Targeted outdoor residual spraying
The insecticide dosage is 25 mg/m2 and is applied by trained operators using a residual hand compression sprayer (Semco). The target spray areas are outer walls covered or partly covered (corridors, staircase, parking, lobby or any corner liable to become a resting spot for mosquitoes) in all floors. We exclude sections of wall close to power plugs, elevator buttons, walls covered with wall paper, oil-based painted walls oil painted walls or tiles. Residents are requested to remove edible plants, hanging clothes, pet cages (e.g. of cats, dogs and birds), and to cover aquaria during the day of the intervention. Parts of corridors with such objects are not sprayed. Based on our findings on the residual effects of K-Othrine® Polyzone [17, 21], TORS is conducted every four months in all intervention areas.
To control the quality of TORS spraying, each spray pump is equipped with a control flow valve. The latter is a device that fits next to the nozzle, and ensures that output remains constant as the pressure in the spray tank decreases. The spray pumps are washed after each use to avoid clotting of the insecticide in the nozzles. All spray cans are washed thoroughly, checked and calibrated once a week. During spraying activities, a quality control officer observes the spraying procedure and technique. Pictures and video recorded randomly by the officer will be used for the quality control evaluation, future reference or records. Training and evaluation of spray and calibration is provided to all operators every two months.
Deployment of auto-dissemination devices
The strategy for ADD deployment in these urban settings was adapted from In2Care standard recommendation (1 trap/400m2), based on results from our pilot study in Johor Bahru, Malaysia [21] and another study evaluating the vertical infestation of Aedes in high-rise building [22]. Smart deployment of ADDs only on floors with higher expected mosquito density was found to be sufficient to reduce the mosquito population in the whole building.
The exact number of ADDs and suitable location is determined at the start of the study (Additional file 2). ADDs are distributed over each floor as evenly as possible and placed according to the manufacturer’s recommendations in semi – indoor environment such as corridors, multipurpose halls, lobby, prayer hall, shops and parking (i.e. in shaded sites).
Community engagement
Community engagement (CE) starts at the time of baseline data collection and is maintained throughout the study period by a team specially trained in CE activities. The main objectives of this activity are to create public awareness and trust, and to increase their participation during and following the deployment of the intervention. The first step of CE activities consists of an appointment with the community leaders in each cluster through phone calls using a standard script in order to schedule a face-to-face meeting. The purpose, objectives, procedure and timelines of the iDEM trial are presented to community leaders followed by a question and answer session. Information on demographic and architectural characteristics such as number of buildings, number of floors per building, number of units per floor, total number of occupants, parking lots, and public facilities (gymnasium, kindergarten, shops, restaurants, swimming pool, and playground) of the cluster are recorded after the briefing. A second meeting involving the residents is organized upon the community leader request. Information, education and communication material such as banners, posters (Additional file 3), and brochures are distributed to explain the objectives of the study and the role of the community during and following the deployment of the intervention. Softcopy of these materials is sent to the building managers through email or WhatsApp message and subsequently disseminated among the community. A notification is sent to the building manager one week before the deployment of the intervention to remind them of the date and details about the intervention.
Outcomes
The primary outcome measure is the estimated dengue incidence rate in both control and treatment clusters. The number of confirmed dengue cases is provided by the e-Dengue system (http://spwd.arsm.gov.my).
Dengue is a notifiable disease in Malaysia. All clinically suspected or confirmed dengue cases are notified by the medical officer to the nearest District Health Office within 24 hours (Additional file 1). Currently, dengue confirmation is based on dengue rapid combo test for 1) qualitative detection and differentiation of specific IgG and IgM antibodies to dengue virus in human serum and plasma, and 2) dengue rapid NS1 test for qualitative detection of dengue NS1 antigen in human whole blood, serum and plasma [23]. Further laboratory tests are performed if necessary (under clinical management not determined by the current trial). The e-Dengue case definition is based on the international classification of diseases ICD (10: A90, A91) and positive diagnostic results of either NS1, IgM or IgG or any combination of test results [24]. For the purpose of this trial, only confirmed dengue cases will be used for statistical analyses.
Secondary outcomes measures include:
- Ovitrap index (OI): number of positive ovitraps divided by the total number of recovered ovitraps.
- Larval index (LI): number of larvae identified divided by the number of ovitraps recovered.
