Residual Malaria Transmission in Select Countries of Asia-Pacific Region: Old Wine in a New Barrel

Abstract Background Despite substantial reductions in malaria burden and improvement in case management, malaria remains a major public health challenge in the Asia-Pacific region. Residual malaria transmission (RMT) is the fraction of total transmission that persists after achievement of full operational coverage with effective insecticide-treated bed nets (ITNs)/long-lasting insecticidal nets (LLINs) and/or indoor residual spray interventions. There is a critical need to standardize and share best practices for entomological, anthropological, and product development investigative protocols to meet the challenges of RMT and elimination goals. Methods A systematic review was conducted to describe when and where RMT is occurring, while specifically targeting ownership and usage of ITN/LLINs, indoor residual spray application, insecticide susceptibility of vectors, and human and vector biting behavior, with a focus on nighttime activities. Results Sixty-six publications from 1995 to present met the inclusion criteria for closer review. Associations between local vector control coverage and use with behaviors of human and mosquito vectors varied by locality and circumstance. Consequently, the magnitude of RMT is insufficiently studied and analyzed with sparse estimates of individual exposure in communities, insufficient or incomplete observations of ITN/LLIN use, and the local human population movement into and from high-risk areas. Conclusions This review identified significant gaps or deficiencies that require urgent attention, namely, developing standardized procedures and methods to estimate risk exposure beyond the peridomestic setting, analytical approaches to measure key human-vector interactions, and seasonal location-specific agricultural or forest use calendars, and establishing the collection of longitudinal human and vector data close in time and location.

countries that are reorienting their programs toward elimination phases (Nepal, Democratic People's Republic of Korea, Vanuatu) or moving toward subnational elimination targets [1,2], many are also facing challenges with varying levels of local RTM and/or emergent or increased zoonotic (simian) malaria transmission. Residual malaria transmission (RMT), defined by WHO as "persistence of parasite transmission even with good access to and usage of ITNs or well-implemented IRS, as well as in situations where ITN use or IRS are not practical, " represents a critical challenge for malaria control and elimination efforts [9].
The urgency of addressing RMT is heightened by the continued transmission of artemisinin-resistant Plasmodium falciparum parasites in countries of the Greater Mekong Subregion (GMS) that should they spread beyond the Myanmar-Northeastern India border [10], threaten global malaria control efforts [11]. RMT, previously referred to as "outdoor malaria transmission, " has been addressed by 2 specialized working groups over the years, the Roll Back Malaria Vector Control Working Group work stream on outdoor transmission (https:// endmalaria.org/until-2015-%E2%80%93-outdoorresidualmalaria-transmission) and the Mekong Outdoor Malaria Transmission Network (https://www.apmen.org/events/7-10november-2016), centered at Kasetsart University in Thailand. In this region, "outdoor transmission" is just one of several components encompassing RMT, owing to a combination of human and vector behaviors when forest workers, dwellers, or periodic, temporary forest exposed populations reside in or visit forest areas or do not sleep in protected dwellings or structures [9,[12][13][14][15].
In addition, local vector species may exhibit behaviors that allow them to avoid LLINs/IRS to which they are physiologically susceptible. According to Killeen [8], the primary vector behaviors that contribute to maintaining residual transmission are "(1) Natural or insecticide-induced avoidance of contact with treated surfaces within houses and early exit from them, thus minimizing exposure hazard of vectors which feed indoors upon humans; (2) Feeding on humans when they are active and unprotected outdoors, thereby attenuating personal protection and any consequent community-wide suppression of transmission; (3) Feeding upon animals, thus minimizing contact with insecticides targeted at humans or houses; [and] (4) Resting outdoors, away from insecticide-treated surfaces of nets, walls and roofs. " The current review focuses on the more obvious and readily surveyed issues where and when compromised exposure and unpreventable transmission risk may occur.
The purpose of this review is to synthesize the current body of evidence on vector and human behaviors that contribute to RMT and illustrate some of the complexities in using programmatic data and findings from research on vector control coverage, general health services, and methods for characterizing human-vector interaction. The review focuses on 2 selected ecoregions affected by persistent malaria transmission because of serious technical and health system problems [16], namely, (1) forest malaria in the GMS and in the Indo-Malaysian Archipelago, with variability in intensity of transmission, vector exophily, length of transmission season, and health system performance; and (2) coastal and lowlands malaria in the Oceania Region, with the potential (at least in principle) for mosquito habitat source reduction.
Within each of these ecoregions, there is considerable variability related to anthropic, natural, and health system factors that affect the magnitude of RMT. It is also appropriate to contextualize the review around the revised RMT definition: "persistence of malaria transmission following the implementation in time and space of a widely effective malaria programme" [17]. This is derived from the 2014 WHO guidance note regarding transmission that occurs even with sufficient access to and usage of ITNs or well-implemented IRS, as well as in situations where ITN use or IRS are not practical interventions.
Depending on local circumstances, these core interventions can be impractical when people are awake and active, with nets sometimes just a meter or two outside of their effective physical reach. LLINs and long-lasting insecticidal hammock nets may readily be used to protect forest workers sleeping outdoors but are generally impractical owing to the strenuous and demanding nature of forest work [12,18] (Lucas Nene, personal communication, July 23, 2020) and represent an additional burden that not all forest workers are willing to accept [19]. Users associate remote settings with increased damage and/or soiling of bed nets, driving the choice of which type of bed net to bring (Lucas Nene, personal communication, July 23, 2020). Sleeping in makeshift lean-to or improvised ground-level shelters, rather than raised structured, often in hammocks, increases exposure of a high proportion of forest workers to malaria vectors [12,20]. Hanging mosquito nets outdoors may be problematic given current commercial mosquito net designs and their reliance on external supporting structures.
In Cambodia, some users hire local tailors to alter and sew multiple nets together into a larger coverage net (Lucas Nene, personal communication, July 23, 2020). User complaints include the difficulty in folding and storing the stiff/rough fabrics of LLINs and the fact that remote sleeping places are often constrained by obstacles and limited space; in remote settings, this promotes improvisation of hanging solutions, such as cutting poles and digging post holes to support free-standing bed nets (Lucas Nene, personal communication, July 23, 2020). Alternatively, in various sleeping areas or household structures, bed nets may be compact in size or hung using internal or complementary supporting structures or alternative "clamping/ tying" systems, increasing their ease of use [21].
To address these concerns, the variables noted above for each ecotype must be defined. There is increased interest in malaria vector bionomic databases and human population movement, with information compiled from published literature [22,23]. While the framework described by Guyant et al [24] has been used to develop more targeted behavior change and outreach interventions for mobile and migrant populations in Cambodia, a programmatic approach based on field observations is adopted herein. This approach is used to define 5 RMT categories: (1) migratory forest-goers of various types; (2) indigenous people who live in the forest, or forest dwellers; (3) village-based people who perform seasonal work in "farm huts"; (4) security, wildlife, and border protection and defense forces; and (5) people living in lowlands and coastal areas where Anopheles punctulatus complex species are present (Australasian Region).
While human Plasmodium knowlesi is primarily regarded a zoonotic pathogen, all indications suggest that human-to-human transmission can occur and probably is occurring in some situations [25]. Recent epidemiological trends in eastern Malaysia (Borneo) suggest that P. knowlesi infections will become more important as P. falciparum and Plasmodium vivax are eliminated [26,27], and the close parallels with residual human malaria parasite transmission suggests the inclusion of "monkey malaria, " a term that is defined within a specific context of transmission epidemiology and involves several sylvatic vectors belonging to the Leucosphyrus group (Anopheles dirus and Anopheles leucosphyrus complexes-Anopheles balabacensis, Anopheles cracens, A. dirus, and Anopheles latens [28]). The human populations at greatest risk of infection are inhabitants of hilly forested areas, particularly ethnic minorities and subsistence farmers who have relatively substandard living conditions and low educational background, and whose normal life activities include forest exploitation and subsistence-level swidden cultivation practices [29][30][31][32].

