Applying the COM-B behavior model to inform the delivery of spatial repellents and insecticide-treated clothing among forest exposed populations in Mondulkiri Province, Cambodia

Background: Southeast Asia is making tremendous progress towards their 2030 malaria elimination goal but needs new interventions to stop forest malaria. This study trials two new vector control tools, a volatile pyrethroid spatial repellent (VSPR) and insecticide-treated clothing (ITC), amongst forest-exposed populations in Mondulkiri Province Cambodia to inform their potential use for eliminating forest malaria. Methods: 21 forest-exposed individuals were given a questionnaire on their perceptions of malaria and preventive practices used, after which they trialed two products sequentially. Mixed methods were used to understand their experience, attitudes, and preferences regarding the products trialed. Quantitative data was summarized and qualitative insights were analyzed using thematic analysis, applying the Capability, Opportunity, Motivation Behavior Change (COM-B) model and Behavior Change Wheel Framework to identify intervention functions to support tailored product rollout amongst these populations. Results: Study participants reported a need for protection from mosquito bites in outdoor and forest-exposed settings and perceived both products trialed to be effective for this purpose. The VPSR product was preferred when travel was not required, whereas ITC was preferred for ease of use when going to the forest, especially in rainy conditions. COM-B analysis identified that key enablers for use of both products included their perceived efficacy and ease of use, which required no skill or preparation. For barriers to use, the odor of ITC was sometimes perceived as being toxic, as well as its inability to protect uncovered skin from mosquito bites, while the perceived usefulness of the VPSR product trialed was limited by its water sensitivity in rainy forest settings. Intervention components to encourage appropriate and sustained use of these products include education about how to use these products and what to expect, persuasion to use them from community leaders and targeted ads, and enablement to guarantee access. Conclusion: The rollout of VPSRs and ITC amongst forest-exposed populations can be useful for eliminating malaria in Southeast Asia. Study findings can be applied to increase product uptake in Cambodia, while research efforts can aim to develop products that are rainproof, easy to use in forest settings, and have favorable odor profiles to target users.

To recruit forest goers and dwellers from these villages, the study team and local health center staff members met with the village chief to inform them of the study, its procedures, and its objectives. The village chief identi ed individuals who would likely be eligible for inclusion for the study, liaised with those individuals, and scheduled meetings between individuals interested to partake in the study and the researcher. During this scheduled meeting, the study team introduced themselves to prospective participants and checked if they met the study inclusion criteria. For those eligible and interested to partake in the study, detailed information about the study was provided, informed consent was sought, and a one-hour meeting for the pretrial questionnaire was scheduled either immediately or later that day.
To recruit forest rangers who lived in or near the villages selected for recruitment, the study team contacted the director of wildlife sanctuary, requesting permission to enter restricted areas to meet forest rangers for research purposes. Permission was granted, after which a time was arranged for the study team to meet with the rangers at a ranger station. At this meeting, information on the study was provided, informed consent was sought, and a one-hour meeting for the pre-trial questionnaire was scheduled either immediately or later that day.

Products piloted
Two new vector control products were piloted in this study: a trans uthrin-based VPSR, and an etofenprox-based insecticide formulation (Perimeter ETO Insect Guard formulation) used for treating clothing, described further below. All individuals enrolled in the study were instructed on the appropriate and safe use of products by the study team, including a live demonstration, before being asked to use each product for 7 days indoors and outdoors, at any time. The VPSR was piloted by all participants rst, after which ITC was then piloted. Participants were instructed that during the trial period, they should use these products along with any vector control tools they would normally use.
The passive VPSR included the odorless active ingredient trans uthrin (manufactured by Bayer) which has an excellent safety pro le in mammals [21] and can prevent mosquito biting and/or induce mosquito mortality [22,23], depending on their levels of exposure to this active ingredient. For the product piloted, trans uthrin was infused in 2 small, rectangular lightweight substrates approximately the size of an A4 piece of paper (size 8.25 x 11.75 inches) and hung up to create an area of protection from insect bites. In this study, this VPSR product was given to users who were told that they should handle the product when wearing latex gloves that were provided with the product. This speci c VPSR product was designed for indoor use, but for the purposes of this research study, participants were told that they could use the product indoors or outdoors as long as they avoided getting the product wet.
The insecticide used to treat clothing in this study was etofenprox, manufactured by Mitsui Chemicals, in a proprietary formulation developed by Pine Belt Processing, a wholly owned subsidiary of Warmkraft, Inc. Etofenprox is a United States (US) Environmental Protection Agency (EPA)-approved insecticide developed to treat clothing worn by the US military. In this study, the researcher and assistant used the product to treat both forest ranger uniforms and the every-day civilian clothing provided by forest goer and dweller participants, providing a demonstration of how to treat the products. Participants observed this process but did not treat their clothes themselves.

