Prof Nancy Devlin provides an overview of the EuroQol Groups’ scientific research agenda for the coming year and plans for the EQ-5D-5L approach to measuring patient-reported outcomes of care.
EuroQol Agenda and Developing the New EQ-5D-5L Value Sets
1. An Overview of the EuroQol Group’s
Scientific Agenda and
Value Sets for EQ-5D-5L
Professor Nancy J. Devlin
Office of Health Economics
Pharmaceutical Oncology Initiative (POI) Meeting
London • 18 January 2013
2. Contents
1. Recent developments with the EQ-5D
2. EuroQol Group: new scientific agenda & Working Groups
3. Valuation of EQ-5D-5L
4. Overview of 5L value set studies
5. Points for discussion
3. 1. EuroQol Group: recent developments
Developments over recent years have focused on:
• Development of a new 5-level version: EQ-5D-5L
• 94 languages; PDA/tablet/web
• Development of a version of EQ-5D for children: EQ-5D-Y
• Provision of an interim value set for the 5L (based on a
crosswalk study)
• Preparation for valuation of the EQ-5D-5L: new methods
for administering, eliciting and modelling values.
4. Final version:
UK English
Herdman, M., Gudex, C., Lloyd, A., Jans
sen, M.F., Kind, P., Parkin, D., Bonsel, G
. and Badia, X. Development and
preliminary testing of the new EQ-5D-
5L. Quality of Life Research.
6. Interim value set: study populations
COPD/asthma (n=342)
6 Countries Diabetes (n=275)
Liver disease (n=426)
Italy
Rheumatoid arthritis/arthritis (n=372)
United Kingdom
(Scotland, Engla Cardiovascular disease (n=251)
nd) Patient study
population Stroke (n=614)
Netherlands
Depression (n=250)
Poland Personality disorders (n=384)
Students (n=443)
Denmark
Other (n=334)
7. A comparison of the distribution of responses
on the 3L and 5L
8. Mobility 5L No Slight Moderate Severe
3L problems problems problems problems Unable to
No problems 1782 119 16 1 4 Van Hout et al (2012) Interim
Some problems 29 552 586 386 23
Confined to bed 1 1 4 30 112
scoring for the EQ-5D-5L:
Self-Care 5L No Slight Moderate Severe Mapping the EQ-5D-5L to EQ-
3L problems problems problems problems Unable to
No problems 2468 82 13 5 0
5D-3L value sets Value in
Some problems 43 408 313 109 6 Health. 15(5), 708-715.
Unable to 3 5 6 35 140
Usual Activities 5L No Slight Moderate Severe
3L problems problems problems problems Unable to
No problems 1382 163 20 9 0
Some problems 42 661 656 274 15 Note: problems with
Unable to 5 7 23 134 239 inconsistencies in self-
Pain/Discomfort 5L
3L None Slight Moderate Severe Extreme reported health in parallel
None 1126 211 21 6 2
Moderate 65 850 837 239 8
fielding pose challenges for
Extreme 1 4 19 159 82 modelling the relationship
Anxiety/Depression 5L
3L None Slight Moderate Severe Extreme
between 5L and 3L health
None 1352 219 30 10 3 states
Moderate 45 841 692 164 6
Extreme 1 3 17 158 93
9. Modelling approach
For each 5L dimension, calculate
probability of response on 3-level
system.
Given those probabilities, calculate
probability of being in each of the 243
(3L) states for a given 5L health state
vector.
This facilitates a 5L “utility” score for
each of the 3125 health
states, calculated as the average utility
score weighted by the probabilities to
be in one of the 243 health states.
11. Interim 5L scores
• 5L value
calculator
available from
the EuroQol
Group website
12. 2. EuroQol Group: new scientific agenda
• Value Sets WG
• Implement valuation protocol for 5L studies
internationally
• Valuation methodology
• Innovative methods for future protocols
• Conceptual basis for ‘bolt on’ dimensions
• Methods for valuation of EQ-5D-Y
• EQ Data Archive
• EQ-5D in Children
• Large scale health applications (PROMs and pops)
• EQ-5D beyond health care
13. 3. Valuation of EQ-5D-5L
• 5L Value set studies underway/planned:
England China
UK Singapore
Ireland Japan
Netherlands South Korea
Spain Taiwan
Germany Middle East
Canada US
Uruguay Nigeria
New Zealand
14. Challenges valuing EQ-5D-5L
• 55 = 3,125 states to value (cf. 243 in
EQ-5D)
• Differences between EQ-5D-5L
Is 55555(5L)
states more subtle than EQ-5D = 33333(3L)?
• Potentially, a greater cognitive (hopefully, ye
s)
burden in differentiating
between/imagining the states
Is 11111
• The 5L more sensitive PRO, but (5L) = 11111
(3L)? (yes)
how does the general public view How does 51111
the decrements in utility? (5L) cf. 31111
(3L)? (unable to
How much ‘milder’ is walk about vs.
12111(5L) cf. 12111 (3L)? confined to bed)
How does 33333 (5L) cf. 22222
(3L)?
15. EQ-5D-5L valuation protocol
1. General welcome
2. Introduction
• Self reported health on the EQ-5D-5L descriptive system
• Self reported health on the EQ-VAS
• Background questions
3. Composite Time Trade-Off
• Instructions and example of TTO task
• TTO valuation of 10 EQ-5D-5L states
• TTO debriefing/structured feedback
4. Discrete Choice
• Instructions and example of DC task
• DC valuation of 10 pairs of EQ-5D-5L states
• DC debriefing/structured feedback
5. General thank you and goodbye
16. EQ-VT (‘valuation technology’)
• EQ-VT software developed to display the tasks; automate
the questioning process; and capture and timestamp
responses
• Administered in face to face interviews, ie computer
assisted personal interviews (CAPI) – not online.
17. Choice of valuation methods
• TTO: widely used and accepted (eg by NICE), so forms the basis
for 5L valuations.
• But problems with conventional approach to TTO in valuation
of states worse than dead.
• Lead time TTO a uniform method that can yield values < 0 and
> 0 – but also not without its problems!
• EuroQol Group has opted for a composite approach:
conventional TTO for > 0; lead time TTO for < 0
• DC: without duration; preferences data used to supplement
TTO.
21. 5L value sets for England & UK
• ‘An EQ-5D-5L Value set for England’: funded by DH via the
NIHR PRP
• Research led by OHE/Sheffield: N Devlin, B van Hout, K
Shah, B Mulhern, Y Feng
• Commenced November 2012; to report July 2013
• ‘An EQ-5D-5L value set for the UK’: additional data
collection in NI, Wales and Scotland funded by Office of
the Chief Scientist, Scotland and the EuroQol Foundation
• Data collection to commence March 2013; to report by
December 2013
• Sample recruitment/interviews for both studies
undertaken by Ipsos Mori
22. What might the value sets look like?
• Too early to tell (< a third of final sample)
But…
• TTO values have a good relationship with the ‘misery
score’ for states.
• Likely that the minimum value (55555) will be higher than
that for 33333 in the MVH
• Improved approaches to modelling will effect the
characteristics of the value set – no N3 term
23. 5. For discussion with POI
• Will the 5L improve the sensitivity of EQ-5D in measuring
QoL for cancer patients?
• Are there any aspects of QoL which might be potential
candidates for exploration of ‘bolt ons’?
• How will the new values effect estimates of QoL
improvement for cancer medicines?