Malaria transmission and prevalence in rice-growing versus non-rice-growing villages in Africa: a systematic review and meta-analysis

Background Rice fields in Africa are major breeding sites for malaria vectors. However, when reviewed in the 1990s, in settings where transmission was relatively intense, there was no tendency for malaria indices to be higher in villages with irrigated rice fields than in those without. Subsequently, intervention coverage in sub-Saharan Africa has been massively scaled up and malaria infection prevalence has halved. We re-examined this rice–malaria relationship to assess whether, with lower malaria transmission, malaria risk is greater in rice-growing than in non-rice-growing areas. Methods For this systematic review and meta-analysis, we searched EMBASE, Global Health, PubMed, Scopus, and Web of Science to identify observational studies published between Jan 1, 1900, and Sept 18, 2020. Studies were considered eligible if they were observational studies (cross-sectional, case-control, or cohort) comparing epidemiological or entomological outcomes of interest between people living in rice-growing and non-rice-growing rural communities in sub-Saharan Africa. Studies with pregnant women, displaced people, and military personnel as participants were excluded because they were considered not representative of a typical community. Data were extracted with use of a standardised data extraction form. The primary outcomes were parasite prevalence (P falciparum parasite rate age-standardised to 2–10-year-olds, calculated from total numbers of participants and number of infections [confirmed by microscopy or rapid diagnostic test] in each group) and clinical malaria incidence (number of diagnoses [ fever with Plasmodium parasitaemia confirmed by microscopy or rapid diagnostic test] per 1000 person days in each group). We did random-effects meta-analyses to estimate the pooled risk ratio (RR) for malaria parasite prevalence and incidence rate ratio (IRR) for clinical malaria in rice-growing versus non-rice-growing villages. RRs were compared in studies conducted before and after 2003 (chosen to mark the start of the mass scale-up of antimalaria interventions). This study is registered with PROSPERO (CRD42020204936). Findings Of the 2913 unique studies identified and screened, 53 studies (including 113 160 participants across 14 African countries) were eligible for inclusion. In studies done before 2003, malaria parasite prevalence was not significantly different in rice-growing versus non-rice-growing villages (pooled RR 0·82 [95% CI 0·63–1·06]; 16 studies, 99 574 participants); however, in post-2003 studies, prevalence was significantly higher in rice-growing versus non-rice growing villages (1·73 [1·01–2·96]; seven studies, 14 002 participants). Clinical malaria incidence was not associated with residence in rice-growing versus non-rice-growing areas (IRR 0·75 [95% CI 0·47–1·18], four studies, 77890). Potential limitations of this study include its basis on observational studies (with evidence quality rated as very low according to the GRADE approach), as well as its omission for the effects of seasonality and type of rice being cultivated. Risk of bias and inconsistencies was relatively serious, with I² greater than 90% indicating considerable heterogeneity.