The study was carried out in Van Duc A village, Bac Lieu province (Mekong Delta), southern Vietnam (9.18353 N, 105.30339 E). In previous studies An. epiroticus, Anopheles sinensis, Anopheles subpictus, Anopheles nimpe, Anopheles campestris and Anopheles vagus were collected in the study site 15. The main species of interest is An. epiroticus, which is also the most abundant. This population of An. epiroticus is resistant against deltamethrin, alpha-cypermethrin, etofenprox and cyfluthrin but susceptible to DDT 9. Additional WHO tube bioassays 16 were done using propoxur 0.1% and malathion 5% at discriminative dosage to obtain a full picture of the insecticide resistance status of the species. In 2008 bioassays were done with deltamethrin 0.05% to assess differences in resistance status between 2005 and 2008.

Six experimental huts adapted from the huts used in West Africa were built using local construction materials (Figure 1) 17. The huts were built on a concrete floor and had a wooden structure. The outside walls were covered with nipa-palm leaves. The lower parts of the inside walls were made of leaves and the upper part of plastic hessian sheeting. The roof was covered with palm tree leaves and inside with plastic sheeting. Each hut was surrounded by a water filled gutter to prevent the entry of scavengers such as ants. The entry side of the huts faced a large brackish water swamp. Entry slits were available in the front panel of the hut: two entry slits (0.75 m long) at each side of the door and one large slit above the door over the entire width of the front panel (3 m). These slits were designed to prevent mosquitoes from escaping once they entered the hut. The slits were covered by a plastic curtain from 5 a.m. till 6 p.m. Each hut was provided with a full screened veranda trap for collecting exophilic mosquitoes. The escape rate from the huts was evaluated by releasing 100 mosquitoes in each hut at 9 p.m. and re-capturing the mosquitoes in the morning from 6 till 8 a.m. The huts were adapted till at least 75% of mosquitoes were re-captured.

Unwashed and washed Permanet 2.0 and 3.0 were compared to a negative control (Table 1). Permanet 2.0 is a polyester LLIN coated with deltamethrin. PermaNet 3.0 is a complex long-lasting insecticidal net designed for the control of insecticide resistant mosquito populations. It comprises deltamethrin-coated-polyester side panels and a deltamethrin plus piperonyl butoxide (PBO) incorporated-polyethylene roof. The multifilament polyester (75 denier) yarns of the side panels have a deltamethrin load of 85 mg AI/m². The fabric of the lower part (70 cm) is denser compared to the higher part resulting in deltamethrin concentrations of 115 mg AI/m². The roof is made of 100 denier polyethylene monofilament containing 4 g AI/kg deltamethrin and 25 g/kg (± 25%) PBO by incorporation. The safety of PermaNet 3.0 has been reviewed by WHO, following the WHO generic risk assessment model for insecticide treatment of mosquito nets and their subsequent use and concluded that use and washing of these treated nets do not pose an undue hazard to the user 18.

Six holes each 4 cm × 4 cm were made in each mosquito net, two in each long side and one at each end, to simulate the conditions of a torn net and to put the emphasis on testing whether the insecticidal treatment, rather than the net, effectively prevents biting of the sleepers.

The design was adapted from the WHO guidelines 19. A base line study was conducted in December 2006 to evaluate the attractiveness of the six huts. The trial lasted for six weeks from 23 September till 28 October 2008 and treatment arms rotated weekly among the six experimental huts. Treatments were assigned randomly among huts in such a way that each treatment was tested once, during one week, in each hut according to a Latin Square Design. Six nets were used per treatment arm and each net was tested during one night of the week. Each morning the nets were removed and stored in their corresponding plastic bag. At the end of the week huts were cleaned and ventilated for one day before moving to another treatment. In each hut, a team of two adult volunteers slept under one net from 7 p.m. to 5 a.m. The six teams of volunteers rotated daily according to the Latin square design described above. Before the volunteers left the net at 5 a.m. they collected the dead and live mosquitoes inside the net. At 6 a.m., live and dead mosquitoes were collected from the hut (room) and from the veranda. Mosquitoes were identified using a standardized key for anophelines of Southeast Asia 20, counted and scored by hut and collection place as dead or alive, blood-fed, unfed or gravid. Live females were put in cups supplied with 10% sugar solution for 24 hours, after which any delayed mortality was recorded.

Ethical consideration: The protocol was approved by the ethical committees of the University Hospital of Antwerp (EC Nr 6/45/221) and of NIMPE Hanoi Vietnam. Volunteers gave written informed consent.

Cone bioassays, following the WHO procedures 19, were carried out on one net per arm before washing the nets, just before the trial in case the nets were washed and after the trial. Bioassays were done at NIMPE-Hanoi using 2-5 days old unfed females of a full susceptible An. dirus s.s. colony strain kept at NIMPE-Hanoi for ten years which allowed to evaluate the bio-availability of the insecticides on the net. For each tested net 10 pieces (2 × 4 side panels, 2 × 1 roof) of net were tested resulting in 50 mosquitoes per net. Mosquitoes were exposed during three minutes. Knockdown was recorded 60 minutes after exposure and mortality 24 h post exposure and compared to the WHO criteria for LLIN i.e. > 80% mortality and/or > 95% knockdown 21. For PermaNet 3.0 washed 20 times, 30 sections were tested (10 on the lower side panels below 70 cm, 10 on the higher side panels, 10 on the roof).

