Millions of people succumb to dengue virus infection each year. It's spread by the Aedes family of mosquitoes and causes a haemorrhagic fever, which can range in severity from mild to lethal. But predicting where the hotspots will be, so that appropriate healthcare provisions and anti-mosquito measures can be supplied to the right places is very difficult. Rachel Lowe, from the Catalan Institute of Climate Sciences in Barcelona, has been tackling the problem with a bit of help from FIFA, as she explained to Chris Smith...
Rachel - We developed a prototype dengue early warning system to produce probabilistic forecast of dengue risk 3 months in advance. We specifically designed this system to predict the dengue risk ahead of the 2014 World Cup in Brazil. So, which areas of the country were at a greater risk of experiencing epidemic levels of dengue.
Chris - So, you had the prediction and then of course, now we’re 2 years on from the World Cup, you can test whether or not your model accurately predicted what was going to happen.
Rachel - Yes. We have now compared the observed dengue incidents rate during the World Cup to our predictions to see in which areas of the country the model worked better and was able to successfully predict high levels of dengue risk.
Chris - And was it any good?
Rachel - We found that in the areas experiencing high levels of risk, the model successfully predicted 57 per cent of those areas. We also compared this to a model just based on the historical seasonal averages of dengue. We found that our model was considering better.
Chris - Some people would say well, 50 per cent-ish, that’s not particularly good. Why is it not 100 per cent?
Rachel - Of the areas experiencing high risk, the forecast model was able to predict 57 per cent of those areas compared to 33 per cent just based on seasonal averages. So, current practices is based on generally knowledge from the past to be able to support decisions. So we found that by incorporating seasonal climate information along with epidemiological information 3 to 4 months in advance, we have this added 24 per cent of areas we’re detecting high risk. Of course, because we’re using quite a long lead forecast several months in advance, it’s very difficult to accurately predict 100 per cent of these areas which depends not only on the climate and early cases of the diseases but also on many other factors such as the types of the virus that are circulating and the susceptibility of human populations and how they're behaving on the ground.
Chris - So actually, it is significantly better than what would be achieved if one just said, “Well, where does it normally produce a hotspot? What's been the seasonal average previously?” what do you attribute the improvement to? What's the deciding factor here?
Rachel - Well, this model is able to take into account variations in the climate in space and time. so by incorporating that information which changes from year to year then you can have a better idea of how the mosquitoes might respond to these particular climate characteristics. And also, by incorporating early cases of dengue from the surveillance system several months in advance, you have some sort of indication of what might happen in the coming season.
Chris - More than a hundred countries worldwide have now got dengue activity and it’s a major disease burden internationally so could I take what you’ve pioneered here in Brazil and apply that to say, Indonesia or the top of northern parts of Australia where we do see dengue activity and use the same parameters to make equivalently good predictions there?
Rachel - You could certainly use the same model framework and it would be wise to take data from those areas and formulate the model based on the dengue incidents and the climate conditions, and other socioeconomic factors particular to that location. But we have certainly developed this model framework for other locations such as Thailand and Ecuador.
Chris - Now obviously, the one thing people are very, very worried about at the moment in the same country Brazil is zika virus because they spread in very similar ways. Could we take this model and also now make predictions about other viruses entirely like zika?
Rachel - Absolutely. this model is a climate base model and many diseases including dengue, chikungunya and zika which are transmitted by the Aedes mosquito are sensitive to climate. So this model could certainly be extended to try and predict the risk of the transmission of other viruses such as chikungunya and zika.