Data collection and analysis
This study had four components: 1. Pre-trial questionnaire (Supplemental Information S1) that established the demographics, perceptions of malaria, working and social habits, and mosquito bite prevention practices and preferences of study participants. The two researchers who sought informed consent were present at each interview; one researcher conducted the interview using the questionnaire in Open Data Kit (ODK) with smart phones to limit data entry errors, and the other took notes. This process was not audio or video recorded. After each interview was completed, data were upload into a cloud server. Once data collection was complete, data were downloaded onto a password secured laptop, and consolidated in Microsoft Excel (Microsoft Corporation Version 2202). A Khmer version and an English version of survey responses were provided, and the English database was sent to a researcher (IC) who analyzed the data using Stata (StataCorp Version 17.0).

2.
Post-trial questionnaire. After using each product for seven days, all individuals enrolled in the study completed a post-trial questionnaire (Supplemental Information S2) on their experiences, preferences, and attitudes toward the use and functionality of that product including tolerability, usability, and user acceptability. We used a Likert scale which presents several options on how much participants agree or disagree with speci c statements, to assess their understanding on the use of the products, their overall usefulness and frequency of use, perceived changes to the number of mosquito bites after use, side effects or discomfort, preference of these products over other products in use, willingness to pay for products monthly, and locations preferred for purchase. The interviews were conducted in Khmer by two members of the study team, one of whom administered the interview and the second of which took notes. Once data collection was complete, analysis was conducted as described for the pre-trial questionnaire. 3. Key informant interviews. After piloting both products, an interview guide (Supplemental Information S3) was used to ask participants about their overall experience and attitudes toward and between the two vector control products trialed, their preferences relative to other mosquito bite prevention tools used prior to the study, and about factors that might in uence use. They were also asked about recommended messaging about each product, and which communication channels they recommended using for disseminating information on malaria prevention and care.
The same two researchers who conducted the questionnaire conducted the interviews, one of whom took notes. The interviews were audio-recorded, and responses to each question were collated in Microsoft Word.
The two interviewers transcribing interviews found results between all 21 interviews to be very similar, and after translating a random selection of 10 interviews into English, agreed that because results among all participants interviewed were so similar, these 10 interviews provided theoretical saturation appropriate for representative analysis of this study population. Those ten interviews were used for thematic analysis, comprising results from 2 forest dwellers, 5 forest goers, and 3 forest rangers. One researcher (IC) transposed the results into Microsoft Excel (Microsoft Corporation Version 2202), reviewed the results, and agreed with the two interviewers that responses amongst participants were similar, and that theoretical saturation had been reached. Themes on the ndings from each question or set of similar questions were identi ed by that researcher, who reviewed the translated results three times to con rm ndings, then highlighted representative quotes from each theme. The interviewer and assistant reviewed results and discrepancies were resolved through review of the original data.
1. COM-B analysis. Results from (2) and (3) were analyzed using the COM-B framework to identify barriers and enablers to intervention uptake, in this case the behaviors were to use the vector control products in addition to any vector control tools they would normally use (e.g., long-lasting insecticide-treated net (LLIN), LLIHN) [16]. Capability refers to the perceived ability to engage in the physical processes and thoughts necessary to use the intervention, opportunity refers to social and environmental in uencers in the settings being studied, and motivation refers to individual beliefs, emotions, and impulses that in uence behavior, but may not be consciously recognized [16]. After barriers and enablers to wearing/using the products were identi ed using COM-B, they were mapped to speci c intervention components to consider, using the Behavior Change Wheel (BCW) [15]. The BCW is an extension of the COM-B model that allows for each potential barrier and enabler to be linked to an intervention function that can allow for a change in behavior to be achieved (e.g., enablement, modeling, education). It also incorporates methods for the selection of contextually appropriate intervention components and options to consider for delivery of the selected approaches [15,16]. We selected priority intervention components to allow for the trial distribution of these products, as well as sustained delivery, due to their relevance for informing implementation approaches.
Sample size calculation: The sample size for this study was based on the quantity of vector control products available. The products piloted were still in development and not yet available at large scale that normally follows commercialization. The quantities available allowed for 21 individuals to pilot these products at this stage of the research program.

