16.2 Overview
A major challenge for malaria dynamics is how to define an state space describing malaria infection and immunity in human populations that captures the essential elements of malaria dynamics well enough to trust for making policies. There are features of malaria infections that have been identified and studied in the past: superinfection; the complex time course of an infection – including fluctuating parasite densities – and the problem of detection; gametogenesis, gametocyte maturation and gametocyte dynamics; fever and disease; development of immunity with exposure including its effects on infection, disease, and infectiousness; treatment, adherence to drug regimens, chemoprotection and infection curing. Over time, these issues have been addressed in various models. We need model that is good enough for policy, but this also means developing a commmon understanding of malaria that can serve as a basis for discussion.
To get to that point, we must start simple and add complexity.
The model for malaria infection that we presented in [Malaria Dynamics] was developed by Ross. In today’s vernacular, it would be called an SIS compartment model. The model is very simple, and it is probably inadequate for every task, but it is useful and it has been used. The model assumes that malaria infections clear at a constant rate regardless of the age of infection or other factors. The persistence of malaria infections over decades tells us that this assumption is clearly false, but it is good enough for some programmatic needs. During the GMEP, Ross’s model was used to characterized the response timelines for the PfPR after the interruption of transmission. Drawing on multiple sources, Macdonald estimated that the duration of infection was around 200 days, which was good enough to use as a basis for monitoring and evaluating the interruption of transmission [84]. The simpler model was used even though Macdonald had already proposed an alternative model that considered superinfection [57]. Despite the simplicity, the model was adequate to the task [85]. An important lesson is that the simplicity has some advantages, and the models that get used in policy tend to be very simple.
The question is how to develop models that are simple and yet are up to policy tasks, which means that the models must (at some point) get validated against research data. Doing so means having sufficient complexity to deal with exposure, infection, detection, immunity, disease, infectiousness, care seeking, and drug taking. Whatever model is selected as a basis for policy, it should be simple enough to understand and yet complex enough to capture the gist of malaria epidemiology. The models, however chosen, must get it right. Sorting through all the complexity to get a model that is good enough is a daunting task. This introduction is mainly historical, but we use it to preview some of the themes. In the following history, we discuss some of the important innovations.