Interventions against Aedes/dengue at the household level: a systematic review and meta-analysis

Summary

Background

Because the evidence for the role of structural housing and combinations of interventions (domestic or peri-domestic) against Aedes mosquitoes or dengue is still lacking, this systematic review and meta-analysis aimed to analyse and synthesize research focusing on the household as the unit of allocation.

Methods

We searched MEDLINE, LILACS, and Web of Science databases until February 2023 using three general keyword categories: (1) “Aedes” or “dengue”; (2) structural housing interventions including “house”, “water”, or “drainage”; and (3) vector control interventions of potential relevance and their combinations. We performed a qualitative content analysis and a meta-analysis for 13 entries on dengue seroconversion data.

Findings

14,272 articles were screened by titles, 615 by abstracts, 79 by full-text. 61 were selected. Satisfactory data quality allowed for detailed content analysis. Interventions at the household level against the immature mosquito stages (21 studies, 34%) showed positive or mixed results in entomological and epidemiological outcomes (86% and 75% respectively). Combined interventions against immature and adult stages (11 studies, 18%) performed similarly (91% and 67%) while those against the adult mosquitoes (29 studies, 48%) performed less well (79%, 22%). A meta-analysis on seroconversion outcomes showed a not-statistically significant reduction for interventions (log odds-ratio: −0.18 [−0.51, 0.14 95% CI]).

Interpretation

No basic research on housing structure or modification was eligible for this systematic review but many interventions with clear impact on vector indices and, to a lesser extent, on dengue were described. The small and not-statistically significant effect size of the meta-analysis highlights the difficulty of proving effectiveness against this highly-clustered disease and of overcoming practical implementation obstacles (e.g. efficacy loss, compliance). The long-term success of interventions depends on suitability, community commitment and official support and promotion. The choice of a specific vector control package needs to take all these context-specific aspects into consideration.

Funding

This work was funded by a grant from the World Health Organization (2021/1121668-0, PO 202678425, NTD/VVE).

Research in context.

Evidence before this study

Mosquitoes are the vectors that transmit such diseases as malaria, dengue, zika, and many others. Vector control is an integral part of disease prevention and includes interventions that reduce contact between mosquitoes and humans. These aim, for example, to eliminate mosquito larvae breeding sites, to put in place physical barriers against adult mosquitoes or to modify houses structurally in order to hinder mosquitoes from entering human dwellings. In the case of malaria control, housing improvements and vector control methods focusing on the house have been investigated systematically. Such reliable evidence is still, however, lacking in the case dengue and its vector, the Aedes mosquito. Formulating high-level policy recommendations on house improvement for dengue control remains difficult.

This systematic review aims to analyse and synthesize all research on the impact of structural housing and domestic vector control interventions against Aedes mosquitoes and dengue, with the household as unit of allocation, and with interventions that can be implemented by the household.

Methodology

A search for peer-reviewed articles was performed in English on three relevant databases (MEDLINE, LILACS, Web of Science) covering publications up to February 2023. The search terms were selected to reflect the following: the use, domestically or peri-domestically, of insecticide sprays, repellents, or insecticide-treated materials, and the treatment, alteration, or elimination of Aedes breeding sites with chemical or biological treatment against adults or larvae, as well as the physical barriers against mosquitoes. The terms included: (1) “Aedes” or “dengue”; (2) “house” or “housing” or “water” or “drainage”; and (3) specific terms about interventions of potential relevance for a single household (e.g. insecticide “Pyriproxyfen”). Inclusion criteria were: 1) focus on Aedes mosquitoes or dengue as a disease; 2) a clear methodology including at least one control element (e.g. randomized control trials); 3) focus either on structural housing aspects or on interventions in or around the house; 4) the house as the unit of allocation, allowing the household members to apply the interventions autonomously for the most part and 5) interventions carried out in a field situation and focused on community effectiveness. Studies were excluded when not meeting the above criteria or when published in a non-peer reviewed journal. We performed a qualitative content analysis on all selected articles and a meta-analysis for 13 entries with data on serological evidence of exposure to dengue virus.

Added value of this study

No articles on housing structure or modification were eligible for selection in this systematic review; rather a range of publications describing interventions with clear impacts on mosquitoes and, to a lesser extent, on the actual disease (dengue) were included. This systematic review showed that approaches targeting the aquatic stages of mosquitoes are more effective and remain functional longer compared to measures targeting adult mosquitoes only. Combined interventions against both immature and adult mosquito stages improved outcomes when implemented thoroughly and regularly. The meta-analysis results showed a small and not-statistically significant effect. This highlights the difficulty of proving effectiveness of control measures when dengue incidence is low or varies considerably (i.e. geographical and temporal hot spots).

