Search

The overall theme of this thesis is to generate innovative health-economic outcomes which are needed prior to the introduction of population-level dengue immunization programs in three dengue-endemic countries: Vietnam, Thailand, and Colombia. This thesis consists of six chapters which are distinctive but inter-related. Following the introductory chapter, where the existing literature search is carried out, in Chapter 2 I begin by conducting a multi-country study of the economic burden of dengue fever based on standardized field studies in the three dengue-endemic countries. In Chapter 3, an Early Warning Signal (EWS) model is developed based upon climate and non-climate factors. In particular, this chapter adopts a theory of elasticity, which is not common in the field of health economics but which is a key notion in economics, to predict geographically-dispersed dengue outbreaks in Colombia. The econometric technique used in the third chapter is further extended to the mixed-effects modeling work in Chapter 4 where high-risk areas for dengue fever are identified at the 5 km by 5 km geographical unit, and the number of dengue vaccine doses is estimated at the population-level in Vietnam, Thailand, and Colombia. The concepts and methodologies applied in both Chapter 3 and 4 are unconventional but clearly underscore the needs for innovative approaches in order to respond to existing public health problems. Based on the findings from Chapter 1 to Chapter 4, the vaccination impacts and cost-effectiveness of dengue vaccination strategies are estimated in the three dengue-endemic countries in Chapter 5. Vaccination impact is derived from a spatially explicit, individual-based meta-population transmission model developed by the Oxford/Exeter group for the WHO Strategic Advisory Group Experts (SAGE) working group exercise. In addition, country-specific health economic parameters are taken from the economic burden field studies in the second chapter of the thesis. In particular, Chapter 5 is designed to reflect the most up-to-date information on the dengue vaccine development: including controversial debates over CYD-TDV, introducing a new vaccine candidate (NVC) with no safety concern identified in CYD-TDV. Lastly, Chapter 6 summarizes the roles of the four core chapters as a country-specific investment case, and underscores the importance of an evidence-based decision-making process prior to implementing vaccination campaigns at the population-level.

Plasmodium vivax is the most geographically widespread of the human malarias and is capable of causing severe debilitating disease. The parasite’s unique biology poses challenges to control of the disease and the understanding of its epidemiology. It is less researched and well understood than the more deadly P. falciparum. In this thesis, spatial relapse patterns and models of endemicity and clinical disease were applied to generate robust estimates of the P. vivax burden to address a key knowledge gap in malaria epidemiology. First, a review of the distribution of the parasite, its vectors and populations at risk found nearly one third of the global population living at risk, and more potential vectors than P. falciparum. In spite of low observed endemicity, the public health impact of P. vivax is likely to have been seriously underestimated in the past. To accurately define the burden of P. vivax it was necessary to improve understanding of one of the parasite’s most unique and challenging aspects, its ability to relapse. A meta-analysis of individual records of relapse showed that relapse periodicity varied systematically by geographic region and could be categorized by nine global regions. The nine regions were applied to a model to quantify the relationship between prevalence of infection and incidence of clinical disease. As relapse would have an influence on both measures, separate relationships were drawn for each relapse zone. The prevalence-incidence model was used to translate maps of predicted endemicity into measures of clinical burden. The evidence-base of P. vivax prevalence was poor in some regions and therefore a burden estimate based on surveillance reports was also derived. Reported cases must be adjusted for parameters such as under-reporting and treatment-seeking behaviours. A model used to fill gaps in treatment-seeking data available from national household surveys was developed to allow burden to be estimated using both cartographic modelling and surveillance reporting methods. To improve fidelity, the results of the two approaches were combined to enumerate P. vivax burden globally. The results and conclusions of these studies are discussed with recommendations for how these findings influence our understanding of P. vivax epidemiology and implications for future control and elimination efforts.