Transcript Pharmacoeconomics

Pharmacoeconomics
An Introduction for the P&T Competition
David Matthews
2012 AMCP P&T Competition National Finalist
The Ohio State University AMCP Chapter
October 9th, 2012
Outline
Types of economic analyses
 Definition of cost-effectiveness
 Determining cost-effectiveness
 Markov modeling

What is Pharmacoeconomics?
Economics is the science of balancing best
outcomes with limited resources
 Pharmacoeconomics applies this concept
to pharmacologic interventions

Types of Economic Analyses
Cost-minimization analysis
 Cost-benefit analysis
 Cost-effectiveness analysis
 Cost-utility analysis

Cost-Minimization Analysis
Compares two interventions considered
equally effective and tolerable
 Determines which intervention costs less
 Costs include more than the price of meds

 Costs
of treatment failure
 Costs of adverse effects
 Drug monitoring or other healthcare services
Cost-Benefit Analysis
Adds up costs associated with intervention
 Compares to monetary benefits of
intervention

 Outcomes
must be converted to dollars
Compares input dollars vs. output dollars
 Determines whether benefits > cost

Cost-Effectiveness Analysis
Usually compares two interventions
 Determines the cost to produce an effect
 Expresses cost of an effect as a ratio:

 Numerator
= cost ($)
 Denominator = clinically appropriate marker,
for example:
mm Hg blood pressure lowering
 mg/dL of LDL lowering
 Quality-adjusted life-years (cost-utility analysis)

Cost-Utility Analysis
Subset of cost-effectiveness analysis
 Determines the cost of adding one year of
perfect health to a patient’s life
 Calculates incremental cost-effectiveness
ratio (ICER)

 Ratio
of cost to effectiveness:
Numerator = cost ($)
 Denominator = Quality-adjusted life-years

Cost-Effective ≠ Cost-Saving!!!
Cost-Saving vs. Cost-Effective

Cost-saving
 An
intervention that has a lower total cost
than an alternative intervention

Cost-effective
 An
intervention that is sufficiently effective
relative to its total cost when compared with
an alternative intervention
Cost-Effectiveness Plane
DOMINATED
cost
NW quadrant:
more costly,
less effective
PERFORM
CEA
NE quadrant:
more costly,
more effective
effect
effect
SW quadrant:
less costly,
less effective
PERFORM
CEA
SE quadrant:
less costly,
more effective
DOMINATES
cost
Adapted from: Smith KJ et al. In: Arnold, RJG, editor. Pharmacoeconomics from theory to practice. Boca Raton: CRC Press; 2010. p. 95-108.
Domination
Occurs when one treatment is both
cheaper and more effective
 Occurs in NW and SE quadrants of plane
 The cheaper/more effective treatment
“dominates” the alternative
 The dominating treatment is the preferred
treatment

Determining Cost-Effectiveness
New intervention in NE or SW quadrant
 Example:

 Drug A is
a new drug
 Drug B is the current standard of care
 Drug A works better than Drug B
 Drug A is more costly than Drug B

Question:
 Using
Drug A instead of Drug B, how much
does it cost us to add one year of perfect
health onto the life of our patient?
Incremental Cost-Effectiveness
Ratio (ICER)
Represents the amount of money spent to
add one year of perfect health onto the life
of our patient
KEY POINT:
The ICER is the single most important
indicator of an intervention’s costeffectiveness.
Its calculation can be complex, and will be
the focus of the next several slides.
Terminology

Utility
 Numerical
estimate of quality of life (QOL)
associated with a disease state or treatment
 Perfect health = 1, Dead = 0
 Anything else…somewhere in between
 Measured using questionnaires
Terminology

Quality-Adjusted Life-Year (QALY)
 Life
expectancy adjusted based on utility
 QALY = time in health state × utility of state
 If patient remains in the state for the remainder
of their life, we can use life expectancy for time
QALY Example

Consider 2 hypothetical chemo drugs
 Standard
of care vs. new therapy
 Both prolong life
 Both cause side effects which reduce QOL
QALY Example

