ADJUSTING
HEALTH STATE UTILITIES FOR USE IN ECONOMIC EVALUATION
Quality of Life
Newsletter 23/1999.
phone: 47 22 04 23
42, fax: 47 22 04 25 95, e-mail: erik.nord@folkehelsa.no
Summary
Health state values from multi-attribute utility instruments such as the
EQ-5D and the HUI need to be transformed before they can be used to estimate
the societal value of different health interventions. Without transformation
the values lead to a very strong overestimation of the value of interventions
for moderate conditions relative to interventions for severe or fatal
conditions. A rough transformation function is offered for a number of MAU
instruments. For a different solution, see paper on cost-value analysis (click).
Cost-utility analysis (CUA)
is a widely used technique for judging whether health technologies and programs
give adequate value for money. In CUA, the value of health care is measured in
terms of the degree to which the care is appreciated personally by the
individuals concerned. This concept of value is operationalised as a product ot
two factors: The increase in quality of life – often referred to as utility
- that follows from an intervention, and the number of years a person gets to
enjoy this increase. The unit of measurement of value in this approach is the
quality adjusted life year (QALY). The underlying idea of the approach is that
medical technologies with low costs per QALY gained should be given priority
over those with high costs per QALY gained (1).
The Canadian Coordinating
Office for Health Technology Assessment recommends cost-utility analysis as one
of two preferred analytical techniques for economic evaluation of
pharmaceuticals (2). The other recommended technique is cost-benefit analysis.
Similar recommendations, although not as strong, have been published by the
Pharmaceutical Benefits Pricing Authority in Australia (3).
A number of so called
multi-attribute utility (MAU) instruments are available for assigning quality
of life scores (utilities) to health states (for a review, see 4). CCOHTA
recommends three of these as particularly suitable for use in QALY
calculations: The Quality of Well-Being Scale (5), the Health Utilities Index
(6) and the EQ 5-D (7).
In spite of the endorsement
of these instruments by CCOHTA, unclarity prevails regarding the meaning of
numbers from MAU-instruments (8). The purpose of this paper is to address a
problem of validity in using such instruments to estimate the societal value of
health programs, and to offer a simple tool that to some degree may remedy this
problem.
The problem
The idea of giving priority
to health care programs according to their cost per QALY is based on the assumption
that society’s appreciation of a health care program is proportional to the sum
of QALY gains, regardless of how these gains are distributed across individuals
in the program. Given this assumption, utilities for health states have
implications for resource allocation decisions in terms of equivalence of
numbers of people treated. For example, saving the life of one healthy person
will be equivalent to curing two people with utility 0.5 and curing ten people
with utility 0.9 (assuming equal duration of benefit). Such equivalence
judgements are often referred to as ’person trade-offs’.
CCOHTA notes that society
may have concerns for fairness that run counter to simple QALY maximisation. To
capture such concerns, it may, according to CCOHTA, be of interest ’in some
studies, for some decision makers to explore the impact of using direct person
trade-off questions … to establish society’s valuation of different health
programs relative to each other’. CCOHTA points out that ’if the approach gives
dramatically different answers to the resource allocation question versus QALY
methods, then a discussion of the reasons could be quite enlightening for the
decision makers’ (2, p.28).
A review of direct person
trade-off preference studies in Australia, England, Norway, Spain and the US
(9) indicated a strong concern for giving priority to the worst off: Life
saving procedures were valued considerably higher than treatments even for
severe non fatal conditions and treatments for severe conditions were valued many
time higher than treatments for moderate conditions. Later studies support
these findings (10-13). If health state values are to be consistent with this
structure of concern, scores for severe problems need to be much higher than
zero, and scores for moderate problems need to be compressed to the upper end
of the 0-1 scale. Table 1, reproduced from (9), uses three example levels of
severity to indicate ’rules of thumb’ for scoring health states in accordance
with the above person trade-off evidence (line 1: ’societal values’). The table
further shows what scores the three example states would roughly obtain if
mapped into and scored by various existing MAU-instruments. (Documentation
released recently on the scoring function of the latest version of the Health
Utilities Index (HUI III) suggests that this instrument yields utilities much
similar to those of HUI II (14)).The general picture is that existing
MAU-instruments lack the compression of states to the upper end of the scale
that is required to encapsulate societal concerns for the worst off.
Table 2 illustrates the
problem in the case of the three MAU-instruments that are recommended by the
CCOHTA. According to the societal rules of thumb, an intervention that prevents
death and instead leaves a person with a considerable problem is valued in the
order of ninety times more than an intervention that eliminates a moderate
problem (assuming the same duration of the benefit). The former intervention
thus justifies in the order of ninety times higher costs per person helped than
the latter. By contrast, according to the QWB, HUI-II and the EQ-5D, the former
intervention would be less cost-effective than the latter if the cost-ratio
exceeds not 90:1, but only 3:1, 9:1 and 2:1 respectively. Similarly, an intervention
that reduces a considerable problem to a moderate problem actually justifies
seven times higher costs than the cure of a moderate problem. According to the
QWB, the HUI-II and the EQ-5D the maximum cost-ratios are 0.5:1, 3:1 and 0.3:1
respectively. The discrepancies are huge.
A suggestion for
economic evaluation of pharmaceuticals
Direct person trade-off
data are still too scarce to allow precise estimates of societal values for
health states for use in resource allocation decisions. On the other hand, the
evidence does indicate roughly in what parts of the 0-1 scale health states
need to be located if they are to be consistent with societal preferences in
such decisions. It seems, therefore, that analysts who wish to use
MAU-instruments in economic evaluations of health programs and technologies may
already at this stage improve their performance by conducting two
analyses: One being a conventional cost-utility study, in which the utilities
from generic instruments are used as they stand, and the other being a study in
which the utilities are transformed into numbers that also encapsulate concerns
for severity. The term ’cost-value analysis’ has been suggested for the latter,
broader approach (15).
Figure 1 is offered as a
simple tool to help conduct the required transformations. The figure uses the
utilities in table 1 and the middle numbers in each of the intervals in the
societal values in the first line of the table. The figure indicates the
functional relationship between utilities and societal value numbers for each
of the MAU- instruments in the table.
Example
Assume that a choice is to
be made between a health program that will cure 100 people of a given condition
A and an equally costly program that will take 30 people from a condition B to
a functional level corresponding to condition A. Assume that life expectancy is
20 years for patients in both programs, and that conditions A and B are
assigned utilities 0.8 and 0.4 respectively if the Health Utilities Index (Mark
II) is used. The former program then yields 100x(1-0.8)x20 = 400 QALYs
(undiscounted), while the latter yields 30x(0.8-0.4)x20 = 240 QALYs. A
cost-utility analysis based on HUI-II thus suggests that the former program
should have priority. However, according to figure 1, HUI utilities of 0.4 and
0.8 correspond roughly to societal values of 0.75 and 0.96. These numbers
encapsulate societal preferences for severity per se. Using these numbers
instead of the simple utilities changes the value score of the former program
to 100x(1-0.96)x20 = 80 and the value score of the latter program to
30x(0.96-0.75)x20 =
Concluding remarks
Clearly figure 1 is a very
rough tool. Considerably more data are needed to estimate the transformation
functions more precisely and for a wider range of the 0-1 value scale. Figure 1
could nonetheless presumably be useful, inasmuch as it is better to try to be
roughly right rather than precisely and systematically wrong when estimating
societal value. By doing a study based on transformed numbers as an add-on to a
conventional cost-utility study, one would be complying with the suggestion
made by CCOHTA, namely to ’explore the impact of using direct person trade-off
questions to establish society’s valuation of different health programs
relative to each other’ (2). If the two analyses give different answers, ’then
a discussion of the reasons could be quite enlightening for the decision
makers’ (2).
REFERENCES
1. Weinstein MC, Stason WB. Foundations of cost-effectiveness
analysis for health analysis and medical practice. New England Journal ofr
Medicine 1977,296,716-721.
2. Canadian Coordinating
Office for Health Technology Assessment. Guidelines for economic evaluation of
pharmaceuticals: Canada. 2nd ed. Ottawa: CCOHTA 1997.
3.Langley PC. The November
1995 revised Australian guidelines for the economic evaluation of
pharmaceuticals. Pharmacoeconomics 1996, 341-352.
4.Nord E. A review of
synthetic health indicators. Background paper for the OECD Directorate for
Education, Employment, Labour and Social affairs, June 1997.
5.Kaplan RM, Anderson JP. A
general health model: Update and applications. Health Services Research,
1988,23,203-235.
6. Feeny D, Furlong W,
Torrance G. The Health Utilities Index: An Update. Quality of Life Newsletter
no 22/1999.
7. De Charro F, Rabin R.
EQ-5D from the EuroQol Group: An Update. Quality of Life Newsletter no 22/1999.
8. Nord E, Wolfson M.
Multi-attribute health state valuations: Ambiguities in meaning. Quality of
Life Newsletter no 21/1999.
9.Nord E. Health status
index models for use in resource allocation decisions. A critical review in the
light of observed preferences for social choice. International Journal of
Technology Assessment in Health Care 1996, 12,31-44.
10.Pinto Prades, José-Luis.
Is the person trade-off a valid method for allocating health care resources?
Health Economics 1997,6, 71-81.
11.Richardson J. Critique
and some recent contributions to the theory of cost utility analysis. Working
paper no 77. Melbourne: Centre for Health Program Evaluation. 1997
12.Ubel PA, Kamlet M,
Scanlon D, Loewenstein G. Individual utilities are inconsistent with rationing
choices: A partial explanation of why Oregon’s cost-effectiveness list failed.
Medical Decision Making 1996,16,108-119.
13.Ubel P. How stable are
people’s preferences for giving priority to severely ill patients? Mimeo.
Philadelphia: Veterans Affairs Medical Center 1997.
14. Furlong W, Feeny D,
Torrance G et al. Multiplicative multi-attribute utility function for the
Health Utilities Index Mark 3 (HUI3) system: A technical report. Paper 98-11.
Hamilton: CHEPA 1998.
15. Nord E, Pinto JL,
Richardson J, Menzel P, Ubel P. Incorporating societal concerns for fairness in
numerical valuations of health programs. Health Economics 1999,8,25-39.
Table 1. Societal values
for health states versus individual utilities from MAU-instruments.
|
Instrument |
Problem level (a) |
|
|
Severe |
Considerable |
Moderate |
|
Societal values |
.65-.85 |
.90-.94 |
.98-.995 |
|
QWB |
.45-.55 |
.65-.70 |
< .80 |
|
HUI 1 |
.10-.20 |
.30-.40 |
< .85 |
|
HUI 2 |
.40 |
.70 |
.90-.94 |
|
EQ-5D |
.20 |
.60 |
.70 |
|
York EuroQol (TTO) |
.20-.25 |
.40-.50 |
.80 |
|
IHQL (3D) |
.50-.70 |
.75-.85 |
.89-.93 |
|
IHQL (complex) |
.70-.75 |
.80-.90 |
.90-.94 |
|
15 D |
.77 |
.86 |
.91-.93 |
|
Rosser/Kind |
.68 |
.94 |
.97-.98 |
(Source: 9)
- The three states were described as
follows:
Severe: Sits in a wheel-chair,
has pain most of the time, is unable to work.
Considerable: Uses crutches for walking, has light pain
intermittently, is unable to work.
Moderate: Has difficulties
in moving about outdoors and has slight discomfort, but is able to do some work
and has only minor difficulties at home.
Table 2. Values of health
improvements according to societal rules of thumb and utilities from three
MAU-instruments.
|
Improvement (better state vs initial
state: |
Societal value |
QWB |
HUI 2 |
EQ-5D |
|
Considerable problem vs
’as bad as dead’ |
0.92 |
0.68 |
0.70 |
0.60 |
|
Moderate vs considerable problem |
0.07 |
0.10 |
0.22 |
0.10 |
|
Healthy vs moderate problem |
0.01 |
0.22 |
0.08 |
0.30 |