Search Strategy and Eligibility Criteria
This review follows the guidelines in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [33]. The search occurred between January and May 2020. All data were extracted by 2 independent researchers, and discrepancies were resolved by consensus.

Data Sources and Search Strategy
Studies were identified by searching electronic databases, scanning reference lists of articles, and consulting with experts in the field. No limits were applied for language in case there was an available English translation. The search was applied to PubMed and the Cochrane Database of Systematic Reviews. The search terms in Figure 1 were applied to all databases.

Inclusion Criteria
In accordance with the 2014 WHO (2014) definition of RMT-"All forms of malaria transmission that persist after full universal coverage with effective ITN and/or IRS interventions has been achieved" [9]-we used the following inclusion criteria to down-select abstracts and publications for the review: data on 1 ownership AND 1 usage indicator (relating to ITN/LLINs) OR data on the IRS indicator OR insecticide susceptibility of primary and secondary vectors OR human and vector behavior observations from qualitative and quantitative surveys, including nighttime human activities.
Indicators for ownership of ITN/LLINs included the percentage of HHs with at ≥1 ITN/LLIN and the percentage of population with access to an ITN/LLIN within the household. Indicators for usage of ITN/LLINs included the percentage of the population reporting having slept last night under an ITN/ LLIN and the percentage of <5-year-olds with reports of having slept last night under an ITN/LLIN. The IRS indicator was the percentage of houses (population) protected by IRS in the last 12 months. Ownership of >1 ITN/LLIN per household and ITN/LLIN use in the target groups of children <5 years of age and pregnant women were key indicators for the Global Fund grants.
(Asia OR Pacific) AND human AND (behavior OR behavior) AND malaria -(Asia OR Pacific) AND "Human Activities" AND malaria (Asia OR Pacific) AND Outdoor OR outside OR residual AND malaria AND (behavior OR behavior) (Asia OR Pacific) AND Malaria AND (outdoor OR residual) AND behavior (Asia OR Pacific) AND Human AND location AND malaria (Asia OR Pacific) AND ("Human behavior" or "human behavior") AND malaria (Asia OR Pacific) AND "Human activities" [Mesh] and malaria and (outdoor OR residual) (Asia OR Pacific) AND Anthropology OR anthropologic AND malaria exposure (Asia OR Pacific) AND Outdoor AND human AND behavior AND night AND Asia-Pacific (Asia OR Pacific) AND Outdoor AND malaria AND ("human behavior" OR "human behavior") (Asia OR Pacific) AND "human exposure" AND malaria (Asia OR Pacific) AND "Human Activities" [MeSH Terms] AND night time (Asia OR Pacific) AND human AND (behavior OR behavior) AND night time (Asia OR Pacific) AND human AND (behavior OR behavior) AND day time Data were obtained from demographic and health surveys or malaria programmatic household surveys. One ITN per 2 people was considered to be sufficient, on average, to protect all individuals in the household [1,2]. Where access to ITN/LLIN data is not available, we reviewed other indicators, such as (1) households with ≥1 net (any kind)/ITN/LLIN; (2) households with sufficient nets (any)/ITNs/LLINs; and (3) households that received IRS in previous 12 months; d) household with sufficient ITNs and/or IRS in previous 12 months.
According to Monroe et al [34], behavior is defined as "The observable response of a man or animal to a situation, " and the term is used broadly in this review to encompass human activities, location, and sleeping patterns. This includes activities occurring within or nearby the home, within the community, or outside of the community. We reported these activities from various publications that reported human behaviors in relation to malaria exposure. Specifically, studies included malariaendemic settings in the 2 Asia-Pacific ecoregions and a description of human behaviors occurring when malaria transmission can occur, that is, when malaria transmitting vectors are active.

Search Results
The search results are illustrated in Figure 2. Initially, 3367 records were identified through database searches and 38 additional records were identified through other sources (eg, "gray" literature-unpublished reports, reference lists). After screening of title and abstracts, the remaining 103 papers were assessed and reviewed in full, after which 66 articles were included ( Figure 2). The most common reason for exclusion in the final stage was the lack of a direct relation with or adherence to the RMT concept and an applicable result in malaria elimination settings.

Study Characteristics
Selected studies were published between 1995 and 2020, proportionally representing Cambodia (15%), Papua New Guinea (14%), Vietnam (14%), Myanmar (12%), Malaysia (9%), China (9%), Solomon Islands (8%), Lao (6%), Thailand (6%), Indonesia (2%), and multiple Asia-Pacific countries (6%). The included studies fit into ≥1 of the 4 criteria: (1) ownership and usage of ITN/LLINs, (2) IRS application, (3) insecticide susceptibility of primary and secondary vectors, and (4) human and vector behavior observations focused on nighttime human activities. The selected articles emphasized the significance of epidemiological, entomological, sociodemobehavioral characteristics and drivers of RMT, and observations of insecticide susceptibility of Anopheles mosquitoes. Several review article on outdoor malaria transmission or RMT were used to complement the review [7,31,[35][36][37][38].   Table 1 provides an overview of 14 studies conducted in the GMS and Asia-Pacific region. Given the study objectives, each example provides some recommended "policy" considerations together with challenges and opportunities. A key theme is the continuing burden and challenges that RMT poses in control areas, regardless if in an elimination phase or not. Mosquito and human behavioral patterns are major contributing factors for sustaining RMT. Across this wide geographic coverage is the inherent spatial and temporal heterogeneity in malaria epidemiology in which human and vector interactions occur. A better understanding of this complex relationship will be needed in order to develop and adopt better tools to combat outdoor transmission not affected by ITN use (or IRS) while promoting other bite prevention tools for those exposed.

