Session 3: Utility measures,
preference elicitation methods
and mapping
- Brendan Mulhern -
Learning objectives

To be aware of the range of generic measures that can be
used to measure utilities, and their similarities and
differences

To understand the methods used to value HRQOL, and
their advantages and disadvantages

To be aware of methods for mapping between nonpreference based and preference based measures
Introduction

A general preference from decision-makers for costutility analyses:


Statement of cost-effectiveness as the incremental cost per
quality adjusted life year (QALY) gained.
QALYs:


Combine length of life (survival) with quality of life.
“Standard” metric that can be compared across indications.
Introduction

“Quality Adjustment” (utility value) reflects value of health
states relative to full health




QALY weight = 1 for “full health”
QALY weight = 0 for death
“Worse than death” health states permissible (QALY weight < 0)
Population based preference weights

Strength of preference for survival and quality of life and
trade-offs between the two.
Introduction

Putting the ‘Q’ in QALYs:

Preference elicitation methods:



Health using utility measures:




Standard gamble, time trade-off, visual analogue scales (cardinal
methods)
Ranking data, discrete choice experiments (ordinal methods)
Generic Multi-Attribute Utility Instruments (MAUI - EQ-5D, SF-6D,
AQOL, HUI)
Condition specific MAUI
Vignettes
Mapping to estimate utilities when direct data not available

E.g. DLQI to EQ-5D
Part 1: Methods for the valuation of health
Health Utility

Reflects the ‘value’ of a health state

How much is someone willing to give up to avoid/achieve a
state?

Scores are weighted by preference (how good or bad person
thinks the health state is)

HRQL represented by a single index value that can be applied
across all health states on a cardinal (ratio) scale

Utility of 0.8 is exactly twice as good as a utility of 0.4
Valuation methods

Health state valuation techniques used to derive utility values
for MAUIs




Visual Analogue Scale
Standard Gamble
Time Trade Off
Discrete Choice Experiments

Selection of health states described by the MAUI valued

Responses modelled to produce a utility value for each state
anchored on a 0 to 1 scale:



1 = Full health
0 = Dead
Negative values = States valued or modelled as worse than dead
Best imaginable
Health state
Visual Analogue Scale


Line with anchors representing best and worst
possible health
States placed on line so intervals reflect differences

To value states on QALY scale, respondents also
value ‘dead’. Values are transformed onto the 0-1
scale

Easy to administer and complete

Does not capture a respondent’s strength of
preference for certain states as no choices are
involved

Prone to ‘response spreading’ and context effects
Worst imaginable health
state
Standard Gamble

Based on expected utility theory

Choice between two alternatives - one certain, and one involving risk:


Alternative 1: Full health for t years (prob. p) OR immediate death (prob. 1 –
p).
Alternative 2: MAUI health state for t years.

Probability of returning to normal health varied until indifference (Hi = P
at indifference)

Good completion and response rates

People generally risk averse and want to avoid death, so higher utilities
produced

Complex choices
Standard Gamble
Time Trade Off

Values for EQ-5D and condition specific MAUIs derived
using TTO

To value a health state:



Life B: 10 years in MAUI health state
Life A: x (< 10) years in full health
x is changed through ‘iterative’ process until indifference

Value for the state is x/10

Responses modelled to produce a value for each health
state
Time Trade Off
Discrete Choice Experiments

Interest in using Discrete Choice methods for health state valuation

Respondents provide ordinal preferences over health profiles

Aggregate choice data modelled to produce a decrement for each
level of each attribute

Different cognitive demands to TTO and SG

DCE utility values need to be anchored on the full health-dead
utility scale


Use external data?
Include duration as an attribute of DCE (DCETTO, Bansback et al, 2012;
Mulhern et al, 2014;Viney et al, 2013)
Discrete Choice examples
Exercise: Valuation methods

Value the health states described using Visual
Analogue Scale, Standard Gamble and Time Trade Off

Remember that preference valuation tasks are about your
individual preferences – there are no right or wrong answers

VAS – You are just ranking the states, you are not ‘giving up’
anything

SG – You are valuing the health state based on how much risk
you are willing to take to avoid that state.

TTO – You are valuing the health state based on how many
years of life you are willing to give up to avoid that state.
Exercise health states
Imagine that you will be in each health state for ten years and then you will die.
F
Full Health
Inflammation of joints which causes pain, stiffness, warmth,
redness and swelling. Ordinary tasks are difficult to undertake.
Simply getting out of bed, dressing, bathing, walking, or opening
A
doors can become major chores. Psychologically, the daily pain
can lead to stress and fatigue. This pain, stress and fatigue can
lead to anger and depression.
D
Immediate Death
Part 2: Generic multi-attribute utility
instruments
Generic multi-attribute utility
instruments

Generic MAUI developed to be used across a wide variety of
health conditions

Examples of generic MAUI:


EQ-5D-3L/EQ-5D-5L

SF-6D (based on SF-36/SF-12)

Assessment of Quality of Life (AQoL)

Health Utility Index Marks 1-3 - HUI, HUI2, HUI3
MAUI have two components:

