Decision Making using Economic Evaluation

Decision Making using Economic Evaluation

Loading
Loading Social Plug-ins...
Language: English
Save to myLibrary Download PDF
Go to Page # Page of 48

Description: Overview of economic evaluation, myths in translating EEs for policy making, Types off EE Studies, Cost Benefit Analysis, Valuing Benefits in a CBA, Cost of Illness/injury (COI) approach, value of statistical life, Myth in Interpreting CBA Results, Cost--Utility Analysis (CUA), Measurement of Health Utilities, Valuation of Benefits in a CUA, Global Burden of Disease Project, Myths in Interpreting CUA Results, Cost Effectiveness Analysis (CEA), Quantifying Outcomes in a CEA, CEA Caveat.

 
Author: Phaedra Corso, PhD (Fellow) | Visits: 1695 | Page Views: 1887
Domain:  Medicine Category: Practice Mngmnt Subcategory: Policy & Economics 
Upload Date:
Short URL: http://www.wesrch.com/medical/pdfME1MS1SKBWKHW
Loading
Loading...



px *        px *

* Default width and height in pixels. Change it to your required dimensions.

 
Contents:
Economic Evaluation for Decision Making
Phaedra Corso, PhD

Centers for Disease Control and Prevention
National Center for Injury Prevention and Control
TM

Objectives
Provide an overview of economic evaluation (EE) methods Define how the benefits measure is calculated in a:
� Cost-benefit analysis � Cost-utility analysis � Cost-effectiveness analysis

Explore myths in translating EEs for policymaking
2

TM

Public Health Approach to Prevention
Problem Identification Program and Policy Development Program and Policy Evaluation Implementation and Dissemination
3

Risk and Protective Factor Identification

Economic Evaluation

TM

Types of EE Studies


Costs only
� Cost-of-illness studies � Cost analyses (intervention/program costs)



Balancing costs and benefits
� Cost-benefit analyses (CBAs) � Cost-utility analyses (CUAs) � Cost-effectiveness analyses (CEAs)

4

TM

Why Use Economic Evaluation (EE) Methods?
Maximizing health outcomes is important Resources are limited, so hard (resource allocation) decisions must be made EE demonstrates the value provided from resources expended
5

TM

Economic Evaluation Economics
Economics seeks to explain choices and behaviors by individuals Economic evaluation (EE) seeks to inform choices made by public policy makers, health care payers

6

TM

Use of EE to Inform Prevention Policy
Tier of Decision Making
US Congress Allocation decision between health, defense, and education. Outcome comparator: $ Director of HHS Allocation decision between violence prevention and cancer screening. Outcome comparator: QALYs Local HD Allocation decision between two interventions designed to reduce child neglect. Outcome comparator: Cases of neglect prevented

CBA CUA CEA

7

TM

Cost-Benefit Analysis (CBA)
� Standardizes both costs and outcomes in $ � Includes health and nonhealth outcomes � Provides an objective summary measure
� Net Benefits (B � C), Net present value (NPV) � Benefit-cost ratio (B / C)

8

TM

Valuing Benefits in a CBA
Cost of Illness/injury (COI) approach
� Direct
Medical, non-medical e.g., hospital stays, medications, travel costs

� Indirect � human capital approach
Time, productivity e.g., time missed from work to care for child Contingent valuation (CV) -

Willingness to Pay approach
9

TM

Willingness-to-Pay
Theoretically better measure than COI
� Includes intangible costs

Types
� Revealed Preferences
Wage-risk studies Consumer valuations of safety products

� Expressed or Stated Preferences
Open-ended CV Binary CV
10

TM

Revealed Preferences
Most often used for value of statistical life (VSL) estimates Wage-risk Studies (Hedonic Wage Models)
� Choice of explanatory variables important � Viscusi lit review ... $3-7 million per statistical life (1990 dollars) � VSL ~ WTP for year of life saved (typical health measure) � Key Variable is Job Risk
Workers' perception of risk & risk aversion
11

TM

Revealed Preferences (cont'd)
Consumer Valuations of Safety
� Product purchased directly for safety
Air bags Smoke detectors Small vs. big cars

� Similar problems as hedonic wage models
Need to collect risk perception data Health care subject to market distortions
12

TM

Expressed/Stated Preferences
Open-ended valuation questions
� Ask directly for max WTP, or min WTA � Ask about all attributes, including costs � conjoint analysis

Difficult to estimate � non-response problems Use aids (bidding game, payment cards)
Improve response Starting-point bias
TM

13

NCIPC Survey

"Now program were a nationally-sponsored child you be willing "If thisimagine we hadavailable to your state, would maltreatment to prevention program that 2 year to sponsor this program?" pay $150 in extra data, was available to 100,000 and this this "Based on national taxes perout of every your state children program was an average of the risk of child day killed due annually, orproven to reduce4 childrenaeverybeing are killed to child maltreatment by 50%. This by parents the caretakers." as�a"Wouldof child willing to pay $225?" that or number of result you be maltreatment means YES children killed on average every day in the U.S. by child NO � "Would you be willingfrom 4 per day to 2 per day." maltreatment is reduced to pay $75?"

