Environmental and anthropogenic factors associated with increased malaria incidence in South‐Kivu Province, Democratic Republic of the Congo

Objective To examine environmental and human factors that affect the spatial and temporal dynamism of malaria in DRC’s South‐Kivu province. Methods In a cross‐sectional study conducted between 1 January 2010 and 31 December 2015, spatial distribution was determined through thematic maps of malaria attack rate. SatScan ™ software and Monte Carlo test were used to identify spatial risk clusters. Temporal evolutions were analysed using the Cleveland algorithm. Generalized Additive Models for Location Scale and Shape and negative binomial regression were used to assess the independent human and environmental factors associated with incident malaria. Results The cumulative annual incidence of malaria increased from 10 968/100 000 in 2013 to 15 501/100 000 in 2015 (P for trend ˂0.001); malaria lethality increased from 0.1% in 2013 to 0.3% in 2015 (P for trend = 0.62). Between 2010 and 2015, 18 of 34 health zones consistently reported the highest attack rates, which ranged from 25 000 to 50 000/100 000. Four risk clusters areas were identified, with relative risk (RR) of 1.2 to 3.0, from which malaria was reported continuously during each year. Factors significantly associated with malaria cases were agro‐pisciculture practices (Incidence Risk Ratio [IRR]: 1.96; 95% CI: 1.23–3.13) and the presence of a lake in the health zone (IRR: 2.48, 95% CI: 1.51–4.42). Conclusions Malaria control in this setting must be intensified in peri‐lacustrine areas and those in which the population is intensively engaged in standing water‐associated activities. Objectif Examiner les facteurs environnementaux et humains qui affectent le dynamisme spatial et temporel du paludisme dans la province du Sud‐Kivu en RDC. Méthodes Dans une étude transversale menée entre le 1er janvier 2010 et le 31 décembre 2015, la distribution spatiale a été déterminée à l'aide de cartes thématiques du taux de cas de paludisme. Le logiciel SatScan™ et le test Monte Carlo ont été utilisés pour identifier les grappes de risques spatiaux. Les évolutions temporelles ont été analysées à l'aide de l'algorithme de Cleveland. Des modèles additifs généralisés pour l'échelle et la forme de l'emplacement et la régression binomiale négative ont été utilisés pour évaluer les facteurs humains et environnementaux indépendants associés à l’incidence du paludisme. Résultats L'incidence annuelle cumulée du paludisme est passée de 10.968/100 000 en 2013 à 15.501/100 000 en 2015 (p pour la tendance ˂0,001); la létalité du pal