1 Health outcome valuation study in Thailand Sirinart Tongsiri Research degree student Health Services Research Unit, Public Health & Policy Department LSHTM Supervisor: Professor John Cairns 17 November 2006
2 Outline Introduction Research question Objectives Methods Budget & Timetable Conclusion
3 Introduction Resources are limited Market failures in Health Economic Evaluation ICER = Cost Outcome
4 Cost-utility analysis (CUA) Outcome in CUA Quality-adjusted Life Year Impact on health: Quality of life & Quantity of life Compare across different health interventions
5 Quality-Adjusted Life Year (QALY) Quality of life (QoL) life expectancy Before treatment After treatment Health interventions 0 1 Q0 Q1 T0 T1 QALY gain = Q1T1 – Q0T0
6 Recommendations from the NICE and the US Panel on Cost- Effectiveness in Health and Medicine A tariff estimated from the general population No tariff estimated from the Thai general population
7 A national tariff for preference-based health measure: Why? UK = Denmark = Zimbabwe = Japan = Thailand ?
8 Research question: A tariff for health outcomes from the Thai perspective
9 How to elicit preferences over health states? Torrance (1986) Prepare health state descriptions Selection of subjects Use a utility measurement instrument
10 Health Complex Encompass many dimensions Individuals perceive differently A number of generic health descriptive systems, e.g. the EQ- 5D, the SF-36 and the HUI The EQ-5D will be used in the research
11 The EQ-5D 5 Dimensions - Mobility - Self-care - Usual activities - Pain and Discomfort - Anxiety and Depression 3 Levels - No problem - Some problems - Severe problems
12 The EQ-5D 5 Dimensions - Mobility - Self-care - Usual activities - Pain and Discomfort - Anxiety and Depression 3 Levels - No problem - Some problems - Severe problems 11223
13 The EQ-5D 5 Dimensions - Mobility - Self-care - Usual activities - Pain and Discomfort - Anxiety and Depression 3 Levels - No problem - Some problems - Severe problems 243 health states
14 Problem 1: Is the EQ-5D an appropriate tool to capture a concept of “health” in the Thai population?
15 Preference, Utility, Value What different between these terms? Different methods to derive preferences, e.g. VAS, SG and TTO Different methods give different values
16 Assumption A fully informed rational person is the best judge of one’s own welfare Individual utility can be aggregated and comparable. An interval scale is needed
17 An interval scale The difference between score 20 and 10 (10) is equal to the difference between 30 and 20 (10). The difference between the state with score 0.4 and 0.3 (0.1) is equal to the difference between the state 0.6 and 0.5 (0.1)
18 How to “quantify” preference? Health states ranking, the VAS and the TTO methods will be used to elicit preferences of respondents in the study
19 Whose preferences should be elicited? Patients or Population Population Aim to use in decision making at the societal perspective Generalizability
20 Debates Whose values should be counted? Preferences are “elicited” or “constructed”? Preferences are “labile”. Simple Heuristics Framing and labelling effects
21 Lenert et al. 2000
22 Problem 2: Do the elicitation methods appropriate for the Thai population?
23 Pre-pilot study in London
24 Pre-pilot study in London
25 Cognitive burden How to minimize cognitive burden of Thai respondents?
26 Respondents can value not more than 13 health states How all 243 health states will be scored?
27 Problem 3: Existing statistical models from various countries Do these models fit with preferences observed from the Thai population? What is an appropriate model for the Thai population?
28 Thailand A majority of population is Buddhist Religious belief 1 : the perfect health in this life guarantee the perfect health in next life Religious belief 2: inferior health results from bad kamma from previous life (no preferences on different inferior health) Does these beliefs influence TTO?
