Societal Preferences for Interventions with the Same Efficiency: Assessment and Application to Decision Making.

BACKGROUND AND OBJECTIVES

Although quality-adjusted life-years (QALYs) may not completely reflect the value of a healthcare technology, it remains unclear how to adjust the cost per QALY threshold. First, the present study compares two survey methods of measuring people's preferences for a specific healthcare technology when each choice has the same efficiency. The second objective was to consider how this information regarding preferences could be used in decision making.

METHODS

We conducted single-attribute (budget allocation) and multi-attribute (discrete-choice) experiments to survey public medical care preferences. Approximately 1000 respondents were sampled for each experiment. Six questions were prepared to address the attributes included in the study: (a) age; (b) objective of care; (c) disease severity; (d) prior medical care; (e) cause of disease; and (f) disease frequency. For the discrete-choice experiment (a) age, (b) objective of care, (c) disease severity, and (d) prior medical care were orthogonally combined. All assumed medical care had the same costs and incremental cost-effectiveness ratio (ICER; cost per life-year or QALY). We also calculated the preference-adjusted threshold (PAT) to reflect people's preferences in a threshold range.

RESULTS

The results of both experiments revealed similar preferences: intervention for younger patients was strongly preferred, followed by interventions for treatment and severe disease states being preferred, despite the same cost per life-year or QALY. The single-attribute experiment revealed that many people prefer an option in which resources are equally allocated between two interventions. Marginal PATs were calculated for age, objective of care, disease severity, and prior medical care.

CONCLUSION

The single- and multi-attribute experiments revealed similar preferences. PAT can reflect people's preferences within the decision-maker's threshold range in a numerical manner.