A health state classification system administered on patients

A utility ‘value set’ for all health states defined by the classification
system obtained from a general population sample
EQ-5D

Health state classification system:




5 dimensions (Mobility, Self-care, Usual activities, Pain/discomfort,
Anxiety/depression
3 levels of severity (None, Some/Moderate, Unable/Extreme)
243 health states described
Utility value set:

Australia (Viney et al, 2011):



198 states valued by general public (n=417) using TTO
Regression analysis was used to estimate value set (range 1 to -0.217;
8% states worse than dead)
United Kingdom (Dolan, 1997):


Values for 45 states elicited from general public (n=3395) using TTO
Value set range 1 to -0.594; > 30 % states worse than dead
EQ-5D
By placing a tick in one box in each group below, please indicate which statements
best describe your own health state today.
Mobility
I have no problems in walking about
I have some problems in walking about
I am confined to bed



Self-Care
I have no problems with self-care
I have some problems washing or dressing myself
I am unable to wash or dress myself



Usual Activities (e.g. work, study, housework, family or
leisure activities)
I have no problems with performing my usual activities
I have some problems with performing my usual activities
I am unable to perform my usual activities



Pain/Discomfort
I have no pain or discomfort
I have moderate pain or discomfort
I have extreme pain or discomfort



Anxiety/Depression
I am not anxious or depressed
I am moderately anxious or depressed
I am extremely anxious or depressed



EQ-VAS
Criticisms of EQ-5D-3L

Descriptive system




Not sensitive to small differences and changes in health (e.g.
change between ‘none’ and ‘some’
Large ceiling effect (respondents reporting best health state
11111
Content misses important dimensions in some conditions (e.g.
fatigue in cancer?)
Valuation process


TTO procedure for states ‘worse than dead’
Modelling of negative values
EQ-5D-5L
SF-6D

Derived from the SF-36/SF-12 (Brazier et al 2002; 2004)

Descriptive system



Value set (UK)



6 dimensions: Physical functioning, Role limitations, Social,
functioning, Pain, Mental health,Vitality
4-6 levels of severity - defines 18,000 health states
249 health states valued by 835 members of UK population using
standard gamble
Range 0.29 to 1
Value set (Australia)


1,017 members of general public completed DCETTO study
Functioning, pain, mental health, and vitality largest drivers of
utilities. Range -0.4 to 1
SF-6D
The following questions ask about different aspects of your health. There are six groups of
statements, each covering a different aspect of health. Please tick one statement in each group
to show the statement which best describes your health over the past 4 weeks.
Physical Functioning
Your health does not limit you in vigorous activities…….
Your health limits you a little in vigorous activities……....
Your health limits you a little in moderate activities…......
Your health limits you a lot in moderate activities…….....
Your health limits you a little in bathing and dressing......
Your health limits you a lot in bathing and dressing……...
Role Limitations
You have no problems with your work or other regular daily activities as a result
of your physical health or any emotional problems.......…………..
Pain
You have no pain……...…………………………
You have pain but it does not interfere with your normal work (both outside the
home and housework) ……..
You have pain that interferes with your normal work (both outside the home and
housework) a little bit…………
You have pain that interferes with your normal work (both outside the home and
housework) moderately…........
You have pain that interferes with your normal work (both outside the home and
housework) quite a bit………...
You have pain that interferes with your normal work (both outside the home and
housework) extremely………...
You are limited in the kind of work or other activities as a result of your physical
health……
Mental Health
You accomplish less than you would like as a result of emotional
problems……...…...........
You feel tense or downhearted and low a little of the time.....
You are limited in the kind of work or other activities as a result of your physical
health and accomplish less than you would like as a result of emotional
problems……
You feel tense or downhearted and low none of the time......
You feel tense or downhearted and low some of the time…..
You feel tense or downhearted and low most of the time…...
You feel tense or downhearted and low all of the time………
Social Functioning
Your health limits your social activities none of the time…...
Vitality
Your health limits your social activities a little of the time.....
You have a lot of energy all of the time…….....
Your health limits your social activities some of the time…..
You have a lot of energy most of the time........
Your health limits your social activities most of the time…...
You have a lot of energy some of the time…...
Your health limits your social activities all of the time……....
You have a lot of energy a little of the time…..
You have a lot of energy none of the time……
Assessment of Quality of Life

AQOL 4D, 6D, and 8D developed in Australia

AQoL- 8D descriptive system:

8 dimensions (35 items):



Independent Living, Relationships, Mental Health, Coping, Pain, Senses,
Self-Worth, Happiness
Generated using exploratory factor analysis and structural
equation modelling
Value set:



TTO scores obtained from 323 patients and 347 members of the
public
Multiplicative and econometric models constructed
Value set range1 to -0.04
Rationale for generic MAUI

Advantages





‘Off-the-shelf’
Cheap, convenient and comparatively easy to use in clinical
trials and other studies
Provide a standard measure across programmes, patient groups
and treatments
Accepted by NICE, PBAC and other agencies in decision
making
Disadvantages

May not be relevant or sensitive to the condition or treatment
effects
How do generic MAUI compare?

