Principles of Economic Evaluation (Cost-Effectiveness in Health Care)
Economic Evaluation & Decision-Making Context
Economic evaluations compare health interventions on both costs and consequences.
Need an incremental analysis: compute an Incremental Cost-Effectiveness Ratio (ICER) and combine it with a decision rule.
Core decision rules covered in this lecture:
Dominance (strong)
Cost-Effectiveness Thresholds (CET) applied to ICERs
Extended Dominance (weak)
QALY League Tables
Recognition of challenges in applying thresholds in practice
Goal: support resource-allocation decisions so that benefits gained exceed opportunity costs.
Worked Population Example: Public-Health Interventions
Interventions (population-level costs & effects)
Stop-smoking, Nicotine Replacement Therapy (NRT): \$200{,}000 → 40 % quit
Stop-smoking, Advertising: \$100{,}000 → 5 % quit
Stop-smoking, Intensive programme: \$900{,}000 → 50 % quit
Park improvements: \$400{,}000 → 50 % ↑ park use
Free swimming (children): \$100{,}000 → 50 % ↑ swimming
Vaccination: \$800{,}000 → 75 % vaccinated
Heating grant (elderly): \$400{,}000 → 10 % ↓ pneumonia
Same table recast with QALYs (illustrative):
NRT: 80 QALYs, Advertising: 10, Intensive: 100, Park: 4, Swimming: 2, Vaccination: 100, Heating grant: 10.
Demonstrates need for comparable outcome metric (e.g., QALY) when ranking very different programmes.
Typology of Economic Evaluation
Cost Minimisation Analysis (CMA)
Costs measured in ; outcomes assumed identical.
Rule: choose least-cost alternative.
Cost-Effectiveness Analysis (CEA)
Costs in ; consequences in natural units (e.g., life-years, mm Hg).
Decision metric: \text{Incremental cost per natural-unit gained}.
Cost-Utility Analysis (CUA)
Consequences valued as utilities (QALYs or DALYs).
Decision metric: \text{ICER = \$ per QALY (or DALY)}.
Cost-Benefit Analysis (CBA)
Both costs and benefits monetised; decision metric: net benefit or benefit-cost ratio.
Average vs Incremental Cost-Effectiveness
Average Cost-Effectiveness Ratio (ACER):
\text{ACER} = \frac{\text{Cost}}{\text{Effect}}
Example: Treatment A: \frac{3000}{0.3}=10{,}000/\text{QALY}.ICER: compares two options A & B
\text{ICER}{B-A}=\frac{CB-CA}{EB-E_A}
Example: Move A→B: \frac{7000-3000}{0.4-0.3}=\frac{4000}{0.1}=40{,}000/\text{QALY}.Graphical depiction on cost-effectiveness plane reinforces incremental nature.
Cost-Effectiveness Plane & Quadrants
Axes: Incremental Effect (x), Incremental Cost (y).
Quadrant meaning:
SE (−ΔCost, +ΔEffect): New dominates (cheaper & more effective).
NW (+ΔCost, −ΔEffect): New dominated by existing.
NE (+ΔCost, +ΔEffect): Trade-off → needs threshold.
SW (−ΔCost, −ΔEffect): Usually rejected on clinical grounds.
Decision Rule 1: Dominance (Strong)
If new intervention both cheaper & more effective, it dominates.
If more expensive & less effective, it is dominated.
Useful for quick eliminations but limited because most new technologies lie in NE quadrant.
Decision Rule 2: Extended Dominance (Weak)
Applies when >2 alternatives.
Steps to identify:
Rank by effectiveness.
Remove strongly dominated options.
Compute ICERs between adjacent options.
Remove any option whose ICER is higher than that of a more effective alternative (i.e., increasing effectiveness does not raise ICER monotonically).
Logic: a mix (weighted combination) of two alternatives can out-perform a middle option.
Example (Costs, QALYs):
Standard: 5{,}000, 1
D: 10{,}000, 2 → ICER 5{,}000/QALY
B: 35{,}000, 4 → ICER 12{,}500/QALY (adjacent).
C: 25{,}000, 3 has ICER 15{,}000/QALY vs standard → higher than B’s ICER; C is extendedly dominated.
After removal: only Standard, D, B, A remain; final frontier ICERs: 5{,}000, 12{,}500, 20{,}000 per QALY.
Decision Rule 3: ICER vs Cost-Effectiveness Threshold (CET)
If neither dominance rule applies, compare ICER with threshold \lambda (lambda, sometimes k).
If \text{ICER} < \lambda → deemed cost-effective.
Interpretation of \lambda:
Opportunity cost of resources displaced.
Equals ICER of last programme funded before budget exhausted.
Proxy for marginal willingness-to-pay for a QALY.
Country practice:
NICE (England): historical implicit range £20–£30 k per QALY; research suggests real NHS opportunity cost ≈ £12.9 k (Claxton 2015).
WHO (historical): <1×GDPpc = highly cost-effective; 1–3× = cost-effective; >3× = not.
• Criticised for ignoring budget impact; guidance now discouraged.Newer empirical estimates: 0.5×GDPpc (LICs) to 1×GDPpc (most settings) based on opportunity-cost methods (Ochalek, Woods, Pinchon-Riviere, 2016–23).
Exercises & Illustrations
Smoking CEA Exercise:
Advertising: 100{,}000 cost, 500 quitters → ACER 200/\text{quitter}.
NRT: 200{,}000, 4 000 quitters → ACER 50/\text{quitter}.
Incremental NRT vs Advertising: \Delta C = 100{,}000, \Delta E = 3{,}500 → ICER ≈ 29/\text{quitter}.
Technology choice (threshold $30 k):
T1: 125{,}000, 8.2 QALYs → front-runner.
T2: 180{,}000, 6.9 → dominated (more cost, fewer QALYs).
Technology choice (threshold $40 k): (T1 8.2; T2 10; T3 13 QALYs)
Establish ICER frontier, compare to $40 k.
Additional exercise (T1, T2, T3 different numbers) repeats extended dominance evaluation.
QALY League Tables
Early approach: rank interventions by /QALY and fund top until budget spent.
Objectives:
Benchmarking: place result in broader context.
Resource allocation: shift money from low-rank to higher-rank interventions.
Example (UK £1990):
Cholesterol diet therapy: £220/QALY
Hip replacement: £1 180/QALY
Haemodialysis: £21 970/QALY
Neurosurgery for malignant tumour: £107 780/QALY
Limitations:
Heterogeneous study methods (discounting, perspective, time horizon, utility elicitation) reduce comparability.
Choice of comparator affects ICER heavily (e.g., A2 vs standard vs A2 vs A1).
Practical & Ethical Challenges with CETs
Threshold may be arbitrary if opportunity cost information lacking.
Budget impact vs affordability not captured solely by ICER.
Equity concerns:
Standard QALY weights everyone equally; may not reflect social preferences for severity or age.
Potential remedies: equity weights, alternative thresholds, equity-informative CEA.
HTA bodies (e.g., NICE) may also weigh:
Not in a qualitative manner
Clinical effectiveness certainty
HRQoL measurement adequacy
Innovation degree
Alternative treatments
Disease severity / end-of-life context
Summary of Decision Rules
Use dominance to cull clearly inferior/superior options.
Use extended dominance to ensure efficiency along the cost-effectiveness frontier.
Compare remaining ICERs to a context-appropriate threshold \lambda.
If \text{ICER} < \lambda and other HTA considerations acceptable, adopt; otherwise reject or request more evidence.
Ongoing research continues to refine thresholds and incorporate equity systematically.