A total of 120,580 Anopheles spp. mosquitoes were collected, with slightly fewer collected in study year two (53,536; 44% of total) compared to year one (67,044; 56%). More Anopheles spp. were collected using HLC (74,201; 60%) than CDC LTs (48,379; 40%). Of those Anopheles spp. for which the sex of the specimen was determined (120,249; 99.7%), more than 97% (117,107) were female and 2.6% (3,142) were male, and more than 92% of the male specimens (2,899) were collected in CDC LTs. The most abundant species morphologically identified was An. squamosus (37,203; 31%), followed by An. funestus s.l. (22,987; 19%), An. tchekedii (22,755; 19%) An. coustani s.l. (19,669; 16%), An. maculipalpis (5,786; 5%), An. tenebrous (4,730; 4%), An. gambiae s.l. (2,721; 2%), and An. gibbinsi (2,312; 2%).
A total of 14,306 specimens from ten different species or species groups were screened for P. falciparum CSP antigens in the head and thorax (Table 1). Hereafter, mosquitoes positive for the P. falciparum CSP antigen are referred to as either being sporozoite positive or as being infectious. The specimens subsampled for ELISA analysis were also identified to species by PCR. There were 165 positive specimens, 96.4% of which (159) were An. funestus s.l. Other P. falciparum positive species included An. coustani s.l. (3), An. gambiae s.l. (2), and An. squamosus (1).
Table 1
Results from the P. falciparum CS ELISAs.
| P. falciparum ELISA results | | | |
Species | Negative | Positive | Crude SP | % of all sporozoite positives (N = 165) | Overall number of female mosquitoes collected a | Estimated number of infectious mosquitoes collected | Estimated % of all infectious mosquitoes (N = 716) |
An. funestus s.l. | 4,609 | 159 | 3.3% | 96.4% | 20,337 | 678 | 94.7% |
An. coustani s.l. | 2,500 | 3 | 0.1% | 1.8% | 19,564 | 23 | 3.3% |
An. gambiae s.l. | 1,122 | 2 | 0.2% | 1.2% | 2,642 | 5 | 0.7% |
An. squamosus | 3,868 | 1 | 0.03% | 0.6% | 36,881 | 10 | 1.3% |
An. gibbinsi | 3 | 0 | - | - | 2,287 | - | - |
An. maculipalpis | 1,979 | 0 | - | - | 5,765 | - | - |
An. pretoriensis | 3 | 0 | - | - | 12 | - | - |
An. rufipes | 41 | 0 | - | - | 163 | - | - |
An. tchekedii | 13 | 0 | - | - | 22,665 | - | - |
An. tenebrosus | 3 | 0 | - | - | 4,645 | - | - |
Total | 14,141 | 165 | - | 100% | 114,961 | 716 | 100% |
SP = Sporozoite Positivity (proportion positive out of all tested); a Total number of females collected in CDC LTs and HLC.
While most infectious bites would have occurred indoors (61.5%), outdoor transmission also likely occurred during the study (Table 2).
Table 2
Location of infectious mosquito capture by species (combined CDC LT and HLC results).
| Estimated number of infectious mosquitoes1 (% of all infectious mosquitoes) |
Species | Indoors N (%) | Outdoors N (%) |
An. funestus | 421 (58.8%) | 257 (35.9%) |
An. coustani | 14 (2.0%) | 9 (1.3%) |
An. gambiae | 5 (0.7%) | 0 (0.0%) |
An. squamosus | 0 (0.0%) | 10 (1.4%) |
Total | 440 (61.5%) | 276 (38.6%) |
1 Location specific proportion sporozoite positive multiplied by the total number of mosquitoes collected in that location |
Within the An. funestus group, 93.5% (3,287) of specimens were confirmed by PCR as An. funestus s.s. – including all CS-positive specimens. Other species identified include An. leesoni (126), An. parensis (90), An. rivulorum-like (6), An. longipalpis (4), An. funestus-like (4), An. rivulorum (3), and An. vaneedeni (2).
Within the An. gambiae complex, 97.7% (1,042) of morphologically identified specimens were confirmed to species by PCR as An. arabiensis, 1.7% (18) were An. gambiae s.s., and 0.7% (7) were An. quadriannulatus. Of the 2 sporozoite positive specimens from this complex, one was An. arabiensis and one was An. gambiae s.s. – both of which were collected indoors. Subsequent entomological results and analysis presented here focus on female An. funestus s.l., the clearly dominant vector species in the study area during the trial, though it should be noted that An. coustani s.l., An. gambiae s.s., An. arabiensis, and An. squamosus may also have played a minor role in malaria transmission as secondary vectors.
Standard WHO tube bioassay tests [21] using locally captured An. funestus s.l. adult females from the study site indicated high levels of resistance to the pyrethroids alphacypermethrin, deltamethrin, and permethrin, with 24-hour mortality ranging from 46–60% [16]. Resistance was not detected to pirimiphos-methyl, nor to dinotefuran using the topical assay [22].
