Protective effect of house screening against indoor Aedes aegypti in Mérida, Mexico: A cluster randomised controlled trial

INTRODUCTION

Arboviral diseases caused by Aedes-transmitted viruses (ATV) such as DENV, CHIKV and ZIKV present a significant public health problem in urban areas worldwide. The widespread distribution of Ae. aegypti, the main ATV vector in the Americas, puts approximately 500 million people at risk of dengue infection [1] and has fuelled the pandemic propagation of novel viruses such as chikungunya [2] and Zika [3,70,71]. Without commercially efficacious and fully licensed vaccines or therapeutics available against many of these arboviral infections, vector control aimed at reducing mosquito vector populations and/or their contact with humans remains the immediate alternative to reduce or prevent ATV transmission [6]. Current vector control strategies of ministries of health (MoH) primarily focus on reducing vector density by either targeting immature stages and their habitats or the adult mosquito population [6]. Although interventions to reduce vector contact with humans are routinely recommended, personal protection and modifications/improvements to the built environment (e.g. mosquito proofing of houses) are seldom implemented as part of MoH programmes.

House screening (HS), covering doors and windows with mosquito nets/screens, is a house improvement and a pesticide-free control approach to reduce human–Ae. aegypti contacts [7, 8]. For decades, people had used netting of different materials to screen their houses and prevent the entry of nuisance (or disease-carrying) insects [9,10]. This approach is particularly prevalent in urban areas, as building structure and economic resources facilitate their adoption. While HS is identified as an example of a housing intervention following the principle of ‘Keeping the vector out’ promoted by WHO [6, 11], this intervention has been largely overlooked by policies and programmes for the prevention and control of ATVs [12, 13]. It was not until 2017 that the WHO Special Programme for Research and Training in Tropical Diseases (TDR) cited HS as a promising vector management approach for the prevention and control of ATVs [14]. Recent meta-analyses and systematic reviews provide evidence of the effectiveness of house screens on external doors and windows in preventing dengue transmission [15, 16]. However, stronger evidence of its efficacy obtained from field randomised trials is recognised as necessary.

In the last decade, projects within the ‘Eco-Bio-social Research’ and ‘Ecohealth’ programmes in Mexico supported by TDR and the International Development Research Centre (IDRC) showed that insecticide-treated screening (ITS, long-lasting insecticide-treated nets fixed with aluminium frames on doors and windows) acts as a physical/chemical barrier that confers sustained protection against indoor female Aedes aegypti infestation [17-21]. Moreover, ZIKV detection in Ae. aegypti during a Zika outbreak was reduced by 85% in clusters with ITS versus untreated control clusters [21]. Although ITS is a widely accepted intervention by the community [18, 22], its accessibility is limited because insecticide-treated nets (ITNs) are not yet commercially available for public use since they are exclusively sold to the Ministry of Health in Mexico [23]. Given that the insecticidal effect of LLINs wanes after a couple of years [16], the sustainability of both HS and LLINs depends on careful evaluations of their cost, scalability, and entomological/epidemiological impacts.

The main goal of this study was to evaluate, in an entomological cluster randomised control trial (CRCT), the efficacy of screening doors and windows with a regular mosquito mesh in reducing infestation with, abundance of and infection by indoor collected Ae. aegypti mosquitoes in the Mexican city of Merida, Yucatan, in 2019. We also assessed the domestic practices implemented by the study participants to reduce mosquitoes and mosquito-borne diseases, as well as the perception and acceptance for HS in intervened households. We hypothesised that by reducing the abundance of Ae. aegypti inside households using HS, a reduction in infection in Ae. aegypti can be achieved.

