Utilising rank and DCE data to value health status on the ‘QALY’ scale using conventional and Bayesian methods John Brazier and Theresa Cain with Aki Tsuchiya.

Slides:



Advertisements
Similar presentations
Elicitation methods Health care demands exceed resource supply Therefore, rationing is inevitable Many ways by which we can ration health care One is economic.
Advertisements

Emma Frew Introduction to health economics, MSc HEHP, October 2012 Outcomes: part II.
Rural Economy Research Centre Modelling taste heterogeneity among walkers in Ireland Edel Doherty Rural Economy Research Centre (RERC) Teagasc Department.
Children’s subjective well-being Findings from national surveys in England International Society for Child Indicators Conference, 27 th July 2011.
Scaling Session Measurement implies “assigning numbers to objects or events…” Distinguish two levels: we can assign numbers to the response levels for.
Research methods – Deductive / quantitative
DEVELOPMENT OF A PREFERENCE-BASED, CONDITION SPECIFIC PATIENT REPORTED OUTCOME MEASURE FOR USE WITH VENOUS ULCERATION Simon Palfreyman 1, John E Brazier.
Modelling Partially & Completely Missing Preference-Based Outcome Measures (PBOMs) Keith Abrams Department of Health Sciences, University of Leicester,
Using a discrete choice experiment with duration to estimate values for health states on the QALY scale Nick Bansback Assistant Professor School of Population.
Utility Assessment HINF Medical Methodologies Session 4.
A METHODOLOGY FOR MEASURING THE COST- UTILITY OF EARLY CHILDHOOD DEVELOPMENTAL INTERVENTIONS Quality of improved life opportunities (QILO)
Common Factor Analysis “World View” of PC vs. CF Choosing between PC and CF PAF -- most common kind of CF Communality & Communality Estimation Common Factor.
Two ways to skin a cat: a comparison of two variants of standard gamble John Brazier and Paul Dolan Prepared for the CHEBS workshop on Elicitation, 9 October.
C82MCP Diploma Statistics School of Psychology University of Nottingham 1 Designing Experiments In designing experiments we: Manipulate the independent.
Elicitation Some introductory remarks by Tony O’Hagan.
Chapter 51 Experiments, Good and Bad. Chapter 52 Experimentation u An experiment is the process of subjecting experimental units to treatments and observing.
1 Modelling valuations for the EQ-5D health states: an alternative model using differences in valuations Jennifer Roberts and Paul Dolan Sheffield Health.
A RECIPE FOR INCOHERENCE: AVERAGING TIME-TRADEOFF OR STANDARD-GAMBLE UTILITIES ACROSS HEALTH ATTRIBUTES Gordon B. Hazen, IEMS Department, Northwestern.
Valuing the SF-6D: a nonparametric approach using individual level preference data Part 1): The SF-6D and its valuation Samer A Kharroubi, Tony O’Hagan,
Modelling Cardinal Utilities from Ordinal Utility data: An exploratory analysis Peter Gilks, Chris McCabe, John Brazier, Aki Tsuchiya, Josh Solomon.
Estimating utilities from individual preference data Some introductory remarks by Tony O’Hagan.
Using ranking and DCE data to value health states on the QALY scale using conventional and Bayesian methods Theresa Cain.
Chapter 7 Correlational Research Gay, Mills, and Airasian
Measuring and valuing health outcome Montarat Thavorncharoensap, Ph.D. 1: Faculty of Pharmacy, Mahidol University 2. HITAP, Thailand.
Week 9: QUANTITATIVE RESEARCH (3)
MAPPING THE DIABETES HEALTH PROFILE (DHP-18) ONTO THE EQ-5D AND SF-6D GENERIC PREFERENCE BASED MEASURES OF HEALTH Brendan Mulhern 1, Keith Meadows 2, Donna.
1 EQ-5D, HUI and SF-36 Of the shelf instruments…..
1 The valuation of disease-specific questionnaires for QALY analysis  To rescue data in absence of an utility measure  Growth hormone deficiency in adults.
Valuing Health Effects of Air Pollution in DevelopingCountries: The Case of Taiwan* JOURNAL OF ENVIRONMENTAL ECONOMICS AND MANAGEMENT 34, 107 ]
1 CADTH Value Methods Panel Using Best Worst Scaling to elicit Values Carlo Marra.
Economic evaluation of health programmes Department of Epidemiology, Biostatistics and Occupational Health Class no. 11: Cost-utility analysis – Part 4.