Net pieces were analysed before washing, after washing before the trial and after the trial. The side panels and roof were tested separately. The chromatographic determination of deltamethrin, deltamethrin R-isomer and piperonyl butoxide was based on the CIPAC method 32+33+345/TK/(M)/3. The chemical analyses were performed by a WHO reference centre (Pesticides Research Department of the Walloon Agricultural Centre, Gembloux, Belgium).

The outcomes measured in experimental huts were: (1) the entry rate or total number of mosquitoes found in the huts. This is used to measure the deterrent effect which is the reduction in hut entry relative to the control huts, (2) the exit rate measured as the proportion of mosquitoes found in the veranda, (3) blood feeding or proportion of blood fed female mosquitoes, (4) the mortality rate calculated as the proportion of dead mosquitoes, immediate and delayed observations combined, found in the huts. The percentage personal protection was calculated as (BF_C-BF_T)/(BF_C) * 100, where BF_C is the total number of blood fed females in the control hut and BF_T the total number of blood-fed female mosquitoes in the treated hut. The overall killing effect of a treatment was calculated as (D_T-D_C)/(T_C) * 100, where D_T is the total number of dead mosquitoes in the treated hut, D_C the total number of dead mosquitoes in the control hut and T_C is the total number of mosquitoes collected in the control hut 19. Statistical analysis of numbers (entry rates) was done by means of a negative binomial regression. The entry, blood feeding and mortality rates were compared between control arm and intervention arms by logistic regression (Stata 9).

The baseline survey lasted for six days in December 2006 during which 1,104 An. epiroticus, 24 An. nimpe, two An. sinensis and one An. campestris were collected. No difference in attractiveness was observed between the huts (Negative Binomial Regression, p > 0.05).

Mosquitoes were collected during 216 hut-nights (6 weeks × 6 days × 6 huts). from September till October 2008. In total 19,784 An. epiroticus, 400 An. nimpe, seven An. sinensis and 7,403 Culicinae were caught. Only An. epiroticus was further considered in the analyses of which the resistance status is given in figure 2.

Before any washing, mortality in cone bioassay was 100% for all treatments and knockdown was higher than 95%. Both PermaNet 2.0 and 3.0 showed mortality equal or higher than 98% after 20 washes and a knockdown of at least 86%. The deltamethrin CTN washed just before exhaustion (five washes) induced a mortality and a knockdown of respectively 90 and 94%. After the trial unwashed PermaNet 2.0 and 3.0 and the 20 times washed PermaNet 3.0 complied with the WHOPES criteria for LLINs, i.e. knockdown ≥ 95% and/or mortality ≥ 80%, whereas Permanet 2.0 was border line with a mortality of 78% and a knockdown of 88%. The deltamethrin CTN washed 5 times induced a mortality and a knockdown of only 24 and 38% respectively.

The active ingredient content for unwashed PermaNet 3.0 (deltamethrin and PBO) and PermaNet 2.0 (deltamethrin) complied with the target doses (± 25%), except for one unwashed net of PermaNet 3.0 for which the average deltamethrin content on the side panels (2.05 g AI/kg) was just below the lower limit (2.1 g AI/kg) of the target dose. For this same unwashed PermaNet 3.0, the relative standard deviation for the side panels was 12% (Table 1). The overall active ingredient retention for PermaNet 3.0 after 20 washes was 55% and 100% on the side panels and roof respectively and 72% for PBO on the roof. The overall deltamethrin retention for PermaNet 2.0 after 20 washes was 50%. The deltamethrin CTN had an initial concentration of 22.2 mg AI/m² which is in line with the target dose of 25 mg AI/m². After five washes this decreased to 5.4 mg AI/m².

The estimated deterrent effect for An. epiroticus varied between 20 and 35% for washed or unwashed PermaNet 2.0 or 3.0, and was only 7% with the CTN washed just before exhaustion (Table 2). In huts with untreated nets, 43% of the mosquitoes were collected in the veranda, and exophily ranged from 35% to 54% in the treatment arms. Blood feeding inhibition varied between 64% and 81% in the treated arms. PermaNet 3.0 washed 20 times did not induce a significantly higher blood feeding inhibition than Permanet 2.0 washed 20 times (Table 2). The overall mortality was relatively high in the untreated control arm (34%), which was mainly due to high mortality of the unfed females. The overall mortality with the PermaNet arms was between 96 and 82% compared to 70% for the CTN. The overall mortality among blood fed in the untreated control arm was low (10%) and relatively high in all treatment arms (79-94%). There were no significant differences between the different PermaNet arms but mortality among blood fed mosquitoes was significantly lower with the CTN. The overall insecticidal effect varied between 30 and 43%, the latter observed with unwashed PermaNet 3.0 (Table 2).