MATERIAL AND METHODS Study site

Merida (20°58′2.532″N; 89°35′33.3096″W) is the capital and the major urban centre of the state of Yucatan, with a population of 921,771 inhabitants living in 284,468 households [24]. Average elevation of the city is 9 metres above sea level and the climate is mainly warm with an annual average temperature of 26–27°C (36°C max to 18°C min). Although there is continuous dengue virus (DENV) transmission throughout the year, two distinct seasons can be clearly identified: a rainy season from May to October and a dry season from November to April. The rainy season is historically associated with mosquito abundance, dengue transmission (increases 80%) and augmented vector control activities [25].

At the national level, Merida is among the cities that have reported the highest proportion of dengue cases in the last 15 years (2.6%) and accounted for >40% of all dengue cases in the state of Yucatan during the last decade [26]. The first cases of chikungunya in Merida and a subsequent outbreak (1531 cases) occurred in 2015 and transmission decreased during the following years (11 cases in 2016, and 0 cases in 2017–2018) [26, 27]. ZIKV transmission was initially detected in May 2016 with 2,199 cases reported, although transmission decreased to 24 cases in 2017 and 28 cases in 2018 [26, 27]. No laboratory-confirmed cases of chikungunya and Zika virus were reported during 2019–2020 [28]. Various neighbourhoods in Merida have been historically identified as hotspots because they produce more cases and consistently demand vector control activities [26, 27, 29-31]. Previous studies in Merida showed that the most important productive container types for Ae. aegypti immatures are disposable containers, buckets/pots and other rain-filled objects left in backyards [30, 32, 33] along with non-residential habitats, such as subsurface catch basins (e.g. drainage systems, storm drains, street drainage) [31, 32].

Experimental design

The study followed a standard two-arm entomological CRCT design, comparing six clusters with the intervention (HS) with another six clusters without HS (as control) during the peak of mosquito abundance, corresponding to the rainy season [18, 21, 25]. As in previous studies [17, 18, 20, 21], we originally planned to carry out the post-intervention evaluation for a second year, but this activity was halted by the COVID-19 pandemic.

Twelve clusters comprising 100 households each (1200 houses in total) in different neighbourhoods of Merida (n = 12) were selected based on their entomological and epidemiological importance according to the local vector control programme (Figure 1a). These 12 clusters were numerically and blindly randomised using an Excel spreadsheet (MS Excel 365, 2019) to generate two groups of six clusters each. On a second round of randomisation, one group was selected to receive the intervention (n = 6) and the other group remained as control (n = 6) (Figure 1, Figure S1). The clusters comprised an average set of 18 city blocks (each block had, on average, 25 premises) located within the areas previously identified as hotspots of Aedes-borne virus transmission [27]. Clusters localisation comprised residential areas, where about 23,330 inhabitants live [24]. Entomological evaluations were conducted on a random sample of 30 houses per cluster (intervention: 180 houses; control: 180 houses) (Figure S1). Not all premises within a block were enrolled in the study because they were small businesses, empty, or householders who declined to participate or were absent at the time of enrolment. Houses included in the study were typically single storey, made of cement-plastered blocks with a closed roof and with no ventilating features (e.g. ventilation bricks, eaves, etc.) other than windows (Figure 1b; Table S1).

image

(a) Study site showing the location of the study areas and randomly selected clusters with and without house-screening (HS) interventions in the city of Merida, Mexico. (b–d) House screening – with regular netting mounted on aluminium frames – installed on doors and windows of houses. (e) Female Aedes aegypti standing on a screen

We powered the study to detect a significant difference in our primary entomological endpoint: the density of Ae aegypti indoors collected after a 10-min Prokopack aspiration session [21]. Based on an expected effect size of 70% in the reduction in Ae. aegypti indoors by ITS [18] from an expected mean baseline number of 4.4 ± 9 [36], an alpha of 0.05 and a power of 80%, we estimated a total of 134 houses per arm (268 total houses) to detect a significant difference between groups (https://clincalc.com/stats/samplesize.aspx). Therefore, our trial size provided enough statistical power to evaluate a difference at even a lower effect size than 60%.

