Mildly cold weather 'more deadly' than heatwaves or very cold snaps

Thursday May 21 2015

"Mildly cold, drizzly days far deadlier than extreme temperatures," The Independent reports. An international study looking at weather-related deaths estimated that moderate cold killed far more people than extremely hot or cold temperatures.

Researchers gathered data on 74,225,200 deaths from 384 locations, including 10 in the UK. The results showed that the days most countries have the fewest deaths linked to temperature are those with warmer temperatures than average.

Therefore, the researchers calculate, the majority of "excess deaths" occur on days that are colder than average. Because extreme temperatures occur on only a few days a year, they have an impact on fewer deaths than the majority of moderately cold days.

Overall, the researchers say, 7.71% of all deaths can be attributed to temperature based on their statistical modelling.

One hypothesis offered by the researchers is that exposure to mild cold may increase cardiovascular stress while also suppressing the immune system, making people more vulnerable to potentially fatal conditions.

The researchers suggest that their findings show public health officials should spend less time planning for heatwaves, and more time thinking about how to combat the effect of year-round lower than optimum temperatures.  

Where did the story come from?

The study was carried out by researchers from 15 universities and institutes in 12 countries led by a team from the London School of Hygiene and Tropical Medicine.

It was funded by the UK Medical Research Council. The study was published in the peer-reviewed medical journal The Lancet and has been made available on an open-access basis, so it is free to read online or download as a PDF.

The media reports focused on the finding that moderately cold weather – such as that experienced in the UK for much of the year – caused more deaths than hot weather or extremely cold weather. The Daily Telegraph gave a good overall summary of the research.

The Independent's claim that "mildly cold, drizzly days" are "far deadlier than extreme temperatures" is an extrapolation, as the study didn't look at drizzle or rain as a risk factor, just temperature.

The Guardian includes a number of reactions from independent experts, such as Sir David Spiegelhalter's, presumably tongue-in-cheek, suggestion that "perhaps they are really saying that the UK climate is killing people". 

What kind of research was this?

This study was a meta-analysis of data on temperatures and deaths around the world to find out what effect temperature has on the risk of death, and whether people are more likely to die during cold weather or hot weather.

The researchers used statistical modelling to estimate the proportions of deaths in the regions studied that could be attributed to heat, cold, and extreme heat and cold. This type of study can tell us about links between variables such as temperature and death rates, but not whether one causes the other. 

What did the research involve?

Researchers collected data on temperature and mortality (74,225,200 deaths) from 384 locations in 13 different countries, during time periods from 1985 to 2012. They used statistical analysis to calculate the relative risk of death at different temperatures for each location.

The countries included were Australia, Brazil, Canada, China, Italy, Japan, South Korea, Spain, Sweden, Taiwan, Thailand, the UK and the US. About one-third of the locations sampled were in the US.

The researchers were not able to adjust figures to take account of the potential effects of other factors, such as income levels in the different countries, although they used air pollution data when it was available.

The researchers divided the temperature data from each location into evenly spaced percentiles, from cold to hot days. This was so the temperatures for the coldest days would be in the lowest percentiles of 1 or 2, while the highest temperatures would be at the top range, 98 or 99.

They defined extreme cold for a location as below the 2.5th percentile, and extreme heat as above the 97.5th percentile. They looked for the "optimum" temperature for each location, being the temperature at which fewest deaths attributable to temperature were recorded.

They calculated the deaths linked to temperatures above or below the optimum, and sub-divided that again to show deaths linked to extreme cold or heat.

The statistical analysis used a complex new model developed by the researchers, which allowed them to take account of the different time lag of different temperatures.

The effects of very high temperatures on death rates are usually quite shortlived, while very cold temperatures may have an effect on deaths for up to four weeks. 

What were the basic results?

Across all countries, colder weather was linked to more excess deaths than warmer weather – approximately 20 times as many (7.29% deaths in colder weather compared with 0.42% in warmer weather). 

For all countries, the optimum temperature – when there were fewest deaths linked to weather – was warmer than the average temperature for that location.

In the UK, for example, the average temperature recorded was 10.4C, while optimum temperature ranged from 15.9C in the north east to 19.5C in London. The optimum temperature for the UK was in the 90th centile, meaning that 9 out of 10 days in the UK are likely to be colder than the optimum.

The proportion of all deaths linked to extremely hot or cold days was much lower than that linked to less extreme hot or cold. The researchers say extreme heat or cold was responsible for 0.86% of deaths according to their statistical modelling (95% confidence interval 0.84 to 0.87).

However, the relative risk of dying at extremes of temperatures was increased, with a sharp increase in deaths at the hottest temperatures for most countries.  

How did the researchers interpret the results?

The researchers say their results have "important implications" for public health planning, because planning tends to focus on how to deal with heatwaves, whereas their study shows that below-optimal temperatures have a bigger effect on the number of people who die.

They say deaths from cold weather may be attributed to stress on the cardiovascular system, leading to more heart attacks and strokes. Cold may also affect the immune response, increasing the chances of respiratory disease.

They say their results show that public health planning should be "extended and refocused" to take account of the effect of the whole range of temperature fluctuation, not just extreme heat. 


Many of the headlines focus on the finding that moderate cold may be responsible for more deaths than extreme hot or cold weather.

Perhaps more interesting is the finding that the optimum temperature for humans seems to be well above the temperatures we usually experience, especially in colder countries like the UK. If this is true, then the finding that most deaths occur on days colder than the optimum is unsurprising, as most days are colder than the optimum temperature.

The relative unimportance of very hot or very cold days in terms of mortality is interesting, because most research and public health planning has focused on extreme weather. However, this depends partly on the definition of extreme temperature.

The researchers used 2.5 upper and lower percentiles to decide on what was extreme for a particular location, so by definition these temperatures are experienced on very few days. Even though the relative risk of death is increased on those days, the absolute number of deaths is nowhere near as high as on the majority of days.

That doesn't mean it's not worth planning for the increased risk of deaths during extreme temperatures. In London, for example, the relative risk of death is more than doubled on days with temperatures below 0C, compared with days at the optimum temperature of 19.5C.

Read more advice about coping with heatwaves as well as very cold snaps.

There are some limitations to the study we should be aware of. First, although it sampled data from 13 countries from very different climates, it didn't include any countries in Africa or the Middle East. This means we can't be sure the findings would apply worldwide.

Second, the study did not take into account some confounders that could affect how many deaths occur in warmer or colder periods – for example, levels of air pollution, whether people have access to shelter and heating, the age make-up of a population, and whether people have access to nutritious food all year round.

This also makes it difficult to know how governments or public health bodies can make plans using this new data, as we don't know whether the effects of moderate cold on mortality could be affected by public health measures.

In the UK, the NHS already plans for more hospital admissions during the winter months, taking account of factors such as the amount of flu-like illness circulating in the population, as well as the temperature.

Read more advice about winter health.

Analysis by Bazian
Edited by NHS Choices