Modeling The Impacts Of Household Environment And Climate Change On Malaria Incidences In Sub-Saharan Africa

Household environmental factors and global environmental factors that impacts malaria occurrence in countries of Sub-Saharan Africa attract wide attention.For one thing,household environmental factors that include distance or nature of water sources,household structures,education level,sanitation status,livestock rearing,bed-nets utilization,indoor residual spray interact with each other,and influence malaria in a complex and non-linear way.On the other hand,global climate change caused by human greenhouse gas(GHGs)emissions directly affects the spatial distribution of mosquitoes,lifestyle behaviors and the life cycle of malaria parasites,thus affecting the malaria infection process while formulating the spatial pattern of future malaria hotspots.Thus,it is of great significance to identify the key household environmental factors affecting malaria incidences and to predict the prevalence of malaria in different GHGs emission scenarios,especially for the effectiveness of malaria prevention.Currently,a large number of studies that use traditional statistical techniques only reveal the influence of a single household environmental factor on malaria incidences.However,knowledge gap still exist.These traditional statistical models lack the capacity to quantify the complex non-linear relationships between household environmental factors and their coupled effect on the incidences of malaria.Meanwhile,future climate change projections under different GHGs emission scenarios have huge uncertainties unavoidably when multiple IPCC models are used to predict malaria patterns.In view of the above problems,this study is based on internationally released databases with children's blood test data,household environmental survey data,GPS data for household clusters,spatial population data of Land Scan,remote sensing observation data and climate change forecast data from IPCC.At household level,Bayesian Belief Network(BBN)model was established to analyze the malaria control strategy from a perspective of household environment management.At regional scale,the GIS-BBN models were developed by combining geographic information system(GIS)and BBN to predict the impact of climate change on future malaria incidences.The results are as follows.(1)Based on the malaria survey database of sub-Saharan Africa,this study constructed a logistic regression model to compare the significance of single household environmental factor on malaria incidences by the parameter of Odds Ratios(OR).The results show that piped drinking water systems(i.e.tap-water,public and private stand piped water)significantly reduced malaria incidences ranging from 30-95%,compared with open water sources(i.e.wells,springs and rivers).The ORs of malaria occurrence decreased from 2.15 to 0.25 when time of households to fetch water from open water sources increased from 0 minutes to 60 minutes.Moreover,livestock as a non-malaria parasite host diverts mosquitoes from biting residents,and thus,reduced malaria incidences by 26-46%.In addition,the use of brick walls,metal roofs and cement on household floors reduced the ORs of malaria incidences by 0.46-0.82,compared to households with thatched roofs and muddy walls.(2)It is difficult to quantify the complex non-linear interactions among all the household environmental factors and rank their importance to malaria incidences by using various traditional statistical models.Based on the blood malaria test data of 141,233 children under five years of age from 10,340 households,this study developed the BBN model to quantify the complex effect of household environment factors on malaria incidences.The accuracy of BBN model is 86.39%to predict malaria incidences,with the receiving operation characteristic curve(ROC)area parameter of 0.82(CI 95%:0.64-0.94,p-value<0.05),logarithm loss of 0.53,square loss of 0.35 and spherical gain parameter of 0.80.By sensitivity analysis of BBN model,environmental factors influencing malaria were quantified and ranked.Entropy reduction of drinking water sources was 3.98%,household wealth 2.33%,wall materials 1.60%.(3)From global environment change perspective,we further developed spatially integrated BBN-GIS models to predict future malaria hotspots using the IPCC projected climate data of 21 CMIP-5 models.Scenarios of RCP 4.5 and RCP 8.5 of 2030,2050 and 2100 were included.The BBN-GIS models had an accuracy of 80.65%to predict four malaria risk categories,namely;no malaria,low,medium and high.Referring to the current malaria pattern,the probability of areas in the high category will reduce by 0.50-0.94%in projected years.For the medium category,malaria will reduce by 1.82-2.49%,but areas with low malaria will increase by 4.01-4.73%.The no malaria category,however,will reduce by 1.35-2.00%,implying that the risk of malaria extension will be unavoidable in any of the two future GHGs emission scenarios.Projected probability maps show that malaria hotspots will shift from the west to the east and southern Africa under RCP 8.5 GHGs emission scenarios.In summary,household environment and global environmental change together contributes to malaria occurrence in SSA in a coupled way.Strategies of household environmental management deserve much attention to achieve win-win goals for both malaria reduction and livelihood improvement in the context of climate change.