Estimating heatwave-induced excess mortality in India's districts

Abstract

India is among the world’s most heat-exposed nations, with hundreds of millions of people facing dangerous temperatures each summer and a mortality burden that remains poorly understood at the district level. We estimate that a single day of extreme heat causes approximately 3,400 excess deaths nationally; a five-day heatwave causes nearly 30,000. Although global studies highlight surging heat-related mortality, granular spatial-temporal data on how heatwaves affect mortality at the district level in India remain inaccessible to common researchers. District-level estimates of heatwave-induced excess mortality covering all of India have not previously been reported in the peer-reviewed literature; this paper provides such estimates using publicly available data and a climate zone-based risk transfer methodology. We adapt findings from a multi-city epidemiological analysis of heat-related mortality across 10 Indian cities to estimate excess deaths across all Indian districts. We integrate district-level mortality rates from the Civil Registration System and population projections for 2024 with city-specific risk coefficients based on the Köppen–Geiger climate classification. We obtain district-level excess death estimates under one-day and five-day heatwave scenarios. Our results suggest that even a single day of extreme heat yields thousands of excess deaths, and a multi-day heatwave results in tens of thousands of excess deaths. By mapping heat-induced mortality risk to individual districts, this study finds that Uttar Pradesh alone accounts for approximately 8,100 excess deaths during a five-day heatwave, and districts such as Ahmedabad, Jaipur, and Surat each exceed 250 excess deaths in a single event. It underscores the need for more localized heat action plans, improved and heat-relevant healthcare infrastructure, and robust early-warning systems. These findings have implications not only for India but also for other countries in South Asia and Sub Saharan Africa facing similar heat vulnerabilities, highlighting the global urgency of heat adaptation measures. With respect to heatwave frequency and temperature thresholds, these estimates are conservative lower bounds. The direction of bias introduced by applying urban-derived risk coefficients to rural populations is uncertain and is discussed in the Limitations section.

Introduction

India's tropical latitude and predominantly low-lying terrain make it vulnerable to high ambient temperatures. Large parts of the country regularly experience prolonged periods of heat and humidity, conditions that place sustained stress on the human body. With increasing global warming, ambient temperatures in many regions are approaching levels at which human thermoregulation becomes increasingly inadequate. Prolonged exposure to high temperatures, particularly in the presence of high humidity, reduces the body's ability to dissipate heat through sweating. When core body temperature rises above its normal range of approximately 37 °C, the risk of heat exhaustion, heatstroke, organ damage, and death increases substantially. Only a small fraction of India's population (about 8%) has access to air conditioning. As a result, a growing share of India's population is exposed to physiologically unsafe heat conditions during the summer months.

The public health consequences of extreme heat are becoming more pronounced worldwide. Globally, heat-related mortality increased by 55% between 2000–2004 and 2017–2021 (1–3). In India, heatwaves are projected to adversely affect the quality of life of up to 600 million people by 2,100 under high warming scenarios (4). Despite this growing risk, the mortality burden associated with heatwaves remains poorly quantified. Official statistics substantially underreport heat-related deaths (5). For example, an online data compilation by Heatwatch and the Veditum India Foundation identified 733 heatstroke-related deaths reported in news media between March and July 2023, and official figures released by the Ministry of Health and Family Welfare reported only 360 heatstroke deaths during the same period. Peer-reviewed epidemiological studies consistently show that the mortality burden attributable to extreme heat is substantially underestimated when surveillance relies on heatstroke or disaster statistics alone, because heat impacts are captured through excess all-cause mortality rather than cause-specific coding (6–8).

The scale of this undercounting has also been highlighted outside the academic peer-reviewed literature. Media reporting has corroborated the pattern of large discrepancies between officially reported heatwave deaths and observed excess mortality—a finding consistent with the peer-reviewed literature on heat-attribution failures (6, 7, 9). While journalistic accounts cannot substitute for epidemiological analysis, such reporting underscores the urgency of developing transparent, data-driven estimates of heatwave mortality using the best available evidence. In particular substantial underreporting of heat deaths will inevitably lead to substantial neglect and substantial underinvestment in addressing the heat threat. We report here our estimate of the size of the heat threat at district-level resolution, based on publicly available data for India.

