Two technologies of mosquito nets were considered in this study.

Coated technology: In this technology a resin based polymer coating is used as the insecticide reservoir for replacement of surface insecticide and this coating is bonded to the surface of the filament [17]. This technology is usually applied to multifilament polyester textile nets. Two types of coated nets were used for this study: Interceptor® nets treated with alpha-cypermethrin (200 mg a.i./m² of net) provided by BASF Chemical Company and PermaNet®2.0 nets treated with deltamethrin (55 mg a.i./m² of net) provided by Vestergaar Frandsen SA.

Incorporated technology: In this technology, the pyrethroid insecticide is directly incorporated into the textile fibers from which the netting is made. This technology can be applied for polyethylene nets [17]. The insecticide diffuses constantly over time to the surface of the yarn and will be regenerated from the reservoir after the surface insecticide is washed off or is lost otherwise. The bioavailability of the active ingredient is designed to be sufficient to kill the mosquitoes [25]. One type of incorporated nets is used in this study: Netprotect® nets. These nets are treated with deltamethrin (79 mg a.i./m² of net) and provided by the Dean Superior Textile Co.

For the laboratory hand washing simulation and the indoor drying, each type of net was cut into 63 pieces of 25 cm×25 cm. For the ISO 6330∶2000 machine washing and the indoor and UV drying, 20 bigger pieces of about 70 cm×70 cm were cut from each type of net.

Two washing procedures were tested to know their impact on active ingredient content of the nets.

Method-1 corresponds to the WHO washing procedure [26], except that the CIPAC standard washing agent is used instead of “Savon de Marseille”. It is further on referred to as laboratory hand washing simulation.

Method-2 is the ISO 6330∶2000 domestic washing further on referred to as the ISO 6330∶2000 machine washing. It is considered to be the more stringent [27] European standard for domestic washing of textiles.

Both methods are standardized. Therefore, it was possible to compare with other studies using these washing procedures.

Two drying procedures were tested to study their impact on active ingredient content of the nets.

Because it was expected that the samples would have low amounts of active ingredient due to the impact of the washing cycles, a multi-residue method able to detect deltamethrin and alpha-cypermethrin in coated and incorporated nets with a high sensitivity was used [33]. The analyses were done at the laboratory of Crop Protection Chemistry of the Ghent University. 300 mg (to the nearest 0.1 mg) of sample was weighed into a 100 mL conical flask. The active ingredient was extracted by heating under reflux during 30 minutes with 40 mL of xylene. The extract solution was allowed to cool down to the ambient temperature and was filtered through a büchner funnel using Whatman™ filter paper into a 50 mL volumetric flask. The filtration cake was rinsed and the extract solution was extended to 50 mL with xylene. After that, 1 mL of the extract solution was diluted into 10 mL with xylene and a portion of this solution was transferred into a vial for chromatographic injections.

The extract was analyzed by gas chromatography with electron capture detection (GC-µECD) using an Agilent Technologies 6890 N equipped with an auto sampler Agilent Technologies 7683 Series injector which was used in split mode. The chromatographic separation was performed with a HP-5 (5% Phenyl Methyl Siloxane) capillary column (30 m×0.250 mm i.d., 0.25 µm film thickness). Helium was used as the carrier gas and kept at constant pressure of 102.7 kPa with a nominal minimal flow of 0.9 mL/min. The split ratio, split flow and total flow were respectively 50∶1, 45.5 mL/min and 49.9 mL/min. The µECD detector temperature was 300°C with nitrogen as make-up gas kept at a constant flow of 60.0 mL/min. For each sample two chromatographic injections were done and the mean was reported as mass of active ingredient per unit mass of netting (g/kg), by comparison with an external standard calibration curve. Then, the value was converted into mg/m² according to the given density of the net. The injection volume was 1 µL and the oven temperature was programmed as: isothermal at 130°C for 1 min, from 130°C to 280°C at 30°C/min and held for 16 min.

According to A. Hill in the report of the 11^th WHOPES Working Group Meeting [34], WHO standard bioassays cannot be used throughout the world for quality control purposes. So, physico-chemical tests must be used. But the use of such tests to define efficacy is problematic. As a minimum for the development of a meaningful specification, it is essential to have a detailed knowledge of the release/retention characteristics of the product. Data were computed to check firstly, if the main characteristic of the LNs could be related to the number of washing cycles. Secondly it was checked which trend or mathematic model could be used to fit the relationship between the release of the active ingredient from the LNs and the number of washing cycles. The data were fitted with the curve estimation program from SPSS 20 and 4 mathematical models (linear, logarithmic, quadratic and exponential) were tested. To check whether the main characteristics of the LNs could be related to the number of washing cycle, the active ingredient retention at each wash was calculated as:with wash_n the active ingredient content after the n wash and Wash₀ the initial active ingredient content of the unwashed net.

