Study area and facilities
Large cage semi-field evaluations were conducted at the Ifakara Health Institute’s (IHI) experimental station, in Kining’ina village (8.11° S, 36.67° E), approximately 6 km north of Ifakara town, in Kilombero district, south-eastern Tanzania. Experiments were conducted inside two different types of large screened cages. The first was a large multi-compartment system covering 553 m2 in ground area, and 4.5 m in height, (Fig. 1a) [49–51]. The experiments were conducted in two compartments inside this large multi-compartment system each covering 36 m2 in ground area, and 4.5 m in height. The second cage was a long, tunnel shaped screened system measuring 110 m long × 2 m wide × 2.5 m high, (Figs. 1b & 1c) [37].
Field experiments were conducted in villages in Kilombero and Ulanga districts, in south-Eastern Tanzania. These study areas experience a wet season between March and June and a dry season in August to October, with a mean annual rainfall of 1200 – 1400 mm and a daily temperature range of 20-32° C [52]. Malaria is endemic and current transmission is primarily mediated by Anopheles funestus sensu stricto (s.s.) and Anopheles arabiensis. The main malaria control intervention used in the study area is LLINs. Pyrethroid resistance is prevalent in both An. arabiensis [53], and An. funestus [54].
Mosquitoes
For the semi-field evaluations, laboratory-reared, pyrethroid susceptible An. arabiensis (Ifakara strain) were used in the semi-field experiments. Larvae were fed on Tetramin® fish food and maintained at temperatures of 28 - 29°C. Pupae were placed in a separate room in emergence bowls inside a 30 ´ 30 ´ 30 cm netting cage and a 10% glucose solution provided for the emergent adults. Temperatures were maintained at 27 ±3°C and relative humidity at 70-90%. The insectary was maintained at a 12:12 (light: dark) photoperiod. The mosquitoes used in experiments were 4-9 days old nulliparous females. The mosquitoes were starved for six hours before each experiment.
Volunteers
Adult male volunteers (18 to 40 years old) participated in the experiments. The volunteers were trained on objectives, benefits and potential risks of the study and recruited only if they provided written informed consent. All volunteers were highly experienced in the procedure of human landing catch, wherein mosquitoes attempting to bite a volunteer’s legs are captured immediately upon landing [55]. The volunteers were instructed not to use any fragranced soap or perfume, tobacco or alcohol throughout the experiment period.
Transfluthrin-treated footwear prototype
Locally manufactured leather sandals were fitted with transfluthrin-treated hessian fabric affixed onto the straps of the sandals. The bands were fitted on the top leather surface of the sandal straps, so that there was no direct contact with the volunteer’s skin (Fig. 2). Hessian fabric was used because it is readily available in Tanzania and has an optimal adsorbent capacity [37]. Transfluthrin, a pyrethroid insecticide recommended by the World Health Organization (WHO) for control of flying insects, such as mosquitoes and flies [56], was selected as spatial repellent. Transfluthrin is a highly volatile pyrethroid with a vapour pressure of 9 x 10-4 Pa at 20°C [56]. This property makes it suitable for use in tropical regions and at low-cost because no additional heating is required to evaporate the chemical [57]. In addition, previous studies carried out at Ifakara Health Institute have demonstrated efficacy of transfluthrin-impregnated hessian fabric against mosquito bites in semi-field and field experiments [37–40] further affirming the potential of transfluthrin-treated hessian substrates in malaria control.
The technical grade transfluthrin used to treat the hessian fabric was donated by SC Johnson (Racine, Wisconsin, USA). Identical sandals treated with axion detergent and water were used as negative controls.
Experiment 1: Semi-field experiments to determine a dose-response relationship for transfluthrin in terms of protection against mosquito bites
Three different amounts of 97% technical grade transfluthrin were used to treat the hessian bands attached to the sandals in the first set of experiments; 0.06g, 0.10g and 0.15g transfluthrin was mixed with 94ml, 90ml and 64ml of AxionÒ liquid detergent (Orbit Chemical Industries Ltd, Kenya), respectively, to enable solubility in water [37]. This emulsion was then mixed with 100 ml of water for both 0.06g and 0.10g transfluthrin and 74 ml of water for 0.15g transfluthrin. These amounts were used to ensure there was no left-over emulsion which might undermine the treatment dose.
Rectangular hessian fabric bands measuring 48 cm2 were dipped in the resulting emulsion containing either of the three amounts of transfluthrin and soaked until complete saturation and all the emulsion had been absorbed after which they were suspended at ambient temperatures indoors, protected from direct sun exposure, and allowed to dry for 24 hours. Three pairs of sandals were then fitted with the treated hessian fabric bands, each containing either of the three amounts of transfluthrin (Fig. 2b). A pair of matching negative controls for each treatment were prepared in the same way, except the treatments were done with only AxionÒ liquid detergent and water without any transfluthrin.