Chapter Eight The Concept of Measurement and Attitude Scales
by B. Zadrozny and C. Elkan
Measuring Health Outcomes
Why use the EQ-5D? What are the alternatives?. What are the alternatives for Direct valuation? Other VAS Time Trade-Off Standard Gamble Willingness to.
Health State Unable to perform some tasks at home and/or at work Able to perform all self care activities (eating, bathing, dressing) albeit with some.
317_L26, Mar J. Schaafsma 1 Review of the Last Lecture Are looking at program evaluation in healthcare Three methods: CBA, CEA, CUA discussed CBA,
Evaluating a Research Report
Statistical analysis Prepared and gathered by Alireza Yousefy(Ph.D)
Chapter Nine
Finding out what people want: a case study of preference elicitation using a multi- criteria methodology David Whitmarsh and Maria Giovanna Palmieri CEMARE,
Evidence and Information for Policy Health as a multi-dimensional construct and cross-population comparability Colin Mathers (WHO) on behalf of Taskforce.
1 EQ-5D, HUI and SF-36 Of the shelf instruments…..
SURVEY RESEARCH.  Purposes and general principles Survey research as a general approach for collecting descriptive data Surveys as data collection methods.
Overview of Health-Related Quality of Life Measures May 22, 2014 (1:00 – 2:00 PDT) Kaiser Methods Webinar Series 1 Ron D.Hays, Ph.D.
Economic evaluation of health programmes Department of Epidemiology, Biostatistics and Occupational Health Class no. 13: Cost-benefit analysis – Part 2.
ITEC6310 Research Methods in Information Technology Instructor: Prof. Z. Yang Course Website: c6310.htm Office:
Chapter 16 Data Analysis: Testing for Associations.
1 Health outcome valuation study in Thailand Sirinart Tongsiri Research degree student Health Services Research Unit, Public Health & Policy Department.
Using a Discrete Choice Experiment to Value the EQ-5D-5L in Canada Nick Bansback Assistant Professor School of Population and Public Health, University.
METHODS Sample: The Institute for Survey Research of Temple University conducted face-to-face interviews for the 1995 National Alcohol Survey (NAS). The.
Question paper 1997.
EQ-5D and SF-36 Quality of Life Measures in Systemic Lupus Erythematosus: Comparisons with RA, Non-Inflammatory Disorders (NIRD), and Fibromyalgia (FM)
HERU is supported by the Chief Scientist Office of the Scottish Executive Health Department and the University of Aberdeen. The author accepts full responsibility.
IMPORTANCE OF STATISTICS MR.CHITHRAVEL.V ASST.PROFESSOR ACN.
Scaling Session Measurement implies assigning numbers to objects or events. In our case, the numbers “weight” responses to questions, so that saying “Yes”
Values Lower Than Death Jan J. v. Busschbach, Ph.D. –Erasmus University Rotterdam institute for Medical Technology Assessment (iMTA) PO box DR.
The measurement and comparison of health system responsiveness Nigel Rice, Silvana Robone, Peter C. Smith Centre for Health Economics, University of York.
Effect of framing of death on health state values obtained from DCEs Dr. Esther W. de Bekker-Grob by Jonker, de.
A. Strategies The general approach taken into an enquiry.
1 Are values cultural determined…..  Many believe that QoL is cultural determined  One of the starting points of the EuroQol group.
Exercise as a Substitute for Sensation Seeking Michele Burbage, Hope Esposito, & Amelia Leonard University of Cincinnati.
Dr.Rehab F.M. Gwada. Measures of Central Tendency the average or a typical, middle observed value of a variable in a data set. There are three commonly.
Feng Xie Department of Clinical Epidemiology and Biostatistics McMaster University.
Methods of multivariate analysis Ing. Jozef Palkovič, PhD.
Developing preference-based measures for diabetes: DHP-3D and DHP-5D
Patient Baseline Assessment
Effect of framing of death on health state values obtained from DCEs
Measuring outcomes Emma Frew October 2012.
Elicitation methods Health care demands exceed resource supply
Presentation transcript:

