Transcript Evidence-based supported employment for people with severe

Economic evaluation of health
programmes
Department of Epidemiology, Biostatistics
and Occupational Health
Class no. 11: Cost-utility analysis – Part 4
Oct 8, 2008
Plan of class
Finish material from last class
Preference-based generic instruments
Construction of QALYs
Limitations of QALYs
QALYs vs DALYs
Preference-based generic
instruments: Purpose
SG, TTO difficult and costly to use
One would prefer a simpler instrument
Administer to subjects in a study to
evaluate their health-related quality of life
as rated by a community sample
Preference-based generic
instruments: 2 steps in use
QUESTIONNAIRE THAT ASKS AT WHICH
OF 3 TO 6 LEVELS RESPONDENT IS ON 5
TO 15 DIMENSIONS
FORMULA FOR CONVERTING
RESPONSES INTO A HEALTH RELATED
QUALITY OF LIFE WEIGHT FROM 0 TO 1
(VALUE OR UTILITY DEPENDING ON
QUESTIONNAIRE)
Questionnaire design
Obvious differences:
 Number and choice of dimensions
 Number of levels for each dimension
Instrument
Dimensions
Quality of
well-being
Mobility, physical activity, social
functioning
27 symptoms/problems
EQ-5D
Mobility, self-care, usual activities,
pain/dicscomfort, anxiety/depression
SF-6D
HUI2
HUI3
Number of Number
levels per
of
dimension states
3
2
945
3
243
Physical functioning, role limitation, social
functioning, pain, energy, mental health
4–6
18,000
(SF-36)
7,500
(SF-12)
Sensory, mobility, emotion, cognitive, selfcare, pain
Fertility
4–5
24,000
Vision, hearing, speech, ambulation,
dexterity, emotion, cognition, pain
5–6
3
972,000
Note: At least two other questionnaires exist: Australian Quality of Life
(AQoL) and the Finnish 15D. Not as widely used as EQ-5D, SF-6D or
HUI2 or HUI3. Not discussed in class.
Health as a spectrum
Disease or
disorder
Impairment
Ability
Participation
Rheumatism
Pain/
Dexterity
Limits
Walking
Role, social
and usual
activities
Macular
degeneration
Vision
Limits ability
to read
Role, social
and usual
activities
WHO international classification of health into disease or disorder,
impairment, ability and participation (with examples).
(Taken from Brazier et al., Measuring and valuing health benefits for
economic evaluation, Oxford, 2007, Fig 4.1 )
SF-6D, EQ-5D, HUI2
HUI3
Disease or
disorder
Impairment
Ability
Participation
Rheumatism
Pain/
Dexterity
Limits
Walking
Role, social
and usual
activities
Macular
degeneration
Vision
Limits ability
to read
Role, social
and usual
activities
Include dimensions relating
to social participation?
“Within skin” aspects of health: avoid
measuring peoples’ choices/preferences
 “purer” measure of health
 No influence of adaptation
But general population values will
underestimate adaptation
Health is a means, social participation part
of its end – this is what matters to patients
Instrument
Country where preferences Valuation technique
obtained
Quality of
well-being
USA (San Diego)
VAS
EQ-5D
Belgium, Denmark, Finland,
Germany, Japan, The
Netherlands, Slovenia,
Spain, UK, USA, Zimbabwe
SF-6D
Hong Kong, Japan, UK,
Australia, Brazil
SG, ranking
HUI2
Canada (Hamilton), UK
VAS transformed into
SG
HUI3
Canada (Hamilton), France
VAS transformed into
SG
TTO, VAS, ranking
Scoring the questionnaires
 In each case, use a method such as SG, TTO,
VAS to value some states, and interpolate
statistically
 Too many states to value them all individually!
 Two approaches to developing scoring methods:
 Multi-attribute utility theory (MAUT): HUI2,
HUI3
 Statistical estimation without restrictive
assumptions of MAUT: QWB, SF-6D, EQ-5D
Multi-attribute utility theory:
The problem
What happens when lottery outcome
yields something that has several
attributes, such as mobility, emotional
state, etc?
How to combine the utilities of separate
dimensions of outcome to generate an
overall utility of the outcome?
Example
How would you combine utilities derived
from an apartment that you might rent?
Attributes include: (a) price; (b) location;
(c) size; (d) quietness; (e) attractiveness;
(f) other factors.
3 most common methods
based on MAUT
In all cases, need to calculate utility
associated with each dimension
Additive, multiplicative, multilinear (see
formulas in book, p. 157)
The simpler the method, the more
restrictive the assumptions
 e.g., additive implies no interactions
HUI2 and HUI3 use multiplicative formula
Exercises
On the EQ-5D, considering Table 6.4,
what does score 11212 mean? What
health-related quality of life weight does
this represent?
On the HUI3, same questions for
22111223.
Choice of instrument matters
 Different instruments yield different results
 Different ranges: -0.4 to 1.0 for EQ-5D, vs. 0.3 to 1.0
for SF-6D
 EQ-5D scores of 11111 can translate to SF-6D scores
as low as 0.56
 Studies comparing scores across instruments for
same patients find significant differences
• Patient group appears to be a factor
 Differences as small as 0.03 should be considered
significant
Reasons for differences
 Differences in coverage
 Capacity vs functioning
 Symptoms, social health, mental health covered
differently
 Sensitivity of dimensions
 Floor effect for physical and social functioning, role
limitations for SF-6D
 Ceiling effect with EQ-5D
 Valuation methods
 Systematic differences depending on method
Choosing a method:
Practicality
Instrument
Comments on practicality – self-administered
Quality of well-being
Somewhat more complex to fill out
EQ-5D
Easiest to administer, very simple
SF-6D
Usually derived from responses to SF-36 or SF-12
which are longer, lower completion rates
HUI2
Easy to administer, license costs
HUI3
Easy to administer, license costs
Choosing a method:
Reliability
All have acceptable test-retest reliability
Different responses depending on whether
patients or health professionals fill out –
need to standardize to whom instrument
administered
Choosing a method: Validity
 QWB based on VAS
 HUI2 and HUI3 based on SG but as
transformation of VAS; not clear this is better
than TTO used in EQ-5D
 Populations from which data for scoring formulas
derived are more or less comprehensive – more
limited for HUIs, very broad for EQ-5D
 Unclear how important this is
Choosing a method:
Conclusions
Differences in dimensions covered,
number of levels, floor and ceiling effects
may make one of the instruments more
suitable for a particular patient group
 Which would you use for assessing cataract
surgery? Antidepressants?
Use HUI3 rather than HUI2
Don’t use QWB
A QALY exercise
With new cancer protocol: 6 months at
HRQOL 0.3, followed by 15 years at 0.95
With standard treatment: 1 year at 0.5,
followed by 7 years at 0.90, then 1 year at
0.8, 1 year at 0.5, then death
How many QALYs does the new protocol
produce?
Limitations of QALYs
(partial list)
Ignore priority often given to helping
people at low initial state
Many small improvements to people at
high initial state can be preferred to saving
a life
Imperfectly measured
Use anyway?
QALYs vs DALYs (Disabilityadjusted life years)
Feature
QALYs
DALYs
Life expectancy measure Context-specific
Longest in world
(Japanese women)
Disability weights
Preferences (public or
patients in study)
PTO scores from a panel
of health care workers
Precision
Continuous scores
7 states in addition to
healthy or dead
Age weights?
No
Yes – lower weights for
young and elderly
DALYs developed to do estimate potential impacts of
possible health interventions in developing countries