During both study implementation years high ATSB station coverage was reported across all intervention clusters: among eligible structures assessed, 93.1% (95%CI 91.6% − 94.7%) had 2 ATSB stations in any condition hanging and 71.5% (95%CI 67.1% − 75.8%) had 2 ATSB stations in good condition hanging [20]. While programmatic data and outcome indicators from the IRS and ITN campaigns cannot be aligned with the study clusters, data from the annual cross-sectional surveys (Ashton et al., in review) confirm that the entomological surveillance clusters were balanced in terms of the proportion of households having received IRS within the previous year, the proportion of households that owned one net per 2 persons, and the proportion of household residents that slept under a net the previous night (Table S1).
A total of 3,131 female An. funestus specimens collected by HLC were successfully assessed for parity status, 1,477 from control clusters and 1,654 from intervention clusters (Table 3). The mean non-parous proportion was 23.0% (95% CI 18.2% − 28.7%) in the control arm and 21.2% (95%CI 18.8% − 23.9%) in the intervention arm, resulting in an unadjusted OR = 1.05 (95%CI 0.82–1.34; p = 0.688). The non-parous proportion and the total number of mosquitoes dissected were both higher in year 2, but there were no significant differences between study arms in either year. Covariate adjusted analysis accounting for the number of months since ATSB deployment, calendar month (as a seasonal adjustment), and location of mosquito capture (indoors or outdoors) did not change the effect estimate of ATSB deployment on An. funestus non-parous proportion (OR = 1.00 [95%CI 0.78–1.24, p = 0.914]).
Table 3
Proportions non-parous for An. funestus collected during HLC activities.
| Control arm | Intervention Arm | Primary unadjusted OR |
| Number of An. funestus assessed | Proportion non-parous | Number of An. funestus assessed | Proportion non-parous | OR | p |
| (95% CI) * | (95% CI) * | (95% CI) |
Year 1 | 181 | 13.3% | 118 | 14.4% | 0.90 | 0.826 |
(6.2% − 26.1%) | (11.0% − 18.6%) | (0.35–2.33) |
Year 2 | 1296 | 24.4% | 1536 | 21.7% | 1.10 | 0.453 |
(19.9% − 29.5%) | (19.4% − 24.3%) | (0.86–1.42) |
Overall | 1477 | 23.0% | 1654 | 21.2% | 1.05 | 0.688 |
(18.2% − 28.7%) | (18.8% − 23.9%) | (0.82–1.34) |
n = number of female An. funestus assessed as non-parous; N = total number of female An. funestus assesses for parity |
*Robust standard errors accounting for clustering effects |
A total of 12,206 female An. funestus were collected in CDC LTs during the study: 7,305 from control clusters and 4,901 from intervention clusters (Table 4). The mean number of An. funestus mosquitoes collected per household per night (indoor and outdoor traps) was 4.8 (95%CI 2.5– 7.0) in the control arm and 3.2 (95%CI 1.5–4.9) in the intervention arm, resulting in an unadjusted RR of 0.65 (95%CI 0.30–1.40, p = 0.267). In general, An. funestus abundance was higher in year two than in year one and indoors compared to outdoors, but there were no significant differences in An. funestus abundance across study arms (Table 4). Covariate adjusted analysis accounting for the number of months since ATSB deployment and calendar month did not result in any change in ATSB effect size or significance (RR of 0.64 (95%CI 0.30–1.40, p = 0.265). Study arm specific light trap densities by cluster, by year, and by month are presented in Figure S1.