Access to, Preference for, and Use of Vector Control Tools
Because access to ITNs is the primary driver of their use [54][55][56][57], we find variations in the reporting and sources of ITN access and use, and IRS coverage. This review did not include other indicators associated with performance of vector control interventions, for example, ITN/LLIN durability, IRS residual efficacy, Larval Source Management effectiveness, intervention coverage, preference, and populations having access to or use of health services. Several studies have found that individuals and household use untreated nets because of their perception that LLINs are made of coarse fabric, are too small, have easyto-break mesh holes that are too large, wrinkle/shrink after washing, and have reduced efficacy over time [39,[58][59][60].
Among the 14 studies, 3 (in Cambodia, Papua New Guinea, Sabah in east Malaysia) use a combination of field research and programmatic surveys, while 11 (78.7%) relied on field research surveys. Eleven studies reported ITN ownership (numbers of nets distributed, persons per net, households with nets, or nets per person), and 3 studies (in Thailand, Vietnam, and Papua New Guinea) reported on population access to ITN in households (defined as the percentage of the de facto household population who could sleep under an ITN if each ITN in the household was used by up to 2 people) (see Table 2).
While the Cambodia Malaria Surveys (  [13]. These outcomes were probably attributed to repeated free ITN distribution campaigns in Cambodia and Papua New Guinea [86] and across Asia-Pacific [36], which led to significant increases in mosquito net ownership and useaccess ratios with few exceptions (Table 2), in combination with increased surveillance and testing, treating, and tracking over the last 5 years. In addition, strong commitment from policy makers and effective partnerships have created a catalytic effect on program success, resulting in 75% and 93% drops, respectively, in reported malaria cases and malaria deaths in GMS countries between 2012 and 2017 [87]. In view of case increase (resurgence and outbreaks) due to drug stock-outs, low use of mosquito nets, disruption to the village malaria workers program, and high influx of people into forest areas, where access to treatment is limited, a combination of approaches is needed to engage these hard-to-reach populations [38,88,89].
Before ITN rollout and mass distribution campaigns during the early 2000s in GMS countries, malaria transmission hot spots occurred within the village in southeastern Thailand, in groves of rubber and fruit trees where A. dirus sensu lato (s.l.) is the primary vector [90], and villages near the Thai-Myanmar border in Tak province [91,92]. In the southeastern region [93], it was reported that transmission probably occurred in forest and rubber plantations rather than rice fields. In the latter region (Mae Hong Son district), entomological and epidemiological data suggest that residential villages, farm hut settings, and forests were transmission sites, but the status of forest foci was inconclusive [94].
For a long time, the consensus has been that mature rubber plantations, especially those next to the forest, provided a suitable habitat for A. dirus [95]. Results of unpublished parasite prevalence surveys conducted sporadically over the last 20 years in Cambodia [96][97][98][99], however, suggest that there is very little transmission in most plantations. Informal field observation by malaria veterans from Thailand and Vietnam indicates that the vast majority of cases among plantation workers are contracted in nearby forests where the workers go to collect forest products to supplement their income (see below). There has actually been an increase in the number of cases among some rubber plantation workers in recent years, but this has coincided with a global reduction in rubber prices, which has affected wages, forcing workers to find additional sources of income-forest products (Sean Hewitt, unpublished data). During the last 10 years, large-scale distribution and rollout of ITNs have provided effective community protection in residential villages that are accessible by central or district teams; however, the current tools are not sufficient because of human or mosquito behavior, and thus malaria transmission in Thailand, Papua New Guinea, and Solomon Islands is not residual according to the current definition, or is probably misclassified.
In situations where malaria transmission can be considered residual, because of the very good coverage of ITNs, as in S118 • jid 2021:223 (Suppl 2) • Hii et al  The ownership of nets was reported without indication of age, condition or whether they were insecticide treated or not  [100] or with both primary and secondary vectors [44]. In such places, the deployment of ITNs needs new approaches, but other vector control tools are also needed. Data on subgroups such as wealth quintile or urban/rural residence, which were not available for this review, may offer ways to identify target groups that do not use their available nets to the fullest degree. Nine of 14 studies (64.2%) reported ITN use, and 6 (42.8%) reported IRS coverage. For example, the 2013-2014 ITN distribution report from Kudat District, Sabah (target of "universal coverage" if the household owned ≥1 net for every 2 people) showed that 325 nets were distributed to 537 household members in 3 villages (Paradason, Limbuak Laut, and Timbang Dayang), giving an ITN access of 121% (650 of 537) which is "too many" because the household owned ≥1 net for every household member (H. Tanrang, personal communication, May 21, 2020). Interestingly, mosquito bed net use did not seem to be protective against P. knowlesi acquisition [30], consistent with findings from the same study area describing earlier peak biting times of A. balabacensis in the early evening (from 1800 to 2000 hours) as mosquitoes adapt to human bed net use later in the night, with biting mainly occurring outdoors [101]. However, given that lower levels of mosquito biting have been reported to continue throughout the night [101], the potential of indoor transmission remains. Consequently, the use of conventional prevention activities supported by social behavior change and communication efforts remains relevant, especially for the large proportion of people who did not use a bed net during travel away from home.
Three countries (Thailand, Vietnam, and Papua New Guinea) reported an ITN/LLIN use-access ratio, which provides an estimate of the proportion of the population using nets, among those who have access to a net within their household. This indicator clarifies whether a gap in net use is related to behavior or to lack of access to nets.
Given the apparent data deficiencies, it is important for National Malaria Control Programs (NMCPs) to monitor intrahousehold access to ITNs/LLINs and indicate whether gaps in protection are due to a lack of access to nets or to human behavior (Peeters Grietens et al 2019, [102]) Free access to LLINs is the primary determinant of participation in bed net distribution campaigns. Preference for larger nets and darker-colored polyester nets over polyethylene nets, owing to wrinkling and shrinking of the latter, has been noted in Solomon Islands [103], while a similar study in Vanuatu found preferences for larger mesh and wider nets [104]. Participants in Timor-Leste did not have enough experience with nets to make preferences revealing, because different brands of LLINs have the same color [14]. The shape of nets can influence their use. In Sri Lanka, the odds of LLIN use were 5.6 times higher for conical nets compared with rectangular nets among 530 LLIN-owning households. The preference vote was evenly split between brands (Yorkool and Olyset), with 505 conical and 155 rectangular nets selected. Respondents cited heat or lack of mosquitoes as the main reason for not using a net [105].
A recent qualitative study among end users in Cambodia showed clear partiality for softer fabrics (in this case polyester over polyethylene) [60]. Because users have access to a variety of bed net options, either from local markets or via free distribution programs, they have come to understand that mass distribution allows them to create whatever value they wish from the free nets. Thus, various alternative uses for distributed net can result, even far from their intended purpose of preventing mosquito-borne pathogens (eg, fishing, protecting fruit and vegetable crops from animal damage). Despite widespread distribution both from local markets to community-wide distribution programs, the product attributes of LLINs confine them to a narrow spectrum of preferred use for prevention of mosquito bites: (1) single-occupant users, often men; (2) use in remote settings, such as forest, mountains, and plantations. where chemical protection is highly valued against a perceived increase in mosquito threats; and (3) use in settings where versatility and durability is important. Whether polyethylene or polyester fabric, nets frequently targeted for household use generally lack the favorable product attributes necessary to provide a more comfortable experience users seek, as recorded in the in Solomon Islands [106], Cambodia [39,58], Vanuatu [104], and Timor-Leste [14]. When these product attributes are coupled with a lack of user education and distribution strategies leading to oversupply, users find new ways of assigning value to the LLINs-often leading to premature abandonment or intentional misuse (Lucas Nene, personal communication, July 23, 2020).
Programs that wish to explore the relationship between various net attributes, preferences, and use rates should include questions from a recent literature review [107] and the "Malaria Matchbox" tool [108]. Survey questions can be complemented with well-designed, site-specific qualitative research for assessing existing products and services under consideration, current usage behavior, and the cultural context behind these behaviors. Irrespective of sample sizes, these more detailed studies can help reveal blind spots for improving future survey designs (Lucas Nene, personal communication, Jul 23, 2020). Such qualitative data assisted Cambodia to receive an exceptional waiver from GFATM to purchase the most preferred products by the community, with future quantitative studies on usage and acceptance to be based on net characteristics that have been suggested to enhance the usage among high-risk groups [59]. The relatively low IRS coverage (42.8%) in our review was not far removed from that reported in the Asia-Pacific region (proportion of structures or household sprayed, 67%; proportion of the population at risk protected by IRS, 56%) [36]. Given the relevance of this indicator in foci investigations, it is recommended that the NMCP should monitor IRS performance to ensure spraying top-ups/mop-ups during each spray round.

