Send to a friend
Quickly detecting, not predicting, malaria epidemics is the key to disease control, says tropical medicine expert, Jonathan Cox.
Over the last decade academics and international agencies, most notably the WHO, have promoted malaria early warning systems (MEWS) as a way of improving how decision-makers manage epidemics, by giving them more time to plan and respond.
The standard blueprint for MEWS includes detecting upsurges in transmission early by surveillance for new cases, and also predicting future transmission, primarily by monitoring environmental data from remote sensing satellites.
Satellites can provide almost immediate estimates of changing rainfall, land surface temperature and vegetation condition. These, in principle, can generate geographically specific epidemic warnings with lead times of several weeks. They also feed into seasonal climate forecasts for more generalised epidemic risk assessments with even longer lead times.
But practical progress in developing and implementing MEWS has been extremely limited — no fully-fledged early warning systems, of the kind envisaged by the WHO, yet exist.
That’s partly because the disease system is inherently complex and difficult to model. But it is also because researchers are focusing on the wrong areas.
Rather than battling with complicated environmental disease models based on remote sensing climate data, researchers, in partnership with government ministries and international agencies, should be trying to solve the more tractable challenges of detecting emerging epidemics early.
For researchers, the idea that well-demonstrated links between climate variability and malaria transmission could be the way to predict epidemics several weeks or months in advance has been tantalising. But translating promising scientific studies on climate–malaria interactions into robust and reproducible models that give early warnings has proved elusive.
A shortage of good quality disease data is hindering the modelling. Several years of data are needed to ‘train’ and test models that use future variations in climate to predict malaria transmission.
There are a few instances where malaria data are good enough to attempt workable models. For example, in semi-arid Botswana, where there is a relatively simple relationship between rainfall and malaria, researchers have developed reasonably reliable predictive models with lead times of up to six months.
But where the epidemiology is more complex, such as the densely-populated highlands of East Africa, climate data have proved less useful in developing models that work.
In such settings, successfully modelling the complex human-vector-environment interactions will probably require sophisticated dynamic systems modelling combined with time-series analysis that includes non-environmental parameters such as a population’s immune status.
Malaria epidemics are sudden and unpredictable, and that poses serious challenges for disease control. In theory, they can be controlled by detecting upsurges in transmission early, mobilising resources quickly and by rapid interventions such as spraying walls and roofs with long-lasting insecticides (indoor residual spraying) or mass drug administration.
In practice, epidemics are rarely, if ever, managed this way. Most effectively run their natural course, often with very high rates of illness and death.
Rather than pursue elusive predictive modelling, countries at risk of malaria epidemics should focus on developing reliable alerts of incipient epidemics — they need appropriate and sustainable case monitoring systems.
Of course, providing these is a far from trivial task, and for many developing countries it will require new specialised, streamlined monitoring systems as well as new or enhanced diagnostic services. It will be a major challenge to introduce mechanisms that do not over-burden already overstretched health systems. Another challenge will be to convince policymakers that the extra resources required for these monitoring systems are justified.
Decision-support mechanisms that quickly translate epidemic warnings into a series of explicit and predeﬁned responses are also essential. Experience from Uganda has shown that without major changes to the way malaria control programmes respond to emergencies, potential benefits offered by MEWS are unlikely to be realised.
An obvious and growing need
Inadequate case monitoring systems represent a serious missed opportunity for making epidemic control efforts more effective. But with a growing number of countries across the developing world now adopting a policy of malaria elimination, the need for systems that can identify outbreaks and facilitate rapid patient follow-up is becoming increasingly obvious.
Now is the time to redress the balance and position MEWS as a standard approach for national surveillance, rather than a tool that is specific to one epidemiological ‘niche’.
Systems that use handheld computers or mobile phones to rapidly send and receive data on cases from remote health units have already been developed in Tanzania, Thailand and other countries. Documenting these pioneering efforts will be important to determine best practice and identify common issues around implementation.
Ironically, once in place, these systems will begin to generate the large amounts of high quality disease data that modellers need to develop and test more reliable predictive early warning models. It‘s high time we put the horse back before the cart.
Jonathan Cox is a senior lecturer at the London School of Hygiene and Tropical Medicine.