Table 4
Mean number of Anopheles funestus females collected in CDC LTs
| Control arm | Intervention Arm | Primary unadjusted RR |
N (Trap nights) | Mean An. funestus per trap night | N (Trap nights) | Mean An. funestus per trap night | RR | p |
(95% CI) * | (95% CI) * | (95% CI) |
Year 1 |
Indoor | 782 | 2.6 | 778 | 1.6 | 0.71 | 0.453 |
(0.9–4.3) | (0.4–2.8) | (0.28–1.74) |
Outdoor | 782 | 1.3 | 778 | 0.6 | 0.50 | 0.198 |
(0.6–2.1) | (0.3–1.1) | (0.18–1.43) |
Total | 1,564 | 3.9 | 1,556 | 2.3 | 0.65 | 0.361 |
(1.5–6.4) | (0.9–3.8) | (0.25–1.64) |
Year 2 |
Indoor | 752 | 4.2 | 745 | 3.1 | 0.64 | 0.213 |
(2.6–6.0) | (1.5–4.7) | (0.31–1.29) |
Outdoor | 752 | 1.3 | 745 | 1.0 | 0.70 | 0.306 |
(0.9–1.8) | (0.6–1.4) | (0.36–1.38) |
Total | 1,504 | 5.6 | 1,490 | 4.1 | 0.65 | 0.232 |
(3.5–7.8) | (2.1– 6.1) | (0.32–1.31) |
Overall |
Indoor | 1,534 | 3.4 | 1,523 | 2.3 | 0.66 | 0.277 |
(1.7–5.1) | (1.0–3.7) | (0.31–1.40) |
Outdoor | 1,534 | 1.3 | 1,523 | 0.8 | 0.61 | 0.242 |
(0.7–1.9) | (0.5–1.2) | (0.27–1.39) |
Total | 3,068 | 4.8 | 3,046 | 3.2 | 0.65 | 0.267 |
(2.5–7.0) | (1.5–4.9) | (0.30–1.40) |
*Robust standard errors accounting for clustering effects
A total of 8,131 female An. funestus were collected during HLC: 3,924 from control clusters and 4,207 from intervention clusters (Table 5). As was noted with CDC LT abundance (Table 4), mosquito human landing rates were higher in year two than in year one, though rates were higher outdoors in year one and indoors in year 2. The mean number of mosquito landings per household per night (indoor and outdoor) was 2.5 (95%CI 1.2–3.8) in the control arm and 2.7 (95%CI 0.59–4.8) in the intervention arm. Despite the highly similar mean landing rates across study arms, the modeled unadjusted RR is 0.68 (95%CI 0.23–2.00, p = 0.479), which, while not statistically significant, does suggest an overall trend towards fewer mosquitos per night in the ATSB arm when accounting for the effects of clustering. These discordant results are likely because of the substantial variation in HLC landing rates observed between clusters (intercluster coefficient of variation = 0.948), months, and years (Figure S2). Covariate adjusted analysis accounting for the number of months since ATSB deployment, calendar month, and HLC collection team did not result in any change in ATSB effect size or significance (RR of 0.65 (95%CI 0.22–1.89, p = 0.426).
Table 5
Female Anopheles funestus landing rates during HLC
| Control arm | Intervention Arm | Primary unadjusted RR |
N (Trap nights) | Mean An. funestus per collection night | N (Trap nights) | Mean An. funestus per collection night | RR | p |
(95% CI) * | (95% CI) * | (95% CI) |
Year 1 |
Indoor | 784 | 0.68 | 782 | 0.34 | 0.60 | 0.534 |
(0.21 − 1.2) | (0.10–0.60) | (0.11–3.04) |
Outdoor | 784 | 1.1 | 782 | 0.69 | 0.49 | 0.286 |
(0.39–1.9) | (0.13–1.2) | (0.13–1.83) |
Total | 1,568 | 1.8 | 1,564 | 1.0 | 0.53 | 0.370 |
(0.61– 3.0) | (0.27–1.8) | (0.13– 2.13) |
Year 2 |
Indoor | 782 | 2.1 | 784 | 2.5 | 0.58 | 0.350 |
(0.96– 3.3) | (0.29–4.7) | (0.18–1.83) |
Outdoor | 782 | 1.1 | 784 | 1.8 | 0.84 | 0.792 |
(0.39–1.7) | (0.38–3.2) | (0.23–3.06) |
Total | 1,564 | 3.2 | 1,568 | 4.3 | 0.64 | 0.475 |
(1.4–5.0) | (0.68– 8.0) | (0.19–2.16) |
Overall |
Indoor | 1,566 | 1.4 | 1,566 | 1.4 | 0.56 | 0.320 |
(0.71–2.1) | (0.22–2.6) | (0.18–1.75) |
Outdoor | 1,566 | 1.1 | 1,566 | 1.3 | 0.82 | 0.730 |
(0.46–1.7) | (0.35–2.2) | (0.27–2.50) |
Total | 3,132 | 2.5 | 3,132 | 2.7 | 0.68 | 0.479 |
(1.2–3.8) | (0.59–4.8) | (0.23–2.00) |
*Robust standard errors accounting for clustering effects |
A total of 159 of 4,767 An. funestus specimens were positive for P. falciparum sporozoites (Table 1) (an overall crude SP = 3.3%). The mean sporozoite positivity was 2.7% (95%CI 1.7% − 4.1%) in the control arm and 4.1% (95%CI 3.0% − 5.5%) in the intervention arm, with an unadjusted OR of 1.55 (95%CI 0.93–2.57, p = 0.090). Covariate adjusted analysis accounting for the collection method and location, as well as the number of months since ATSB deployment and the calendar month, did not result in any change in ATSB effect size or significance (OR = 1.57 [95%CI 0.94–2.63, p = 0.088]).
To compare EIRs across the study arms, cluster-specific estimates of SP and human landing rate for An. funestus were used to independently calculate estimated EIRs for each cluster (Table S2). The overall monthly EIR for An. funestus was 4.6 (95% CI 2.4–6.7) infectious bites per household per month in the control arm and 7.4 (95%CI 3.0–11.7) in the intervention arm, a non-significant difference of 2.8 infectious bites per household per month (95%CI -7.3–1.7, p = 0.216).