Human and Vector Interaction
Human behavior is a central component of RMT owing to the overlap with vector behavior to allow better targeting of vector control to the human-vector contact point, especially where transmission occurs away from home (eg, the farm hut, forest, and forest fringe [39,44,45,70,71,109,110]). Among the 14 studies, 5 (in Cambodia, Lao PDR, Thailand, and Vietnam) used a combination of entomology and mixed methods (eg, in-depth ethnography, rapid participatory assessment, qualitative and quantitative, observations, and transect walks); 4 (in Lao PDR, Thailand, Myanmar, and Vietnam) used cross-sectional household interviews and entomology; 1 (in Solomon Islands) used movement diaries, observations, and entomology; 1 (in Papua New Guinea) used observations, household interviews, census of humans and animals, and entomology; 1 (in Sabah and Malaysia) used household surveys and pretested questionnaires in a case-control study of P. knowlesi; and 2 (in Vietnam) used entomology (see Table 2). Twelve studies (92.8%) conducted entomological surveys using all-night outdoor and/ or indoor human landing collections, animal-baited and exit traps knockdown collections, or light traps; only 1 study (Lao PDR) conducted 24-hour mosquito collections using a human double-net trap.
Ten studies integrated human behavioral and entomological data to provide a quantitative estimate of human-vector interaction occurring indoors and outdoors (Table 2). Of these, 4 studies used early human-biting proportion calculated as the percentage of vectors biting before 2200 hours (or the locally appropriate sleeping time) by vector species (1 for Cambodia and 3 for Vietnam) and the entomological inoculation rate for the respective parasite (P. falciparum or P. vivax) (2 in Vietnam). Four studies(3 in GMS and 1 in Solomon Islands) integrated human and vector data estimates of indoor and outdoor vector biting as well as the distribution of people indoors and outdoors for each hour of the night to produce a weighted estimate of exposure occurring indoors and outdoors. With the exception of Lao PDR, which used assumptions of community sleeping time and outdoor activity, this analytical approach was successfully used to quantify human-vector interaction by Killeen et al in rural Tanzania [111] for Myanmar, Thailand, and Solomon islands. Interestingly, the unusual late nocturnal, indoor-feeding behavior of A. dirus s.l. [8] is due to premass distribution of ITN/LLIN, which in the prolonged presence of impregnated nets, contributed to 24.4% and 26% of earlier outdoor biting in forested villages of Lao PDR (Vythilingam et al 2005, [112]) and Cambodia (Durnez et al 2013, [40]), respectively before sleeping time. Finally, 2 of the 10 studies documented odds of exposure from sleeping outdoors in the forest or plantation (Sabah) and habitat types (Lao PDR).
Two studies (in Thailand and Vietnam) used entomology, observations, and/or interviews to determine location-specific hot spots. One study reported the effects of pre-ITN and post-ITN distribution on human biting rates (Papua New Guinea), and 1 entomological study (Vietnam) did not quantify human exposure to vectors.
Although IRS increased the outdoor biting rate of A. dirus s.l. [113,114], and Anopheles minimus s.l. in forested and foothill regions in Thailand [113,115], A. dirus displayed avoidance behavior by not entering sprayed houses and resting on sprayed surfaces [114]. Twice as many A. dirus exited via window traps after spraying compared with DDT prespray structures [113], owing to the pronounced excitatory-repellent effects of DDT, and resulted in an overall greater reduction in human landing activity [116]. If mosquitoes are killed quickly, they may not have the time or ability to exit, so high mortality rates may result in apparent reduced exiting time [117]. DDT reduced the likelihood of attempted mosquito blood feeding by more than half compared with mosquitoes in the presence of deltamethrin, by causing a greater increase in the rate of mosquitoes exiting experimental huts or local houses (ie, owing to excitorepellency action) [116]. However, this behavioral avoidance action did not translate into an equivalent elevated risk in malaria transmission for A. minimus s.l. and A. dirus, both outdoor biting species. [113,118], in forest fringe and rubber plantations, respectively.
In the absence of knockdown or mortality data of mosquitoes entering experimental huts or houses, it is speculated that a higher survival of exiting mosquitoes that evaded contact with treated surfaces may sustain persistent transmission outdoors. Very few studies (in Thailand, Vietnam, Lao PDR, and Solomon Islands) recorded human behavior during human landing collection of the number of people present or actively using interventions at the collection site (either outdoors or indoors). In forested areas in the GMS and coastal islands in the southwest Pacific, measurements were made during the same periods and locations as vector biting with concurrent observations on hourly human activity, movement and use of ITN. Analysis of ITN use every hour and/or IRS status coupled with vector behavior observations and/or insecticide resistance status are useful to identify gaps in protection [119].
These 4 studies provide adjusted human biting rate (calculated as the product of human biting rate by proportion of humans observed inside vs outside, awake vs asleep, with vs without a ITN), which is useful to analyze human behavior together with vector behavior and use of vector control interventions [119].
For example, for an unprotected individual, comparing the proportion of vector bites occurring indoors with proportion of bites occurring outside provides an idea of relative exposure risk and is useful for characterizing residual transmission in a programmatic context.
Similarly, very few studies (see above) recorded human behaviors and activities by location, resulting in the lack of "gap" in protection, not only before sleeping time, but also for people who remain outdoors during the night (Figure 3). Quantifying and characterizing gaps in personal protection against mosquitoes, defined as the proportional reduction in biting exposure an individual experiences as a direct result of personal using a protection measure, requires information on the behaviors of vectors and humans, as well as when and where they intersect. Vector-human interaction is useful for determining how this gap needs to be tackled by additional vector control measures and an integrated perspective on relevant indicators of humanvector interactions.
By applying Global Positioning System trackers (Thailand, Vietnam) or movement diaries (Solomon Islands) across time and relevant geographic areas to track population movement (eg, sleeping in villages vs sleeping at farms), it is possible to specify the exact location of transmission hot spots. Reports from Institute of Pasteur Cambodia of 20% daytime biting by Anopheles females during 24-hour static landing catches have implications for the protection of workers and among those moving from place to place, foraging for forest products between dawn and dusk. Similarly, reports from Lao PDR suggest that a recent outbreak in Nong District in Savannakhet Province, which affected males and females from all age groups, was the result of entire families foraging in the forest during the daytime for scarce but high value medicinal leaves to sell to Vietnamese buyers (Sean Hewitt and Bouasy Hongvanthong, unpublished information).
In summary, the deficiencies identified in this review that will improve the characterization of RMT are (1) sleeping or awake time (proportion of individuals asleep vs awake, indoors vs outdoors by hour during biting times) and (2) adjusted human biting rate (calculated as human biting rate times the proportion of humans observed inside vs outside, awake vs asleep, with or without a ITN/LLIN). Given that approximately 20% of Anopheles vectors were caught during the daytime in forested areas of Cambodia (Amelie Vantaux, personal communication, April 20, 2020), consideration should be given to daytime biting by opportunistic vectors, especially by moving landing catches that aim to replicate foraging behavior [15].

