Predicting the cost-effectiveness of strategies for case management of "plasmodium falciparum" malaria in Sub-Saharan Africa

Malaria is an important cause of death and illness in children and adults, particularly in the tropics. The World Health Organization (WHO) estimated that, worldwide, there were 655,000 malaria deaths in 2010, of which 91% were in Africa, and 216 million cases, of which 91% were due to Plasmodium falciparum (P.falciparum). However, case estimates are particularly uncertain, due to the ambiguous definition of a malaria case and methods used for their quantification.
Efficacious interventions against malaria exist, but it is not clear what their full impact will be or how they could be most efficiently implemented. A cornerstone of malaria strategies is case management, which consists primarily of administering drug treatment to cure the disease, and was the focus of this thesis. Currently, the aim of most countries in sub-Saharan Africa is to control malaria and reduce the disease burden by increasing coverage of effective preventive and curative interventions. However, in some places successes in reducing disease burden have lead countries to consider whether and how local interruption of malaria transmission could be achieved and maintained. In these settings, improved surveillance is critical, but it is not clear what it should consist of. It is important to consider the long-term effects of intervention and intervention combinations, such as the dynamic effects on population immunity, which are not captured within the time frame of intervention trials, and their impact in real health systems. Mathematical models can offer guidance in these situations.
In 2006, Smith and colleagues presented individual-based stochastic simulation models of the biology and epidemiology of P. falciparum malaria. As part of this project, a model for the case management of malaria was developed which permitted simulation of the dynamic effects of treatment on transmission. For this thesis, these models were extended to low-transmission settings and used to predict the levels of passive case detection and treatment that would be needed to prevent local re-establishment of transmission in different settings. We assessed the uncertainties in model predictions resulting from stochastic variation and from the assumptions in our model formulations. We found that, even at rather low levels of receptivity, case management alone could not reliably prevent re-establishment of P. falciparum malaria transmission in the face of medium to high importation rates. Model assumptions regarding rates of decay of natural immunity resulted in significantly different odds of transmission re-establishment, highlighting the urgent need for research in this area.
We also developed a literature-based estimate of the per-person cost of screening an entire population for P.falciparum infection using diagnostic tests. We used this cost estimate along with simulation model outputs to analyse the cost-effectiveness of mass screening and treatment (MSAT) as a burden-reducing intervention, relative to the cost-effectiveness of scaling up case management or insecticide-treated net (ITN) coverage. We found that MSAT may be a cost-effective strategy at medium to high transmission levels and at moderate ITN coverage. This finding is in contrast to the current focus on MSAT as an intervention for low or near-elimination settings. Future analyses comparing the cost-effectiveness of case management with that of preventive interventions should include both disability and deaths averted (expressed in DALYs) as an outcome measure. The analysis also highlighted the need for alternative measures of uncomplicated malaria burden to capture the impact of case management in simulation models of its cost-effectiveness. An approach to do this, using data available in community surveys, is presented in this thesis.
Finally, the previous case management model was extended to allow a finer-grained simulation of health systems and a drug action model was integrated to allow simulation of the effects of case management on parasite densities. The development and parameterization of the new case management model, and its potential future uses and limitations, are presented in the last sections of this thesis.