Two groups of samples were considered from each type of LN. The first consisted of 10 pieces of about 70 cm×70 cm. This group was washed following the ISO 6330∶2000 machine washing procedure (method-2) up to 20 times. The second group consisting of 60 pieces of 25 cm×25 cm was washed following the laboratory hand washing simulation procedure (method-1) up to 20 times. Between the washes, both groups were dried indoor and stored at the same condition before the next wash. To check whether the two washing procedures were significantly different, the active ingredient content of the 1, 5, 11, 15 and 20 times washed samples for both groups was analyzed by GC. Data obtained were analyzed with two-way ANOVA using the washing procedure as fixed factor with 2 levels (laboratory hand simulation and ISO 6330∶2000 machine washing) and the number of washes as fixed factor with 5 levels.

The effect of 2 drying methods was compared on all types of nets washed according to method-2.

To assess the effect of the drying process on the release of the active ingredient from the nets, two groups of samples were considered from each type of LN. Each group contained 10 samples of about 70 cm×70 cm washed following the ISO 6330∶2000 machine washing procedure up to 20 times. After each wash, one group of samples was dried during 24 hours indoors as shown in Figure 1. The other group was dried during 24 hours with true-light lamps as shown in Figure 2.

To check whether the drying process had a significant impact on the active ingredient content of LNs, the samples from each group were subjected to GC analysis.

The software SPSS version 20 package was used to run all the statistic tests. Statistical analysis was performed using Student’s t-test and ANOVA. Before performing ANOVA and using obtained data, assumptions (homogeneity of variance, normality of observations) underlying the analysis of variance were tested. Comparisons were made between the number of washes, the type of washing process, the type of drying methods and the type of impregnation technology. To find the best relationship between the release of active ingredient from the LN and the wash numbers the curve estimation program was used.

Table 1 shows the average of active ingredient content in LN after different washing cycles. Many studies confirmed that the washing process affects significantly the release of active ingredient content from the nets [12], [34], [35]. Here, independently to the drying process, it was checked whether the laboratory hand washing simulation using the CIPAC washing agent and the ISO 6330∶2000 machine washing as described above affected significantly the total active ingredient content on the different nets. Data obtained are shown in Table 2.

The Table 2 shows that there is a very significant difference in the concentration of the active ingredient between the washing cycles of the coated nets (Interceptor® and PermaNet®2.0) (0.001<p<0.01). This difference is very highly significant for the incorporated nets (Netprotect®) (p<0.001). The type of washing and the number of washing cycles both have effect on the release of the active ingredient content of the net.

Chemical analyses showed that the initial active ingredient concentration from Interceptor® and PermaNet®2.0 nets were respectively 4.72 g/kg (189 mg/m²) and 2.00 g/kg (60 mg/m²). These concentrations were slightly different compared to the target dose of 5 g/kg (200 mg/m²) for Interceptor® nets and 1.83 g/kg (55 mg/m²) for PermaNet®2.0 nets but within the specifications of ±25% [36]. The analysis of Netprotect® nets showed 2 peaks for deltamethrin with a proportion of about 26/74. The first eluting peak was the R-alpha isomer of deltamethrin, a non relevant impurity of deltamethrin. The initial deltamethrin concentration of Netprotect® was 1.12 g/kg.

The Table 1 shows that the retention of ≥50% of the active ingredient after 5 standard washes according to each type of net is met [34]. That would indicate a retention index of ≥90% for reservoir behavior or ≥87% for free migration behavior. This was confirmed in Table 3, where the active ingredient retention at each wash was calculated.

From Figure 3, it can be noticed that the percentage of alpha-cypermethrin remaining on Interceptor® LN after 20 washes cycles was 21.9% (1.03 g a.i./kg, n = 3) for the laboratory hand washing simulation (indoor drying), 32.9% (1.55 g a.i./kg, n = 3) for the ISO 6330∶2000 machine washing (indoor drying) and 31.5% (1.49 g a.i./kg, n = 3) for the ISO 6330∶2000 machine washing (UV drying).