Nighttime Activities, Personal Protection, and Transmission Risk Factors
Among 10 studies that covered some description of daytime and nighttime activities occurring during times when local malaria Islands [2 studies]) gave varying detail regarding locations of people and/or activities taking place in the peridomestic setting (inside and directly outside the home), as well as away from home, throughout the night. This included routine household chores and entertainment occurring in the evening hours before bed, routine livelihood activities that lasted throughout the night or part of the night, such as hunting, forest work, gathering forest products, logging, rubber tapping, supervision of local agroindustry, security, and sociocultural events (eg, funerals that lasted throughout the night) ( Table 3). Methods used in these 10 studies document and characterize human behavior included participant observations, in-depth interviews, rapid participatory rural appraisals, structured observation surveys, questionnaires, focus group discussions, mixed methods, transect walks, triangulation of qualitative and quantitative data, daily movement diaries, and Global Positioning System trackers. Some studies often looked at specific nighttime or daytime activities, as well as the impact of these activities on use of malaria prevention tools. Risk factors can include occupational exposures and other behaviors outside of households (eg, forest-going, farming, and cooking). Mobility of individuals and/or population groups may greatly vary (daily, weekly, seasonally), which in turn may affect the impact and effectiveness of vector control interventions, and consequently malaria risk.
Given the paucity of literature on observations of nighttime activities, we provided some informal field observations by malaria veterans below. Informal field observation 1 is from a Khmer refugee camp, Thailand-Cambodia border (1980)(1981): … when I worked in the "Kampot" Khmer refugee camp at Pong Nam Ron, right across the border from Pailin, 1980-81, I was hired to control a malaria outbreak in the camp, 17 000 souls, by setting up ULV sprayer coverage, which I did. But I also did some epidemiology and found that the assumption of domestic transmission in the camp was wrong. First of all, most of the cases were adult men, even though everyone slept in bednets in 3-sided houses, went to bed at 9pm (curfew) and got up at the same time in the morning. Second, the hypothesized vector was dirus, flying into the camp from a nearby forest, however, there were no identifiable breeding sites, and nighttime collections caught no dirus. Third, and definitive, when I went out to find and followup on patients for recrudescence, they could not be found, but/and when my Khmer microscopist when out, he reported that the patients (mostly adult male) were not in the camp, which directly contradicted the firm assertions by camp officials that no one left the camp, due to barbed wire fences and armed Thai soldiers. The Khmer left the camp at night, by truck, to work in Thai fruit orchards, and came back to camp when they got malaria. Another example of non-traditional transmission, ie, not in the home. This was a blind spot for all of us for many years. I could have had a rich and full career talking about mobile and migrant malaria, starting in the 1980s, if I had recognized it at the time (Steven Bjorge, personal communication, July 15, 2020).
Informal field observation 2 was from Hilltribe Village in Mae Hong Son (2000): Malaria transmission was occurring among school children staying in a boarding school in a forested community. It was determined that the infected children had been regularly watching TV soap operas in the (unscreened) library after dinner from 8-9 p.m., prior to sleeping in a screened dormitory (Jim Hopkins, personal communication, July 15, 2020).

Informal field observation 3 was from Karen Hilltribe Village in Tha Song Yang, Tak Province (2005):
As a member of the Global Fund Round 2 Malaria Evaluation Team, I visited a village of bamboo houses with thatched leaf roofs on the Thailand-Myanmar border with the highest incidence of MDR falciparum malaria in Thailand. ITN coverage was more than 100% and the village had a full-time malaria volunteer post with RDTs and radical treatment drugs. Each house had a bamboo pole antenna, a solar panel, and a car battery connected to a television. A quick survey indicated that villagers used the solar-powered battery to watch television for 2-3 hours from 6-9 p.m. through the prime time soap opera, prior to setting up their ITNs for sleeping (Jim Hopkins, personal communication, July 15, 2020).

Informal field observation 4 was from market places in Southern Palawan (2019):
Southern Palawan with hilly and forested terrain is the last hold-out of endemic malaria in the Philippines. Transmission persists despite LLINs, IRS and curative services. As elsewhere in Southeast Asia, this has been ascribed to deep forest farming from unprotected dwellings, and additional LLINs have been distributed for those situations. It is increasingly accepted that there is also some malaria risk at marketplaces, where whole families often stay overnight, the children enjoying TV shows and videogames thanks to generators. Some representatives of indigenous people's organizations have opined that the malaria risk at those market-places could be higher than in the home or at the deep forest plot-huts. One barangay Residual Malaria Transmission in Asia-Pacific Region • jid 2021:223 (Suppl 2) • S127

Table 3. Continued
Residual Malaria Transmission in Asia-Pacific Region • jid 2021:223 (Suppl 2) • S129 captain (elected community leader) had just decided, last time I was there, in November 2019, to close the market at 6 pm to prevent malaria. I pointed out that people would then be exposed when walking home, and advised to close at 4 pm or not close but put emphasis on the giant collective insecticide-treated nets, which were being made available for those settings by the national malaria programme and its partners (Allan Schapira, personal communication, Jul 17, 2020).
Informal field observation 5 was from touch points in forested areas of Pursat, Cambodia (2017): As many people who live in or near the forest do not have access to electricity, they often gather together with family or friends in the evening. There are no days off, so each day is similar, with people often being physically exhausted after working all day in the farm or forest. They often want to just chat with friends over dinner, and live or visit in structures which are not closed, do not sit inside nets while eating dinner or visiting friends, and do not wear long sleeves due to the heat. They sometimes gather at a nearby house or village if someone has a generator to watch TV, however listening to the radio is the most common activity especially among the older generations. The most popular and available vector control products (other than ITNs) are mosquito coils and insecticidal sprays, although they often do not have enough money to use them all the time (John Hustedt, personal communication, July 21, 2020).

