Background:

Theobjectiveofthisstudyistoanalyzethespatialandtemporalpatternsofmalariaincidenceastodeterminethemeansbywhichclimaticfactorssuchastemperature,rainfallandhumidityaffectitsdistributioninMaputoprovince,Mozambique.

Methods:

ThisstudypresentsamodelofmalariathatevolvesinspaceandtimeinMaputoprovince-Mozambique,overatenyearsperiod(1999-2008).Themodelincorporatesmalariacasesandtheirrelationtoenvironmentalvariables.Duetoincompletenessofclimaticdata,amultipleimputationtechniqueisemployed.Additionally,thewholeprovinceisinterpolatedthroughaGaussianprocess.Thismethodovercomesthemisalignmentproblemofenvironmentalvariables(availableatmeteorologicalstations-points)andmalariacases(availableasaggregatesforeverydistrict-area).MarkovChainMonteCarlo(MCMC)methodsareusedtoobtainposteriorinferenceandDevianceInformationCriteria(DIC)toperformmodelcomparison.

Results:

ABayesianmodelwithinteractiontermswasfoundtobethebestfittedmodel.Malariaincidencewasassociatedtohumidityandmaximumtemperature.Malariariskincreasedwithmaximumtemperatureover28°C(relativerisk(RR)of0.0060and95%Bayesiancredibleinterval(CI)of0.00033-0.0095)andhumidity(relativerisk(RR)of0.00741and95%BayesianCI0.005141-0.0093).Theresultswouldsuggestthatadditionalnon-climaticfactorsincludingsocio-economicstatus,elevation,etc.alsoinfluencemalariatransmissioninMozambique.

Conclusions:

Theseresultsdemonstratethepotentialofclimatepredictorsparticularly,humidityandmaximumtemperatureinexplainingmalariaincidenceriskforthestudiedperiodinMaputoprovince.Smoothedmapsobtainedasmonthlyaverageofmalariaincidenceallowedtovisualizemonthsofinitialandpeaktransmission.Theyalsoillustrateavariationonmalariaincidenceriskthatmightnotberelatedtoclimaticfactors.However,thesefactorsarestilldeterminantformalariatransmissionandintensityintheregion.