Descriptive system differences:




Dimensions – content and description
Severity – range (e.g. floor effect in SF-6D; ceiling effect in EQ-5D-3L)
Sensitivity – number of levels (e.g. EQ-5D-3L for less severe problems)
Valuation differences:


Preference elicitation technique – e.g. SG known to give higher values than
TTO
Variant of preference elicitation technique (e.g. different iterative versions
of TTO – Ping pong vs. titration)

Generic measures may not be sensitive to all of the HRQOL
impacts of a particular condition

Would a condition specific MAUI be more useful?
Part 3: Mapping
Mapping

Process of ‘mapping’ non-preference based measures onto
generic preference-based measure to predict utilities from the
non-preference based measure

E.g DLQI onto EQ-5D.

Requires both measures to be used in the same patient sample

Uses regression methods to predict mapping function
Advantages:
 It can be quick and in some circumstances it may be adequate
 Accepted by NICE (2013) and some other agencies
Mapping process
Estimation sample:
1.
•
Clinical and demographic characteristics cover same range as ‘target’
Dependent variable:
2.
•
•
Utility index
Dimension level scores (response mapping)
Independent variables:
3.
•
Non-preference based measure of health (total, dimension or item
score), clinical measures and socio-demographic characteristics
Model selection, specification and performance:
4.
•
•
OLS, Tobit, CLAD, Two part or various mixture models
Assess RMSE/MSE, error across severity, range penalised likelihood
(AIC/BIC)
Uncertainty testing:
5.
•
Sensitivity analysis
Predictions of EQ-5D from generic and
condition specific measures
Mean
1.00
EQ-5D
Inpatients
GLS RE
0.75
Gray et al.
(2006)
0.50
Franks et
al (2004)
0.25
0.00
-0.25
-0.50
EQ-5D state (ordered by severity)
Actual out-of-sample mean EQ-5D
values, predicted values and errors
1.0
0.8
EQ-5D scores
0.6
0.4
0.2
0.0
-0.2
Actual EQ-5D
Predicted EQ(1)
error
Mapping summary

Mapping is useful when actual utility data are not available and
is reasonably widely used across a range of conditions

However, generic measures are not appropriate in all
conditions/there may not be a psychometric relationship
between the measures. This increases the error of the
predictions

Important dimensions may not appear in the generic measure
(e.g. vitality has an insignificant coefficient when EQ-5D
mapped onto SF-36 dimension scores)

Another solution is to directly value the content of condition
specific measures
Any questions?
For further information contact:
[email protected]
Further reading/resources

General references and QALYs



Health measurement/valuation









Brazier J, Ratcliffe J, Solomon J, Tsuchiya A. (2016). Measuring and Valuing Health Benefits for Economic Evaluation
2nd Edn. Oxford: University Press.
Drummond MF, Sculpher MJ, Torrance GW, O’Brien BJ, Stoddart GL. (2005). Methods for the economic evaluation
of healthcare programmes. Oxford: University Press.
Bansback N, Brazier J, Tsuchiya A, Anis A (2012). Using a discrete choice experiment to estimate societal health
state utility values. Journal of Health Economics, 31(1):306-318.
Brazier, J., Roberts, J., Deverill, M. (2002). The estimation of a preference-based measure of health from the SF-36. J
Health Econ; 21: 271-92
Dolan P. Modeling valuations for EuroQol health states. Medical care 1997;35(11):1095-108.
Herdman M, Gudex C, Lloyd A, et al. (2011) Development and preliminary testing of a new five level version of the
EQ-5D – the EQ-5D-5L. Quality of Life Research; 20(10): 1727-36
Longworth L, Yang Y, Young T, Mulhern B, Hernandez-Alava M, Mukuria C, Rowen D, Tosh J, Tsuchiya A, Evans P.
(2014). Use of generic and condition specific measures of health related quality of life in NICE decision making:
systematic review, statistical modelling and survey. Health Technology Assessment, 18:9.
Oppe M, Devlin NJ, van Hout B, Krabbe PFM, de Charro F. (2014). A program of methodological research to arrive
at the new international EQ-5D-5L valuation protocol. Value in Health, 17: 445-53.
Szende A, Oppe M, Devlin N. EQ-5D value sets: inventory, comparative review and user guide. Dordrecht: Springer;
2007.
Viney R, Norman R, Brazier J, Cronin P, King M, Ratcliffe J, Street D. (2013) An Australian discrete choice
experiment to value EQ-5D health states. Health Economics. 23: 729-742.
Mapping



Brazier JE, Yang Y, Tsuchiya A, Rowen DL. (2010) A review of studies mapping (or cross walking) non-preference
based measures of health to generic preference-based measures. Eur J Health Econ; 11(2):215-25.
Longworth L, Rowen D. (2013). Mapping to obtain EQ-5D utility values for use in NICE health technology
assessments.Value Health. 16(1):202-10
Mapping study database: https://www.herc.ox.ac.uk/downloads/herc-database-of-mapping-studies