Background
Remarkable progress in malaria control over the last decade has been attributed to massive deployment of malaria control interventions including long-lasting insecticidal nets (LLINs), indoor residual spraying (IRS), and case management with artemisinin-based combination therapy (ACT) [
1‐
3]. Studies of multiple malaria indicators in Tororo, Uganda, documented dramatic reductions in the incidence of malaria, prevalence of parasitaemia, test positivity rate (TPR), and annual entomological inoculation rates (aEIR) that began after IRS spraying commenced in December 2014 [
4‐
6]. Prior to that, only a change in TPR had been observed in the months after a mass LLIN distribution in November 2013 [
6]. The entomological data from Tororo suggests the effectiveness of vector-control interventions may have been modified by insecticide resistance [
7]. Here, changes in vector populations and behaviour were investigated using human landing catches before and after initiation of vector control.
Responses to vector control can be affected by coverage of the intervention, by the composition of vector species and by insecticide resistance [
8,
9]. Changes in the species composition of vector populations are commonly observed in response to vector-control interventions [
10‐
12]. The changes in composition commonly occur when the vector species that are more sensitive to a specific vector-control measure become less common, leaving vector species that are less sensitive. These changes in composition are most obvious when the dominant species is reduced, although minor vectors may also increase, decrease or remain unchanged.
Changes in mosquito physiology and behaviour result to ineffectiveness vector-control interventions [
13‐
15]. Physiological resistance involves changes in the sensitivity to the insecticides used in the control of the vectors, while behavioural resistance refers to changes that cause mosquitoes to avoid exposure [
9,
16,
17]. Examples of behavioural adaptations include changing from late to early feeding, shifting from indoor to outdoor feeding, avoiding resting on LLINs or the walls sprayed with insecticides, and developing a preference to feed on non-human blood [
13,
18‐
21].
The World Health Organization (WHO) recommends increased surveillance for changes in vector biology, physiology and behaviour following implementation of IRS and LLINs [
22]. In response to these recommendations, scientists working in Uganda, where pyrethroid resistance is prevalent and likely to undermine the impact of LLINs, have extensively studied the physiological adaptations of mosquitoes to insecticides [
7,
16,
23‐
25]. Findings from these studies have been used to inform the vector control policy in the country, including the rotation of insecticides to avoid prolonged selection for resistance, and the deployment of LLINs that have piperonyl-butoxide (PBO) in addition to pyrethroid insecticides to enhance their activity [
26]. However, little has been done regarding surveillance for behavioural adaptations and changes in malaria vector species composition following the roll-out of vector-control interventions in the country. Effects of LLINs and IRS on
Anopheles transmission intensity, biting behaviour and species composition in Tororo, a historically high malaria transmission area in Uganda were investigated. Results from this study will further inform roll-out of vector-control interventions in Uganda as the country moves towards achieving malaria control and eventual elimination as stipulated in the global technical strategy [
27].
Discussion
Large-scale deployment of LLINs and IRS in an area with historically high levels of malaria was associated with an eightfold reduction in vector densities as measured by human landing catches, an increased proportion of vector biting outdoors and a shift in vector dominance from An. gambiae s.s. to An. arabiensis. Vector control, including high coverage of LLINs, 3 rounds of IRS using bendiocarb, a carbamate insecticide, and 3 rounds of IRS using Actellic, an organophosphate insecticide, resulted in an 88% reduction in the HBR. There was a substantial reduction in both indoor and outdoor Anopheles HBR after over 5 years of mass distribution of LLINs and repeated rounds of IRS.
The combination of the interventions resulted in a 93% reduction in indoor biting by malaria vectors and a 49% reduction in outdoor biting. The observed reduction is explained by a Bayesian spatio-temporal model supporting the hypothesis that the sharp decline in vector numbers was associated with high LLIN coverage and IRS [
36,
37] although it is unclear whether the effect was due to LLINs or IRS or both. In a related study, no changes in malaria were observed in the aftermath of mass distributions of LLINs, but then sharp declines were observed after IRS [
6]. Indeed, both bendiocarb and Actellic are persistent and effective insecticides for use against mosquitoes [
38,
39]. However, although LLINs alone have been shown to reduce biting rates of
Anopheles and consequently transmission intensity in Uganda [
40,
41], LLINs alone were not sufficient to suppress malaria in northern Uganda after withholding IRS [
42]. It is possible the LLINs enhanced the effect of IRS, but it is impossible to say from these data.
