Prediction Of Malaria Transmission Risk Based On Environmental Factors Of Small-scale Spatial And Temporal Patterns

BackgroundMalaria is caused by a parasite called Plasmodium, which is transmitted via thebites of infected mosquitoes. The disease results from the multiplication ofPlasmodium parasites within red blood cells, causing symptoms that typically includefever and headache fever, headache, and vomiting, and usually appear between10and15days after the mosquito bite. Four species of malaria can infect and be transmittedby humans,including P. vivax, P. falciparum, P. malariae and P. ovale. In2010, about3.3billion people-almost half of the world's population-were at risk of malaria.Every year, this leads to about216million malaria cases and an estimated655000deaths. People living in the poorest countries are the most vulnerable.Historically, malaria was also a serious health threat in Anhui Province, easternChina, and the number of malaria cases was up to1.12million, accounting for33.9%of the malaria cases reported in the whole China in1980. Great attention was paid toanti-malarial programs by the Central Government, including policy planning andanalysis, funding allocations and the development of technical guidelines for malariacontrol and prevention. The number of cases decreased dramatically and endemicareas have been diminished after the active implementation of malaria controlmeasures since1980s. Many counties in Anhui Province reached the standard of thebasic malaria elimination (i.e. the incidence was less than1/10000) in the late1990s.Meanwhile, considerable success has been achieved for malaria control, and theannual incidence rate of malaria in Anhui Province was reduced from52.0/100000in1990to1.3/100000in1999. Since2000, malaria re-emergence was reported in somelocal areas along Huanghuai River Basin in Anhui, Henan and Jiangsu Provinces,central China. Especially in Anhui Province, both the number of malaria cases and theincidence were highest in the county, where34984malaria cases were reported in 2006, accounting for54.5%of the total number of malaria cases in mainlandChina.Even though malaria resurgence was evident in some areas, Anhui Province,what drove the resurgence of malaria is still unclear now. Although malariatransmission is affected by various factors, it is widely accepted that climaticconditions may influence Anopheles abundance and the frequency they bite, extrinsicincubation period (EIP) of the parasites within mosquitoes, human behaviors, whichcould significantly impact on malaria transmission, predictive models were developedto forecast malaria distribution under different scenarios of climate change. However,the impact of climate change on malaria transmission in some areas of Africa is stillhotly disputed, and especially it was focused on the link between global warming andworldwide increase in malaria. Despite this known association between malariatransmission dynamics and climate factors, there is still much uncertainty about thepotential impact of climate factors on malaria at local and global scales, and remains aconsiderable interest for studies. Also, these studies mostly involved Plasmodiumfalciparum malaria; it is unclear on the impact of climate factors on Plasmodiumvivax malaria, which has different vectors and characteristics of transmission. AnhuiProvince is vivax malaria endemic area, with resurgent malaria in certain parts ofAnhui; this provides a unique opportunity to study the association between vivaxmalaria and climate factors. Therefore, the priority of measures is the elimination ofinfection source, which is, detecting and treating malaria patients as much as possible.So it is very important to understand the spatial and temporal distribution and identifythe hotspots of malaria epidemics to target interventions against outbreaks of malaria.In this study, we aimed to describe the distribution of malaria epidemics over time andspace and explore the potential effect of climate variables on malaria transmissionacross different regions in Anhui Province, using long-term historical surveillancedata.ObjectivesUnderstanding where and why resurgence has occurred historically can helpcurrent and future malaria control programmes avoid the mistakes of the past. Toclarify endemic trend and the temporal and spatial distribution patterns of malaria in Anhui Province using long-term historical surveillance data over past20years, and toidentify the distribution of the hot spots;To analysis association between malariatransmission and climate change,environmental factors across different regions inAnhui Province and the predictive model of malaria transmission risk was constructedbased on climate factors; Results of studies may facilitate the development of earlywarning systems and forecasting system for reducing the incidence of malaria, andprovide scientific clues and theoretical evidence for the prevention and control offuture malaria outbreak. At the same time, spatial analysis technology (GIS, GS, GPS)and statistical methods (PDL model, Panel data regression model) were applied tomalaria research fields, which provide the method reference for the this wide-spreaddisease and other mosquito-borne diseases.MethodsThe monthly reported cases from1990to2009for each county were obtainedfrom Anhui Centre for Disease Control and Prevention; The monthly meteorologicaldata for the same period were extracted from monitoring stations in