Over the past 50 years, Europe has experienced a significant increase in temperatures, with the average temperature rising by 1.7 °C since the pre-industrial era. According to the European State of the Climate 2020 report, the 2010s were the warmest decade on record in Europe, and there has been a global increase in the frequency of warm days and nights (Copernicus Climate Change Service, 2020). At the same time, Europe's population is ageing rapidly, with the percentage of those aged 65 and older projected to reach 27% by 2050. This demographic trend poses significant challenges for pension systems, healthcare services and long-term care. Understanding the relationship between extreme temperatures and health in the older population is crucial to public health in ageing societies (Kovats and Hajat, 2008). Climate change could increase future temperature-related mortality and morbidity, particularly among the older population (Bunker et al., 2016; Harper, 2019; Leyva et al., 2017; Schneider and Breitner, 2016).

Extreme temperatures put the body under stress, requiring it to exert extra effort to maintain a comfortable internal temperature of 37 °C (Cheshire, 2016). The weakening of the physiological response to the environment with age increases the likelihood of a failure in thermoregulation following exposure to heat and cold, which could lead to mortality and multiple morbidity from causes such as heat stroke and respiratory diseases (Ye et al., 2012).

Several studies have found that heat and cold increase mortality (e.g. Conte Keivabu, 2022; Gasparrini and Armstrong, 2011). A large multi-city study (Gasparrini et al., 2015) attributed roughly 8% of mortality to temperature, with most casualties following days colder than the optimum, and a much smaller number of casualties following days warmer than the optimum. Overall, however, the contribution of extreme temperature days to mortality appears to be comparatively low. Research on temperature-related morbidity is more outcome-, exposure- (Bhaskaran et al., 2009; Cicci et al., 2022; Phung et al., 2016; Ryti et al., 2016; Turner et al., 2012; Ye et al., 2012) and context-dependent (Michelozzi et al., 2009). A minority of these studies have focused specifically on the older population (see Åström et al., 2011; Bunker et al., 2016).

The objective of this study is to enhance our understanding of the impact of extreme temperatures on morbidity, particularly the risk of hospitalisation, medication usage for heart conditions, and experiencing the onset of cardiovascular diseases (CVDs) among older adults in Europe. We also account for how these effects differ depending on geographical location and climate conditions, given the tendency to adapt to the local climate (Kovats and Hajat, 2008); and depending on socio-demographic factors, including gender (Gifford et al., 2019), education (Gronlund, 2014), the urban-rural divide (e.g. Kovats and Hajat, 2008) and partnership and parenthood status (e.g. Conte Keivabu, 2022). These factors are known to influence individuals’ exposure and vulnerability to climate.

Previous studies relied on vital statistics, census data and hospital records, which offer detailed information on health-related events. However, many of these data sources are not publicly accessible to researchers, do not provide extensive information on socio-economic characteristics or are limited to multiple cities (Michelozzi et al., 2009) or single cities/countries (e.g. Fonseca-Rodríguez et al., 2021; Linares and Díaz, 2008). In the present study, we use data from the Survey of Health, Ageing and Retirement in Europe (SHARE) (Börsch-Supan et al., 2013), which have, to the best of our knowledge, not been previously employed for research on this topic. In the European context, the SHARE data offer researchers a unique opportunity to compare older populations living in several European regions over the span of a decade based on a wide range of health and socio-economic characteristics.

There is a vast body of literature on the relationship between temperature and health. The bodily stress caused by uncomfortable temperatures triggers physiological responses designed to keep the body’s internal temperature at about 37 °C, the human temperature of comfort. Failures in thermoregulation can lead to hypothermia in response to prolonged exposure to cold, and to hyperthermia in response to prolonged exposure to heat (Cheshire, 2016). Hypothermia occurs when, due to failed thermoregulation, the body temperature drops below 35 °C, leading to symptoms such as shivering, drowsiness and, ultimately, death if the condition is not addressed promptly (Osilla et al., 2023). Conversely, hyperthermia (i.e. abnormally high body temperature), which can result from excessive heat exposure, occurs when the core body temperature reaches 40 °C or higher. Some of the symptoms related to hyperthermia are sweating, rapid pulse, dizziness and nausea, which can progress to heatstroke, and then to death (Cheshire, 2016). In addition to triggering these acute conditions, uncomfortable temperatures can cause other illnesses affecting the cardiovascular and respiratory systems and mental health (Achebak et al., 2019; Mullins and White, 2019). Importantly, some socio-demographic groups are more vulnerable than others to the impact of extreme temperatures.