Several recent studies have estimated the overall mortality burden associated with rising temperatures. Zhao et al. (8) estimate approximately 111,000 heatwave-related deaths annually across South Asia using a multi-country, grid-based modelling framework. Xu et al. (10) project a sharp increase in heat-related mortality under future warming scenarios by identifying a narrowing human climate niche, beyond which human populations are exposed to unprecedented heat conditions under which they cannot survive long term. Although these studies provide important regional and global perspectives, their spatial resolution is too coarse to inform local preparedness and adaptation efforts. In India, the absence of district-level estimates of heatwave-related mortality limits the ability of policymakers to identify high-risk areas, allocate resources effectively, and design targeted interventions. This gap is compounded by limited public access to high-resolution temperature time-series data and comprehensive, district-level mortality time-series data.

A recent multi-city study by de Bont et al. (11) provides the most robust to date epidemiological assessment of heatwave-related excess mortality in 10 cities in India. The authors studied daily mortality and temperature data from 10 cities across diverse climatic zones in India for a recent multi year interval (2008–2019). They quantified the increase in all-cause mortality associated with consecutive days of extreme heat and estimated the fraction of excess deaths attributable to heatwaves. Building on this work, we extend the analysis to generate district-level estimates of excess mortality associated with short- and multi-day heatwave events across India. By transferring city-specific heatwave–mortality risk estimates, stratified by climate zone, to all districts, this study provides the first India-wide estimates of heatwave-induced excess mortality, with district level resolution, using publicly available data.

These estimates represent the most comprehensive assessment currently possible in the absence of systematic, district-level mortality and exposure datasets. While additional information from government sources, including more recent demographic data, occupational exposure profiles, and cause-specific mortality records, would allow for more refined estimates, the present conservative estimates establish a lower-bound, evidence-based benchmark for heatwave mortality in India. Future research should build on this framework by incorporating updated population data, improved temperature exposure metrics, and differential vulnerability among groups such as outdoor workers (12, 16), older adults, and those with pre-existing health conditions.

In the existing scholarly literature, estimates of heat-related mortality in India are available at two scales: global multi-country studies that include India as a single national unit (6–8), and city-level analyses covering at most ten urban centers (11). Studies at either scale are inadequate to supports subnational planning and action. This work fills that gap by extending city-level risk coefficients to all districts using Köppen–Geiger climate classification, producing district-resolution excess mortality estimates for India derived from peer-reviewed epidemiological data.

Methods

We build our estimates on the de Bont et al. (11) study. We extrapolate in five steps their empirically estimated heatwave–mortality relationships for 10 Indian cities to all districts in India, using district-level demographic data and climate classification. We outline these five steps below.

Step-1. Obtain data for baseline mortality

District-level mortality rates were obtained from the Civil Registration System (CRS) of India for the year 2020, the most recent year for which district-level mortality data are publicly available (13). District-level population projections for 2024 were compiled using official population estimates and standard demographic projections. Using these inputs, we calculated the baseline daily all-cause mortality for each district as the product of the district's population and its annual death rate, divided by 365 days. Excess mortality and officially attributed heat deaths measure different quantities: the former captures all deaths above the expected baseline during heatwave periods regardless of stated cause, while the latter requires explicit heat attribution at the point of death registration. The two figures are therefore not directly comparable.

This approach assumes that district-level mortality rates remained stable between 2020 and 2024, and have no seasonal variation. While incremental changes in mortality patterns may occur due to improvements in healthcare access or demographic shifts, such changes are expected to be modest over this short time horizon. Under these assumptions, the baseline all-India mortality rate was estimated at approximately 23,000 deaths per day in 2024.

