Interpretation of Cost Effectiveness Analysis & Introduction to Cost Benefit Analysis

Part 1: MCDA and Economic Evidence in Policy

  • Topic focus (Page 1–5): review of league tables; brief introduction to multi-criteria decision analysis (MCDA); introduction to cost-benefit analysis (CBA).
  • Learning context (Page 4): aim to understand how economic evidence informs policy decisions and to consider what else society cares about beyond health gains via MCDA.
  • Core idea: MCDA extends beyond cost-effectiveness analysis to include multiple objectives and equity considerations; CBA aggregates benefits and costs in monetary terms to assess overall welfare impact.

Part 2: Policy-Relevant Economic Evidence and MCDA Basics

  • Economic evidence informs policy decision-making using decision rules in economic evaluation (Page 5).
  • MCDA broadens evaluation to include distributional concerns (e.g., prioritizing worse-off populations, access, socioeconomic disadvantage).
  • Examples of MCDA in public/private sectors: transport, immigration, education, environment; allows prioritization across multiple health and social objectives.
  • Key motivation: decision-makers may prioritize health gains for groups with worse health, lower access, or socioeconomic disadvantage.

Part 3: MCDA in Practice (Overview)

  • What MCDA offers: an explicit, structured framework to incorporate multiple criteria beyond cost per health outcome.
  • Core elements (summarized):
    • Define decision problem and stakeholders.
    • Select criteria reflecting valued objectives (health gains, equity, accessibility, etc.).
    • Measure performance of each option on each criterion.
    • Score alternatives against criteria; elicit stakeholder preferences.
    • Weight criteria to reflect trade-offs among criteria.
    • Aggregate scores to derive overall value for each option.
    • Address uncertainty with sensitivity analysis.
    • Present results clearly (graphs/tables) and interpret results for decision making.

Part 4: MCDA Guidelines (ISPOR, Thokala et al., 2016)

  • 1. Defining the decision problem
    • Understand and define the problem; identify stakeholders (patients, clinicians, payers, regulators, general population).
  • 2. Selecting and structuring the criteria
    • Choose criteria to measure alternatives; can be identified via focus groups/workshops.
  • 3. Measuring performance
    • Gather performance data for each alternative on each criterion (systematic reviews, meta-analyses, expert opinions).
  • 4. Scoring alternatives
    • Score alternatives according to preferences for change within each criterion.
    • Methods: compositional vs decompositional approaches.
  • 5. Weighting criteria
    • Elicit stakeholder trade-offs; weights reflect relative importance of criteria.
  • 6. Calculating aggregate scores
    • Compositional: multiply each criterion score by its weight and sum.
    • Decompositional: use valuation functions from experiments (DCE, conjoint) to estimate utilities.
  • 7. Dealing with uncertainty
    • All MCDA steps involve uncertainty; use probabilistic sensitivity analysis where possible.
  • 8. Interpretation and reporting
    • Results can be graphical or tabular; alternatives ranked by importance; total scores can be monetized to inform resource allocation.

Part 5: MCDA in Practice: Baltussen & Niessen (Baltussen & Niessen, 2006)

  • MCDA example setup (Performance Matrix):
    • Criteria include: effectiveness (e.g., US$ per DALY), severity of disease, disease of the poor, age groups.
    • Examples of options: antiretroviral treatment in HIV/AIDS, treatment of childhood pneumonia, inpatient care for acute schizophrenia, plastering for simple fractures.
  • Scoring approach (illustrative):
    • Weights (e.g., 100, 50, 0 etc.) and scores for each criterion are combined to form total scores.
    • Example provided shows costs, QALYs, and ICERs for various options; some options are extendedly dominated, dominated, or non-dominated depending on their ICERs and costs.
  • Example insight: when applying MCDA, decision rules can identify which option offers best trade-off given stakeholder preferences and distributional concerns.
  • Key note: decision-makers may assign scores for cost-effectiveness, severity, disease of the poor, age, etc., with weights to reflect societal preferences.
  • Visual cue: scoring table demonstrates how different options can be ranked when multiple criteria are considered.

