Health Economics II –2010 Health Economic Evaluations Part III Lecture 2 Cost-effectiveness analysis QALYs and cost-utility analysis Nils-Olov Stålhammar

Type of analysis - the outcome measurement Type of analysis Costs Effect Cost-minimization Money Not measured Cost-effectiveness Money Natural units; Life years gained, relapses Cost-consequence Money Several disease specific measures Cost-utility Money Combining length and quality of life, QALYs Cost-benefit Money Money

Cost-effectiveness analysis CEA is of most use when operating with a given budget and there is a limited range of options within a given field However, the measure of effectiveness must be appropriate with respect to the objective But can also guide broader resource allocation, but with a ‘narrow’ measure of effectiveness (compared to CUA more is left to the decision maker’s judgment )

Decision rules of CEA/CUA Average cost-effectiveness ratio CERa=Ca/Ea CERb=Cb/Eb Incremental cost-effectiveness ratio ICER= C b a E

Decision rules of CEA/CUA Possible outcomes in CEA(CUA) Comparison of two alternatives, A and B Alternative B may fall in area I, II, III or IV Cost IV Higher cost Poorer effect I Higher cost Better effect A CA III Lower cost Poorer effect II Lower cost Better effect Effect EA

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996* Assumptions Three patient groups of equal size, 1000 homogeneous pts per group Mutually exclusive treatments are available, i.e. a patient can only receive one of the treatments Constant returns to scale, i.e. C and E per patient are independent of the number of patients treated C and E in one group are not dependent on choice of treatment in another group Fixed budget, i.e. a given amount of resources is available *) Karlsson, Johannesson. The decision rules of cost-effectiveness analysis. Pharmacoeconomics, 1996;9:113-20

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 Patient group I Patient group II Patient group III C E C/E A 100 10 F 200 12 17 K 5 20 B 14 G 400 16 25 L 8 300 19 H 550 18 31 M D 21 500 Average cost-effectiveness ratios do not provide relevant information Incremental cost-effectiveness ratios have to be calculated!

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 ICER is estimated between treatment alternatives within the same patient group, i.e. a comparison with the next most effective mutually exclusive alternative Patient group I Patient group II Patient group III ΔC ΔE ΔC/ΔE A 100 10 F 200 12 17 K 5 20 B 4 25 G 50 L 3 33 C 2 H 150 75 M D E 1

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 Dominance A treatment is dominated if it is more costly and less effective than another treatment; d is dominated by a Extended dominance A treatment can also be dominated when its ICER is higher than that of the next more effective treatment; b should be ruled out because of extended dominance C c A combination of a and c, i.e. some patients get a and some patients get c, will give more effectiveness at the same cost or same effectiveness at a lower cost, than b alone b d a E

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 ICER is estimated between treatment alternatives within the same patient group, i.e. a comparison with the next most effective mutually exclusive alternative Patient group I Patient group II Patient group III ΔC ΔE ΔC/ΔE A 100 10 F 200 12 17 K 5 20 B 4 25 G 50 L 3 33 C 2 H 150 75 M D E 1

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 Dominated alternatives are excluded and ICERs are recalculated Patient group I Patient group II Patient group III ΔC ΔE ΔC/ΔE A 100 10 F 200 12 17 K 5 20 B 4 25 G 50 M 7 29 D 40 H 150 2 75 E 1

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 G D M C L F B K A

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 Treatment alternatives to be implemented can be determined by The size of the budget available The maximum price we are willing to pay for a unit of effectiveness will then be determined implicitly or The maximum price we are willing to pay for a unit of effectiveness The size of the budget will then be determined implicitly

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 The budget as the decision rule Budget Treatment MC Effectiveness Group I Group II Group III 0-99 900 Some: 0 Some: A 10 0-9 990 100 000 A 17 10 000 100 200- 299 800 Some: F 10 012- 21 988 300 000 F 20 22 000 300 100- 399 900 Some: K 22 005- 26 995

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 In I some get A some get B All in I get B All in I get A

Decision rules of CEA/CUA Example from Karlsson and Johannesson, 1996 G D M B K F A Area under MC curve = total cost A specific price (threshold) implies a specific budget