Step-2. Classify climate and assign mortality risk coefficients to each district

Each of the ten Indian cities studied in de Bont et al. (11) was assigned its relevant Köppen–Geiger climate classification, and the study estimated city-specific heatwave–mortality risk coefficients corresponding to those climate zones with 95% confidence intervals. To apply these estimates nationally, we assigned each Indian district to one of the study cities based on climatic similarity.

Specifically, each district was first classified according to its Köppen–Geiger climate zone. Districts were then matched to the study city within the same or closest climate classification. In cases where multiple cities shared a similar climate category, geographic proximity and similarity elevation were used to identify the most appropriate representative city. Through this assignment each district inherited the heatwave–mortality risk coefficients estimated for its matched city (see Figure 1) @ (11).

Figure 1

11). Each district is coloured according to the study city whose excess heatwave–mortality risk estimates were applied in the district-level extrapolation. Blue markers indicate the geographic locations of the ten study cities. District assignments are based primarily on Köppen–Geiger (2024) climate classification, with geographic proximity and elevation used to resolve overlaps across climate zones. The boundaries and names shown and the designations used on this map do not imply any opinion regarding the legal status of any country, territory, city or area, or concerning the delimitation of its frontiers or boundaries.

This procedure assumes that the populations of districts sharing similar climate characteristics experience similar physiological and epidemiological responses to extreme heat, in the absence of district-specific mortality–temperature time series data. It also makes the simplest possible assumption that the excess heat-mortality in urban populations of the 10 cities is also the best estimator of the excess heat-mortality in corresponding rural population [ignoring the effects of differences in income, occupations (16), access to health care, and demographics such as overall health, nutritional status, age, and gender (12)].

Step-3. Define heatwave events

Heatwave events were defined using a percentile-based approach consistent with de Bont et al. (11). Daily temperature thresholds were calculated separately for each district using the same historical baseline period, from 2008 to 2019.

Step-4. Estimate excess mortality for each district for a heatwave event

We estimated excess heat-mortality for each district and heatwave event by applying the relative increase in daily mortality associated with heatwave exposure, as reported by de Bont et al. (11), to the district's baseline daily mortality. Specifically, the expected number of excess deaths was calculated as the product of baseline daily mortality, the heatwave-associated percent increase in daily mortality, and the duration (in days) of the heatwave event. The one-day and five-day scenarios each use their respective duration-specific coefficient from de Bont et al. (11); these are distinct empirical estimates and are not derived by multiplying a single daily coefficient by the number of days. To characterise uncertainty in the national estimates, we apply the pooled 95% confidence intervals from de Bont et al. (11) proportionally to our city-specific point estimates. This is an approximation: it scales each city-specific coefficient by the ratio of the pooled CI bounds to the pooled point estimate, rather than drawing from city-specific distributions. City-specific 5-day confidence intervals from the source study would allow a more precise propagation and are recommended for future work.

One-day and five-day heatwave scenarios were evaluated separately, yielding district-level estimates of excess deaths attributable to short-duration and prolonged heat exposure. National and state-level totals were obtained by aggregating district-level estimates.

These estimates represent the excess mortality expected during heatwave periods relative to non-heatwave periods, under the assumption that the heatwave–mortality relationships observed in the study cities apply to climatically similar districts.

Step-5. Conduct sensitivity analysis for district-city assignments

To assess the robustness of the district-to-city assignment, we conducted sensitivity analyses incorporating additional geographic criteria. Districts whose average elevation differed substantially from that of their initially assigned city, or that were located at large geographic distances, were reassigned to alternative study cities with closer elevation profiles or geographic proximity. Excess mortality estimates were recalculated under these alternative assignments and compared with the primary results.