Demographics
Participants of this study were mostly men, aged 26-35, consisting of a mixture between Khmer and Bunong ethnic groups (Table 1). All participants lived in or near the forest, with forest dwellers and goers mostly being farmers who cultivate rice, cassava, cashews, and other crops as their main source of earnings. For ethnicity, all forest rangers were of the Khmer ethnic group, whereas forest dwellers and forest goers comprised both Khmer and Bunong ethnic groups. All participants spoke the Khmer language uently and most could read or write it. Most could also speak Bunong uently, although only three forest goers could read or write in this language, the written form of which has only existed in recent decades. For levels of education, the forest rangers enrolled generally had higher levels of education, while forest dwellers had less. Pre-trial questionnaire Perceptions of malaria and care-seeking behavior When asked about their perceptions of malaria, individuals were presented with options on whether they agreed or disagreed with a series of questions. Responses were similar between all participant groups; all agreed that mosquito bites are dangerous, that they worried about mosquito bites, and that mosquito bites cause itching and make ugly marks on the skin. Most participants agreed that mosquito bites can cause severe disease needing hospitalization. The reasons for worrying about mosquito bites included getting sick with malaria and dengue, as cited by forest rangers and goers, as well as the costs of going to the hospital, as cited by forest dwellers, the group that described the most substantial barriers to accessing care.
When asked about malaria diagnosis and care-seeking behavior, the most common method cited for diagnosis was using microscopy or a malaria rapid diagnostic test. Care was sought in private clinics (76%), referral government hospitals (71%), government health centers (48%), or from the village malaria worker (VMW, 5%). Forest rangers typically visited government hospitals, while goers and dwellers did not have a main source of care, reporting that they went to a variety of these facilities.
When asked about what they did to recover from malaria, all mentioned taking medicine, with 6 participants mentioning that in addition to medicine, they received intravenous therapy.

Working and social habits
All participants stated that they spent time in the forest, with forest rangers spending four to ve days a week in the forest, presumably for work, while forest dwellers and goers reported working or travelling in the forest three to four days a week. All participants reported going to the forest with others; forest rangers and dwellers went with co-workers, while forest goers reported going with neighbors. Participants seldom reported going deep into the forest with their family members.
When asked about where they spent their time during the day, all individuals reported waking up before 7AM and going to sleep around 9PM, spending time inside the house, outside of the house, and away from the house throughout the day (Fig. 3).
Responses were similar among all participant groups, with rangers reporting more time spent away from their house (the ranger station) during all of the times that they were asked about.

Mosquito biting frequency, times, and locations
When asked about being bitten by mosquitoes inside or outside of their houses, all participants stated they had been bitten by a mosquito recently, with approximately half reporting being bitten inside the house the day they were interviewed, and the other half saying they were bitten the day before. These self-reported biting frequencies were similar inside the house and outside the house, with biting times shown in Fig. 4.
Perceived mosquito biting times were similar amongst all participants in the study, with the most commonly cited biting time being in the evening (6-9PM), occurring both inside the house and outside of the house. At this time, most forest goers and dwellers reported being inside the house, with rangers being inside, outside, or away from the house (Fig. 3). The second most commonly reported biting time was 5AM-9AM, when forest goers and dwellers reported mostly being inside the house and sometimes away from the house, and rangers reporting being inside, outside, or away from the house. The third most commonly reported biting time was late afternoon (3PM-6PM), where participants reported being in all locations, with more rangers reporting that they get bitten by mosquitoes during this time.