The older population is among the most vulnerable to elevated temperatures and cold spells (Cicci et al., 2022; Phung et al., 2016; Ryti et al., 2016; Ye et al., 2012). The factors that contribute to this vulnerability are both medical and socio-behavioural. Physiologically, the ageing process can lead to oxidative stress, inflammation and myocardial deterioration, which may increase the risk of developing health conditions such as high blood pressure, hypertension, coronary heart disease, arrhythmia (Rodgers et al., 2019) and diabetes (Wang et al., 2021). These conditions may be further exacerbated by extreme temperatures, which could trigger the onset of cardiovascular diseases or worsen pre-existing conditions (Ratter-Rieck et al., 2023). Furthermore, pre-existing health conditions can impair thermoregulation (Osilla et al., 2023), which may, in turn, increase susceptibility to temperature-related illnesses. In addition, socio-behavioural factors, such as having a mental disorder, living alone, feeling fatigued, being sleep deprived (Minor et al., 2022; Teyton et al., 2022) and being confined to bed, can alter coping strategies of heat and cold avoidance (Åström et al., 2011).

Most studies included in literature reviews and meta-analyses have reported a significant relationship between ambient temperature and all-cause and specific-cause morbidity. Hot spells usually have short-term effects lasting a few days, while the effects of cold spells can unfold over several weeks (Åström et al., 2011; Phung et al., 2016; Turner et al., 2012; Ye et al., 2012). Ryti et al. (2016) found that cold spells are associated with increased morbidity, particularly among people >65 years old. The authors observed that while the results of studies on causes of mortality are generally consistent, the substantial heterogeneity of the findings of morbidity studies makes it hard to quantitatively summarise the evidence. Bunker et al. (2016) found that a 1 °C reduction in temperature increases the risk of cold-induced pneumonia and respiratory morbidity, whereas a 1 °C increase in temperature increases the risk of cardiovascular disease, respiratory disease, diabetes mellitus, genitourinary conditions, infectious disease and heat-related morbidity. Bhaskaran et al. (2009) reported a short-term increase in the risk of myocardial infarction following exposure to low temperatures, and an increase in this risk at hot temperatures. Sun et al. (2018) found an immediate association between myocardial infarction and heat exposure and heatwaves, and a delayed association between myocardial infarction and cold exposure. Morbidity causes related to heat and cold spells are usually measured via hospital records, which implies that individuals underwent (emergency) hospitalisation. In a study of European cities, Åström, Bertil and Joacim (2011) found that while respiratory admissions increased during hot days and heatwaves, there was no or a slightly negative association between cardiovascular and cerebrovascular admissions and high temperatures. Phung et al. (2016) reported an increase in cardiovascular hospitalisations related to cold exposure, heatwaves and increases in diurnal temperatures, but not to heat exposure. Similarly, Turner and colleagues (2012) found no association between increased ambient temperature and cardiovascular morbidity. Leyva et al. (2017) also reviewed the effects of cold and heat events on fluid and electrolyte disorders, ischemic heart diseases, and infectious diseases. In addition, Cicci and colleagues (2022) observed no association between high temperatures and ischemic heart disease (IHD), heart failure, dysrhythmia and some cerebrovascular-related hospital encounters. However, the authors also found evidence of a relationship between high temperatures and emergency department visits and hospitalisations related to total CVDs, hyper/hypotension, acute myocardial infarction (AMI) and ischemic stroke. Overall, the literature reviews and meta-analyses mentioned above highlight the importance of considering the whole spectrum of temperatures and morbidities, as well as the heterogeneity in the effects of temperatures on morbidity outcomes.

In examining the relationship between extreme temperatures and morbidity, it is important to consider that this association may differ depending on climatic conditions. As the sensitivity of populations to cold and heat stress varies geographically, the effects of exposure in one area may diverge from those in another area. Consequently, it is important to analyse the impact of temperature in different climatic areas. Although multi-city studies on this topic exist, to our knowledge, only Michelozzi et al. (2009) has compared cities with very different latitudes and climates, namely, 12 European cities. The authors found that heat exposure increased the risk of respiratory admissions, but had no effect on CVD admissions. The results also indicated that the heat impact was greater in Mediterranean cities, which suggests that adaptation to the local climate (by both individuals and institutions such as healthcare providers) is not sufficient to avoid climate-induced morbidity.

