Linking Health Risk Assessment and Economics: Designing and Interpreting Health Studies to Inform Ec

1
Linking Health Risk Assessment and Economics
Designing and Interpreting Health Studies to
Inform Economic Analysis

• Dan Greenbaum, President
• Health Effects Institute
• Institute of Medicine Roundtable on Environmental
  Health Sciences, Research, and Medicine
• Washington, DC
• November 14, 2006

2
Translating Health Science into Economics

• The Basic Approach
• Applying health data in complex models
• The Results
• Examples from recent EPA proposals and rules
• The big picture across federal agencies
• Some challenges
• How do we predict the baseline?
• Are all health effects equal (e.g. asthma
  attacks)
• How soon after exposure do effects occur?
• DALYs vs. deaths
• Concluding Thoughts

3
Integrating Health Benefits into Decisions (The
simple version)
Model Changes in ambient air concentrations of
pollutants
Model Changes in Health Effects
Select Pollution Reduction Strategy
Model Changes in Emissions
Model Changes in human exposure
Model Value of Health Benefits
4
The Complex Version Cost Benefit Analysis
Analytical Framework (Courtesy Bryan Hubbell, EPA)
INCIDENCE ESTIMATION
Ambient air monitoring
County-, regional-, national-level health
surveillance data
Grid-based air modeling
Background disease incidence and prevalence
(mortality, morbidity)
Census data
Ambient air quality change in simulation year
Population growth projection model
Epidemiological studies
Projected demographic (population count) data
Change in health effect incidence (counts) for
simulation year
Concentration-response function
VALUATION
Willingness to pay (WTP) and cost of illness
(COI) metrics
Income growth
Income elasticity
Monetized benefits for simulation year
5
Key Steps
Establish Baseline Conditions (Emissions, Air
Quality, Health)
Estimate Expected Reductions in Precursor
Pollutant Emissions
Model Changes in Ambient Concentrations of Ozone
and PM
Estimate Expected Changes in Human Health Outcomes
Estimate Monetary Value of Changes in Human
Health Outcomes
6
Impacts of PM2.5 on Heart and Lung
Mortality(Source HEI Reanalysis of the American
Cancer Society Study (Krewski 2000)
7
How do we value improvements in public health?

• Cost of illness
• Hospital admissions, work loss days
• Captures the direct dollar savings
• Ignores the value of reduced pain and suffering
• Generally a lower bound
• Willingness to Pay
• Premature death, chronic bronchitis, Respiratory
  symptoms
• Measures a more complete value of avoiding a
  health outcome
• Relies on either revealed or stated preferences
  for risk reductions
• Generally more uncertain than COI

8
Current mean values for health effects (2000 )

• Premature death 6.3 million (WTP)
• Chronic bronchitis 340,000 (WTP)
• Heart attacks 67,000-141,000 (COI)
• Hospital admissions 7,000 - 18,000 (COI)
• ER visits 300 (COI)
• Acute bronchitis 360 (WTP)
• Respiratory symptoms 15 - 50 (WTP)
• Asthma attacks 40 (WTP)
• Work loss days 100 (COI)
• School loss days 75 (COI)

9
A Recent ExampleAssessing Health Benefits of
US Diesel Fuel and Technology Rules

• New US EPA rules to reduce diesel fuel sulfur and
  engine emissions
• Fuel sulfur from 500 ppm to 15 ppm in 2006
• Reduced PM and NOx emissions in 2007, 2010
• EPA conducted extensive Regulatory Impact
  Analyses (RIA 2000)
• Accepted by US Office of Management and Budget

10
Impacts of PM2.5 on Heart and Lung
Mortality(Source HEI Reanalysis of the American
Cancer Society Study (Krewski 2000)
11
Estimates of benefits(Source US EPA RIA, 2000)
Number of Annual Cases for All of US2030

8,300
Mortality
Hospital Admissions
5,600
Emergency Room Visits
2,100
New cases of chronic bronchitis
5,500
New cases of bronchitis in children
17,600
361,400
Acute asthma attacks
Acute respiratory symptomse.g. new cases of
croup, pneumonia
386,000
9.5 million
Restricted activity days
12
Comparing Costs and Benefits US Highway Diesel
Rule(Source US EPA RIA 2000)

• Current administration has maintained and
  extended these rules

13
Results of other recent analyses

• Nonroad Diesel Engines
• By 2030, reduces NOx emissions by over 800,000
  tons and diesel PM by over 126,000 tons
• 12,000 premature mortalities avoided
• 5,600 cases of chronic bronchitis avoided
• 15,000 nonfatal heart attacks avoided
• Millions of acute respiratory symptoms and work
  loss days avoided
• Valued at over 80 billion

14
Results of other recent analyses

• Clear Skies Legislation
• Reduction in utility NOx and SO2 emissions of 2.5
  million tons and 4.9 million tons respectively
• 12,000 premature mortalities avoided
• 7,400 cases of chronic bronchitis avoided
• Over fifteen million acute respiratory symptoms
  and work loss days avoided
• Valued at 96 billion

15
Applying These Approaches Across the Federal
GovernmentThe Office of Management and Budget
Annual Report
16
A Surprising Result ?EPA Net Benefits exceed all
other federal agencies together (2004 OMB report)
17
A Challenge How do we predict the baseline?
18
Predicting the baseline

• It is not enough to predict what will happen with
  a new rule
• We also have to estimate what would have happened
  without the rule
• i.e. would reductions have occurred for economic
  reasons even without any intervention?
• What would those have meant for health?

