Damage Function Models

1
Damage Function Models
Dave Stieb
2
And it was so typically brilliant of you to have
invited an epidemiologist
3
Quantifying Health Benefits
Cleaner fuel
Improved ambient air quality
Improved public health
/QALY
Value to society
Reduced vehicle emissions
Reduced population exposure
4
Representing the weight of evidence How do you
choose a parameter value?

• Expert judgment
• Flexible, streamlined
• Can be seen as arbitrary
• Systematic overview and meta-analysis
• Rigorous, comprehensive
• Rigid, cumbersome
• Structured consensus process (eg. Delphi),
  expert elicitation
• Middle ground

5
Dealing with uncertainty

• WHAT IF the parameter were different?
• Sensitivity analysis
• Run the analysis a few times and see how the
  results change
• Probabilistic analysis
• Run the analysis thousands of times (iterations)
• For each iteration, pick a different value for
  each parameter, from an input distribution
• The results are also presented as a distribution
  most likely value of benefits is y, but could be
  as low as x and as high as z

6
(No Transcript)
7

• 1996
• Numerous policy applications
• Acidifying emissions
• AQOs
• Sulphur in gas
• CWS
• Climate change co-benefits
• Peer reviewed

8
Sulphur in Gasoline
9

• Canadian gasoline sulphur levels ranged from lt10
  to 1000 ppm with national average of 340 ppm
• In 1999, federal regulation took effect to reduce
  average sulphur content of gasoline to 30 ppm by
  2005

10
(No Transcript)
11
Findings of Health Panel for seven cities only,
in 2001- Reducing sulphur to 30 ppm improves the
health of Canadians
12
The doctor will bill you now
13
Benefits for Year 2001 by type for 4 alternative
scenarios central, low and high estimates
14
Health Benefits vs. Refinery Costs
( 45 of CanadianPopulation )
Benefits and costs based on 30 ppm (total
population)
15
To do list

• Alternative methods of mapping monitoring data to
  populations (eg. geostatistical methods)
• Update baseline air quality data
• Alternative status quo air quality (dont just
  assume constant)
• Alternative population projections (allow
  different assumptions about birth rate, mortality
  rate, immigration and internal migration)
• Update to 2001 census
• Update risk coefficients
• Derive risk coefficients more systematically
• Allow risk coefficients to differ by geographic
  area
• Allow baseline morbidity, mortality rates to
  differ by geographic area
• Allow non-linearity in concentration response
• Permit multi-year analysis
• More flexible approach to uncertainty analysis
  (not just discrete three point distributions)
• Alternatives to valuation (eg. QALYs)

16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
ICAP

• Ontario version released in 2000
• Expanded nationally
• Peer reviewed
• Currently being updated

20

• Includes database of concentration response
  functions, valuations
• Peer reviewed
• Available online

21
Key points

• Reduced (increased) air pollution results in
  health and environmental benefits (damages)
• Alternative approaches to representing weight of
  evidence, uncertainty
• Improved health (and environmental quality) have
  value to society
• Variety of tools available to assess benefits

22
Extra slides
23

• Damages
• Burden on society from negative effects of air
  pollution
• Benefits
• Gain to society from reducing negative effects of
  air pollution

24
How do we quantify the impact of reduced
(increased) exposure?
25
Monetary Valuation

• AKA
• monetization
• assigning a monetary value to a change in health
  status
• most appropriately measured as willingness to pay
  to improve health or willingness to accept
  compensation to worsen it

26
Image Problem - I

• Unseemly

• Valuation implicit in innumerable private and
  public decisions
• Either do it implicitly or explicitly, but cant
  avoid it
• Balances singular focus on industry compliance
  costs

27
Image Problem - II

• Simply asking people their preferences is too
  hypothetical

• Considerable effort in developing valid and
  reliable measures
• Reality checks

28
Image Problem - III

• Lives worth more in rich countries

• Distributional/ equity issues can be taken into
  account

29
Quality/ Disability Adjusted Life Years
Quality Adjusted Life Years (QALYs)
Quality of Life (0-1 scale)
Time (years)

x
OR
30
Shortcomings

• People like QALYs/DALYs because simple, intuitive
• Huge following in clinical/ public health domain
• BUT
• Some economists hate QALYs/ DALYs
• Dont measure what they purport to measure
  (peoples preferences among health states)
• Ethical concerns
• Discriminates against elderly, disabled
• Big question left unanswered
• - how much should we spend to gain 1 QALY
• - 40,000 (?) see also Harvard Cost-utility
  Analysis Database

31
1991
1992
1993
1994
1995
1996
1997
1998
32
Sulfate and Quality Adjusted Life Expectancy

• Risk functions by age, sex, education, from
  re-analysis of ACS cohort
• Applied to population of Canada over lifespan
• Change in life expectancy estimated
• Quality adjustment based on health utilities
  index from NPHS

33
Results
Study Life Years Lost QALYs Lost
Brunekreef 0.19 (6 cities ACS)
Nevalainen and Pekkanen 0.17 (6 cities) 0.09 (ACS)
Wolfsson (smoking) 0.5 (females) 0.9 (males)
Coyle 0.05 (ACS) 0.01 (females) 0.07 (males)
34

• Risk coefficients from American Cancer Society
  cohort study applied to Canadian life tables
• Quality of life ratings from National Population
  Health Survey (Health Utilities Index)
• Substantial impact of sulfate on quality adjusted
  life expectancy
• Investment of over 1 billion/yr. would be
  warranted if it reduced sulfate concentrations
  by 1 µ g/m3.

35
(No Transcript)