19
Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Embed Size (px)

DESCRIPTION

Malaria biology: malaria parasite (plasmodium) and vector (mosquito) Sporogonic cycle

Citation preview

Page 1: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Making useful climate-based predictions of malaria

Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Page 2: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Introduction to malaria

Weather & climate links to malaria

Current global state and outlook

Using seasonal climate forecasts to anticipate epidemics

Results for Botswana

Page 3: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Malaria biology: malaria parasite (plasmodium) and vector (mosquito)

Sporogonic cycle

Page 4: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Malaria dynamics depend on temperature# days for parasite to develop in

mosquito (sporogonic cycle)

Sporogonic cycle length > mosquito life cycle

Mosquitos take more frequent blood meals

(50% survive each blood meal: high temp = lower

mosquito rates)

Mosquito survival

Page 5: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Malaria dynamics depend on rainfall

Egg to adult takes 10 days on average (gonotrophic cycle)

Needs water!

Page 6: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

2015 statistics:

214m cases, 839,000 deaths (9 out of 10 in Africa)

Since 2000:~50% countries reduced

incidence by >75%Malaria mortality decreased

globaly by 60%

Millenium Development Goal 6C “to have halted and begun

to reverse the incidence of malaria” achieved

[source: WHO World malaira report 2015]

Current state of the world

Page 7: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

1900

2007

2007-1900

Endemicity class and change since pre-intervention (Gething et al 2010)

Intervention works!

Any increases in malaria due to climate change so far have been outweighed by impact of interventions & other factors

But what about the future?

Page 8: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Projections for 2080 [Caminade et al 2014]

• Warm/cold colours indicate longer/shorter transmission• Hatched area where models agreement on sign of change• Unquantified uncertainties remain…

Page 9: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Outlook (personal opinion!)

• Continuation of anti-malaria initiatives can deal with increased risk from climate change (climate is just one factor)

• Far future is uncertain (runaway climate change? Parasite mutation?)

Taking the shorter route (Washington et al 2006)

• Malaria epidemics are happening now!• Adapt to climate-related changes by anticipating variability

Use short-term forecasts to anticipate seasonal epidemics and mitigate the worst

Page 10: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Taking the shorter route- with seasonal climate forecasts

• Impossible to predict day-to-day changes beyond a week

• Slow fluctuations in surface conditions influence long-term average weather (e.g. El Niño)

Page 11: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Linking seasonal forecasts to malaria

• Seasonal forecasts indicate departures from normal temperature and precipitation, months in advance

• How to link temperature & precipitation anomalies to malaria?

Page 12: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Linking seasonal forecasts to malaria- the Liverpool Malaria Model

Page 13: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Validating climate-driven malaria forecasts

• Seasonal climate forecast + LMM = malaria forecast• But how good is it?

Hindcasting• Forecast as if we were in the past• Compare ‘forecast’ with observed data• Repeat for all available observations

• Not a lot of season average malaria data!• 1 data point per year• Botswana data (Thomson, 2003)• Clinical observed malaria cases, over January-May, 1982-

2003

Page 14: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Creating and validating climate-driven malaria forecasts- a recipe for Botswana

1. Create a seasonal climate forecast using System 4 (ECMWF seasonal climate model) – initialized separately at the start of every November 1981-2002

2. Use forecast precipitation & temperature to drive LMM3. Take ‘# infected humans’ from LMM and average across

January-May, and across Botswana4. Compare with observed malaria cases (Jan-May 1982-2003)

Page 15: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Validation of seasonal forecasts over Botswana

ObservationsSystem 4 seasonal forecast

Page 16: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Validation of seasonal malaria forecasts over Botswana

Observations + LMMSystem 4 seasonal forecast + LMM

Malaria incidence climatology

Page 17: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Validation of seasonal malaria forecasts over Botswana

Forecast probability of higher than normal malaria incidence

Page 18: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Implications

• In the long term forecast we beat the house…but• Impact of a forecast bust. Boy who cried wolf!

• Decisions to inform? • Preplacement & allocation of resources, funding appeal• Who takes responsibility? Less individual/institutional risk

in playing it safe

• Imperfect data• Uncertainty in validation• ‘Invisible skill’: is the model doing things well which we

can’t validate? e.g. timing of first outbreak of the season?

Page 19: Making useful climate-based predictions of malaria Dave MacLeod, Francesca di Giuseppe, Anne Jones, Cyril Caminade & Andy Morse

Recommendations

• More data!

• Better seasonal forecasts!

• Co-design: more involvement of end-users

See MacLeod et al 2015 Demonstration of successful malaria forecasts for Botswana using an operational seasonal climate model, ERL, OPEN ACCESS

Contact me: [email protected]