Review and Characteristics of Nighttime Activities
We included 10 studies (with multiple authors from companion articles) of some descriptions of evening activities occurring during times when local malaria vectors are active (Table 3). These studies identified activities taking place in the peridomestic setting, which are quite similar to those in the study by Monroe et al [34]: for example, activities "inside and directly outside of the home as well as away from home throughout or part of the night. " Similar patterns regarding "routine household chores and entertainment occurring in the evening hours before bed, routine livelihood activities that lasted throughout the night such as security and fishing, and large-scale sociocultural events, such as funerals that might last throughout the night" [34] were also seen in the Asia-Pacific region.

Influence of Insecticides and Behavior of Vectors
One of the major threats for control of RMT is the development and spread of insecticide resistance (IR) in vector populations expressed as either physiological or behavioral resistance [128][129][130]. Care should be exercised not to overstate the risk of insecticide resistance in forest settings where selection pressure is almost entirely absent. Many NMCPs are doing active monitoring for insecticide phenotypes to insecticides used in control programs, usually through limited sentinel site surveillance rather than in response to increased malaria burden, so it is unlikely that resistance will be detected at an early stage. Consequently, funding partners with limited understanding of entomological issues (and understandably) tend to fund what they know: "vector bionomics and insecticide resistance in sentinel sites, " rather than what would be most useful to the program in a resource limited environment [59]. In addition, data to determine actual mechanisms (ie, metabolic and target-site) responsible for resistance are rarely collected with only 14% of Asia-Pacific countries evaluating insecticide resistance mechanisms; [36]. Because this is an essential prerequisite for maintaining intervention effectiveness by enabling more proactive decision making management for IR management [131], and despite "a high prevalence of countries monitoring IR phenotypes, not all countries monitoring IR have strategic frameworks for responding to the data (eg, IR management plans) and among those with a management plan, not all report using their data as a basis for maintaining or selecting alternative insecticides for programme use" [36].
The presence of insecticides can not only act to suppress susceptible vector populations as toxicants but can also modify normal behavioral responses, such as host seeking and biting activity regardless of susceptibility status (ie, physiological resistance). So-called behavioral resistance, as reported in the literature [50,78,[132][133][134], is likely more a reflection of a stimulus-response (eg, avoidance or deterrence) to the presence of a chemical operating as contact locomotor excitation agent and/or as spatial repellent [135][136][137][138]. In particular, DDT and many pyrethroid class chemicals used to treat bed nets and IRS possess such behavioral excitorepellency responses. In some instances, they has been reported to have "changed" (or modified) the temporal and spatial blood feeding behavior of the vector population, typically a "shift" to earlier evening biting hours and a higher proportion of the population feeding outdoors away from residual insecticides present indoors (ie, sprayed walls or ITNs). This is the case for Anopheles farauti in Papua New Guinea and Solomon Islands and A. dirus in Thailand, which impose an enormous challenge for malaria control owing to behavioral avoidance of insecticide-treated surfaces (Table 4). Interestingly, the widespread use of IRS begun in the 1960s has nearly eliminated Anopheles koliensis and A. punctulatus, species that are mainly endophagic, in the Solomon Islands [133], with replacement by Anopheles hinesorum, a relatively poor vector, if not nonvector [139], and A. farauti, a more exophagic and earlybiting mosquito, often present in high densities [7,133].
Currently, the primary malaria vectors in Papua New Guinea appear to remain susceptible to pyrethroids [56,57]; however, despite mass distribution of ITNs throughout malaria-endemic areas of the country, transmission continues to persist at high levels. Entomological studies conducted by NMCP in partnership with Rotary Against Malaria and the Papua New Guinea Institute of Medical Research have revealed that very low bioefficacy of LLINs procured from 2014 onward [110] may have been among the contributory factors in the malaria resurgence, including poor adherence to case management [152]. In addition to outdoor blood feeding, persistent transmission may be due, in part, to the more generalist, broader range of host selection by the primary vector species. As a species group, the Punctulatus assemblage is typically opportunistic in feeding (varying anthropophilic to zoophilic), based on host availability and ITN usage [134]. Although vectors readily feed on humans, opportunistic host selection, partly a function of relative host availability and a propensity for using alternative animal hosts (eg, domesticated pigs, dogs) has been well described [153,154].
In Papua New Guinea, it was reported that hosts were not selected in proportion to their abundance, but rather were either underselected or overselected by the mosquitoes [133]. Four species, A. farauti s.s., A. punctulatus s.s., A. farauti no. 4 and Anopheles longirostris, overselected humans in villages with low LLIN usage but overselected pigs in villages with high LLIN usage. A. koliensis consistently overselected humans despite high LLIN usage, and Anopheles bancroftii overselected swine [133]. A lack of selectivity for human blood allows these vectors to escape exposure or greatly minimize contact with indoor treated surfaces or nets entirely, thus precluding sufficient selection pressure promoting resistance. A recent review concluded that zoophagic mosquitoes "with moderate vectorial capacity often respond poorly to LLIN or IRS interventions because the technologies are designed to target the stereotypical behaviors of the smaller number of more potent human-specialized species that mediate most, but by no means all, of the global malaria burden" [8]. Killing or reducing mean survival in these mosquitoes with livestock-based interventions, such as topical or systemic insecticides or endectocides ,could play a complementary role in reducing RMT [8,155,156].
