경제성 평가 방법론 연구설계와 효과평가snu-dhpm.ac.kr/pds/files/130514...
TRANSCRIPT
경제성 평가 방법론
Economic evaluation
건국의대 예방의학교실
이 건 세
Outline
• 경제성 평가 배경 및 필요성
• 정의, 단계• Design, modeling• CEA• CUA• CBA• Discounting outcome• Sensitivity Analysis
• Decision, Implication
한국의 건강보험의 특성과 과제
• 특성• 빠른 제도의 정착 –복합적 문제 직면
• 국가 및 중앙의 정책적 영향 - 정치쟁점화
• 민간중심의 공급체계 –비용 증가
• 과제• 급여 확대
• 비용절감
• 효율성 향상
• 불평등 감소
• 3 waves in Health care• Universal coverage and equal access
• Controls, rationing and expenditure caps
• Incentive and competition
한국 건강보험의 과제-비용절감
• 약제비 통제• 약가 통제 (일반명의약품 권장), 이윤율 통제, 약제비 통제
• 참조가격제 ; 독일, 네델란드
• 이윤제한 : 영국
• 품목별 통제 : 프랑스, 스위스, 이태리 등
• 소비자 부담 증가(참조가격제, 상황대상 품목 제한)
• 보험 상황 대상 결정 : positive list • 의약품의 비용효과를 제대로 분석하여 우수한 의약품을 선별해 보
험적용
• 비용효과 평가의 기준이 되는 경제성 평가방안 등 연구
• 요양기관의 의약품 저가구매 유도 및 거래내역 투명화 확보
건강보험의 급여 기준
• 신약의 보험급여 등재결정 및 가격결정
• 의약품을 포함한 신의료기술에 대한 경제성 평가• ‘국민건강보험 요양급여의 기준에 관한 규칙’• 요양급여대상 또는 비급여대상으로 결정되지 아니한 새
로운 행위․약제 및 치료재료• 요양급여대상 여부의 결정을 보건복지부장관에게 신청
할 때 비용효과에 대한 자료를 포함
Why economic evaluation?
• Best choices
• given their limited resources
• prevention or cure?
• antibioitics or kidney transplantation?
• HBP: Diuretics or Ca blocker?
How EE can help?
• Burden of disease • signifies size of problem (prevalence +severity)
• No answers : what can be done to reduce the problem
• Cancer : major problem, but should most resources be spend on anti-cancer tx?
• Clinical trials• shows effects of treatment
• not say anything on the value(cost) per treatment
• Kidney transplantation save a life. should spend resources for this first?
경제성 평가의 정의 및 특성
• Economic evaluation• 비용(costs)과 결과(consequences) 두 가지 관점에서 활동(치
료, intervention)의 대안들을 비교 분석하는 것으로 정의
• 경제성 분석의 특징• Inputs과 outputs(costs and consequences) 두 가지에 관심• 의사결정 자체에 관심
• 자원의 제약• 합리적인, 근거에 입각한 의사결정 기준(explicit criteria)
Economic Evaluation
new
health care
intervention
old
health care
intervention
Health status
after
intervention
Resources
consumed
Question:
Does the improved health status justify the
additional resources required for the new
intervention as compared to the old intervention?
Literature on cost
of illness, burden
of illness
Most
randomized
clinical trials
투입, 결과에 따른 보건의료의 평가
Trends in economic evaluation_____________________________________
Decade approach cost outcome_________________________________________________
60’s cost-benefit analysis $ $
70/80’s cost-effectiveness analysis $ natural unit
80/90’s cost-utility analysis $ QALY
00’s outcome measurement: $ QALY or
utilities, conjoint analysis, $ (WTP)
willingness to pay
___________________________________________________________
Phases in an economic evaluation
Phase 1: design of the study
Phase 2: measuring and valuing costs
Phase 3: measuring and valuing benefits
Phase 4: discounting
Phase 5: sensitivity analysis
Phase 6: applying a decision rule
Phase 1:design of the study
• perspective (societal/third party payer)
• selecting the alternative
• experiment / model
• outcome parameters
• time horizon
Perspective
• Which perspective a CEA should take?
• comprehensive societal perspective is advocated
• All costs and effects should considered, regardless of who bears the cots and who get these effects
• Third party payer (insurer) perspective
• Convenient
• May ignores real societal costs & effects
Choice of comparator
• most efficient alternative
• standard treatment (volume, market share)
• consider “no treatment”
Design of the study
Prospective data-
collection?
experiment ?
modelling (claims data, meta-analysis,
expert opinion)
no
yes
piggy back EE (external
validity?)