From EPA Costs and Benefits of the Clean Air Act
1970 - 1990
19
A Real World Example Air Quality and Health
Changes in the Harvard Six Cities Study (Laden,
et al 2006)
20
A ChallengeValuing Different Effects on
Different People
21
The Challenge of Valuing Different Effects

• We have extensive studies on relationships
  between pollution and many effects
• In many cases we have sufficient data to estimate
  concentration response relationships
• But there are fine points to translating those
  estimates into economic values
• E.g. the case of Asthma (courtesy Bryan Hubbell,
  EPA)
• And for many health endpoints we do not always
  have adequate data

22
What does EPA currently use for values?

• Asthma ER visits 290 (cost of illness)
• Asthma hospital admission 6,600 (cost of
  illness)
• Asthma symptom day 42 (willingness to pay)
• Work loss day National average of 110
• School absence day 75 (from lost parent wages)

23
Can all asthma attacks be valued the same?

• Asthma exacerbations or attacks are not well
  defined endpoints and may have very different
  impacts depending on the underlying severity of
  asthma in those affected
• Two key issues
• Distribution of asthma severity in the population
  and
• distributions of severity of asthma episodes
  conditional on asthma severity
• Yet most air pollution epidemiology studies
  cannot distinguish among individuals with
  different levels of severity

24
Of the adult asthmatic respondents to the 1999
NHIS, 73 were classified as mild intermittent,
7 as mild persistent, and 20 as moderate or
severe persistent. For asthmatic children, 78
were classified as mild intermittent, 8 were
classified as mild persistent, and 14 were
classified as moderate or severe persistent.
25
Within the severity classifications, there was
considerable variability in the number of asthma
episodes. Some mild intermittent asthmatics had
over 95 episodes in a year, while some
moderate/severe asthmatics had only 1 episode in
a year. However, the severity of these episodes
varies.
26
Most mild intermittent asthmatics indicated
little to no activity restrictions, while most
moderate/severe asthmatics indicated fair to
moderate or a lot of activity restrictions due to
their asthma episodes.
27
The Broader Challenge What health effects do we
quantify?
28
Other, In Some Case Emerging, Public Health
Impacts that We Dont Quantify

• Infant mortality/low birth weight
• Decreased lung development
• Cancer
• Doctor visits
• New incidence of asthma
• Ozone mortality
• Not quantified due to
• Lack of appropriate baseline incidence rates
• Not enough weight of evidence
• Not easily monetized or characterized in terms of
  public health significance

29
A ChallengeWhen do the benefits occur?
30
The Steubenville CaseIf we saw improvements in
health, when did they occur relative to when the
pollution dropped?
31
Why could this be important?

• Future economic values are normally discounted
  back to todays dollar values
• i.e. a dollar in 2030 will not be worth as much
  as a dollar today
• If health benefits quickly follow after a
  pollution reduction, then this is a non-issue
• Benefits would be counted in todays dollars
• But, if the benefits of an action taken today are
  not felt until 20 years from now
• Then the estimates of the value of those benefits
  will be substantially smaller
• The database to examine this question is limited
• HEI currently funding some attempts to look at
  this in the American Cancer Society database.

32
A Challenge Disability Adjusted Life Years
(DALYs) vs. Deaths
33
Accounting for Death and Disease

• Politically charged debate in US over the use of
  deaths vs. measures of life years lost
• Number of people dying prematurely is simple to
  grasp and generates largest benefits estimates
• 6 million per death
• Life Years Lost may better reflect extent of
  impact (especially when measures of disability
  or quality are included) but normally results
  in lower benefits estimates
• Criticized as devaluing the elderly
• But numbers of deaths also may devalue longer
  term disabilities for children

34
Accounting (continued)

• Life Years Lost is extensively used in assessing
  medical interventions
• And in environmental health estimates by WHO
• May give a better integrated measure of heath
  impact
• But requires enhanced epidemiologic data and
  better understanding of relationship of time of
  exposure to effects (and benefits)

35
The WHO Global Burden of Disease Project (Lancet
October, 2002)
36
Annual Burden of Disease from Urban Outdoor Air
Pollution in China, 2000Estimated from WHOCohen
AJ, Anderson HR, et al, 2004.
37
Concluding Thoughts

• We have reasonably well-established methods for
  estimating health benefits of environmental
  interventions
• Quality depends, however, on breadth and quality
  of underlying epidemiology
• We have methods for economic valuation of those
  benefits
• Each with their own strengths and weaknesses
• There are a number of challenges in making such
  estimates
• In the absence of better data, these challenges
  require sensitivity analysis and caution
• There is a continuing need for enhancing the
  underlying health and economic science,
  especially after-the-fact assessments of benefits
  and costs

38
One Way ForwardEnhanced Studies Measuring the
Benefits and Costs After-the-Fact

• The Need
• Quality studies of the actual impacts of
  intervention on health and cost
• One Approach HEIs Accountability Initiative
• Expert monograph on the best approaches
• Now 8 studies underway of short- and long-term
  interventions in US, Europe, Asia

39
Thank You!Dan Greenbaumdgreenbaum_at_healtheffects
.orgwww.healtheffects.org