Literature reviews and meta-analyses have compared the effect sizes reported in research carried out in different climatic regions. Åström et al. (2011) found very heterogeneous results by geographical location, with respiratory admissions increasing for individuals aged 75+ in both Mediterranean and Northern-Continental cities, but not for individuals aged 65–74 in Mediterranean countries. Hajat and Kovats (2010) explained that populations who experience higher summertime temperatures (and are closer to the equator) have a higher heat threshold or minimum mortality temperature (MMT), defined as the temperature above which mortality/morbidity risks start to increase (see also Tobías et al., 2021). To cope with the local climate, individuals may implement multiple adaptation measures that can be classified as behavioural (e.g. avoiding outdoor leisure during hot periods of the day, see Fan et al., 2023), cultural (e.g. painting buildings white) or institutional (e.g. placing healthcare systems on higher alert or providing cooling areas during hot days). In addition, individuals may experience long-term physiological adaptation to climate conditions (Rai et al., 2022). Similarly, Turner et al. (2012) found evidence of an association between increased heat exposure at higher latitudes (colder climates) and the risk of cardiovascular hospitalisation. This latitude effect in colder climates was attributed to the populations in these climates having a lower adaptive capacity, as they are often less acclimatised to high temperatures, live in houses that are not suitable for hot weather and lack adaptive measures such as air conditioning. Bhaskaran et al. (2009) also observed that locations with higher mean temperatures are more vulnerable to cold exposure. However, these results contradict those of Phung et al. (2016) and Sun et al. (2018). Phung and colleagues examined the relationship between exposure to temperatures and the risk of hospitalisation for CVDs at different latitudes, and found that in countries at higher latitudes (i.e. colder countries), the effect of diurnal temperatures on the risk of CVD hospitalisation is lower. Meanwhile, Sun and colleagues (2018) studied the effects of heat exposure and cold exposure on the risk of MI hospitalisation, and found that in countries at a higher latitude, the effects of both cold and heat exposure on MI hospitalisation are weaker. While the weaker impact of cold exposure can be due to adaptive capacities at both at the physiological and the behavioural level (e.g. adequate housing), the weaker impact of heat exposure can be related to the lower frequency and strength of extreme heat.

In addition, evidence on the effects of heat and cold on morbidity has been mixed depending on the socio-demographic characteristics of the populations studied. There are three main mechanisms to explain these differences: differential exposure, differential vulnerability (sensitivity and adaptive capacity) and differential access to good quality medical care and social support.

Beyond the previously mentioned relationship between temperature-related morbidity and age (Åström et al., 2011), heterogeneity by gender and socio-economic status (SES) has been widely studied. As gender differences in temperature-related morbidity are extremely outcome- and context-specific, they are very difficult to assess (see, for example, Cicci et al., 2022; Hajat and Kosatky, 2010; Ye et al., 2012). Some studies have highlighted physiological differences that are linked to differentials in sensitivity to temperatures. For example, compared to men, women may have a lower tolerance for heat because of their higher core temperature and the cyclical changes in oestrogen that affect their thermoregulation (Cicci et al., 2022). However, a meta-analysis on the topic showed that at all ages, men are at greater risk of heat illness than women. This may be due to sex-related behavioural differences in exposure. For example, compared to women, men may take fewer protective measures or engage in more risk-taking behaviours, and they might be more exposed to heat stress due to their work and outdoor leisure activities (Gifford et al., 2019).

In terms of socio-economic status, a review on heat-related health effects found that low SES and lower educated individuals are more vulnerable to heat (Gronlund, 2014). Single-country studies have also reported SES disparities in cold-related illnesses, e.g. for South Korea (Min et al., 2021). In terms of sensitivity to extreme temperatures, temperature vulnerability can be affected by pre-existing medical conditions and CVDs. The existence of an educational gradient in health in later life, with low educated individuals being disadvantaged, is well-established (Corna, 2013). In terms of exposure, low SES individuals tend to live in the hottest neighbourhoods of cities, whereas highly educated individuals tend to be more resourceful and able to afford better housing and thermally controlled indoor environments (Gronlund, 2014). In addition, individuals with lower SES or education are more likely to be employed in strenuous and outdoor occupations (e.g. miners, construction workers, farmers) (Gronlund, 2014). Lastly, access to good quality medical care and social support can protect individuals from the consequences of extreme temperatures (Leyva et al., 2017; Masiero et al., 2022). In Europe, social care programmes tend to be targeted to low SES individuals. However, low SES individuals may be less likely to have private health insurance, while high SES individuals might be able to access higher quality medical care.

