Nature News interview with Simon Hay

by Declan Butler, Nature News

An international team of scientists today published the most detailed global maps of worldwide malaria prevalence ever compiled; maps they will update annually to track the progress of efforts to fight the disease. The maps show that nearly three-quarters of the 2.4 billion people at risk of infection with the deadliest form of malaria live in low risk areas, where the disease could be easily eliminated by using modest control measures.

Nature News asks Simon Hay, of the Malaria Atlas Project at the University of Oxford, UK, how the new maps were built, and how they could help policy-makers.

What makes these new maps so special?

Last year we published maps that split the world into three categories - areas where there was no malaria, areas where the disease was unstable and might be feasible to eliminate, and places where it was stable, or entrenched. Those maps already showed hope, as they demonstrated that whereas some 2.4 billion people live in places where they risked infection by Plasmodium falciparum - which causes the most deadly form of the disease - almost 1 billion people live in places where transmission of the disease is so low that bringing it under control would be eminently feasible, using existing tools such as bed nets and effective drugs.

But those maps were a rough cut, the hors d'oeuvre; the new map is the main course. We've spent two years compiling and modelling data, to quantitatively map malaria prevalence in 2007 - from 0 to 100% - across the entire planet down to 5 x 5 kilometre resolution. So you can even download maps of any country, and see the sub-national geographical distribution. That's never been done before.

The maps took a year of computing time on a supercomputer. The high resolution is computing intensive; it took months just to compute a 5 x 5 kilometre grid of data for Africa.

To speed things up, have you thought of using cloud computing - access via the Internet to a 'cloud' of powerful servers?

Funny you should mention that. We'd thought about it but figured it would be too expensive. But on Friday, Amazon confirmed that they would work with us, and have opened us a free credit line of $12,000 dollars computing time on its system.

So what do the maps show?

Look at the Americas. Almost all the 40 million people there at risk of malaria are living in areas where transmission is below 5%. So technically the obstacles to malaria elimination in the region are low. It's not rocket science; it says that if you can roll out prevention measures, you know they will work, so it's purely a question of resources and logistics, nothing else. We can say the same for vast swathes of Central and South East Asia, although there are pockets of higher transmission areas such as Myanmar and lowlands of Papua New Guinea which are going to require more effort. The maps should of course be valuable guides in those efforts.

Even in Africa, where you have some 660 million people at risk, some 100 million live in areas where the maps show prevalence of less than 5%, so again there are excellent prospects for eliminating malaria from these areas, such as the northern extent of the Sahara, and the Horn of Africa. Again it's the logistical obstacles that will be the rate-limiting step.

Then you can see that many countries and regions in Africa are in the middle with prevalence rates somewhere between 5% and 40%; these are the regions where it is going to take scale up of bednets and other measures to universal coverage to have an effect. But the modelling suggests that if you do and you sustain that coverage, it should be possible to eliminate malaria from many of these regions.

Do the maps reveal any surprises?

Yes. The Horn of Africa, Kenya and Tanzania had much lower prevalences than I would have thought; on a finer scale you can also see many success stories around Africa. I was also surprised that prevalence in west Africa remained so high. We clearly have a huge job to do there. Another thing to notice when you look at the maps is the uncertainty layers that go with them - in some countries we really have a very poor idea of the malaria levels - these maps will also help us prioritize the collection of new information.

Where did you get your data?

The maps are based on parasite prevalence data. These are collected in national and research surveys in which doctors and scientists go in and take blood tests in a population and calculate the percentage of people who are carrying parasites. We've contacted every ministry of health, and every malaria-control project manager we could find, in every malaria endemic country in the world, to ask them for their data. They have been fantastically generous.

The new maps for 2007 are based on over 8,000 surveys, and next year's maps will have over 13,000. We have made the maps open source and are designing the infrastructure to make the underlying entire database open source, so anyone can download it.

Now we have the maps, what's next?

The maps aren't an end in themselves. The idea is to generate them annually so that we can track progress in malaria intervention efforts. Look at Mozambique, where the 2007 map shows a high prevalence; the new data we have coming in already show that recent interventions are having a huge effect, and you'll be able to see these reductions in the 2008 map. The same would be true for Cambodia. It's only the first year of the map, but the data we have in hand also shows that in east Africa, and wide areas of Kenya, as well as the Americas we are seeing real decreases. We intend to make it easy for scientists and policy-makers to see which areas are making progress and which are not, to better evaluate progress towards international and national targets, and control efforts.

http://www.nature.com/news/2009/090324/full/news.2009.184.html