Excess mortality attributable to high temperatures during the summers of 2021 to 2024 in Spain: description of the MoMo real-time monitoring system

Abstract

Introduction

Spain stands out as one of the countries most affected by rising heat-related mortality in Europe. The MoMo system is a daily all-cause mortality monitoring system that provides estimates of excess mortality attributable to high temperatures in Spain. This study describes the recent evolution of these estimates during the summers of the period 2021–2024.

Method

We conducted an observational descriptive study using data from MoMo. Population data by province, sex, and age group were obtained from the Spanish National Statistics Institute. Provincial daily maximum and minimum temperature data were provided by the Spanish National Meteorological Agency, following the standard MoMo methodology. Estimates are based on mixed generalised additive models and were described by sex and age groups for the overall country. Geographic patterns were explored by calculating standardized mortality ratios.

Results

Over the four summers analysed, 11,684 excess deaths attributable to high temperature were estimated by the model, 59.2% occurred in women and 40.8% in men. Excess mortality progressively increased with age, particularly among individuals over 74 years old. The highest number of excess deaths was estimated in 2022. By regions, in 2021, 2022 and 2024 excess of mortality were mainly located in inland provinces, while coastal regions generally showed lower than average risks, except in certain provinces in the East and South.

Conclusions

MoMo has shown the impact of rising temperatures on excess mortality attributable to high temperatures. It is essential to implement public policies focused on climate change mitigation and adaptation, especially in the most vulnerable populations and in the at-risk areas.

Introduction

Exposure to extreme heat is one of the most significant environmental and occupational risks for morbidity and mortality worldwide, along with unsafe water and air pollution [1]. The number of people exposed to this risk is increasing as a result of climate change, which is causing a rise in average temperatures and increasingly intense, frequent and prolonged heatwaves [2], as well as a progressive increase in night time minimum temperatures [3, 4]. The Spanish National Meteorological Agency (AEMET) defines a heatwave as an episode lasting at least three consecutive days, during which a minimum of 10% of weather stations in the country register maximum temperatures exceeding the 95th percentile of their series of daily maximum temperatures for the months of July and August in the 1971–2000 period [5].

Heat vulnerability is determined by both physiological factors [6, 7], affecting especially people over 65 years old, infants under one year old, and pregnant women, and exposure factors, such as occupation and socioeconomic conditions [8, 9]. People with chronic diseases [10] or who are under certain pharmacological treatments (antidepressants, benzodiazepines, antipsychotics, anticholinergics), as well as people with disabilities or those who use drugs (alcohol, MDMA, cocaine, heroin) [11] are also at increased risk of heat-related adverse events.

Within Europe, Spain stands out as one of the countries most affected by rising heat-related mortality, with an increase of 39.9 deaths per 100,000 inhabitants—substantially higher than the European average and far exceeding the minimal changes observed in northern countries such as Iceland (1.0 deaths per 100,000 inhabitants) [12]. As in most other European countries, population with vulnerable conditions is increasing, and this trend is expected to continue in the coming decades due to population ageing and improvements in survival of populations with several pathologies. In Spain, 40% of men and 44% of women aged 65 and over report having multiple chronic conditions, which is higher than the EU averages of 32% and 40%, respectively [13].

The all-cause daily mortality monitoring system (MoMo, available at https://momo.isciii.es/panel_momo/) is managed by the National Centre of Epidemiology (CNE). MoMo offers a population-based model for timely monitoring of short-term mortality impacts, having detected excess deaths during heat and cold waves, influenza, COVID-19 [14, 15] and acute emergencies such as the Madrid train bombings and the DANA event in Valencia [16]. The daily, up-to-date data enable rapid assessment and support timely public health decision-making, illustrating the added value of real-time surveillance for intervention planning.

This study aims to describe the effects of high temperatures on excess mortality in Spain during summers of the period 2021–2024 using the MoMo real-time monitoring system. It presents the evolution of excess mortality attributable to high temperatures by age and sex, and geographical location.