A large number of studies have also explored the modifying effects of urbanicity, mainly on mortality. While there are only a handful of studies on these effects on morbidity, there is no reason to believe the mechanisms regarding mortality to be different from those at play for morbidity. Urban areas are at risk of becoming urban heat islands, as higher temperatures are likely to be found inside the city. Night-time temperatures are higher in cities than in the rural surroundings due to the retention of heat by concrete and asphalt and air pollutants, which can also intensify heat perception (Antal and Bhutani, 2022; Harper, 2019). The effects on morbidity and mortality of heat islands – and, in general, of living in an urban rather than in a rural context – are hard to assess, as they can vary depending on the housing conditions and socio-economic status of the population (Uejio et al., 2011), as well as on the city’s ability to cope with high temperatures (e.g. it may be higher in Southern than in Northern Europe) (Kovats and Hajat, 2008). Nevertheless, compared to their urban counterparts, rural residents tend to have less access to healthcare services and facilities, less access to strong social networks for social support and more difficulties implementing strategies for temperature relief (especially in the case of extreme heat) due to a lack of transport options. These factors may lead to an increase in temperature-related health conditions among rural residents (Williams et al., 2013). In light of these competing dynamics, the existing empirical evidence on the urban-rural divide in mortality and morbidity is mixed. Some studies have reported a higher risk of heat-related mortality in urban areas, e.g. in densely populated areas of Berlin (Gabriel and Endlicher, 2011). By contrast, other studies have found no protective effect of vegetation or imperviousness in, for example, Philadelphia (Uejio et al., 2011) and Worcester (Madrigano et al., 2013); or differences between rural villages and the provincial capital (see, e.g. for Spain Martínez-Navarro, 2004; for the Czech Republic Urban et al., 2014) (for exhaustive reviews, see Kovats and Hajat, 2008; Gronlund, 2014). Moreover, no differences between rural and urban areas were found in a study of the effects of heat on morbidity in New York (Adeyeye et al., 2019) and Vienna (Wanka et al., 2014). Conversely, a study conducted in the Zhejiang province of China found that rural areas are the most vulnerable to heat and cold exposure (Hu et al., 2019). Interestingly, in South Korea, a U-shaped curve between urbanisation and temperature-related morbidity has been observed, with the highest risks being found in rural areas and in the most densely populated parts of the city, largely due to the lower availability of hospital beds (Lee et al., 2022).

Social networks could provide important help to individuals aged 65 and older during cold spells and heatwaves. There is long-standing evidence of a protective effect of marriage on health (by encouraging a healthier lifestyle and health monitoring, and by providing socio-emotional and material support) (Rendall et al., 2011). For example, widowed men and divorced women in Turin, Italy were found to have a higher risk of heat-related mortality (Ellena et al., 2020). Similar findings showing a higher risk of temperature-related mortality for unmarried individuals have been reported for the Czech Republic (Vésier and Urban, 2023), Scotland (Wan et al., 2022) and Spain (Conte Keivabu, 2022). In addition, social programmes providing support to individuals aged 75 and older have been shown to be effective in decreasing heatwave-related risks (Liotta et al., 2018). Notably, the strength of social networks during heatwaves has been found to be a major factor stratifying mortality risks in the Chicago Heatwave of 1995 (Klinenberg, 2002; Klinenberg et al., 2020). To the best of our knowledge, no previous study has explored the potential protective factor of older individuals having adult children as part of their support network during episodes of extreme temperatures.

In conclusion, the present study will explore the effects of experiencing extreme temperatures on morbidity, while also taking into consideration subsequent mortality, among people aged 65 and older across a range of location-specific climates and socio-demographic individual characteristics, including gender, educational level, area of residence (urban/rural) and partnership and parenthood status.

In this article, we investigated how exposure to extreme temperatures affects morbidity, i.e. the risk of being hospitalised and of using medications for heart problems, in individuals aged 65 and older living in several European regions; and whether this relationship is moderated by the local climate and individual level socio-demographic characteristics. We made use of SHARE data (2004–2015), the only publicly available survey data in Europe that allow us to take into consideration the health outcomes of the older population across a variety of climatic contexts, along with their socio-demographic characteristics.

