RESUMO
BACKGROUND: Although the recent decline of malaria burden in some African countries has been attributed to a scale-up of interventions, such as bed nets (insecticide-treated bed nets, ITNs/long-lasting insecticidal nets, LLINs), the contribution of other factors to these changes has not been rigorously assessed. This study assessed the trends of Plasmodium falciparum prevalence in Magoda (1992-2017) and Mpapayu (1998-2017) villages of Muheza district, North-eastern Tanzania, in relation to changes in the levels of different interventions and rainfall patterns. METHODS: Individuals aged 0-19 years were recruited in cross-sectional surveys to determine the prevalence of P. falciparum infections in relation to different malaria interventions deployed, particularly bed nets and anti-malarial drugs. Trends and patterns of rainfall in Muheza for 35 years (from 1981 to 2016) were assessed to determine changes in the amount and pattern of rainfall and their possible impacts on P. falciparum prevalence besides of those ascribed to interventions. RESULTS: High prevalence (84-54%) was reported between 1992 and 2000 in Magoda, and 1998 and 2000 in Mpapayu, but it declined sharply from 2001 to 2004 (from 52.0 to 25.0%), followed by a progressive decline between 2008 and 2012 (to ≤ 7% in both villages). However, the prevalence increased significantly from 2013 to 2016 reaching ≥ 20.0% in 2016 (both villages), but declined in the two villages to ≤ 13% in 2017. Overall and age specific P. falciparum prevalence decreased in both villages over the years but with a peak prevalence shifting from children aged 5-9 years to those aged 10-19 years from 2008 onwards. Bed net coverage increased from < 4% in 1998 to > 98% in 2001 and was ≥ 85.0% in 2004 in both villages; followed by fluctuations with coverage ranging from 35.0 to ≤ 98% between 2008 and 2017. The 12-month weighted anomaly standardized precipitation index showed a marked rainfall deficit in 1990-1996 and 1999-2010 coinciding with declining prevalence and despite relatively high bed net coverage from 2000. From 1992, the risk of infection decreased steadily up to 2013 when the lowest risk was observed (RR = 0.07; 95% CI 0.06-0.08, P < 0.001), but it was significantly higher during periods with positive rainfall anomalies (RR = 2.79; 95% CI 2.23-3.50, P < 0.001). The risk was lower among individuals not owning bed nets compared to those with nets (RR = 1.35; 95% CI 1.22-1.49, P < 0.001). CONCLUSIONS: A decline in prevalence up to 2012 and resurgence thereafter was likely associated with changes in monthly rainfall, offset against changing malaria interventions. A sustained surveillance covering multiple factors needs to be undertaken and climate must be taken into consideration when relating control interventions to malaria prevalence.