Bj. Cohen, ASSIGNING VALUES TO INTERMEDIATE HEALTH STATES FOR COST-UTILITY ANALYSIS - THEORY AND PRACTICE, Medical decision making, 16(4), 1996, pp. 376-385
Cost-utility analysis (CUA) was developed to guide the allocation of h
ealth care resources under a budget constraint. As the generally state
d goal of CUA is to maximize aggregate health benefits, the philosophi
cal underpinning of this method is classic utilitarianism. Utilitarian
ism has been criticized as a basis for social choice because of its em
phasis on the net sum of benefits without regard to the distribution o
f benefits. For example, it has been argued that absolute priority sho
uld be given to the worst off when making social choices affecting bas
ic needs. Application of classic utilitarianism requires use of streng
th-of-preference utilities, assessed under conditions of certainty, to
assign quality-adjustment factors to intermediate health states. The
two methods commonly used to measure strength-of-preference utility, c
ategorical scaling and time tradeoff, produce rankings that systematic
ally give priority to those who are better off. Alternatively, von Neu
mann-Morgenstern utilities, assessed under conditions of uncertainty,
could be used to assign values to intermediate health states. The theo
retical basis for this would be Harsanyi's proposal that social choice
be made under the hypothetical assumption that one had an equal chanc
e of being anyone in society. If this proposal is accepted, as well as
the expected-utility axioms applied to both individual choice and soc
ial choice, the preferred societal arrangement is that with the highes
t expected von Neumann-Morgenstern utility. In the presence of risk av
ersion, this will give some priority to the worst-off relative to clas
sic utilitarianism. Another approach is to raise the values obtained b
y time-tradeoff assessments to a power a between 0 and 1. This would e
xplicitly give priority to the worst off, with the degree of priority
increasing as a decreases. Results could be presented over a range of
a. The results of CUA would then provide useful information to those h
olding a range of philosophical points of view.