Standard of care treatment:
 Prolongs
life by an average of 1 year
 Estimated utility of 0.65 due to side effects

New treatment:
 Prolongs
life by an average of 1.5 years
 Estimated utility of 0.5 due to side effects
Standard of Care QALYs
QALY = Life expectancy × utility
= 1 year × 0.65 utility
= 0.65 QALYs
The standard of care is expected to add 0.65
quality-adjusted life-years to our patient’s life.
New Treatment QALYs
QALY = Life expectancy × utility
= 1.5 years × 0.5 utility
= 0.75 QALYs
The new treatment is expected to add 0.75
quality-adjusted life-years to our patient’s life.
Calculating ICER
ICER = difference in cost
difference in effectiveness
Or…
ICER =
C2 – C1 $’s
E2 – E1 QALYs
Back to Our Chemo Drugs…

Suppose a full course of treatment costs…
 $12,000
for standard of care
 $15,000 for new treatment
ICER of Chemo Drugs
ICER =
C2 – C1
E2 – E1
ICER =
$15,000 – $12,000
0.75 QALY – 0.65 QALY
ICER =
$30,000/QALY
Interpretation of ICER
On average, it costs us $30,000 to add
one year of perfect health onto the life of
our patient.
So is this considered cost-effective?
Threshold of Cost-Effectiveness
Subjective
 $50,000/QALY commonly reported in studies
 WHO recommends 3x per capita GDP for a
given country

 Would

be around $150,000/QALY in USA
National Institute for Health and Clinical
Experience (NICE) recommends
£30,000/QALY ($48,396/QALY)
Dasbach EJ et al.. In: Arnold, RJG, editor. Pharmacoeconomics from theory to practice. Boca Raton: CRC Press; 2010. p. 119-143.
World Health Organization. Available from: http://www.who.int/choice/costs/CER_thresholds/en/index.html
McCabe C et al.. Pharmacoeconomics. 2008;26(9):733-44. Review.
Problems with Oversimplification
Much more complex than “averages” in
the real world
 Some people will tolerate the drugs better
or worse than others
 Patients do not remain in one health state
 Each individual experiences different
quality of life, incurs different costs, etc.

Markov Models
Common in pharmacoeconomic research
 Used to calculate the entire cost and
QALYs gained for a population
 Uses a hypothetical cohort of patients
 Patients move between health states
 Each state has associated probabilities,
costs, and utilities

Components of Markov Models
Expected health states
 Probabilities related to treatment failure,
side effects, etc.

 Normally

from probabilities seen in studies
Cycle length
 How
frequently would patients be expected to
transition through health states?
Utility and cost estimates for each state
 Time horizon

Simplified Example
New treatment for a terminal illness
 More costly, more effective than standard
of care
 Patients whose disease progresses incur
greater costs

 Hospitalizations
 More
treatments
Example Markov Model
Cycles patients through health states
based on preset probabilities
 Example model:

 Healthy
 Sick
 Dead

Each state is assigned its own utility and
cost
Summary of Therapies
Therapy
Standard of care New treatment
Cost of treatment, one
month
$800
$1,500
Progression from
healthy to sick per
month
8%
4%
Cost of tx + disease
progression per month
$2,500
$3,200
Progression from sick
to death per month
20%
10%
Markov Model Example
Standard of Care
Healthy
0.92
0.08
0.8
Sick
0.2
Dead
Markov Model Example
New Treatment
Healthy
0.96
0.04
0.9
Sick
0.1
Dead
Health State Utilities