Part 6: Cost-Effectiveness vs MCDA: QALY League Tables and Budget Constraints

  • QALY league table and decision rules (Pages 7–9):
    • Decision rules in cost-effectiveness analysis (CEA) involve dominance and extended dominance and ICERs.
    • ICER: ICER=extIncrementalCostextIncrementalQALYs=riangleCostriangleQALYsICER = \frac{ ext{Incremental Cost}}{ ext{Incremental QALYs}} = \frac{ riangle Cost}{ riangle QALYs}
    • Thresholds: e.g., $50{,}000/ ext{QALY} is a common upper bound; decisions hinge on whether ICER is below/above threshold.
    • Dominance: a new treatment that is more effective and less costly dominates the existing treatment.
    • Extended dominance: a treatment that is not on the efficient frontier and can be excluded because a combination of other options provides a better cost-effectiveness profile.
  • Cost-effectiveness plane interpretation:
    • Axes: Incremental Cost (horizontal) and Incremental Effect (vertical).
    • Quadrants indicate whether an option costs more/less and is more/less effective.
  • Example questions from slides (illustrative):
    • If threshold is $50,000/QALY, which intervention is selected given incremental costs and QALYs?
    • With a budget constraint (e.g., $65,500), which interventions fit within budget and what is implied threshold?
  • Takeaway: ICERs and dominance rules guide selection under budget and threshold constraints; MCDA adds further dimensions when multiple criteria beyond QALYs are important.

Part 7: Using MCDA in Decision-Making (Broader View)

  • Practical workflow (from slides 11–18):
    • Define decision problem and stakeholders.
    • Select and structure criteria.
    • Measure performance data for each alternative.
    • Score alternatives according to preferences for each criterion.
    • Weight criteria to reflect trade-offs.
    • Calculate aggregate scores (compositional vs decompositional).
    • Address uncertainty with probabilistic sensitivity analysis.
    • Interpret results and report findings; rank alternatives; total scores can be monetized if desired.
  • Uncertainty considerations:
    • Every aspect of MCDA is subject to uncertainty; use probabilistic methods to reflect parameter uncertainty.
  • Reporting and interpretation:
    • Results can be displayed graphically or in tables; alternatives ranked by total value; monetization of total scores can facilitate budgetary decisions.

Part 8: Welfare Economics and Foundations for CBA (Part 2)

  • Aim of CBA in welfare economics: evaluate whether benefits to society exceed costs, using monetary terms to express all benefits and costs.
  • Welfare economics (Welfarism) foundations (Page 24):
    • Proposition 1: Social welfare should be based on individuals’ welfare (utility).
    • Proposition 2: Individuals are the best source of information on their own welfare (consumer sovereignty).
    • Proposition 3: Markets are assumed to be competitive and current income distribution is appropriate (ability to pay is not an issue).
  • Welfarism combines individual utility with a consequentialist view; underpinning theory for CBA.

Part 9: Pareto and Kaldor-Hicks Principles (Pages 25–26)

  • Pareto improvement (Actual Pareto):
    • A policy makes at least one person better off and makes no one worse off.
    • Allocative efficiency achieved when no Pareto improvement is possible; very restrictive.
  • Potential Pareto improvement (Kaldor-Hicks):
    • Some winners and losers, but if winners could compensate losers and remain better off, society would benefit.
  • Kaldor-Hicks criterion in CBA:
    • Use willingness to pay (WTP) and compensation tests to operationalize the criterion.
    • Ezra Mishan emphasized monetary compensation: willingness-to-accept (WTA) and willingness to pay (WTP) as comparative measures.
  • WTA vs WTP:
    • WTP: amount people are willing to pay for improved health; WTA: amount people would accept as compensation for worse health.
    • WTA tends to be higher than WTP due to income effects, diminishing marginal utility of income, and other biases.

Part 10: Willingness to Pay (WTP) and Valuation Methods (Pages 27–33)

  • WTP concepts and definitions (Page 28):
    • WTP definitions for health benefits: certainty-based, uncertain outcomes, and future uncertain outcomes.
    • Components of program benefit that can be valued via WTP: future health care cost savings, production gains, income effects.
  • WTP measurement approaches (Page 31–32):
    • Market values (e.g., safety equipment) and revealed preference methods (e.g., wage premia for dangerous jobs).
    • Expressed or stated preference methods (contingent valuation, CA, DCE).
  • Contingent valuation (CV) and stated preference methods (Page 32):
    • CV surveys present hypothetical scenarios and value health outcomes via WTP/WTA.
    • Valuation techniques include: Value of a Statistical Life (VSL) and Value of a Change in Health; conjoint analysis (CA) and discrete choice experiments (DCE).
  • Value of a Statistical Life (VSL) and VSLY (Page 33):
    • VSL: society’s value of reducing the risk of death; life is treated as a statistical life, not an individual case.
    • VSLY: value of a statistical life year.
    • Example figures (2024 AUD): VSL ≈ 5.7imes1065.7 imes 10^6; VSLY ≈ extAUD245,000ext{AUD } 245{,}000.
    • Note: VSL and VSLY often used in CV and CV-based health benefit valuation.

