Fillable Printable Consumers Use Cost Benefit Analysis In Order To Maximize What
Fillable Printable Consumers Use Cost Benefit Analysis In Order To Maximize What
Consumers Use Cost Benefit Analysis In Order To Maximize What
1
Cost-Effectiveness Analysis
Henry A. Glick, Ph.D.
Pharmacoeconomics
September 18, 2013
Outline
•Introduction to cost-effectiveness analysis (CEA)
•Choice criteria for CEA
•The cost-effectiveness frontier
•Net benefits (a transformation of CEA) and choice
criteria
•Additional topics
Cost-Effectiveness Analysis (I)
•Estimates costs and outcomes of intervention
•Costs and outcomes are expressed in different units
– If outcomes aggregated using measures of
preference (e.g., quality-adjusted life years saved),
referred to as cost utility analysis
www.uphs.upenn.edu/dgimhsr/fda2013.htm
2
Cost-Effectiveness Analysis (II)
•Results meaningful if:
– Compared with other accepted and rejected
interventions (e.g., against league tables), or
– There exists a predefined standard (i.e., a threshold
or maximum acceptable cost-effectiveness ratio or an
acceptabilitycriterion) against which they can be
compared
• e.g., $50,000 per year of life saved might be
considered maximum acceptable ratio, or
– Can define utility curves that trade off health and cost
Cost-Effectiveness “History”
•$/Life saved
•$/Year of life saved (YOL)
•$/Quality adjusted life year saved (QALY)
•??? Outlawing QALYs ???
Why CEA Rather Than CBA?
•Not precisely clear
– Potential difficulties in measurement
– Discomfort with placing a dollar value directly on a
particular person's life (rather than years of life in
general)
– QALYs / life years more equally distributed than
wealth
– Health more a “right” than a commodity
• Implies 1 person 1 vote may be more appropriate
than 1 dollar 1 vote
• Cost-effectiveness analysis uses 1 QALY/year
1 vote
3
Cost-Effectiveness Ratios
•Cost-effectiveness ratio
•A ratio exists for every pair of options
– 1 option (case series), no ratios calculated
– 2 options, 1 ratio
– 3 options, 3 ratios (option 1 versus option 2, option 1
versus option 3, andoption 2 versus option 3)
•In “efficient” selection algorithm, don’t necessarily
calculate all possible ratios
12
12
Costs - Costs
Effects - Effects
Average Cost-Effectiveness Ratio
•Some dispute about definitions
– e.g., Some use “average cost-effectiveness ratio” to
refer to practice of dividing therapy’s total cost by its
total effect (including Treeage, a fairly ubiquitous
piece of decision analysis software)
•Don’t use this definition of average CER
•Recommend against dividing a therapy’s total cost by its
total effect
– These ratios provide little to no information
Dividing a Therapy’s Costs by Its Effects is
“Generally Uninformative”
CostEffectRatio
Example 1
Rx1500.02520,000
Rx2780.02630,000
Example 2
Rx1500.02520,000
Rx21500.0530,000
4
Dividing a Therapy’s Costs by Its Effects is
“Generally Uninformative”
CostEffectRatio
Example 1
Rx1500.02520,000
Rx2780.02630,000
(780-500) / (.026-.025) = 280,000
Example 2
Rx1500.02520,000
Rx21500.0530,000
(1500-500) / (.05-.025) = 40,000
Average Cost-Effectivemess Ratio (2)
•Definition: Comparison of costs and effects of each
intervention with a single option, often "do nothing" or
usual care option
# Guaiac
TestsCostCases Detected
Avg Cost/ Case
Detected *
17.75.00659469--
210.77.007144245495
313.02.007190048852
414.81.0071938511,783
516.31.0071941714,279
617.63.0071942016,480
* (C
i
– C
1
) / (E
i
– E
1
)
Example: Average Ratios and Sixth Stool Guaiac
•Neuhauser and Lewicki, NEJM, 1975;293:226-8.
