Health Economics Metrics

Topic 22 of 76 · Economic analysis types

Cost-Benefit Analysis (CBA)

CBA values both costs and outcomes in money. It is the only analysis type that can answer "is this worth doing at all?" — not merely "which option is best?" — because monetized benefits can be compared directly against costs.

Why it matters

CBA is the standard of the UK's HM Treasury Green Book for all public spending appraisal, health included when outcomes can be monetized. Where CEA/CUA stop at "cost per unit of health," CBA prices the health itself (QALY × threshold value) and everything else — time, travel, carbon — and reports one net figure. Every full NHS digital business case contains a CBA-shaped economic case.

The math

NPV (net present social value) = Σ_t [ (Benefits_t − Costs_t) / (1 + r)^t ]
BCR (benefit-cost ratio)       = PV(benefits) / PV(costs)

Adopt if NPV > 0 (equivalently BCR > 1); rank by NPV, not BCR.
r = 3.5% (Green Book social time preference rate)

Health effects can enter monetized as QALYs × λ (see willingness-to-pay thresholds). The Green Book also mandates optimism-bias adjustments — uplifting cost estimates and haircutting benefits by evidence-based percentages, because appraisals are systematically rosy.

Worked example

An e-referral system, 5-year horizon, 3.5% discount:

Costs:    build £1.2M (year 0), run £300k/yr (years 1–5)
Benefits: admin savings £250k/yr, avoided duplicate diagnostics £280k/yr,
          patient time saved 40,000 hrs/yr × £15 = £600k/yr → £1,130k/yr

PV costs    = 1,200k + 300k × 4.515 (annuity factor) = £2,555k
PV benefits = 1,130k × 4.515                          = £5,102k

NPV = 5,102 − 2,555 = +£2,547k     BCR = 2.0

Apply Green Book optimism bias (say +40% on build cost, −20% on benefits): PV costs ≈ £3,035k, PV benefits ≈ £4,082k, NPV ≈ +£1,047k — still positive, which is the point of the adjustment: cases should survive their own optimism.

Software engineering connection

Engineering business cases are informal CBAs. The Green Book upgrades worth stealing:

  • Optimism bias as a standard uplift — engineers underestimate migration cost as reliably as ministries underestimate infrastructure cost; apply a stated uplift instead of pretending this time is different.
  • Monetize the dominant benefit honestly or not at all — patient/user time is monetized at defensible rates; "brand value" is not.
  • NPV ranks, BCR doesn't: a tiny project with BCR 5 can matter less than a big one with BCR 1.6.

Pitfalls

  • Monetizing the unmonetizable to inflate benefits (morale, "strategic alignment") — keep those qualitative, per cost-consequence analysis.
  • Counting transfers as benefits: money moving between public bodies nets to zero at the societal perspective.
  • No counterfactual: benefits are measured against the do-minimum option, not against zero.

Sources