Two volunteers were asked to wear knee-length shorts to standardize the area of the lower limbs exposed. They sat on low chairs in experimental compartments measuring 6 x 6 m inside the semi-field facility (Fig. 1a). One compartment of the semi-field system was used for testing the treatment while the other tested the control sandal. The experiments were conducted one at a time starting with the sandal fitted with the hessian fabric that had the lowest amount of transfluthrin. Each night, 200 female An. arabiensis mosquitoes were released at the centre of each experimental compartment at 1800 hrs, with the volunteer sitting approximately 10 m from each other. The volunteers collected all mosquitoes attempting to bite the exposed lower limbs for 45 minutes of each hour and rested for 15 minutes. Mosquito collections were done from 1800 hrs to 0600 hrs the next morning. Each volunteer was given a head torch and siphon for aspirating the mosquitoes. The mosquitoes were kept in separate paper cups for each hour of collection. At the end of each night of experiment, the recaptured mosquitoes were killed using petroleum ether, counted and recorded for each hour.
The experiments were conducted in a binary cross-over design, where each amount of transfluthrin (sandal pair) was tested for 8 consecutive nights against the control. Only one sandal pair, containing either of the three treatment amounts was tested per night. On the first night of each experiment testing the different transfluthrin doses, the treatment and control sandals were randomized to the experimental compartments. Volunteers were then rotated each following night between the two experimental compartments of the semi-field system. The experimental compartments were separated by a similar-sized chamber in between that acted as a buffer in the event of spillover effects of the treatment. The treatment and control sandals remained in the same compartment throughout the experiments to minimize the potential impact of residual effect of transfluthrin.
Experiment 2: Semi-field experiments to evaluate the of impact of hessian fabric surface areas on mosquito bites
Another set of experiments was conducted to assess the impact of surface area of treated hessian fabric on mosquito bites. Hessian fabric pieces measuring 48 cm2 and 240 cm2 were each treated with 0.10g and 0.15g of transfluthrin using the same methodology described above, affixed onto sandals and their efficacies against mosquito bites evaluated.
The same binary cross-over design as experiment one above was used, and the experiments were replicated 8 times (nights) for each treatment.
Experiment 3: Experimental field evaluation of the efficacy of transfluthrin-treated sandals against bites from wild mosquitoes
A binary crossover design, similar to the semi-field experiments was used in the two study villages. The 48 cm2 hessian fabric pieces treated with 0.15g transfluthrin were evaluated in these experiments as they demonstrated significant efficacy against mosquitoes. The field tests were conducted by two volunteers who carried out the experiments over 12 nights in each study village. The tests were conducted outdoors, next to rice fields and away from human dwellings, and these sites, each with two positions, were fixed throughout the experiments. The experiments were conducted on separate nights in the two villages.
Other mosquito genera caught were stored in batches of five per micro-centrifuge tube and later identified by experienced entomologists [60].
Experiment 4: Assessment of whether transfluthrin-treated sandals divert host-seeking mosquitoes to persons not wearing the sandals
This experiment was carried out to determine whether the use of transfluthrin-treated sandals would put nearby non-users at a greater risk of being bitten by diverting mosquitoes to non-users. In this experiment, a volunteer wearing transfluthrin-treated sandals sat at one end of an experimental compartment while another volunteer wearing untreated sandals sat at the other end of the same compartment, approximately 10 metres away. This was the treatment compartment. Four hundred laboratory-reared female An. arabiensis mosquitoes were released from a small cage midway between the two volunteers. The number of mosquitoes attempting to bite the volunteers were caught, throughout the night. Another pair of volunteers sitting in a comparative compartment (control compartment), replicated this experiment using untreated sandals. To establish diversion, the number of mosquitoes caught by the volunteer wearing the untreated sandals in the treatment compartment were compared to the number of the mosquitoes caught by the volunteers in the comparative compartment to determine if the volunteer who sat next to transfluthrin-treated sandals caught more mosquitoes compared to either volunteer in the comparative compartment with untreated sandals only. Similar to the semi-field experiments above, the sandals remained in the same position in each respective chamber and only the volunteers rotated between positions to control for individual attractiveness. The limitations of this design have been outlined in the limitations section of the manuscript.