•naturalistic study (internal
validity?)
•other observational designs
(case-control, cohort)
yes
no
Time Horizon
• choose time horizon • all consequences in terms of costs and outcomes are
taken into account
• if observation period of a clinical trial is shorter
• modelling may be appropriate to study long term consequences
Phase 2: Costing (3 phases)
• identification of resource items
- heavily depend on ‘perspective’ taken.
• measurement of resource use
- valid amount of medical services consumed
- eg. Diaries, expert opinions, prospective survey etc.
Valuation of the resources
- Unit costs of each items.
Total Cost
= Resource uses (measurement) x unit cost(valuation)
Categories of costs
Direct costs • within health care
physician time, pharmaceuticals, hospital costs, etc.• outside health care
travel costs(patient), informal care cost, etc.Indirect costs
• within health care :controversial
health care during prolonged survival
• outside health care
loss in production due to morbidity or mortality
Productivity costs: what & when?
• Productivity cost measurement aims at quantifying the impact of illness on production of individuals in monetary terms
• Waiting, undergoing treatment, recovering
• Being ill at work -> reduced productivity
• Being ill and absent from work
• Disability
• Premature death
How to include these costs?
• Four methods are discussed:
Transfer payments
Human capital method
Friction cost method
US Panel method (Gold et al.,1996)
• The human capital method is most popular, but…
Phase 3: outcome
• dependent on relevant outcomes
• CEA: Natural units
• CUA: QALY
• CBA: monetary valuation
Outcome measures
• CEA and CUA• CEA: single, program-specific, unvalued, natural units
• CUA: single or multiple, general, valued
• CBA• value benefits in monetary terms : worthwhile?
• benefit : not only cost savings
• but also monetary value of health outcome
Outcome measures 1: CEA
• Choice of effectiveness measure• final output. life-years gained
• intermediate output : cases found, patients treated.
• admissible intermediate output• link between intermediate and final output
• some values in itself. diagnosis. provide reassurance.
• Effectiveness data : How should be obtained?• availability of data: crucial
• major source: medical literature
• quality, relevance
Modelling complementary to prospective approach
• intermediate to final outcome
• beyond trial duration
• beyond indications trial patients
• beyond trial setting (costs and outcomes)
• compliance patients and physicians
Measuring the cost-effectiveness of lipid-lowering drugs
Structure of a decision analytic model
• Definition of the patient ‘population’
• Comparison of alternative strategies
• Definition of probabilities of (clinical) events and related costs
• Patient health state defined as final endpoint as a result of a clinical pathway
Example decision tree
fatal
0.1
not fatal
0.9
complications
0.03
no complications
0.97
symptomatic ulcer
0.011
fatal
0.1
not fatal
0.9
complications
0.03
no complication
0.97
asymptomatic ulcer
0.099
no ulcer
0.89
NSAID
NSAID & misoprostol
OA patient
Markov modelling
• A patient’s situation may vary over time
• The states that can be distinghuised are different regarding costs and value of health
Questions and answers:
• What is the duration that a patient will be in the specific health states
• What is the total value (costs, health) of the patient’s time in the different health states
HEALTHY ILL DEAD
HEALTHY ILL DEAD
CYCLETIME
Markov structure
HEALTHY ILL DEAD
Outcome measures 2: CUA
• Why CUA?• to compare a broad set of interventions
different interventions: different effects
• to incorporate a large number of outcomes
life extension, quality changes, side effects
• to weight the different outcomes
important: more valued
consumer preference
CUA : when ?
• Health-related quality of life is an important outcome
• A health care programme affects both morbidity andmortality
• To compare programmes that have a wide range of differentkinds of outcomes (resource allocation decisions)
• To compare with programmes evaluated by CUA in the past
Quality Adjusted Life Year (QALY)
Health Related Quality of Life (HRQoL)
• Subjective
• The patient reports (not the doctor)
• a multi-dimensional construct Dimensions:• Physical well-being
• Social well-being
• Emotional well-being
• Usual activities (Self care, Housekeeping and Paid and unpaid work)
• Pain
• Symptoms
Quality of life instruments
• Generic:• EuroQol instrument EQ-5D
• Health Utility Index
• Quality of Well-Being
• SF-36
• Nottingham Health Profile
• Sickness Impact Profile
• Disease / Condition Specific:
• Developed for a specific diseases(Parkinson disease) or group ofconditions (Cancer)
• Contain detailed questions ondimensions of health that areaffected by the disease concernedand its treatment
EQ-5DMobility
I have no problems in walking about 1
I have some problems in walking about 2
I am confined to bed 3
Self-Care
I have no problems with self-care 1
I have some problems washing or dressing myself 2
I am unable to wash or dress myself 3
Usual Activities (e.g. work, study, housework, family or leisure activities)
I have no problems with performing my usual activities 1
I have some problems with performing my usual activities 2
I am unable to perform my usual activities 3
Pain/Discomfort
I have no pain or discomfort 1
I have moderate pain or discomfort 2
I have extreme pain or discomfort 3
Anxiety/Depression
I am not anxious or depressed 1
I am moderately anxious or depressed 2
I am extremely anxious or depressed 3
Your Own
Health State Today
100
0
9 0
6 0
5 0
7 0
8 0
4 0
3 0
2 0
1 0
Best Imaginable Health State
Worst Imaginable Health State
We made a rating scale (comparable with a
thermometer) to help people rating how good
or bad their health is. At the scale “100” refers
to the best health state you can imagine and
“0” to the worst health state you can imagine.