Mosquito bite prevention habits
When asked about which mosquito bite prevention methods participants used, all participants reported using methods to prevent malaria inside their house or ranger station, and in the forest, and most (76%) stated that they used methods to prevent malaria outside the house / ranger station. Some forest goers and dwellers reported not knowing which bite prevention methods to use in forest.
A summary of mosquito bite prevention methods used is in Fig. 5 with results shown by participant group in Table 2. The most common methods used indoors were mosquito nets and coils, which were mostly used by forest goers and dwellers. Forest rangers reported using approximately twice the number of mosquito prevention methods as dwellers and goers. These rangers explained they did not use mosquito nets often because they could not be hung in the bunk beds at the ranger station. Instead, rangers described using skin repellents, hammock nets, and insecticide spray which is an aerosol spray that rapidly kills mosquitoes and other insects.
Methods used outside of the house (but near the house) and in forest sites (away from the house) were similar although fewer methods were used in forest sites as compared to other settings. Forest goers and dwellers most commonly reported wearing long-sleeved clothing and burning coils both day and night, outside of the house and in the forest. For forest rangers, methods used during the day were most commonly topical skin repellents, followed by wearing long sleeves and insecticide spray. At night, skin repellents were most commonly used outside the house and in the forest, followed by burning coils, wearing long sleeves, and using insecticide spray. The main notable difference between forest ranger prevention methods used outside of the house vs in the forest at night were hammocks, which were more commonly used in the forest as compared to outside of the house.  When asked about their ownership and usage of bed nets, forest goers and dwellers were similar, all of whom reported that their household owned one to three bed nets. These were obtained from the government free of cost or purchased for 20,000 and 50,000 Riel ($4.86 and $12.15 USD) per net from retail shops or markets 36 to 60 km from their homes (Table 3). Reported bed net usage was high, with all 14 forest dwellers and goers reporting that they used bed nets the last time they stayed in the house, with multiple people sleeping under nets. The questionnaire did not specify whether the bed nets were treated with insecticides or not. The majority of forest goers and dwellers also reported owning at least one hammock net, most of which they believed were treated with insecticides (based on the presence of a National Center for Parasitology, Entomology, and Malaria Control (CNM) or Global Fund-procurement logo/label). These hammock nets were either provided for free from the government or purchased for 50,000 to 60,000 Reil (USD $12.15 to $14.58) between 15 to 60 km from their homes. Only two hammock nets were reported to be used the night before the interview, inside the house.
For forest rangers, all owned one hammock, and only one used a bed net obtained from a market 15km away from the ranger's home. The hammock nets were obtained free of charge from their work, which are typically a hammock zip net which allows for a rain y to be hung over the top. It was unclear whether these hammock nets were treated with insecticides. Forest rangers reported the frequent use of hammock nets, with the vast majority stating they used the net the night before they were interviewed, both in the ranger station in the forest and in the forest itself.
Post-trial questionnaire After the pre-intervention questionnaire, participants piloted the VPSR for seven days and were then administered a questionnaire. After that questionnaire, participants tried the ITC for seven days and were given the same questionnaire.
Results are summarized below in Table 4 and described for VPSR, followed by ITC. VPSR All participants used the VPSR product during the piloting phase and found it useful or very useful and perceived reductions in mosquito bites after using it. They reported using the VPSR indoors, in the bedroom or in the main living area, and some participants, mostly forest rangers, brought the VPSR to the forest with them in their backpack. Those who did not carry the VPSR product to the forest stated that they were concerned about the product getting wet from the rain. Generally, participants liked the odorless quality of the VPSR as well as how the product looked. Only one side effect was reported by a forest ranger who touched the product without gloves, mentioning that he felt pain on his skin similar to a needle injection.
When asked about whether they preferred the VPSR to other mosquito bite prevention methods they were using, most preferred the VPSR to LLINs, hammock nets, and skin repellents or mosquito coils, due to its perceived ability to chase mosquitoes away, both indoors and in the forest. Other preferable aspects of the VPSR included that it was more comfortable to sleep with compared to LLINs which can be hot, and did not create the bad smell and smoke generated when using mosquito coils which participants perceived to be harmful to their health. On the other hand, some participants mentioned preferring LLINs to the VPSR product due to their long duration of e cacy and the physical barrier they provided from mosquitoes. They also preferred hammocks and skin repellents due to their ability to withstand rain, which they mentioned was particularly helpful when spending time in the forest. All participants stated that they would recommend the VPSR product to others and would be willing to pay for it, with willingness to pay most commonly amounting to 5,000 or 10,000 Reil ($1.22 or 2.44 USD) monthly.

ITC
Most participants (81%) used the ITC; those who did not pilot the product were concerned about its smell, which they believed meant that the product was unsafe to use. Those who used ITC reported wearing it to the forest and some found the product to be useful, perceiving moderate reductions in mosquito bites during its use. However, several side effects were noted while wearing ITC. Most participants mentioned its unpleasant odor, and some also noted that wearing the treated clothes caused itchiness, skin irritation, and/or dizziness.
When asked about whether they preferred the ITC as compared to LLINs, hammocks, coils, and skin repellent, responses varied widely. Forest rangers preferred treated clothes to LLINs and hammocks, mentioning their usefulness when spending time in the forest. Those who preferred LLINs and hammocks believed that these tools offered better protection than ITC, had longer durations of e cacy, were more useful for sleeping, and did not have the bad smell that the treated clothes had. Forest rangers generally agreed that they would recommend this product to others, while other participant groups gave mixed responses. Those who would recommend the product expressed that they would be willing to pay on average 11,000 Reil ($2.71 USD) monthly, while those unwilling to pay stated their reasons as a perceived lack of e cacy, bad smell, and that the product caused itchy skin.

