Scenariobased Health Risk Assessment: Modelling Tony McMichael NCEPH Wednesday 1st October 2003

1
Scenario-based Health Risk AssessmentModelling
Tony McMichael NCEPHWednesday 1st October 2003
2
Estimating Future Health Impacts Scenario-based
Health Risk Assessment

• Expert Judgement
• Simple extrapolation
• Integrated Assessment Modelling

3
Predicting Future Temperature-related Deaths in
Delhi
Relative mortality ( of daily average)
Daily mean temperature /degrees Celsius
4
But . . .What Else Might Change?

• Material living conditions
• Housing
• Urban landscapes
• Literacy, preparedness
• Prevalence of potentially fatal diseases
• Cardiovascular disease
• Chronic respiratory disease
• Etc.

5
Modelling. I

• Choose between biological (process-based) models
  and statistical-empirical models.
• Biological models assume that basic relationships
  between meteorological variables and biology of
  vector organism and parasite are sufficiently
  well-known.
• Statistical-empirical models eschew such
  assumptions, and work from current observable
  relationship between the local meteorological
  conditions and the regional occurrence of
  disease.

6
Modelling. II

• Models simplify a complex real world. Balance
  needed between clarity and computability and the
  comprehensive inclusion of meteorological-biologic
  al detail.
• Argument as to how much we should, and can,
  horizontally integrate foreseeable future
  changes in population characteristics (such as
  wealth levels, technology and education) that
  affect vulnerability to disease.

7
Modelling. III

• The validation of predictive integrated models
  is, by definition, difficult.
• Analogues may be found in recent experience,
  where climate variability or change appears to
  approximate future climatic changes.
• However, such opportunities are limited.

8
Diagram depicting annual number of heat-related
deaths attributable to temperature increase per
se and ageing, for Brisbane in 2050, under the
mid climate-change scenario and CSIROMk2 model
NCEPH/CSIRO/BoM, 2002
9
GHG EMISSIONS

• CLIMATE CHANGE

Geophysical characteristics - Land-use -
Hydrological engineering - Flood storm
protection
WEATHER MANIFESTATIONS, EVENTS
ENVIRONMENTAL/ ECOSYSTEM IMPACT
Distribution density of human settlements Built
environment infra-structure Emergency and
health-care systems
IMPACT ON HUMAN HEALTH
10
UVR gradient over Australia in Winter
11
Age-standardised MS prevalence and Malignant
Melanoma incidence in relation to ambient
ultraviolet radiation, Australia
80
70
Multiple sclerosis
60
50
MM incidence per 100,000
Malignant melanoma
Age-standardised MS prevalence (per 100,000)
40
30
20
10
0
2
3
4
5
6
Average annual ultraviolet radiation (kJ/m per
day), 1979-92
12
1500
Excess cases of skin cancer per million per year
1250
1000
US baseline rate 2000 cases/m/yr
750
500
250
100
1975
2000
2025
2050
2075
2100
1950
Year
Slaper et al, Nature 1996
13
20
19
Temperature (OC)
18
17
16
Modelled temperature change
15
14
The coming century
13
1900
2100
2000
2050
1950
1860
Year
14
2020s
GCM-Modelled increases in temperature under the
B2 SRES Scenario (UK Hadley Centre CM3 climate
model)
2050s
2080s
15
Climate change impacts on rain-fed cereal
production, 2080, Fischer et al, 2001
16
Africa
Parry et al., 1999
Reference Scenario

Climate Change Scenario
400
-7
300
? Additional Number of Malnourished People
Cereal production (mmt)
200
100
0
2020s
2050s
2080s
2020s
2050s
2080s
17
Malaria Dengue Fever Tick-borne
encephalitis Lyme Disease West Nile Fever Ross
River Virus
VECTOR-BORNE INFECTIOUS DISEASE
18
The changing distribution of malaria
a) The current distribution according to WHO
b) The distribution of malaria in 1850-1870
19
Modelling Climate Change Impacts on Future
Malaria Occurrence
Biologically-based models gtgt Where malaria
COULD be transmitted (solely in terms of
climate). Statistical models gtgt Where malaria
probably WOULD occur. Latter assumes current
social/economic/technological constraints will
apply in future. But conservative? Both have
been used primarily to focus on areas of
new/eliminated risk (cf. changes in annual
incidence). Change in total cases not (yet)
estimated.
20
Vectorial Capacity
Number of potentially infectious bites by a
mosquito population per infective person per day
ma2pn
VC
-ln(p)
m mosquito population density a human
biting index p daily mosquito survival rate
n incubation period of parasite in mosquito
21
Malaria Temperature and Biology
22
Proportional changes in Falciparum Malaria
Transmission Potential 2020s and 2080s versus
baseline climate scenario
Martens, Kovats, McMichael, et al., 1999
(UK)HadCM2 climate scenarios, within vector limits

23
Statistical model to predict falciparum malaria
in 2050 under BAU climate change (Rogers
Randolph, 2000)
Current distribution
Predicted change by 2050 (change in pop. at risk
25 mill.)
24
The End