13 Research methods, statistics, critical review and evidence-based practice13.6 Economic analysis

13.6 Types of economic evaluation

Creator: Resident Doctor Committee
Published: 2026-10-14T21:05:44.632Z

4 min read3 figures

Written by Resident Doctor Committee · Published 14 October 2026

Try it

Cost-effectiveness plane

Interactive

Willingness to pay for one QALY

Below the threshold: worth funding (round marks)Above the threshold: not worth funding (square marks)

Threshold: £20,000 per QALYNICE usually considers £20,000 to £30,000 per QALY.
Worth funding: 1 of 4

Illustrative figures, not taken from any published evaluation. Net monetary benefit is the threshold times the extra QALYs, less the extra cost.
Option	Extra cost	Extra QALYs	ICER	Net monetary benefit	Verdict
1. Group CBT vs usual care	£1,500	+0.06	£25,000 per QALY	−£300	More effective, more costly. Buys health, but at more than the threshold is willing to pay.
2. Digital self-help vs GP follow-up	−£300	+0.02	Dominant	£700	More effective, less costly. Dominant: more health for less money. Adopt it.
3. New antidepressant vs sertraline	£2,500	+0.05	£50,000 per QALY	−£1,500	More effective, more costly. Buys health, but at more than the threshold is willing to pay.
4. Closing a day service	£400	−0.01	Dominated	−£600	Less effective, more costly. Dominated: less health for more money. Reject it.

A negative ICER turns up in two opposite quadrants, so the ratio cannot decide anything on its own. South-east of the origin an option is dominant and always adopted; north-west it is dominated and always refused. In the south-west quadrant the rule reverses and a high ICER is good news, because the saving per QALY given up is larger than what the payer would have spent to gain one. Net monetary benefit of zero or more says the same thing in all four quadrants, which is why it is the safer test.

At £20,000 per QALY, group CBT (£25,000 per QALY) sits just above the line. Move the threshold to £30,000 and it becomes acceptable, while the new antidepressant (£50,000 per QALY) stays out. The digital option is dominant and the day-service closure dominated at any threshold.

Full and partial evaluations

A full economic evaluation compares at least two alternatives on both costs and consequences.
A partial evaluation lacks a comparator or leaves outcomes unpriced, so it cannot by itself show comparative value.

Partial design	What it lacks
Cost-of-illness study	An alternative and any priced health gain
Cost description	A comparator and an outcome
Cost-outcome description	A comparator
Efficacy or effectiveness evaluation	Any costs

The full evaluations

All four count costs the same way. They differ in how they handle outcomes.

Type	Outcome measured as	Result	Main limitation
Cost-minimisation	Shown or assumed equivalent, so only costs are compared	Which option costs less	A non-significant difference does not prove equivalence
Cost-effectiveness	One shared outcome in natural clinical units	Extra cost per extra unit of effect	It compares only interventions sharing that outcome
Cost-utility	Utility-weighted health, usually quality-adjusted life years (QALYs) or disability-adjusted life years (DALYs)	Cost per QALY gained	It depends on how health is measured and valued
Cost-benefit	Money, for costs and consequences alike	Net benefit or benefit-to-cost ratio	Putting money on health raises method and equity problems

Cost-minimisation analysis

Cost-minimisation analysis compares costs alone, whereas every other full evaluation requires outcomes to be measured.
Equivalence needs a prespecified, clinically acceptable range with the confidence interval inside both bounds, so failing to find a significant difference is not enough.

Four point estimates with confidence intervals, labelled A to D, plotted on a mean difference axis from -1.0 to 1.0, with dashed equivalence bounds at -0.5 and 0.5. — Equivalence holds only when the whole confidence interval lies inside both prespecified bounds (dashed), as for A and C; D is not significantly different yet crosses a bound, so it is not equivalent. Lakens, CC BY-SA 4.0, via Wikimedia Commons

Cost-effectiveness analysis

Natural units include cases avoided, remissions, admissions avoided and symptom-free days, and they are never converted to money.
Cost per depression-free day cannot be compared with cost per psychotic relapse avoided.

