Making Decisions in Health Care: Cost-effectiveness and the
Value of Evidence
Karl Claxton Centre for Health Economics,
Department of Economics and Related Studies,
University of York,
NICE Appraisals Committee
Overview
• What decisions need to be made?• Should a technology be adopted?• How uncertain is this decision?• Is more evidence needed?• What can and should NICE do?
What are the decisions?
• Should a technology be adopted given existing information?– Which clinical strategies are worthwhile? – For which patient groups?
• Is current evidence sufficient to support use in NHS?– Do we need more evidence?– What type of evidence is required?– What additional research should be conducted to provide this
evidence?
What are the decisions?
• Should a technology be adopted given existing information?– Which clinical strategies are worthwhile? – For which patient groups?
• Is current evidence sufficient to support use in NHS?– Do we need more evidence?– What type of evidence is required?– What additional research should be conducted to provide this
evidence?
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 1 2 3 4 5 6 7 8 9 10 11 12
Life years
Qua
ility
of
Life A
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 1 2 3 4 5 6 7 8 9 10 11 12
Life years
Qua
ility
of
Life
B
Is it worthwhile?• What is an improvement in health?
– Gain in life expectancy– Improvement in quality of life
Does it improve health?
Quality adjusted life years (QALYs) A = 4.2 QALYs B = 7.7 QALYsHealth Gain = 3.5 QALYs
But what about costs?
QALYs gained
Cost
2
£20,000£10,000 per QALY
£40,000£40,000 per QALY
1
£20,000 per QALY
£20,000
2 QALYs =
= 2 – £20,000
£20,000
Is it cost-effective?Is it worthwhile?
Is the ICER less than the cost-effectiveness threshold?
If the cost-effectiveness threshold is £20,000 per QALY, B is cost-effective
Is net benefit positive? Net health benefit = QALYs gained – QALYs lost
Net money benefit = £ value of QALYs gained – additional costs
= 2 x £20,000 – £20,000
Additional cost
QALYs gained ICER = = £10,000 per QALY
= 2 – 1 = 1 QALY
= £20,000 = 1 QALY
What do we need?
• Estimate QALYs gained and costs– Over time (often patient’s life time)– For each alternative– For each patient group
• Relevant evidence?– Clinical evidence of effect– Progression of disease and events– Quality of life– Resource use and costs
Need to Combine evidence
Clinical effect
Disease Progression
QALY
Costs
Rand
om
s
ampl
ing Asymptomatic Progressive
Dead
Treatment A
Asymptomatic Progressive
Dead
Treatment B
Model Structure
Treatment A
QALY Cost
Treatment B
QALY Cost
1 £10,000
2 £30,000
0 £ 5,000
3 £20,000
2 £15,000
4 £40,000
1 £10,000
3 £30,000
= 2 – £20,000
£20,000
Should a technology be adopted?
Treatment A
QALY Cost
Treatment B
QALY Cost
2 £30,000
3 £20,0004 £40,000
1 £10,000
0 £ 5,0002 £15,000
1 £10,000
3 £30,000
Additional cost
QALYs gained ICER =
£20,000
2 QALYs = = £10,000 per QALY
Is the ICER less than the cost-effectiveness threshold?
£10,000 per QALY < £20,000 per QALY, B is cost-effective
Is net benefit positive?
Net health benefit = QALYs gained – QALYs lost
= 2 – 1 = 1 QALY
Net money benefit = £ value of QALYs gained – additional costs
= 2 x £20,000 – £20,000 = £20,000 = 1 QALY
What are the decisions?
• Should a technology be adopted given existing information?– Which clinical strategies are cost-effective? – For which patient groups?
• Is current evidence sufficient to support use in NHS?– Do we need more evidence?– What type of evidence is required?– What additional research should be conducted to provide this
evidence?
How uncertain is a decision?
What’s the best we can do now? But we are not always right
Choose B and expect 13 QALYs Chance that B is the best = 3/5 = 0.6
Chance that A is the best = 2/5 = 0.4
Chance that C is the best = 0/5 = 0
So if we adopt B the probability of error = 0.4
How things could turn out
Net Health Benefit
Best choiceTreatment A Treatment B Treatment C
Possibility 1 9 12 8 B
Possibility 2 12 10 9 A
Possibility 3 14 17 11 B
Possibility 4 11 10 10 A
Possibility 5 14 16 12 B
Average 12 13 10
How uncertain is the decision?
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
£0 £10,000 £20,000 £30,000 £40,000 £50,000 £60,000
Cost-effectiveness threshold
Pro
ba
bili
ty c
ost
-effe
ctiv
e
B
A
C
Choose A Choose B
ICER = £25,000 per QALY
Why does uncertainty matter?