Utilising rank and DCE data to value health status on the ‘QALY’ scale using conventional and Bayesian methods John Brazier and Theresa Cain with Aki Tsuchiya and Yaling Yang Health Economics and Decision Science, ScHARR, Health Economics and Decision Science, ScHARR, University of Sheffield, UK University of Sheffield, UK Prepared for the CHEBS Focus Fortnight

Outline  Concerns with current cardinal methods for valuing health states  Problems in using ordinal data  Application of rank and DCE methods to valuing Asthma health states using conventional methods  Application of Bayesian methods to analysing DCE data  Implications for research and policy

Problems with cardinal methods for valuing health states TTO and SG seen to be cognitively complex tasks that may be too difficult for some (e.g children, very elderly) TTO and SG seen to be cognitively complex tasks that may be too difficult for some (e.g children, very elderly) TTO values contaminated by time preference, standard gamble by risk attitude and rating scales by end point bias (among other things) TTO values contaminated by time preference, standard gamble by risk attitude and rating scales by end point bias (among other things)  Role for ordinal methods (rank and discrete choice)

Ordinal tasks: Ranking and discrete choice experiments Ranking respondents asked to order a set of health states from best to worst - traditionally used as a warm up exercise prior to VAS/SG/TTO based preference elicitation Ranking respondents asked to order a set of health states from best to worst - traditionally used as a warm up exercise prior to VAS/SG/TTO based preference elicitation Discrete choice experiments (DCE) - typically asks respondents to choose between two health states (A and B) Discrete choice experiments (DCE) - typically asks respondents to choose between two health states (A and B)

Problems with using ordinal data to value health for QALYs DCE and rank models estimate a latent health state utility value, but with arbitrary anchors DCE and rank models estimate a latent health state utility value, but with arbitrary anchors QALYs require health states to be valued on the full health (one) and being dead (zero) scale QALYs require health states to be valued on the full health (one) and being dead (zero) scale Key problem is linking results of DCEs to the full health-dead scale Key problem is linking results of DCEs to the full health-dead scale

Previous work using ordinal data Ranking Early application of Thurstone’s method by Kind (1982) Early application of Thurstone’s method by Kind (1982) Use of conditional logit on rank data by Salomon (2003) on EQ-5D and McCabe et al (2005) on SF-6D and HUI2 – some success Use of conditional logit on rank data by Salomon (2003) on EQ-5D and McCabe et al (2005) on SF-6D and HUI2 – some successDCE DCE applications in health economics mainly concerned with relative weight of different attributes of health care rather than to valuing health per se DCE applications in health economics mainly concerned with relative weight of different attributes of health care rather than to valuing health per se DCE considered unsuitable for assessing cost effectiveness (because utility scale is not comparable between studies) DCE considered unsuitable for assessing cost effectiveness (because utility scale is not comparable between studies)

Past attempts to apply DCE to valuing HRQoL Hakim and Pathak (1999) applied DCE to valuing EQ-5D states Hakim and Pathak (1999) applied DCE to valuing EQ-5D states - used ‘pick one’ from 12 choice sets (each containing 3 states plus dead) - Exploratory and did not produce weights McKenzie et al (2001) estimated weights for asthma symptoms McKenzie et al (2001) estimated weights for asthma symptoms - no link to full health-dead scale Viney et al (2004) included attributes for HRQoL and survival – but did not estimate health state values Viney et al (2004) included attributes for HRQoL and survival – but did not estimate health state values