Cost-utility analysis

The QALY combines survival with preference-weighted quality of life, and its common unit lets results be set against treatments for other conditions.
The National Institute for Health and Care Excellence (NICE) generally uses QALYs for this reason.
Cost-consequences analysis lists costs beside several separate outcomes and never combines them, whereas cost-utility analysis combines them into one measure.

Step plot of health-related quality of life from 0 (dead) to 1 (perfect health) against time, with a blue area A under the curve without intervention and a larger gold area B with intervention, which also ends in death later. — The area under the quality of life curve is the QALYs, so the intervention adds area B by both raising quality of life and extending survival, combining the two into one measure. Chris Sampson, CC BY-SA 4.0, via Wikimedia Commons

Cost-benefit analysis

Because everything is in money, cost-benefit analysis can compare interventions with unlike outcomes on one scale.
A study whose final result is in money is a cost-benefit analysis, even if it measured QALYs along the way.
The benefit-to-cost ratio is monetised benefits divided by costs: above 1, benefits exceed costs, and below 1, costs exceed benefits.
In drug treatment, net benefit is costs before treatment minus costs after, less the cost of treatment, and it may be worthwhile in one subgroup and not another.

Incremental analysis

The incremental cost-effectiveness ratio (ICER) is the extra cost of an intervention over its comparator divided by the extra effect it produces.
The ICER is calculated as (cost of A minus cost of B) divided by (effect of A minus effect of B).
An average cost-effectiveness ratio looks at one arm alone, dividing that arm's total cost by its successful outcomes, whereas the ICER compares arms.
Incremental effectiveness is the difference in health outcome in natural units per unit time, such as bed days saved per patient per year. Figures in pounds belong to the cost side.
Total cost saving is the per-patient saving (bed days saved times unit cost), multiplied by the number analysed and divided by the years of follow-up.

These illustrative figures cover two years of treatment for schizophrenia.

Option	Cost per patient	QALYs	Relapses
Oral antipsychotic with standard follow-up	£8,400	1.10	0.70
Clozapine with haematological monitoring	£13,200	1.34	0.30

The extra £4,800 divided by 0.40 relapses avoided gives £12,000 per relapse avoided, which is a cost-effectiveness result.
The same £4,800 divided by 0.24 QALYs gained gives £20,000 per QALY, which is a cost-utility result.

The cost-effectiveness plane

On the incremental cost-effectiveness plane, north means dearer and east means more effective than the comparator.

Quadrant	New option is	Verdict
Southeast	Cheaper and more effective	It dominates, so no ratio is needed
Northeast	Dearer and more effective	Judge the ICER against a threshold, which NICE sets at £25,000 to £35,000 per QALY
Northwest	Dearer and less effective	It is dominated and discarded
Southwest	Cheaper and less effective	The saving is traded against health loss

Cost effectiveness plane with incremental QALYs on the horizontal axis and incremental cost in pounds on the vertical axis, four labelled quadrants, lines for £20,000 and £30,000 per QALY, and one point at £1,200 for 0.06 QALYs. — The comparator sits at the origin, with north dearer and east more effective; the illustrative therapy costs £1,200 more for 0.06 more QALYs, an ICER of £20,000 per QALY, the same ratio as clozapine in the table. Drawn for the MRCPsychiatrist statistics course

Perspective and valuing outcomes

The perspective of an evaluation is defined by whose costs are counted.
A narrow perspective counts public health and social care budgets and leaves out productivity, informal care and out-of-pocket costs.

Method	How it values a health state
Standard gamble	It varies the probability p of full health against death (1 minus p) until the person is indifferent between the gamble and a certain chronic state
Time trade-off	It varies years of life instead of probability
Willingness to pay	A stated-preference method that asks the most money a person says they would give up for a health gain

Revision material for the MRCPsych examinations. Not clinical guidance, and not for use in patient care.