1. A. Introduction to Health Economics
Dr Alan Haycox
Reader in Health Economics
Health Economics Unit
University of Liverpool Management School

2. Introduction to Health Economics – Programme
The programme will be broken down into four sections:
Introduction to health economics
Economics modelling: Theory & Practice
Value of new drugs including new cancer drugs: Scottish Medicines Consortia (SMC) Scotland
Value of new drugs including new cancer drugs: NICE (England)

3. Methods of economic evaluation and other techniques
The four types of health economic evaluation are:
CMA
CEA
CUA
CBA
We will also cover measuring health related
quality of life as well as economic modelling

4. The Four Methods of economic evaluation
Cost Minimisation Analysis (CMA)
Cost Effectiveness Analysis (CEA)
Cost Utility Analysis (CUA)
Cost Benefit Analysis (CBA)

5. Cost Minimisation Analysis (CMA)
Simplest of all methods of economic evaluation
Does not mean benefits are ignored – they have to be proven to be equivalent
Once benefits have been proven to be equivalent, analysis needs only to consider costs

6. Example - CMA of generic formulations and different treatments
Two drugs with exactly the same pharmaceutical components with differing costs, e.g. different formulations of paclitaxel
Two approaches to cancer surgery with similar outcomes but different costs, i.e. one approach more invasive requiring more extensive analgesia

7. Cost-Effectiveness Analysis (CEA)
Health benefits are measured in natural units reflecting a single dominant therapeutic goal
Reduction in blood pressure (treatment)
Increase in cases detected (screening)
CEA is only useful and undertaken if a single dimension dominates the health outcome to be compared

8. Example - CEA of alternative Approaches to cervical screening
How much more does the more effective screening system cost? (incremental costs)
How many more cases are detected by the more effective screening system? (incremental effectiveness)
What is the incremental cost-effectiveness ratio (ICER)?
ICER = incremental cost/incremental effectiveness

9. The Cost-effectiveness Plane
Cost-effectiveness ratio
(additional cost per additional success)
Existing technology
dominates
(-) Incremental costs (+)
Cost-effectiveness ratio
(cost saved per reduced success)
New technology dominates
Figure 9.1 The cost-effectiveness plane
(-) Incremental effectiveness (+)

10. Probability cost-effective
Incorporating cost-effectiveness thresholds (CEAC’s) for decision-making 1
0.8
Probability cost-effective
0.5
Figure 9.4 Cost-effectiveness acceptability curve
0.3
£30,000
£50,000
£20,000
Ceiling ratio

11. Cost Utility Analysis (CUA)
Incorporates the effects on morbidity (quality of life) and mortality (quantity of life)
The most commonly used index is the quality- adjusted life-year (QALY)
A QALY is calculated by aggregating the number of years gained from a health care intervention, weighted by the relative value attached to each future health state
Issues underlying outcome analysis for CUA are explored in detail in the next session

12. Measuring Quality of Life (QoL)
QoL weights reflect the subjective level of wellbeing experienced in different health states; the more preferable a health state the higher will be its associated ‘value’
Perfect health = 1
Death = 0
We will return shortly to the methods used to help determine QoL. In the meantime, give a brief overview to determine QALY gains

13. Preference elicitation methods
There are three main methods for direct measurement used in cost utility analysis.
Visual Analogue Scale (VAS)
Standard Gamble (SG)
Time Trade-off (TTO)

14. Visual Analogue Scale (VAS)
Individuals are asked to indicate where on the line between the best and the worst imaginable health states they would rate a pre-defined health state

15. Your own health state today
VAS
100
We would like you to indicate on this scale how good or bad is your health today, in your opinion. Please do this by drawing a line from the box below to wherever point on the scale indicates how good or bad your current health state is 90 80 70 60 50 40 30 20
Your own health state today 10

16. With probability p: Full health, H2 With probability (1-p): Death, H3
Standard Gamble
Alternative 1:
Health state H1
with certainty
Choice
With probability p: Full health, H2
Figure 10.5 The standard gamble method for a chronic health state preferred to death
Alternative 2:
Gamble
With probability (1-p): Death, H3

17. Standard Gamble board Choice A 90 10 % Chance % Chance PERFECT DEATH
HEALTH
DEATH
Choice B
Figure 10.5 The standard gamble method for a chronic health state preferred to death – visual aid used to help respondents with the standard gamble task
Some problems in moving about
No problems with self-care
No problems with usual activities
Moderate pain or discomfort
Not anxious or depressed

