Discounting in cost-utility analysis of healthcare interventions: reassessing current practice.

Cost-utility analysis (CUA) is a technique that can potentially be used as a guide to allocating healthcare resources so as to obtain the maximum health benefits possible under a given budget constraint. However, it is not clear that current practice captures societal preferences regarding health benefits. In analyses of healthcare interventions providing survival benefits, the market rate of interest is the sole empirical variable that reflects societal preferences. This approach is based on the assumptions that: (i) healthcare interventions should be ranked using cost-effectiveness (CE) ratios; (ii) the discount rate for costs in CUA should be equal to that used in cost-benefit analysis (CBA); (iii) the discount rate in CBA should be the market rate of interest on long-term government bonds; and (iv) the Keeler-Cretin paradox is applicable to CUA of healthcare interventions, so that the discount rate for benefits in CUA should be set equal to the discount rate for costs. This approach ignores a fundamental difference between CBA and CUA, namely that CUA assumes that a budget constraint has been specified prior to the analysis. It starts with the assumption that a given amount of funds have been withdrawn from the economy to fund healthcare, so there is no opportunity cost to consider. For that reason, the principles on which the choice of discount rate rests differ in the two techniques. Furthermore, use of CE ratios to rank interventions assumes that the budget constraint can be expressed as a single constraint. But healthcare budgets are multiyear budgets that are roughly constant from year to year. A more realistic model would involve multiple constraints and would require linear programming for solution. This can be reduced to a series of single constraints, thereby allowing use of the simpler CE ratio approach, if we assume that the budget being allocated is intended for one cohort at a time, i.e. all people for whom a new funding decision must be made in a given year. In general, we assume that future cohorts will be allotted comparable funding. However, the Keeler-Cretin paradox depends on the assumption that cohorts are competing with each other for resources, and is therefore not applicable to CUA of healthcare. Other approaches are therefore needed to assign utilities to healthcare interventions providing survival benefits. Methods should be developed that allow analyses to reflect a range of philosophical approaches through sensitivity analysis.