House screening

The installation of HS ran from July to August 2019. Regular fiberglass net (brand Herralum®, available in 30 m length x 1.50 m width rolls, colour grey, mesh light 0.6 x.07 mm and density 0.32mm) was mounted in aluminium frames custom fitted to doors and windows of houses (Figure 1b–d) in collaboration with a third-party local small business (Vidrios y aluminios Bojorquez S.A) as described in Manrique-Saide et al. [21, 31] and Che-Mendoza et al. 2015, 2018, [17, 18]. During the installation, at least one person in every household received information about the proper use and maintenance of HS from the research staff. The main recommendation to the householder to keep the door closed as much as possible. The average cost of the HS per house was US $141.66. This average price included screening of two doors and seven windows of a typical 75 m2 household (floor area) (Figure 1b–e).

Both areas received routine vector control as part of national policy in response to dengue outbreaks and entomological risk indices [37]. The activities during 2019 included: outdoor spraying with organophosphates (malathion), fast-acting pyrethroids (transfluthrin) and neonicotinoids plus pyrethroids (imidacloprid +prallethrin); and indoor space spraying with carbamates (propoxur and bendiocarb) and larviciding with spinosyns, bacterial insecticides (Bti), insect growth regulators (methoprene and pyriproxyfen) and organophosphates (pirimiphos-methyl and temephos).

Entomological studies

Two cross-sectional entomological surveys were conducted in intervention and control clusters as in Manrique-Saide et al. [21, 31] and Che-Mendoza et al. 2015, 2018 [17, 18]. Indoor adult mosquito collections were performed in a randomly selected subsample of 30 houses from each cluster. From a list of participating houses ordered numerically in each cluster, random numbers were generated until the 30 houses were completed (Figure S1). The baseline survey was completed in May 2019 and was followed by a post-intervention (PI) survey during the wet season (September to October) in 2019. The primary outcome measures were indoor Aedes mosquito density and Aedes aegypti infection with Aedes-borne viruses.

Indoor adult mosquitoes were collected with Prokopack aspirators [38] during a 10-min period per house. Collections within each cluster were performed by three teams of two skilled collectors each on the same day between 09:00 and 12:00 hrs. Considering HS from our study is easily recognisable, entomological collections could not be blinded to the intervention. All mosquitoes collected were identified to species and sex and stored for molecular detection of viral infection.

Detection of DENV and ZIKV infection in Aedes mosquitoes

The study included the detection of DENV and ZIKV genome in female Ae. aegypti collected from the same sample of houses [n = 183 houses divided into HS (n = 80 houses) and control (n = 103 houses)], in which we performed the entomological collections for baseline and post-intervention surveys.

A total of 194 pools (1 to 6 mosquitoes per pool) of field-collected female Aedes mosquitoes were preserved in Eppendorf tubes containing RNA stabilisation reagent (RNAlater; Thermo Scientific). Samples were initially stored at −20°C at the Collaborative Unit for Entomological Bioassay (UADY), then transported to the Virology Laboratory of the Regional Research Center ‘Dr. Hideyo Noguchi’ (CIR-UADY) for further analysis. Pools were processed for RNA extraction followed by molecular detection of viral RNA genome using an in-house endpoint RT-PCR assay. Briefly, each pool of female Ae. aegypti mosquitoes was initially disinfected with 70% ethanol at room temperature for 2 h. Then, samples were mechanically homogenised in 150 μl of sterile PBS1X using a sterile pestle and electric homogeniser as previously described [39]. RNA extraction was performed using a commercial QIAamp Viral RNA Mini kit (QIAGEN) following the manufacturer's instructions. RNA extract was eluted in nuclease-free water (Ambion) and quantified using a nanodrop (Thermo Scientific). Finally, extracts were stored at −80°C until further analyses.

