 | | | 
 | 
Sponsored by
the Natural Environment Research Council
and
the Medical Research Council | | 
 |
|  | 
|  |  | | (This debate is closed and is a read-only archive.) | The World Health Organisation estimates that 150,000 deaths each year can be attributed to the effects
of climate change. Will this figure rise as a result of global warming? What do we
know about the likely effect of climate change on our health? And how should
we manage new risks?
| Unpicking the evidence
[1-Mar-2005]
| | 'Tougher tests are required before we can conclude that greenhouse gases are killing people.' |  | | | The World Health Organisation (WHO) has estimated that climate change was responsible for approximately 2.4 per cent of worldwide diarrhoea, and six per cent of malaria in some middle-income countries, in the course of the year 2000, contributing to a total of 150,000 deaths worldwide (1). Does this mean we can say that human influence on climate is already killing 150,000 people per year? While this line of reasoning sounds plausible, the WHO is careful to stress that pointing the finger of blame (which it calls 'causal attribution') is complicated, so we should be cautious about jumping to conclusions.
| Human societies adapt to the climate to which they are accustomed, so any change in climate is likely to cause stress, and in extreme cases an increase in mortality. So 'climate change kills' is a truism, if it refers to any form of climate change, be it caused by a volcano, greenhouse gases or an incoming asteroid. But this is just pedantic: by 'climate change' most people mean human-induced climate change due to rising greenhouse gas levels. Before we can blame deaths on a specific external factor such as greenhouse gases, we need to understand how that factor affected the climatic conditions directly responsible for those deaths. And this is where it gets complicated.
| Many studies of the impact of climate change on health begin by showing that mortality (for example) is associated with a particular weather variable, like wet season precipitation. This association is normally based on studying the health impacts of year-to-year fluctuations in that weather variable that are nothing to do with human influence. If a significant trend is observed in that weather variable, of the size and direction we should expect due to human influence on climate, one might be tempted to conclude that human-induced climate change is causing some increase in mortality. Should we accept this logic?
| I would argue we should not, and that tougher tests are required before we can conclude that greenhouse gas emissions are killing people. The signal may just now be becoming strong enough for these tougher tests to be passed, but it is crucial for the credibility of the whole science of attribution that we don't blow the whistle too early.
| One problem is that the health impact of a change in the weather depends not only on how large that change is, but on how long it lasts. An agricultural society might be able to survive a single year's drought without significant adverse health effects, but would be much more severely affected by 20 years of below-average rainfall. Conversely, societies can adapt (or simply migrate) in response to very slow climatic changes that take place over centuries without discernible adverse health effects. Rapid, quasi-permanent changes that take place over a few decades, such as the cooling that ended the Medieval climate warm period and may have been linked to the onset of the Black Death in Europe, are much more traumatic. Of course, a rapid, quasi-permanent change taking place over only a few decades is precisely what we are facing at present. So finding an association on one timescale and assuming it holds on another will always be problematic.
| A second problem is that finding a trend in some climate variable, and showing that this trend is consistent with the trend we would expect to be due to human influence on climate, is not enough to attribute that trend to human influence. We also have to show that the trend is not consistent with plausible alternative explanations, like internally generated climate variability or naturally driven climate change. This is relatively difficult, given the many uncertainties in climate modelling, and so far has only convincingly and consistently been achieved for temperature changes averaged over relatively large (sub-continental) regions.
| So far, there is no consistent evidence of significant human influence on either mean precipitation (rain and snowfall) or the size of precipitation extremes, although there have been significant naturally driven fluctuations in precipitation over the past few decades. Models predict that precipitation will change in many regions in response to climate change in the future, so if we observe a trend in a health problem that we know is linked to precipitation, we might be tempted to blame that trend on human influence on climate, but we would be wrong (so far).
| So what are these tougher tests that I would advocate before we start claiming that greenhouse gases kill? To answer this question, we have to get to grips with the difference between weather and climate.
| Greenhouse gases affect climate, or the weather we should expect, on average, along with its variability, at a given time of year. In general, it is extreme weather events (floods, droughts, cold snaps, heat waves) that cause problems for human health. As Ed Lorenz, the father of modern chaos theory, put it: 'Climate is what you expect, weather is what you get.' In the twenty-first century, climate is what you affect, and weather is what gets you. So how can we make the link? Because the climate is a chaotic system, there will almost always be some chance of a particular weather event occurring in an unmodified, pre-industrial climate, so we are unlikely ever to be able to say that 'but for' human influence on climate, that weather event would never have occurred. What we can quantify is how much human influence on climate may have increased the risk of such a weather event occurring in a particular year, or 'loaded the weather dice' in its favour. Is this enough to claim that human-induced climate change is already killing people?
| For example, the heatwave in the summer of 2003 is estimated to have caused between 20,000 and 30,000 'excess deaths' (above what would be expected in a typical August) across Europe (2). The epidemiological evidence is reasonably clear, and there was no marked decrease in mortality following the heatwave, so these were not deaths due to a simple 'harvesting' effect that were brought forward by only a few weeks: many of the victims would otherwise have survived for years. In a paper published in December 2004 (3), we argued that past human influence on climate is likely to have increased the risk of such a heatwave by at least a factor of two, more likely a factor of four or more. Our paper focused on large area averaged temperatures over the whole of Southern Europe, since that was the kind of region we were able to simulate realistically with current climate models. More work will be required to establish if the same levels of human contributions to risk apply on smaller (eg, provincial) scales, but this is an example of how the evidence is going at the moment.
| Statistical associations will always be open to challenge, because of the timescale problem and because of the difficulty of distinguishing a trend that is consistent with human influence on climate from a trend that is attributable to human influence on climate. Instead, to quantify the impact of human-induced climate change on health, we need to focus on specific weather events that have well-documented health impacts, and quantify the contribution of human-induced climate change to the risk of those events occurring.
| So I would say that the correct interpretation of the WHO statistic quoted at the beginning is that climatic changes, to which human influence may have contributed although we cannot yet say how much, were responsible for 150,000 deaths in the year 2000. This doesn't sound particularly dramatic. But suppose regional studies focusing on provincial scales drew similar conclusions to those we drew for Southern European area-averaged temperatures regarding the contribution of human influence to the risk of extreme temperatures observed in the summer of 2003. Then we will be able to say, in a statistical sense, that past emissions of greenhouse gases killed several thousand elderly Europeans that summer. And that has some serious implications.
| Myles Allen is head of the Climate Dynamics Group at Oxford University.
| (1) Climate change and human health - risks and responses - summary, World Health Organisation
(2) Hot news from summer 2003, Christoph Schär and Gerd Jendritzky, Nature, 2 December 2004 [pdf format: 129kb]
(3) Human contribution to the European heatwave of 2003, Peter Stott, D Stone and Myles Allen, Nature, 2 December 2004 [pdf format: 217kb]
|
|
| | |
|
|
|
