|
In utility elicitation
with SG, the respondent is asked to compare a known, intermediate
state to a gamble, and to choose which of the two is preferable.
The gamble has one
of two outcomes:
- The
best health state under consideration (often "perfect health"),
with probability p, or
- The
worst health state under consideration (often "dead"),
with probability 1-p.
In decision models,
these outcomes can be described as any outcomes suitable to the
question. In health utility assessments, these outcomes are conventionally
set as perfect health versus immediate and painless death.
The probability in
the gamble is varied until the decision maker is indifferent between
the gamble and the intermediate health state: the probability
at which the decision-maker is indifferent is the utility of the
health state (Sox, Blatt et al., 1988). For example, if the subject
is willing to take up to a 20% risk of death in exchange for an
80% chance of perfect health rather than accept the intermediate
state for certain, then the utility of the intermediate state
is 80% or 0.8.
The figure
below illustrates the standard reference gamble for assessing
the utility for a lifetime of back pain. The left side of the
figure shows the "sure thing" - lifetime of back pain
- for which the utility is to be determined. The right side shows
the gamble involved in undergoing a risky procedure with a risk
of perioperative death. The probability, shown as p in the figure,
is varied until the respondent is indifferent between the "sure
thing" and the risky procedure.
Figure
7-5 Standard Reference Gamble for Assessment of the
Utility for a Lifetime of Back Pain
|
|---|
 |
|
The
SG is the rating method best grounded in VNM expected utility
theory. However, the SG can be difficult to understand, because
it requires mental comparison of a probabilistic outcome with
a certain outcome.
The TTO is a direct
question about how much life expectancy an individual would trade
off to improve quality of life. It attempts to present the respondent
with a task that some believe is simpler than the SG task, while
preserving an element of trade-off in the assessment (Torrance,
1982). The TTO utility equals one minus the maximum proportion
of time that the subject is willing to trade off. For example,
if she is willing to give up up to 1/10th of her life expectancy
in return for perfect health, her utility for the intermediate
health state is 1-0.1, or 0.9.
Health State Classification Systems
Three common health
state classification systems are characterized in the table below.
Similarities and difference in these methods are detailed in the
discussion that follows.
Additional information
about these systems is available at the following website: http://www.medicine.ucsd.edu/fpm/hoap/instruments.html.
Relationship of
Health Status and Utility
The relationship between health status and utilities derived from
health state classification systems is by definition strong: Utilities
are assigned based on the health state.
In contrast, numerous
studies have demonstrated that the correlation between one's current
state of health and one's TTO or SG utility for that health state
is at best modest (Tsevat, 2000). For example, for two people
who are blind, one may consider blindness to be a severe problem
associated with a low utility, while the other may consider that
blindness is not such a severe problem, with a higher utility.
For individual decision-making,
since health status does not correlate well with utility, a health
state classification system may not be optimal, and it may be
preferable to use directly assessed utilities.
|