Alternative approaches to using DCE The latent utility scale needs to be anchored on the full health-dead scale and there are a number of different ways: The latent utility scale needs to be anchored on the full health-dead scale and there are a number of different ways: Value PITS state externally by TTO/SG (Ratcliffe and Brazier, 2005) Value PITS state externally by TTO/SG (Ratcliffe and Brazier, 2005) Include a dead state in the pair wise choice set* Include a dead state in the pair wise choice set* Using the question ‘is this a state worth living’ in the best-worst scaling method (Flynn et al, 2005) Using the question ‘is this a state worth living’ in the best-worst scaling method (Flynn et al, 2005) * Method used in this study

Background to AQLQ study  Asthma Quality of Life Questionnaire (AQLQ) developed by Professor Juniper is a condition specific measure with 32 questions with 7 levels each covering 4 dimensions  A simplified health state classification was developed from the AQL-5D based on a sample of items on 5 domains: concern, breathlessness, pollution and environment, sleep and activity

AQL-5D  Feel concerned about having asthma [1]None of the time [2]A little or hardly any of the time [3]Some of the time [1]None of the time [2]A little or hardly any of the time [3]Some of the time [4]Most of the time [5] All of the time [4]Most of the time [5] All of the time  Feel short of breath as a result of asthma [1]None of the time [2]A little or hardly any of the time [3]Some of the time [1]None of the time [2]A little or hardly any of the time [3]Some of the time [4]Most of the time [5] All of the time [4]Most of the time [5] All of the time  Experience asthma as a result of air pollution [1]None of the time [2]A little or hardly any of the time [3]Some of the time [1]None of the time [2]A little or hardly any of the time [3]Some of the time [4]Most of the time [5] All of the time [4]Most of the time [5] All of the time  Asthma interferes with getting a good night’s sleep [1]None of the time [2]A little or hardly any of the time [3]Some of the time [1]None of the time [2]A little or hardly any of the time [3]Some of the time [4]Most of the time [5] All of the time [4]Most of the time [5] All of the time  Overall, the activities I have done have been limited [1] Not at all [2] A little [3] Moderate or some [1] Not at all [2] A little [3] Moderate or some [4] Extremely or very [5] Totally [4] Extremely or very [5] Totally

Health state Feel concerned about having asthma some of the time [3] Feel concerned about having asthma some of the time [3] Feel short of breath as a result of asthma a little or hardly any of the time [2] Feel short of breath as a result of asthma a little or hardly any of the time [2] Experience asthma symptoms as a result of air pollution some of the time [3] Experience asthma symptoms as a result of air pollution some of the time [3] Asthma interferes with getting a good night’s sleep most of the time [4] Asthma interferes with getting a good night’s sleep most of the time [4] Overall, totally limited with all the activities done [5] Overall, totally limited with all the activities done [5]

Valuation survey: sampling and interview Representative sample of adult general population invited to participate Representative sample of adult general population invited to participate At the interview: Ranked health states from best to worst (7 AQLQ health states, full health (i.e. best AQLQ state), the worst AQLQ state and immediate death) Ranked health states from best to worst (7 AQLQ health states, full health (i.e. best AQLQ state), the worst AQLQ state and immediate death) Time trade-off (York MVH variant) of 8 AQLQ health states against shorter time in full health Time trade-off (York MVH variant) of 8 AQLQ health states against shorter time in full health 100 health states valued in this way 100 health states valued in this way

Methods: postal follow-up Approx 4 weeks after interview respondents received DCE questionnaire in post Approx 4 weeks after interview respondents received DCE questionnaire in post Optimal statistical design for DCE based upon level balance, orthogonality and minimum overlap was produced by programme in SAS (Huber and Zwerina, 1996) Optimal statistical design for DCE based upon level balance, orthogonality and minimum overlap was produced by programme in SAS (Huber and Zwerina, 1996) 12 pair wise comparisons were produced and randomly allocated to two versions of questionnaire with 6 choices in each 12 pair wise comparisons were produced and randomly allocated to two versions of questionnaire with 6 choices in each Two additional pairs presented to respondents containing with AQL-5D states vs. dead. Two additional pairs presented to respondents containing with AQL-5D states vs. dead.

