Study site

The study was carried out in Sare Alpha village (13°21' 38 N, 13°58' 50 W), in the Upper River Region of The Gambia, an area of open Sudanian savannah. There is a single rainy season (June to October) with malaria transmission mainly from August to December. The villagers live in delineated compounds with an average of 28 people and 10 dwelling rooms. Most houses are made of mud or cement with interior walls cemented and matt painted (67%) or left bare. Roofs are either thatch (51%) or metal and nearly every house has closed eaves (93%). The village received LLIN from the government countrywide distribution in March and June 2011 and the project provided additional LLIN (Olyset® Nets, Sumitomo Chemical Company, Japan) in July 2011 to achieve 97% (1754/1808 sleeping places) coverage.

Insecticide application

The 104 residential compounds were stratified into large (those with 11-30 rooms) and small (5 to 10 rooms) and random sampling was used to select 16 compounds in each strata. Eight compounds were then selected randomly, four in each strata, for spraying with one of the three insecticides or left untreated. Insecticides were applied at concentrations recommended by WHO [8]: 0.4g/m² bendiocarb (Ficam W, Bayer (Pty.) Ltd., Nigel, South Africa), 2g/m² DDT (DDT 75% WP, Hindustan Insecticides Ltd, New Delhi, India) and 1g/m² pirimiphos methyl (Actellic 300 CS, Syngenta, Seneffe, Belgium). IRS was conducted, 6-8^th July 2011, using Hudson X-pert sprayers at a rate of five seconds per 2m wall, by experienced spray-men from the Gambian National Malaria Control Programme under supervision and according to WHO guidelines [8].

During IRS, insecticidal sprays were sampled in 9-10 houses for each insecticide on Whatman No. 4 filter papers and the insecticide concentration was estimated by High Performance Liquid Chromatography, using Thermo, Fisher Scientific (Hemel Hempstead, UK) equipment and Chromeleon Version 6.80 SR11 software [15]. The results were expressed as grams of active ingredient /m² by reference to calibration curves of each insecticide.

Insecticide susceptibility testing

The susceptibility of wild caught An. gambiae s.l. was assessed using tube tests, viz. WHO bioassays [7]. Immature stages were collected from August to November 2011 in Sare Alpha and neighbouring Sare Juldeh (13°19' 32 N, 14°3' 36 W), transported to the insectary and reared to adults. Exposures were conducted at 29 ± 4°C and 63% relative humidity (range 45-75%). Approximately ten, three to five day old adult female mosquitoes were placed in holding tubes for 30 minutes and damaged mosquitoes removed. Remaining mosquitoes were then exposed for 60 minutes to insecticide-impregnated or appropriate control papers: (1) 0.1% bendiocarb, batch BE63; (2) 4% DDT, batch DD117; (3) 0.75% permethrin, batch PE190; (Vector Control Research Unit, University Sains Malaysia, Penang, Malaysia) (4) 0.025% pirimiphos methyl (Syngenta, Seneffe, Belgium). Insects were returned to the holding tube, given 10% glucose solution and mortality recorded after 24 hours. Live mosquitoes were those able to fly and any knocked down mosquito that had lost legs or wings was considered dead. Members of the An. gambiae complex were identified to species, M and S forms and kdr resistant markers by PCR [16,17,18].

Assessment of insecticide residual activity

Persistence of insecticides on walls was measured using WHO cone tests [7] in four houses with bare-mud walls and four with matt-painted cement for each treatment arm; four unsprayed houses with mud walls served as controls. A breeding stock of insecticide susceptible An. gambiae s.s. M form, originating from Yaoundé, Cameroon was obtained in June 2011 from the Pasteur Institute in Dakar and maintained in an insectary at 27 ± 2°C, 70-80% relative humidity and a 12: 12 h light : dark photoperiod. Adult mosquitoes were maintained with 10% glucose solution ad libitum. Larvae were fed daily with Tetramin^® Fish Flakes (Melle, Germany). Adults from the colony, 20 mosquitoes/tube test, were tested for sensitivity to permethrin, deltamethrin and DDT in July 2010 and were 100% sensitive.

Tests were conducted two months post-IRS and then at monthly intervals for a further three times. Cages of three to five day old, female mosquitoes were taken to the field and transferred to the cones in batches of 20, exposed to walls for 30 minutes and knock down was recorded. Cones were attached to walls furthest from the door, to minimize risk of abrasion, at 70cm below the roof, 70cm above the floor and in the middle. The positions were marked and repeated were measurements made at the same place. Mosquitoes were then transferred into holding cups in a cool box, taken to the insectary and mortality recorded 24h post-exposure. The average temperature during exposures was 32°C (range 27-37°C) and relative humidity decreased with time: 75% (range 68-85%) in the second and third month post-IRS, 61% (range 19-75%) for the fourth month and 12% (range 2-25%) in the fifth month.