Implications of all the available evidence

The decision to adopt a specific vector control package needs to take in account context specific issues, such as community sensitization and existing routine vector management. When properly implemented, classical approaches applied at household level, as in many of those presented here, contribute to the reduction of Aedes infestation and dengue cases but do not ensure full mosquito control. In addition, overcoming obstacles to practical implementation (e.g. efficacy loss, insufficient compliance) is still a challenge. The long-term success of interventions depends on suitability, community commitment and official support and promotion.

In the future, alternative approaches or technical innovations, if effective and sustainable, may help to further depress dengue incidence. Basic research on the potential of a change in housing structure leading to a reduction in dengue incidence is lacking and may be a beneficial topic to pursue.

Based on the results of this systematic review, we recommend for further research the following:

• i.Outcome measures should not be limited to entomological indicators but should include epidemiological indicators (such as serological conversion or disease cases) where possible;
• ii.Background dengue endemicity needs to be reflected in the study protocol;
• iii.The effectiveness of intervention should be measured for at least 12 months and preferably longer, and the means of maintaining it over time (e.g. three years) to ensure sustainability should be documented;
• iv.There must be further technical innovation and logistical availability of technical supply, serving and acknowledging residents' needs;
• v.Community-level research on housing structure and house modification focusing on the reduction of Aedes vector and its transmitted diseases is still lacking; here, considering vector adaptation and biology might be relevant.

Introduction

Dengue is an arboviral disease, mainly transmitted by mosquito species Aedes aegypti, Aedes albopictus and, in specific regions, by other species of the genus Aedes. It extends predominantly throughout Asia and South America but is becoming an increasing risk in many parts of Africa and the Middle East.^1,2 The global burden of dengue represents almost 400 million infections each year, of which nearly 100 million manifest apparent symptoms.³ The economic loss was estimated as high as US$ 8–9 billion in 2013.⁴ Until today there are only symptomatic treatment options for dengue cases available. Dengue vaccine has been licensed in some endemic countries; however, safety and effectiveness continue to be challenging. Additionally, Aedes mosquitoes are competent vectors of other diseases such as Zika, Chikungunya and yellow fever; against this background Aedes vector control interventions continue to remain crucial.5, 6, 7

Since 2004 the World Health Organization (WHO) has advocated the use of integrated vector management (IVM)⁸ instead of single interventions. After nearly two decades of IVM implementation, vector control still falls short of expectations for effective, sustainable disease control. Consequently, IVM is now positioned as a basic building block of a larger process that emphasizes strengthening of infrastructure and human capacity, and is linked to sustainable development goals.⁹ In this context, WHO emphasizes the need for “evidence-based recommendations on housing and vector-borne diseases”.¹⁰ For malaria, housing improvements and vector control methods focusing on the house as a unit of allocation have been investigated systematically, thus providing good evidence of basic research as well as summary evidence and policy recommendations.¹¹ Several interventions that aim at protecting the house against vectors have also been shown effective against dengue and other neglected tropical diseases, such as Chagas disease and lymphatic filariasis, while housing improvements have shown no, or weaker, effects.¹² Regarding dengue specifically, there is some basic research and some summary evidence on the efficacy and community effectiveness of vector control interventions focusing on the house as the unit of allocation.¹³ However, evidence on the role of structural housing aspects and combinations of interventions is still lacking and formulating high-level policy recommendations remains difficult.

Therefore, this systematic review aims to analyse and synthesize all research on structural housing aspects or domestic or peri-domestic vector control interventions against Aedes mosquitoes or dengue, with the household as the unit of allocation, and with interventions applicable by the household. Predefined outcome variables are stated in the methodology section.

Methods

Search strategy, databases and search terms

This systematic review followed the updated PRISMA guidelines for systematic reviews and meta-analyses.¹⁴ The searches were performed from May to July 2021 with an update done in February 2023, using the following databases: 1) Lilacs; 2) MEDLINE (PubMed); 3) Web of Science. No search restrictions on publication time, language, setting or population were applied. The search was carried out in English. A subsequent manual search was conducted by inspecting all reference lists of articles eligible for inclusion.

The search terms were derived to reflect the following: The use, domestically or peri-domestically, of insecticide sprays, repellents, or insecticide-treated materials, and the treatment, alteration, or elimination of Aedes breeding sites with chemical or biological treatment against adults or larvae, as well as physical barriers that hinder mosquitoes from entering the house, or any structural house modifications. and categorized as follow: (1) “Aedes” or “dengue” (2) “house” or “housing” or “water” or “drainage” and (3) vector control interventions of potential relevance for a single household. Terms were searched in titles and abstracts and by MeSH-terms if applicable. For the detailed search performed in PubMed see Supplementary Table S1.

