26
HERU is supported by the Chief Scientist Office of the Scottish Executive Health Department and the University of Aberdeen. The author accepts full responsibility for this talk. Health Economics Research Unit, University of Health Economics Research Unit, University of Aberdeen Aberdeen Using DCEs to estimate utility Using DCEs to estimate utility weights within the framework of weights within the framework of QALYs QALYs Professor Mandy Ryan Professor Mandy Ryan

Using DCEs to estimate utility weights within the framework of QALYs Professor Mandy Ryan

  • Upload
    ashton

  • View
    38

  • Download
    0

Embed Size (px)

DESCRIPTION

Using DCEs to estimate utility weights within the framework of QALYs Professor Mandy Ryan. Structure. What DCEs are and background to their use in Health Economics Application – developing a utility index in the area of glaucoma anchoring between 0 and 1 (John and Theresa) - PowerPoint PPT Presentation

Citation preview

Page 1: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

HERU is supported by the Chief Scientist Office of the Scottish Executive Health Department and the University of Aberdeen.

The author accepts full responsibility for this talk.

Health Economics Research Unit, University of Health Economics Research Unit, University of AberdeenAberdeen

Using DCEs to estimate utility weights Using DCEs to estimate utility weights within the framework of QALYswithin the framework of QALYs

Professor Mandy Ryan Professor Mandy Ryan

Page 2: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

StructureStructure

• What DCEs are and background to their use in Health Economics

• Application – developing a utility index in the area of glaucoma

anchoring between 0 and 1 (John and Theresa)distinguishing ‘weight’ from ‘scale’ (Terry)assumption and analysis issues (Jorge, John +

Theresa)

Page 3: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Discrete choice experimentsDiscrete choice experiments

• Attribute based hypothetical survey measure of value

• Origins in mathematical psychologyDistinguish from conjoint analysisAlso known as ‘Stated preference discrete choice modelling’

• Increasingly used in environmental, transport and health economics

Page 4: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Can’t have the best of everything!Can’t have the best of everything!

Legroom

Food and drink

Entertainment

Reclining chair

Ticket priceCheck-in service

Page 5: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Example of binary - Yes/No response Example of binary - Yes/No response

Place of Screening Type of Screening

Cost to you of Chlamydia Screening

Chance of Pelvic Inflammatory Disease (PID) if not screened.

Type of Information and Support when

you are given Screening Results

Family Planning Clinic

Full Pelvic Examination

£5 10% None

Family Planning Clinic

Perineal Swab £10 0% None

Genito Urinary Medicine (GUM)

ClinicUrine Test £10 10%

Support of Trained Health Advisor

At Home Perineal Swab £5 5%Support of Trained

Health Advisor

At Home Urine Test Free 0% None

At GP ClinicFull Pelvic

Examination£20 0%

Support of Trained Health Advisor

I would have Test

I would not have

Test

Choice 1

Choice 3

Choice 4

Choice 5

Choice 6

Choice 2

Page 6: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Example of generic multiple choice – Example of generic multiple choice – including a neither optionincluding a neither option

Question 4 Clinic A Clinic B Length of wait 28 weeks 6 weeks Time with doctor 15 mins 45 mins Pain Management Service

No Specialist Team

Specialist Team

Cost to you £60 £60 Which Clinic would you prefer (tick one box only)?

Prefer Clinic A

Prefer Clinic B

Neither

Page 7: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Discrete choice experimentsDiscrete choice experiments

• Attribute based hypothetical survey measure of value

• Origins in mathematical psychologyDistinguish from conjoint analysisAlso known as ‘Stated preference discrete choice modelling’

• Increasingly used in environmental, transport and health economics

Page 8: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

DCEs – their use in HEDCEs – their use in HE• Pre 1970 - cost-benefit analysis

human capital approach willingness to pay

• 1970s - cost-effectiveness analysis e.g. cost per life year

• 1980s - cost-utility analysis e.g. cost per Quality Adjusted Life Years (QALYs) Standard gamble and time trade-offs

• 1990s - cost-benefit analysis health, non-health and process attributes Contingent valuation method and discrete choice experiments

• 2000 forward the importance of factors beyond health outcomes NICE

• WTP for a QALY• Estimation of utility weights

Page 9: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

HERU is supported by the Chief Scientist Office of the Scottish Executive Health Department and the University of Aberdeen.

The author accepts full responsibility for this talk.

Health Economics Research Unit, University of Health Economics Research Unit, University of AberdeenAberdeen

Eliciting a health state utility index using a Eliciting a health state utility index using a discrete choice experiment: an application discrete choice experiment: an application

to Glaucomato Glaucoma

Funded by Ross FoundationFunded by Ross Foundation

Jen Burr, Mary Kilonzo, Mandy Ryan, Jen Burr, Mary Kilonzo, Mandy Ryan, Luke ValeLuke Vale

Page 10: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Case Study - GlaucomaCase Study - Glaucoma

• chronic eye disease - progressive damage to optic nerve

• does not reduce length of life but associated with impaired quality of life

• outcomes - intraocular pressure reduction and measures of visual function

• do not capture impact of condition or treatment on emotional and physical functioning or lifestyle

• Standard gamble and time trade-off not appropriate

Page 11: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Conducting a DCEConducting a DCE

• Stage 1 - Identifying attributes and levels

• Stage 2 - Experimental design to determine choices

• Stage 3 - Collecting data Principles of a good survey design

• Stage 4 - Data analysis Discrete choice modelling

• Conditional logit model and developments– nested logit, random parameter logit

Page 12: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Attributes and LevelsAttributes and Levels