The overall percentage of alpha-cypermethrin remaining on Interceptor® LN after 20 wash cycles was 28.7% (95% CI [22.7, 34.8]).

From the Figure 3, it can be noticed that for the PermaNet®2.0 LN, deltamethrin remaining on the net was 25.5% (0.51 g a.i./kg, n = 3) for the laboratory hand washing simulation (indoor drying), 26.3% (0.53 g a.i./kg, n = 3) for the ISO 6330∶2000 machine washing (indoor drying) and 28.9% (0.58 g a.i./kg, n = 3) for the ISO 6330∶2000 machine washing (UV drying).

The overall percentage of deltamethrin remaining on PermaNet®2.0 LN after 20 washes was 26.9% (95% CI [24.7, 29.1]). These data confirmed those of the study of Atieli [19] which showed that PernaNet®2.0 retained about 18 to 27% of the deltamethrin after 20 washes. Using standardized WHO protocol washing [26], the study of V. Corbel [37] showed also that PermaNet®2.0 loses between 60 to 85% of deltamethrin after 20 washes.

From the Figure 3, it can be noticed that for the Netprotect® LN, the percentage of deltamethrin remaining in the net after 20 washes cycles was 63.9% (0.72 g a.i./kg, n = 3) for the laboratory hand washing simulation (indoor drying), 69.8% (0.78 g a.i./kg, n = 3) for the ISO 6330∶2000 machine washing (indoor drying) and 70.5% (0.79 g a.i./kg, n = 3) for the ISO 6330∶2000 machine washing (UV drying).

The overall percentage of deltamethrin remaining in Netprotect® LN after 20 washes was 68.0% (95% CI [62.5, 73.6]). This was also shown by Skovman et al. [38] (77% of deltamethrin remained on Netprotect® after 20 washes) and in the 11th WHOPES report [34] (77.6% remains). The high percentage of deltamethrin retention for Netprotect® is due to the incorporation technology compared to the coated one. This may also mean that the active ingredient from the Netprotect® net is more inaccessible to aqueous surfactant and is also unlikely to be accessible for mosquitoes walking on the surface [34]. So, from our observation and based on this fact, Netprotect® net might perform less against mosquitoes compared to PermaNet®2.0 and Interceptor®. This assertion has to be confirmed by bioassay.

As this study did not involve bioassay tests, the data were compared with data from the literature on active ingredient content on the net (mg/m²) with the corresponding mortality percentage according to the number of washes. To assess the efficacy of mosquito nets, bioassays are recommended by WHOPES, even if results in different parts of the world are varying. A net maintains its efficacy if it produces more than 80% mosquitoes mortality in a bioassay cone test after 20 washes [26]. The variation was in general due to the formulation of the insecticide, the type of insecticide, the mosquito species, the susceptibility level of mosquitoes, the time of exposure, the texture of bed net, the quality of different batches and the type of tests. Some studies revealed that there was no significant difference in the mortality of species exposed to different types of LNs washed by machine or by hand [10], [16], [25], [39]. Table 4 compares information about the efficacy of the nets submitted to the 3 wash cycles used during this study. It shows that the active ingredient content is close to values published in the literature studies until 20 washes for all the 3 types of nets tested. Indeed, the PermaNet®2.0 nets after 20 washes should provide 70% mortality [19] with around 15 mg a.i/m², while the Netprotect® and Interceptor® nets should provide respectively a mortality of 30% [19] with 30 mg a.i/m² and 15% [10] with around 60 mg a.i/m². Thus, even if the high percent of retention seems to be less favorable for Netprotect® nets in term of availability of the insecticide over the net, the Table 4 infirm this trend.

Table 5 summarizes the F test of model fit and the estimated parameters of the models. The significant value of the F statistics on each brand of LN is less than 0.05 for the 4 models (linear, logarithmic, quadratic and exponential). This means that the variation explained by each model was not due to the chance. R, the multiple correlation coefficient, is the linear correlation between the observed and model-predicted values of the insecticide content. Its large value indicates a strong relationship. That was also seen by the R-Square. R-Square statistic, a measure of the strength of association between the observed and model-predicted values of the dependent variable (concentration of active ingredient of the nets), was large in general for all the nets, except for the modeling of the Netprotect® nets. Indeed for Netprotect®, the washing cycle’s data did not properly fit any curve model. For the others types of net the large R-Square values indicated strong relationships for exponential, logarithmic and quadratic models. A comparison between the 3 cycles showed that the R-Square for the logarithmic model were larger for PermaNet®2.0 nets while, for the Interceptor® nets it appeared that the exponential model better follows the trend of the data (Figure 4 and 5). This is confirmed by the findings of A. Hill [34].