14

TM

Myth in Interpreting CBA Results
Only implement programs with + NB or NPV (benefits � costs > 0)

15

TM

� NPV > $0, net savings--argument for investment

� NPV < $0, net cost--need additional reasons for investment (ethical, moral, political)

� Program with highest NPV should be funded first.

16

TM

Cost-Utility Analysis (CUA)
� Combines
� Length of life (survival), and � Quality of life

� Compares disparate outcomes in terms of utility � preferences for health states � Derives a ratio of cost per health outcome
� $/Quality-adjusted life year (QALY) � $/Disability-adjusted life year (DALY)

17

TM

What is a Utility?
Utility is a numerical measurement of the desirability of, or preference for, a health state Grounded in tenets of Expected Utility Theory
(von Neumann and Morgenstern)

Utility is (most often) anchored by
� 1, for perfect health � 0, for death
18
TM

Measurement of Health Utilities
Direct utility elicitation� very intensive
� A description of the health state � Surveys to measure preferences: Rating scales, timetradeoffs, standard gambles, or person-tradeoffs (for DALYs)

Indirect: Use of multi-attribute utility functions
� Administer instrument with description of health to "classify health state" � Preference weights obtained from population-based samples

For some conditions, utility weights are available "off the shelf"
� Risky approach, may not be transferable
TM

19

First: Define the health state What the person can do
Physical ability Self-care Role/Social

How the person feels
Pain Energy Emotional well-being
20

TM

Second: Elicit preference weights for the health states
Question type Response method Certainty "value" Scale Choice Rating Scale Time-Trade Off Standard Gamble Uncertainty "utility"

21

TM

Rating Scale (RS)

Blindness in both eyes Worst possible state Best possible state

0

50

100

22

TM

Time Trade-off (TTO)
Utility
healthy

U(healthy) = 1.0

blind both eyes

U(blind both eyes) = ?

Dead 0

Years

12

20

23

TM

Standard Gamble (SG)
blind both eyes Utility= ? prob=? prob=1.0 1 - prob perfect vision Utility= 1.0 death Utility= 0

U (blind in both eyes) = prob (perfect vision)
24

TM

Utilities (preference-based), not to be confused with...
Profile-based measures:
SF-12 SF-36 HRQOL-4 from BRFSS

25

TM

26
TM

Utilities: Indirect Measurement
Health Utilities Index (HUI) EuroQOL (EQ-5D) Quality of Well-Being (QWB) SF-6D
27
TM

Example: HUI-III
Attributes (number of levels): VISION (6) HEARING (6) SPEECH (5) AMBULATION (6) DEXTERITY (6) EMOTION (5) COGNITION (6) PAIN (5)

Number of unique health states: 6 x 6 x 5 ... x 5 = 972,000
28

TM

A Comparison of Scales
# of health states HUI 972,000 Valuation technique RS, transformed into SG RS Method of extrapolation MAUT* Sample Country

504 (general) 866 (general) 3395 (general) 611 (general)

Canada

QWB

1,170

Statistical

US (San Diego) UK

EQ-5D

243

TTO and RS

Statistical

SF-6D

18,000

SG

Statistical

UK
29

* Multi-attribute utility theory
TM

Panel* Recommendations


Preference measures should be generic. Health-state classification system should reflect important domains for problem at hand. Community weights most appropriate.
* U.S. Panel on Cost-effectiveness in Health and Medicine
TM





30

Source of Preference Weights, 1976 � 1997*
Source of preference Author Clinician Patient Community n 74 59 55 52 % 32.4% 25.9% 24.1% 22.8%

* Source: Harvard SPH, CEA Registry
TM

31

Utilities for Selected Health States
Health State Perfect Health (reference case) Chronic stable angina (TTO, patients) Chronic stable angina (SG, patients) HIV without symptoms (TTO, patients) HIV without symptoms (TTO, physicians) Home dialysis (TTO, patients) HIV without symptoms (TTO, public) Chronic stable angina (RS, patients) Home dialysis (TTO, public) Death (reference case) Utility 1.00 0.93 0.87 0.87 0.69 0.64 0.63 0.59 0.54 0.00

Sources: Nease et al. JAMA 1995;273:1185 Owens et al.Quality of Life Research 1997;6:77 Torrance and Feeny. Int J of Tech Assess in Hlth Care 1989;5:559.

