—TABLE 3—
Depression Health State Utility Scores
(Standard Gamble)
Utility Score
95%
Confidence
Interval
Health State Mean
Temporary states—Six-month duration
Mild depression 0.59
Moderate depression 0.32
Severe depression 0.09
Chronic states—Lifetime duration
Self-reported health state 0.79 0.74–0.83
Severe depression 0.04 0.01–0.07
Source: Bennett, Torrance, Boyle and Guscott, Cost-Utility Analysis
in Depression, Psychiatric Services, Sept. 2000, Vol. 51, No. 9, pp.
1171–1176.
0.55–0.62
0.29–0.34
0.05–0.13
—TABLE 4—
Benefit-Cost Table of Health Initiatives
QALYs
Saved
Cost
(in dollars)
Cost per QALY
(in dollars)
100
1,800,000 18,000
100 5,000,000 50,000
500 1,000,000 2,000
100 2,200,000 22,000
100 1,200,000 12,000
500 2,000,000 4,000
100 10,000,000 100,000
200 1,200,000 6,000
150 4,500,000 30,000
50 800,000 16,000
250 2,000,000 8,000
Program
A
B
C
D
E
F
G
H
I
J
K
Choosing a Weight
In general, the meanings that people attach to the terms
preference and utility depend on context, a practice that’s prone to
confusion. For this reason, economists insist on precise definitions that allow for the calculation of utils, the generic unit.
The type of utility favored in economic analyses, called
cardinal utility, is such that the magnitude of differences in
preferences is significant. Think about temperature measurements where 0 degrees corresponds to freezing water and 100
degrees to boiling water: It’s correct to assert that the gain in
temperature in going from 10 degrees to 20 degrees is twice
as much as going from 80 degrees to 85 degrees. Analogously, if the survival rates for surgeries A and B are, respectively,
40 percent and 60 percent, it would be correct to state that
the chances of survival are 50 percent higher for B than for A.
The analogy with the QALYs of the standard gamble method is
straightforward: Utilities are expressed as probabilities (hence
they are cardinal), 0.00 represents death, 1.00 perfect health,
and consequently, gains or losses in QALYs are equivalent as
long as their magnitudes are the same.
Preferences derived from the time trade-off and rating scale
methods are not cardinal utilities because the magnitudes of
the respondents’ preferences are significant only in that a higher value is more desirable than a lower one. Their differences
are meaningless. Consider the utility associated with rest. By
the law of diminishing returns, the benefit gained by increasing
sleeping time from four to six hours is greater than the benefit
derived by sleeping eight instead of six hours. That is:
u( 6) – u( 4) > u( 8) – u( 6).
The type of utility in which the magnitude of differences is
meaningless is known as ordinal utility.
Using cardinal utilities as a building block, von Neumann
and Morgenstern developed a theory that allows us to draw
inferences about the choices rational people make. Their model
is regarded as the best approach for decision-making under uncertainty, which explains why the standard gamble technique
is the best methodology to calculate QALY weights. (See Page
52 for a list of the fundamental axioms of classical utility theory
and an informal statement of the expected utility theorem.)
Empirical Studies
Empirical studies supply some insights into people’s
preferences.
First, as a rule,
u(Standard Gamble) > u(Time Trade-Off ) > u(Rating Scale).
Source: Torrance, Socio-Econ. Plan. Sci. 10:129-136; O’Leary et al.,
Med Dec Making 15:132-137; Stiggelbout et al., IJTAHC 12:291:298