Key informant interviews and COM-B analysis
Data from a convenience sample of ten key informants comprising two forest dwellers, ve forest goers, and three forest rangers is presented below, with their demographics shown in Table 1. We summarize themes on perceptions of malaria (Table 5) and experience trialing VPSR and ITC (Table 6), including application of the COM-B framework to code for barriers and enablers to product use.

Health beliefs and perceptions of malaria
Consistent with results found from the pre-trial questionnaire, participants expressed that they were worried about malaria, recognizing the risks it posed to themselves and their community due to the amount of time they spend in the forest, which is an enabler to product use (Table 5). Additional enablers to product use included their ability to protect vulnerable populations, such as children, and that preventing malaria infection could avert treatment seeking costs which are a particular challenge owing to the remote locations in which these participants live. We need methods to protect ourselves from mosquito bites when spending non-resting time in the forest.
"I go to the forest and elds in the forest where there are lot of mosquitoes, I do not have proper protection, only from sleeping in the hammock." -Male forest goer, age 26 Enabler: psychologic motivation and automatic motivation Products piloted are seen as meeting a need that is not met by the other prevention products available because they offer protection from malaria and dengue fever during non-resting time in the forest.

Experience with piloting VPSR and ITC
Participants identi ed positive attributes associated with both products trialed. Common themes applying to both products trialed are described in Table 6, while attributes where the products differed, as well as participant comparisons between the products are described in Table 7. Enablers to the use of both products included psychological capabilities, such as their perceived ability to offer protection from malaria to one's family and community, as well as their perceived e cacy where participants could immediately observe the absence of mosquitoes near the VPSR product and a reduction in mosquito bites when wearing ITC. Ease of use was another major enabler to product uptake, mapped to physical capability and automatic motivation. The odor pro le of the products was an important factor affecting user experience, where participants equated product odors with e cacy and safety. This presented a barrier to uptake affected by psychological capability and automatic motivation, where users need to understand that the VPSR is odorless yet effective, and that the smell of the ITC are safe. A barrier to uptake is also the ability of users to purchase the products, as they may not have the resources available to do so. Misperception about odor, such that there was discomfort and concern that smelly products might be toxic and that odorless ones might not be effective.
People do not like spending money on preventive products for malaria.
" [People] don't like spending money, for example buying mosquito coils." -Male forest goer, age 26 Barrier: physical opportunity Resources to buy the products may not be available.
When comparing the two products piloted, participants perceived distinct advantages that drew insights on use case scenarios for each, with participants concluding that the VPSR product was best for staying in one place, particularly in indoor settings, while ITC were best for use in the forest due to ease of use for mobile work and continued e cacy in rainy conditions (Table 7). Drawing on differences between the products, the VPSR had more visible perceived e cacy from the lack of mosquitoes around the product as compared to ITC, which only prevents mosquito landing on treated clothes. Therefore, barriers to ITC include the need for psychological capability of users to understand that these clothes do not protect uncovered skin, as well as a need for physical opportunity to access additional protective measures for uncovered areas, such as topical repellents.
The ITC had a smell and reported side effects, which target users need the psychological capability to understand are temporary and not serious. Users also need to understand the need for periodic retreatment of clothes which participants believed would be easy to do based on watching study team members do so. Despite these differences, the ITC was cited to be easier to use in mobile forest settings due to ease of use and its ability to withstand rain. The VPSR would need to be carried when being mobile and could not get wet because it was designed for indoor use, presenting physical capability and opportunity as barriers for nding a way to keep the product dry in rainy conditions. Enabler for treated clothes: physical capability and physical opportunity ITC were easy to use in the rainy outdoor conditions common for target users, and suitable for the mobile outdoor nature of their work.