A literature search was done to find out some data coming from Interceptor®, PermaNet®2.0 and Netprotect® nets washed following the laboratory hand washing simulation and using the previous detergent “Savon de Marseille” which had to be replaced. Table 6 presents a comparison between literature data and data obtained in this study. Both data show that the percentage of the insecticide decreases after washing, and in function of the number of washings. When the results obtained in this study are compared to those of the literature, small to big differences are found. Explanation for this can be given by the fact that firstly only a few data in literature are available and, secondly the baseline concentration of the nets, the net materials and the soap used might have been different. Because of this one can not say that washing is the only determining factor explaining the differences between the literature and the observed data.

Coming back to the current study, the retention/release and the concentration of the active ingredient of LNs washed following the laboratory hand washing simulation (WHOPES standard washing procedure with the proposed CIPAC washing agent) were compared with the LNs washed with the domestic washing procedure (ISO 6330∶2000 machine washing).

The Figures 6–8 show the graphs of the active ingredient content from the LNs in relation with the number of washes for both washing procedures. Figure 6 and 7 show no difference of the effect of both washing procedures on active ingredient content of Interceptor® and PermaNet®2.0 nets. This was not the case for the Netprotect® nets (Figure 8). For these nets, the concentration of the active ingredient after washing with the laboratory hand simulation procedure was lower than after washing with the ISO 6330∶2000 machine procedure. All these observations were confirmed by statistics shown in Table 7.

So the influence of the washing procedure depended on the type of LNs. As the P-value was more than 0.05 for the Interceptor® and PermaNet®2.0 nets, the laboratory hand washing simulation and the ISO 6330∶2000 machine washing had statistically the same effect on removing active ingredient form those nets. For Netprotect® the P-value obtained from the statistical test (0.001<P<0.01) showed a highly significant difference between the washing procedures. The laboratory hand simulation washing removed more active ingredient from the Netprotect® nets than the ISO 6330∶2000 machine washing. That shows that insecticides coated on the net seem to be removed easily with a soft washing, while this was not the case for the insecticides incorporated into the nets. This confirmed the fact that washing is considered to be a more important loss mechanism for coated net than incorporated nets [40].

The ISO 6330∶2000 machine washing procedure that is considered to be the more stringent [27] (European standard for domestic washing procedure for textiles testing), seems to have less impact compared to the laboratory washing procedure which is considered as a hand simulation washing procedure [41]. So applying the ISO 6330∶2000 machine washing with the IEC A* detergent to the nets underestimates the loss of active ingredient due to washes. At the same time it seems better to wash LNs using this procedure compared to washing by hand. The explanation of this difference might be based on the soap used (chemical action) and/or the shaking process (mechanical action) applied in both methods. The chemical impact might be due to the emulsification actions of surfactants while the mechanical action is caused by the textiles impacting and rubbing against one another [42]. Mechanical action plays an important role in washing process [43], it is seen to be responsible for textile wear; so, it might be also the wear cause of removal insecticide from the net. The washing machine involves important mechanical forces: the normal/impact force and the abrasion/friction force (Figure 9). The abrasion force is related to the drum rotation speed during the main washing. Considering the impact force, it is likely to be higher for the laboratory hand washing simulation than the ISO 6330∶2000 machine washing procedure. This might be why the laboratory hand washing simulation using horizontal shaking affected more the release of the active ingredient from net.

Even if it was found here that the laboratory hand washing simulation removes more insecticide from the nets, it still underestimating the real impact of the washing by hand [19]. Some studies [41], [44] show that the dose of the new CIPAC detergent should be 5.0 g/L in order to reach the expected effect of hand washing. The same studies show that 4.0 g/L CIPAC agent was close to 2.0 g/L Marseille soap in the same washing conditions.

Data obtained from the GC analysis were clustered in graphs (Figure 10, Figure 11, Figure 12) and statistically analyzed with two-way ANOVA using drying processes as fixed factor with 2 levels (indoor and UV drying) and number of washes as fixed factor with 5 levels.