Considering the Biological Context to Improve Models for Estimating the Global Disease Burden of Yellow Fever

Yellow fever is an infectious disease that poses a significant threat to global public health. This dissertation focuses on improving current models used to estimate yellow fever burdens by considering both biological and social contexts. Three key refinements to the existing burden projection model for yellow fever were proposed.

Firstly, a hierarchical Bayesian analysis was conducted to estimate the misclassification in vaccination status. These findings indicated substantial misclassification of yellow fever vaccination status overall, with considerable variation across countries. These estimates can be utilized to correct biases and errors in vaccine information reports, ultimately improving burden estimates. Accounting for misclassification, adjustments to reported vaccination coverage were made, revealing potential underestimation in 10 out of 20 countries and overestimation in 5 out of 20 countries.

Secondly, I aimed to explain the age distribution of yellow fever cases in endemic settings in Africa using a catalytic model that incorporates age-dependent exposure and immunity from natural infection or vaccination. Bayesian inference was used to estimate the model's parameters using age-stratified serological and case data. The results indicated that the highest exposure to yellow fever virus occurs among adults (median: 32.4 (95% posterior predictive interval [PPI]: 0 - 64.5)), with their risk of exposure being four-fold higher than the youngest and oldest individuals. This work refined the assumption of homogeneous reservoir-human contact rates among age groups in current models.

Thirdly, I further improved the current models used for projecting yellow fever burden by incorporated the human-to-human transmission mode. Using this model, we have estimated that vaccination has prevented 735,358 deaths worldwide from 1980 to 2022 (95% posterior predictive interval [PPI]: 82,768 - 1,367,529). This represents a reduction in estimated deaths by 55.8% (95% PPI: 27.9% - 93.1%), underscoring the impact of immunization efforts in reducing yellow fever mortality. When comparing the results between the model without human transmission and our projections, we observe a more significant impact of vaccination. This has led to a substantial reduction in disease burden and effectively prevented urban outbreaks. The analysis identified three distinct patterns, each with unique interpretations and policy implications. Of particular concern is the third pattern, which highlights the potential for unpredictable human-to-human outbreaks in high-risk countries.

In conclusion, this dissertation proposes three refinements to improve the estimation of yellow fever burdens that can enhance our understanding of yellow fever epidemiology and contribute to more precise burden projections and effective public health interventions.