- Container index (CI): proportion of water-holding surveyed containers infested with Aedes larvae or pupae.
- Adult survey: mean number of female adults Aedes collected in sticky ovitraps.
- Resistance to deltamethrin: measurement of knock-down and mortality rates in wild strain adult mosquitoes population
- Residual activity of deltamethrin measured by wall deposit bioassays [25]
Sample size
The sample size determination was adapted from the bootstrap approach of Kleinman & Huang [26] for cluster randomized trials. To take into account the regression to the mean due to the selection of the localities with the highest incidence rates in 2015-2016, sample size determination was based on the incidence rates of the years 2017-2018. For the purpose of sample size calculation, the estimated number of inhabitants in each cluster was obtained as follows: number of blocks (buildings) x number of floors x number of apartments in each floor x 4.9 (i.e. estimated number of individuals in each apartment is based on National Health and Morbidity Survey 2019 and Department of Statistics). The mean of the cluster-level incidence rates was 0.34% per year (range: 0 to 4% per year). The power calculation used resampling with replacement to generate plausible incidence distributions during the future study period, in each arm. The resampling was repeated 5000 times. Each time, the resampled data from the clusters underwent a simulated randomization to the intervention or control arm. For those in the control arm, the inferred dengue incidence rates were calculated from the confirmed dengue cases observed during the years 2017-2018. For clusters in the intervention arm, the inferred incidence rates were calculated from the observed incidence rates multiplied by the expected effect of the intervention. The effect of the intervention was estimated from this simulated dataset using a negative binomial regression model. Finally, the power was estimated as the proportion of analyses with a p-value ≤ 0.05 (two-sided). If we assume that the minimum effect that we are able to detect is a reduction in the incidence by 33% in the intervention areas (0.34% and 0.23% estimated mean incidence rates per year in the control and intervention arms, respectively), the inclusion of 280 eligible clusters (140 per arm) achieves a power of 85%. This means that, if the rates are as assumed, the study has 85% probability of establishing a benefit of the intervention, using a two-sided significance level of 5%. This power was considered acceptable and this sample size was used for the trial.
Allocation
The random allocation of the clusters to the control or the intervention arm (1:1 ratio) was carried out using the PLAN procedure of SAS software, version 9.4 (Copyright (c) 2002-2012 by SAS Institute Inc., Cary, NC, USA), and was stratified on the economic level of the clusters (low or medium) because this is associated with dengue incidence.
In each stratum, clusters were randomly numbered and block randomization were carried out with block sizes of 10 and 16 in order to allocate the clusters to the intervention or control arm. Randomization was carried out after the protocol was approved by the IMR Research Committee.
This trial is not blinded. All clusters were allocated at the start of the trial so an allocation concealment mechanism is not required. The allocation sequence has been generated by a statistician of the Service of Biostatistics of Hospices Civils de Lyon, France. The code written by this first statistician was double-checked by a second statistician in regard to the conformity to the randomization protocol. The list of clusters in each arm was saved in a text file and transmitted to the investigators (IMR-Malaysia).
Data collection
Epidemiological data
No active recruitment of dengue cases is planned, as the information on the incidence of dengue during the overall study period in both control and intervention clusters is extracted from the national e-Dengue surveillance database. For each confirmed dengue case, demographic data, signs and symptoms at first medical visit, date of diagnosis, results of the rapid dengue COMBO test, hospitalization and vital status at discharge are extracted.
Entomological data
Entomological data for the assessment of secondary outcomes are collected in 12 clusters randomly selected from each arm (24 in total) and stratified by socio-economic status (medium versus low cost).
Climatic parameters
Common climatic parameters (rainfall, temperature, and humidity) have been reported to influence the intensity and magnitude of dengue incidence [27-29]. These parameters will be requested from the Malaysian Meteorological Department, Ministry of Energy, Science, Technology, Environment and Climate Change for two points: Petaling Jaya and Kuala Lumpur International Airport.