Duplicates were removed within Zotero (www.zotero.org). Selection of articles based on title and the abstract screen was carried out independently by three researchers (CAMQ, VRL and SRR) using Rayyan (https://www.rayyan.ai) and discrepancies in article inclusion were resolved concordantly. The full text of the remaining articles was assessed by application of all inclusion and exclusion criteria (Fig. 1).

Fig. 1.

PRISMA flow diagram for study selection, adapted from Page et al., 2020; numbers include original search done + update of searches done from January to February 2023.

Inclusion criteria were: 1) focus on Aedes mosquitoes or dengue as a disease; 2) follow a clear methodology including studies containing at least one control element (e.g. randomized control trials (RCT), cluster randomized control trials (c-RCT), before-after studies); 3) focus either on structural housing aspects or on interventions in or around the house; 4) consider the house as the unit of allocation, allowing the household members to apply the intervention(s) for the most part autonomously and 5) consider interventions carried out in field situations and focused on community effectiveness (as defined by Burches and Burches).¹⁵ Studies were excluded when not meeting the criteria above or when published in a non-peer reviewed journal.

Data extraction, management, analysis and synthesis

Data were extracted, using a pre-piloted data extraction form on the following a priori defined topics: 1) year of publication, geographical area, study aim; 2) type of study, sample size (number of houses), type and details of interventions, 3) rural or urban area, start date, duration, follow up period (if any), 4) entomological outcomes: vector species, method(s) of mosquito collection, vector indices of immature stages and adult mosquitoes; 5) disease outcome: number of disease cases or seroconversion and 6) reported entomological and disease outcomes, statistical significance, reported impact of intervention, limitations, and conclusions. Studies were subsequently grouped by intervention types into categories for 1) interventions on the structural parts of the house beyond windows and doors (e.g. roof, eaves, gutters, construction material, differences of constructions), 2) interventions against immature mosquito stages (e.g. source reduction, biological predators, chemical larvicides), 3) interventions against adult mosquitoes (e.g. container covers, treated fabrics, insecticides, adult traps), and 4) interventions against both immature and adult mosquito stages (e.g. adult traps and ovitraps, source reduction combined with screen covers and biological predators). Interventions involving curtains and screens were classified as interventions against adult mosquitoes. As this is a systematic review following qualitative data analysis, statistical significance provided in the publications was used as one outcome variable to qualitatively categorize and illustrate the effect of the interventions under investigation. Heterogenous backgrounds such as different vector control methods investigated, different epidemiological backgrounds and seasons, different study designs, and varying statistical analyses and outcome measures precluded any attempt to pool data for quantitative effects size analysis or to perform a meta-analysis. Intervention sustainability was assessed by the duration of the observed effect in the short-term (2–5 months), mid-term (6–12 months), and long-term (>12 months).

Meta-analysis

Studies that included epidemiological outcomes were considered for a meta-analysis because of their public health relevance. Too few studies (only 4) dealt with actual dengue cases but 14 presented results on dengue seroconversions (IgM for recent or IgG for new exposure, or both). Eleven studies provided adequate data to be included. Two studies provided two separate entries by including distinct geographical areas or time points.^16,17 The studies were comparable but with notable variations in the study population (children only or a mix of adults and children), and serological approaches (blood or saliva samples; IgM, IgG or both). Therefore, a random effect model was used. The meta-analysis was performed using STATA/BE 17.0.

Quality assessment

The CONSORT checklist (www.consort-statement.org/) was used for quality assessment expressed as a percentage of the maximum score. For RCTs and c-RCTs we applied all 25 items of the checklist. For non-RCTs the checklist was reduced to 19 items, excluding the items specifically dealing with the randomization out.

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Results

Descriptive analysis

Search results

The initial search identified 20,936 articles; 5785 additional articles were identified with the update of the search. After removing duplicates, 14,272 were screened by title, 615 by abstract, and 79 articles by full text to finally include 61 articles (Fig. 1). No additional article was selected after screening the reference lists of the included studies.

Study characteristics

The 61 studies were conducted in a total of 28 countries (Fig. 2) and included a multi-centre study presenting data from eight countries. Most studies took place in the Americas (11 countries) or Asia (12 countries), while three countries were represented in Africa and two in Europe.

Fig. 2.

Geographical locations of selected studies, with number of studies per country.

Most included studies were published in the last 20 years and the variety of interventions increased after 2004 (Fig. 3). Over half of the studies (36, 59.0%) were conducted in urban areas, of which 3 took place in high-rise buildings.