18. The time trade-off method
QOLA=1
Value of health
QOLB
Figure 10.6 The time trade-off method – for a chronic state preferred to death
LOLA
LOLB
Years
QALYB
QALYA =

19. Cost-benefit analysis
This requires all costs and benefits to be measured in the same unit – money
In cost-benefit analysis an activity should be undertaken if the sum of the benefits are greater than the sum of the costs
The difficulties of converting all benefits (pain, anxiety, disability, death) to a monetary equivalence implies that CBA is rarely used in health economic analyses

20. Summary Which tool for which analysis?
What is the context of the analysis?
What is the nature of the comparison being made?
What is the nature of the ‘outcome’ arising from the competing options?

21. Conclusion - What Health Economics aims to achieve
Efficiency: Does the allocation of scarce resources maximise the achievement of health outcomes?
Equity: Is the sharing of health care resources fair between people?
The manner in which we are attempting to
achieve these aims is explored in the following
presentations

22. What is Health Related Quality of Life?
A multi-dimensional concept that encompasses the physical, emotional and social components associated with an illness or its treatment

23. Measuring Health-Related Quality of Life (HRQoL)
Pain P4 X B P3 P2
A = P0D0 = Normal health
B = P4D3 = Total disability
& severe pain
Figure10.2 Measuring health-related quality of life: profile of ill-health, relative ranking and absolute values P1 A P0 X D0
Disability D1 D2 D3

24. What are Q of L ‘weights’?
Such weights reflect the subjective level of wellbeing experienced in different health states; the more preferable a health state the higher will be its associated weight.
Perfect health = 1
Death = 0

25. Measuring Health Gain in Theory1
 Prognosis with intervention
QUALITY
OF LIFE
= Health gain
INTERVENTION
 Prognosis without intervention
Figure 10.3 Measuring the health gain from interventions
DEATH
ONSET OF
ILLNESS
TIME

26. Two ‘types’ of Measure Generic instruments
Designed to have broad application across a wide range of disease states
eg sickness impact profile, Nottingham health profile, EuroQol
Disease specific instruments
designed to assess the impact of specific disease states
eg arthritis impact measurement scale, back pain disability questionnaire

27. Calculating QALYs – A Simple Example
Survival and associated health states
With treatment ‘X’ years in improved health
Without treatment ‘X’ 8 years in
poorer health
Preference weights for health states
With treatment ‘X’ 0.7
Without treatment ‘X’ 0.5

28. QALY Analysis for Treatment ‘X’
With treatment X Survival = 10 years Q of L = 0.7 QALYs = (10 X 0.7) = 7.0
Without treatment X Survival = 8 years Q of L = 0.5 QALY = (8 X 0.5) = 4.0
QALY gain = 3.0 Q.A.L.Y’s ( )
Cost of intervention = £45,000
Cost per QALY = £15,000

29. QALYs – For and against their use in health economic evaluations
Generic multi-dimensional
Easy to apply
Provides practical guidance in allocating health care resources between very different therapeutic interventions
Against
Too superficial to measure the full benefits from health care?
Insufficiently sensitive to capture small changes in the patient’s Q of L
Can we really measure quality of life in only five questions?

30. Measuring gains from different types of intervention
= Health gain
Improved survival
(increased length of life) only
Quality
of life
Figure 10.4 Measuring gains from different types of intervention - IMPROVED SURVIVAL (INCREASED LENGTH OF LIFE) ONLY
Time

31. Measuring gains from different types of intervention
= Health gain
Improved quality of life only
Quality
of life
Figure 10.4 Measuring gains from different types of intervention - IMPROVED QUALITY OF LIFE ONLY
Time

32. Measuring gains from different types of intervention
= Health gain
Improved survival and improved quality of life
Quality
of life
Figure 10.4 Measuring gains from different types of intervention - IMPROVED SURVIVAL AND IMPROVED QUALITY OF LIFE
Time

33. Measuring gains from different types of intervention
= Health gain
= Health loss
Quality
of life
Improved survival at expense of decreased quality of life
Figure 10.4 Measuring gains from different types of intervention - IMPROVED SURVIVAL AT EXPENSE OF DECREASED QUALITY OF LIFE
Time

34. Conclusion of this section
Accurate health outcome measurement is vital in determining the value and hence priority that should be placed on competing healthcare interventions. For cancer, this includes screening, initial management (adjuvant treatment, surgery, radiotherapy), management of advanced disease and end of life
The need for sensitivity and practicality may pull in different directions
QALYs assume that all health interventions aim either to make us live longer (quantity) or live better (quality)

35. B. Economic Modelling Theory & Practice

36. Therapeutic interventions are messy and complex
Limited understanding of how things work
Disease/Treatments/Services
Limited evidence of effectiveness
A better treatment? How much better and is it better for all patients?
Evidence limited in time and place
Are RCTs valid for other situations and in other countries?
Variable quality and limited availability of evidence
How to fill gaps?
What is the comparative value of RCTs, observational data and ‘expert’ opinion?