National and state-level findings were consistent across alternative city-district assignment rules, confirming that results do not depend on any particular matching decision. We therefore characterize the sensitivity to city allocations as modest, as it does not alter the key quantitative features of the results. Specifically, we tested the only climatically defensible alternative assignment in this framework: swapping Pune- and Hyderabad-assigned districts, the sole pair sharing similar elevation, climate classification, and heat stress profiles on the Deccan Plateau. This affects 53 of 765 districts (∼8%), confined to Maharashtra and Telangana. The national five-day excess death total changes by less than 0.1% (29,967 to 29,946), and the five highest-burden states are unchanged. We note that this ten-city framework is designed for national-scale estimation; sub-national figures are indicative decompositions of the national total and carry greater uncertainty at finer geographic resolution. A map of the alternative assignment (Supplementary Figure S1) and a state-level percentage-change table (Supplementary Table S1) are provided in the Supplementary Information.

Results

Our analysis indicates that extreme heat events are associated with substantial excess mortality across India, even when heatwaves are defined conservatively using historical temperature thresholds from the 2008–2019 baseline period. Because many regions in India experienced record-breaking temperatures in 2023 and 2024, the estimates presented here should be interpreted as lower-bound estimates of heatwave-related mortality under current and future climate conditions (owing to the high likelihood that the intensity and duration of heat waves will continue to increase for India in the coming years.) Heatwave events are defined using dry-bulb temperature thresholds consistent with de Bont et al. (11). In humid coastal districts, some high-heat-stress days will not breach the dry-bulb 97th percentile, and therefore genuine heatwave events may be missed in these areas.

National excess mortality

Under the one-day heatwave scenario, defined as a single day on which daily temperature exceeds the district-specific 97th percentile, we estimate approximately 3,400 excess deaths nationwide. This finding indicates that even a one-day exposure to extreme heat can produce a significant increase in mortality at the national scale, much larger than the all-India heat-related excess mortality reported in the press and by government agencies, which remains about 800 per year.

Under the five-day heatwave scenario, defined as five consecutive days exceeding the same threshold, excess mortality increases sharply. We estimate approximately 30,000 excess deaths nationwide during a single five-day heatwave. The nearly ninefold increase between the one-day and five-day scenarios reflects two compounding factors: the five-day risk coefficient is empirically larger per day than the one-day coefficient, and that larger per-day risk applies across five days of exposure rather than one. Applying the 95% confidence intervals reported by de Bont et al. (11) proportionally to our city-specific coefficients yields national ranges of approximately 2,400 to 4,400 excess deaths for the one-day scenario and approximately 18,000 to 43,000 for the five-day scenario. The order of magnitude of both central estimates is therefore robust across the full range of coefficient uncertainty.

Spatial concentration of mortality burden

Excess mortality associated with heatwaves is distributed highly unevenly across India. A small number of populous states account for a disproportionate share of the national burden. In particular, Uttar Pradesh alone accounts for approximately 8,056 excess deaths during a five-day heatwave, followed by Bihar (approximately 3,615 deaths), Madhya Pradesh (approximately 2,964 deaths), Rajasthan (approximately 2,664 deaths), and Gujarat (approximately 2,354 deaths). Together, these five states (comprising 43% of India's population) account for more than 60 percent of total national excess mortality under the five-day heatwave scenario.

This concentration reflects the combined effects of large population size, higher baseline mortality, and exposure to extreme heat conditions. The results suggest that heatwave-related mortality is driven not only by climatic intensity but also by underlying demographic vulnerability.

District-level and state-level mortality variation

The spatial distribution of district-level excess deaths during a five-day heatwave is shown in Figure 2.

Figure 2

At the district level, excess mortality varies substantially. Several districts experience mortality surges exceeding 180–300 deaths for a five-day heatwave. For a five-day heat wave, the highest estimated excess deaths are expected for Ahmedabad (approximately 307 deaths), Jaipur (approximately 265 deaths), and Surat (approximately 261 deaths). Other high-burden districts include Prayagraj, Patna, Lucknow, Kanpur Nagar, Azamgarh, Agra, and Bareilly, each with expected excess deaths exceeding 180.

