I’m returning to an old subject because the Guardian won’t leave it alone. In Losing the Plot I criticised a potentially misleading Guardian headline (“Extreme temperatures kill 5 million people a year with heat-related deaths rising, study finds”) by pointing out that the study referenced by the Guardian in that story actually demonstrated that a human inhabitant of planet earth is almost 10 times more likely to die of cold than of heat. Another study on the same topic has just been published, and unfortunately for the Guardian’s climate doomsayers it also undermines the narrative that rising temperatures (this time specifically in Europe) will lead to more deaths.

What, then, is to be done? Why, the answer is obvious. It’s to do what the Guardian always does in these cases, namely to choose a headline that spins the study’s findings in the desired direction, and then to cherry-pick the findings so that the required impression is created in the mind of the reader. Thus we find an article with the heading “Heat deaths in Europe may triple by end of the century, study finds” and the sub-heading “Countries in south most at risk, with rise likely to outstrip fall in cold-related deaths if global heating hits 3C or 4C”.

Having got off to this good start, the article proceeds to build upon these foundations. As always, it’s cleverly done, and the data cited do appear in the study. The problem is the way in which they are presented, which creates a different impression from the one justified by an objective reading of the study. Thus the Guardian concentrates on soundbites such as these:

Heat deaths in Europe could triple by the end of the century, with the numbers rising disproportionately in southern European countries such as Italy, Greece and Spain, a study has found.

If global heating reaches a catastrophic 3C or 4C, the researchers concluded, the rise in heat deaths will greatly outstrip the fall in cold deaths.

Deaths from warm weather could kill 129,000 people a year if temperatures rise to 3C above preindustrial levels. Today, heat-related deaths in Europe stand at 44,000. But the yearly death toll from cold and heat in Europe may rise from 407,000 people today to 450,000 in 2100 even if world leaders meet their global warming target of 1.5C, the study found.

That last one is particularly clever, with its talk of 44,000 heat-related death, followed by a neat segue into ten times as many cold and heat related deaths by 2100.

The research comes on the back of a series of scorching heatwaves that have wreaked havoc across the continent. Its results challenge arguments from climate deniers that global heating is good for society because fewer people will die from cold.

And so, job done. The claims of evil climate deniers are undermined very successfully. Even modest warming will cause more heat-related deaths. More people will die. If temperatures rise by 3C or 4C more deaths from heat will outstrip the reduced deaths from cold. Whichever way you look at it, warming is bad news. Game over. Take that, deniers everywhere.

There’s just one problem – the detailed findings of the study itself. Shall we take a look at what it says?

Probably the main thing to note is the methodology:

Mortality was projected for present conditions observed in 1991–2020 and for four different levels of global warming (1·5°C, 2°C, 3°C, and 4°C increase) by regions, and subregions using an ensemble of 11 climate models produced by the Coordinated Regional Climate Downscaling Experiment-CMIP5 over Europe, and population projection data from EUROPOP2019.

The key findings are helpfully summarised near the beginning of the study:

Our results highlight regional disparities in temperature-related mortality across Europe. Between 1991 and 2020, the number of cold-related deaths was 2·5 times higher in eastern Europe than western Europe, and heat-related deaths were 6 times higher in southern Europe than in northern Europe. During the same time period, there were a median of 363 809 cold-related deaths (empirical 95% CI 362 493–365 310) and 43 729 heat-related deaths (39 880–45 921), with a cold-to-heat-related death ratio of 8·3:1. Under current climate policies, aligned with 3°C increase in global warming, it is estimated that temperature-related deaths could increase by 54 974 additional deaths (24 112–80 676) by 2100, driven by rising heat-related deaths and an ageing population, resulting in a cold-to-heat-related death ratio of 2·6:1. Climate change is also expected to widen disparities in regional mortality, particularly impacting southern regions of Europe as a result of a marked increase in heat-related deaths.

Let’s pause for a moment. Between 1991-2020 across Europe as a whole an average of 8.3 people died from cold for every one who died from heat. If temperatures rise by 3C by 2100 then that ratio shifts to 2.6 cold deaths for every heat death, and more people will die overall from a combination of heat and cold, but this will be due both to temperature changes and an ageing population. In fairness, the Guardian report does mention this latter (and very significant) factor, but does so almost in passing and mentions it only once, while the entire article is dedicated to climate alarmism.

The detail within the study offers some curious insights. For instance, between 1991-2000, in the Netherlands cold-related mortality was 25 per 100,000 deaths, while in Bulgaria it was 300 per 100,000 (in other words, cold-related mortality was 12 times higher in Bulgaria than in the Netherlands). No doubt this is related to lots of factors (sadly, not explored by the study, so far as I can see, but I assume wealth/poverty and access – or not – to central heating, played a substantial part in those results). Perhaps surprisingly the lowest numbers of cold-related deaths were found to be in central Europe (due to wealth again?) and in parts of southern Europe (presumably due to a warmer climate).

Heat-related deaths were six times more common in the south than in the north. Not surprisingly the risk of heat-related deaths in Scandinavia was low, as it was in the British Isles too (which isn’t a surprise if the non-existent summer of 2024 is anything to go by). Highest risk values from heat were to be found in southern areas and in Croatia.