37. Hence we need to model in order to…
Extrapolate beyond the results of a trial
Link intermediate clinical endpoints to final outcomes
Generalise to alternative settings
Synthesise head-to-head comparisons where relevant trials do not exist

38. 1. Extrapolating beyond the results of a trial
Economic evaluations require long term analyses to comprehensively assess the costs and benefits arising from an intervention
Techniques
A range of techniques are available to extrapolate outcome data into the future e.g. constant benefits or linear extrapolation

39. 2. Linking intermediate endpoints to final outcomes where necessary
Where RCTs only report intermediate clinical endpoints e.g.
Hypercholesterolaemia (changes in HDL/LDL)
Response rates to length of survival
Disease free progression to length of survival
Economic evaluations in comparing cost-effectiveness attempt to consider ‘harder’ outcomes
Life-years gained
Techniques
Logistic equations and other methods are used to try and determine impact on length of survival

40. 3. Generalising to alternative settings
Costs
Costs differ from one setting (e.g. country) to another
Techniques
Adapt analyses to take account of local unit costs, comparators and patterns of care
Efficacy
Patients are carefully selected in clinical trials
Compliance in trials is artificially high
Develop an ‘impact model’ that identifies factors underlying the success of a healthcare intervention and dichotomise between ‘locally specific’ and ‘generalisable’

41. 4. Synthesising head-to-head comparisons
RCTs do often compare an active drug vs. Placebo; alternatively an ‘add-on’ drug to an existing regimen and not a replacement. Clinicians need to know whether a new drug is superior to existing therapeutic interventions – not as an ‘add on’ especially when scarce resources
Techniques
Modelling allows for the results of more than one trial to be incorporated thus facilitating indirect comparisons between drugs

42. Stages in developing an economic model
Define the problem and your objective
Identify all relevant factors and how they inter-relate
Search for data/information to quantify those relationships
Choose an appropriate methodology/structure
Construct and calibrate the model
Test/validate model
Revise/correct model (return to stage 5 as required)
Apply model results to problem/decision

43. Knowledge requirements for modelling
Epidemiological:
Population at risk, mortality, effects
Medical:
Nature of the disease and how well do the treatment and comparators work?
Economic
Resources consumed at each stage of the treatment process

44. Data requirements for modelling
Parameter estimates for each possible outcome or health state
Probabilities of occurrence of each outcome or health state
Cost for each resource consumed during the process of care provision

45. Types of model 1. Decision Tree
Model all possible treatment paths and outcomes
Each alternative is shown as a branch
Each branch is connected by a decision (choice) node
Outcomes are connected to branches by probability (chance) nodes
Terminal health states / outcomes totalled for costs & benefits

46. Types of model 2. Markov Chain
Based on movements between defined health states caused by events
Individuals may enter the system at one or more source states
Individuals progress from one state to another according to a set of transition probabilities
Transitions occur at predetermined intervals (cycle period)
Model may include one or more sink or terminal states (no exit)

47. Example of a simple Markov Model
1-p1-p3
1-p2 1
Asymptomatic
disease
Progressive
disease
Patient
Death p1 p2 p3
pn = transitional probability

48. How ‘robust’ are health economic analyses?
Issue to be addressed:
Do limitations in either the quality or availability of evidence affect the recommended decision?
If the decision is not altered despite ‘reasonable’ variations in key assumptions/parameters, then the analysis can be considered to be ‘robust’
Two types of uncertainty:
Structural (is the model design correct?)
Parameter (are the values correct?)