What’s the best we can do now? Could we do better?
Choose B and expect 13 QALYs If we knew we get 13.6 QALYs
Maximum benefit of more evidence is 0.6 QALYs
But is it worth it?
How things could turn out
Net Health Benefit Best we could do if we knew
What we could loseTreatment A Treatment B Best choice
Possibility 1 9 12 B 12 0
Possibility 2 12 10 A 12 2
Possibility 3 14 17 B 17 0
Possibility 4 11 10 A 11 1
Possibility 5 14 16 B 16 0
Average 12 13 13.6 0.6
Do we need more evidence?
£0
£5,000,000
£10,000,000
£15,000,000
£20,000,000
£25,000,000
£0 £10,000 £20,000 £30,000 £40,000 £50,000 £60,000
Cost-effectiveness threshold
Ma
xiu
m b
en
efit
of
evi
de
nce
.
Choose A Choose B
Cost of research
Cost of research
£0
£5,000,000
£10,000,000
£15,000,000
£20,000,000
£25,000,000
£30,000,000
£35,000,000
£0 £5,000 £10,000 £15,000 £20,000 £25,000 £30,000 £35,000 £40,000 £45,000 £50,000
Threshold for cost-effectiveness
Po
pu
latio
n E
VP
I
WT
Ritalin
Zanamivir
Riluzole
Orlistat
Do we need more evidence?
£0
£2,000,000
£4,000,000
£6,000,000
£8,000,000
£10,000,000
£12,000,000
£14,000,000
£16,000,000
£18,000,000
£20,000,000
WT Ritalin Zanamivir Riluzole Orlistat
Po
pu
latio
n E
VP
PI
Effectiveness
Utiltities
Cost
What type of evidence?
Quality of life
Is current evidence sufficient?
• Summary– Uncertainty matters because we might need more evidence– Value of evidence (information)
• How uncertain is the decision?• Consequences of getting the decision wrong• Number of patients who could benefit
– Costs of getting more evidence
Decisions in a joined up world?
• Adopt technologies if we expect them to be cost effective based on existing evidence
But only if we simultaneously address question:Is the evidence sufficient?
• Demand or commission further research to inform this choice in the future
In a fragmented world?
• Publicly funded research?– Separation of the remit for adoption and research commissioning– NICE can’t control research prioritising and commissioning
• Some limited influence• Prioritising and commissioning not consistent with adoption
decisions
• Sponsored research? – No powers to demand research (or disclosure or access to ipd)
• A remit for ‘coverage with evidence’?• Could it be enforced?
What can NICE do?
• Separation of adoption and research decisions– Adoption decisions without accountability for impact on
future research– Research decisions without accountability for relevance to
adoption decisions • Dangers
– Adoption decisions undermine evidence base for practice• Incentives and ethics
– Commissioned research does not inform decisions• Adoption becomes the only policy instrument
£0
£10
£20
£30
£40
£50
£60
£70
£80
£0 £5,000 £10,000 £15,000 £20,000 £25,000 £30,000 £35,000 £40,000 £45,000 £50,000
Cost-effectiveness threshold
Op
po
rtu
nity
loss
Account for the cost of uncertainty
What we loose if we accept technology
What we loose if wereject a technology
£0
£10
£20
£30
£40
£50
£60
£70
£80
£0 £5,000 £10,000 £15,000 £20,000 £25,000 £30,000 £35,000 £40,000 £45,000 £50,000
Cost-effectiveness threshold
Op
po
rtu
nity
loss
Clear signals and incentives
Provide more evidence!
£0
£10
£20
£30
£40
£50
£60
£70
£80
£0 £5,000 £10,000 £15,000 £20,000 £25,000 £30,000 £35,000 £40,000 £45,000 £50,000
Cost-effectiveness threshold
Op
po
rtu
nity
loss
Clear signals and incentives
Reduce price (but don’t tell)
Why only in research?
• Clear signals – No because it is not a cost-effective use of resources– No because there is currently insufficient evidence to justify
NHS use– Spell out the key evidence needed (not the research)
• Clear incentives– If and when additional evidence is made available then
considered for early review– Incentives to sponsors (evidence and price)– Incentives for others stakeholders to lobby for publicly
funded research– Clear signals to research commissioners
• Appraisal process– Already generates much of the analysis and information– Explicit consideration of which uncertainties are most
important – Clear consideration of the evidence (not the research)
needed
• STA makes this the most pressing issue– Issuing guidance when evidence base is least mature– Piecemeal nature of STA guidance
What should NICE do?
• Real danger– Potential damage to evidence base for current and future
NHS practice– Costs to the NHS of changing guidance
• Real opportunity – Address evidence needs of the NHS– Provide clear signals and incentives
Dangers and opportunities?