Recommended communications channels
We asked participants on how they would convince other people in their village to use either the VPSR product or ITC and mapped these onto BCW intervention functions (Table 8). Participants focused on accessing in uencers, such as health center staff, village health workers, and elders. Communications channels or touchpoints recommended for sharing the bene ts of the products with others in their village were community meetings led by the village chief, and rangers also mentioned their monthly meetings being a useful place to introduce and promote new products. Billboards and mobile motor announcements were also suggested, while radio and mobile channels were not recommended given limited receptivity. We have many in uencers on malaria prevention practices, including using traditional methods, watching what other people in the community do, and advice from elderly community members and health center staff.
"[I learned] from the health center staff who educated me, I watch others and follow." -Male forest goer, age 23 "I used to raise cattle; I used to burn the leaves or rice hay to get the smoke to prevent the mosquitoes from biting the cattle. I know this method from the old people." -Male forest ranger, age 55 Modeling: watching others use the product.
Persuasion: Listening to elders.
Education: Learning about additional malaria prevention practices.
Community or ranger meetings are a recommended forum to introduce products particularly if in uential leaders or experts, e.g., VMWs or ranger team leaders are involved.
"[I recommend you share information on these products at a] community meeting, invite villagers to join, let them ask questions and rest up their concerns. When they know the product well, they will use it." -Female forest dweller, age 40 "The best way [to introduce products] is for VMW to organize meetings with villagers to introduce products. People will use it when they know it is effective and trust VMW." -Male forest goer, age 46 "[Introduce the project at] monthly meetings because everyone joins is and reports to the chief ranger. This is a great time to share and get everyone used to it." -Male forest ranger, age 37 Persuasion/Education: leaders introduce products to community in open forums, citing their bene ts to health.
Targeted mass communication channels that are accessible to target communities can be useful to inform the village of new products "If there is an ad on the mobile motor(bike) or announcement would be good. Everyone in the village knows." -Male forest goer, age 38 "Broadcasting on a radio or the big billboard is good because everyone can see it, not only rangers but everyone also knows it." -Male forest ranger, age 24 "For radio, some will listen and some not because it is not possible in the forest." -Male forest ranger, age 51 Persuasion/Education: Provide information so that the entire community can be engaged The COM-B components for areas identi ed for key informant perceptions of malaria (Table 5) and of the products piloted (Tables 6 and 7) and recommended communications channels mapped onto intervention functions using the BCW (Table 8) were used to inform the selection of behavior change techniques to address these target behaviors as well as modes of delivery for introducing the products. For this, we considered the pilot scale used in this study, as well as the hypothetical delivery of those products through longer term efforts which would require sustained use ( Table 9).
The priority COM-B components to leverage enablers and address barriers to trial use of these products included psychological capability, physical opportunity, and re ective motivation, which we mapped to modeling, education, and persuasion. For purposes of piloting the products, the behavior change techniques used were effective, comprising of instructions on how to use the product and information on its bene ts delivered through face-to-face sessions. Village and ranger leaders were involved in the recruitment process, allowing for effective persuasion that study team members providing the products were trustworthy.
For sustained use of these products, similar to trial use priority COM-B components should include psychological capability, mapped to education to ensure that target populations understand what the products are for and how to use them. Different from trial use however, social opportunity and automatic motivation should be leveraged to encourage regular use, which we recommend centering around their ability to protect community members, including vulnerable groups such as elders and young children. For these COM-B components, BCW intervention functions for use can include modeling and persuasion through targeted advertisements and trusted members of society could encourage uptake, enablement through service provision would ensure accessibility, and training focused on the retreatment of ITC could be necessary. For these intervention functions, behavior change techniques from our results above and modes of delivery drawing from Table 8 could include multi-media campaigns using outdoor billboards, printed or digital graphic tools, mobile motorbike announcements, and community events led by health center staff, village malaria workers, village chiefs, and lead forest rangers.