49. Techniques for handling uncertainty
Structural: scenario analysis
Re-run the analysis with alternate assumptions and model structures
Parameter: sensitivity analysis (SA)
Re-run the analysis with different parameter values
One-way SA,
Multi-way SA,
Extreme values SA,
Probabilistic SA

50. Presentation of results of sensitivity analysis 1
Presentation of results of sensitivity analysis 1. Cost-Effectiveness Plane
3000
2000
1000
Incremental Cost
-0.05
0.05
0.1
0.15
-1000
-2000
-3000
-4000
-5000
Incremental QALY

51. Presentation of results of sensitivity analysis 2
Presentation of results of sensitivity analysis 2. CE Acceptability Curve
0.2
0.4
0.6
0.8 1 £0
£10,000
£20,000
£30,000
£40,000
£50,000
£60,000
Value of ceiling ratio
Probability cost-effective

52. Using the results of modelling
A model simply provides a structure (good or bad) that organises complex relationships and data enabling them to be interpreted and manipulated
By predicting and comparing costs and outcomes of competing interventions, it enables decision-makers to address problems in a more systematic manner

53. Good economic modelling practice
A good model provides a structure that allows data to be interpreted and used. However, to maximise the value of the model, certain principles should be followed:
Keep analyses simple
Keep analyses transparent
Make explicit the quality of the underlying data
Keep a focus on uncertainty
Compare the results obtained in your model to others

54. Conclusion - converting ‘numbers’ to ‘knowledge’
Remember:
Numbers are meaningless
Data = numbers with meaning and a source of integrity
Information = data interpreted
Knowledge = information in action

55. C. Value of new drugs including. new cancer drugs:
C. Value of new drugs including new cancer drugs: Scottish Medicines Consortia (SMC), Scotland

56. Only limited number of new products having reasonable health gain
SMC recently analysed their guidance for 281 new products and indications (all drug classes) issued between April and September 2008
Data extracted from base case QALY gain estimates provided by the manufacturers showed:
Overall median health gain QALY
Mean health gain QALYs (standard deviation 1.72)
This broken down as:
22% offered no benefit
28% offered >0 – 0.1 QALY
25% offered > QALY
13% offered > QALY
12% offered >1 QALY
Ref: Andrew Walker and Ailsa Brown EACPT 2009

57. Recent examples of new drugs not recommended by SMC as economic concerns
Disease
Reason for rejection
Cost/ QALY
Sunitinib (SUTENT)
GIST and mRCC
Economic case not proven
£ £81000
Pemetrexed (ALIMTA)
Metastatic NSCL cancer
Economic case not proven
Up to £53,000
AVASTIN and ERBITUX
Metastatic ca colon/ rectum
Economic case not proven
£24000 –£93000
Rimonabant (ACCOMPLIA)
Obesity
Economic case not proven
Not assessed - no comparator
Aliskiren (RASILEZ)
Essential hypertension
High costs with comparable efficacy
£11-14/ year (generic ACEi) vs. £
Ref: SMC website 57

58. SMC and new anti-cancer medicines recently reviewed
61 cancer medicines reviewed
36 for advanced/metastatic cancer
25 for earlier/adjuvant treatment
Median QALY gain (over current treatment)
0.38 for advanced cancer
0.30 for earlier/adjuvant treatment
Mean QALY gain (over current treatment)
0.52 for both groups

59. What do these ‘mean and median’ QALY gains imply in reality?
Median health gain
6 months with quality of life 70% of normal
Mean health gain
8-9 months with QoL 70%
Only 6 drugs (10%) offered ≥1 QALY
22 drugs (36%) offered ≤0.2 QALY
= ≤3 months at 70% of normal QoL
Overall

60. Some individual cancer drugs had considerable health gain
Some of the greatest health-gains are with really innovative drugs:
Trastuzumab – 2.4 QALYs
Nilotinib – 2.1 QALYs
Bortezomib – 1.1 QALYs
Even if these are expensive, they may offer good ‘value-for-money’
The issue subsequently becomes affordability and opportunity costs (workshop)

61. Health gain with cancer drugs similar to other disease area
Anti-cancer drugs are much like new drugs for other disease areas
Musculoskeletal (11) – 0.66 QALY
Infections (33) – 0.11 QALY
Endocrine (24) – 0.07 QALY
Cardiovascular (33) – 0.05 QALY
CNS and pain (55) – 0.04 QALY
Overall new drugs in general do not appear to be as valuable as many would like to think!