Attributes

• Central and Near Vision

• Lighting and glare

• Mobility

• Activities of daily living

• Local eye discomfort

• Other effects of glaucoma and treatment

Levels

• No difficulty

• Some difficulty

• Quite a bit of difficulty

• Severe difficulty

Page 13: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Experimental designExperimental design

• Fractional factorial design of 32 choicesMain effects no interactions

• PropertiesOrthogonalityLevel balanceMinimum overlap

Page 14: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Example of a DCE choice – respondents were Example of a DCE choice – respondents were asked what they think is WORSEasked what they think is WORSE

SITUATION A SITUATION B

No difficulty with:Central and near visionLighting and glareMobilitySome difficulty with:Activities of daily livingEye discomfortOther effects of glaucoma and its treatment

No difficulty with:Central and near visionSome difficulty with:Lighting and glareQuite a lot of difficulty with:Activities of daily living Other effects of glaucoma and its treatmentSevere difficulty with:MobilityEye discomfort

(Tick one box only) Situation A Situation B

Page 15: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Rationality testsRationality tests

Dominance tests too easy and may question credibility of experiment

Sen’s expansion and contraction rationality tests used

Page 16: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Data collectionData collection

• Subjects from 4 hospital-based clinics and 1 community-based glaucoma clinic across two eye centres in the UK (Aberdeen and Leeds) received questionnaire (n=225)

• Also recruited volunteers from the International Glaucoma Association (IGA) (n=248)

Page 17: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Analysis of DCEAnalysis of DCE

• QWij = ∑dlXdl + e + u • where

QWij is the quality weight for outcome state i as valued by individual j

Xdl is a vector of dummy variables • where d represents the attribute from the profile measure • l the level of that attribute

Page 18: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Estimating utility weightsEstimating utility weights

• summation of the coefficients associated with the best level for each attribute

• Rescaled between zero (worse level of all attributes) and 1 (best level of all attributes)

Page 19: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Response rates and rationalityResponse rates and rationality

• 289 subjects responded to DCE questionnaire

• 3 respondents failed both consistency tests

• Analysis performed on 286 respondents

• Analysed according to severity

Page 20: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Results of the DCEAttributes and levels Coefficient

Central and near vision tasks

No difficulty 1.254

Some difficulty 0.852

Quite a lot of difficulty 0.526

Lighting and glare

No, some and quite a lot of difficulty 0.272

Mobility

No difficulty 0.921

Some difficulty 0.577

Quite a lot of difficulty 0.349

Visual judgement for activities of daily living

No difficulty 0.999

Some difficulty 0.720

Quite a lot of difficulty 0.431

Eye discomfort

No difficulty 0.241

Some and quite a lot of difficulty 0.134

Other effects

No difficulty 0.202

Some and quite a lot of difficulty 0.169

Page 21: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Quality weights Quality weights Dimension Index

Central and Near Vision

No difficulty 0.322

Some difficulty 0.219

Quite a lot 0.135

severe 0

Lighting and glare

No difficulty 0.070

Some difficulty 0

Quite a lot 0

severe 0

Mobility

No difficulty 0.237

Some difficulty 0.148

Quite a lot 0.090

severe 0

Dimension Index

Activities of daily living

No difficulty 0.257

Some difficulty 0.185

Quite a lot 0.111

severe 0

Eye discomfort

No difficulty 0.062

Some difficulty 0.035

Quite a lot 0.035

severe 0

Other effects

No difficulty 0.052

Some difficulty 0.043

Quite a lot 0.043

severe 0

Page 22: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Utility score for BEST health stateUtility score for BEST health state

Situation description Qualityweights

UtilityScore

You have no difficulty with central and near vision 0.322 1

You have no difficulty with lighting and glare 0.070

You have no difficulty with mobility 0.237

You have no difficulty with activity of daily living 0.257

You have no difficulty with local eye discomfort 0.062

You have no difficulty with other effects of glaucoma and its treatments

0.052

Page 23: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Utility score for WORSE health stateUtility score for WORSE health state

Situation description Quality weights

Utility Score

You have severe difficulty with central and near vision

0 0

You have severe difficulty with lighting and glare 0

You have severe difficulty with mobility 0

You have severe difficulty with activity of daily living

0

You have severe difficulty with local eye discomfort 0

You have severe difficulty with other effects of glaucoma and its treatments

0

Page 24: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Utility score for intermediate health stateUtility score for intermediate health state

Situation description Qualityweights

UtilityScore

You have some difficulty with central and near vision 0.219 0.737

You have some difficulty with lighting and glare 0

You have some difficulty with mobility 0.148

You have no difficulty with activity of daily living 0.257

You have no difficulty with local eye discomfort 0.062

You have no difficulty with other effects of glaucoma and its treatments

0.052

Page 25: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Some general pointsSome general points

• One of few studies to estimates utility weights from DCEs (though appears to be increasing)

• Programme specific!

• Response rate 62% good for DCE, though issues of generalisability are important

• Preferences differed according to severity

Page 26: Using DCEs to estimate utility weights within the framework of QALYs  Professor Mandy Ryan

Points for DiscussionPoints for Discussion

• Weights for use in programme specific QALYWhat if want to generate generic QALY weights (anchored between

DEATH and PERFECT HEALTH)• How value DEATH?

• Distinguishing weight (importance of attribute) from scale (importance of attribute levels)

• Econometric analysisAssumptions of logit model

• Errors terms independent, irrelevance of alternatives and heterogeneityDecision making heuristics

• Do individuals trade across attributes