HTA - lecture 6 - quality of life (B)

Question 1

methods to estimate utility values from disease-specific questionnaires

Answer 1

1. Valuing health states
   VAS/TTO/SG using disease specific health states
2. Mapping
   Predict EQ-5D* values based upon responses on the disease-specific measure
   *or any other generic questionnaire

Question 2

econometric modelling

Answer 2

Regression analysis:
Predict utility score (y) based on the levels of each dimension (x1, x2, xn)
y = β0+β1x1+β2x2+…+ βnxn

à Inconsistent results are possible
à More health states need to be valued
à But prediction errors are smaller

Question 3

multi - attribute utility theory

Answer 3

1. Determine the relative preference of each level within one dimension
2. Determine the relative preference of each dimension
   o What is most important for health? Mental? pain?
3. Combine these preferences in one model
   o Estimate utility score for all possible health states

à Ensures consistent results; it is impossible to give a bad outcome a higher score than a less worse outcome
à Fewer health states need to be valued
à But larger prediction errors
à Requires structural independence between dimensions

Question 4

generic

Answer 4

enables comparison across health conditions –> desirable for reimbursement decisions

Question 5

disease specific

Answer 5

more specific questions for the disease of interest –> better able to measure quality of life –> assumption! needs to be tested

Question 6

reliability

Answer 6

degree to which the measurement is free from measurement error

the extent to which a measure provides the same results on repeated measurements, assuming that the characteristics being measured do not change

Question 7

validity

Answer 7

degree to which the instrument measures the construct(s) it purports to measure

Question 8

responsiveness

Answer 8

ability of a measure to detect changes in health

Question 9

content validity

Answer 9

Extent to which the content of the questionnaire is adequate for the specific disease
Which domains are most important for them: that is the question asked at patients in the study of sclerosis

Related to content, SF-6D is best

Question 10

construct validity

Answer 10

The degree to which scores on an instrument are consistent with hypothesis

Question 11

convergent validity

Answer 11

correlation with similar measures (for example other HRQOL instruments)

Question 12

discriminant validity

Answer 12

differences between known groups (for example severity)

The degree to which scores on an instrument are consistent with hypothesis

Question 13

feasibility

Answer 13

if someone does not fill in the whole questionnaire it is not valid as well. So you need to miss as little data as possible

Question 14

floor effects

Answer 14

substantial group of patients reporting lowest possible score

Question 15

ceiling effects

Answer 15

substantial group of patients reporting best possible score