Our results show that exposure to extreme temperatures below the 1^st percentile of the regional temperature distribution slightly increases the risk of being hospitalised and of using heart medications. Conversely, we do not observe any substantive effect of heat exposure on health outcomes. These results align with those of some studies on cause-specific hospitalisation and the health effects of heat exposure (Åström et al., 2011; Cicci et al., 2022; Phung et al., 2016; Turner et al., 2012), but contradict those of several others (Bhaskaran et al., 2009; Bunker et al., 2016; Ye et al., 2012), which highlights the importance of the outcome that is considered as well as the measure of exposure that is employed.

When looking at climate-specific heterogeneity, we find that extremely cold days affect the risk of hospitalisation in both the coldest and the warmest European regions, while extremely hot days affect the risk of hospitalisation (but just slightly) in the warmest regions only (as was found by Michelozzi et al., 2009). The effects of extreme cold in both the coldest and the warmest regions may be attributable to two different mechanisms: on the one hand, despite their higher preparedness, one additional cold day strikes more populations living in very cold environments because of the strength and the frequency of these events. This also holds for the effects of one additional hot day in the warmest regions. On the other hand, one additional cold day in the warmest regions could affect the risk of hospitalisation because of the lower adaptive capacity of populations who are not used to and are not well equipped for below-zero temperatures.

The effects for taking medications for heart problems follow the same pattern as for cold days, with the effects being much reduced in size and not statistically significant. No noteworthy relationship is found between exposure to heat and the risk of taking heart medications.

Since we employed a prospective longitudinal survey, our results could be affected by mortality bias. The relationship between temperature and mortality is well-established in the literature (Gasparrini et al., 2015; Gasparrini and Armstrong, 2011; Tobías et al., 2021). Individuals can report on their health problems and hospitalisations only if they are still alive. By contrast, if individuals were hospitalised or developed a heart condition following exposure to extreme temperatures and subsequently died, they would not be present in the survey. This could greatly reduce our effect size. SHARE includes end-of-life interviews that we employed in a complementary analysis, despite the difficulties of assessing the region of residence and the timing of the onset of cardiovascular diseases for deceased individuals. These limitations notwithstanding, our results including deceased respondents show a much larger effect of exposure to extreme cold on the risk of both being hospitalised and experiencing CVDs (even though the effects remain limited in size). There also appears to be an effect on health outcomes of moderate-to-extreme heat, namely, of temperatures between the 75^th and the 95^th percentile. The lack of evidence for exposure to extremely hot temperatures (above the 95^th percentile) could be due to the diversity of temperatures across the NUTS-2 regions (e.g. the 99^th percentile ranges from 17 °C in Tyrol, Austria to 30 °C in Extremadura, Spain). Indeed, the interaction between temperature exposure and location-specific climate (i.e. as measured by the median temperature) indicates that any detrimental effects of exposure to extremely hot temperatures on morbidity and mortality are limited to the warmest regions.

The effects of exposure to heat on the risk of both being hospitalised and suffering from CVDs appear to be stronger in Europe’s warmest regions (with median temperatures above 17 degrees). This could suggest that the greatest mortality bias lies in heat-related morbidity. Interestingly, albeit to a different extent, the effects of exposure to cold on the risk of both being hospitalised and experiencing CVDs are observed across European regions.

In terms of heterogeneous impacts, we observe age differences in the risks related to cold (but not heat) exposure, with older individuals being at higher risk of being hospitalised and of suffering (and subsequently dying) from CVDs. A similar pattern is found for men compared to women in relation to the risk of hospitalisation. Moreover, for these two outcomes, we find that lower educated individuals are at greater risk than their highly educated counterparts following exposure to extreme cold. No sizeable differences emerge regarding the other socio-demographic characteristics considered. Indeed, if any, there seems to be a relationship between social vulnerability and vulnerability to temperatures, with individuals who have no partner and no children and who live in a rural area being at slightly higher risk of being hospitalised following exposure to extreme cold. Again, these differences are very limited in size.