Monitoring Aedes population densities
The density of Aedesaegypti and Ae. albopictus populations will be monitored using larval and adult collection methods in a sample of 12 clusters per arm. Larval surveys are carried out using a minimum of 40 ovitraps per cluster. Each ovitrap consists of a 250 ml cylindrical, black plastic container (7.5cm diameter, 9.0cm height, 300ml volume) filled with tap water to a depth of 5.5 cm and is equipped with a removable oviposition paddle made from a thin strip of brown hardboard (10 cm x 2.5 cm x 0.3 cm). Larval surveys start five weeks before the start of the intervention to measure the baseline density. Ovitraps are positioned randomly in spaces with minimum human, physical, and environmental disturbance; a) in the semi-indoor environment – defined as the area outside of the housing units but still sheltered by the roof (e.g. shared corridor and stairway); b) in partially or totally shaded outdoor areas (e.g. near bushes, small plants, and temporary structures); c) indoor (under the sofa, dining table, sink, bed, working / study desk) only in the apartments of those who volunteer to have them.
To determine the OI, all ovitraps are collected once a week and replaced by new ones with fresh tap water and egg-free oviposition paddles. The collected ovitraps are brought back to the IMR laboratory (Kuala Lumpur) for further processing. Ovitrap contents and oviposition paddles are transferred into plastic containers and labelled with the name of the cluster and date of collection. Larvae emerged from the eggs laid in the ovitraps undergo species identification [30, 31] using a compound microscope (Nikon Eclipse® E100, Japan).
To determine the CI, a minimum of 20 suspected mosquito breeding containers within 50 m radius of the study cluster are examined once a month in each cluster. Larval samples from the positive containers are collected in a sample bottle and brought back for identification. The larvae are allowed to emerge into adults for morphological identification.
To monitor adult mosquito densities, monthly surveys are carried out with sticky ovitraps, i.e. a modification of conventional ovitraps as described by Lau et al [32]. The inner wall of the ovitrap is lined with a 5.5 cm x 24 cm transparent plastic sheet. The inner plastic sheet is covered with insect glue (Neopeace, ACM, Malaysia) attached to the container using adhesive tape on both sides of the top of the container. A hole is drilled about 3cm above the bottom to avoid flooding of the trap with rain water. A total of 50 – 200 sticky ovitraps are deployed in the study area, in every alternate floor with emphasis on dark and humid areas such as garbage rooms, emergency exits, shoe racks and vegetation considered as Aedes resting places. The adult samples are isolated in a single vial and send for identification.
Monitoring mosquito resistance
Adult Bioassay Test
We will assess measure susceptibility levels to deltamethrin in the field population. The test will follow the WHO protocol [33]. Pyrethroid resistance monitoring is conducted on sugar-fed adult female mosquitoes aged three to five days against a diagnostic concentration (0.03% deltamethrin, 0.25% permethrin, 0.15% cyfluthrin, and 0.03% lambda-cyhalothrin), which is used to impregnate paper at the recommended exposure time. Five replicates of 20 adult field strain mosquitoes from the 24 clusters selected for entomological endpoints are prepared. Two replicates of susceptible mosquitoes from the insectary laboratory of IMR are used as control strains. The cumulative knockdown counts are recorded every minute within the exposure period of one hour, or until 90% knockdown is observed. After the exposure period, all the tested mosquitoes (live and knocked down) are transferred into holding tubes for a 24-hour recovery period and supplied with 10% sugar solution in cotton balls. Adult mortality is then recorded. If the mortality rate in the control group is between 5% and 20%, the percentage of mortality in the exposed mosquitos is corrected by using Abbot’s formula [33]. Surviving mosquitoes are transferred into Eppendorf tubes and kept at -80°C for enzyme micro-assay.
Enzyme Micro-assay
Three enzyme assays, namely mixed function oxidase [34, 35], esterase [35, 36], and insensitive acetylcholinesterase [35] are conducted in sugar-fed female mosquitoes, aged 3–5 days, to determine the enzyme activity. The female mosquitoes are raised from eggs collected in the field. The same homogenate of an individual mosquito is used to assess all three enzymes assays.
Monitoring quality of TORS and ADDs post-application
To assess post-application quality of K-Othrine® Polyzone spray, wall deposit bioassays are conducted 48 hours after spraying and repeated every three weeks for a period of four months [25]. Standard WHO bioassay cones are firmly positioned onto the TORS treated wall in a perpendicular position using masking tape. Ten adult female mosquitoes (sucrose-fed, 3–5 days old, non-blood fed) are introduced into the bioassay cone for 30 minutes through the aperture using a battery-operated aspirator, then collected back and observed for 24 hours mortality. Three biological replicates of 10 mosquitoes are prepared for laboratory and field strains of Aedes aegypti. Residual effect of K-Othrine® Polyzone is determined by evaluating the knock down time and mortality rate. Knock down is observed for 30 min at 1-minute intervals. After exposure, the mosquitoes are aspirated out, transferred to paper cups, held at 27 ± 2°C with 75 ± 10% relative humidity and sugar-fed. The mortality is recorded 24h after testing. The level of resistance is evaluated using the WHO criteria [33].