Part 11: Contingent Valuation and CV Question Types (Pages 34–42)

  • CV exercise and hypothetical scenarios (Page 34): comparing two airline choices with different safety and price attributes; illustrates WTP for risk reductions.
  • CV in environmental evaluations (Page 35): CV is widely used where markets are absent or incomplete, e.g., water/air quality, recreation, wildlife preservation; used to set payments for environmental damage/pollution.
  • Open-ended CV questions (Page 36):
    • Respondents name the maximum amount they would pay for a given health outcome or service (high cognitive burden).
  • Payment card CV questions (Page 37):
    • Respondents select from a range of bids; capture maximum willingness to pay; advantages include easier recall but potential range biases.
  • Close-ended CV questions and bidding formats (Pages 38–39):
    • Take-it-or-leave-it bids; large sample sizes needed; estimates depend on bid range and starting point.
  • Example: bidding game for a new BMD test (Page 40):
    • Step 1: Describe the test.
    • Step 2: Use a bidding sequence: $5 → $7 → $9; if no, $3 → $1.
  • Principles of stated preference for WTP (Page 41):
    • Make scenarios resemble market context; be cautious with zero/high bids; choose bid format and payment vehicle; minimize cognitive burden; careful scenario development.
  • Limitations and biases in WTP/WTA (Page 42–45):
    • WTA is cognitively harder; few examples of WTA use in health care; theory suggests WTA ≥ WTP due to income effects; scope insensitivity; strategic bias; compliance bias; WTP limited by ability to pay (ATP) and distributional considerations.

Part 12: Income, Ability to Pay, and Distributional Considerations (Pages 44–47)

  • Relative impact of income on WTP/WTA (Page 44):
    • Two individuals with identical welfare gains may state different WTP due to differing wealth and diminishing marginal utility of income.
    • Higher WTP by wealthier individuals may reflect wealth differences, not greater valuation of the benefit.
  • Marginal utility of income (Page 45):
    • The marginal utility of each additional unit of income declines as wealth increases: U = f(Y).
  • Influence of ATP (Page 46):
    • WTP is shaped by ability to pay; consider distribution of preferences across income groups; potential need for distributional weights or reflecting society’s preferences about equity.

Part 13: Using CBA in Decision Making (Pages 47–52)

  • Core decision rule in CBA (Page 48):
    • Express all benefits in monetary terms to obtain Present Value of Net Benefits (NSB).
    • NSB = Present Value of Benefits − Present Value of Costs; choose the program with NSB > 0 and the highest NSB.
    • PV concept: discount future benefits/costs to present value using discount rate r and time horizon t.
  • Formula intuition (Page 48):
    • Net Societal Benefit (NSB) can be written as:
    • NSB=extPV(extBenefits)−extPV(extCosts)=extBenefits<em>t(1+r)t−extCosts</em>t(1+r)textsummedovertNSB = ext{PV}( ext{Benefits}) - ext{PV}( ext{Costs}) = \frac{ ext{Benefits}<em>t}{(1+r)^t} - \frac{ ext{Costs}</em>t}{(1+r)^t} ext{ summed over } t
  • CBA vs CEA/CUA (Pages 49–50):
    • Type of efficiency: CBA emphasizes allocative efficiency (value-for-money across society); CEA/CUA emphasize technical efficiency (how best to design/implement within health care).
    • Scope: CBA is broader (societal perspective) and can inform decisions across sectors; CEA/CUA are often sector-specific (health care).
    • Application: CBA more common in transport/environment; CEA/CUA more common in health care.
  • Central notion and underpinning theory (Page 50):
    • CBA is based on welfare economics; CEA/CUA are based on a decision-maker philosophy (not necessarily welfare economics).
  • Usefulness and examples of CBA (Page 51):
    • Provides a firm theoretical foundation; can answer whether to fund a project; values benefits not always considered in CEA/CUA (productivity gains, process utility, externalities, option value, etc.).
    • Widely used in public policy areas like transport, environment, health-related topics with various interventions.
  • Potential problems with CBA (Page 52):
    • Monetary valuation of life, health, psychological effects is problematic.
    • WTP is influenced by Ability to Pay; scope insensitivity; various biases (strategic, protest responses);
    • Uncertainty needs thorough sensitivity analyses; risk of double counting; challenging to standardize time horizons and discount rates.

Part 14: Ethical and Practical Implications in CBA and MCDA

  • Ethical considerations (Pages 24–25, 52):
    • Welfarism emphasizes aggregate welfare and individual utility; distribution and equity concerns may require weights or alternative criteria.
    • Pareto/Kaldor-Hicks frameworks raise questions about compensation and distributive fairness; practical policy implications depend on feasibility of compensation and political acceptability.
  • Practical implications for decision-makers (Across sections):
    • MCDA provides a structured method to incorporate multiple objectives and equity concerns; useful when health gains are not the sole objective.
    • CBA offers a monetized framework that can compare interventions across sectors; useful for broad societal allocation but requires careful handling of value judgments, equity, and uncertainty.