The Punctulatus group vector species commonly exist in sympatry throughout most of the malaria-endemic areas of Papua New Guinea [22,24,40,41,44,[55][56][57]. While ITNs may affect the anthropophilic species (eg, A. koliensis), transmission is still sustained by the more opportunistic and behaviorally "plastic" species that will feed on other animals as availability permits. This condition, when combined with increased outdoor and early-evening biting observed in some vector population of Papua New Guinea ( [78,134]), presents a challenge to the ITN program in Papua New Guinea, as well as the rest of the Pacific region, where the more opportunistic species A. farauti s.s. and A. punctulatus are the primary regional vectors. By maintaining high population size, the proportion that feed on humans outdoors can sustain residual transmission despite high ITN usage in the village [157].
In central Vietnam, the A. minimus population has virtually disappeared after the introduction of ITN [158] and remained absent for the next 18 years (Ron Marchand, unpublished data). In Assam, northeastern India, A. minimus mosquitoes were not seen resting inside human dwellings after an initial 3 years of continuous ITN distribution [160]. The ITN-based intervention not only deterred entry of A. minimus species, but also served as personal guard against infective mosquito bites corroborated by data on human mosquito landing catches and declining trends of malaria transmission [161]. The use of public health insecticides in Nepal eliminated A. minimus s.l. [162,163] and significantly reduced populations in the Thailand peninsula and central plains, although they did remain abundant in hilly forested areas [164,165]. In malaria-endemic areas of China, ITN use resulted in a higher decrease in the endophilic and anthropophilic Anopheles lesteri (Anopheles anthropophagus) [166] and A. minimus s.l. [167] relative to more exophagic and zoophilic Anopheles sinensis, a species that also shows high levels of multiple insecticide resistance. (Table 4).
Vectors can have a repertoire of behavioral actions to external stimuli that can effectively avoid contact with insecticides [130,168], which may be a result of innate, preexisting variability (plasticity or resilience), resulting from an instantaneous stimulus-response in a species or particular population or possibly the genetic selection over time of a population to avoid the real or potential presence of inimical chemicals in the environment that expresses an evolutionary advantage ("resistance"). For example, behavioral resistance can be expressed as changes in either the location where vectors seek blood meals (ie, shifting from indoor to outdoor biting) or the time when blood meals are taken (ie, feeding earlier in the evening or in the early morning hours when people are outside their houses and not protected by ITNs or IRS) [36]. Examples of observed shifts in species population behavior (eg, the near-elimination of A. koliensis and A. farauti in Solomon Islands; decrease of A. dirus s.l. relative to A. minimus s.l.; species shift from A. minimus to Anopheles harrisoni in Thailand; rebound of Anopheles fluviatilis s.l. and Anopheles maculatus s.l. after IRS in Nepal), shifts to outdoor biting or early biting periods (eg, A. farauti and A. koliensis in Papua New Guinea and Solomon Islands; A. dirus and A. minimus in Thailand and Vietnam), shifts to zoophily (A. farauti in Papua New Guinea, A. minimus in Thailand, Anopheles culicifacies in India) or to exophily (A. farauti in Solomon Islands, A. culicifacies in India) are reported in the Asia-Pacific region associated with the use of routine ITNs and IRS [7].
Given the significant limitations in the scope and methods of historical literature and sparse data in response to LLINs and IRS, it is suggested that findings accurately describe behaviorally resilient rather than resistant mosquito taxa, which have always exhibited evasive traits [138]. Because the precise mechanisms driving these genetical shifts have not yet been fully elucidated, the vast majority of behavioral response variation to stimuli and conditions is due to inherent "plasticity" in the mosquito. Thus, it is inappropriate to infer selection of new behavior patterns when mosquitoes show an inherent plasticity in feeding after being frustrated in accessing their hosts [170]. More research on methods, the natural heterogeneity of vector populations, and the role of heritable traits in behavioral resistance are needed to understand better the spectrum of changes induced by intensive insecticide use for future policy discussions [170]. Because "[m]ost NMCPs are unlikely to detect the emergence of behavioral resistance with their existing vector surveillance programs as only 31% of countries monitor the indoor-outdoor biting ratio with 34% tracking changes in peak biting time, the 2 most common expressions of behavior resistance" [36], it is important to know which humans to target and when and where to target humans who are exposed to mosquito bites. For example, survey data on human behavior and high-risk populations can be analyzed together with data on vector bionomics and intervention efficacy, which may help determine gaps in protection and local drivers of transmission, including drivers of residual transmission. While there is a growing research agenda on this topic [44,45,171] program-oriented methods are currently available for national malaria programs to consider using [119]. Integrating vector behavior and insecticide resistance data with human behavioral observations will demonstrate where and when people are exposed to mosquito bites, as well as potential gaps in protection indicating that supplemental tools may be needed. The effects of insecticide resistance and behavioral responses of vectors on malaria transmission can vary, depending on a range of different epidemiological factors, being species and location specific [172]. Regardless of the mechanism involved, movement of vector populations to early biting times, when more people are active outdoors, is an important consideration for applying control measures and an obstacle for achieving elimination.
Given the complete susceptibility (100% mortality rate) of fieldcollected populations of A. harrisoni (Minimus complex species) and A. dirus.to transfluthrin in Thailand, [173], there is potential for deploying highly volatile pyrethroids in situations where conventional control methods are not readily accessible and where repellent devices may complement other currently deployed methods of protection (eg, topical repellents, LLIN, and IRS) [174,175]. Sangoro et al [175] demonstrated that using transfluthrin-treated chairs and ribbons reduced outdoor-biting malaria vectors in peridomestic spaces, and also elicited significant mortality among pyrethroidresistant field-caught malaria vectors. The use sandals treated with 0.05 g of transfluthrin sandals reduced exposure to Anopheles gambiae s.l. landings by at least 50% and 40% against Culex mosquitoes in the field. From the public health impact, the combined use of metofluthrin-impregnated spatial repellent devices and LLINs (Olyset Plus) reduced the infection rate in children as well as the number of pyrethroid-resistant vector mosquitoes in malariaendemic villages in southeastern Malawi [176].

