This is a preliminary, time-boxed investigation. It is not comprehensive and I have likely made judgement calls a domain expert would disagree with. I am posting this to make a first-pass argument into the open and to invite correction. I particularly welcome pushback from people with direct experience of HAP implementation, the LMIC heat-health funder ecosystem, or heat-mortality epidemiology.
Summary
In this post I investigate “Heat Action Plan Implementation support” in LMICs as a candidate for philanthropic funding. A Heat Action Plan is a set of frameworks designed by cities or regions to mitigate health impacts of extreme heat events. A HAP can have several elements:
- Early-warning systems: colour-coded alerts triggered by city-specific temperature thresholds
- Inter-agency coordination (city health department with hospitals, ambulance services, meteorological department, schools, labour departments, and media)
- Capacity-building among health professionals
- Public outreach
In India, after the first HAP was implemented in the city of Ahmedabad, the HAPs have been replicated to 100 cities in 23 states already. However, most HAPs in India are far from functional. The main reasons for this are lack of funding and limited technical capability.[1] This indicates that the governance / institutional infrastructure and motivation exist, but HAPs lack the means to be implemented effectively. This is a high-leverage opportunity for impact: Philanthropic funds will support the implementation of HAPs in cities with high heat-burden; the city government will absorb the recurring running cost; and the benefit will recur every heat season for as long as the plan persists.
In this post:
- I use the ITN framework to evaluate the intervention
- Build a cost-effectiveness model and run sensitivity analysis
- List major sources of uncertainty for future investigation
My base-case cost-effectiveness lands at roughly $25 per DALY averted (about 3967x against Coefficient Giving's published GHW bar of $100,000 per DALY). The pessimistic scenario, with large discounts on effectiveness and attribution, comes out to roughly 700x.
Importance
Mortality
Heat mortality faces challenges in attribution and is systematically undercounted because deaths from cardiovascular, renal and respiratory causes during heat waves are not counted under heat mortality. Despite that, in the Indian city of Ahmedabad alone, 1344 excess deaths were attributed to the May 2010 heatwave.[2] Excess-mortality modelling consistently suggests true heat mortality in South Asian cities is several-fold higher than coded deaths. The Lancet Countdown 2025 estimates an annual average of 546,000 heat-attributable deaths globally, with the rate having risen 23% since the 1990s.[3]
Productivity
According to Lancet Countdown on Health and Climate Change, heat exposure resulted in a record-high 639 billion potential work hours being lost in 2024, 98% above the 1990–99 average. The hours lost in 2024 resulted in potential losses worth US$1.09 trillion: almost 1% of global domestic product.[3]
I have not included productivity loss in the cost-effective modelling for the intervention in this post.
LMIC concentration and trajectory
The burden is concentrated, unsurprisingly, in tropical and sub-tropical urban populations with low air-conditioning coverage, large number of outdoor workers and high-density housing and informal settlements. India’s urban population exceeds 500 million and is projected to reach 770 million by 2040.
Two salient characteristics of the heat problem are important to note:
- The threat to health from heat is worsening rapidly
- There are geographically concentrated pockets of impact (urban/ suburban with outdoor workers) where interventions can reach scale.
I do not have a clear number on heat-related mortality in LMICs per year. For cost-effectiveness modelling I have used deaths averted numbers from the Ahmedabad HAP pilot evaluation paper mentioned later in the post.[4] The uncertainty on this number can be reduced by a more in-depth investigation of GBD heat-attribution dataset. I plan to do this as a follow-up post or as an update to this one.
Neglectedness
Funders
There has been increasing philanthropic attention on Heat resilience recently. The Climate and Health Funders Coalition earmarked $300M at COP30 in 2025, and ClimateWorks' Adaptation and Resilience Fund committed $50M in 2025 for heat resilience and adjacent areas. However, the specific intervention for HAP implementation remains underfunded.[1] NRDC India, a key technical partner for Indian cities, operates on a relatively small budget.
Non-functional HAPs
In India context, the other aspect of neglectedness is the on the ground reality of Heat Action Plans. Based on the Centre for Policy Research's 2023 assessment of 37 HAPs across 18 states, only 2 of 37 plans explicitly conducted vulnerability assessments, only 3 of 37 identified funding for their proposed measures, and 11 of 37 HAPs discussed funding sources, eight of which asked implementing departments to self-allocate resources, pointing at a serious funding constraint. None of the plans are legally notified.[1]
This matters because the “number of cities with HAPs” suggests low scope for counterfactual impact, but the “number of HAPs that actually reduce mortality and economic loss” suggests the opposite. This is the gap that philanthropic funding can address: taking existing plans and making them functional. The Ahmedabad case study below is a good example of what “functional” HAP looks like in practice.