This study has several limitations. First, the respondents were surveyed at different months, and not at regular intervals. Our decision to measure temperature exposure in the 12 months before the interviews ensures that all the respondents have experienced the same seasons. Moreover, this time span matches that for the morbidity outcomes (hospitalisation and heart medications intake in the previous 12 months; and CVDs experienced since the last wave). The drawback of this approach is that it cannot detect the short-term effects of temperature exposure. Second, as the hospitalisation variable includes all causes, we cannot distinguish between hospitalisations for respiratory diseases and for CVDs. Including all-cause hospitalisation could have offset the evidence on the effects of temperature exposure. Still, the results employing the measure of the onset of CVDs are in line with the results on hospitalisation, even though they are measured over a two-year time span, making the timing issue described above more problematic. Third, in terms of our research design, the NUTS-2 region of residence is reported only at survey entry. Thus, to adopt a longitudinal approach, we are forced to retain only individuals who did not change residence between two waves. Changing residence could be a coping strategy to escape extreme temperatures; following this reasoning, individuals who stay may be worse-off, with fewer resources to deal with extreme temperatures, than those who left, making our estimates upwardly biased. To check for this potential bias, we have compared the group of movers (deleted from the analytical sample) with the group of non-movers (included in the analytical sample) in terms of their socio-economic background (education, income) and family network (having a partner, number of children). Small differences by educational level emerge, with the group of movers being slightly more likely to be tertiary educated than the group of non-movers. This gives us confidence that if this bias exists, it is small in size.

Despite its limitations, SHARE remains the only survey that allows us to compare older individuals’ health outcomes at the European level with an adequate sample size across several climates, and in a longitudinal perspective. At the same time, SHARE allows to explore the heterogeneity of these outcomes based on socio-demographic and economic factors. Recently, the SHARE survey has been expanded with the creation of SHARE-ENV dataset (Midões et al., 2024), which was not yet available at the time of the present analysis. SHARE-ENV complements the information on SHARE respondents’ life conditions, health histories, healthcare use and working lives with indicators of respondents’ cumulative exposures to different environmental hazards. This harmonised source of data will be crucial for gaining new knowledge about the health hazards of climate change for the older population.

In conclusion, several best practices have been identified, and have often been implemented, to mitigate the effects of climate change on older people’s health (Schifano et al., 2012). For example, healthcare providers have been developing the competencies of GPs and nurses to identify health problems related to climatic events, adjusting and monitoring the care plans of older people (changing diet, adjusting medications) and implementing educational interventions aimed at encouraging preventive behaviours (Montoro-Ramírez et al., 2022). Nevertheless, the success of these strategies may be complicated by individuals with social disadvantage facing difficulties in accessing these interventions, even though people with fewer socio-economic and cultural resources are at greatest risk of experiencing social isolation (no children, no partner) and of suffering from temperature-related illness.

Therefore, the ageing-climate change nexus must be addressed in more structural terms, as it will be among the defining relationships of the coming decades. For example, the prevention of temperature-related illness should be included in policies, such as those aimed at promoting active, healthy and successful ageing (Sowa et al., 2016; Urtamo et al., 2019; World Health Organization, 2002). While there are several different definitions of healthy ageing, they all point to the need to sustain older people’s health and well-being via an active and healthy lifestyle, and through engagement with others and society at large. These frameworks can find a place in the movement for age-friendly cities, which seeks to make urban environments supportive and inclusive for older people (e.g. services are easily accessible and reachable), especially in light of the growing share of older people living in urban areas (Antal and Bhutani, 2022).

Research has recently shown the importance of what has been called “social infrastructure” during disasters and episodes of extreme temperatures (Klinenberg, 2002, 2018; Klinenberg et al., 2020). Social infrastructure comprises “the physical places and organisations that shape the way people interact” (Klinenberg, 2018, p.12); e.g. accessible gathering places such as libraries, community gardens and parks, restaurants and bars, and beauty parlours and barbershops. These places affect the formation of social capital in everyday life. When public spaces offer people the opportunity to engage in casual but sustained and recurrent interactions, particularly during activities they enjoy, they develop bonds and social cohesion. Strong interpersonal relationships and networks foster contact, mutual support and collaboration. This social infrastructure can lead to people caring for and checking in on each other, which is particularly crucial when the weather is very hot or very cold, especially for vulnerable individuals such as the oldest old (ibidem).

In conclusion, the design of age-friendly cities with a healthy social infrastructure takes on particular importance in a world that is increasingly affected by extremely hot temperatures due to climate change. Conceiving of ageing and climate change as related challenges could greatly benefit European societies. For example, the creation of more green outdoor spaces can be seen as supporting older people’s health and social life, while also mitigating heat perception (van Hoof et al., 2021).

Supplementary file 1. Tables S.1–S.8, Figures S.1–S.3.