Compound mortality impacts from extreme temperatures and the COVID-19 pandemic

The study addresses the previously unquantified compound health impacts of the co-occurrence of extreme temperatures and the COVID-19 pandemic. Motivated by mounting evidence that extreme weather events and pandemics can overwhelm health systems when occurring in parallel, the authors investigate England and Wales, where detailed epidemiological models and regionally resolved mortality data are available. The research quantifies temperature-related (heat and cold) mortality alongside COVID-19 mortality between 30 January 2020 and 31 December 2022, aiming to reveal how these concurrent crises contribute to excess mortality and to inform climate adaptation, pandemic preparedness, and public health planning.

The paper notes that while the links between climate and conflict are relatively well studied, intersections between climate change and other simultaneous crises—such as obesity, artificial intelligence, and pandemics—are understudied. During the COVID-19 public health emergency (January 2020–March 2023), multiple unprecedented climate-related extremes occurred globally (e.g., UK’s 40.3 °C heat, extreme North American heatwaves, Pakistan floods, Horn of Africa droughts). Prior UK studies show that moderate cold contributes substantially to mortality across most days of the year, and literature has reported mixed findings on weather (temperature/humidity) effects on COVID-19 transmission. The authors reference previous warnings about compounded risks from climate hazards and pandemics and highlight media and public health messaging disparities that may downplay heat and cold risks compared with COVID-19.

Study area and period: England and Wales from 30 January 2020 (date of first COVID-19 death certificate) to 31 December 2022. Outcomes: daily temperature-attributable mortality (heat- and cold-related) and daily COVID-19 deaths (per death certificates). Data: Daily all-cause deaths by region from the Office for National Statistics (ONS); COVID-19 deaths identified from death certificates and removed from all-cause mortality to isolate temperature-related mortality; temperatures from HadUK-Grid gridded observations aggregated to regions; regional populations from ONS (mid-2021) for per-100,000 scaling. Modeling: For each region, the authors estimated non-linear associations between daily average temperature and daily mortality for 1981–2022 using quasi-Poisson time series regression with Distributed Lag Nonlinear Models (DLNMs). Temperature–mortality functions used natural cubic splines with internal knots at the 10th, 75th, and 90th percentiles of the regional temperature distribution. Lag-response was modeled with natural cubic splines across lags up to 21 days to capture delayed heat and cold effects; seasonality and long-term trends were controlled via smooth functions of time. Associations were centered at the Minimum Mortality Temperature (MMT), defined per region. Estimation of daily temperature-attributable deaths during the study period used observed daily mean temperatures, estimated relative risks from DLNMs, and the average number of non-COVID all-cause deaths by calendar day (1981–2022). Attribution to heat vs cold was based on whether daily mean temperature exceeded or fell below the regional MMT. Uncertainty: 95% empirical confidence intervals obtained via Monte Carlo simulations (10,000 draws) of spline coefficients assuming multivariate normality. Extreme events: Heatwave and cold snap days during 2020–2022 were identified from UKHSA Heat Mortality Monitoring Reports and Level 3 Cold Health Alerts (10 heatwaves totaling 70 days; 8 cold snaps totaling 70 days). For comparison with 2010–2019, equivalent definitions were applied: heatwaves via UKHSA Level 3 alerts or Central England mean temperature >20 °C plus ±1 day; cold snaps via periods with regional average temperature <2 °C for ≥2 consecutive days. Results were expressed as deaths per 100,000 using regional mid-2021 populations (and 2015 populations for the 2010–2019 comparisons).

• Temperature-related mortality frequently exceeded COVID-19 mortality during substantial portions of the study period (June–October 2020; March–August 2021; September 2021–December 2022), driven by heat in summer and cold in later periods after vaccine rollout.
• In South West England, cumulative temperature-related mortality exceeded COVID-19 mortality by 8% over the 2020–2022 period.
• During the study’s 10 heatwaves (70 days), temperature-related deaths outnumbered COVID-19 deaths in 9 of 10 regions (ratios from 1.1 to 2.7; London 2.7; East of England and South East 1.7; North West 0.8).
• During the 8 cold snaps (70 days), COVID-19 deaths exceeded temperature-related deaths in all regions (ratios around 0.8 for temperature-related to COVID-19 deaths), influenced by large winter COVID-19 surges.
• Compound mortality (temperature-related + COVID-19) during heatwaves ranged from 19 (95% CI: 16–22) deaths per 100,000 in North West England to 24 (95% CI: 20–29) in Wales; during cold snaps from 80 (95% CI: 75–86) in Yorkshire and the Humber to 127 (95% CI: 123–132) in East of England. These are at least twice the levels observed during comparable events in 2010–2019.
• Peak daily heat-related mortality occurred on 19 July 2022 (the day 40.3 °C was recorded), and peak daily cold-related mortality reached 531 (95% CI: 493–574) on 15 December 2022.
• The highest daily COVID-19 mortality was 1382 deaths on 19 January 2021; total COVID-19 deaths recorded on death certificates during the study period were 194,480 in England and Wales.

The analysis demonstrates that extreme temperatures and COVID-19 can jointly create substantial compound mortality burdens capable of straining health systems. Even amidst a pandemic, heatwaves posed serious and sometimes greater mortality risks than COVID-19 in many regions, underscoring that temperature extremes should be treated as major public health threats comparable to pandemics. Cold-related mortality also constituted a significant burden, especially in winter periods coinciding with COVID-19 surges. The findings indicate that reducing temperature-related deaths—through targeted public health messaging and structural interventions—could free capacity within health services to better respond to pandemics. The temporal patterns observed (e.g., exceedance of temperature-related over COVID-19 deaths post-vaccination) highlight the dynamic interplay between non-pharmaceutical interventions, vaccination coverage, seasonal climate variability, and healthcare system capacity. These insights are directly relevant for integrated climate adaptation and pandemic preparedness planning.

This study quantifies, for the first time, the compound mortality impacts from the co-occurrence of extreme temperatures and the COVID-19 pandemic in England and Wales (2020–2022). It shows that temperature-related mortality frequently rivaled or exceeded COVID-19 mortality during specific periods, and that compound mortality during heatwaves and cold snaps was at least double that of comparable events in the previous decade. The results advocate for elevating extreme heat and cold to the same level of urgency as pandemics in public health planning and for implementing evidence-based interventions (e.g., urban greening, reflective/green roofs, external shutters, air conditioning where appropriate, improved home heating and ventilation) to enhance resilience. Future work should: (i) explicitly model temperature–mortality associations using COVID-era data to assess confounding; (ii) conduct finer-scale analyses to guide local adaptation investments; and (iii) better characterize meteorological effects on pathogen transmission using rigorous epidemiological designs that address lags, seasonality, and confounders.

• Disentangling temperature-related and COVID-19 mortality is challenging due to shared risk factors and potential confounding; attributing deaths to a single cause has inherent uncertainty.
• Only two to three years of pandemic-era data limits the ability to derive robust, COVID-era-specific temperature–mortality functions; longer time series are needed to refine estimates.
• Identification of historical (pre-2016) heatwaves and cold snaps required proxy criteria due to limited availability of formal alerts, which may introduce classification uncertainty.
• Behavioral changes, public health interventions, and healthcare-seeking patterns during the pandemic may have altered exposure–response relationships and are difficult to fully account for.
• Regional aggregation may mask local heterogeneity in exposure, vulnerability, and healthcare capacity.