To assess ADD quality post-application, the presence/absence and estimated quantities of mosquito larvae in the water inside the ADDs are recorded at each service. Every month, 250 ml water samples are taken from three randomly selected deployed ADDs that are positive for Aedes larvae and brought back to the laboratory to confirm the presence and activity of the pyriproxyfen larvicide. In each 250 ml ADD water sample, 25 third instar (L3) larvae from the IMR rearing colony of Aedes aegypti are added [33]. Emergence of mosquito adults is monitored every day until all larvae have emerged or died, or all larvae from control emerge into adult. An emergence rate of adult mosquito less than 10% is indicative of adequate pyriproxyfen quality in the ADD water sample. The samples are collected at the same time as the Container Index sampling and are used for quality control of the ADD and as positive control for the auto-dissemination testing.
To assess auto-dissemination efficacy of the ADDs, water samples of nearby containers are assessed for pyriproxyfen content. The Aedes-positive samples from natural containers collected for the CI (as explained above) are used to monitor larval development. Per 10ml water sample collected, one susceptible strain larvae (L3, Aedes aegypti) is added to the water. Emergence of adults is monitored every day until all larvae have emerged or died. Significant reductions in adult mosquito emergence rates (i.e. high rates of larval/pupal mortality) in the intervention cluster samples compared to the control cluster samples is indicative of pyriproxyfen autodissemination from the ADDs.
Datamanagement
Epidemiological data are extracted from the e-Dengue system by the staff of the Vector Borne Disease Sector, Disease Control Division, Ministry of Health Malaysia. Confirmed dengue patients are assigned a unique anonymous code. Entomological data are reported on paper-based forms and subsequently entered into an Excel file by data managers at IMR. A second data manager checks the accuracy of the Excel file by comparing the paper-based forms and data entered in the Excel file. Access to the data is password protected and restricted to authorized study investigators and data management staff.
Both epidemiological and entomological data are sent to the department of Biostatistics in Lyon-France on a monthly basis. Data are controlled for consistency by the team in Lyon. The list of inconsistent or erroneous values (i.e. data queries) is sent to the data managers in IMR for verification. A database in SAS software is prepared for final statistical analyses. A common identifier for clusters is used in order to link the entomological and epidemiological databases.
The original paper-based entomological forms are stored in secured locked cabinets at IMR. As for epidemiological data, access to the data is password protected and restricted to only authorized study investigators and data management staff. The anonymized databases sent to the service of Biostatistics of Hospices Civils de Lyon are stored in the study folder on a secured computer server. Only the biostatisticians in charge of the analyses have access to the data. A daily backup of the study folder is made automatically.
No human samples will be collected for, or held by, this study. Entomology specimens will be stored at IMR. This study will be carried out in compliance with the Malaysian Personal Data Protection Act 2010, stipulating full protection of an individual’s personal information.
All data, documents, the database, and the intervention procedure will be available for audits by regulatory and independent authorities appointed by funders or IMR.
Plans to promote participant retention and complete follow-up
Our previous study showed that regular contacts between the study population and the field workers creates public trust [21]. To maintain adherence to the interventions, a liaison officer assigned to each cluster contacts (via phone call or messaging) the building manager regularly to collect all complaints and feedback from the population. The pest control organisation in charge of TORS and ADDs deployment provides also regularly feed back to the CE team on any concerns faced in the field. These are discussed and taken into consideration in order to improve adherence to the intervention.
Statistical methods
Statistical analysis will be performed using the statistical analysis software SAS® version 9.4. or higher. Missing data will be reported and their impact on the outcomes may be explored. No interim analyses are planned.
