A phase III trial to evaluate the efficacy, fabric integrity and community acceptance of Netprotect® using a recommended long-lasting insecticidal net as positive control

Present study was conducted between 2009 and 2013 in Veal Veng district, a rural forested area in the province of Pursat, in Cambodia at the border with Thailand. In this area, 1.5% malaria prevalence is detected based on Rapid Diagnostic Tests/microscopy (Cambodia Malaria Survey 2010) [8]. The main malaria vectors are Anopheles dirus sensu stricto, Anopheles minimus s.s., Anopheles maculatus and Anopheles barbirostris. All vectors are fully susceptible to pyrethroids [9].

In Cambodia, the overall proportion of people sleeping under an ITN/ LLIN “the previous night” increased the last years to more than 50%. About 63% of households owned sufficient amount of mosquito nets but only 23% of the 3,164 households (HHs) interviewed during the Cambodia Malaria Survey (CMS) 2010 [8] owned sufficient LLINs. This net coverage contributed to the decrease of the overall malaria prevalence in Cambodia from 4.4% in 2004 [10] and 2.6% in 2007 [11] to 0.9% in 2010.

A phase III field trial, designed as a prospective longitudinal study, was set up to study the performance of Netprotect® using conventionally-treated nets (CTN) as a positive control. The study protocol followed the WHO guidelines for phase III LLIN field trials [7] except that a reference LLIN (PermaNet® 2.0) was used as an additional positive control. The unit of observation was the household (HH), in which one net was assigned for follow up. CTNs and their use were monitored for one year and the LLINs (Netprotect®, PermaNet® 2.0) for three years.

The LLIN Netprotect®, manufactured by Bestnet Europe LTD, is a LLIN made of polyethylene with deltamethrin incorporated into the 118-denier monofilament with a target dose of 1.8 g active ingredient (a.i.)/kg, corresponding to 68 mg a.i./m² (weight 35 g/m²). The nets distributed had a standard size of 160 cm W × 180 cm L × 150 cm H. The PermaNet® 2.0 was bought from Vestergaard Frandsen. This is a WHOPES fully recommended LLIN [12] made of knitted poly-filament polyester fibres coated with deltamethrin with a target dose of 1.4 g a.i./kg for a 100 denier yarn, or 55 mg a.i./m². The PermaNet® 2.0 nets had the same colour (blue) and size as the Netprotect® nets. The conventionally-treated nets (CTNs) were blue but slightly wider (190 cm W × 180 cm L × 150 cm H) than both LLINs with a fabric weight of 31.1 g/m². The deltamethrin treatment (Deltamethrin EC 10 or 10,0 g a.i./L) of these nets, as per the standard WHO instructions [13], was carried out by two trained dippers provided with adequate protective gear, at a target dose of 25 mg a.i./m².

Prior to the net distribution, a HH census was conducted in the study area to obtain the demographic and socio-economic characteristics of each HH. The HH census data were entered in an Access database to form a master list. From this master list, all HHs in the selected villages were randomly allocated to one of the three study arms in a 2:2:1 ratio, for respectively Netprotect®, PermaNet® 2.0 and CTN, and this in each village. For HHs assigned to Netprotect® or PermaNet® 2.0 nets, enough nets of the same type were distributed to assure full coverage (i.e. one net per two inhabitants), but only one of these nets was assigned as net to follow up. In HHs assigned to CTN, only one CTN was distributed, and additional Netprotect® nets were distributed to cover all residents. In December 2009, a total of 148 HHs were provided with a CTN, 305 HHs with Netprotect® nets and 309 HHs with PermaNet® 2.0 nets.

All distributed nets received a unique Net ID number, written with wash-resistant ink on the factory label. The net with the highest code number within each HH was retained for follow up.

For every sampling round a HH randomization was made to sample 30 nets of each study arm, evenly distributed over the villages. As people were not always present during the survey or nets were lost, 60 HHs per study arm were randomized. The HHs were visited according to the randomization order, until the number of samples were achieved per village and type of net brand. Nets selected and withdrawn were replaced with a new LLIN without ID number. HHs where people were absent (for more than three consecutive days) were maintained in the master list for the next randomization, while the HHs where the net was sampled were removed from the master list. In the first year of the study the HHs that recorded loss of their net for follow up were also removed from the master list. However, because of the high number of nets lost for follow up, it was decided after one year that only the HHs that were sampled or moved away would be removed from the master list. Nets randomized for the second time and still lost for follow-up, were recorded as lost and removed from the master list. From that point on though, the second highest net ID in that specific HH would become the new net to follow up for the upcoming sampling surveys.