These districts combine high baseline daily mortality with climate-zone-specific heatwave risk coefficients inherited from their representative study cities. The results demonstrate that individual districts may already experience excess mortality burdens during a single heatwave that rival or exceed the annual heat-related mortality reported for entire states in official statistics (the low values of the latter are likely an artifact of misallocation of cause of death).

We quantified inequality in heatwave mortality burden across India using two Lorenz curve analyses: a population-weighted curve at the district level (n = 765) and a GDP-weighted curve at the state level (n = 36), Figures 3, 4. At the district level, the population-weighted Gini coefficient of 0.215 indicates moderate inequality in the distribution of excess deaths relative to population. The top 100 districts—comprising 31% of India's population—account for 44% of projected 5-day heatwave excess deaths, with the highest-burden quintile (Q5) bearing 57% of deaths from 41% of the population. At the state level, we computed total state GDP as the product of state population and per-capita GSDP (2023–24 current prices), then constructed a Lorenz curve with cumulative GDP share on the x-axis. The resulting GDP-weighted Gini coefficient of 0.432 double the population-weighted value reveals that heatwave mortality is heavily concentrated in economically weaker states. The five highest death-burden states (Uttar Pradesh, Bihar, Madhya Pradesh, Rajasthan, and Gujarat) account for 66% of national excess deaths while contributing only 29% of India's GDP, representing a 2.3-fold disproportion between mortality burden and economic capacity. The highest quartile (Q4) bears 82% of deaths from 59% of GDP, whereas the lowest quartile (Q1) contributes just 0.4% of deaths from 1.1% of GDP. This divergence between population-weighted and GDP-weighted inequality demonstrates that heatwave mortality risk is not merely proportional to population size but is structurally concentrated in states with lower economic output precisely those with the least fiscal capacity to invest in adaptation.

Figure 3

Figure 4

Role of climate and altitude

Sensitivity analyses incorporating district elevation and geographic proximity into the city–district assignment process indicate that the overall spatial pattern of excess mortality is robust to alternative matching assumptions. Low-lying and humid districts consistently exhibit higher excess mortality across assignment scenarios. At the same time, districts located at higher elevations also experience elevated mortality when local temperatures exceed historically extreme thresholds, despite lower average temperatures.

These findings hint that relative deviations from local temperature range, rather than absolute temperature levels, might play an important role in shaping heatwave-related mortality risk. This is a weak hint, since it comes from the current results which are based on numerous approximations that need to be reassessed and improved. We feel it should be left for further study.

Interpretation of risk magnitude

The excess mortality estimates rely on the assumption that heatwave–mortality risk coefficients derived from urban study cities apply to climatically similar districts. Under this assumption, incremental increases in temperature above extreme thresholds translate into substantial increases in mortality. The contrast between the one-day and five-day heatwave scenarios highlights the disproportionate impact of prolonged heat exposure. The one-day and five-day scenarios use separate duration-specific risk coefficients from de Bont et al. (11), estimated from city-level mortality data for those respective durations. The five-day coefficient is empirically larger per day than the one-day coefficient, consistent with the nonlinear accumulation of physiological heat stress across days. The near-ninefold difference in national excess deaths between the one-day and five-day scenarios therefore reflects both the larger per-day risk associated with longer heatwave events and the greater number of days of exposure.

India is expected to experience heat waves of higher intensity and longer duration. As an illustrative scenario only, if each district were to experience five heatwaves of five-day duration each summer — a round-number planning assumption, not an empirically derived frequency — the implied national excess mortality would reach approximately 150,000 deaths per year. This figure is not an annual forecast; it is provided solely to convey the order-of-magnitude implications of repeated heat exposure at the national scale. The actual number of heatwave events per district per summer varies substantially by region and year; a more precise annualisation would require district-specific observed heatwave frequency data, which lies outside the scope of this study.

