Expanding Quantitative Methods of Value Assessment: Multi-Criteria Decision Analysis
Cost-effectiveness analysis (CEA) and cost-utility analysis (CUA) have become the de facto approaches for assessing the value of healthcare interventions and informing healthcare decision-making. However, they are not without their flaws. For example, standard CUA estimates the incremental cost-per-quality-adjusted life year (QALY) which is prohibited under the Inflation Reduction Act (IRA) as QALYs have been argued to be discriminative against elderly, disabled or terminally ill individuals who have a lower maximum utility and thus reduced capacity to benefit from new interventions.1 Further information on the implications of the IRA can be in a separate commentary found here (‘The IRA is Law – Now What?’). The development of the Generalised Risk-Adjusted Cost-Effectiveness (GRACE) Framework aims to address the purported shortcomings of standard CEA by incorporating the concept of diminishing marginal utility with respect to health into CEA and reflecting varying willingness-to-pay thresholds by disease severity. Our detailed commentary on the GRACE framework can be found in a separate write-up here (‘Analytical Methods & Technical Topics’).2 However, beyond the above criticism, it has been argued that only a fraction of the value an intervention provides can be captured in the incremental cost-per-QALY metric. New methods, such as Extended CEA and Distributional CEA, have been proposed to account for benefits such as financial risk protection and equity within CEA.3, 4 Nevertheless, incorporating novel elements of value like those presented in the ISPOR value flower (e.g. real option value, value of hope) and accounting for disparate preferences across stakeholders (patients, providers, payers, etc.) within the healthcare system in CEA remains challenging.
Multi-Criteria Decision Analysis
Phelps et al. (2019) have proposed using Multi-Criteria Decision Analysis (MCDA) as a supplement to CEA for evaluating healthcare interventions.5 At this year’s ISPOR International, there were multiple sessions involving discussion on the use of MCDA in healthcare decision-making.6, 7 By eliciting the preferences of decision makers, MCDA can theoretically better capture the value associated with each alternative intervention that may be difficult to reflect through CEA. The goal of MCDA is not to replace the role of CEA in healthcare decision-making but to expand the value assessment framework and present stakeholders with additional evidence of value for relevant healthcare interventions together with findings from the CEA. A typical MCDA usually includes the following steps:
A simplified example of MCDA is presented in Table 1.
Table 1. Example of MCDA
| Criterion |
Weight |
Score of Drug A |
Score of Drug B |
| Cost of therapy |
0.5 |
100 |
50 |
| Administration route |
0.3 |
60 |
70 |
| Number of pills |
0.2 |
80 |
80 |
| Total |
1.0 |
84=(100*0.5+60*0.3+80*0.2) |
62=(50*0.5+70*0.3+80*0.2) |
In this example, the results suggest that Drug A is the optimal choice since it has a higher weighted sum compared with Drug B. Note that methods for conducting an MCDA are not limited to the example presented above, and there are many weighting and scoring methods available. This diversity in methods and the user complexities associated with them lead to challenges in the adoption of MCDA. ISPOR’s Emerging Good Practices Task Force has published a report on good practices for conducting MCDA and provided guidance for choosing the appropriate MCDA methods, in an effort to improve MCDA’s usability.8, 9 Overall, MCDA represents a promising methodology to account for additional elements of value that may not be captured by traditional CEA, but we are yet to see an emerging dominant method or use case, despite guidance from ISPOR. With increasing use, the appropriate application of this methodology may become clearer.