Fig. 3.

Timeline of studies with number and type of interventions indicated. A study may include more than one intervention.

The included studies tested interventions that aimed either at the immature stages of mosquitoes (21 studies, 34.43%), at the adult mosquito stage (29 studies, 47.54%), or both simultaneously (11 studies, 18.03%). No study specifically investigating structural housing parts was eligible for inclusion. The most employed interventions were insecticide treated fabrics and curtains (20 studies, 32.8%), followed by biological predators (21.3%), then source reduction and chemical larvicides (both 12 studies, 19.7%), covering of water containers (18.0%), and traps (16.4%). The remaining three types of interventions were insecticide sprays or emanators (11.57%), Bti (8.2%) and autodissemination (3.3%) for two studies including one using pyriproxyfen (Fig. 4). Mosquito species targeted during interventions were Aedes aegypti in 41 studies (67.2%), Aedes albopictus in 2 studies (3.3%), or both in 12 studies (19.7%). In six studies (9.8%) the species was not specified. RCTs were conducted in 29 studies (43.4%). Community involvement was not considered an intervention in itself but an enabling factor. It was used for supporting the interventions in 18 studies (29.5%), mostly in interventions against aquatic mosquito stages (61.1%) and in combined interventions (27.8.7%), but rarely in interventions against adult mosquitoes only (11.1%) (Fig. 4).

Fig. 4.

Overview of the characteristics of the included studies. ∗Venezuela, Mexico, Peru, Kenya, Thailand, Myanmar, Vietnam, Philippines. Colours are visual aids with no additional significance; ^a RCT = randomized control trial, Iv.C = intervention vs control, B&A = before and after intervention, Q-ex = quasi-experimental; ^bAedes: 1 = aegypti, 2 = albopictus, 3 = both, 4 = not specified; ^c Rural/urban: 1 = rural, 2 = semi-rural, 3 = urban, 4 = urban-high rises; ^d Intervention type: 1 = immature stages (eggs, larvae, pupae), 2 = adult mosquito. 3 = both; ^e Results (reductions are expected): 0 = no statistical differences between intervention and control, −1 and 1 = statistical differences, respectively with worse or better outcome in intervention group, 0&1 = mixed results depending on indicator or geographical area, NDt = not detected; ^fSustainability: 0 = effect no longer observed, 1 = effect still observed; X = measured but not applicable because treatment was ineffective; Other categories: 1 = “yes” or in the study. The total indicates the number of occurrences in a column regardless of values and the percentage is computed for N = 61.

All studies except one¹⁸ used entomological indicators to measure the intervention outcome but only 16 (26.2%) used seropositivity or dengue cases as endpoints. The most common way of assessing intervention success was to estimate the abundance of immature mosquito stages (49 studies, 80.3%). Adult mosquitoes were used as outcome indicators in 21 studies (34.4%). Few studies used epidemiological indicators as outcome measures with 14 studies (23.0%) reporting serological test results and four (6.6%) reporting the number of dengue cases.

Quality analysis

Nearly two thirds of studies (40/61) scored above 65% in the CONSORT quality analysis (Fig. 4), providing reliable data and results. Low scores reflected early or simple studies and did not impede content analysis. Studies used in the meta-analysis had a higher median score (76% vs. 68%) with all but one (10/11) scoring above 70%.

Content analysis

Studies were categorized according to the type of vector intervention at the household level.

Single or combined interventions exclusively targeting the immature mosquito stages

A simplified extraction table of these data is presented in Supplementary Table S2.

Source reduction

Seven studies used larval source reduction as the sole intervention, supported by community involvement. Four studies19, 20, 21, 22 showed a reduction in larval and pupal indices. The three other studies23, 24, 25 showed a trend in intervention areas but did not achieve significant reduction levels.

Biological predators

All ten studies using aquatic predators except two^26,27 were single intervention studies. Nine studies showed statistically significant positive outcome. Tikasingh and Eustace showed a significant decrease in Ae. aegypti catches using Toxorhynchites larvae.²⁸ Tun-Lin et al. reported data from Myanmar on the use of dragonfly nymphs and larvivorous fish, which reduced Aedes aegypti larvae to a very low level in the first two to three weeks and suppressed them until the trial ended.²⁶

Four additional studies used larvivorous fish and significantly reduced entomological indices.

This included reducing the house index, container index and Breteau index,²⁹ decreasing the container index,³⁰ and a sharp decrease in the Breteau index.³¹ Wu et al. additionally noted the lack of reported dengue cases in the intervention areas, while the control areas experienced an outbreak between the study and publication time.