Consistency with existing literature

The district-level estimates presented herein align with an increasing body of epidemiological and modeling evidence concerning heat-related mortality in South Asia and India. Global and regional assessments have consistently identified South Asia as one of the world's most vulnerable regions to extreme heat. Utilizing a three-stage modeling framework, Zhao et al. (8) estimate approximately 111,000 annual heatwave-related fatalities across South Asia, with India contributing approximately 20.7% of global heatwave-related mortality between 1990 and 2019, representing the largest national share. Similarly, Vicedo-Cabrera et al. (14) and Gasparrini et al. (6) document a pronounced increase in heat-attributable mortality within low- and middle-income regions, with South Asia experiencing a particularly elevated burden. Furthermore, long-term projections by Xu et al. (10) suggest that, under high-emissions scenarios, heat-related mortality in India could escalate several-fold in the coming decades as substantial populations are exposed to thermal conditions exceeding historically safe limits.

Limitations

Five simplifying assumptions underlie these estimates. First, risk coefficients from ten urban centers are applied to all districts including rural ones, where more participation in outdoor labor, limited healthcare access, and poor housing conditions suggest higher vulnerability than the urban populations studied. Second, baseline mortality rates are drawn from 2020, when COVID-19 introduced two countervailing distortions: elevated all-cause mortality in severely affected states (which could inflate baseline rates and overstate heatwave-attributable excess deaths) and suppressed registration completeness in others (which could depress baseline rates and understate the burden). Banerjee et al. (15) document substantial state-level variation in pandemic-related excess mortality using the same CRS data, confirming that this uncertainty is spatially heterogeneous rather than uniformly directional. We therefore treat the 2020 baseline as a source of state-varying uncertainty rather than a consistently conservative bias. Third, the 97th percentile heatwave threshold is calibrated to historical temperatures and likely is an underestimate of the recent 97th percentile, owing to the higher intensity of recent heatwaves. Additionally, heatwave events are defined using dry-bulb temperature, consistent with de Bont et al. (11), rather than wet-bulb temperature or heat index. In humid coastal districts, some high-heat-stress days will not breach the dry-bulb 97th percentile, and therefore genuine heatwave events may be missed in these areas. Fourth, the analysis assumes uniform risk across seasons and does not account for variation in acclimatization across months. Simplifications one, three, and four point toward underestimation: rural populations likely face higher heat vulnerability than the urban populations studied, historical thresholds understate current heatwave intensity, and excluding seasonal acclimatization variation is conservative. Simplification two, the use of 2020 baseline mortality, has uncertain direction and is discussed above. On balance, the estimates are most appropriately interpreted as conservative with respect to heatwave frequency and temperature thresholds, while carrying directionally uncertain bias from the urban-to-rural coefficient transfer and the 2020 baseline.

A complete formal uncertainty analysis, one that simultaneously propagates uncertainty from the risk coefficients, the baseline mortality estimates, the population projections, and the district-to-city matching, is beyond the scope of this study. As a first-order approximation, we apply the pooled 95% confidence intervals reported by de Bont et al. (11) proportionally to our city-specific coefficients. For the 97th percentile one-day definition, de Bont et al. report a pooled estimate of 12.2% (95% CI 8.5; 15.9); for the five-day definition, 19.4% (95% CI 11.5; 27.9). Applying these confidence interval ratios proportionally yields national ranges of approximately 2,400 to 4,400 excess deaths for the one-day scenario and approximately 18,000 to 43,000 for the five-day scenario, confirming that the order of magnitude of the central estimates is robust across the full range of coefficient uncertainty. Finally, de Bont et al. (11) do not report risk coefficients beyond five consecutive days; the five-day estimates should therefore be treated as conservative for heatwave events of longer duration.

Discussion

The order of magnitude agreement between the sum total of our district-level results and broader low-resolution regional estimates from prior authors, supports the general validity of the extrapolation approach and highlights the added value of subnational resolution for heat preparedness and policy planning.