During the surveys, HH owners or net users were interviewed by trained staff about their net use (method and frequency), washing habits (frequency and method of washing and drying) and the occurrence of adverse events. The net to follow up was sampled for submission to physical integrity tests and cone bioassay tests. Chemical analysis were done only on the nets sampled 1 week, 6 months, 12 months and 36 months after net distribution. At the end of the first study year, besides the normal sampling of the three net types, the remaining CTNs were collected and replaced with a Netprotect® or PermaNet® 2.0.

In the annual surveys (at the end of months 12, 24 and 36) in addition to the randomized HHs, all households remaining in the master list were visited door-to-door. Whereby the physical presence/absence and the information on people’s net perceptions and practices were recorded, to help estimate the annual attrition rate.

All nets sampled during the surveys were examined for their physical integrity. The integrity of the nets was determined counting the number of tears and holes as described in [14]. Hole sizes were categorized in three groups; holes allowing a thumb to pass through, holes between a thumb and a closed fist and holes bigger than a closed fist. A proportionate hole index (pHI) [15, 16] was calculated by making the sum of the holes weighted by size for each net. For these groups, the weights used to calculate the pHI were 1, 9 and 56 as described in [15, 16]. To better translate the hole index to an integrity status (net condition) for each sampled net, the pHI is categorized into good (%25), fair (25–174), mediocre (175–299) and poor (>299) net condition [16]. The dirtiness of the nets was determined by comparing the nets to reference samples (clean, bit dirty, dirty, very dirty and never used).

The CTN and LLIN functional survivorship rates were calculated from the records made during the surveys at the end of each study year. The functional net survivorship accounts for loss due to damage only and is not influenced by nets that are removed from the study area (stolen, given away or family moved) as they still may be functional [20].

All data were recorded on standard forms. The hard copy data was single entered into the main Access database. Data entry was checked by cross checking all entered records of adverse events, physical integrity and bioassays with the hard copy files. All data were imported into R, version R-3.0.2 [21], for statistical analysis.

Descriptive analyses were done on the demographic household census data, to define the household characteristics. Data collected on net use and net washing habits were analysed using a chi-square test (χ²) for significance of deviation between the net types.

For the analysis of net performance (physical condition, chemical content, bioassay) all sampled nets were used except nets that were still in the original plastic bag (i.e. never used) and nets that were completely torn and thrown away.

For each net chemical content and bioassay performance were calculated from all samples apart from position 1, as the lowest part of the net may be exposed to excessive abrasion in routine use. For chemical analysis an inter-net variation was expressed by relative standard deviation (RSD). For the bioassays, each of the four samples were tested individually and an assessment of between-net variability was done with a RSD. The within-net variability was expressed as the difference of samples of the same net to the mean expressed as percent of the mean and then averaged over the sample [15].

A Chi-Square test over multiple strata was conducted (Mantel Extension Test [22] on the acceptance of the nets (proportions of LLINs that fulfil the WHO criteria (≥95% knockdown or ≥80% mortality).

A generalized linear mixed model (GLMM) was fitted for chemical content, bio-efficacy and net integrity as response variable. As fixed effect explanatory variables the net type, survey (time point), net condition, dirtiness of the net, net use intensity, number of washes, drying and washing method and socioeconomic status (SES) were included. The net ID was included as a random effect. Per outcome the model was optimized by removing the insignificant variables using a likelihood ratio test [23] or by comparing the AIC (Akaike information criterion) values [24].

Ethical approval for the study was obtained from the Ethics Review Committee of WHO for two arms (CTN and Netprotect®) and from the Ethical Committee of Cambodia, the Institutional Review Board of the Institute of Tropical Medicine and the legal ethical committee of the University of Antwerp for all three arms (CTN, Netprotect® and PermaNet® 2.0).

An informed consent was signed by impregnators and by HH owners. All participants were informed about the benefit of using an insecticide impregnated net and the additional advantages of using a LLIN. The information sheet addressed also the possible adverse effects and their remediation for all users.

The random allocation of each HHs to one of the net types was in all 7 villages according to a 2:2:1 ratio, for respectively Netprotect® (305 HH), PermaNet® 2.0 (309 HH) and CTN (148 HH). All HHs together comprised of 3,832 individuals of which 45% were younger than 17 years and 26% were between 17 and 30 years old. The average age of the head of the HHs is 42 years, 74% of them had no education, 23% a primary education, and 87% of head of HHs are farmers. 82% of the houses were built on stilts compared to 18% built on the ground, the walls are made of wood (78%) or thatch (21%) and the roofs are covered with sheets of corrugated iron (61%) or thatch (32%) (Additional file 1).

