One evidential question, different institutional uses
The same prevention evidence can reach different decisions. What changes is the decision owner, the decision at stake and the cost of relying on it wrongly. Each use gets its own decision scope.
Pricing and assumption setting
- The question
- When is prevention evidence robust and transferable enough to change a pricing or projection basis?
- The protocol asks
- Is the effect supported by a valid comparison? How does it carry to expected claims on this exposure? What remains uncertain, and which experience would reopen it?
- The institution decides
- Whether to change the basis, how to weigh the evidence alongside credibility and judgement, and the approval route.
Reinsurance renewal and underwriting
- The question
- What may prevention evidence contribute to the expected-loss view for a declared treaty decision?
- The protocol asks
- What is the supported effect, the claims impact range and the permitted adjustment? Which monitoring terms would reopen the view? A fall in attritional hospital frequency can matter to a quota share and little to a high excess-of-loss layer.
- The institution decides
- Whether to write, retain or cede the risk; price, structure, limits, capacity and terms.
Outcome-linked contracting
- The question
- How much prevention-linked payment can current evidence support, and what happens if that evidence weakens?
- The protocol asks
- Is the outcome defined and attributable? How does the payment mechanism map to the payer’s exposure? What are the ceiling, duration and review triggers?
- The institution decides
- The funding arrangement, payment terms, safeguards and governance.
Public purchasing In development
A public purchaser may need to fund, scale or redesign a programme before long-term outcomes are observable. The current protocol is written for insurance decisions; this application follows the same logic with its own impact assessment in place of claims impact.
The protocol keeps four quantities apart: what was claimed, what the evidence supports, what that support may change, and what the owner decides. A case may support an adjustment, call for more evidence, find no material consequence, or confirm that ordinary governance already suffices.
A hypertension programme, from claim to decision
The insurers, partners and figures are invented, drawing on published ranges for treatment cost and blood-pressure response. Every real case rests on its own evidence and its own decision scope.
Why India
India brings together the conditions in which recognition has the most at stake. About one adult in four has hypertension, and few have it under control. Health insurance mostly pays for hospitalisation, while medication and follow-up are paid out of pocket. Protocol-based treatment with generic medicines costs little; many health insurers operate close to breakeven on underwriting; and health IDs make linking programme records to claims possible, though demanding.
The situation
A private health insurer covers millions of lives, mostly through group policies that pay for hospitalisation. A care partner offers a frugal programme: screening, protocol-based medication through generic pharmacies, teleconsultations, and follow-up by community health workers.
After eighteen months, the partner reports that participants had 20% fewer cardiovascular hospitalisations than similar members who did not take part. The insurer must set next year’s group pricing now.
Step by step
- Define the decision. The head of group pricing records the decision: next year’s pricing basis for the group health book. The assumption that may move is the expected rate of cardiovascular hospitalisations among enrolled members. She sets an adjustment limit in advance. Reserves, capital or reinsurance capacity would each need a decision scope of their own.
- Agree the analysis plan. Linkage through the health ID or policy number works for about eight in ten participants, and the evaluator judges the data usable with stated limitations. The participant definition, comparison group, endpoint, diagnostics, fallback and review triggers are frozen before the evaluation result is known.
- Review the evidence. The first check fails: participants’ blood pressure was already falling faster than the comparison group’s before enrolment, because more engaged members joined first. The agreed fallback, comparing trends rather than levels, applies. Retained participants show systolic blood pressure about 9 mmHg lower; after the residual bias assessment, 7 mmHg is supported. The 20% hospitalisation headline is recorded, but at eighteen months claims are too few and too immature to bear weight.
- Assess the claims impact. Half of hypertensive members enrol, and about three in five stay in care. For them, 7 mmHg corresponds to roughly 14% fewer major cardiovascular events; across all participants, about 8%. Hypertensive members account for about 40% of the book’s claims and cardiovascular admissions for about 30% of theirs, so around 6% of claims are in scope. The result: about 0.5% of expected claims. The headline at face value would have meant about 1.2%.
- Set the use limits. Up to the claims impact, within the adjustment limit, for next year’s group pricing basis only, for twelve months, while retention and programme delivery hold. The evidence may be shared with the reinsurer; any use in treaty pricing needs its own review. The pricing actuary includes 0.4%, below what is permitted, because retention beyond the first year is uncertain. The insurer also decides to cover hypertension medication for participants.
- Monitor experience. With medication covered, more participants stay in care and blood-pressure control holds. Maturing claims sit within the expected range. The retention review reopens the claims impact, support increases, and the allowance rises at renewal.
The trail
On a book with ₹10,000 crore of annual claims, the supported effect is about ₹50 crore a year, roughly €5 million. Across some 25 lakh participants, that is about ₹200 per participant a year, close to the cost of a year of protocol-based hypertension treatment in India’s public sector. In claims terms: about 40 cardiovascular admissions per 1,000 participants a year at an average of ₹60,000, of which the supported effect avoids about 3.4.
A different market, a different ending
A digital programme with home monitoring and nurse coaching, on a health book with €500 million of annual claims. The provider reports 18% fewer cardiovascular admissions.
- Decision scope. Next year’s pricing basis for a group health product, with an adjustment limit set in advance.
- Analysis plan. Participants, comparison group, endpoint and review triggers fixed before the evaluation result.
- Evidence review. Selection takes the 18% headline to 13%; 9% is supported.
- Claims impact. About 0.6% of expected claims (€3 million), against 1.2% (€6 million) at face value.
- Use limits. The actuary includes 0.5% (€2.5 million), as programme funding is secured for one year only.
- A year later. Participation falls after an app change; support is reduced and the allowance lowered.
Six lessons about recognition
The percentage travels; its meaning does not
Both cases land near half a percent of claims. What that buys differs: a share of an intensive European programme, much of a frugal Indian one.
Recognition matters most where prevention and events have different payers
Where cover pays for the stroke but not the medication, recognition gives the insurer a credible basis for funding prevention.
Recognition can change what it measures
Value that funds medication can raise retention; the next assessment then treats the programme as modified.
Support moves in both directions
It fell in Europe and rose in India, each time against conditions agreed in advance.
Thin margins raise the stakes both ways
Where underwriting runs near breakeven, half a percent is material, and so is a headline taken at face value.
The hardest step depends on the market
In India the demanding steps are evidential: linkage, selection and endpoint maturity. In Europe they are economic: claims impact and programme cost.