Methods

We conducted an observational descriptive study using data from MoMo. Population data by province, sex, and age group were obtained from the Spanish National Statistics Institute (INE). Provincial daily maximum and minimum temperature data were provided by the AEMET, following the standard MoMo methodology [17]. To estimate the number of excess deaths attributable to high temperatures, MoMo uses mixed generalised additive models (GAMs) [18] at the provincial level based on Poisson regression. The dependent variable of the model is the number of excess deaths in a given day, sex and age group, and the independent terms are trend (rigid cubic spline, k = 1), annual seasonality (cyclic spline, k = 6), and excess temperatures measured by accumulated thermal overcharge (ATO) and population as offset. ATO [19] is defined as the number of degrees by which the observed temperature exceeds heat-impact trigger thresholds previously estimated for each province [20, 21], modulated by the temperature in the previous seven days to take into account the possible impact of sustained exposure to high temperatures. The model excludes outliers (observations above the 99% confidence interval) and uses a fixed period of time of 10 years in its fit (excluding 2020 due to the abnormal increase in mortality caused by COVID-19, and the current year). Further details about the model can be found at (https://momo.isciii.es/panel_momo/). Excess deaths attributable to high temperatures are calculated as the difference between number of deaths estimated by this model, and number of deaths estimated by the same model but without the temperature term. In this study, the estimated number of excess deaths attributable to high temperature are presented by sex, age group (all ages, 0–14, 15–44, 45–64, 65–74, 75–84, and > 85 years), summer month (June, July, August and September), and year.

Observed daily mortality and estimated expected number of daily deaths with and without consideration of excess of temperature were represented in graphs for each year and sex during summer months. Periods corresponding to heatwaves were also represented. Additionally, this excess mortality was estimated by associating it with the mean maximum and mean minimum daily temperatures during the summers of the period 2015–2024 in order to allow for a comparison with the most recent decade.

To explore geographical patterns in excess mortality attributable to high temperatures throughout the study period, heat-attributable Standardized Mortality Ratios (SMR) were calculated for each province and year and represented in maps. For this purpose, observed deaths were divided by the expected deaths, which were calculated by multiplying the provincial population in the corresponding stratum by the observed national mortality rate, using the Spanish population of 2022 (the midpoint of the study period) as the standard population. The expected deaths were then summed across all strata. The resulting SMRs can be interpreted as the relative risk of excess heat-attributable mortality in each province compared to the reference population — the Spanish population in 2022.

The analyses were performed using R software, version 4.3.2.

Results

During the summers of the period from 2021 to 2024, a total of 11,684 excess deaths attributable to high temperatures were estimated: 6,911 (59.2%) in women and 4,773 (40.8%) in men (Table 1). Excess deaths progressively increased with age: the most affected age groups were those over 74 years old, with 91.4% and 76.7% of the excess deaths concentrated in women and men, respectively (Fig. 1). The year with the highest number of excess deaths was 2022 (4,732), more than twice the figures estimated for 2021.

Table 1.

Excess mortality attributable to high temperature by sex, summer month and year from 2021 to 2024 in Spain

	Women				Men				Total
	2021	2022	2023	2024	2021	2022	2023	2024	2021	2022	2023	2024
June	22	503	94	19	17	325	48	13	40	828	142	32
July	345	1363	428	459	215	855	258	312	560	2217	686	771
August	492	966	1229	810	286	642	761	577	778	1607	1990	1387
September	28	39	128	11	20	40	58	7	48	79	186	18
Total	888	2871	1879	1298	538	1861	1124	909	1426	4732	3003	2207

Totals may not add up exactly in some columns due to rounding

Fig. 1.

Excess mortality attributable to high temperature by sex and age group in Spain from 2021 to 2024

There was a positive linear association of both mean maximum (R² = 0.63) and mean minimum daily temperature (R² = 0.48) with excess mortality attributable to high temperature during the summers of the period 2015–2024 (Fig. 2). Summer 2022 was the warmest summer and registered more excess deaths.

Fig. 2.

Excess mortality attributable to high temperature associated with mean maximum temperature (top) and mean minimum temperature (bottom) during the summers of the period 2015–2024

Figure 3 shows the evolution of daily observed and estimated all-caused and excess heat-attributable mortality by sex during the studied summer months, showing the increased excess mortality attributable to high temperatures observed during heatwaves in Spain.

