Modelling the transmission potential for vectorborne disease: Dengue Monday 29th September 2003 Simo

1
Modelling the transmission potential for
vector-borne disease DengueMonday 29th
September 2003Simon Hales, Neil de Wet, John
Maindonald, and Alistair Woodward.

• Ecology and Health Research Centre, Department
  of Public Health, Wellington School of Medicine
  and Health Sciences, Wellington, New Zealand
• International Global Change Institute, Hamilton,
  New Zealand
• Statistical Consulting Unit, Australian National
  University, Canberra, Australia.

2
Dengue

• Dengue fever is the worlds most important viral
  vector-borne disease.
• Affects hundreds of millions of people each year
• Transmitted predominantly by a single species of
  mosquito, Aedes aegypti.
• This species is adapted to living near to human
  habitation, feeds during the day and prefers
  humans to other species.

3
Vector biology

• Mosquitoes require standing water in order to
  breed
• Ambient temperature is also critical in
  determining adult feeding behaviour, mortality
  rates of larval development and virus replication
• If too cold, viral replication is slow
  mosquitoes are unlikely to survive long enough to
  become infectious

4

• Suitable climate is a necessary but not
  sufficient factor for disease transmission also
  need
• source of infection
• vector populations
• susceptible human population
• Can assume that all of these are present, then
  model transmission potential using a mathematical
  model based on vector biology

5

• Such models have been criticised on the grounds
  that they do not adequately account for
• rainfall
• interactions between climate variables
• socio-economic factors
• An empirical model can (in principle) allow for
  such factors

6
(No Transcript)
7

• Dengue fever was defined as present if any
  outbreak of the disease had been reported to WHO
  between 1975-1996
• Observed 1961-1990 average climate data obtained
  from UEA Climate Research Unit grids of 0.5
  degree latitude and longitude resolution
• Spatial averages of observed temperature,
  rainfall, humidity etc. estimated within
  administrative regions.

8
Model selection

• Logistic regression used to model the presence
  (1) or absence (0) of dengue
• Explanatory variables temperature, rainfall,
  humidity, singly or in combination
• Cross validation Leaving out each twentieth
  part in turn, we compared observed values for the
  omitted part with predictions from a model that
  is fitted to the remaining data.

9
Model fitting baseline

• The model (selected from previous step) was
  fitted using the original gridded climate data
• Modeled baseline risk of dengue transmission
  (between zero and one) for each 0.5 degree grid
  cell, on the basis of observed 1961-1990 climate.
• Model accuracy assessed by comparing the results
  with the known distribution of dengue fever.

10
Results

• Vapour pressure was found to be the single best
  climatic predictor of dengue transmission (OR
  1.3 SE 0.003 plt0.001).
• When areas at risk of dengue were defined by a
  cut-off probability of 0.5, the model classified
  89 of the grid cells accurately.

11
Vapour pressure
Predicted probability of dengue transmission
12
Model of baseline transmission potential
(1961-1990 climate)
13
Future risk

• We then used the regression model to estimate
  risk of transmission potential in a future world,
  based on GCM projections of climate

14
Model of future transmission potential (2080s
climate)
15
Population at risk

• Overlay of population data in GIS
• Estimated distribution of population in 1990 and
  projected population in 2055 and 2085
• Estimate effects of population increase alone,
  and effects of population increase and climate
  change together

16
Population at risk

• If humidity remained at baseline (1961-1990)
  levels, on the basis of projected increases in
  population
• 3.2 billion people (or 34 of the total) at risk
  in 2055
• 3.5 billion people (or 35 of the total) at risk
  in 2085.

17
(No Transcript)
18
Population at risk

• Using four alternative GCMs, 5 - 6 billion at
  risk of dengue by the 2080s (of a total
  population of 10 billion)

19
Conclusions

• Other factors being equal, dengue would be
  substantially worsened by climate change
• Uncertainties much greater than for traditional
  environmental health issues
• Need to try to account for effects of some of
  these other factors in future work