32

TM

From Utilities to QALYs
Quality-adjusted life year = a year of life in a health state adjusted by the utility associated with the health state. QALYs summarize expected health profiles over time QALYs are a measure of the difference in value of two profiles Tradeoff between health and longevity
33

TM

Valuation of Benefits in a CUA:

Combining Length of Life with Quality of Life

QUALITY OF LIFE (weights)

1.0

with intervention

without intervention 0.0 birth LENGTH OF LIFE (Years) death death'
34

TM

What is a DALY?
Global Burden of Disease Project (WHO/Harvard) Disability Adjusted Life Years (DALYs) a composite measure of disease burden YL lost due to premature mortality + YL lived with disability (adjusted for severity) Utility weights (called disability weights) are based on preferences elicited using the person trade-off (PTO) technique administered to a panel of experts.
35

TM

DALYs � Utility Assessment
PTO1 Project A can extend life of 1,000 healthy persons for 1 yr. Project B can extend life of X blind people by 1 yr. What does X have to be for the two projects to be =? PTO2 Project A can extend life of 1,000 healthy persons for 1 yr. Project B can give Y people their eyesight back for 1 yr. What does Y have to be for the two projects to be =?

Adjust responses so that X = Y
36

TM

DALYs � Pros and Cons
Pros
Useful for monitoring disease burden Some reasonable assumptions
Time as a unit of measurement Using age and sex to calculate disease burden

Cons
Age-weighting Whose values are elicited?
37
TM

U.S. Burden of Disease - Men
Deaths
Major depression Self-inflicted injuries Pulmonary disease Homicide and violence Cerebrovascular disease Alcohol use HIV/AIDS Lung/tracheal cancer Motor vehicle crashes Ischemic heart disease

7 4 9 3 8 2 5 1

0

2

4

6

8

10

12

Percentage of Total DALYs Lost, 1996
Source: Michaud, Murray, Bloom. Burden of disease: Implications for future research. 38 JAMA 2001;285(5):535-539.
TM

U.S. Burden of Disease - Women
Deaths
Motor vehicle crashes Diabetes Dementia Pulmonary disease Breast Cancer Osteoarthritis Lung/tracheal cancer Cerebrovascular disease Major depression Ischemic heart disease

9 6 8 4 5 3 2 1

0

1

2

3

4

5

6

7

8

Percentage of Total DALYs Lost, 1996
Source: Michaud, Murray, Bloom. Burden of disease: Implications for future research. 39 JAMA 2001;285(5):535-539.
TM

Myth in Interpreting CUA Results

Only implement programs less than $50,000/QALY saved (or some other threshold)

40

TM

$$/QALY: What is a good "value"?
Intervention
Pneumoccocal vaccine vs no vaccine in 65 cohort Nicotine patch vs no patch in 25-29 males Mammography screening vs no screening in 45-69 women Driver side airbag vs no airbag in driving population Hypertension screening/treatment vs no screening in asympt. 20 yo women

Median $/QALY
2,500 5,200 18,000 27,000 61,000
41

TM

Cost-Effectiveness Analysis (CEA)
� Expresses outcomes in natural units.
� (e.g., number of cases prevented or lives saved)

� Compares results with other interventions affecting the same outcome. � Derives a ratio of cost per unit of outcome.
� $/case prevented � $/life saved � $/life year saved
42

TM

Quantifying Outcomes in a CEA
Intermediate outcomes
Increased pro-social skill development Enhanced school climate

Final outcomes
Aggressive behavior prevented Pro-social behavior increased Academic achievement improved

43

TM

CEA Caveat
Outcomes cannot be combined, so one or two of the most important effectiveness measures should be considered for the CEA. The number of summary measures depends on the number of outcomes chosen.
If 2 outcomes, A and B, are considered the most important for evaluation, then
Cost/outcome A Cost/outcome B
44

TM

Myth in Interpreting CEA Results

Cost effective = Cost saving

45

TM

Cost-effective = Cost-saving
25000 $/LY saved 20000 15000 10000 5000 0 -5000 0% 20% 40% 60% 80% 100%

% people vaccinated
46

TM

Final Thoughts
Economic evaluation is both art and science It should involve multiple disciplines
� � � � Epidemiology Statistics Economics Health outcomes research

Published analyses often not authored or reviewed by people with expertise in methods Results should be interpreted with caution
47

TM

...and
Economic evaluation is valuable to decision making and for setting health policy. Yes, YOU too can do an EE at home.
Read some key references Haddix, Teutsch, Corso (OUP, 2003) Gold et al. (OUP, 1996) Seek advice and consultation Have fun!
48

TM

Subscribe
x