5
Incremental Cost-Effectiveness Ratios
•Average ratios not important when making selection from
among all candidate therapies
•ICER = comparison of costs and effects among
alternative options (i.e., excluding comparator used in
calculation of average cost-effectiveness ratios)
•When there are only 2 options being evaluated, average
and incremental cost-effectiveness ratios are identical
•Neuhauser and Lewicki, NEJM, 1975;293:226-8
.
Guaiac Average and Incremental Ratios
# Guaiac
testsCost
Cases
Detected
Average
CER *
Increm
CER **
17.75.00659469----
210.77.0071442454955495
313.02.00719004885249127
414.81.0071938511,783469,816
516.31.0071941714,2794,687,500
617.63.0071942016,48044,000,000
* (C
i
– C
1
) / (E
i
– E
1
)
** (C
i
– C
i-1
) / (E
i
– E
i-1
)
Cost-Effectiveness Plane
•Axes
•Origin
•Average
ratios
•Incremental
ratios
Alternative
therapy dominates
Alternative therapy more
effective but more costly
New therapy more
effective but more costly
New therapy
dominates
(-) Difference in Cost (+)
(-) Difference in Effect (+)
oo
oo-oo
-oo
6
Good and Bad Value
Choice Criteria For Cost-Effectiveness Ratios
•Choose options with acceptable average and
incremental cost-effectiveness ratios (i.e., whose ratios
with all other options are acceptable)
•Subject to:
– Budget Constraint?
– Acceptable Ratio?
•Not accounting for uncertainty around ratios
•Consider 3 mutually exclusive options and a willingness
to pay of 50k
Choice Criteria, Example 1
Option 1Option 2Option 3
Expected Costs10,000135,000270,000
Expected QALYs202530
RatiosOption 2Option 3
Option 125,00026,000
Option 2--27,000Adopt?
7
Choice Criteria, Example 2
Option 1Option 2Option 3
Expected Costs10,000135,000235,000
Expected QALYs202526
RatiosOption 2Option 3
Option 125,00037,500
Option 2--100,0000Adopt?
Choice Criteria, Example 3
Option 1Option 2Option 3
Expected Costs10,000210,000230,000
Expected QALYs202121.5
RatiosOption 2Option 3
Option 1200,000146,667
Option 2--40,000Adopt?
Multitherapy Example
•Suppose 6 screening strategies have the following
discounted costs and life expectancies:
TreatmentCostYOLS
No screening (S1)105217.348
Sig Q10 (S2)128817.378
Sig Q5 (S3)153617.387
U+Sig, Q10 (S4)181017.402
C Q(10) (S5)202817.396
U+Sig, Q5 (S6)203417.407
FrazierAL, et al. JAMA. 2000;284:1954-61.
8
Choice Among Screening Strategies
•Which therapyshould be adopted if the acceptability
criterion is $40,000 / YOL Saved? $50,000 / YOL
Saved?
•In what follows, demonstrate 2 (of 4) methods for
selecting a single therapy from among these candidates
– Methods all based on selecting therapy with an
acceptable ratio
– Methods transformations of one another --use same
information in slightly different ways -- and all yield
identical choices
Method 1: Efficient Algorithm (MEA) for
Choosing among Multiple Therapies (I)
•Suppose 6 therapies have thefollowing discounted costs
and life expectancies
TreatmentCostYOLS
No screening (S1)105217.348
Sig Q10 (S2)128817.378
Sig Q5 (S3)153617.387
U+Sig, Q10 (S4)181017.402
C Q(10) (S5)202817.396
U+Sig, Q5 (S6)203417.407
Efficient Algorithm: Step 1
•Rank order therapies in ascending order of either
outcomes or costs (final ordering of nondominated
therapies unaffected by variable chosen)
TreatmentCostYOLS
No screening (S1)105217.348
Sig Q10 (S2)128817.378
Sig Q5 (S3)153617.387
C Q(10) (S5)202817.396
U+Sig, Q10 (S4)181017.402
U+Sig, Q5 (S6)203417.407
9
Efficient Algorithm: Step 2
•Eliminate therapies that are strongly dominated (i.e.,
have increased costs and reduced effects compared with
at least one other alternative)
TreatmentCostYOLS
No screening (S1)105217.348
Sig Q10 (S2)128817.378
Sig Q5 (S3)153617.387
C Q(10) (S5)202817.396
U+Sig, Q10 (S4)181017.402
U+Sig, Q5 (S6)203417.407
Efficient Algorithm: Step 3
•Compute incremental cost-effectiveness ratios for each
adjacent pair of outcomes (e.g., between options 1 and
2; between options 2 and 3; etc.)
