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 pay Difficult… –Paired comparisons –DCE etc

Visual Analogue Scale VAS –Also called “category scaling” –From psychological research “How is your quality of life?” “X” marks the spot –Rescale to [0..1] Different anchor point possible: –Normal health (1.0) versus dead (0.0) –Best imaginable health versus worse imaginable health Dead Normal health X

Time Trade-Off TTO Wheelchair –With a life expectancy: 50 years How many years would you trade-off for a cure? –Max. trade-off is 10 years QALY(wheel) = QALY(healthy) –Y * V(wheel) = Y * V(healthy) –50 V(wheel) = 40 * 1 V(wheel) =.8

Standard Gamble SG Wheelchair Life expectancy is not important here How much are risk on death are you prepared to take for a cure? –Max. risk is 20% –wheels = (100%-20%) life on feet –V(Wheels) = 80% or.80

Willingness to pay Cost benefit analyis Revealed preferences –Look in market how much subject are willing to pay –Different situations give different results –Weighted by in income Conclusion: –the validity of cost benefit analysis is not sufficient

Alternatives for indirect measurements 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

Validated questionnaires

The Rosser & Kind Index

The Rosser & Kind index Criticism on the Rosser & Kind index –Sensitivity (only 30 health states) New initiatives –Higher sensitivity (more then 30 states) EuroQol Group –EQ-5D-3L and the EQ-5D-5L McMaster University –Health Utility Index 2 & 3 SF-36 –SF-6D

Health Utility Index Developed from pediatric care –Strong proxy versions Symptom driven: –“Outside the skin” instead of “inside the skin” EQ-5D: “problems with daily activity” HUI: “Unable to read ordinary newsprint…” Commercial –All user have to pay 35 Translations

HUI 2

HUI 3

Increasing number of health states

No longer value all states Impossible to value all health states –If one uses more than 30 health states Estimated the value of the other health states with statistical techniques –Statistically inferred strategies Regression techniques EuroQol, Quality of Well-Being Scale (QWB) –Explicitly decomposed methods Multi Attribute Utility Theory (MAUT) Health Utility Index (HUI)

Statistically inferred strategies Value a sample of states empirically Extrapolation –Statistical methods, like linear regression –11111 = 1.00 –11113 =.70 –11112 = ?

Statistically inferred strategies EuroQol –EQ-5D: 5 dimensions of health –245 health states Quality of Well-Being scale (QWB) –4 dimensions of health –2200 health states plus 22 additional symptoms SF-36 –SF-6D: 6 dimensions of health – health states

Explicitly Decomposed Methods Value dimensions separately –Between the dimensions –What is the relative value of: Mobility…... 20% Mood……….. 15% Self care.….24% Value the levels –Within the dimensions –What is the relative value of Some problems with walking…..80% Much problems with walking……50% Unable to walk………..……………….10% = 1 - (0.20 x ( )) = 0.96

Explicitly Decomposed Methods Combine values of dimensions and levels with specific assumptions –Multi Attribute Utility Theory (MAUT) Mutual utility independence Structural independence

Explicitly Decomposed Methods Health Utilities Index (Mark 2 & 3) –Torrance at McMaster –8 dimensions –Mark 2: health states –Mark 3: health states The 15-D –Sintonen H. –15 dimensions –3,052,000,000 health states (3 billion)

More health states, higher sensitivity ? (1) EuroQol criticised for low sensitivity –Low number of dimensions Development of EQ-5D plus cognitive dimension –Low number of levels (3) Gab between best and in-between level

More health states, higher sensitivity ? (2) Little published evidence –Sensitivity EQ-5D < SF-36 Compared as profile, not as utility measure –Sensitivity EQ-5D  HUI Sensitivity  the number of health states –How well maps the classification system the illness? –How valid is the modelling? –How valid is the valuation?

More health states, more assumptions General public values at the most 50 states The ratios empirical (50) versus extrapolated –Rosser & Kind1:1 –EuroQol1:5 –QWB1:44 –SF-361:180 –HUI (Mark III)1:19,400 –15D1:610,000,000 What is the critical ratio for a valid validation?

Conflicting evidence sensitivity SF-36 Liver transplantation, Longworth et al., 2001

SF-36 as utility instrument Transformed into SF6D SG N = 610 Inconsistencies in model – health states –regression technique stressed to the edge Floor effect in SF6D

Collapsing levels SF-6D Many levels are taken together –If PF=2decrement: –If PF=3decrement: –If RL=2decrement: –If RL=3decrement: –If RL=4decrement:

SF-6D loses a lot of levels Levels clas.system and actual levels –PF6 5 –RL 42 –SF 55 –PN 65 –MH 54 –VI 53 Levels in clas. system: –6x4x5x6x5x5 Actual levels: 480 –5x2x5x5x4x3

Some levels in the SF-6D do not work…

EQ-5D Strong punts –Very sensitive in the low –Measures subjective burden (inside the skin) –Low administrative burden –Many translations –Cheap –Most used QALY questionnaire –Most international validations Weak points –Only there levels per dimensions –Insensitive in the high regions

HUI Strong punts –Sensitive –Measures objective burden (outside the skin) –Well developed proxy versions –Well developed child versions Weak points –Expensive –Only a few valuation studies

SF-6D Strong punts –Probably sensitive in the high regions –Often already include in trials (SF-36) –Many translations Weak points –Insensitive in the low regions –Only a few validation study –Might be expensive

Conclusions More states  better sensitivity The three leading questionnaires –have different strong and weak points