Discussion
Forest malaria challenges malaria elimination efforts in many locations, and is responsible for many remaining pockets of transmission in Southeast Asia [13]. This is the rst academic study to characterize at-risk groups that spend time in the forest in Mondulkiri Province, Cambodia and examine their experiences and perceptions after use of a VPSR and ITC in Cambodia. We furthermore apply the COM-B model and BCW to analyze qualitative insights related to use of these products, to inform forest malaria elimination programming strategies in Cambodia. Our study found that the forest goers, dwellers, and rangers who participated in this study all understood mosquito bites to be dangerous and had gaps in protection during waking hours, especially in forest settings, and furthermore found both products piloted to be e cacious for preventing mosquito bites. For use cases, participants preferred the VPSR product when travel was not required, while ITC was preferred when going to the forest due to its ease of use in not having to carry anything separately, and its e cacy in rainy conditions.
COM-B analysis identi ed that key enablers for use of both products included psychological capability and automatic motivation about their perceived e cacy and ease of use, requiring no skill or preparation. Barriers to use included psychological capability and re ective motivation that the odor of ITC was sometimes perceived as being toxic and that it was unable to protect uncovered skin from mosquito bites, while for the VPSR product physical capability and physical opportunity would be necessary to nd ways to keep the product dry if being used in rainy forest settings. Mapping these COM-B components to BCW intervention functions, education can be used to explain that the smell of ITC is safe and that additional protective measures are necessary on exposed skin, that the VSPR needs to be kept dry, and that both products offer protection from malaria and dengue for target users and members of their community, saving time and money seeking care as well as risks posed to employment. Enablement will be needed to ensure products are accessible and affordable to target users (willingness to pay was between $1.44 and $2.71 monthly), and modeling and persuasion through respected members of the community such as village leaders, ranger team leads, and targeted advertisements will likely be necessary for longer term use of these products.
We suggest that our results be applied to the design of social behavior change communication (SBCC) strategies to support the introduction and uptake of these products, recommending the complete and continuous use of multiple mosquito bite prevention products as the best way to avoid losing valuable time and money due to a possible serious illness. To enable this level of protection, the continued use of LLINs, long-lasting insecticide treated hammock nets, and other preventive measures should be promoted alongside efforts to motivate the use of VPSRs inside and outside the house, and ITC when leaving the house to go to the forest. If the target population chooses to treat short-sleeved shirts or short pants with insecticides, the use of topical repellents to protect uncovered parts of their body should be encouraged [24]. Our results also suggest possible delivery channels for SBCC to achieve these communication objectives, which include face-to-face sessions reinforced with illustrated print or digital materials in local language, as well as multi-media campaigns using accessible channels to target audiences e.g., outdoor billboards, mobile motorbike announcements, and community events. Across channels, campaign spokespeople or characters featured in SBCC content could include respected elderly community members and leaders and/or health worker staff.
With regard to generalizability, our ndings are consistent with broader themes identi ed through other research on forest malaria, including poor access to healthcare as a result of mobility, and association with poverty [25,26]. Our results reinforce the assessment that efforts to eliminate forest malaria would bene t from the use of multiple vector control interventions, especially those that can be used to reduce outdoor mosquito biting [9], that could also be used in combination with chemoprophylaxis [13]. Our ndings on the lifestyle habits of forest goers, dwellers, and rangers in Mondulkiri province however, are speci c to those within the speci c district and province in Cambodia where the research was conducted, with similar occupation, gender, and ethnic groups present, as con rmed by similar ndings in a 2018 study conducted amongst 4,200 forest workers in the same province [5]. Even in villages in Mondulkiri province, where reported mosquito peak biting times were similar, participants reported to be in different locations during those times [27], suggesting they should be a different target group than those in our study. Other risk pro les in Cambodia could include illegal loggers, who had very different forest-going habits compared to our study participants and did not use mosquito coils due to the fear of being detected [6], as well as mobile populations that create temporary encampments in the forest such as those described in Stung Treng province [28].
For speci c products piloted in our study, many other studies on the use of VPSRs and ITC have been published. For VPSRs, one study investigated their use in Mondulkiri Province, Cambodia, in 2013 for indoor use amongst Bunong villagers living near the border with Vietnam, with similar user acceptability to our ndings where almost all participants and users of the VPSR product perceived it to be useful and would be willing to use it again [27]. That study found that socioeconomic status did not have a signi cant effect on willingness to pay for VPSRs, which is consistent with our ndings where forest rangers, who despite having a higher income and socioeconomic status than forest goers and dwellers, had a lower willingness to pay for these products perhaps because vector control tools are provided to them for free through their jobs. VPSRs were also preferred to mosquito coils because they are long lasting and do not require frequent user activation [9], and furthermore do not cause smoke which can irritate end users and may be a risk factor for lung cancer [29,30].