Primary outcome
The primary endpoint is to measure the effectiveness of the IVM approach on dengue incidence. The average of the cluster-level incidence rates will be compared between the control and intervention arms. In each cluster, the incidence rate will be estimated as the ratio of the number of dengue cases registered during the trial (starting from the first cycle of spraying and ending three months after the last spraying cycle), divided by the number of person-years. The number of inhabitants per cluster, as obtained for the sample size calculation, is checked and corrected if needed during the site visits for the purpose of calculation of incidence rates.
A negative binomial regression model will be used to compare the incidence between control and intervention clusters. This will yield the rate ratio and its 95% confidence interval, taking into account overdispersion. The response variable for this analysis will be the number of cases, and the logarithm of the person-years will be included as an offset. Stratification (low versus medium income) will be included as a covariate.
Secondary outcome
The main secondary outcome will be the OI. The latter will be estimated over the baseline and intervention period in the 24 clusters selected for entomological monitoring. To quantify the effect of the interventions on the proportion of positive ovitraps, a modified ordinary least squares regression model using a robust standard error estimator will be implemented [21]. Using this model, the effect of the interventions will be quantified by the estimation of a difference of risk with its 95% confidence interval. This analysis will be carried out for both species overall and for each of the two species of mosquito separately.
Larval count per ovitrap (LI) will be estimated in the 24 clusters selected for entomological monitoring over the baseline and intervention period. A negative binomial regression will be used with the number of larvae as the response variable. The analysis will be adjusted by the baseline measurement of the outcome. Using this model, the effect of the interventions will be quantified by the estimation of a ratio of means with its 95% confidence interval. This analysis will be carried out for both species overall and for each of the two species of mosquito separately if possible.
The adult population will be estimated by counting the number of adult female Aedes mosquitoes collected with sticky ovitraps during the study. The total number of adult females will be divided by the number of sticky ovitraps used to obtain the mean number of adult females.
The emergence rate, defined as the proportion of larvae that emerge as adult mosquitoes, will be estimated in each arm and compared between the two arms using the chi-squared test or the Fisher exact test. This analysis will be carried out for both mosquito species combined, and for each species separately.
Methods in analysis to handle protocol non-adherence and any statistical methods to handle missing data
The main analysis will be by intention to treat. The clusters will be analysed in their randomisation arm. As explained above, the primary outcome is provided by the e-Dengue system and will be consequently available for the overall included clusters during the study period. A secondary per protocol analysis will be carried out excluding the clusters with any major deviations to the protocol. Major deviations will be defined by the technical committee (trial governance members) during the data review.
Trial governance
Implementation of the field work is coordinated by IMR (Malaysia). The Malaysian Ministry of Health provide technical expertise and is in charge of dengue data extraction from the e-Dengue database. The Departments of Epidemiology and Public Health, and of Biostatistics (Claude Bernard University and Hospices Civils de Lyon-France) host the project leader and are responsible for statistical analyses. The trial is governed by a technical committee composed of academic/private partnership with skills and knowledge in the epidemiology of infectious disease in particular vector-borne diseases, entomology, new vector control technologies, policy makers, and statistics. The committee defines the study design, review available data, and make decisions on trial conduct.
Adverse event reporting and harms
The product for spraying outdoor walls is already registered in Malaysia and has been in use for vector control management. Both bio-actives of ADDs have short half-lives and are classified as low risk for non-target organisms. However, instructions about precautionary measures to be taken before and after the intervention are provided in the posters (supplementary material appendix 2) and during community engagement meetings. In addition, a dedicated phone number is provided by the research team to be contacted in the case of an event which might be linked to the interventions. We do not plan to set up a data monitoring committee for the purpose of this cRCT. The trial interventions do not include drugs, and no interim analysis is planned. On ethical review in Malaysia, the study was assessed as no more than minimal risk.
Plans for communicating important protocol amendments to relevant parties
Any modification in the protocol that significantly affect the scope or the scientific quality of the investigation will be submitted to the ethical committee with an amendment containing a verbatim description of the changes.
Dissemination plans
The results of the trial will be disseminated through international scientific conferences and submitted to highly ranked peer-reviewed journals for publication. The findings will also be shared with regional, national, and international stakeholders and partners. Reporting will follow the guidelines in the Consolidated Standards of Reporting Trials (CONSORT) Statement [37]. Authorship will follow guidelines established by the International Committee of Medical Journal Editors (http://www.icmje.org/) which require substantive contributions to the design, conduct, interpretation, and reporting of a trial.