Transcript Plan of class

Economic evaluation of health
programmes
Department of Epidemiology, Biostatistics
and Occupational Health
Class no. 7: Cost-effectiveness analysis –
Part 2
Sept 24, 2008
Plan of class
Review Question 1 from assignment no 1
Finish material from previous class
Quality of life scales
Extended dominance
Net benefit vs ICER
Relevance of each
perspective
Perspective
Relevance
Patient
Rarely reported; can help to anticipate patient
choices
Health and social care system Usually makes decision whether to fund
intervention
Government
Decision to fund may have wider impact, may
be relevant to government as a whole
Society
Broadest perspective, ideally the one on which
decision would be based
Perspective of analysis: Which costs
to include
Cost
Patient
Health care
system
Government
Society
Direct health care
costs
Physician visits
Psychologist visits
Medications
x
x
x
Out-ofpocket
(if any)
If public
If public
Total cost
(public or
private)
Out-ofpocket
costs only
Cost borne
by RAMQ (if
any)
Cost borne
by RAMQ
(if any)
Total cost
x
x
x
Any public
Any public
Total cost of
gym
membership or
equipment
Hospitalisations
Other direct costs
(exercise
intervention)
Any out-ofpocket
costs
Perspective of analysis:
Which costs to include
Cost
Patient
Health care
system
Government Society
Time costs
Physician and
psychologist visits,
any
hospitalisations (total
time including travel)
Time to exercise,
self-administer
therapies, etc.
Travel costs
Time
cost over
and
above
what is
reflected
in
personal
income
Time over and
above what is
counted in
productivity
losses
x
x
Perspective of analysis:
Which costs to include
Cost
Patient
Productivity losses
(or gains)
Any
changes
in
personal
income
Changes in tax
revenues
Changes in welfare
payments
Any
changes
in
personal
income
Health care
system
Government
Society
x
x
Administrative
costs only
x
Administrative
costs only
Time horizon decision
Should be long enough for consequences
directly related to intervention to play
themselves out
Do the costs of the 4 interventions have
different time profiles?
Depression known to influence physical
health care costs (several mechanisms)
Longer follow-up costly; use modeling
study
CEA or CUA?
Turtle soup was tangy
Tables were attractively decorated
Service was prompt and attentive
Salmon was ordinary
Decor was so-so
Price was moderate
VS.
Overall value for money: 4/5!
CEA
or
CCA
Need for good effectiveness
data
Efficacy vs effectiveness
Study protocols may influence outcome
 Adjust if possible
Selective use of studies?
If no evidence, use sensitivity analysis
Intermediate vs final
outcomes
Intermediate outcomes: medication
adherence, blood pressure, cholesterol
levels…
Usefulness of results depends on strength
of evidence linking intermediate and final
outcomes
Discounting benefits
Controversy whether to also discount
benefits
But logical inconsistencies arise if benefits
and costs not discounted at the same rate
So in practice best to discount at the same
rate (report results with 5%, 3%, 0% for
both)
See book for more detailed discussion
Quality of life scales
Specific measures (e.g., Wisconsin QOL
for people with severe mental illness)
General health profiles (e.g., SF-36, GHQ)
Preference-based measures
To be discussed as part of Cost-utility analysis
Specific
measures
General health
profiles
• More
responsive to
change
• May
• More acceptable to
patients and clinicians
• May be less
acceptable to
patients and
clinicians
• Do not yield results that
can be compared across
disease domains
be less
responsive to change
• May yield results
comparable across
disease domains
Extended dominance
Alternative
Cost (C)
Change in
C (∆C)
E (lifeyears)
∆C/∆E
Change in
E (∆E)
A
100
100
5
5
20
B
200
100
7
2
50
C
300
100
12
4
25
350
300
250
200
150
100
50
0
0
2
4
6
8
10
12
14
Assume 100 patients are to be treated and that the 3
treatments may be used (e.g., 1/3 get A, 1/3 B, etc.).
What treatment(s) should the 100 patients receive to
maximize the number of life-years gained?
Suppose you have a budget limit - $20,000. Can a
combination of A and C yield more life-years than B?
New
Tx
costs
more
0
Increased
threshold
ratio
A
Existing
threshold
ratio
New Tx
more
effective
Net benefit instead of ICER
∆C/∆E < RT
NMB = RT ∆E - ∆C > 0
or
NHB = ∆E - ∆C/ RT > 0
Example
∆C= $1,000; ∆E = 10 life years
∆C/∆E =100 $ per life-year
Suppose RT = $50 per life-year
Then ∆C/∆E > RT
NMB = 50 x 10 - 1000 = -500 < 0
or
NHB = 10 - 1000/ 50 = -10 < 0
Intervention is too costly for the life-years it
provides