This study provides the first-ever nationwide, district-level estimates of excess mortality associated with extreme heat events in India. By extending empirically estimated heatwave–mortality relationships from a multi-city study to all districts, we quantify the scale and spatial concentration of heatwave-related deaths under conservative assumptions. The findings indicate that even short-duration heatwaves can result in thousands of excess deaths nationally, while prolonged heat events pose risks comparable to large-scale public health emergencies. Importantly, the mortality burden is not evenly distributed. A small number of populous states and districts account for a disproportionate share of excess deaths, reflecting the combined influence of baseline mortality, population size, and climatic exposure.

The Gini analysis reveals a profound and troubling environmental injustice. The five states bearing the highest heatwave mortality burden account for 66% of national excess deaths while contributing only 29% of India's GDP. This 2.3-fold disproportion between mortality burden and economic capacity means that the states least able to finance adaptation are precisely those facing the greatest heat mortality risk. This finding has direct and urgent implications for how India designs and funds its heat resilience architecture. India's National Heat Action Plans have historically been developed and resourced at the city level, with Ahmedabad's 2010 plan serving as the national model. This city-centric approach, however well-executed, does not address the structural concentration of heat mortality risk in states with low fiscal capacity. The 2.3× GDP disproportion documented here provides a quantitative basis for arguing that federal adaptation investment, including funding under the National Disaster Management Authority and the National Action Plan on Climate Change, should be weighted toward high-burden, low-GDP states rather than allocated in proportion to population or administrative capacity. Failing to account for this disproportion when sizing and directing heat resilience investments will systematically under-resource the populations most at risk. This finding warrants a dedicated policy analysis, which we intend to pursue as a follow-on to the present work.

Several directions would strengthen this line of research. First, rural-specific heat mortality studies in India are needed to replace the urban-derived risk coefficients applied here. Second, updated Civil Registration System data beyond 2020 would improve baseline mortality estimates. Third, Monte Carlo simulation drawing on the confidence intervals in de Bont et al. (11) would allow estimates to be reported with explicit uncertainty ranges rather than as point estimates. Fourth, incorporating wet-bulb temperature metrics would improve hazard characterization in humid coastal districts where dry-bulb temperature understates physiological heat stress. The current dry-bulb threshold definition, while internally consistent with de Bont et al. (11), may undercount high heat-stress days in humid areas where wet-bulb temperature or heat index would identify additional dangerous events not captured by the dry-bulb 97th percentile. Addressing this limitation would require re-estimating source coefficients using humidity-adjusted exposure metrics, which we identify as a priority for future work. Finally, evaluating the mortality impact of Heat Action Plans using the excess mortality framework developed here would provide evidence on intervention effectiveness.

These results highlight the inadequacy of current heatwave mortality surveillance and the risks of relying on coarse national or regional estimates for preparedness planning and resource allocation. District-level estimates, even based on these approximations and simplifications, are highly valuable for identifying high-risk areas, allocating resources appropriately and efficiently, and designing locally appropriate heat action plans. Public access to district-level daily mortality data in India would transform the quality of heat mortality research, enabling direct estimation of local risk coefficients and eliminating the need for the urban-to-rural extrapolation that is the primary source of uncertainty in this analysis.

As extreme heat events become more frequent and intense under climate change, failure to act on such evidence is likely to result in continued large and avoidable loss of life. Strengthening mortality surveillance, improving access to high-resolution temperature data, and integrating heatwave preparedness into district-level public health and disaster management systems are critical steps toward reducing preventable deaths from extreme heat in India.

References

• 1.

  RomanelloMDi NapoliCDrummondPGreenCKennardHLampardPet alThe 2022 report of the lancet countdown on health and climate change: health at the mercy of fossil fuels. Lancet. (2022) 400(10363):1619–54. 10.1016/S0140-6736(22)01540-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2.