During 8 field surveys spread over the three study years, 1435 HHs were visited. In the first two years of the study, more than 80% of the HHs declared to use the net year-round and every night (Additional file 2). The remaining 20% of the HHs only used their net seasonally, did not use them or did not provide an answer. No significant differences in net use intensity were found after one year between the three study arms (χ² test: P = 0.173, df = 12). After three years the net use intensity dropped to about 60% for both Netprotect® and PermaNet® 2.0. Moreover, 34% of Netprotect® and 20% of PermaNet® 2.0 were recorded as not used anymore and stored away. Those net users replaced these nets with up-to-that-point unused spare nets. During the first year 30/601 nets of all three net types were kept as spare nets. This occurred because of the over saturation of nets or because most families slept with more than two individuals under one net. Similarly no significant differences in net use intensity were found between the two LLINs after three years (χ² test: P = 0.213, df = 3).

Out of 522 HHs in which the net to follow up was present at the end of each year, 250 (48%) HHs recorded to have washed the net (Additional file 3). For net that have been washed, the average number of washes per year reported by the HH for Netprotect and Permanet 2.0 was 3.4 and 3.5 respectively for year one; 1.7 and 2.1 for year two, and 1.3 and 2.4 for year three. The accuracy on the reported washing behaviour is however questionable as the reported washes declined for the last survey. Irrespective of the net type, almost all nets were washed with cold water (98.8%) and with a local detergent or soap (96%). In 8% of cases the nets were rubbed against rocks. A total of 65% of the washed nets were dried outside in the sun, 23% were dried outside in the shade and 10% were dried inside.

Both impregnators declared to experience a bad smell during impregnation, no other adverse events were recorded during or after impregnating the nets. Net users from all study arms declared some adverse events during the first 6 months of use, while from 12 months post net distribution and onwards no more adverse events were declared. The most frequently declared adverse events were bad smell, itching skin and eye irritation. Analysing the declared adverse events among the three study arms within each survey, no significant differences were found between the three net type users.

The functional survivorship stayed above 90% for both LLINs in the first two years but after three years this dropped to about 60% for both Netprotect® (58.1%) and PermaNet® 2.0 (61.2%) (Table 1).

During the three years of follow up, a total of 507 nets were sampled and all tested for their physical integrity. However 42 nets were excluded from analysis, of which 33 had never been used and nine had been thrown away because being completely torn (one CTN, three Netprotect® and five PermaNet® 2.0).

The proportion of nets with at least one hole increased rapidly within the first year up to 89%, 50% and 63% of respectively CTN, Netprotect® and PermaNet® 2.0. The increase continued up to two years of use for both LLINs to 86%. Finally after three years 83% of Netprotect® and 93% of PermaNet® 2.0 had at least one hole. Similarly the proportionate hole index (pHI) increased rapidly over time, but important variation exists between nets for respectively all net types. Consequently the median and interquartile range ([IQR] 0.25–0.75) were used to present the pHI (Figure 1 and Additional file 4). After one year of use the median pHI was 59 (IQR 8–137) for CTN, 0.5 (IQR 0–3) for Netprotect® and 3 (IQR 0–27) for PermaNet® 2.0. After three years, the pHI increased to 68 (IQR 14–186) and 127 (IQR 41–220) for respectively Netprotect® and PermaNet® 2.0, of which 25% of Netprotect® nets. By the end of the study, the percentage of nets categorized as in mediocre or poor condition (pHI ≥ 175) were 25% and 30% for Netprotect® and PermaNet® 2.0, respectively. Over the entire three-year follow-up no significant difference was found between both LLIN’s physical durability (Odds ratio [OR] = 0.82, 95% CI 0.49–1.38 - GLMM, P = 0.45). Net integrity of both Netprotect® and PermaNet® 2.0 were affected by the dirtiness, with significantly more holes recorded in dirty and very dirty nets (OR = 39, 95% CI 17–92 - GLMM, P % 0.001 and OR = 122, 95% CI 47–315 - GLMM, P % 0.001, for Netprotect® and PermaNet® 2.0 respectively).