Tractability
I use Ahmedabad's HAP as an example of a tractable model. As the first HAP in South Asia, it was developed by a coalition of Ahmedabad Municipal Corporation, the Indian Institute of Public Health at Gandhinagar, the Public Health Foundation of India, and the Natural Resources Defense Council (NRDC).
A pilot evaluation study computed HAP’s effect on all-cause mortality using a distributed lag non-linear model to compare 2014-15 (post-HAP) to a 2007-2010 baseline (pre-HAP). The relative risk of mortality at maximum temperature 47 °C fell to 1.25 (95% CI 1.02 to 1.53) from 2.34 (95% CI 1.98 to 2.76). The paper estimates 1,190 (95% CI 162 to 2218) annualised deaths avoided per year[4]. The wide CI comes from the structural limits: heatwaves are sporadic, existence of large confounders, and small sample size (one city).
Implementer landscape
Several credible implementers exist with funding absorptive capacity:
- NRDC
- CEEW (India’s first ward-level HAP working with Thane Municipal Corporation)
- Indian Institutes of Public Health (IIPH)
- Public Health Foundation of India
The implementer base outside India needs to be explored. I expect it to be limited and would need development through funding.
Cost-Effectiveness
The cost-effectiveness modelling demonstrates the leverage mechanism: the philanthropic funding pays for one-time technical assistance (protocol development, threshold definitions, vulnerability mapping, training, implementation support); the city absorbs the recurring running costs (messaging, staff, training etc.), and the benefits recur every heat season.
Full CEA spreadsheet: here
I modelled the burden and effectiveness of the intervention on the Ahmedabad HAP pilot evaluation study [4]and scaled it down to an average mid-size Indian city (2M population) that is heat-vulnerable. I have used an attribution discount of 0.6 (to adjust for wide CI in excess-mortality modelling) and an effectiveness transfer discount of 0.35 (taking Ahmedabad as better than average case study).
The adjusted deaths averted per year is 69; at 20 DALYs per heat death and a 10-year recurring benefit timescale, this is roughly equal to 13,883 DALYs per city.
The Technical assistance cost is a weak input in my model. I use $275,000 as a one-time plus $75,000 multi-year support cost. There is no published per-city HAP TA cost that I could find. The sensitivity analysis tests robustness from $200k to $750k.
I run three scenarios and compare the $ per DALY to Coefficient Giving’s valuation of $100,000 per DALY:
| Scenario | $ per DALY | CG Multiple |
| Pessimistic | $144 | ~ 700x |
| Base | $25 | ~ 3967x |
| Optimistic | $4.5 | ~ 22,300x |
In the model, I show that the sensitivity analysis for two weak inputs (Technical Assistance cost and effectiveness transfer) gives a multiple of above 2000x in 10 out of 12 scenarios.
What philanthropic fund support could look like
City prioritisation: Taking high heat-burden, governance and institutional appetite, and presence of implementers as prioritisation criteria, funders can focus on South Asian cities first (India, Pakistan, Bangladesh), expanding to West African and Southeast Asian cities.
Grant structure: Three potential grants types:
- Turning paper HAPs into functional HAPs through technical assistance and implementation support
- Smaller grants to bring HAPs to cities without one in place
- Pooled technical resources and scaling capacity building to reduce the marginal cost of additional city / functional HAP roll-out.
Major sources of uncertainty
The Technical Assistance cost: I have used $275,000 one-time plus $75,000 multi-year light support per city. I could not find a published figure for HAP implementation. One way to resolve this uncertainty is to reach out to NRDC India and CEEW for guidance.
Effectiveness transfer factor: I use 0.35 as a conservative guess. The transfer factor could plausibly be anywhere between 0.2 and 0.8. The cost-effectiveness stays promising for most of this range, but could weaken if technical assistance value goes high.
Attribution: My 0.6 attribution discount is a judgement call. This is a partial overlap with the wide CI in capturing uncertainty and a more rigorous analysis would use the lower CI bound directly.
Counterfactual: The fund's counterfactual value rests on the quality argument (most existing HAPs are weak) and on geographic expansion (I am uncertain about the infrastructure outside India). Both claims need more validation and research.
Climate trajectory: The fund's value increases substantially under high-warming scenarios because heat burden grows. I have not modelled this and the CEA effectively assumes static burden.
What I would do next to reduce uncertainty
- Roughly in priority order: Resolve the Technical assistance cost question and effectiveness transfer discounting question through direct conversations with NRDC India, CEEW
- Pull GBD heat-attributable DALY data by age band to refine the DALY/death averted input
- Build a structured funder-mapping of heat-health adaptation across India, South Asia, and one or two African cities (in Kenya, Ghana). This would refine neglectedness claim.
Conclusion
Heat Action Plan implementation support is worth an in-depth investigation and attention from funders. Broadly there are three reasons that make this intervention promising:
- A real leverage mechanism exists
- Implementer ecosystem that can absorb capital
- The health impact of heat is worsening and is expected to stay on this trend. Funding it now goes a long way.