08/09/09

Climate complexities stoke disease controversies

Climate models — and thus disease models — are full of uncertainty Copyright: VETS/UCAR

Send to a friend

The details you provide on this page will not be used to send unsolicited email, and will not be sold to a 3rd party. See privacy policy.

Modelling how climate change might affect insect-borne disease is hugely complex — and increasingly controversial, explains Justine Davies.

It’s a compelling idea with far-reaching implications: climate change could spread some of the most deadly infectious diseases to new places, increasing their burden in the developing world.

The theory has been around for decades — and devising models to predict where infectious diseases will spread, and how much they will increase, is keeping scientists increasingly busy. They published 4,000 papers on the subject in 2008 alone — and some researchers have published evidence that suggests the invasion has already begun.

"The idea is intuitively appealing and has spread," says Kenneth Wilson, of Lancaster University in the United Kingdom (UK).

But this year dissenting voices have made themselves heard, asking difficult questions about the quality of the models scientists are using, the data going into them, and whether even improved models could ever predict how diseases will react to climate change.

Growing dissent

When Kevin Lafferty submitted a paper to Ecology (the journal of the Ecological Society of America), questioning the orthodoxy, and suggesting some diseases might even diminish under climate change, his reviewers expressed huge concern. Opinions were so strong that the society took an unusual decision.

Lafferty — from the University of California, Santa Barbara in the United States — would get his paper published. But the "extreme and contrasting views", as Wilson has described the responses from Lafferty’s reviewers, would also be published. The result was 46 pages of discussion (see Debate erupts over effects of climate change on disease).

The journal believes that the debate is profound and has implications that stretch far beyond the world of science, touching professionals, conservationists and policymakers.

"This is because of the funding implications and political fallout that might be generated by questioning the association between climate change and infectious diseases," says Wilson, in an introduction to the journal’s debate.

Double model trouble

Kissing bugs can transmit Chagas disease when they bite

Flickr/gauchocat

Researchers are particularly keen to model climate change’s implications for those infectious diseases that are transmitted by vectors. These include insect-borne diseases such as malaria, dengue fever, sleeping sickness, Chagas disease and leishmaniasis. They are passed to humans when mosquitoes, tsetse flies, kissing bugs and sandflies feed on blood. Sometimes the insects transfer infectious diseases between humans, sometimes between humans and other animals.

The problem is two-fold. Scientists do not yet understand the myriad ways that weather patterns influence this transmission — so the models contain many assumptions, says Simon Hay, a specialist in the statistics of infectious diseases at the UK’s Oxford University. In addition, he points out that the models use future climate scenarios, which are also predictions.

"So there are two large potential sources of error," he says.

Scientists have developed two types of model. Statistical models use data about climates that have supported such diseases in the past, in order to predict where they could occur in the future.

Biological models, on the other hand, consider how the changing climate may affect the complex factors involved in disease transmission, for example the biting frequency and lifespan of the vector, or the time needed for the virus or parasite to develop inside it.

Without a full understanding of how weather patterns affect the twists and turns on this path to disease transmission, both types of model suffer.

Baffling biology and simplistic statistics

Scientists constructing biological models have mostly focused on malaria. Their models must reflect how an increase in temperature or rainfall might affect mosquito populations. Rising temperatures also cause a mosquito to feed more often, increasing the chance that it will pass on any infection that it carries, says Menno Bouma, of the London School of Tropical Medicine and Hygiene, in the UK.

But it would be simplistic to conclude that malaria will inevitably increase with global warming. The parasite is complex. It has separate development phases in both the mosquito and human, and we don’t yet understand how climate affects these.

Krijn Paaijmansa, a specialist in the dynamics of infectious diseases at Pennsylvania State University in the United States, has recently shown that even temperature fluctuations within a single day have an impact, and should be taken into account in models.

Where climate change brings big differences between day- and night-time temperatures, the parasite may spend too little time in its temperature ‘comfort zone’ to develop quickly — and if the parasite develops more slowly inside the mosquito, it’s possible that the short-lived mosquito will die before it can pass the infection on.