We like to ask you to rate how good or bad your
own health state is on the rating scale. Draw a line
from the box below to the point at the scale that
indicates how good or bad your own health state
is today.
How to assess the relative desirability (preference) of each possible health state?
Response method
Question framing
Certainty
(Values)
Uncertainty
(Utilities)
Scaling Visual Analogue Scale
Choice Time Trade-Off Standard Gamble
QALY Analysis
• Value (V) of quality of life (Q)• V(Q) = [0….1]
• 1 = healthy
• 0 = dead
• Adjust life years (Y) for quality of life • QALY’s = Y * V(Q)
• Y: numbers of life years
• Q: health state during life years
• V(Q): the value of health state Q
Methods to value health state (TTO)
Healthy 1.0
State i hi
Death 0
x t Time
Alternative 2
Alternative 1
Value for state i = hi = x/t
Methods to value health state (SG)
Healthy
Death
State i
Probability p
Probability 1-p
Alternative 2
Alternative 1
Preference score (utility) for state i = hi = p
The valuation task
• Ranking of 13 cards with EQ-5D health state descriptions
Card for state 11111(Life A)
Card for state 12121(Life B)
• Valuation of the 13 health states using a Visual Analogue Scale
• TTO Valuation exercise
No problems in walking
No problems with self care
No problems performing usual activities
No pain or discomfort
Not anxious or depressed
No problems in walking
Some problems with self care
No problems performing usual activities
Moderate pain or discomfort
Not anxious or depressed
EuroQol: TTOStart with a value 1.
For any state not 11111 decrement: - .081
Mobility at level 2 decrement: - .069
Mobility at level 3 decrement: - .314
Self Care at level 2 decrement: - .104
Self Care at level 3 decrement: - .214
Usual Activities at level 2 decrement: - .036
Usual Activities at level 3 decrement: - .094
Pain/Discomfort at level 2 decrement: - .123
Pain/Discomfort at level 3 decrement: - .386
Anxiety/Depression at level 2 decrement: - .071
Anxiety/Depression at level 3 decrement: - .236
Any dimension at level 3 decrement: - .269
Factors influencing the values of health state
• Valuation method • SG, TTO, VAS
• whether or not in combination with the descriptive system of the health states to be valued
• Perspective • Patient
• General public
• Operational definitions: Interview bias
• Country: Culture?
• Socio-economic factors:• Age, gender (hardly any influence)
• Education (small influence)
• Religion and beliefs about life after death
Outcome measures 3: CBA
• Decision making based on monetary value
CEA/AUA: QALY league table
• Broader in scope. health and non-health
• Allocative efficiency
CEA/AUA: production efficiency
• Quantify externalities(spill over effect)
CEA/AUA: narrowly client-focused
willingness-to-pay technique
• conforms more closely to Welfare Theory
Contingent valuation
• Appraches to the monetary valuation of healthoutcomes
• human capital
• revealed preferences
• willingness-to-pay(contingent valuation)
• Asking individuals for their maximum willingness to pay (WTP) for a gain in health
• Fits in Cost Benefit analysis (CBA)
Example WTP question
• Are you prepared to pay … for a drug that
reduces the risk of getting a migraine attack
by 50%?
Critique on CV-WTP
• WTP depends on ability to pay -> equitable?
• Scope effects: WTP responses tend to be undersensitive to the magnitude of benefit
• Budget constraint bias: WTP inflates valuations of intervention asked about.
When asked for an intervention in isolation, WTP is far in excess of WTP when intervention is considered in relation to other interventions
Key issues in design
• how is information presented?
• type of payment vehicle for WTP
• commodity valued under uncertainty?
• what time period for valuation?
• how survey administered?
• WTP or WTA?