Whilst the actual number of bites outdoors and indoors declined following vector control, the relative abundance of mosquitoes collected biting outdoors increased from 11.6% before vector-control interventions to 49.4% after vector-control interventions. There was a shift in the relative abundance of species, as well; 71.5% of mosquitoes were
Anopheles gambiae sensu stricto (s.s.) before vector control, whilst after vector control 79.4% were
An. arabiensis. In 2001–2002 and 2011–2012, the major malaria vector species reported in the area, in order of dominance, were
An. gambiae s.s.,
An. funestus and
An. arabiensis [
16,
28,
29]. However, the massive killing associated with vector control resulted in only one
An. gambiae s.s. being collected and no
An. funestus. Other studies have also shown that
An. funestus, which is highly endophilic, is susceptible to IRS [
43‐
45]. This shift in vector composition has been reported previously with the massive deployment of LLINs or IRS [
10‐
12,
46] and is associated with the preferential killing of the highly endophilic and anthropophilic
An. gambiae s.s., which is replaced by the more exophilic and zoophilic
An. arabiensis [
47]. In East Africa it has been shown that massive deployment of LLINs has resulted in a change in the species composition of the vectors, with the once dominant indoor vectors
An. gambiae s.s. being replaced by
An. arabiensis [
13,
15,
48]. In these cases, whilst the overall number of vectors declines, there are proportionately more
An. arabiensis than
An. gambiae s.s. Overall the number of
An. arabiensis biting outdoors declines, but they make up a larger proportion of the vector population after vector control than they did before.
Anopheles arabiensis is an efficient malaria vector and is capable of maintaining malaria transmission [
21,
49,
50], and is likely to do so in the study area even at low levels.
Both LLINs and IRS target mosquitoes that enter houses, and will also kill a proportion of outdoor-biting mosquitoes that enter houses, as indicated by the decline in number of outdoor-biting mosquitoes during the study. Thus, at least a proportion of mosquitoes collected outdoors are likely to enter houses during their lifetime. Although mosquitoes were not collected outdoors throughout the night, there was an indication that indoor-biting vectors were biting earlier in the evening after the intervention than before the intervention. This slight shift to early evening has been associated with the massive deployment of LLINs in sub-Saharan Africa [
51], with 21% of biting occurring before the time people are in bed, a percentage higher than previously recorded.
After vector control, none of the 243 potential malaria vectors tested positive for
P. falciparum sporozoites, compared to 1.8% sporozoites rate before the interventions were deployed, suggesting that vector control markedly reduced infectivity of mosquitoes with
falciparum sporozoites. These results are consistent with other studies that have shown a sharp reduction in sporozoites rates after vector control. For example, after 3 rounds of spraying with insecticides in Bioko Island, Equatorial Guinea, the sporozoite rate dropped from 8.3% before spraying to 0% after spraying with pyrethroid and carbamate insecticides [
52]. Similarly in western Kenya, the sporozoite rate dropped from 3.4% before intervention to 0.8% after intervention with permethrin-treated bed nets [
53]. These results suggest that high coverage and combinations of LLINs and IRS, where the mode of action of the insecticides on the walls and nets differ, are required to reduce the sporozoite rate.
Vector control was highly effective in the study area, but residual transmission begs the question, ‘What more needs to be done to eliminate malaria from the study area’? Addressing the issue of outdoor biting would help to further reduce malaria transmission, so mass drug administration or additional vector-based interventions that target outdoor biting could be considered [
14,
54,
55], such as environmental management and larviciding, the use of insecticides on cattle, toxic sugar baits traps, spatial repellents, or transgenic mosquitoes [
56,
57].
Limitations of the study
The study was not designed to evaluate the effectiveness of the interventions in comparison with absence of such interventions since no similar entomological data were collected in other areas without such interventions.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.