Healthy
 Utility

Sick
 Utility

= 0.8 (not 1.0 due to side effects)
= 0.4
Dead
 Utility
=0
10,000 Patient Cohort:
New Treatment
Healthy
10,000 pts
0.96
0.04
Sick
0.9
0.1
Dead
After 1 month
Healthy
9,600 pts
0.96
0.04
Sick
400 pts
0.1
Dead
0.9
COST: 9,600 x $1,500
=$14.4M
QALY: 1/12 x 9,600 x 0.8
=640 QALY
COST: 400 x $3,200
=$1.3M
QALY: 1/12 x 400 x 0.4
=13 QALY
After 2 months
Healthy
9,216 pts
0.96
0.04
Sick
744 pts
0.1
Dead
40 pts
0.9
COST: 9,216 x $1,500
=$13.8M
QALY: 1/12 x 9,216 x 0.8
=614 QALY
COST: 744 x $3,200
=$2.4M
QALY: 1/12 x 744 x 0.4
=25 QALY
After 3 months
Healthy
8,847 pts
0.96
0.04
Sick
1,039 pts
0.9
COST: 8,847 x $1,500
=$13.2M
QALY: 1/12 x 8,847 x 0.8
=590 QALY
COST: 1,039 x $3,200
=$3.3M
QALY: 1/12 x 1,039 x 0.4
=35 QALY
0.1
Dead
114 pts
And so on until all patients are
in the “absorbing state” (death)
Markov Model Results
Model continues until all patients in
absorbing state or time horizon complete
 Patients accrue QALYs and costs each
cycle
 Separate models run for new treatment
and standard of care
 Once complete, ICER is calculated

 (difference
in cost) / (difference in QALYs)
Markov Models in the Real World
Theoretically, models could be completed
by hand
 Real life models become much more
complex

 More
health states
 Ability to move more freely through states
 Account for issues such as adverse events

Computers solve complex models
Real Life Example
Shaheen NJ et al. Gut. 2004 Dec;53(12):1736-44.
Problems with Markov Models
Complex models are difficult to understand
 Validity of model depends upon utility and
cost estimates

 Sensitivity
analysis to account for variability
Sensitivity Analysis
The scenario based off initial estimates is
called the “base case scenario”
 Real life probabilities and costs may be
higher or lower than predicted
 Adjust assumptions upward and
downward and recalculate ICER
 Provides a range of possible economic
outcomes

Conclusion
New interventions are usually more
effective but at a higher price
 Cost-effectiveness analysis helps
determine if a new intervention is effective
enough to be worth our limited resources
 ICER is a numerical value that
summarizes cost-effectiveness
 Markov models are used to calculate ICER

Questions?
References








McGhan WF. Introduction to pharmacoeconomics. In: Arnold, RJG, editor.
Pharmacoeconomics from theory to practice. Boca Raton: CRC Press; 2010. p. 1-16.
Haycox A. What is cost-minimization analysis? In: Arnold, RJG, editor. Pharmacoeconomics
from theory to practice. Boca Raton: CRC Press; 2010. p. 83-94.
Smith KJ and Robers MS. Cost-effectiveness analysis. In: Arnold, RJG, editor.
Pharmacoeconomics from theory to practice. Boca Raton: CRC Press; 2010. p. 95-108.
Dasbach EJ, Insinga RP, and Elbasha EH. Cost-utility analysis: a case study of a
quadrivalent human papillomavirus vaccine. In: Arnold, RJG, editor. Pharmacoeconomics
from theory to practice. Boca Raton: CRC Press; 2010. p. 119-143.
Beck JR. Markov modeling in decision analysis. In: Arnold, RJG, editor.
Pharmacoeconomics from theory to practice. Boca Raton: CRC Press; 2010. p. 47-58.
World Health Organization. Choosing interventions that are cost-effective [Internet].
[Geneva]: WHO; c2012 [cited 7 Oct 2012]. Available from:
http://www.who.int/choice/costs/CER_thresholds/en/index.html
McCabe C, Claxton K, Culyer AJ. The NICE cost-effectiveness threshold: what it is and what
that means. Pharmacoeconomics. 2008;26(9):733-44. Review.
Shaheen NJ, Inadomi JM, Overholt BF, Sharma P. What is the best management strategy
for high grade dysplasia in Barrett's oesophagus? A cost effectiveness analysis. Gut. 2004
Dec;53(12):1736-44.