Type of analysis - the outcome measurement Type of analysis Costs Effect Cost-minimization Money Not measured Cost-effectiveness Money Natural units; Life years gained, relapses Cost-consequence Money Several disease specific measures Cost-utility Money Combining length and quality of life, QALYs Cost-benefit Money Money

Quality Adjusted Life Years Health index or Utility or QALY weight Without treatment 1 0.9 0.8 With treatment 4 5 Life expectency, years QALYs without treatment = 4 x 0.9 = 3.6 QALYs with treatment = 5 x 0.8 = 4.0

How to estimate QALY-weights Rating Scale, Category Scaling, Visual Analogue Scale Note; respondents are not faced with a ’choice’ SG – Standard Gamble A choice between a certain intermediate health state and a treatment with two possible outcomes, dead and healthy TTO – Time Trade Off A choice between living for a given time in a poor health state and living for a shorter time in full health

P ( 0<P<1) is used as the index of QoL Standard Gamble STATE i HEALTHY DEAD Alternative 1 Probability p Probability 1-p Alternative 2 P ( 0<P<1) is used as the index of QoL

Time Trade Off QoL Alternative 2 Alternative 1 Time t T HEALTHY 1.0 Alternative 1 STATE i hj DEAD 0 Time t T The value of t for which the respondent is indifferent is used to construct an index of QoL for the state i: h = t T

Some quality weights (utilities) for health states Full health (reference state) 1.00 Life with menopausal symptoms (RS) 0.99 Side effects of hypertension treatment (RS) 0.95-0.99 Mild angina (RS) 0.90 Kidney transplant (TTO, patients) 0.84 Moderate angina (RS) 0.70 Hospital dialysis (TTO, Hamilton, patients) 0.59 Hospital dialysis (TTO, St John´s, patients) 0.57 Hospital dialysis (TTO, general public) 0.56 Severe angina (RS) 0.50 Anxious/depressed much of the time (TTO) 0.45 Being blind or deaf or dumb (TTO) 0.39 Dead (reference state) 0.00 Source: Torrance (1987)

Preferences, values and utilities Response method Question framing Certainty (values) Uncertainty (utilities) Scaling Rating scale Category scaling VAS Choice TTO Person trade-off *) SG *) Person Trade Off measurement; How many pts in a certain state of health should have their lives extended by one year in order to be equivalent to extending the lives of 100 healthy individuals with 1 year?

To estimate QALY-weights – some remarks VAS less ’attractive’ from a theoretical point of view SG can be expected to result in higher weights than TTO due to: People are risk averse People have a positive rate of time preference SG may be preferable compared to TTO since it involves uncertainty However, the ability to map utilities can be questioned for both methods Who is to give the weights? Patients know what it is like, but cope General public bear the costs and are potential patients

Multi-attribute health status classification Health states are described with respect to a number of attributes (e.g. mobility, pain..), each with a number of levels Utility scores are derived with VAS, SG or TTO, but necessary to interpolate some of the utilities Composite approach Respondents value a subset of states Econometric modelling is used to give values to all states Decomposed approach Involves asking respondents to value each level within a dimension/attribute assuming that all other dimensions are constant Composite health states are generated with a multi-attribute function Stringent conditions for the multi-attribute function

Multi-attribute health status classification EuroQoL; EQ-5D Five dimensions; mobility, self-care, usual activities, pain/discomfort, anxiety/depression Three levels for each dimension Also a VAS 243 health states plus ‘unconscious’ and ‘dead’ Values derived for different countries, different population groups and with different techniques (VAS and TTO) http://www.euroqol.org/home.html

EQ-5D Mobility Self-Care 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 and 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 Cont…

EQ-5D…Cont Pain/Discomfort Anxiety/Depression 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 extreme anxious or depressed

Multi-attribute health status classification The McMaster Health Utility Index HUI3 has eight attributes; vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain 5-6 levels per attribute Defines 972 000 unique health states Utility scores derived with VAS and SG among 504 adults, Hamilton, Ontario (replicated in French population with similar results) http://www.healthutilities.com/

Guidelines’ views The Dental and Pharmaceutical Benefits Agency, (Tandvårds och läkemedelsförmånsverket, TLV): “QALY weightings based on appraisals of persons in the health condition in question are preferred before weightings calculated from an average of a population estimating a condition depicted for it.” NICE: “Information on changes in HRQL as a result of treatment should be reported directly by patients (and directly by carers when the impact of treatment on the carer’s health is also important). The valuation of changes in HRQL reported by patients should be based on public preferences elicited using a choice-based method in a representative sample of the UK population.”