There are even more problems to consider for other diseases, such as yellow fever, which has another reservoir in monkeys.

Monkeys are a reservoir for diseases such as yellow fever

Flickr/teague_o

Statistical models face different problems, many of them to do with untangling the effects of climate from other influences.

"As temperature has increased in the past, so have other factors that may contribute to vector-borne diseases," says Sarah Randolph, a parasite ecologist at the UK’s Oxford University. The population may have grown; drug resistance may have arisen. If scientists cannot isolate the effect climate has played in the past, they cannot predict what it will do in the future.

Climate uncertainty

The second part of the modellers’ ‘double problem’ lies with the uncertain climate models themselves.

"The amount by which individual areas will warm is difficult to predict," says Andy Morse, of the UK’s Liverpool University. For example, the tropics may experience much less warming relative to more northern latitudes.

Meanwhile, says Morse, future rainfall scenarios, drawn up by the Intergovernmental Panel on Climate Change (IPCC), are even less robust than temperature scenarios. Some predict more, and some less rainfall — but on average there may be little change.

What’s more, adds Madeleine Thomson of the International Research Institute for Climate and Society, the IPCC scenarios show simply a long-term trend within which fluctuations are likely. There may, for example, be a decade or so of cooling superimposed on a long term trend of warming.

And there are other factors that may lie outside the scope of many models. A simple but powerful one is that a key player in disease transmission may not be able to reach an otherwise perfect setting. North America had an environment suitable for the mosquito Aedes albopictus, says Randolph: but the continent remained empty of the insect until it hitched a lift from Japan in consignments of used tyres.

How well can models assess the protective effect of using bednets?

Flickr/Vestergaard Fransen

Even if the protagonists are present, and the models accurately predict whether disease could emerge, there is another layer of complexity: vulnerability to infection. Does the model assess levels of HIV infection, which makes people more susceptible, asks Thomson?

And can the model assess the protective effect of using bednets, spraying and drugs? Lafferty, author of the controversial paper in Ecology, even argues that infections could fall if malaria shifts north into more developed countries that are well-armed against the disease.

Populations vary in density too — something some models fail to consider. Many models, such as one Randolph co-authored in the journal Science, suggest malaria will disappear from some areas and invade others, so there will be little overall change in area affected.

But if malaria manages its predicted move into the African highlands, it will encounter people living up to 100 times more closely packed than in dry areas. "There would be a dramatic increase in malaria cases," says Bouma.

If malaria spreads to the African highlands, it will encounter people living at up to 100 times greater density

Flickr/Kakenyi

Too specific to be useful

Some scientists say the perfect model is probably impossible, and may not even be desirable, being relevant to such a small area that it would not be very useful.

"If too many factors are included in a model it will be very accurate, but too specific," says Andrew Dobson, of Princeton University in the United States.

So by the time models become robust for all purposes, will it be too late? Dobson rejects this idea. "It is never too late," he argues. "The science of developing models for vector-borne disease is relatively new, and is attracting more and more scientists," he says.

Some analysts point out that a model’s robustness depends on the question being asked.

Thomson says: "We need to start by knowing what question needs to be answered. On a long term global scale, models predicting change in vector-borne diseases based on the IPCC scenarios are useful to prompt countries to do something to mitigate climate change."

She would like to see two changes in the approach to modelling.

"The scientists need to agree on what constitutes evidence and work closely with climatologists so that they can understand and interpret the data."

She also stresses the role of developing countries: "International scientists often only use global climate products. National meteorological agencies and local scientists often have much better local information".

As for Wilson, he ends his paper by highlighting issues on which scientists mostly agree: that climate change is altering the distribution and incidence of some infectious diseases and will continue to do so; that isolating how climate influences the spread of disease is a tough challenge; that data and modelling approaches must be improved; and that factors intrinsic to specific diseases — such as evolution and immunity — will also have a role.

The thorny question of how important climate change is in changing distributions of infectious diseases needs two ingredients he concludes: informed discussion and the passage of time.