Fig. 3.

Observed deaths (grey), excess mortality attributable to high temperatures (red), estimated deaths without the effect of high temperatures with 99% CI (blue), and heatwaves (orange) by sex in Spain from 2021 to 2024

A dynamic pattern of excess mortality attributable to high temperature emerged across Spain, showing both consistent and evolving trends between coastal and inland regions (see Figs. 4 and 5). In the summers of 2021, 2022 and 2024, there seemed to be a coastal-inland divide, with the highest SMRs consistently recorded in inland provinces. Coastal provinces, on the other hand, generally exhibited lower risks throughout this period, with areas such as the Balearic and Canary Islands, and the northern coast maintaining lower SMRs. However, exceptions to this coastal-inland pattern were observed, particularly in the eastern coast where some provinces consistently showed elevated risks. In 2021, Alicante and Girona had SMRs of 1.81 and 1.80, respectively, and this pattern persisted in 2022, with those SMRs along with Castellón. The southern coastal regions generally maintained lower risks. By 2023, there was a shift in the geographical pattern. The previously distinct coastal-inland divide began to blur, particularly with an increase in SMR along the northern coast and a more moderate and uniform distribution of risk in the interior regions, with fewer extreme SMR values compared to earlier years. The eastern coast still showed elevated risks in provinces such as Girona, Alicante and Castellón. The highest SMRs were also observed in inland provinces in the summer of 2024, with some areas recording elevated SMRs compared to previous years. Palencia and Toledo emerged as the provinces with the highest SMRs, recording values of 5.42 and 3.08, respectively. Huesca, which had consistently been among the provinces with elevated risks, continued to show a high SMR of 2.61, along with Pontevedra, which had an SMR of 2.67. Despite a general reduction in risk in most coastal areas, the Levante region once again demonstrated increased mortality, with Girona, Alicante, and Castellón remaining consistent in their elevated risks over the years.

Fig. 4.

Age and sex heat-attributable standardized mortality ratio (SMR) by province from 2021 to 2024 in Spain

Fig. 5.

Map of Spanish provinces

Discussion

Over the four summers analysed, the highest excess mortality attributable to high temperature was estimated in 2022 with 4732 deaths followed by 2023, 2024 and 2021 with 3003, 2207 and 1426 deaths, respectively. 2021 was the year with the fewest recorded deaths, which could also be explained by the comparatively lower summer temperatures observed, as shown in Fig. 2. The results showed that the heatwaves in July and August had a relevant impact on mortality in Spain. Previous studies have confirmed this excess mortality during heatwaves, particularly due to an increase in deaths caused by cardiorespiratory diseases [22]. In this regard, it is important to highlight the direct relationship between extreme temperatures and the increase in mortality, and how these weather conditions, observed often during the summer months, exacerbate pre-existing health conditions. The summer of 2022 was extremely hot, being classified by AEMET as the hottest summer in Spain since records began in 1961 [23], and therefore, the hottest summer of the 21st century so far. In July and August 2022, unprecedented temperature peaks were reached and several intense and prolonged heatwaves occurred, coinciding with the highest excess mortality during the studied period. This rise in temperatures is directly linked to an increase in mortality, especially during heatwaves, and could be exacerbated by the population’s vulnerability. It is crucial to consider these results in the context of long-term climate change trends, which foresee an increase in the frequency and intensity of heatwaves [2]. There were estimated to be around 47,700 heat related deaths in Europe in 2023, and around 61,700 in 2022 (an estimate for 2024 is not yet available) [24].

This study estimates a higher number of excess heat-attributable deaths among people over 74 years. A greater susceptibility in older people has been consistently observed in previous studies [25] and this could be explained by physiological factors, such as a reduced capacity for thermoregulation in older adults, in addition to the higher prevalence of chronic diseases and the use of medications that may disturb the body’s response to heat [7]. On the contrary, diverse patterns have been found regarding the modifying effect of sex. In a study carried out in Rome, women were found to be more susceptible than men [26], while in Stockholm, men were found to be more vulnerable than women [10], and in Catalonia, there were no sex difference after age was taken into account [7]. This study shows that women had a higher number of excess heat-attributable deaths in Spain. More studies are needed to investigate the modifying effect, not only of sex, but also of the underlying causes.