The average active deltamethrin content of each of the three net types measured one week post distribution was within the given target dose, though for CTN and Netprotect® only marginally (Table 2A). Only 14% of the sampled CTNs were actually within its chemical target dose and this with a high between-net variation (RSD: 78%) showing a low homogeneity of treatment (Figure 2 and Table 2A). Regarding the LLINs, 57% of Netprotect® nets (RSD: 10%) and 69% of PermaNet® 2.0 nets (RSD: 15%) were within the target dose. The remaining 43% of Netprotect® nets were below the threshold (Figure 2). For PermaNet® 2.0 nets, 4% were below the tolerance limit and 27% were above the maximum allowed deltamethrin content. After six months of use, all sampled Netprotect® nets dropped below the initial chemical target dose while 48% and 33% of PermaNet® 2.0 nets sampled, respectively after one and three years, were still within the limits of the initial target dose. Despite this difference, and taking into account the insecticide content after one week as baseline, the deltamethrin retention at 6, and 12 months was similar for the two LLINs (81% and 62% retention for Netprotect®; 86%, 65% for PermaNet® 2.0). Though a difference was noted after three years, 36% retention for Netprotect® and 49% for PermaNet® 2.0. The CTNs sampled after 12 months of use retained only 37% of its original content.

Additionally to the content of active S-alpha isomer (deltamethrin), the content of the deltamethrin R-alpha inactive isomer was measured per survey and net type. In the baseline CTN samples 3% of the deltamethrin content counted for the R-alpha isomer. This amount remained below 3% in the samples after six months but did increase to 20% in the samples collected after one year (Table 2B). For Netprotect® the deltamethrin R-alpha isomer content in the baseline samples was 33% of the deltamethrin content; this high percentage even increased in the samples collected after 6, 12 and 36 months up to respectively 51%, 53% and 69% of the deltamethrin content. For PermaNet® 2.0 the deltamethrin R-alpha isomer content in the baseline samples was 7%; this amount remained low in the samples collected after 6, 12 and 36 months, respectively 3%, 6% and 3% of the deltamethrin content.

Results from WHO cone bioassays are presented in Table 3. Mortality in the negative control never exceeded 4%, this occurred only in two bioassay tests and in another eight tests control mortality reached 2% while no mosquito mortality was recorded in all other 451 bioassay tests. The CTNs were fully effective after six months of use, but both the mosquito knockdown and mortality rates of CTNs dropped sharply after one year of use to 56% and 64% respectively. For both LLINs the average knockdown and mortality were close to 100% up to six months of use. After 12 months a small drop of about 10% for both knockdown and mortality rates was recorded. Despite this small drop and the constant loss of deltamethrin over time, these average rates were maintained up to three years with an average mortality and knockdown around or above 80%. Both LLINs performed significantly better than the CTNs after one year of use (GLMM mosquito mortality, P % 0.001) while after three years both LLIN average mortality rates did not differ (GLMM, P = 0.17). However when mortality rates over the entire three years are compared, PermaNet® 2.0 does obtain significantly higher mosquito mortality than Netprotect® (GLMM, P = 0.023). For each net type a multivariate analysis was carried out (GLMM) to assess the relative role of net use intensity, washing, net condition, dirtiness, chemical content and time of observation on the bio-efficacy of each net type. Time of observation showed a significant association with the bio-efficacy of all three net types (GLMM, P = 0.01, P % 0.001 and P = 0.006 respectively for CTN, Netprotect® and PermaNet® 2.0). CTN bio-efficacy was negatively correlated with bad net condition (GLMM, P = 0.03) and its chemical content (GLMM, P = 0.004), while for both LLINs net use intensity, number of washes, washing methods and the chemical content was not correlated to their corresponding bio-efficacy. For both Netprotect® and PermaNet® 2.0, the between- and within-net variations of knockdown and mortality follow similar trends (Table 3).

The seven Netprotect® and four PermaNet® 2.0 nets that failed the bioassay criteria after 36 months were subjected to the tunnel test. All tunnel tests were valid, with an mortality in the control of less than 6% and more than 76% were blood fed. None of the tested nets passed the tunnel test criteria of 80% mortality or 90% blood feeding inhibition (see Additional file 5).

The proportions of LLINs that fulfil the WHO criteria (≥95% knockdown or ≥80% mortality) are reported in Table 4. Only 42% of the CTNs complied with these criteria after one year of use, while 73% and 85% of Netprotect® and PermaNet® 2.0 did. The proportion of Netprotect® nets that met the efficacy criteria varied throughout the study and was below the WHOPES threshold of 80% after 12, 24 and 36 months (respectively 73, 72 and 71%), however the 80% value was still within the confidence intervals. Whereas PermaNet® 2.0 performance was consistent over the entire three study years with a success rate of more than 80%. The acceptance of nets is entirely based on nets complying with the mortality criteria, as there are no nets that pass the 95% knockdown rate and then fail the 80% mortality rate. The acceptance over the period of one to three years was significantly higher for the reference LLIN than for Netprotect® (Mantel-Haenszel Test: OR = 2.091 [95% CI 1.071 – 4.083]; p = 0.0044).