Indoor sampling of mosquitoes

Eight sentinel rooms, in 32 separate residential compounds where a consenting adult slept under a bed net, were sampled for each intervention and the control arm using CDC light traps. Traps were positioned once a month, from 2^nd August to 2^nd December 2011, at the foot of the bed with the light 1m above the floor and operated throughout the night. Potential risk factors known to affect mosquito densities in The Gambia [19] were recorded at each capture. Specimens were frozen before morphological identification using established keys [20,21]. Members of the An. gambiae complex were identified to species, M and S forms, and kdr resistant markers by PCR as indicated above.

Focus Group Discussions

Given the invasive nature of IRS, its possible effect on bednet acceptance and usage, the different odours of the three insecticides and odour duration, we set out to explore the perceptions, attitudes and practices of adult residents of IRS-houses using Focus Group Discussions (FGDs) in Sare Alpha five months post-IRS. The FDG guide was prepared collectively by the principle co-authors and centred on the impact of the insecticides on mosquito abundance, house appearance, net use and the acceptability of the spray process. Three FGDs of approximately seven participants were held for each insecticide; a male group, a younger female and an older female group (31 years or older). A trained moderator with experience in qualitative research methods whose first language was Fula, the dominant language of the village, and who was familiar with local dialects, customs and values helped the groups discuss using the pre-prepared topic guide following a standard method [22]. With the consent of the participants the discussions were tape-recorded, and later transcribed into Fula, translated into English and typed into a word package.

Statistical analysis

Data were double-entered into Access databases, verified and validated by consistency checks. The analysis of the susceptibility tests followed WHO recommendations [7]. Fishers exact test was used to compare the susceptibility of mosquitoes to insecticides, the effect of the interventions on the presence of An. gambiae s.l. in light traps and to identify confounders. The Clopper-Pearson method was used to calculate 95% confidence intervals for proportions, unless 97.5% is stated. General Estimating Equations (GEE) population-averaged models with exchangeable correlation structure were fitted to ranked mosquito mortality from the cone-tests to compare insecticide residual activity and the influence of confounders. Random-effects logistic regression models were used to compare the presence or absence of mosquitoes in light traps by intervention arm and the influence of confounders on this. Statistical analysis used STATA (ver. 12, College Station, Texas, USA). FGDs were examined by content analysis using the pre-defined topics to create a code list and matching respondents comments to these, first within each FGD (Table S2) and then by insecticide. Where comments were congruent across the treatment arms these were presented together and in case of differences these were explored separately.

Ethics statement

The study proposal was reviewed and approved by the Scientific Coordinating Committee of the Medical Research Council Unit, The Gambia, and subsequently ethical approval was granted by the Gambian Government / Medical Research Council Unit, Joint Ethics Committee and the London School of Hygiene and Tropical Medicine Ethics Committee. Verbal, informed consent was provided by the village leaders following information meetings and documented, verbal, informed consent was provided by household heads for compounds enrolled into the study. Room owners provided written, informed consent for the mosquito collections and the cone tests.

Insecticide susceptibility

258 female An. gambiae s.l., raised from wild-caught larval specimens, were exposed to insecticides in tube tests at an average of 9.9 mosquitoes / tube. No mosquito (0/56; CI 0-6) died in the control tubes during the tests. The susceptibility of mosquitoes from Sare Alpha and neighbouring Sare Juldeh to DDT were similar (22/44 and 11/28, p=0.469) so the results were combined. Mortality 24h post-exposure was 46% (33/72; CI 34-58%) for DDT, 31% (16/52; CI 19-44%) for permethrin, 88% (37/42; CI 75-96) for bendiocarb, and 100% (26/26; 97.5% CI 87-100) for pirimiphos methyl.

More than half (58%, 148/257) of the female An. gambiae s.l. were An. gambiae s.s. and the rest were An. arabiensis (42%, 109/257); 90% (133/147) of the An. gambiae s.l. were S-form. Kdr 1014 genotypes were similar between villages but varied by species (Table 1). The 1014S (serine) mutation was only present in An. arabiensis, and at low frequencies, whilst the 1014F (phenylalanine) was present in >95% of An. gambiae s.s. S form but only in <20% of An. arabiensis (Table 1).

