Mortality Modelling and Malaria Cost Estimations in Kenya

Abstract

Mortality and morbidity burdens continue to pose significant challenges to public health and development in low- and middle-income countries, particularly in sub-Saharan Africa. Malaria remains a major cause of morbidity and mortality, disproportionately affecting vulnerable populations, such as children and those living in poverty. Accurate data and reliable models are crucial for effective planning, resource allocation, and decision-making in healthcare, especially to achieve sustainable development. This thesis addresses these challenges by developing methodologies to improve mortality modelling and forecasting in Kenya and estimating the costs of treating malaria. To do this, the thesis is organised into two parts: Part I focuses on developing methodologies to model mortality and address data availability challenges, while Part II concentrates on estimating the costs of treating malaria in Kenya. In Part I, we propose a cause-of-death (CoD) mortality modelling technique to improve mortality forecasting in Kenya, utilising data from Kenya. Notably, a novel constrained penalised splines (CPS) model is applied to individual causes of death, providing insights into the critical drivers of mortality. Forecast reconciliation approaches ensure coherence between individual causes and aggregate mortality projections. The CPS model's flexibility allows the incorporation of expert judgment, leading to improved forecasts. We also demonstrate that varying constraints through the CPS model can potentially improve forecast accuracy, especially for causes that have historically experienced significant changes in mortality rates. Comparisons with other models, including the Lee-Carter (LC) and penalised splines (PS) models, demonstrate the CPS model's versatility. Projections using the CPS model indicate an increase in average life expectancy to 73.3 years, compared to the United Nations' estimate of 70.0 years, which has implications for future healthcare costs and life insurance product costing. Part II of the thesis focuses on estimating the health and economic costs of malaria in Kenya using microsimulation models (MSMs). Accurate estimation of age-specific malaria costs is crucial for informed funding allocation to prevention and treatment strategies. These models simulate births, deaths, and migration to project the population until 2040, considering malaria incidence and parasite resistance. The models are calibrated using the Approximate Bayesian Computation (ABC) rejection algorithm and validated internally and externally. The calibrated MSMs perform well and provide accurate predictions for the Kenyan population. Using the calibrated MSMs, we estimate the health and economic costs of malaria in Kenya from 2010 to 2040. The results provide insights into malaria treatment costs and the cost-effectiveness of various treatment strategies, including evaluating a vaccination program. Notably, over a 30-year modelling period, the estimated average discounted medical costs per under-five and over-five individuals were $87$ and $65$, respectively. Treatment costs increase significantly when uncomplicated malaria progresses to severe malaria. Cost-effectiveness analyses of various treatment options reveal the superiority of reducing stock-outs of rapid diagnostic test kits (RDTs) compared to introducing vaccination at reported efficacy rates. Notably, reducing the stock-out of RDTs from 9 in 28 days to 1 in 28 days yielded superior health and economic outcomes compared to introducing vaccination at the reported efficacy rates. Therefore, this research highlights critical malaria health challenges in Kenya and the larger sub-Saharan Africa region. It also offers a roadmap for improved decision-making and resource allocation, ultimately working towards sustainable development goals and better malaria healthcare outcomes for populations in Kenya and beyond. Show more