Published in:
Open Access 01-12-2014 | Research
Testing a multi-malaria-model ensemble against 30 years of data in the Kenyan highlands
Published in: Malaria Journal | Issue 1/2014
Login to get accessAbstract
Background
Multi-model ensembles could overcome challenges resulting from uncertainties in models’ initial conditions, parameterization and structural imperfections. They could also quantify in a probabilistic way uncertainties in future climatic conditions and their impacts.
Methods
A four-malaria-model ensemble was implemented to assess the impact of long-term changes in climatic conditions on Plasmodium falciparum malaria morbidity observed in Kericho, in the highlands of Western Kenya, over the period 1979–2009. Input data included quality controlled temperature and rainfall records gathered at a nearby weather station over the historical periods 1979–2009 and 1980–2009, respectively. Simulations included models’ sensitivities to changes in sets of parameters and analysis of non-linear changes in the mean duration of host’s infectivity to vectors due to increased resistance to anti-malarial drugs.
Results
The ensemble explained from 32 to 38% of the variance of the observed P. falciparum malaria incidence. Obtained R2-values were above the results achieved with individual model simulation outputs. Up to 18.6% of the variance of malaria incidence could be attributed to the +0.19 to +0.25°C per decade significant long-term linear trend in near-surface air temperatures. On top of this 18.6%, at least 6% of the variance of malaria incidence could be related to the increased resistance to anti-malarial drugs. Ensemble simulations also suggest that climatic conditions have likely been less favourable to malaria transmission in Kericho in recent years.
Conclusions
Long-term changes in climatic conditions and non-linear changes in the mean duration of host’s infectivity are synergistically driving the increasing incidence of P. falciparum malaria in the Kenyan highlands. User-friendly, online-downloadable, open source mathematical tools, such as the one presented here, could improve decision-making processes of local and regional health authorities.