Alternatives to QALYs HYE - Healthy Years Equivalent The number of years in perfect health that are equivalent to a particular profile of health states over remaining life time SAVE- Saved Young Life Equivalents Based on Person Trade Off measurement; extending lives of 100 healthy individuals with 1 year = extending lives of X patients in designed health state with 1 year DALY – Disability Adjusted Life Years Measures time lost due to premature death (compared to the greatest reported national average life expectancy) and due to disability (scores on a discrete scale with 7 categories, assigned by health care workers as PTO-scores)

Growth in CUA- literature From Neuman PJ, et al. Value in Health 2005;8(1):3-9

Quality of CUA- literature From Neuman PJ, et al. Value in Health 2005;8(1):3-9

’Old’ exam question. Part III a) Give two reasons why QALY is an often recommended measure of outcome in a health economic evaluation. b) Describe in detail VAS, SG and TTO as methods for eliciting preference weights (QALY weights). (Use illustrations.) c) Why can SG be expected to result in higher QALY weights than TTO?

’Old’ exam question. Part III Assume the following: - There are three patient groups with 1000 homogeneous patients per group - A number of mutually exclusive treatments are available for each patient group (see table below), i.e. a patient can only receive one of the treatments - There are constant returns to scale, i.e. cost (C) and effect (E) per patient are independent of the number of patients treated - C and E in one group are independent on choice of treatment in another group - In addition to the 11 alternatives listed in the table, there also exist ‘do nothing’ alternatives which are assumed to have zero cost and zero effectiveness - The fixed budget that can be used for treatment in these three patient groups is 700 000. Cont.

’Old’ exam question. Part III Cont. Patient group I Patient group II Patient group III Alternative C E A 100 10 F 200 12 K 5 B 14 G 400 16 L 8 300 H 550 18 M D 19 500 20 Cont.

’Old’ exam question. Part III Cont. a) Which treatments should be chosen, given that the objective is to maximise the effectiveness of the resources spent on treatment for these 3 patient groups? (You must describe and illustrate the necessary steps in the calculations needed to arrive at your answer.) b) What is the total effectiveness and the marginal cost of producing effectiveness when the budget is being optimally used, i.e. when the budget is allocated in accordance with your answer to question a).

’Old’ exam question. Part III While discounting of costs occurring in the future is generally accepted, discounting of future health effects is sometimes questioned. Describe two often given reasons for not discounting future health effects and give two arguments in favour of discounting future health effects.

’Old’ exam question. Part III Explain the difference between ‘utilities’ and ‘values’. Give a plausible explanation to the common empiric finding that the utility for a specific health state is higher than the corresponding value.

’Old’ exam question. Part III Regarding the estimation of QALYs   a) Describe in detail Visual Analogue Scale, Standard Gamble and Time Trade Off as methods for eliciting preference weights (QALY weights). Support your description by using illustrations. b) An alternative way of measuring the effect of, for instance, a treatment on health outcomes is to use a pre-scored multi-attribute health status classification system. Give a brief description of what this means.

’Old’ exam question. Part III According to an assessment of two treatment alternatives, alternative A results in a cost per patient of 1 MSEK and a life expectancy of 10 years. The corresponding numbers for alternative 2 are 1,5 MSEK and 12 years.   a) Calculate average and incremental cost-effectiveness ratios. b) Is it possible to decide which alternative that is most cost-effective? Explain. c) Suggest cost and effectiveness for a hypothetical third alternative that would be dominated by extended dominance. Explain the concept of extended dominance.