Copepods were used in four studies: Three studies from Vietnam32, 33, 34 achieved Aedes larval reductions in treated containers. Kay et al. registered a decrease in the number of adult mosquitoes by between 30 and 100%.

Only one study used copepods³⁵ and showed no significant impact on Aedes populations.

Bacillus thuringiensis israelensis (Bti)

Tun-Lin and team present data from Thailand on Bti-use combined with interventions against larvae (source reduction and Temephos).²⁶ The differences were statistically significant and more pronounced when looking at the pupae per person index (PPI) with a reduction of 14.8% under targeted control vs. 48.6% under non-targeted control.

Chemical larvicides

Four studies employed chemical larvicides as an intervention against Aedes larvae and showed statistically significant positive outcomes. Two studies from Laos,^18,36 used pyriproxyfen in the form of SumiLarv 2MR disks and reported significant reductions in a) the number of containers infested with Aedes aegypti larvae (baseline container infestation of 20.2% and a decrease, during intervention to 4.1% and later to 2.8%.) and b) reduction of anti-DENV IgG antibodies in the sera of volunteers. A study in Cambodia³⁷ administered pyriproxyfen strands to storage jars and the achieved inhibition of adult emergence was 0% and 2.0% in the intervention and control areas at baseline and reached 99.8% in treated areas. The multi-site study²⁶ equally investigated the use of chemical larvicides (Temephos, pyriproxyfen) at four sites, Venezuela, Peru, Kenya, and Thailand in combination with different interventions.

Autodissemination

Autodissemination targeting the larval stages was used in one study,³⁸ where the investigators fitted modified ovitraps with pyriproxyfen. A statistically significant reduction in the ovitrap index from 90 to 33% was observed by week 20, and the number of reported dengue cases reduced from 53 in 2013 to 13 in 2014 but increased again after cessation of the trial.

Single or combined interventions exclusively targeting the adult mosquito stage

A simplified extraction table of these data is presented in supplementary Table S3.

Container covers

Five studies used container covers as an intervention and all produced statistically significant positive outcomes. Chuang et al.³⁹ delivered untreated fine net screens to farmers to cover rainwater buckets. The intervention led to a significant reduction in the container index (from 5.88 to 1.63 and from 2.33 to 1.2). Kusumawathie et al.⁴⁰ similarly employed untreated covers for ground level water storage cement tanks. Significant reductions in the mean number of tanks containing Aedes mosquito larvae in the net-covered tanks were observed. In Cambodia, jar covers distributed free of charge led to a decrease in PPI from 6.9 at baseline to 2.1 after 13 weeks, and increased to 2.8 after 22 weeks.⁴¹ The corresponding rates in the control areas were 7.3, 6.4, and 3.9, respectively. Two studies^42,43 used long lasting insecticidal nets (LLIN) covers for water containers in conjunction with insecticide treated curtains (see section on treated fabrics).

Treated fabrics

Most interventions against adult mosquitoes were in the form of insecticide-treated cloths and curtains, investigated in 15 studies. All but two^17,44 were single interventions.

Ten studies showed statically significant positive outcomes (three with mixed results). The use of LLIN in Haiti led to a decrease in entomological indices and in serological indicators (+Dengue IgM from 33.7 at baseline to 18.5 after one year of intervention)⁴⁵ as did a study in Venezuela.¹⁷

Che-Mendoza et al. 2018⁴⁶ employed Duranet LLINs mounted in aluminium frames, custom fitted to doors and windows. A significant reduction in the presence (odds ratio (OR) = 0.48) and abundance (incidence rate ratio (IRR) = 0.45), of indoor Ae. Aegypti was achieved. The team around Manrique-Saide conducted three studies in Mexico using impregnated nets mounted in frames to cover doors and windows. In one study,⁴⁷ the number of treated houses infested with Ae. aegypti adult females were significantly lower at 5 months postintervention than the number of control houses (OR = 0.38). The later study⁴⁸ achieved significant reductions in the total abundance of adult female Ae. aegypti (incidence rate ratio, IRR = 0.12). A third study using fiberglass nets mounted in aluminium frames showed significant reduction of indoor Aedes mosquito density and Aedes aegypti infection with Aedes-borne viruses (DENV and ZIKV) (OR = 0.29, 95% CI 0.10–0.86, p = 0.02).⁴⁹

Quintero et al.⁴² tested insecticide treated curtains on windows and doors showed a reduction in entomological indices in intervention clusters compared to controls, with only significant differences in the Breteau index. After adding LLIN covers for water containers, a significant reduction in PPI was recorded, showing a decline of 71% compared to 25% in the control group. In India, Ansari et al. used deltamethrin-treated curtains⁵⁰ showing significantly reduced indoor resting density of Ae. aegypti, Anopheles stephensi, and Culex quinquefasciatus. Densities of Ae. aegypti were reduced by 91.9% after the first intervention and 96.3% after the reimpregnation.