ChinaMeteorological Data Sharing Service System. Long-term historical surveillance dataof malaria in Anhui Province from1990-2009and maps with administrative boundarywere applied to plot the temporal and spatial distribution of malaria incidence;Spearman correlation analysis was employed to examine the crude association and lageffects between climate variables and monthly malaria incidence for each of differentgeographic regions; To examine the contribution of climatic factors to malariatransmission, a polynomial distribution lag (PDL) model was constructed by takinginto account the autocorrelation of the lagged effects; Akaike information criterion(AIC) was used to measure goodness-of-fit of the PDL models to the data; Theadjusted R-squared and residual of fitting were calculated, and Ljung-Box Q test wasconducted to verify whether the residual were white noise sequence; The predictivevalidity of the models was evaluated using the root mean square error, and Ljung-BoxQ test was performed to verify whether the residual were white noise sequence;Temporal scan statistical method,spatial autocorrelation analysis and spatial clusteranalysis were used to identify temporal and spatial distribution pattern of malaria and determine regions of hot spots at the county level; Each malaria case from January2005to December2010was geo-coded according to residential address and linked toa digital map using geographical information system(GIS) technologies in ArcGIS9.2software. A grid map was created using GIS technique, and each grid included an areaof25km2(5km×5km). The panel data analysis was utilized to analyzeenvironmental factors associated with malaria incidence.Results1. A total of198875cases were reported from1990to2009, distributed in allcounties in Anhui Province. Malaria incidence significantly decreased in all threegeographic regions during the1990s, however, since2000, a rapid increase of malariaincidence was reported, and then the annual incidence peaked in2006, reaching90.54per100000peoplein northern Anhui Province. Nevertheless, the annual incidencesremained at relatively low level in other two geographic regions, with the mean of3.47/100000±1.29, and1.09/100000±0.54in mid and southern Anhui Province from2000to2009, respectively. The seasonal influence in the incidence of malaria wasobvious, and June to November as peak times was founded. The map also showed ashift of the hotspots of malaria from counties in three geographic regions in1990s tocounties in northern Anhui Province in2000s over the past two decades. The spatialvariation in malaria epidemics over counties showed that annual average incidenceranged from0.45to88.65per100000peoplewith a median of10.45per100000people from1990to2009.2. The results of spatial autocorrelation showed that malaria incidence wassignificantly clustered at the county level. Appling the maximum spatial cluster sizeof <20%of the total population, the results of spatial cluster analysis identified a mostlikely cluster including Suixi, Suzhou, Lingbi, Xiaoxian, Huaibei, Guzhen,Mengcheng, Woyang, Lixin counties, which all located in the north of Huai River,center coordinates/radius:(33.678555N,116.699142E)/76.98km; the secondaryclusters including Mingguang, Wuhe, Fengyang counties, which all located in the eastof Anhui Province, center coordinates/radius:(32.8076N,118.1115E)/52.28km. 3. Only rainfall with lags of1-2months was significantly associated with themalaria incidence in all three regions(β=1.18,1.51,1.23). No significant associationwas found with temperature, humidity and MEI. The validation of these PDL modelsby using2008-2009data showed a good fit between observations and predictions, andhigh predictive powers of these models were achieved using the two-yearobservations in three regions. The predicting residuals were white noise sequence,RMSE was1.63,1.86,1.75, in all three regions respectively.4. The univariate panel data analysis showed the monthly relative humidity,monthly amount of rainfall, monthly average temperature and monthly NDVI withtheir time-lag variables from1to3months separately, elevation and land cover weresignificantly positive-correlated with monthly malaria incidence. Multivariate paneldata analysis demonstrated that monthly relative humidity with1-monthlag(IRR=1.29, P<0.01), monthly NDVI with2-month lag (IRR=1.05, P<0.01),monthly cumulative rainfall with1-month lag (IRR=1.11, P<0.01) and monthlyaverage temperature with1-month lag (IRR=3.53, P<0.01), rain-fed land wassignificantly associated with malaria incidence in Anhui Province (IRR=1.23,P<0.01).ConclusionWe observed substantial spatial heterogeneity in malaria epidemics in differentregions of Anhui Provinces and found that rainfall with lags of1-2months appearedto be the most important factor for the malaria re-emergence in Anhui Province.Meanwhile, this study identified the temporal and spatial distribution of hot spots atthe county level in Anhui Province fome1990to2009. Multivariate panel data modelanalysis demonstrated that monthly average temperature, monthly relative humidity,monthly cumulative rainfall, monthly NDVI, rain-fed land was significantlyassociated with malaria incidence in Anhui Province. The results of this studyfounded small-scale temporal and spatial distribution pattern and identified the mostimportant environmental factors on malaria transmission in the study areas, which would be beneficial in the prevention and control of malaria epidemic.