Health Economics Metrics

Topic 61 of 91· HTA frameworks and prevention economics

Carbon Footprint per QALY

Carbon per QALY is an efficiency ratio — an intervention's carbon emissions (or emissions avoided) divided by the QALYs it delivers — directly analogous to cost per QALY, letting an intervention's carbon efficiency be assessed alongside its cost efficiency. A "carbon-adjusted net monetary benefit" goes a step further, monetizing the carbon impact using the UK Green Book's official non-traded carbon values and netting it against the standard net monetary benefit.

Why it matters

NICE and NHS England now expect environmental impact to be considered alongside cost and QALYs. The NHS has a public net-zero commitment: net zero for its direct emissions by 2040, and net zero for its full supply-chain footprint by 2045. NICE's health technology evaluations manual (PMG36) references environmental sustainability as an emerging consideration in technology appraisal. For a digital health product, this means carbon is becoming a fourth pillar of the value case, alongside cost, QALYs, and dominance on the efficiency frontier — not a replacement for any of them, but a dimension a well-built business case increasingly needs to report.

The maths

Carbon per QALY = total_emissions_tonnes_co2e / total_qalys
  (a negative value means net emissions AVOIDED per QALY gained — a
  double win: better health and lower carbon)

Monetized carbon impact = emissions_tonnes_co2e × carbon_value_per_tonne
  (negative emissions × positive value = a negative cost, i.e. a benefit)

Carbon-adjusted NMB = net_monetary_benefit − monetized_carbon_impact

This extends the cost/QALY efficiency-frontier idea with a second axis — carbon per QALY — the same "plot every option and see what's dominated" logic as dominance and the efficiency frontier, applied to carbon instead of cost.

Worked example

A telehealth service replaces in-person visits, avoiding 5,000 car journeys/year at roughly 8kg CO2e each — 40 tonnes of CO2e avoided, represented as a negative emissions figure (−40.0 tonnes), and it delivers 25 QALYs/year:

Carbon per QALY = −40.0 / 25.0 = −1.6 tonnes CO2e avoided per QALY gained

Using the Green Book's non-traded carbon value (illustrative figure, 2023 non-traded central value ≈ £269/tonne CO2e — the Green Book updates carbon values annually, re-verify before citing in a live analysis):

Monetized carbon impact = −40.0 × £269 = −£10,760

A −£10,760 "cost" is a £10,760 benefit. If the intervention's standalone net monetary benefit is £500,000:

Carbon-adjusted NMB = £500,000 − (−£10,760) = £510,760

The carbon saving adds to the case rather than detracting from it — the double win the negative-emissions framing is meant to surface.

Software engineering connection

This is a live, current intersection with AI/cloud economics: the compute carbon footprint of training and running an AI model is now a real line item in NHS procurement, since NHS supplier contracts above certain thresholds require a Carbon Reduction Plan. Cloud unit economics already tracks cost per unit of compute output; carbon per QALY is the natural template for a future "carbon cost per inference" metric extending that module and inference unit economics into the environmental dimension, though that metric doesn't exist yet.

Pitfalls

  • Scope-boundary gaming: counting only direct (Scope 1) emissions and excluding supply-chain (Scope 3) emissions, which are usually the majority for a digital health product's actual footprint.
  • Using a stale carbon value: the Green Book updates its non-traded carbon values annually, so any cited £/tonne figure must be dated, not quoted as a fixed constant.
  • Treating "carbon efficient" as a substitute for "cost effective": a low-carbon, low-value intervention is still a poor use of NHS resources. Carbon is a fourth pillar alongside cost and QALYs, not a replacement for either.

Sources