Issues in the measurement and valuation of health

1
Issues in the measurement and valuation of health

• Paul Kind
• Principal Investigator
• Outcomes Research Group
• Centre for Health Economics
• University of York
• York, England

2
Evaluation
Type of economic evaluation ?
Cost-utility
Cost-effectiveness
QALY
Outcome measure
Health status
Natural units BP Weight etc
EQ-5D
Survival / Life expectancy
HUI 15-D QWB ......
3
HrQoL and public policy analysis

• The principal problem is the mismatch between
  preference-based HrQoL and available data
• Primary data sources typically lack HrQoL data in
  a viable form
• Need to adapt
• Crosswalks between data source and HrQoL measure
• General problem commonly encountered in clinical
  trials

4
Quality-adjusted life expectancy
5
Existing public policy applications

• Non-fatal road traffic accidents Sweden
• Helicopter evacuation - Netherlands
• Riskbenefit analysis of food (ILSIE) - EU
• Health and Safety Executive - UK
• Environmental hazard of lead UK
• Health impact of changes in retirement age UK

6
Design issues

• All HrQoL instruments comprise two related
  component
• Descriptive system
• Operationalises the underlying concept of health
• Always involves a trade-off between
  comprehensiveness / complexity and practice
• Valuation system
• Assigns values to elements of the descriptive
  system according to some rule
• Valuation starts with the specification of the
  descriptive system
• exclusions count gt they have an imposed zero
  weight

7
Fundamental questions
X

• How determined ?
• Arbitrary assignment
• Mortality rates
• Karnofsky gt SF-36
• Explicit preference elicitation
• Paired comparisons
• Ranking
• Rating (VAS)
• WTP
• TTO
• SG
• How aggregated ?

• Whose descriptions / values ?
• Patients
• Providers
• (tax)Payers
• Politicians
• Health economists

8
Value and valuation

• How much value do is placed on each dimension ?
• How much value is attached to level changes
  within dimension ?
• How should / might we obtain estimates of those
  values ?
• Who should determine those values ?

9
Measuring value for quality-adjustment

• There is no compelling case for the exclusive use
  of utility weights as the quality-adjustment
  factor in computing QALYs
• Any system could be used so long as it conforms
  with key measurement desiderata
• Single index
• Generic
• 0 1 metric
• Values represents relevant constituency of
  interests

• There is no standard method for eliciting utility
  weights
• TTO is at best a proxy for SG
• VAS-based methods fail the trade-off
  requirement
• All variants yield different numerical estimates
• There is no ex post test of utility status
• conferred by virtue of elicitation method

10
Valuing EQ-5D health states

• The rank order and value of (EQ-5D) health states
  is generally not known
• The EuroQoL Group adopted a 20cm graduated visual
  analogue scale (VAS) as its standard valuation
  method
• The VAS scale is calibrated so that 100 and 0 are
  assigned respectively to the best and worst
  imaginable health state

• The location and value assigned to dead is not
  pre-defined
• All national studies conducted by EuroQoL Group
  members include VAS
• Local experimentation is encouraged different
  valuation methods such as ranking, paired
  comparisons, magnitude estimation, Standard
  Gamble, Time Trade-Off, PTO

11
Values estimates computed using paired
comparisons model
12
Strengths and weaknesses

• 18 years of RD
• Huge volume of survey and clinical trial data
• Gets the job done !
• 60 seconds
• Little missing data
• Suitable low-cost candidate for new prospective
  data acquisition
• Calibrated in terms of US social preferences
• Normative population data

• Conflated dimensions
• Pain/discomfort
• Anxiety/depression
• Asymmetric levels
• text descriptions
• examples (3)

13
Improving HrQoL measurementGeneral

• Developmental research underpinning HrQoL
  measures is seldom revisited
• Review basis of descriptive systems
• Seek endorsement of components from the
  relevant constituency
• Treatment and valuation of dead
• Non-trivial issue assumed away by transforming
  observed valuation data
• Investigate valuation stability over time
• Do valuations remain unchanged ?

14
Improving HrQoL measurementEQ-5D specific

• Increasing the number of levels
• Decision in principle to extend the number of
  levels from 3 to 5
• Empirical investigation over past 3 years
  including several population studies
• Extending coverage to children
• Child-friendly version already in existence with
  normative results for UK
• Concurrent studies in Sweden, Germany

• Facilitating data capture
• Web-based solutions
• Computer-assisted interview methods
• Improving standard valuation protocol
• Develop more appropriate/efficient survey methods
  (non-postal)
• Establish additional preference elicitation
  methods as adjunct to VAS