Discrete choice question

Statistical model for rank and DCE data General model: µ ij = f(ß’x ij + ΦD+u ij ) General model: µ ij = f(ß’x ij + ΦD+u ij ) Where µ ij is the latent utility function of respondent i for state j denotes dimension α=3, level λ = 2. x is a vector of dummy explanatory variables for each level of each dimension of the classification. For example, x 32 denotes dimension α=3, level λ = 2. D is a dummy variable for the state of being dead which takes the value 1 for being dead or otherwise zero. D is a dummy variable for the state of being dead which takes the value 1 for being dead or otherwise zero.

Modelling health state values Modelling: TTO: individual level model (random effects) TTO: individual level model (random effects) DCE: random effects probit model DCE: random effects probit model Ranking: rank ordered logit model Ranking: rank ordered logit modelRescaling: Re-scale by dividing ß coefficients on each dimension level by the coefficient for being dead. Re-scale by dividing ß coefficients on each dimension level by the coefficient for being dead. These rescaled coefficients provide predictions for health state values on the same scale as TTO valuations although the predicted values for health states may not necessarily be the same as those obtained using the TTO technique. These rescaled coefficients provide predictions for health state values on the same scale as TTO valuations although the predicted values for health states may not necessarily be the same as those obtained using the TTO technique.

Results of valuation survey Rank/TTO interview:  308 respondents (response rate 40% )  Representative in terms of gender, age, education  2455 TTO valuations across 100 health states DCE  168 returned questionnaires (response rate 55%)  1336 pair wise comparisons

Results - impact of dimension level on TTO scores (Individual level Random Effects model with main effects)  Concern *  Concern *  Concern *  Concern *  Breath  Breath *  Breath *  Breath * * statistically significant in 0.05 level Dependent variable: TTO values Dependent variable: TTO values MAE = MAE =  Pollution  Pollution  Pollution *  Pollution *   Sleep  Sleep  Sleep *  Sleep *  Activity  Activity *  Activity *  Activity *

Comparison of ßs

Spearman rank Correlations (n=100) TTO pred Rank pred DCE pred Rank pred..918 DCE pred TTO Observed

Predicted health state valuations

Comparisons of models TTO RE Rank DCE warm DCE cold Negative ßs 20/2020/2015/2019/20 Inconsistencies0031 MAE/MAD > Mean error/difference Scale range

Overall comparison TTO model predicts observed TTO values best (lowest MAE) TTO model predicts observed TTO values best (lowest MAE) Rank model predicts observed TTO values nearly as well as TTO model Rank model predicts observed TTO values nearly as well as TTO model DCE model is associated with largest difference from observed TTO values and seems to have a steeper gradient (i.e. more extreme values) DCE model is associated with largest difference from observed TTO values and seems to have a steeper gradient (i.e. more extreme values)

Research questions 1. Is DCE really easier than TTO/SG or VAS? 2. Does DCE produce different estimates from TTO and SG? 3. Theoretical basis for using DCE rather than conventional TTO or SG 4. Basic DCE design issues 5. Analysis – mixed logit or Bayesian models 6. Does the dead dummy solve the problem?

Does including dead solve the problem? A more natural solution is to include survival as an attribute – but this has a multiplicative relationship to QoL and so would require a far larger design A more natural solution is to include survival as an attribute – but this has a multiplicative relationship to QoL and so would require a far larger design Using ‘dead’ requires the ‘pits’ health state of the classification to be considered worse than dead by some respondents – so not suitable for milder classifications Using ‘dead’ requires the ‘pits’ health state of the classification to be considered worse than dead by some respondents – so not suitable for milder classifications What about those who do not think any state is worse than dead (85% in this sample)? What about those who do not think any state is worse than dead (85% in this sample)? For those who do not think any state is worse than dead, then their data tells us nothing about their strength of preference for QoL compared to quantity of life For those who do not think any state is worse than dead, then their data tells us nothing about their strength of preference for QoL compared to quantity of life Are the 85% all none traders? SF-6D (67%), HUI3 (33%) and EQ-5D (14%) Are the 85% all none traders? SF-6D (67%), HUI3 (33%) and EQ-5D (14%)