Vanlerberghe et al. tested insecticide treated materials in Thailand, and Venezuela.^43,51 Both studies deployed ITCs but only the one in Venezuela additionally tested container covers. Both studies found a sustained decrease in the Breteau index. However, this reduction was dependent on the coverage of insecticide-treated curtains. No attributable effect was found from insecticide treated jar-covers.⁴³ In the study in Thailand, the protective effect declined and was no longer associated with the intervention after 18 months when ITC coverage had decreased to 33.2%.⁵¹

Six studies showed no change.

The effect of impregnated mosquito bed nets on mosquito biting rates was investigated in the Democratic Republic of Congo.⁵² Biting rates for Ae. aegypti were zero but had been very low prior to intervention.

The deployment of insecticide treated curtains (ITCs) in Thailand and Peru did not lead to a significant effect on vector indices,⁵³ nor did it offer protection against exposure to dengue virus infections.⁵⁴ Two studies conducted in Cuba used long lasting ITC,⁵⁵ one of them a combination of ITC and residual insecticide treatment.⁴⁴ The setting was of low Aedes infestation and neither study reduced mosquito levels.

Loroño-Pino et al.¹⁶ compared ITCs with untreated, but otherwise identical, window curtains. In some trial areas dengue virus infection prevalence tended to be lower in ITC homes (18.6%) compared to homes with non-treated window curtains (24.2%) but was not statistically significant.

Insecticides

Four studies investigated the use of insecticides that the household can deploy as single intervention. All four showed statistically significant positive outcomes but one⁵⁶ was very short-lived. Devine et al. used passive metofluthrin emanators, small impregnated pieces of net hung in rooms to release active volatiles into the air.⁵⁷ The abundance rate ratio for Ae. aegypti adults averaged over all seven deployment cycles was reduced 2.57-fold (average reduction of 60.19% for adults). In Malaysia researchers used metofluthrin impregnated polyethylene plastic strips.⁵⁸ In intervention houses mosquito densities were significantly lower after treatment with an effect less than eight weeks. Morrison et al. used transfluthrin passive emanator special repellent (SR) but found no statistical difference between incidence in the SR and placebo. Significant reduction in indoor female Ae. aegypti abundance and blood feeding was found.⁵⁹ One study⁵⁶ showed a short-term statistically significant reduction in adult Ae. albopictus abundance using commercially available space sprays based on pyrethroids available to individual households. However, mosquito populations recovered to pre-treatment abundance 10 days post-treatment.

Adult traps and ovitraps

Six studies employed adult or ovitraps and all were single interventions. Two studies showed statistically significant mosquito reduction. Barrera et al.⁶⁰ used the autocidal gravid ovitrap (AGO trap) for adult vector control. Reductions in the number of female Ae. aegypti captured in Biogents traps and sentinel AGO (SAGO) traps attributable to the presence of the AGOs were 53 and 70%, respectively. In a study conducted in Manaus, Brazil,⁶¹ intervention households received one Biogents Sentinel trap. The Ae. aegypti adult density significantly decreased from 1.35 to 0.62 (reduction of 54%) in the intervention arm. Reductions were observed immediately after installing the traps. The detection of recent dengue measured by serological conversion showed a non-significant reduction.

Three studies showed mixed results. A study placing lethal ovitraps in homes in Brazil,⁶² observed significant reductions in the number of containers containing Ae. aegypti larvae, and in the pupal densities (number/house). The number of adult Ae. aegypti females per house were generally higher post-treatment. Alarcón et al.⁶³ used ovitraps with Bti to control and survey Aedes aegypti populations in Colombia. Significant differences were found between the household index in treated and control neighbourhoods, however, the container index did not show this clear trend and the Breteau index, for one month, was even higher in the intervention households. Ligsay et al. placed Ovitraps laced with PPF (Ovi-PPF) in selected households at the Philippines. Decrease in CI was significant post intervention, not so ovitrap index and seroconversion rate.⁶⁴

In Brazil, Degener et al.⁶⁵ used MosquiTRAP traps to reduce Ae. aegypti. Impact evaluation through MosquiTRAP mass trapping collected significantly more female Ae. aegypti than in the control arm.

Integrated vector management approaches, targeting immature and adult mosquito stages simultaneously

A simplified extraction table of these data is presented in supplementary Table S4.