DENV and ZIKV infections in Ae. aegypti mosquitoes were examined by an end-point one-step RT-PCR. Primers were designed to target a ~200 bp fragment of the viral gene NS5 of DENV (DENV-F: ACAAGTCGAACAACCTGGTCCAT; DENV-R: GCCGCACCATTGGTCTTCTC) [40], or a fragment of ~100 bp of the viral E gene of ZIKV (ZIKV-F: CCGCTGCCCAACACAAG; ZIKV-R: CCACTAACGTTCTTTTGCAGACAT) [41]. The RT-PCR protocol was performed using a Mastercycler EP Gradient-Thermal-Cycler (Eppendorf) and the OneStep RT-PCR Kit with a master mix including the following components: QIAGEN OneStep RT-PCR Buffer (5×), dNTP Mix (10 mM each), QIAGEN OneStep RT-PCR Enzyme Mix, Q-solution (5×), forward and reverse primers (10 µM), RNAse free-water and extracted RNA template (100–200 ng per reaction). Amplification parameters were established as follows: initial reverse transcription step at 50°C for 30 min, followed by an initial PCR activation step at 95°C for 15 min and 40 cycles of denaturation at 95°C for 1 min, Tm annealing at 53°C for 1 min and extension at 72°C for 1 min; and final extension at 72°C for 5 min. Viral RNA extracted from DENV and ZIKV strains grown in C6/36 cells (Ae. albopictus, from the CDC [USA]) were used as positive controls. RNA extracted from a laboratory-reared Aedes aegypti strain from Yucatan was used as negative control. Amplicons were visualised using agarose gel (1.5%) stained with Syber safe (Thermo Scientific) under UV excitation. Screening for arboviral infections was blindly performed at C.I.R-UADY.

Social assessment

As in previous studies, our team performed a social assessment focused on the community initial response during the enrolment (pre-intervention) and a post-intervention acceptance and perceived efficacy survey among the participants [21, 22]. In February–March 2019 (during the enrolment process and before the intervention), face-to-face household surveys were conducted among 150 heads of family randomly selected from houses in intervened clusters to address the social response of the project. Topics included knowledge, attitudes and practices (KAP survey) on mosquito-borne diseases as well as personal and domestic preventive measures.

In June 2020 (post-intervention), a second household survey was applied to 100 family heads interviewed during the enrolment to evaluate the social acceptance and the perceived efficacy of the intervention. Topics explored were the acceptance of intervention, opinion on the installation process, perception of temperature increase associated with HS, perceived reduction in mosquitoes inside the houses, positive cases of DEN/CHIK/ZIK reported by the families after the installation of the mosquito screens and recommendations for scaling-up the HS method. Because of the COVID-19 contingency, the questionnaires were applied through telephone calls to guarantee the safety of participants and the scientific team.

Data analysis

From indoor Prokopack adult collections, we calculated: (a) Houses positive to at least one female Ae. aegypti (%); (b) Houses positive to blood-fed female Ae. aegypti (%); (c) Number of females per house and (d) Number of total blood-fed females per house. We also report the prevalence of positive houses to indoor female Ae. aegypti with arbovirus infection (houses with at least one pool of Ae. aegypti females positive to the presence of arboviral RNA genome [e.g. DENV and ZIKV]).

Logistic regression models (for presence–absence mosquito data) and negative binomial models (for count data) accounting for each house's cluster (cluster-robust SE calculation) were performed for each cross-sectional entomological evaluation survey. Odds ratios (OR) and rate ratios (RR) with 95% CI were assessed and significance expressed at the 5% level. Analyses were performed using STATA 13.0 (Stata Corp, College Station, TX, USA).

Values from the infection calculation were used to estimate a measure of epidemiological efficacy, as HSeff = (1−OR) × 100 [42]. This value ranks between 0 and 100 and indicates the proportional reduction in Ae. aegypti infection in the intervention arm compared to the control arm.

Ethics statement

This study was approved by the ethical committee of CCBA-UADY (CB-CCBA-I-2019–003). Written informed consent was obtained for each participating household (householder over the age of 18) at the beginning of the study.