Experiment 5: Semi-field experiments to assess the efficacy of transfluthrin-treated sandals on laboratory-reared Anopheles and Aedes mosquitoes
Eight different sandal prototypes (Fig. 3), treated with 0.05g of transfluthrin were later developed and tested in the semi field system against the following laboratory reared mosquito species; A (male design 1) - An. arabiensis and Aedes aegypti; B (male design 2) - An. arabiensis; C (male design 3) - An. arabiensis and Ae. aegypti; D (male design 4) - An. arabiensis; E (female design 1) – An. gambiae s.s and Ae. aegypti; F (female design 2) - An. arabiensis and Ae. aegypti; G (female design 3) - An. arabiensis and Ae. aegypti and H (female design 4) - An. arabiensis. All sandals developed could not be tested against all mosquito species because of logistical and cost implications. Different prototype sandals were therefore tested against randomly selected mosquito species. This was done to select the sandal prototype that offered the best protection against the respective randomly selected mosquito species. The results from these experiments were later pooled to evaluate the impact of transfluthrin-treated sandals on different mosquito species.
The surface areas of the hessian fabric used for the different sandal designs were as follows; A – 395 cm2; B – 327 cm2; C – 330 cm2; D - ≈400 cm2; E – 346 cm2; F – 780 cm2; G – 640 cm2, and H – 325 cm2.
All the designs developed were treated with 0.05g of 97% technical grade transfluthrin (Shenzhen Sunrising Industry Company, China) following risk assessments of transfluthrin-treated sandals. The treatments followed the same methodology described above.
A binary cross-over design similar to the previous experiments was used to evaluate the impact of the sandals against their respective randomly chosen mosquito species. The experiments were replicated for 6 days for each sandal design. In experiments where Anopheles species were used, experiments were conducted from 1800 hrs. to 00:00 hrs. (midnight). When Aedes species were used, experiments were conducted from 0600 hrs. to 1200 hrs. as they are day-biting vectors.
One hundred mosquitoes were released in two batches of 50 mosquitoes after every 3 hours in each experimental compartment. The volunteers were asked to wear a customized clothing that covered their whole body but left their lower limbs exposed by pulling the lower parts of their overalls to their knees. Mosquitoes were released into the two experimental compartments at the same time and left to acclimatize for 15 minutes before the volunteers entered their respective compartments assigned randomly before the start of the experiments. At the end of the hourly collections, the paper cups holding the mosquitoes were placed in a cool box until the next morning when the mosquitoes were killed using 70% ethanol. Field technicians counted and the sorted the mosquitoes into anophelines and culicines. Each anopheline was stored in a 5 mL micro-centrifuge tube (Eppendorf tubes) containing silica gel and later taken to the laboratory for further species identification. Other mosquito genera caught were stored in batches of five per micro-centrifuge tube and later identified by experienced entomologists.
Experiment 6: Field experiments to assess the efficacy of sandals treated with transfluthrin on wild mosquito bites
The field experiments for the two prototype designs of sandals were conducted in Minepa and Lupiro villages in Ulanga district, south-Eastern Tanzania. However, in these experiments, instead of HLCs, the mosquito electrocuting trap (MET) [61,62] was used to test to the efficacy of transfluthrin-treated sandals against wild mosquitoes.
Similar to the semi-field experiments, a binary cross over design was used when assessing the efficacy of the transfluthrin-treated sandals using the MET in the field. Two METs were placed at least 20 metres apart, next to rice fields and away from human dwellings. The volunteers wearing transfluthrin-treated or untreated sandals were randomly assigned to their positions only on the first day of the experiments and on subsequent nights rotated between the two fixed trapping points. The volunteers sat with their lower limbs exposed inside the MET from 18:00 to 00:00 hours for the first 45 minutes of each hour and collected electrocuted mosquitoes during the last 15 minutes of every hour. These experiments were conducted for a total of 12 days, 6 days for each prototype. These two prototypes (male and female sandals design 4) were selected because they provided the best protection against mosquito bites in semi field experiments above.
Statistical analysis
Data from both the semi-field and field experiments were recorded in a spreadsheet showing date of data collection; name of volunteer; whether the volunteer wore treated or untreated sandals; position of volunteer and the number of mosquitoes caught in each hour.
The effect of transfluthrin-treated sandals on the risk of exposure to mosquito bites was quantified by fitting a generalized linear mixed effects model with a negative binomial distribution to account for the over dispersion of mosquito count data. To account for day to day variation, date was included in the model as a random effect. The treatments on the sandals were included in the model as independent variables and the number of mosquitoes caught of those released as the dependent variable. Variations associated with fluctuations in temperature, humidity, wind direction and speed were assumed to be captured by the date random effect. Date, together with volunteers and hour and were treated as random effects.
To assess whether treated sandals divert mosquito to non-users, the number of mosquitoes caught by the untreated sandals in the two experimental compartments were compared. Diversion was modelled using three different values; the untreated sandal that was set up in the same compartment as the treated sandal compared with the two untreated sandals that were used in the comparative compartment. The model was used to derive incidence rate ratios (IRR) for numbers of biting mosquitoes in each of the three scenarios; untreated sandals in the treatment compartment and the two untreated sandals used in the comparative compartment.