The ratio of cold-related deaths to heat-related deaths in 1991-2020 is confirmed – approximately 8.31:1 across Europe, but with significant disparities, from 3.3:1 in Slovenia to 132.5:1 in Ireland.

Next follows a very interesting table analysing projected deaths separately from cold and from heat by individual country, in the four scenarios considered (temperatures rising by 1.5C, 2C, 3C and 4C) compared to deaths from those causes in 1991-2020. This is where it gets really interesting, because it illustrates (without separating out and separately identifying) the extent to which an ageing population leads to more deaths from extreme temperatures (both cold and heat). Taking the first few countries alphabetically, the point rather jumps out. The number of cold-related deaths in Austria is projected to be higher under every increased temperature scenario than in 1991-2020, with the greatest increase being from 3,742 (1991-2020) to 5,018 (1.5C), then declining steadily as the temperature rises, but still (at 4C) projected to be 3,846, i.e. more than without any temperature rise. To this curious bystander, it seems obvious that an ageing population is a significant factor in the projected increase in temperature-related deaths, if the number dying from cold is significantly higher after a 1.5C increase in temperature, and is still higher after a 4C increase. Naturally, the numbers projected to die from heat increase with every rise in temperature. The same is true for Belgium, though with a slight decline in cold-related deaths only once the 4C threshold is broached. Bulgaria and Croatia show declining deaths from cold under all temperature scenarios, but Cyprus shows significant increases. The narrative does make the point about ageing, but so far as I can see, the relative importance of ageing and temperature changes aren’t stripped out:

The number of deaths will be highest among people aged 85 years and older. This age group, in addition to having greater vulnerability to extreme temperatures compared with other groups, will also increase in size substantially by 2100 due to the expected rise in life expectancy. The overall net effect on mortality (increase in heat-related deaths minus reduction in cold-related deaths) is expected to remain highly positive for this age group, especially in warmer climate scenarios. [My emphasis].

A follow-on paragraph, related to the above analysis, makes interesting reading:

The direction of the projected change in cold-related mortality risk is mixed depending on the area considered. Under the 3°C increase scenario, standardised death rates due to cold would vary from 80 fewer to 80 additional deaths per 100 000 people compared with the 1991–2020 scenario. Moderate declines in deaths would be observed in regions in eastern Europe (eg, –62·1 in Yambol Province, Bulgaria), with mild reductions in some parts of Germany (eg, –5·4 in Hannover), France, Italy, and Portugal. Moderate to high increases are projected in Poland, Czechia, Ireland, and some Scandinavian regions. In contrast, heat-related standardised rates would univocally increase in all European regions, with increases ranging from 0·5 to 92 additional deaths per 100 000 people in regions of Spain. For heat-related deaths, the pattern of the projected increases follows a north–south gradient, where southern European regions will have the highest increase in excess deaths attributable to heat.

The curious thing is that the population will have aged successfully, despite the supposed threat of growing age-related mortality due to extreme temperatures. This is another fascinating (and rather curious) paragraph:

The role of climate is projected to vary depending on the considered temperature risk, showing a negative contribution to cold-related deaths and positive contribution to heat-related deaths. The size of contribution of warming differs by geographical location, with net negative effects (ie, lower mortality) observed in the north and intensified positive effects (ie, higher mortality) in southern latitudes. The contribution of demographic forces was strongly driven by the ageing component and was much more pronounced for cold-related mortality, since the relative risk of cold is notably higher than that of heat for all age groups, especially for older age groups. Overall, a small negative effect (2·4 fewer deaths per 100 000 people) on the cold-related death rate would be expected in Europe, since the positive impact of ageing on the risk will be outweighed by the negative contribution of climate. Notable increases in cold-related death rates by country are expected in Ireland (39·5 additional deaths per 100 000 people), Slovakia (19·0 additional deaths per 100 000 people), and Norway (17·5 additional deaths per 100 000 people), all of which have a strong ageing population. Conversely, cold-related mortality rates would decrease in Bulgaria (39·2 fewer deaths per 100 000 people) and the Baltic countries (35·8 fewer deaths per 100 000 people in Latvia and 29·7 fewer deaths per 100 000 people in Lithuania), due to a stronger decline in total mortality rates that would compensate for the ageing component.

One more long quote from the study (this time with respect to heat-related deaths) again illustrates the extent to which an ageing population is a factor:

We identified areas of future heightened risk of heat-related mortality (termed hotspots) where marginally greater susceptibility (ageing), greater hazard increase (warming), or a combination of the two are expected to result in a larger increase in mortality risk by 2050. These hotspots are primarily concentrated in southern latitudes, particularly in regions of Spain, Italy, and Greece, but also extend to more northern areas, significantly impacting a substantial part of France. The most eastern regions will be affected by an intensification of warming, but total death risk attributed to heat will be mitigated mainly by a strong decline in total mortality in these areas. In northern Europe, mean heat during summer will increase but not enough to cause additional deaths. However, the expected ageing of the population will make this area more susceptible to extreme heat episodes

The reference to a strong decline in total mortality in eastern Europe suggests in addition, to my untutored eye, that assumed growing wealth (or reduced poverty) in those regions might have a role to play. My suspicion in that latter regard is reinforced by a sentence in the discussion section of the paper, which refers to “the positive association between mortality risks and regional income (a proxy for these risk-moderating drivers), particularly evident in the case of cold mortality.”