RESULTS Impact of HS on indoor adult mosquitoes

A total of 897 adult indoor resting mosquitoes (413 males, 484 females) were collected during the whole study period. Ae. aegypti was the most abundant species, representing 76% of the total collection (682 [320males, 362 females]), followed by Culex spp. (23%, 206/897) and a few Ochlerotatus taeniorhynchus (1%, 9/897).

Entomological indicators are summarised on Table 1. During the pre-intervention survey (dry season, May to June 2019), adult-based entomological indicators showed similar seasonal patterns of house infestation in both study arms (Table 1). Indoor Ae. aegypti females at different feeding stages were collected among 20–30% (36–54/180) houses in both study arms.

TABLE 1. Entomological indicators for control and HS intervention surveys during dry and rainy seasons Survey Treatment Mean SEM OR/IRR 95% CI p value Houses positive for Aedes females Dry season 2019 Control 0.27 0.03 HS intervention 0.24 0.03 0.86 0.47–1.58 0.64 Rainy season 2019 Control 0.30 0.03 HS intervention 0.19 0.03 0.56 0.33–0.97 0.04* Houses positive for blood-fed Aedes females Dry season 2019 Control 0.27 0.03 HS intervention 0.23 0.03 0.81 0.45–1.45 0.48 Rainy season 2019 Control 0.28 0.03 HS intervention 0.17 0.03 0.53 0.28–0.97 0.04* Number of female Aedes per house Dry season 2019 Control 0.41 0.07 HS intervention 0.57 0.10 1.41 0.71–2.79 0.32 Rainy season 2019 Control 0.69 0.12 HS intervention 0.34 0.06 0.50 0.30–0.83 0.01* Number of blood-fed female Aedes per house Dry season 2019 Control 0.39 0.07 HS intervention 0.51 0.1 1.30 0.67–2.51 0.44 Rainy season 2019 Control 0.65 0.12 HS intervention 0.31 0.06 0.48 0.27–0.85 0.01* House positive for Aedes females infected with arboviruses (pools) Dry season 2019 Control 0.18 0.03 HS intervention 0.11 0.02 0.55 0.19–1.58 0.27 Rainy season 2019 Control 0.20 0.03 HS intervention 0.07 0.02 0.29 0.1–0.86 0.025* House positive for infected Aedes DENV (pools) Dry season 2019 Control 0.13 0.02 HS intervention 0.08 0.02 0.58 0.19–1.71 0.32 Rainy season 2019 Control 0.19 0.03 HS intervention 0.06 0.02 0.28 0.09–0.85 0.024* House positive for infected Aedes ZIKV (pools) Dry season 2019 Control 0.14 0.03 HS intervention 0.07 0.02 0.42 0.17–1.08 0.07 Rainy season 2019 Control 0.17 0.03 HS intervention 0.06 0.02 0.28 0.09–0.91 0.034* Note Comparison between intervened-treated (HS) and untreated (control) arms on indoor female Aedes-based entomological indicators (n = 180 houses per arm) in Merida, Mexico. Odds ratios (OR) and rate ratios (RR) with 95% confidence intervals are showed for presence–absence data and count data, respectively, for each cross-sectional entomological survey by arm. * Statistical significance is indicated in bold (p < 0.05). Abbreviation: HS, house screening.

After the intervention (rainy season 2019), adult Ae. aegypti abundance was significantly lower in the houses protected with HS than in the houses not protected with HS (Table 1). Houses with HS had significantly lower risk of having Ae. aegypti female mosquitoes (OR = 0.56, 95% CI 0.33–0.99) and blood-fed females (OR = 0.53, 95% CI 0.28–0.97) in comparison with unscreened households. Indoor abundance of Ae. aegypti also showed significantly fewer adult females in houses protected with HS (RR = 0.50, 95% CI 0.30–0.83) and fewer blood-fed females indoors (RR = 0.48, 95% CI 0.27–0.85) (Table 1).