The intervention in eleven studies targeted both immature and adult vector stages, in an integrated approach. The most widely used method was the treatment of breeding sites with chemical larvicides, combined with various other interventions against immature and adult stages.

Eight studies showed statistically significant results. A study in Spain⁶⁶ used four complementary interventions and recorded a significant reduction in mosquito eggs by ovitraps. Barrera et al. achieved through source reduction in combination with a one-time chemical larviciding and AGO traps a reduction in adult mosquito densities in houses of between 84.3 and 92.4%.⁶⁷ In Mexico Duranet screens fitted to doors and windows together with the treatment of water tanks with the larvicide Natular resulted in significantly fewer infestation of houses with Ae. aegypti adult females (OR = 0.38); 12 months post-intervention for adult females still (OR = 0.41) and males (OR = 0.41) but not for blood-fed females (OR = 0.51).⁶⁸ Kittayapong et al. (2008) combined source reduction through clean up campaigns with screen covers for water jars, copepods and Bti and permethrin-treated ovitraps. The proportion of IgG/IgM positive students in the treated areas was reduced from 13.46% to 0% whereas those from untreated areas increased from 9.43% to 19.15%.⁶⁹ Chemical larvicides (pyriproxyfen) were used in conjunction with container covers and insecticide treated window curtains in Mexico.⁷⁰ Even though the acceptance rate for pyriproxyfen was low the Breteau index fell from 60% to 7% (P < 0.001) in the intervention group and from 113% to 12% in the control group (P = 0.02). In Vietnam, impregnated container lids and pyriproxyfen were used in water containers.⁷¹ The container-index and house-index for immature Ae. aegypti fell steeply one month after treatment in the trial area, however, the dengue seroconversion rate was not changed. A study in Malaysia⁷² tested the combined effect of autodissemination and targeted outdoor residual spraying. As compared to the control site, the overall outdoor and semi-indoor ovitrap index was lower in the autodissemination intervention sites (−8.3%, P = 0.04). Bigio et al. investigated a combination of predators (guppy fish), gravid ovitraps, solid waste management, containers lids, and community engagement and achieved a significant reduction of entomological indicators.⁷³

Two Studies showed moderate or mixed results. Wai et al.⁷⁴ applied multiple interventions and the PPI decreased from 0.34 at the first evaluation to 0.23 at the last evaluation (32% reduction) in intervention clusters and from 0.33 to 0.15 (54.5% reduction) in routine service clusters. The container index decreased from 27.7 to 19.4, the Breteau index from 49.7 to 27.9, and the household index from 6.3 to 3.7. The same level of reduction was detected in the clusters where routine larviciding activities with Temephos were carried out. Rizzo et al.⁷⁵ used the chemical larvicide Temephos together with insecticide treated curtains. Additionally, water drums were covered with nets and buckets were emptied and disposable items eliminated. The impact of the interventions on vector abundance was found to be moderate. Stegomyia (Aedes) indices were about the same in intervention and control arms.

The study from Ocampo et al. in Colombia combining Bti briquettes with lethal ovitraps showed no significant differences in entomological indices in the untreated and treated areas.⁷⁶

Statistical significance and sustainability

Most studies (50, 82.0%) reported a statistically significant improvement measured by reducing at least one outcome indicator in the intervention area compared to the control area. In the 21 Interventions exclusively against the mosquito aquatic stages all but three studies (85.7%) showed significant reduction of entomological indicators and the three out of four studies that measured epidemiological outcomes reported decrease. Among the 29 studies that used interventions exclusively against adult mosquitoes, 23 (79.3%) showed some significant reduction in the entomological indicators, while only two in 9 (22.0%) demonstrated some decrease in the epidemiological indicators. Two studies (6.9%) showed a negative effect of the intervention. Combined interventions showed a reduction both in entomological (10/11, 90.9%) and epidemiological (2/3, 66.7%) indicators (Fig. 4, Table 1).

Table 1.

Overview of study results.

Category indicators	Type of vector intervention at household level
	Immature stage only				Adult mosquitoes only				Combined
	Entomol.		Epid.		Entomol.		Epid.		Entomol.		Epid.
Number and % of studies	N	%	N	%	N	%	N	%	N	%	N	%
Outcome
Positive	17	81%	3	75%	21	72%	1	11%	7	64%	2	67%
Mixed	1	5%	1	25%	2	7%	1	11%	3	27%
No change	3	14%			5	17%	6	67%	1	9%	1	33%
Negative					1	3%	1	11%
Total studies	21		4		29		9		11		3
Sustainability (studies reporting positive outcomes in the period)a
2–5 months	17	81%			20	69%			8	73%
6–12 months	15	71%			12	41%			4	36%
>12 months	6	29%			6	21%			4	36%
Total studies	21				29				11

Percentages are computed vertically over the number of studies by type of vector intervention.

