The cold curves of mortality

Though it’s still not exactly warm out there, in the UK we’ve at last come to the end of the coldest winter for 30-odd years. This was due to a more southerly-than-normal-for-the-time-of-year jet stream, as the somewhat controversial Paul Hudson explains here. He has an interesting if not entirely convincing (to this admitted non-meteorologist/climatologist) explanation which links the cold snap to the recent lack of sunspot activity. But that’s by-the-by.

The snow was photogenic and fun in the beginning, but persistent cold temperatures come with a price attached. This price is paid disproportionately within the population— and interestingly, disproportionately between populations, too.

Cold weather has a number of health effects, which are not fully understood. In fact, the effect of temperature on mortality is a surprisingly complex interaction which remains under very active investigation.

Unusually high temperatures lead to an immediate increase in deaths from cardiovascular mortality— heart attacks and the like— although it is believed that a sizeable proportion of this effect is down to a phenomenon named ‘mortality displacement’— or more grimly, ‘harvesting’— in which vulnerable people who were destined to die in the near future anyway have their deaths brought forward by a few days, with a corresponding slight reduction in mortality a few days after the heatwave passes.

Most (if not all) cold-related deaths, on the other hand, are believed to be genuinely avoidable. The obvious cause of cold-related deaths is hypothermia, but this only accounts for a small and declining proportion of winter deaths, and other factors, both known and unknown, are at play: for example, low temperatures increase the tendency of blood to clot—promoting heart attacks, pulmonary emboli and the like— as well as increasing mortality from respiratory diseases. The effect of cold is more insidious than that of heat: it takes about two weeks for it to fully manifest in increased mortality rates. Complicating all of this is the seasonal nature of some infectious diseases, such as influenza, although epidemiologists take great pains to account for such confounders in their analyses.

A 2006 time series study showed that the elderly, women, nursing and care home residents, and deprived populations (though only in rural areas) are at particular risk from ambient temperature-related deaths in England and Wales. The lower two graphs below look at cold-related mortality (by incorporating the averaged temperature of the previous 13 days) by age group and cause of death:
(CVD: cardiovascular disease. Resp: respiratory disease)

What’s really interesting, though, is what happens when you look at temperature related deaths across different populations in different parts of the world. There’s usually a roughly U-shaped pattern of temperature-dependent mortality, with a ‘just-right’ temperature at the bottom of the U at which the fewest people die. You can see this in some of the graphs above, and lots of studies have reproduced similar shaped graphs in different settings. This ‘temperature of minimum mortality’ can be expressed as a range, because the base of the U is usually quite flat. Yet there’s considerable variation from country to country and city to city— both of the minimal mortality temperatures, and the width of this ‘comfort zone’.

The 2008 ‘ISOTHURM’ study looked at twelve cities in middle income countries. The temperature-mortality curves are below.

Again, the lower group of graphs (b) looks at cold mortality, with temperatures averaged over the previous 13 days. Compare say, Sofia and Cape Town: it’s clear that in Sofia, which has cold winters, the temperature threshold for rapidly increasing cold mortality is lower than that in the warmer Cape Town— although there’s a gradual increase in mortality below 20 degrees C, it only jumps upwards below 0 degrees. In Cape Town, the threshold for rapidly increasing cold-mortality is well-defined at a much higher temperature: about 19 degrees celsius. People dress, behave, build and insulate according to their local climate, so perhaps the difference in thresholds is not so surprising (this finding has been used to argue that we don’t need to worry about heat-related mortality from global warming). But the differences in the overall shape of the graphs— the widths of the ‘comfort zone’ temperature bands, and the gradients of the slopes of increasing mortality at both temperature extremes— suggests that it’s not just about temperature and uniformly stable, local climate-adapted human populations. Although differences in climate or climate variability unrelated to just latitude-dependent temperature (such as arid continental vs maritime climates) may account for some of these differing patterns, perhaps the adaptive capacity of the local population has something to do with it too.

Further reading:
2006 study identifying subgroups vulnerable to ambient temperature-related mortality in England and Wales
Declining vulnerability to cold in London over the 20th century
Study of temperature vulnerability in 50 US cities (uses case-crossover design rather than time series) and
Some recent-ish blog discussion on this sort of thing, with links to yet more studies


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: