Basic issues in valuing health outcomes in terms of QALYs

Issues in valuing health
outcomes in terms of QALYs
Group B:
Norman Daniels, Mark Kamlet, Alistair
McGuire, Erik Nord, George Torrance,
Milton Weinstein
The QALY
• Metric for measuring or estimating the
value of health scenarios or outcomes.
• Equals the value of a healthy life year or
an outcome equivalent to that.
Ambition
• To value all health outcomes on a common
scale, whatever the type of health problem
involved and whatever the effect on length
of life.
• Aid decision making. Does not set aside
the need for fair, democratic procedure.
The standard QALY procedure
.. consists in:
(1) weighting life years according to their quality
(healthwise) and proximity in time,
(2) valuing a multi-year life scenario as the sum of the
weighted life years in the scenario and
(3) estimating the value of one scenario relative
to another as the difference between the
sums of weighted life years.
Each of these steps raise methodological questions. The
most salient ones, which we first address, are:
Most salient
methodological questions
Re step 1 in the QALY procedure:
What are the most appropriate methods for
weighting for quality of life?
Should there be additional weighting for
severity?
Most salient
methodological questions (ctd)
Re step 2 in the QALY procedure:
Is the value of a multi-year scenario simply the sum of
the weighted years in the scenario?
HYEs question this. But also:
Is the value of two years really twice that of one year
(even if the quantity is twice as large)?
Most salient
methodological questions (ctd)
Re step 3 in the QALY procedure:
Is the value of a movement from one scenario to another simply the
difference between the two scenarios valued independently?
Most important: The assumption implies that saving the life of people with
chronic illness or disability carries less value than saving the life of
otherwise healthy people.
But also:
Does SG-disutility(A) = 2 x SG-disutility(B) necessarily imply
utility(A=>B) = ½ x utility(A=>Healthy)?
(Could change of viewpoint/expectations mean that the second equation
(viewpoint of the ill) does not necessarily follow from the first (viewpoint of
the healthy)?)
Another important question
• The discount rate for distance in time
Less pressing questions
• Age weighting
• Altruism
The answers
to the methodological questions
depend on which questions
QALYs are meant to address
Main purposes of QALYs
• Aid individual decision making
• Aid societal resource allocation
Questions in
individual decision making:
• Which treatment should I choose?
• How should I prioritise in my insurance
package?
Questions in societal resource
allocation
•
How much value does the average person assign personally to avoiding different
types of health losses in the future (quality of life and life years)?
Use: Valuing health programs ex ante in terms of ’the sum of self interest’.
•
How much do people with different conditions actually suffer, and how much value do
they place ex post on different treatments?
Use: Judging ’value for money’ in ongoing activities/programs (’societal
audit’).
•
When thinking about both efficiency and fairness, how highly does the general public
value programs for different categories of patients? (Note: The point is inclusion of
fairness. PTO is not necessarily only about ’others’.)
Use: Valuing health programs ex ante in terms of population
preferences for priority setting.
The various questions call for QALYs based on
different concepts of value, different respondents and
different health state valuation techniques
Which treatment should I choose?
Ex ante desirability, community
members, SG/TTO/RS
How should I prioritize in my insurance
package?
Same
Aggregate personal valuation of
avoiding different health losses in the
future.
Same
Evaluating ongoing activities: What is
actual suffering and what are actual
effects of treatments?
Experienced utility, patients/disabled
people, SG/TTO/RS
How much do people value programs
for different groups, given concerns for
both efficiency and fairness?
Overall societal value, community
sample, PTO
Dependency
between valuations
Experienced utility
Ex ante desirability
Overall societal value
(Judgments of ex ante desirability should be informed by
experienced utility, and overall societal value judgements
should be informed by both the former.)
• The two previous slides may serve as our first
proposals for a consensus in Philadelphia.
• We may now address the methodological
questions raised in slides 5-7 in more detail
within the context of each of the three concepts
of value: Ex ante desirability, experienced utility
and overall societal value. We first focus on
societal resource allocation (rather than
individual decision making).
Methodological questions
in societal resource allocation (slides 5-7):
Summary of what reflection and evidence suggests.
Value type: Aggregate
Experienced
personal ex ante utility
Question:
Overall
societal value
Appropriate health
state valuation
technique?
SG, TTO
RS (due to
insensitivity in
SG and TTO)
PTO
Severity weights
needed?
No
No
Covered by
PTO
Value proportional
to duration?
Probably not
(fresh evidence)
Uncertain
No
Value of life saving Clearly not
proportional to
value of end state?
Clearly not
Clearly not
Value proportional
to ’size’ of health
gain?
Perhaps not
No
Perhaps not
(fresh evidence)
A closer look at weighting
techniques
• SG, TTO, RS, PTO
Multi-attribute utility instruments
• Which are they?
• Variation in values for the same states.
• What to do about that.
A closer look at assumptions about
the value of changes in health
Effect and value (1)
• Societal decisions about resource allocation
1
A
B
0
•
Daniels and Nord, Amsterdam 1992: If the arrows represent the maximum
obtainable (potentials) in the two groups, then V(B)~V(A).
Effect and value (2)
• Gained years
• A
• B
20
10
• V(A) probably not = 2 x V(B), even disregarding
discounting for distance in time. E.g. Dolan and
Cookson, 1998.
Effect and value (3)
• Life saving
1
0
A
B
V(B) = V(A) both from societal and individual view point
Effect and value (4)
• Health improvements
1
0
A
B
C
D
If V(B) = V(A) also in individual utility (U), perhaps U(C)
not much greater than U(D)
Effect and value (5)
• Valuing improvements X and Y vs judging the disutility of conditions
A and B from the viewpoint of the healthy
QALYs: State A vs B
Actual question of interest: V(X) vs V(Y)
• 1
B
X
Y
A
• 0
•
In QALYs, the result from A vs B is used to value X vs Y. But the viewpoints are
different. So even if V(A)=0.6 and V(B)= 0.8 given healthy, V(X) may be less than 2
x V(Y) given state A.
Effect and value (6)
• Implications
• To say that the value of saving life should count
as 1 ’regardless’ can lead to inconsistency:
V(0 =>A) + V(A =>Healthy) > V(0 => Healthy)
But desire for mathematical consistency should not
trump the need to model the real world correctly.
Effect and value (7)
•
Proposed solution in cost value analysis: Transform utilities. Difference between
vertical arrows < difference between horisontal arrows. Note: No inconsistency.
•
Values for valuing change
• 1
• 0
1 Utilities from the viewpoint of healthy
A
B
Effect and value (8):
Useful to distinguish between
the following viewpoints
1. Ex ante to illness: Preferences for insurance
2. Ex ante to treatment: Strength of desire for
treatment
3. Ex post to treatment: Satisfaction/increase in
well being.
In (1), it is possible that feelings in (2) and (3) will
be anticipated.
Other issues
•
•
•
•
Discounting for distance in time
Age weighting
Altruism
Negative health states, maximal endurable
time
Individual decision making
• Shall we spend time on this?
Some tentative conclusions so far
• Experienced utilities are needed as inputs
in all salient types of valuation.
Conclusions ctd
• In the model for aggregate ex ante self interest
valuation (standard QALYs) of gains in health,
the importance conventionally assigned to the
disutility of health states and ’effect size’ needs
to be reconsidered.
• Most importantly, the valuation of life saving
needs to be separated from the valuation of
health states.
Conclusions ctd
• Also, the problem of value variation across
MAUs needs to be adressed.
Conclusions ctd
• The three main types of societal economic evaluation
should have different names: Cost-utility ex post, costutility ex ante and cost value (concerns for fairness is not
utility).
• All three types of evaluation can provide useful
information for decision makers.
• Since the three types of evaluation may yield quite
different results, any cost-per-QALY ’league table’ must
be specific to one type of evaluation.
• Journals should require of papers that the type of
evaluation be specified and that comparisons be made
within types only.
Final remark
• Speculation: Based on similarities in
values obtained in experienced utilities,
some ex ante standard gambles and PTO
suggest a structure of values with upper
end compression that could serve as a
rough guide in the field?