  RomanelloMWalawenderMHsuS-CMoskelandAPalmeiro-SilvaYScammanDet alThe 2024 report of the lancet countdown on health and climate change: facing record-breaking threats from delayed action. Lancet. (2024) 404(10465):1847–96. 10.1016/S0140-6736(24)01822-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3.

  RomanelloMWalawenderMHsuS-CMoskelandAPalmeiro-SilvaYScammanDet alThe 2025 report of the lancet countdown on health and climate change: climate change action offers a lifeline. Lancet. (2025) 406(10521):2804–57. 10.1016/S0140-6736(25)01919-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4.

  DebnathRBardhanRBellML. Lethal heatwaves are challenging India’s sustainable development. PLOS Clim. (2023) 2(4):e0000156. 10.1371/journal.pclm.0000156

  • CrossRef
  • Google Scholar

• 5.

  RavindraKBhardwajSRamCGoyalASinghVVenkataramanCet alTemperature projections and heatwave attribution scenarios over India: a systematic review. Heliyon. (2024) 10(4):e26431. 10.1016/j.heliyon.2024.e26431

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6.

  GasparriniAGuoYHashizumeMLavigneEZanobettiASchwartzJet alMortality risk attributable to high and low ambient temperature: a multicountry observational study. Lancet. (2015) 386(9991):369–75. 10.1016/S0140-6736(14)62114-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 7.

  ZhaoQGuoYYeTGasparriniATongSOvercencoAet alGlobal, regional, and national burden of mortality associated with non-optimal ambient temperatures from 2000 to 2019: a three-stage modelling study. Lancet Planet Health. (2020) 5(7):e415–25. 10.1016/S2542-5196(21)00081-4

  • CrossRef
  • Google Scholar

• 8.

  ZhaoQLiSYeTWuYGasparriniATongSet alGlobal, regional, and national burden of heatwave-related mortality from 1990 to 2019: a three-stage modelling study. PLoS Med. (2024) 21(5):e1004364. 10.1371/journal.pmed.1004364

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9.

  DasA. India’s heat waves are deadlier than the official toll suggests. The New York Times (2025). Available online at:https://www.nytimes.com/2025/11/16/world/asia/india-heat-wave-deaths.html(Accessed November 16, 2025).

  • Google Scholar

• 10.

  XuCKohlerTALentonTMSvenningJCSchefferM. Future of the human climate niche. Proc Natl Acad Sci USA. (2020) 117(21):11350–5. 10.1073/pnas.1910114117

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11.

  de BontJNori-SarmaAStafoggiaMBanerjeeTIngoleVJaganathanSet alImpact of heatwaves on all-cause mortality in India: a comprehensive multi-city study. Environ Int. (2024) 184:108461. 10.1016/j.envint.2024.108461

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12.

  YasobantSLekhaKSTrivediPKrishnanSKatorCKaurHet alImpact of heat on human and animal health in India: a landscape review. Dialogues Health. (2025) 6:100203. 10.1016/j.dialog.2024.100203

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13.

  AdairTLopezAD. The civil registration system is a potentially viable data source for reliable subnational mortality measurement in India. BMC Med. (2020) 18:82. 10.1186/s12916-020-01539-6

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14.

  Vicedo-CabreraAMScovronickNSeraFRoyéDSchneiderRTobiasAet alThe burden of heat-related mortality attributable to recent human-induced climate change. Nat Clim Chang. (2021) 11(6):492–500. 10.1038/s41558-021-01058-x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15.

  BanerjeeAMahapatraPKaranANguyenHDaoA. Estimating excess mortality during the COVID-19 pandemic in India using the civil registration system. PLOS Global Public Health. (2022) 2(12):e0001099. 10.1371/journal.pgph.0001099

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16.

  IoannouLGFosterJMorrisNBPiilJFHavenithGMekjavicIBet alOccupational heat strain in outdoor workers: a comprehensive review and meta-analysis. Temperature. (2022) 9(1):1–44. 10.1080/23328940.2022.2030634

  • CrossRef
  • Google Scholar