In essence, the statistics seem to be heavily driven by projections of an ageing and growing population (of course more people will die every year if there are more people and if a growing proportion of that increased population is older than now):

The evolution of cold-related mortality suggests a more complex interplay between the effect of warming and demographic shifts, which can push cold mortality risk downward or upward (eg, –20·5 expected deaths in Croatia, benefiting from lower overall mortality rates, compared with +39·5 deaths in Ireland, disadvantaged by a marked ageing of its population). The total effect for each region will depend on which of the forces dominates.

Also this:

There are two main factors driving these shifts: climate and demographic changes. Climatic drivers relevant to this impact assessment include the generalised increase in average temperatures, coupled with the proliferation of longer, more intense, and more frequent extreme heat episodes.

Demographic shifts in Europe are predicted to be characterised by three processes: (1) an overall mild decrease (about –5%) in the total European population by the end of the century, subject to strong spatial variability; (2) a general and sustained population ageing process affecting all countries and regions, with the European share of the population aged 85 years and older increasing from 2·9% (1991–2020) to 9·3% in 2100; and (3) a projected increase in life expectancy, indicated by a decrease in the total mortality rate across all age groups. Ageing and warming will be generalised and will extensively affect all the studied regions. However, these effects will be particularly pronounced in the identified hotspots, which can serve as focal points for targeted interventions and actionable adaptive measures. Other factors, such as the evolution of socioeconomic developments or other secondary consequences of climate change on demography—such as the effects of net migration—could influence total temperature-related deaths. Consequently, it should be noted that the projections presented in this study should be interpreted as conditional on current adaptation policies and socioeconomic characteristics.

As the authors of the study acknowledge, their analyses have several limitations, not least of which is the extent to which their results will be distorted by the urban heat island effect:

The results presented are based on observations from a sample of the urban population living in medium to large cities. Consequently, not all regions were adequately sampled (particularly in eastern countries), and results do not cover rural populations. Urban populations typically face higher levels of temperature stress, particularly from heat.

Consequently, they acknowledge that their results might be “slightly overestimated”. In addition, they took no account of possible acclimatisation to warmer summers on the part of European populations, nor did they take any account of possible adaptation measures that might be put in place. With 76 years to go until 2100, my money would be on both of those being significant omissions from the analysis.

Be that as it may, it seems to me that the best way of stripping out the distorting effect of an ageing population on the numbers is to see whether, under any of the hypotheses, more people are projected to die from heat than from cold in Europe by 2100. The answer to that question is a resounding “no”. Across Europe as a whole, the ratio between cold-related and heat-related deaths unsurprisingly drops with every increase in temperature, but the results stubbornly point to more cold-related deaths than heat-related ones under every scenario (6.7:1 if temperatures rise by 1.5C); 4.9:1 if temperatures rise by 2C; 2.6:1 of temperatures rise by 3C; and 1.4:1 if temperatures rise by 4C). Only in Malta are more people projected to die from heat than from cold with a 3C increase (and even then the ratio is just 0.9:1). With a 4C increase, a few extra countries fall into that category, namely Croatia, Cyprus, Greece, Italy, Slovenia, Spain and Switzerland. Even with a 4C increase in temperatures, the numbers dying from cold are still projected massively to outstrip deaths from heat in many countries. For instance, in that scenario, the cold/heat deaths ratio would remain 4.7:1 in Estonia, Latvia and Finland (and 4.8:1 in Lithuania); it would be 7.4:1 in Ireland; 5.3:1 in Norway; 3.8:1 in Sweden and 4.9:1 in the UK. Stripping out demographic factors, it’s obvious that we have an improving position all the way to a 1:1 ratio.

Finally, I note that although the study refers to the effects of rising temperatures in Europe, a number of European countries appear to be excluded from the analysis, notably (but not exclusively) in eastern Europe. Admittedly, political issues might have been an issue in respect of some of them, but I suspect that the omission of such countries distorts the narrative towards that of heat becoming more of a problem than cold. The study never mentions, so far as I can see, Ukraine, Belarus, Russia, Moldova, Bosnia, Serbia, Albania, North Macedonia, Montenegro, Iceland or the Faroe Islands. In all of those countries, whether because of climate or poverty or both, I suspect that cold is more of a problem than heat, such that rising temperatures will benefit their mortality rates.

In short, the study provides us with a great deal of interesting information and food for thought. It has shortcomings, however, some of which its authors acknowledge and others of which they may not be aware. None of them are mentioned by the Guardian, with its fixation on the “climate crisis” narrative. My conclusion is that it remains legitimate for sceptics (who are not “deniers”, Guardian please note) to point to the beneficial effect of reducing numbers of cold-related deaths as temperatures rise. This narrative holds good in Europe specifically, and across the world more generally.