29 Are Buddhism beliefs influence preferences on health of the Thai general population ? The study by Chirawatkul (2005)
30
31 Objectives 1. Elicit preferences on health states from a Thai general population 2. Identify appropriate statistical models to explain respondents’ preferences over health states 3. Whether the Thai EQ-5D adequate to capture health concept of the Thai general population
32 Methods Objective 1: - Health states ranking - Visual Analog Scale - Time trade-off Pre-testing and piloting the survey questionnaire and process
33 Pre-testing the questionnaire What, from Thais, are “usual activities”, “self-care”?
34 Pilot interview To test the interview procedure Cognitive burden
35 Sample Randomly selected from the Thai general population Household registration database The National Statistical Office, Thailand Health Welfare Survey (addresses and maps are included) Regional level: 5 provinces in the central region
36 1 region (Central region) 5 provinces: Ratchaburi, Phetchaburi, Nakorn-Nayok, Nakorn-Pathom and Prachuab Kirikhan Multi-stage sampling Sample size = 1,000
37 Interview procedure Replicate from the Measurement and Valuation of Health (MVH) study in the UK (Dolan et al 1995)
38 Interview procedure Complete the EQ-5D with own health Ranking own health Ranking 15 different health states Score each state using the VAS Score each state using the TTO Personal details: age, gender, education and socioeconomic status
39 Thai EQ-5D questionnaire Mobility Self-care Usual activities Pain/Discomfort Anxiety/Depression
40 Thermometer scale The best health imagination The worst health imagination Your health today
41 Example of health state card ข้าพเจ้า ไม่สามารถไปไหนได้และจำเป็นต้องอยู่บนเตียง มีปัญหาในการอาบน้ำหรือการแต่งตัวบ้าง ไม่สามารถทำกิจกรรมที่ทำเป็นประจำได้ ไม่มีอาการเจ็บปวดหรืออาการไม่สุขสบาย รู้สึกวิตกกังวลหรือซึมเศร้ามากที่สุด Moderate ข้าพเจ้า มีปัญหาในการเดินบ้าง ไม่สามารถอาบน้ำหรือแต่งตัวด้วยตนเองได้ มีปัญหาในการทำกิจกรรมที่ทำเป็นประจำอยู่บ้าง มีอาการเจ็บปวดหรืออาการไม่สุขสบายมากที่สุด รู้สึกวิตกกังวลหรือซึมเศร้าปานกลาง Severe 23232
42 Health states ranking anchor anchor Bisection method
43 Time trade-off question 1. Imagine that you live in a state for 10 years and die 2. If you can choose to live in healthy life and die sooner than 10 years, how many years you would sacrifice? Preference is subjective. To compare preference between states, Years of life in perfect health will be compared The shorter duration in perfect health, the less preferred state (use years of life to “buy” a better state)
44 Time trade-off score transformation Duration of life (yrs) Health status X Preference = x 10 Better than death
45 Time trade-off score transformation Duration of life (years) Health status 10 X 1 0 Value of health state: -x (10-x) Worse than death
46 Statistical Modelling To estimate preferences for 243 health states from the observational data of 42 health states Econometrics methods Use existing models to fit new data STATA 9
47 Estimate preference from TTO Better health states have higher preferences is “better” than Overall preference is the result of the addition of sub-preference in each dimension
48 Example of models Dolan 1997 R 2 = 0.46 Mean absolute difference = 0.46 Dolan et al 2002 R 2 = 0.55 Mean absolute difference = 0.03
49 Estimate preference from health states ranking Salomon (2003) Parameters are predicted using the conditional logit regression model
50 Timetable ActivitiesWhen? 1. Proposal, budget and questionnaire preparation November 2006 – January Preference elicitation interview February – June Identify appropriate modelling to predict preferences from TTO and VAS July - August Qualitative surveySeptember – December 2007
51 Budget Preparation 10,000 baht Preference interview 737,000 baht Qualitative survey 18,400 baht Total: 765,400 baht
52 Potential funding organizations The International Health Policy and Programs, Thailand The Health promotion for the Disabled project, Thailand
53 Conclusion Can the EQ-5D health description system capture the concept of health in Thais? A tariff of the Thai EQ-5D to be used in the economic evaluation in Thailand How existing models can fit the new data
54 How Buddhist beliefs influence preference on health states Contribution of preference scores to a new version of the EQ-5D
55 Acknowledgement Prof. John Cairns Louise Longworth Dr.Viroj Tangcharoensathien Dr.Wachara Riewpaiboon My fellow PhD students My family