India’s public health architecture has a geography problem. National policy is set in Delhi. State budgets are allocated in state capitals. But the actual delivery of health services — the primary health centres, the frontline workers, the cold chains, the drug supply — operates at the district level. The District Collector sits at a desk, manages a budget, and makes the operational decisions that determine whether a pregnant woman gets an ultrasound or walks home without one.
A climate and health fund working across Indian districts faces a choice: design the programme around states, around districts, or around villages?
Why districts
States are too large. A state health director in Jharkhand manages 24 districts with wildly different health profiles. An intervention that works in Ranchi does not necessarily work in Khunti. State-level design smooths over the variation — and the variation is where the problem lives.
Villages are too small. A funder cannot design a grant around 600,000 individual villages. The administrative unit is wrong for the funding cycle, the reporting cycle, and the monitoring cycle.
The district is the sweet spot. It is small enough for a programme manager to know the terrain — the specific health centres, the specific staffing gaps, the specific road conditions that affect vaccine cold chain integrity. It is large enough for a funder to allocate a meaningful budget. And it is the level at which the government officer with actual decision-making power — the District Collector or the Chief Medical Officer — sits.
What this meant in practice
The fund designed its entire architecture around the district. Each of four districts — Chamarajanagar (Karnataka), Dhubri (Assam), Khunti (Jharkhand), and West Singhbhum (Jharkhand) — received a base paper: a detailed profile of the district’s health system, climate vulnerabilities, demographic patterns, and institutional capacity. These base papers were not academic exercises. They were operational documents designed to tell a programme manager: here is what you are working with, here is what is broken, here is where the intervention should land.
The fund then built a fellowship programme within each district — placing fellows in the district health system to work alongside government officers on specific problems (heat action plans, vector-borne disease surveillance, nutrition during extreme weather events). The fellows were embedded, not external. They attended district health meetings. They used district data systems. They reported to district officers.
The learning layer
Most development funds evaluate at the end. The fund designed a learning layer that operated in real time — producing evidence during the fund’s lifecycle, not after the money was spent.
The learning layer tracked three things: what the fellows were doing (activity data), what the district health system was changing (process data), and what health outcomes were shifting (outcome data). The three streams were cross-referenced monthly. When activity data showed that a fellow had successfully introduced a new heat vulnerability screening tool, the process data could show whether the district health system adopted it, and the outcome data could eventually show whether heat-related morbidity patterns shifted.
This is the opposite of the standard evaluation model, where a consultant arrives after the programme ends, collects data, writes a report, and leaves. The learning layer was built into the fund’s governance — the programme committee used the learning data to adjust priorities mid-cycle.
Where this leads
The district-level approach forced a different kind of honesty about what health systems change actually looks like. A state-level programme can report aggregate improvements even when specific districts are getting worse. A district-level programme cannot hide. The variation is the whole point.
This is the same argument the Measurement Checklist makes about indicators: the number you choose to report shapes the story you tell. A state average tells one story. A district profile tells another. The second story is harder to compile, harder to fund, and harder to spin. It is also more useful to the District Collector trying to allocate her budget on a Monday morning.