TreatmentCostYOLSICER
No screening (S1)105217.348--
Sig Q10 (S2)128817.3787867
Sig Q5 (S3)153617.38727,556
C Q(10) (S5)202817.396Dom
U+Sig, Q10 (S4)181017.40218,267
U+Sig, Q5 (S6)203417.40744,800
Efficient Algorithm: Step 4
•Eliminate therapies that are less effective (costly) but
have a higher cost-effectiveness ratio (weakly
dominated) than next highest ranked therapy
•Rationale: Rather buy more health for a lower cost per
unit than less health for a higher cost per unit
– e.g., eliminate S3 (sig,Q5), because:
• S3 is less effective than next higher ordered S4
(U+sig,Q10) [17.387 YOLS vs. 17.402] AND
• Incremental ratio for moving from S2 to S3
(27,556) is greater thanincremental ratio from
moving from S3 to S4 (18,267)
– Implies that moving from S2 to S4 is more cost-
effective than is moving from S2 to S3
10
Efficient Algorithm: Step 5
•Recalculate ICERs (e.g., between options 2 and 4)
– Repeat steps 4 and 5 if necessary
TreatmentCostYOLSICER
No screening (S1)105217.348--
Sig Q10 (S2)128817.3787867
Sig Q5 (S3)153617.38727,556
C Q(10) (S5)202817.396Dom
U+Sig, Q10 (S4)181017.40221,750
U+Sig, Q5 (S6)203417.40744,800
Efficient Algorithm: Step 6
•Identify acceptable therapy
Maximum WTPTherapy
<7867S1
7867 to 21,749S2
21750 to 44,799S4
44,800+S6
Full Cost-Effectiveness Table
TreatmentCost∆CYOLS∆ YICER
S1 No screening1052--17.348----
S2 Sig Q10128823617.3780.0307867
S3 Sig Q51536--17.387--WD
S5 C Q(10)2028--17.396--SD
S4 U+Sig, Q10181052217.4020.02421,750
S6 U+Sig, Q5203422417.4070.00544,800
SD = strong dominance; WD = weak dominance
11
Reduced Cost-Effectiveness Table
TreatmentCost∆CYOLS∆ YICER
S1 No screening1052--17.348----
S2 Sig Q10128823617.3780.0307867
S4 U+Sig, Q10181052217.4020.02421,750
S6 U+Sig, Q5203422417.4070.00544,800
Introduction to Method 2: Net Benefits
•A composite measure (part cost-effectiveness, part cost
benefit analysis), usually expressed in dollar terms, that
is derived byrearranging cost-effectiveness decision
rule:
W > ∆C /∆Q
where W = willingness to pay (e.g., 50 or 100K)
Net Benefits (II)
•Two forms of net benefit expression exist depending on
rearrangement of expression
– Perhaps most naturally for economists, net monetary
benefits can be expressed on cost scale (NMB)
(W * ∆Q) - ∆C
– OR net health benefits (NHB) can be expressed on
health outcome scale:
∆Q - (∆C / W)
• Potential disadvantage: NHB undefined when WTP
equals 0