Insecticide residual activity

The estimates for concentrations of insecticide sprayed were higher than expected for bendiocarb (target dose= 0.4g/m², sprayed mean dose= 0.98g/m², CI 0.83-1.12) and DDT (target dose= 2g/m², sprayed mean dose= 3.44g/m², CI 1.25-8.92), but within the expected range for pirimiphos methyl (target dose= 1g/m², sprayed mean dose= 1.12 g/m², CI 1.03-1.21).

Mortality in control cone tests did not vary significantly between cone position nor with time after IRS (median 3.02%, inter quartile range = 5.60, p=0.07 from linear regression on log transformed percentages). At the first sampling time, two months post-IRS, residual activity varied with wall surface being 100% on walls with matt paint and lower on mud walls (Figure 1). Tests conducted three months post IRS showed similar values for the two surfaces and by the fourth and fifth months there was a tendency for residual activity to be lower on walls with matt paint. GEE population-averaged models of mosquito mortality averaged across the three wall positions (Table S1) showed significant interaction between the wall surface and time after spraying (p<0.001) but there was no significant difference between the insecticides. Similar results were found using mortality from the three wall positions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Insecticide residual activity during the first five months after IRS by insecticide.

https://doi.org/10.1371/journal.pone.0074351.g001

Sampling of adult mosquitoes

An. gambiae s.l. females were collected in 33% of traps; 36% in the control rooms (14/39), and 30% in the IRS rooms (36/117). Most An. gambiae s.l. (98%, 128/130) were identified to subspecies; 41% were An. arabiensis and 59% An. gambiae s.s., with the S form predominating in the latter (75%, 56/75). Kdr 1014 genotypes were identified for 53 An. arabiensis and 74 An. gambiae s.s. The results were similar to those of the larval caught specimens; the 1014S (serine) mutation was only present in An. arabiensis (gene frequency 20%, 21/106, CI 13-29%) and the 1014F (phenylalanine) predominated in An. gambiae s.s. (74%, 109/148, CI 66-81%), with only 13% in An. arabiensis (13/106, CI 7-20%).

The impact of IRS on presence of female members of the An. gambiae complex in traps was examined in a two-way table, but the proportions of traps with An gambiae s.l. were not significantly different in the presence or absence of IRS either when the catches from interventions were compared individually to control or when they were combined (Fishers exact test). However, several factors were significantly associated with presence of An. gambiae females in rooms: roofing material (metal or thatch), the interior wall surface (bare cement, bare mud or matt paint) and the number of people sleeping in the room. A formal multivariable analysis was performed using logistic regression with compound as a random-effect to examine these possible confounders and their interactions, together with those of open or closed eaves and the presence of a tethered horse. There was a significant interaction between roofing material and the different insecticides. Table 2 shows the odds ratios for IRS compared to control for metal and thatched roofs for each insecticide and adjusting for the number of people sleeping in the room. The effect was only significant for DDT in the presence of thatched roofs (Table 2), although a similar trend was apparent for all three insecticides

Focus Group Discussions

Spraying with any of the three insecticides was generally perceived as a necessary nuisance (Table S2). Participants disliked moving their furniture, having sprayers work unattended in their homes, waiting outside for two hours and the stains on the walls. The majority, however, said they were pleased with the spraying and were ‘rewarded with peace and health’. A few participants did not want their house sprayed, noting that ‘the mosquitoes did not die and the number of insects has slowly increased’. Most villagers did not state an intervention preference, but a few preferred bed nets: ‘[bed nets] last longer than spraying” and “can be taken outside for use’ and a few preferred IRS as 'in the hot season it is too hot to sleep under a net'. Insecticide odour was discussed frequently. The smell of DDT three days after spraying was perceived as intense but this waned and eventually disappeared: several observed that ‘The smell has gone and so mosquitoes return’. Participants from homes treated with pirimiphos methyl also noticed the smell, which ‘got less over time’ but ‘when you come close to the wall, you can smell it still now’. One villager observed: ‘because of the smell most mosquitoes are gone, but not all of them and they are coming back again’. Some residents also said that the number of rats and geckos had decreased. Many participants said that they continued to use their bed nets, for example in the in pirimiphos methyl-treated groups many slept under a net because ‘we can still hear mosquitoes’. Others in the DDT and bendiocarb groups mentioned that since IRS there were no mosquitoes they did not use the LLIN: “Even children slept without nets, as no mosquitoes were in the house. Mosquitoes are however starting to come back” and “We only bring down the net to protect us from the cold”.