Long-lasting insecticidal nets provide protection against malaria for only a single year in Burundi, an African highland setting with marked malaria seasonality

Abstract

Background: Long-lasting insecticidal nets (LLINs) are one of the key interventions in the global fight against malaria. Since 2014, mass distribution campaigns of LLINs aim for universal access by all citizens of Burundi. In this context, we assess the impact of LLINs mass distribution campaigns on malaria incidence, focusing on the endemic highland health districts. We also explored the possible correlation between observed trends in malaria incidence with any variations in climate conditions.

Methods: Malaria cases for 2011-2019 were obtained from the National Health Information System. We developed a generalised additive model based on a time series of routinely collected data with malaria incidence as the response variable and timing of LLIN distribution as an explanatory variable to investigate the duration and magnitude of the LLIN effect on malaria incidence. We added a seasonal and continuous-time component as further explanatory variables, and health district as a random effect to account for random natural variation in malaria cases between districts.

Results: Malaria transmission in Burundian highlands was clearly seasonal and increased non-linearly over the study period. Further, a fast and steep decline of malaria incidence was noted during the first year after mass LLIN distribution (p<0.0001). In years 2 and 3 after distribution, malaria cases started to rise again to levels higher than before the control intervention.

Conclusion: This study highlights that LLINs did reduce the incidence in the first year after a mass distribution campaign, but in the context of Burundi, LLINs lost their impact after only 1 year.

Keywords: Control strategies; Descriptive study; Epidemiology; Malaria.

Figure 1

Topography of Burundi and indication of the health districts.

Figure 2

Median malaria incidence per 1000 population per health district in function of the elevation at the yearly malaria peak (2011–2019). The solid black line represents the median malaria incidence in function of elevation, as estimated by a generalised additive mixed model. Dotted lines represent boundaries where health districts were selected.

Figure 3

Weekly malaria incidence (cases per 1000 people) for endemic highland health districts between 1200 and 1850 m elevation in Burundi. The envelope represents the interquartile (dark red) cases for the health districts with minimum-maximum values (light red). Vertical dotted lines represent time periods when LLINs were distributed in the different health districts. LLINs, long-lasting insecticidal nets.

Figure 4

The effect of different time components on malaria incidence (cases per 1000 population) in the endemic highland health districts of Burundi as estimated by a generalised additive mixed model. (A) Seasonal fluctuations during the year. (B) The effect of LLIN distributions in 2014 and 2017. (C) The trend during the entire study period. The envelope represents the 95% CI on the average malaria incidence estimated over all time periods and health districts. LLINs, long-lasting insecticidal nets.

Figure 5

Estimated mean climate variables as function of seasonal and continuous time components as predicted by generalised additive mixed models for the endemic highland health districts in Burundi. (A, B) Show rainfall as function of time, (C, D) show changing day temperatures as function of time and (E, F) show changing night temperatures as function of time. The black dotted line in figure (E, F) represents the minimal required temperature for Plasmodium falciparum’s sporogony. The red envelope represents the 95% CI on the average climate variable estimated over all time periods and health districts (see online supplemental figure S2 for the full time-series model).