DISCUSSION
The current review identified several areas of importance related to RMT. The first highlights the inefficiency of current  measurement tools, which produce data often confounded by human and/or mosquito behavior and results in a misclassification of "residual" transmission. In situations where malaria is considered residual, as in Vietnam and Cambodia, primarily because of the very high population coverage of ITNs, the persistence of malaria transmission has moved from villages to farm plots and forested areas, often involving primary and/or secondary vector species [44,45,100]. The lack of high-quality coverage data on ITN access and use at subnational levels in many countries presents a significant challenge and hampers a full understanding of RMT dynamics. Where such data exists at the district level or at lower administrative levels (eg, the health facility or village level), not only does the deployment of ITNs require new approaches, but other vector control tools are also needed. The second area relates to when (time of day) and where (indoors vs outdoors) people are exposed to malaria vectors, which was evaluated in just 5 studies (in Thailand, Vietnam, Cambodia, Solomon Islands, and Papua New Guinea) [42,50,53,121,126]. The third aspect is night activities that may increase people's contact with malaria vectors. Where data is available, a good understanding of human behavior is crucial for targeting contextappropriate vector control interventions across specific settings. Given the inherent heterogeneity of malaria transmission across different landscapes, it is imperative that data and information be "local" to identify the relative importance of specific activity categories and target groups based on the entomological, human behavioral, and epidemiological context. To do so requires a highly coordinated, interdisciplinary approach involving input from NMCP managers, social scientists, and public health entomologists to help prioritize and interpret data on the access to and use of ITNs and to clarify the gaps in protection.
The fourth aspect is gender-related vulnerabilities and barriers for accessing malaria services that fuel persistent malaria transmission from cross-border or intraborder mobility of illegal migrants and indigenous populations. The available data suggest that outside forest communities, where entire families resides inside the forest, it is predominately younger males who are more frequently accessing the forest [42,53,177]. The acceptability and use of ITNs are strongly linked to cultural norms of sleeping patterns, in which sex and age play important roles. In Shan state, Myanmar, young children of ethnic groups and Chinese inhabitants shared nets with adults, presumably with their mothers, and are therefore protected by their mothers' ITNs, if available [178]. However, a follow-up study in 2016 covering 32 052 people, including 1514 forest-goers from every state and region (except Chin State) in Myanmar, showed that age and sex had little effect on net or ITN/ LLIN use [59].
Another study highlighted poor ownership and use of ITNs among migrants in the Regional Artemisinin Initiative project areas of Myanmar, including identified barriers to their ownership and use [179]. In mapping the population, there is a need to consider sex-related behavioral differences, especially for the reproductive age group, because sharing and sleeping under a single net together is inappropriate [179]. For instance, this would apply for implementing behavior change interventions in migrant plantation workers [180]. An example from Nigeria showing sex disparity in ITN use, with males less likely to use ITNs (particularly in those aged 15-25 years), illustrates the need for age-sensitive and gender-sensitive messaging during a universal distribution campaign to ensure that males benefit equally from such communications and activities [181], which may be applicable in the Asia-Pacific region [182].
It is essential that social and gender determinants of ITN ownership and usage be explored in high-risk communities, and health communication messages should stress the need for everyone entering the forest or in nearby areas with malaria transmission to sleep under ITNs. The Malaria Matchbox [108] is a useful tool for program managers and implementers to identify barriers that people encounter in trying to access and use healthcare services, particularly those related to malaria prevention and care. Identifying those barriers, whether sociocultural, financial, physical, or related to gender norms, is an essential step to match people's specific needs to responses that are person centered, rights based, and gender responsive [108].
In the current review, various activity categories ( Table 3) are provided that serve as a useful framework for informing context-specific research on the relative importance of these activities that can drive locally appropriate interventions. Incorporating this framework and appropriate methods with the Entomological Surveillance Planning Tool (ESPT) [117] is the next step to develop standardized approaches to guide programmatically relevant data collection on nighttime and daytime activities and categories to interpret ITN use and access information [119]. This review is limited by the possibility that studies that would have met inclusion criteria were not identified in the review process. Particularly, it did not consider factors (or possible "drivers" of transmission) that could influence transmission dynamics, such as changes in malaria receptivity and/or importation risk (ie, vulnerability), including new land development, population movement associated with seasonal agricultural pursuits, periodic political, cultural or religious events, civil unrest, internally displaced populations, environmental parameters such as mean and/or total rainfall by location, and accessibility to malaria diagnosis and treatment. For example, from 2006 to 2016 in Cambodia, the number of confirmed malaria cases decreased 84% from 143 758 to 23 492 [177], while from 2000 to 2014 the amount of dense forest decreased from 35% of land cover to 16% [183].
Geographers have noted that the reduction in forest is directly related to the surge in economic land concessions, and that the annual rate of forest loss was between 29% and 105% higher in land concessions than in comparable land areas outside concessions [184]. This probably results in lower species heterogeneity in rubber monoculture plantations than in secondary forests, which provide a higher diversity of aquatic habitats and consequently a higher heterogeneity of mosquito species [185][186][187]. While all important topics to consider, they were outside the scope of this focused objective. However, with exclusion of population movement as a search term and inclusion criterion, it is possible that relevant articles could have been missed. Nonetheless, a comprehensive and structured process was used.
In addition to identified constraints and limitations of available data (published or otherwise), it was not possible to compare study findings directly owing to differences in study design and methods. However, the general conclusions of the studies in this review unequivocally suggest that a majority of malaria vector biting occurs outdoors during early evening hours, before people retire indoors, followed by indoor transmission for unprotected individuals not using a bed net (treated or not). This is evident even in contexts where unadjusted human biting rates are higher outdoors than indoors. However, when viewing ITN users, we estimated that roughly half of the malaria exposure likely occurred outdoors in some settings, indicating a clear gap in protection.
One of the most relevant indicators for understanding RMT is the protective efficacy of ITNs, defined as the proportion of human exposure to malaria vectors prevented by ITN use out of total exposure (ie, compared with a nonuser) [120]. Protective efficacy, which is the overall reduction in nightly biting rate for an ITN user compared to a nonuser, ranged from 88% to 93% in 2 settings where ITN coverage was suboptimal (Lao PDR and Thailand), with even lower estimates (<50%) for settings beset with primarily exophagic and/or opportunistic malaria vectors despite high ITN coverage (Myanmar) [46,75]. Understandably, the fraction of exposure occurring outdoors and indoors during nonsleeping hours poses a significant obstacle to malaria control and elimination efforts [188].
The importance of human behavior for understanding malaria transmission dynamics cannot be underestimated, and yet, even with this commonly acknowledged link with persistent malaria, relatively few studies were identified that included specific human activities as risk factors. For those studies that have identified human factors, different methodological approaches were used across studies, greatly limiting the comparisons for this review.
Here we have identified gaps/deficiencies that have been addressed by ESPT guidelines: (1) standard operating procedures and methods to best estimate risk exposure away from the peridomestic setting; (2) a standard analytical approach to measure human-vector interaction, which should account for outdoor sleeping as well as segments of the population that may spend most or all of the night away from home; and (3) collection of human and vector data close in time and location, and across time points (longitudinally), to reflect changes in vector and or human behavior between seasons as proposed by Monroe et al [120] and Killeen et al [189,190]. In rural agrarian and swidden agriculture, there is a need to develop locationspecific seasonal agricultural or forest calendars to reflect normal activities during the year [125].
Our review identified the ESPT as a standard approach to validate the estimates used for collecting human behavioral and entomological data-an important next step for developing a more uniform understanding of RMT and risk factors in general for those vulnerable populations subjected to persistent malaria. Since linking human activities with entomological parameters of transmission is generally lacking, these parallel domains need to be carefully integrated if we are to make sound epidemiological sense of the transmission dynamics at work. Finally, only with a better understanding of the gaps in protection can other appropriate vector control interventions be introduced or developed, over time and across settings, to address these challenges. To do so must take into account the health systems enablers, such as capacity to maintain supply chains of necessary commodities, maintain high-quality, timely, and responsive heath information systems and develop human resource capabilities as well as adaptive capacity at the district to community level. This type of data will be needed to ensure access to and encourage use of LLINs and vector control products among at risk populations.
In conclusion, perhaps the most important finding of this review pertains to the current state of the published research literature, which was limited at best. Standardized procedures and methods to estimate human exposure to mosquitoes indoors for unprotected residents at high vector control coverage within and beyond the peridomestic setting was not provided in 8 of the 14 survey-based studies included in this review. This omission rendered it impossible to reliably interpret the relative importance of the findings in the respective studies. Reliable interpretation of entomological and human behavioral data was often undermined by inadequate description of the study design and methods. Examples included the frequent failure to report how data were obtained (eg, in response to a structured checklist or guidelines provided by ESPT). The omissions described above likely reflect that the focus of this review (reasons for not integrating entomological and human behaviors as reported by study participants) was rarely a primary focus of RMT studies as identified by the search method. On the other hand, well-documented descriptive analysis of nighttime human activities and use of preventive measures were reported in 16 studies among 8 countries of interest: Cambodia (3 studies) [19,62,121], China (1 study) [122], Indonesia (1 study) [127], Lao (1 study) [42], Malaysia (3 studies) [30,47,48], Papua New Guinea (1 study) [50], Solomon Islands (1 study) [53], Thailand (3 studies) [44,123,124], and Vietnam (2 studies) [45,126].
Taken together, the omission of important data and information in many of the survey findings, the lack of a clearer understanding of RMT concept, and the dearth of dedicated mixed methods (eg, quantitative and especially qualitative investigations) seriously compromised the findings cited in this review. The seemingly clear patterns evident in the reviewed data and the recommended interventions should, therefore, be considered cautiously and highly tentative until greater numbers of well-designed studies are available in the literature. The current evidence-based review is not sufficient in scope or quality to reliably inform personal protection or preventative promoting interventions or broader community campaigns targeted at individuals who possess but do not (reliably) use mosquito nets or long-lasting hammock nets.