• (questionnaire format)
Phase 4: Discounting
• Principle:• Effects in terms of costs or health gains are weighted less when they
occur later in time.
• Reasons:
• time preference• Impatience
• diminishing marginal utility of income
• uncertainty
• opportunity cost of capital• the existence of a positive rate of return implies that one resource
unit in the future is valued less than one now!
Discounting Health Effects
• Should costs and benefits be discounted at the same rate?
• Same rate • Assumption: trade-off between health and money
• Value of health will increase as society gets richer
• Keeler and Cretin paradox• Discounting costs at higher rate than benefits leads to
infinite delay
Discounting Health Effects
• Keeler and Cretin paradox• Discounting costs at higher rate than benefits leads to
infinite delay
Now 1 Year
Costs(5%) $100,000 $95,238
Effects(3%) 10 QALYs 9.7 QALYs
Ratio $100,000 per QALY gained
$9,809 per QALY gained
Discounting Health Effects
• Different discount rate• Change in value money different from change in value
health over time
• Discount health differently (lower) from money (empirical evidence)
• For the sake of future generations
• Health : less easily transferable in time
• Health and wealth not always perfectly substitutable
Discounting Health Effects
• NL : 4% for costs and health effects
• UK : costs 6%, health effect 1.5%
• US: 3% per year for costs and health effects (Gold)
• Advise• Perform sensitivity analysis at least show results for 0%
discount rate
Phase 5: uncertainty
• The values used in cost-effectiveness
analysis are estimates
• Uncertainty is associated with all estimates
• Quantifying uncertainty through• Sensitivity analysis
• Statistical analysis
Sensitivity vs. statistical analysis
• Sensitivity analysis• Quantifies uncertainty when values are:
• Guessed
• Determined from secondary sources
• Approximated
• Statistical analysis• Quantifies uncertainty when values are estimated from a sample of
a population such as in a randomised clinical trial
Types of Sensitivity analysis
• Goal is to find out how sensitive ICER is to changes in parameter.
• Univariate sensitivity analysis• vary only 1 parameter at the time
• Multivariate sensitivity analysis• vary 2, more or all parameters at the time
• ‘worst case’ & ‘best case’
• threshold
• Probabilistic
• Probabilistic sensitivity analysis
Probabilistic analysis
• Most informative method, since it presents extreme outcomes, but also likelihood of outcomes
• Define probability distribution for each variable
• Where possible, base distribution on trial data
• Draw random number from each distribution and calculate ICER
• Repeat many times (1000-5000)
Results :probabilistic analysis
-5000
0
5000
10000
15000
20000
-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
additional effects
ad
dit
ion
al c
osts
Phase 6: Decision rule
Cost-effectiveness of strategy 2 versus strategy 1
cost (2) - costs (1)
benefits (2) - benefits (1)
• benefits in natural units (e.g. life years gained, healthy babies)
• benefits in QALY’s
QALY league table
GM-CSF elderly with leukemia $235,958
EPO in dialysis patients $139,623
Lung transplantation $100,957
End stage renal disease $53,513
Heart transplantation $46,775
Didronel in osteoporosis $32,047
Statins in high cholesterol $18,151
PTA with Stent $17,889
terbinafine in onychomycosis $16,843
Breast cancer screening $5,147
Viagra $5,097
Congenital anorectal malformation $2,778
Totaal
Dutch Experience: Priority setting
• Defining Basic healthpackage
• Dunning criteria• Necessity
• Effect
• efficiency(cost-effectiveness)
• individual responsibility
Explicit cost per QALY threshold?
• No explicit threshold; also other criteria considered (budget impact, severity of illness, own responsibility, etc.)
• Canada: Can$ 20,000-100,000 (Laupacis)
• NICE/UK: £ 30,000
• Netherlands: guideline cholesterol: € 18,000
Reimbursement decisions in Australia
0
50000
100000
150000
200000
250000
300000
0 1 2 3 4
Funding decision
cost
/QA
LY
(A
ust
r $)
Recommended
at price
Recommended at lower
price
Rejected
Role of economic evaluation?
• EE is NOT an easy prescription for decision making in health care, it is an aid
• Efficiency is an important goal in health care
but certainly NOT the only one
• In some areas health gain does not play a role
whatsoever (e.g. care for dying)
• Other criteria play a role as well!
Implication of Pharm. EE
• Fourth hurdle • evidence of quality, safety, and efficacy • requirement of effectiveness and cost-
effectiveness data for drug coverage policy decisions
• Cost containment vs Expanding insurance coverage
• Cost containment is not an appropriate objective of public policy
• Resource allocation (efficiency)• 3 waves ?