Previous studies have reported substantially higher estimates of excess heat-attributable mortality in Spain. This contrast is largely explained by methodological differences in the definition of heat exposure. Many epidemiological analyses quantify the heat-mortality association starting at the minimum mortality temperature (MMT), moderately warm on most locations (e.g. MMT around 13 °C daily mean temperature in Madrid, 45th centile [27]), not relevant from an operational public-health perspective. In contrast, the MoMo system focuses specifically on mortality associated with high temperatures and actionable measures, using thresholds measuring statistically significant mortality increases (e.g., around 36 °C daily maximum temperature in Madrid, 86th centile [21]). When the comparison is restricted to mortality attributable to genuinely high temperatures, for instance taking the 95th percentile of MMT as a basepoint, the magnitude of the estimates becomes much more similar across studies.

A second source of divergence relates to the modelling framework. Most studies characterize temperature-mortality relationships using distributed lag non-linear models (DLNMs), a well-established epidemiological tool [28]. In MoMo, by contrast, estimates are generated using generalized additive models (GAMs), a modern and flexible statistical approach that allows for automated, near–real-time surveillance and inference [18]. The GAM-based formulation incorporates delayed and cumulative heat effects through the ATO variable, which measures exposure to anomalously high temperatures over the preceding seven days. This captures the same fundamental features that DLNMs are designed to model (nonlinearity and lagged responses) while ensuring robustness and contributing to a broader body of evidence.

Climate change is now regarded as the greatest global threat to human health by the World Health Organization, with its effects varying significantly based on the type of territory, as well as factors like age, sex, and socioeconomic status. Geographical differences, such as living in coastal or inland areas, play a crucial role in determining exposure to extreme climate events and the vulnerability of affected populations. In this study, we found geographical differences in mortality attributable to high temperature, revealing distinct risks in coastal and inland provinces. The persistent inland concentration of elevated SMRs highlights the vulnerability of these regions to mortality attributable to high temperature. This phenomenon could be attributed to both climatic such as urban heat island effect [29, 30] and socioeconomic factors. Coastal regions often benefit from the moderating effect of the sea [31], which helps reduce extreme temperatures and lowers the risk of prolonged heat exposure. In contrast, inland areas tend to experience more extreme temperatures, increasing the risk of mortality.

The lower SMR observed in the southern provinces of Spain are explained by the province-specific heat thresholds used in the MoMo system. These thresholds, defined by the Spanish Ministry of Health, are set according to each province’s historical temperature distribution. As a result, provinces with a long-standing exposure to higher temperatures—such as those in southern Spain—have higher baseline thresholds. This reflects a greater population acclimatization to heat, which reduces the relative excess mortality detected above those local thresholds. This rationale also explains why non-coastal southern provinces show lower SMR values: although they experience high absolute temperatures, these values often remain closer to their historically established thresholds. Indeed, the spatial patterns of SMR align more closely with the provincial summer temperature anomalies than with maps of absolute temperature levels [32], which is consistent with a threshold-based approach such as that used by MoMo.

One of the main limitations of this study is the aggregation of data at the provincial level, as it may hide important local specificities, where different realities exist in terms of climatic and socio-economic conditions. This phenomenon does not occur uniformly across the entire city. Neighborhoods with lower-income residents tend to face higher levels of vulnerability [33] because of the absence of green spaces, amount of paved surface area, low degree of street ventilation, traffic levels, urban pollution, poor thermal insulation and low-quality building materials. Additionally, the model used to estimate excess mortality attributable to high temperatures depends on temperature thresholds set for each province, which may not accurately reflect the conditions of thermal stress in different population subgroups or in rural in comparison with urban areas. Another limitation of this study is the absence of uncertainty measures for the estimated excess mortality, as the MoMo surveillance system provides point estimates by design and does not allow for the reliable derivation of confidence intervals.