Impact of HS in houses with pools of female Ae. aegypti positive for arbovirus

Among 360 houses from both arms sampled during the study, a total of 26% (93/360) and 25% (89/360) were positive for Ae. aegypti females during the dry and rainy season respectively. A total of 194 female Ae. aegypti pools (mean of 1.06/house positive to females) were analysed for DEN/ZIK infection. A total of 99/194 pools (51%) were positive for arboviruses, from which specifically 42% (82/194) and 40% (79/194) were positive for DENV and ZIKV respectively.

At baseline (dry season), no significant differences were observed between study arms on the prevalence of arbovirus-positive pools (Table 1). After HS implementation, having screens was significantly associated with fewer houses with indoor female Ae. aegypti positive for either arbovirus (OR = 0.29, 95% CI 0.10–0.86, p = 0.02). Although we continued detecting indoor Ae. aegypti females with DENV and ZIKV in houses from both study arms, the proportion of houses with HS positive for Ae. aegypti females with arbovirus was lower (7%) than in unprotected houses (20%). Based on these data, the estimated intervention effectiveness of HS in reducing arbovirus infection in Ae. aegypti was HSeff = 71%.

Knowledge of ATV and preventive practices

The demographic characteristic of surveyed population is included in the supplementary material (Table S2). Participants were already familiar with HS, although none of the houses had HS installed prior the intervention, mainly because of the cost (70%, 105/150), a perceived difficulty for its maintenance (20%, 30/150) and because they could move to another house (10%, 15/150).

Most respondents associated mosquito bites with the infection/transmission of DENV (91%, 134/150), CHIKV (88%, 129/150) and ZIKV (88%, 129/150). They were aware of some clinical manifestations, which were cited as differentially associated with each disease (Table S3). For example, fever was perceived as the main symptom of DENV but not for CHIKV and ZIKV, while joint pain was the most mentioned symptom associated with CHIKV and ZIKV; however, nobody mentioned that ZIKV could be asymptomatic, and respondents were overall less aware about this disease.

Regarding preventive practices, about half of householders reported the use of topical repellents (49%, 71/150) and commercially available insecticide products (68%, 100/150) as the main domestic preventive measures to avoid mosquitoes indoors. The main reason reported by repellents users was the efficacy of the product, while the non-users said that they could not afford the products. People also used commercially available household insecticides because their perceived efficacy but some people did not use them due to health-related concerns, for example, having asthmatic relatives at home and the perceived toxicity of the product.

Social acceptance and perceived efficacy of HS

All interviewed participants reported acceptance of the intervention (HS), high expectations on its efficacy and recommended the scaling-up of the intervention to other areas of the city. The main reasons for acceptance were to avoid mosquitoes at home (77%, 77/100), concerns about ATV (63%, 63/100) and the free cost of the intervention (54%, 54/100) (Table 2). The majority (94%, 94/100) did not recall having any family member sick from any ATV at home after the installation of HS and most of them (92%, 92/100) believed that HS helped prevent their families from mosquitoes-borne diseases.

TABLE 2. Reasons for acceptability and the perceived efficacy of house screening among the participants of the study Topics addressed N = 100 Reasons for acceptance of house screening To avoid mosquitoes at home 77% (n = 77) Concerns that Aedes-borne diseases could impact their families 63% (n = 63) The free cost of the intervention 54% (n = 54) Impact perceived Reduction in mosquitoes indoors after the intervention No mosquitoes indoors 66% (n = 66) Reduced number of mosquitoes 29% (n = 29) No reduction in mosquitoes indoors 5% (n = 5) Cases of DEN/CHIK/ZIK reported by the families after the intervention No 94% (n = 94) Yes 6% (n = 6) Perception of temperature increase due to house screening Did not acknowledge any increase in indoor temperature 80% (n = 80) A light overheating was reported but associated with specific day-hours (mid-day) 18% (n = 1

留言 (0)

沒有登入
gif