For the use of ITC among forest-going populations, only one other publication is available, which explored the use of an odorless permethrin-based product in 2015 amongst 234 rubber tappers in the Mon State in Myanmar [31]. The study population was comprised of migrants staying in plantation lodges overnight, who used similar malaria preventive measures to those reported in our study, including insecticide-treated nets, mosquito coils, and long sleeves. Although the Myanmar study used an odorless product with a different active ingredient for which resistance is increasingly reported globally [2], user acceptability assessments had similar ndings to those in our study, with participants citing that they liked the ease of use and comfort of using ITC. This Myanmar study also made suggestions on SBCC strategies that were consistent with our ndings in Cambodia, suggesting that these strategies incorporate promotional and educational messages delivered through billboards, with products becoming available in health centers and shops.
Importantly, the products piloted are not yet available for widespread distribution for public health use. VPSRs may soon be available for this indication, with a recent cluster randomized trial in Indonesia showing that a VPSR provided 31.3% protective effect for incident malaria infections, offering even higher levels of protection in clusters with higher malaria endemicity [32]. A cluster randomized trial amongst 2,907 households in Iquitos, Peru using a VPSR showed a protective e cacy of 34.1% against Aedes-borne diseases [33]. Unitaid is sponsoring a large multi-country trial of VPSRs, and additional evidence on its e cacy in preventing malaria and dengue fever is forthcoming [34]. Insecticide treated clothing has long been used by military populations, and there is evidence available on their e cacy [35]. A cluster randomized trial on the effect of permethrin-based clothing treatment with topical repellents on the incidence of malaria is underway [36]. Although most studies on ITC have used permethrin, etofenprox, the active compound used in our study, has better wash resistance than permethrin [36].
Our study also supports ndings in our wider research project, that topical repellents could be useful when wearing ITC due to the latter's inability to protect uncovered areas of skin. Topical repellents are already in use by many rangers, suggesting its acceptability amongst this at-risk population in Mondulkiri province. They have also been widely distributed in a study conducted in Ratanakiri Province, Cambodia, which showed that regular use can be challenging [37,38] and may need to be encouraged using SBCC-type approaches. Topical repellents have otherwise shown to be e cacious, reducing the odds of malaria infection in a large study amongst 116 villages in Myanmar [24], and reducing both malaria infection rates and mosquito density when distributed to a refugee camp in Northeastern Kenya [39].
The growing evidence base on VPSRs and ITC suggests that new products within these classes may be available for public health use in the coming years. In the meantime, two areas of research should be prioritized. First, to accelerate Cambodia's malaria elimination goals and following evidence of entomological protective e cacy among local vectors in Mondulkiri province, products within these classes can be delivered to more at-risk individuals using SBCC strategies, collecting implementation outcomes on factors that enable or prevent sustained uptake. These bite prevention products can be given with several other protective measures against malaria including chemoprophylaxis [40,41], to inform the evidence base on eliminating forest malaria. Second, our ndings that the odor of these products in uences user acceptability should be applied to the development of new mosquito bite prevention products or optimization of existing products, using ingredients and formulations that have an acceptable or desirable smell to target populations.
This study had several limitations. We had a small sample size, we did not pilot topical repellents together with other products trialed, and our distribution of ITC involved a member of the study team treating clothes for participants, which could be done for large-scale implementation but would be much more expensive than teaching participants how to treat clothes. However, our sample size was likely su cient given consistent ndings by study participants and similarities in responses on user experience amongst a subset of only 10 of these participants. Our study also suggests that forest goers and dwellers might be suitable for integrating into a single risk pro le since their results were similar; larger studies can explore the potential for this, as well as the possibility to further segment these groups by gender, socio-economic status, and more. For ITC distribution, future work can focus on identifying potential barriers to treating clothing effectively, and how to encourage retreatment at appropriate times periodically.
In conclusion, malaria elimination strategies for Cambodian forest-exposed populations could bene t from the use of VPSRs, ITC, and topical repellents to protect exposed skin not covered by treated clothes. Study ndings can guide the design of tailored SBCC strategies, incorporating education and training on the ability of these products to protect people from malaria, how to correctly use them, and what to expect from their use including their odors or lack thereof, enablement to ensure access, and persuasion from targeted advertisements and trusted members of society. Rollout strategies should be iteratively assessed using implementation science frameworks, and in parallel, research and development efforts should prioritize products that are rainproof and easy to use in forest settings, with favorable odor pro les for target users.

Declarations
Ethics approval and consent to participate: The study protocol was approved by the University of California's Human Research Protection Program Institutional Review Board (IRB 20-30660) and the Cambodia Ministry of Health National Ethics Committee for Health Research (NECHR 296). Informed consent was sought by each participant prior to enrollment.
Availability of data and materials: All data generated or analyzed during this study are included in this published article and its supplementary information les.

Competing interests:
The authors declare that they have no competing interests.