62. D. Value of new drugs including new cancer drugs: NICE (England)

63. What does NICE mean by cost-effective?
More effective and less costly
More effective and more costly AND additional effect is worth the extra cost
Less effective and less costly AND the cost saving is large enough to compensate for the loss of effect
What is the cost-effectiveness threshold for acceptance?
NICE ‘does not use a precise ICER threshold above which a technology would automatically be defined as not cost effective or below which it would’

64. Why do NICE use a cost-effectiveness threshold?
“The appropriate threshold to be used is that of the opportunity cost of programmes displaced by new, more costly technologies”
If most plausible estimate is below £20,000 per QALY gained: cost effective use of NHS resources
Above £20,000: are there benefits not captured by the QALY? Has quality of life aspect been adequately measured?
Above £30,000 “…need to identify an increasingly stronger case for supporting the technology as an effective use of NHS resources”

65. End of life care: The NICE criteria
Introduced 5 January 2009, revised July 2009
Three criteria in order to qualify:
The treatment is indicated for patients with a short life expectancy, normally <24 months
There is sufficient evidence to indicate that the treatment offers an extension to life, normally of at least an additional 3 months, compared to current NHS treatment
The treatment is licensed or otherwise indicated for small patient populations

66. End of life care: The NICE process
For eligible treatments, the Committee will consider:
The impact of giving greater weight to QALYs achieved in the later stages of terminal diseases, using the assumption that the extended survival period is experienced at the full quality of life anticipated for a healthy individual of the same age
The magnitude of the additional weight that would need to be assigned to the QALY benefits in this patient group for the cost-effectiveness of the technology to fall within the current threshold range
Committee requires that the assumptions used in the reference case economic modelling are plausible, objective and robust

67. End of life care: Specifying the comparator
The comparator for the technology being assessed is very important because the choice to a large extent determines the incremental costs and incremental effects (and thus the cost per QALY)
Relevant comparators might include:
Therapies routinely used in the NHS
Current best practice
What is expected to be replaced (SMC)
‘Do nothing’ (e.g. best supportive care)

68. End of life care: Measurement of health benefit
The incremental QALYs as a result of a treatment have two components:
Changes in survival
Changes in health-related quality of life
The main challenge with estimating changes in survival arises because the data on clinical effectiveness typically means that long-term overall survival must be extrapolated from short-term progression-free survival data
Two challenges recur with quality of life data
The absence of data
Unsatisfactory measure of quality of life

69. Case study: Cetuximab for locally advanced squamous cell cancer of head and neck
Cetuximab with radiotherapy versus radiotherapy alone in patients considered unsuitable for chemotherapy
RCT showed significant improvement in duration of locoregional control, overall and progression-free survival, and overall response rate for the combination than for radiotherapy alone (Bonner et al, NEJM 2006)
Manufacturer estimated a cost per QALY of £6,390
Committee rejected the submission highlighting uncertainties regarding the clinical evidence (e.g. RT regimens used in trial not typical of UK current practice, high proportion of patients in trial suitable for chemotherapy, and no clinical benefit demonstrated in patients with poor performance status)

70. End of life care: The importance of sub-groups
Cost-effectiveness generally varies across sub-groups
Important because ICER for entire patient group may be above the threshold but there may be sub-groups for whom the intervention is cost-effective
Similarly, an ICER below the threshold for the patient group as a whole may hide ICERs for particular sub- groups above the cost-effectiveness threshold
RCTs often under-powered to assess treatment effects in sub-groups

71. Additional analysis presented following appeal
Karnofsky performance status
Hazard Rate
Confidence Interval
Cost effectiveness
100
0.61
0.28 to 1.31
£13,200 90
0.58
0.39 to 0.88
£4,500 80
1.11
0.69 to 1.77
£58,200 70
1.22
0.53 to 2.78
RT dominant
<70
3.41
0.65 to 17.7
£37,000

72. NICE recommendation (June 2008)
The Committee concluded that Cetuximab in combination with radiotherapy is clinically and cost-effective in patients with locally advanced squamous cell cancer of the head and neck who have a Karnofsky performance status score of 90% or greater and for whom platinum-based chemoradiotherapy treatment is contraindicated

73. NICE evaluation: A summary
ICERs and cost-effectiveness
Understanding the economic model
Key elements to watch out for:
Appropriate comparators
Relevant sub-groups
Measurement of health benefit
Analysis of uncertainty

74. The importance of HTA: Conclusion
No health system can afford to fund all new healthcare interventions so we inevitably have to prioritise and choose
HTA simply attempts to identify the healthcare interventions that provide sufficient clinical benefit to justify their cost
HTA enables health systems to optimise the amount of patient benefit obtained from the limited resources available to the healthcare system
HTA also enables an informed debate to be undertaken with the industry concerning the importance of linking drug pricing to drug effectiveness