|
Utilities to measure
health states are used in medical decision analyses and in cost-effectiveness
analyses as described below.
Clinical Application
of Utility Assessment
Utility assessment
may be integrated into medical decision making with individual
patients. For a description of applications of utility assessment
"at the bedside," please see "Applying Utility
Assessment at the Bedside" in Decision Making in Health
Care: Theory, Psychology, and Applications (Goldstein
and Tsevat, 2000).
A detailed discussion
of patient preference assessment may be found in (Stiggelbout,
2000). Stiggelbout provides a detailed discussion of issues
in choice of assessment method. She also discusses scaling problems
that arise when comparing utilities elicited on a scale with "optimal
health" or "perfect health" as the top of the scale
(utility = 1) with utilities elicited on a scale with "absence
of a particular disease" as the top of the scale. The chapter
includes an approach to adjusting utilities to take account of
scaling differences.
For an introduction
to using decision models and utility assessment in genetic analysis,
see (McConnell
and Goldstein, 1999).
Utilities in Cost-effectiveness
Analysis
A cost-effectiveness
analysis calculates the ratio of additional net costs of a health
care intervention to additional effectiveness (benefits) associated
with the intervention compared with the next-best alternative
(Weinstein
and Stason, 1977). The health effects may be measured in life-years
alone, or may include the quality of life in each year as a weighting
factor, yielding quality-adjusted life years (QALYs).
Note that some authors refer
to a CEA that uses QALYs as an outcome measure as a "cost-utility analysis."
A quality-weighting factor (utility rating) of 1
indicates that a health state is equivalent to full health, while
a quality-weighting factor of 0 indicates that a health state
is equivalent to being dead. The QALYs associated with an intervention
are estimated as the sum of the future expected life years weighted
by the quality of life (expected utility) in each time interval.
An intervention can increase the number of QALYs by changing the
quality weighting (utility) even if it has no effect or a negative
effect on survival; an intervention that improves symptoms can
increase the expected utility.
The incremental cost-effectiveness
ratio is calculated as the ratio between the incremental difference
in costs associated with two alternative treatments to the incremental
difference in QALYs associated with the alternatives. This ratio
is defined only if the more expensive intervention is also more
effective, since otherwise one choice would dominate the other.
The challenge of incorporating
quality of life effects into CEAs arises from the difficulty in
measuring the utility associated with the health states. The results
of CEAs can be highly sensitive to the methods used to calculate
utility.
The estimation of the
quality weight for a given time period and treatment requires
successfully completing two tasks:
- Measuring the impact
of the intervention on the distribution of health states, which
requires completely characterizing the health states that are
influenced by the treatment, and
- Assessing the preferences
(utilities) for these alternative states of health.
The two tasks are related
but logically distinct. For example, specialized geriatric care
may make the health state of dependency in an activity of daily
living (ADL), such as incontinence or difficulty in transferring,
less likely after an illness. Health status measurement quantifies
the severity of the ADL impairment after an illness. Utility assessment
quantifies the degree of desirability with which individuals view
life with these ADL impairments. Some people view them as bothersome
inconveniences, while others see them as devastating. Measuring
these differences in valuation of the health states is the task
of preference assessment.
For a discussion of
variability in responses by method of assessment, see Hornberger
et al. (Hornberger,
Redelmeier et al., 1992).
|