From a public health perspective, this study illustrates the value of real-time mortality monitoring systems for responding to the health impacts of extreme heat. Despite these limitations, a key strength of the analysis is the use of daily up-to-date death monitoring, which allows for an immediate assessment of the impact of heatwaves on mortality. This surveillance approach is crucial for planning rapid interventions, supporting situational awareness during extreme events and informing the development of public policies designed to mitigate the adverse effects of extreme heat on the population.

MoMo combines near–real-time all-cause mortality data with routinely available meteorological information, enabling timely detection of excess mortality associated with heat events at a fine spatial scale. Compared with traditional mortality statistics, which are often available only after substantial delays, this timeliness represents a major advantage for early warning systems, evaluation of action plans and risk communication. The Spanish experience provides a practical and scalable example for other countries seeking to strengthen climate–health surveillance, as MoMo is based on existing civil registration and meteorological infrastructures and does not require cause-of-death coding.

The findings also underscore the need for sustained and integrated public health policies. These policies should aim to create habitable places that are resilient, sustainable, and adaptable to rising temperatures, where urban design, infrastructure, and green spaces help reduce heat exposure and vulnerability. Sustained efforts are also required to improve access to healthcare and energy resources, especially for the most vulnerable populations. As climate change continues to increase the frequency and intensity of heatwaves, investing in timely, integrated mortality monitoring systems such as MoMo should be considered a core component of national and international climate change adaptation strategies.

Conclusions

High temperatures increase overall mortality in the population of Spain, where most excess deaths are expected to occur among those over 74 years old and particularly in women. In the summers of 2021, 2022 and 2024, there seemed to be a coastal-inland divide, with the highest SMRs consistently recorded in inland provinces. The lower SMR were observed in the southern provinces of Spain. It is essential to implement public policies focused on climate change mitigation and adaptation, with particular attention to differences in the vulnerability among population groups. MoMo is a useful surveillance system both for detecting and assessing the impact in mortality during extreme weather events such as heatwaves. Its ability to estimate excess mortality and support public health decision-making in potentially threatening situations has been shown.

Authors’ contributions

Raquel Barba Sánchez (RBS): Participated in shaping the study design based on an initial idea by DGB and ILG. Acquired and prepared the data for analysis. Performed the statistical analysis and generated the tables and figures. Drafted the first version of the manuscript and coordinated all revisions. Acted as the corresponding author throughout the submission and review process. Diana Gómez Barroso (DGB): Co-developed the initial idea and study design; guided the application of the MoMo model; provided support for data analysis; critically revised all manuscript drafts, contributing to interpretation of results and improving the clarity and relevance of the study. Inmaculada León Gómez (ILG): Co-developed the initial idea and study design; provided methodological guidance on the use of the MoMo model and statistical tools; reviewed and commented on all manuscript drafts with special focus on structure and interpretation. Lucía Pérez Marín (LPM): Provided a critical reading of the manuscript during advanced drafting stages; helped refine the language and argumentation; contributed additional literature and suggestions to improve overall clarity. David García García (DGG): Reviewed the methodology in detail and contributed to clarifying its presentation; suggested improvements to the manuscript’s coherence and supported interpretation of meteorological and geographical results; contributed relevant bibliographic references. All authors approved the final version of the manuscript and agree to be accountable for all aspects of the work, including the accuracy and integrity of any part of the research.

Funding

Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature. This research did not receive any funding.

Data availability

The analyses were conducted using anonymised, aggregated mortality and temperature data obtained from publicly available sources at https://momo.isciii.es/panel_momo/#section-datos), managed by the National Centre of Epidemiology (CNE) of the Carlos III Health Institute.

Declarations

Ethics approval and consent to participate

This study did not require ethical approval as it did not involve human participants, personal data, or biological specimens. The analyses were conducted using anonymised, aggregated mortality and temperature data obtained from publicly available sources at https://momo.isciii.es/panel_momo/#section-datos), managed by the National Centre of Epidemiology (CNE) of the Carlos III Health Institute.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.