Part 15: Key Equations and Concepts (Summary of Formulas)

  • Incremental Cost-Effectiveness Ratio (ICER):
    • ICER=riangleextCostriangleextQALYsICER = \frac{ riangle ext{Cost}}{ riangle ext{QALYs}}
  • Decision rules with ICER and threshold (illustrative):
    • If ICER < threshold, intervention is cost-effective; if ICER > threshold, not cost-effective; dominance concepts apply.
  • Cost-Effectiveness Plane (illustrative):
    • Axes: Incremental Cost vs Incremental Effect.
    • Quadrants indicate combinations of higher/lower cost and higher/lower effectiveness; threshold lines help classify cost-effective options.
  • Net Societal Benefit (NSB) in CBA:
    • NSB = ext{PV(Benefits)} - ext{PV(Costs)} = igg( rac{B0}{(1+r)^0} + rac{B1}{(1+r)^1} +
      rac{B2}{(1+r)^2} + \,igg) - igg( rac{C0}{(1+r)^0} + rac{C1}{(1+r)^1} + rac{C2}{(1+r)^2} + \,igg)
    • Alternatively: NSB=extPV(Benefits)−extPV(Costs),NSB = ext{PV(Benefits)} - ext{PV(Costs)}, with r the discount rate and t the time horizon.
  • Aggregate MCDA score (compositional method):
    • Totalextvalue<em>i=∇</em>jw<em>jimess</em>i,jextsummedovercriteriajTotal ext{ value}<em>i = \nabla</em>j w<em>j imes s</em>{i,j} ext{ summed over criteria } j where
    • wjw_j = weight of criterion j;
    • si,js_{i,j} = score of option i on criterion j.
  • Value of a Statistical Life (VSL) and VSL Year (VSLY) (illustrative values):
    • VSL ≈ 5.7imes106extUSD5.7 imes 10^6 ext{ USD} (AUD, 2024 guidance).
    • VSLY ≈ 2.45imes105extUSDperyear2.45 imes 10^5 ext{ USD per year} (AUD, 2024).
  • Willingness to Pay (WTP) and Willingness to Accept (WTA):
    • WTP/WTA concepts used to monetize health benefits; WTP often tied to ability to pay; WTA tends to be higher due to compensation considerations and loss aversion.
  • Human Capital Approach (for health-related benefits):
    • Value of a life = present value of expected future earnings; depends on age, gender, occupation, unemployment status, etc.; debate exists about whose perspective (employee vs employer) and which costs to include (replacement/friction costs).

Part 16: Practical Takeaways for Exam Prep

  • MCDA is a structured approach to decision-making when multiple objectives matter; it explicitly incorporates stakeholder preferences and trade-offs.
  • CBA quantifies all benefits and costs in monetary terms to assess whether a project increases overall social welfare; it requires careful valuation of health and non-health benefits and robust sensitivity analysis.
  • Know the core differences between CBA and CEA/CUA: scope (societal vs sectoral), objectives (allocative vs technical efficiency), and usecases (broad policy vs health program design).
  • Understand sources of bias in WTP/WTA: ability to pay, scope-insensitivity, strategic reporting, and potential distributional distortions; consider distributional weights when equity is a concern.
  • Be comfortable with key formulas: ICER, NSB, and MCDA total value; understand how discounting affects PVBenefits and PVCosts over time.
  • Recognize ethical dimensions: welfare economics foundations (individual utility, consumer sovereignty) and compensation principles (Kaldor-Hicks) that underpin CBA practices.

Appendix: Quick Reference (Key Terms)

  • ICER: Incremental Cost-Effectiveness Ratio
  • QALY: Quality-Adjusted Life Year
  • NSB: Net Social Benefit
  • WTP: Willingness to Pay
  • WTA: Willingness to Accept
  • VSL: Value of a Statistical Life
  • VSLY: Value of a Statistical Life Year
  • CV: Contingent Valuation
  • CA: Conjoint Analysis
  • DCE: Discrete Choice Experiment
  • MAU: Multi-Attribute Utility
  • ATP: Ability to Pay
  • MCDA: Multi-Criteria Decision Analysis
  • ISPOR: International Society for Pharmacoeconomics and Outcomes Research
  • Baltussen & Niessen: Foundational MCDA example in health priority setting
  • Kaldor-Hicks: Compensation principle for potential Pareto improvements
  • Welfarism: Welfare economics foundation for CBA