Health plan coverage and reimbursement decisions for new medical technologies typically include efficiency (cost-effectiveness) criteria. Alongside cost-effectiveness, other criteria are sometimes considered, related to either clinical (e.g., burden of disease, benefit-risk impact) or broader societal (e.g. productivity, equity) effects. However, methods to aggregate and measure the trade-offs among these different criteria are diverse. The aim of this study is guided by recommendations of the ISPOR Special Task Force on US Value Assessment Frameworks. We compare the resource allocation implications from an augmented cost-effectiveness analysis (ACEA), which includes novel elements of value under the net monetary benefit (NMB) metric, with the use of multi-criteria decision analysis (MCDA).

We explore the idea of what could be called a “QALY-anchored” MCDA and compared with a form of AECA where elements of value are either classified as health or consumption attributes—and ultimately monetised. The model considers the role of health insurance through its smoothing effect on consumption. Trade-offs in ACEA are derived as willingness to pay for a QALY from the expected utility model. In MCDA, value trade-offs between consumption and health are derived from a multi-attribute utility model which includes interactions between health and consumption.

We find that trade-offs change with income, medical spending, severity of disease, and risk attitudes. Changes in income and medical spending affect these trade-offs similarly in ACEA and MCDA, with increased trade-off with income, and decreased trade-off with medical spending. Severity of disease has a positive effect on the ACEA willingness to pay but an ambiguous effect on the MCDA value trade-off. For multiplicative MCDA models, the consideration of complementary benefits between health and consumption creates a positive interaction term.

Trade-offs derived from an ACEA and a QALY-anchored MCDA model are similarly affected by budget changes. Therefore, these are aligned and would result in similar decisions