector-management intervention on host L. infantum infections at a household level within a local community. By employing these statistical models, this research contributes to the understanding of VL spread and provides insights for targeted public health interventions.
- Academic Unit
- Biostatistics
- Record Identifier
- 9984454540502771
Metrics
45 Record Views