Health - Malaria may kill far more people than we thought

Malaria kills almost twice as many people worldwide as the World Health Organization estimates, according to a major study that is causing a stir this week. It claims a staggering eight times as many adults die of the disease in Africa as the WHO says.

The report has not convinced everyone, though. Malaria experts on a blue-ribbon committee that is meeting this week for the first time at WHO headquarters in Geneva, Switzerland, to advise on fighting the disease, say that conclusion is flawed.

It is difficult to estimate how many people die of malaria because stricken countries are too poor to produce reliable health statistics. The new analysis, by the Institute for Health Metrics and Evaluation (IHME) in Seattle, Washington, tries to overcome this problem with innovative modelling techniques.

It claims that malaria killed 1.2 million people worldwide in 2010, nearly twice the WHO's estimated 655,000. Compared with the WHO, it found 1.3 times more deaths among children under 5 years old in Africa, and 1.8 times more deaths for all ages outside of Africa.

Adult immunity

However, the difference between these IHME and WHO numbers may not be that significant, given the statistical uncertainty in all estimates, says Richard Cibulskis of the WHO's Global Malaria Programme and a member of the committee meeting in Geneva this week. What is different is the IHME's estimate for deaths in children over 5 and adults in Africa, which is a massive 8.1 times as high as the WHO's.

"We were surprised by that," says Steve Lim of the IHME, an author of the report. "The standard belief is that in places with a lot of malaria, adults have enough immunity to prevent severe disease."

Because of this, says Lim, doctors in malarial regions underdiagnose severe malaria in adults. The IHME factored this into its analysis and counted many adult deaths recorded as "fever" as malaria.

Cibulskis disputes this interpretation: he says that recent studies show that adults in Africa diagnosed with malaria in fact often have something else, so numbers may be lower than official statistics suggest, not higher.

Model behaviour

The rest of the increase in African adult deaths is down to IHME's modelling, says Lim. Studies of malaria deaths are mostly surveys of small areas for short times. The team looked at how various "drivers" of malaria – population growth, issues around mosquito control, rainfall, resistance to drugs and others – correlated with those deaths, and derived statistical models to predict how many deaths there should be everywhere, all the time, as a result of these drivers.

Various sets of malaria drivers were used to create several models with different degrees of uncertainty, which made different predictions of malaria deaths. These were averaged to obtain a single estimate. "We tested how well models based on part of a malaria data set predicted the rest of that data set," says Lim, and selected the cluster of models that did best.

The WHO uses simpler modelling and considers only population growth and mosquito control as drivers, says Lim.

Bob Snow, head malaria epidemiologist at the University of Oxford's research unit in Nairobi, Kenya, and another member of the committee meeting in Geneva, agrees that the WHO underestimates malaria deaths. However, he says there is good evidence that adults rarely die of malaria in Africa, and criticises the IHME for applying statistics to poor data, saying that the institute has not understood the disease.

"We simply do not know what the real global malaria burden is," he says. The answer is better disease surveillance, he says, so "disease burden estimation is anchored in what we know, not what we can model".

Journal reference: The Lancet, vol 379, p 413

http://www.newscientist.com/