Health Economics Metrics

Topic 47 of 76 · HTA frameworks and prevention economics

Prevention Economics

The economics of intervening before disease occurs or progresses. The headline finding is counterintuitive: most prevention does not save money — it buys health at a good price. Cohen, Neumann and Weinstein's landmark NEJM analysis found fewer than 20% of preventive interventions are net cost-saving; the rest are cost-effective at best.

Why it matters

"Prevention saves money" is the most repeated false claim in health policy, and business cases built on it get demolished by health economists. The honest structure: prevention costs money now (screening whole populations, treating risk factors in people who would never have gotten sick) and returns health later — usually at a good cost per QALY, occasionally at a saving, sometimes at a terrible price. Knowing which regime you're in is the analysis. The distinction matters commercially: a prevention product sold as "saves the NHS money" invites an audit it will fail; sold as "buys QALYs at £4,000" it can win on the same facts. See earlier intervention for the within-pathway version.

The math

Net cost of prevention (per person) =
    intervention cost × everyone treated
  − downstream costs avoided × the few who would have progressed
  (both discounted — the avoided costs are years away; see
   discounting-and-time-preference.md)

Cost-saving requires: intervention cost < P(progression) × avoided cost × discount factor
Cost-effective requires only: net cost / QALYs gained < threshold

The prevention paradox: intervention cost multiplies over the whole population; benefits accrue only to the counterfactual few.

Worked example

A hypertension-management app offered to 100,000 at-risk adults, £25/person/year. Over 10 years it prevents 400 strokes (each costing £45,000 discounted, and 3 QALYs lost).

Cost:    100,000 × £25 × 10 yrs (discounted ≈ ×8.3) ≈ £20.8M
Offsets: 400 × £45,000 = £18.0M
Net cost ≈ £2.8M — NOT cost-saving

QALYs gained = 400 × 3 = 1,200
Cost per QALY = 2.8M / 1,200 ≈ £2,300/QALY — outstandingly cost-effective

Same program, both truths: it loses £2.8M in cash and buys health at a tenth of the NICE threshold. Fund it on the second number; never promise the first.

Software engineering connection

Shift-left quality is prevention economics, caveat included. Reviews, tests, and static analysis apply cost to every change to catch issues in the few that would have progressed to production incidents. The defect-cost curve (10–100× by stage) plays the role of stroke costs — and the honest conclusion mirrors health: shift-left is usually cost-effective, not automatically cost-saving, because most flagged issues would never have become incidents (the counterfactual few problem). Compute it: total gate cost per period vs incidents actually avoided × incident cost — the same worked-example structure, with NNT as the per-catch unit.

Pitfalls

  • Claiming cost savings when the evidence supports cost-effectiveness — the defining error of prevention advocacy in both domains.
  • Undiscounted future offsets: benefits 15 years out at face value.
  • Ignoring overdiagnosis/overtreatment costs: prevention finds pseudo-disease too — see screening economics.

Sources