Increased proportions of outdoor feeding among residual malaria vector populations following increased use of insecticide-treated nets in rural Tanzania

Abstract

Background: Insecticide-treated nets (ITNs) and indoor residual spraying (IRS) represent the front-line tools for malaria vector control globally, but are optimally effective where the majority of baseline transmission occurs indoors. In the surveyed area of rural southern Tanzania, bed net use steadily increased over the last decade, reducing malaria transmission intensity by 94%.

Methods: Starting before bed nets were introduced (1997), and then after two milestones of net use had been reached-75% community-wide use of untreated nets (2004) and then 47% use of ITNs (2009)-hourly biting rates of malaria vectors from the Anopheles gambiae complex and Anopheles funestus group were surveyed.

Results: In 1997, An. gambiae s.l. and An. funestus mosquitoes exhibited a tendency to bite humans inside houses late at night. For An. gambiae s.l., by 2009, nocturnal activity was less (p = 0.0018). At this time, the sibling species composition of the complex had shifted from predominantly An. gambiae s.s. to predominantly An. arabiensis. For An. funestus, by 2009, nocturnal activity was less (p = 0.0054) as well as the proportion biting indoors (p < 0.0001). At this time, An. funestus s.s. remained the predominant species within this group. As a consequence of these altered feeding patterns, the proportion (mean ± standard error) of human contact with mosquitoes (bites per person per night) occurring indoors dropped from 0.99 ± 0.002 in 1997 to 0.82 ± 0.008 in 2009 for the An. gambiae complex (p = 0.0143) and from 1.00 ± <0.001 to only 0.50 ± 0.048 for the An. funestus complex (p = 0.0004) over the same time period.

Conclusions: High usage of ITNs can dramatically alter African vector populations so that intense, predominantly indoor transmission is replaced by greatly lowered residual transmission, a greater proportion of which occurs outdoors. Regardless of the underlying mechanism, the residual, self-sustaining transmission will respond poorly to further insecticidal measures within houses. Additional vector control tools which target outdoor biting mosquitoes at the adult or immature stages are required to complement ITNs and IRS.

Figure 1

Kilombero and Ulanga districts (8.1°S and 36.6°E) in Tanzania showing Njagi and Lupiro villages.

Figure 2

The hourly indoor and outdoor biting profile of Anopheles gambiae s.l. (A - C) and Anopheles funestus (D - F) in the Kilombero Valley, Tanzania during 1997, 2004 and 2009. The grey shading represents the proportion of the human population indoors. The value of 1 was added to each mean to avoid zero values for presentation using a log scale.

Figure 3

The hourly indoor and outdoor profile of human contact with Anopheles gambiae s.l. (A - C) and Anopheles funestus (D - F) bites in the Kilombero Valley, Tanzania during 1997, 2004 and 2009. This stacked line graph presents estimates of human indoor and outdoor contact rates, taking into consideration the movement pattern of people by weighting the mean indoor and outdoor biting rates throughout the night by the proportion of humans that are typically indoors or outdoors at each time period [3]. The value of 1 was added to each mean to avoid zero values for presentation using a log scale.

Figure 4

Graphical comparison of the historical and recent estimates of the proportion of human contact with Anophelines occurring indoors (π_i). The proportion of human contact with mosquito bites occurring indoor (π_i) was calculated by taking into consideration the movement pattern of people using two methods: (A) by weighting the mean indoor and outdoor biting rates throughout the night by the proportion of humans that are typically indoors or outdoors at each time period and (B) using the formula: (I_{21→05 hrs})/(I_{21→05 hrs} + O_{05→21 hrs}).

Figure 5

The sibling species composition of the Anopheles gambiae s.l. complex (A) and monthly rainfall (B) between 2002 and 2009. For (A), the plotted line represents the predicted fit (β) of a GLM with a binary distribution and a logistic link function (Solid line = fitted; Dashed lines = se). References: 2002: [19]; 2003: [30]; 2005: [31]; 2007: [26,32]; 2008: [33] and Moore et al. Unpublished data; 2009: [34] and current data.