^aPercentages for each sustainability period are computed over the total per study type.

The sustainability of the intervention, defined as its effectiveness over time, was estimated for the short-term (2–5 months), mid-term (6–12 months) and long-term (more than 12 months). Nearly all studies (58, 95.1%) checked for effectiveness of the outcome measure in the short-term, 47 (77.0%) in the mid-term and 25 (41.0%) in the long-term (Fig. 4). Evaluation of sustainability was not applied to the eight studies (13.1%) where the intervention proved ineffective. Five of the 45 remaining studies (11.1%) failed to show effectiveness in the short-term (Fig. 4). Half of the studies (31, 50.9%) demonstrated their sustainability between 6 and 12 months and about a quarter (16, 26.2%) for more than 12 months.

Meta-analysis

A meta-analysis of seroconversion cases included 13 entries that were sorted by ascending values of dengue incidence in the control arm to visualise dengue transmission pressure (Fig. 5). The overall result shows a not-significant reduction on the intervention side (log odds-ratio: −0.18 [−0.51, 0.14 95% CI]). Studies with higher dengue incidence in the control arm were more likely to show a seroconversion reduction. Several studies experienced high clustering of dengue seroconversion cases in time^17,65,69 or space^16,17 making it more difficult to show intervention effectiveness during low incidence episodes. Another factor documented by Lenhart 2020⁵⁴ was a false sense of security brought by a treated fabric intervention when the insecticide treatment became ineffective.

Fig. 5.

Forest plot with 13 entries from 11 studies sorted by ascending values of dengue incidence in the control arm (indicated in parenthesis).

Discussion

There is an increasing amount of (summary) evidence of Aedes vector control interventions available. Vector control interventions are mainly carried out by public health officials, or by the community itself, or by public health officials supported by the community. In this systematic review and meta-analysis, we pursued the hypothesis that effective vector control might benefit from empowerment of the smallest unit of allocation, namely the household. We considered community involvement as a means of delivery and an enabling factor rather than an intervention on its own. Here we focused on measures that can be taken at individual house level and by household residents. Acknowledging the crucial role of the household in the long chain of Aedes vector control, we addressed the question: “what can individual households do to effectively reduce dengue risk?”

The studies eligible for inclusion considered a variety of interventions against Aedes mosquitoes and their aquatic stages. When the impact of the individual interventions is considered, no single, clear frontrunner can be identified. Our results showed that interventions against the immature mosquito stages at household level (21 studies, 34.4%), all but three being single interventions, seemed to perform best in term of entomological and epidemiological outcomes, (85.7% and 75.0% respectively). Interventions against adult mosquitoes (29 studies, 47.5%) performed less well (79.3% and 22.0%, respectively). In other studies evidence suggested that multipronged interventions were generally more effective than single interventions. Our results showed that combined interventions (11 studies, 18.0%) performed similarly to (mostly) single aquatic interventions for entomological and epidemiological outcomes (90.9% and 66.7%, respectively) (Table 1).

These findings can be discussed in the light of several factors. Many interventions targeting adult mosquitoes rely on impregnated materials that lose effectiveness over time or require meticulous implementation, such as covering breeding containers. Interventions against eggs, larvae, or pupae also require consistent and repeated implementation, but generally appeared to be less susceptible to disruption of routine. Focusing on implementing one or few approaches well may be more effective than trying to split the effort among several strategies competing for resources. Novel approaches targeting adult mosquitoes, such as autodissemination traps, may still need to be optimized for routine use.

Mid-term and long-term sustainability of the effect of the interventions against immature stages was demonstrated respectively with 71% and 29%), against adult stages with 41% and 21%; and using combined interventions with (36% and 36%). Beyond the achieved immediate impact, sustained results depend on important such factors as simplicity, interval between required follow-ups, community acceptance and public health advocacy. Interestingly, sustainability was limited especially for interventions against adult mosquitoes and for several integrated approaches, while interventions against the aquatic stages performed better. Although the need for reapplication is expected in areas of dengue endemicity, sustainability will be favoured by interventions remaining effective longer. Determining the frequency of required follow-ups (e.g. on semi-annual or annual basis) and their acceptability by the community are crucial elements for the long-term success of a given intervention.

Demonstrating significant statistical reduction is easier in a context of high mosquito infestation, especially if the intervention is short-term and focused on entomological indicators. The meta-analysis shows that studies performed in areas of higher background endemicity were also more likely to prove effectiveness of interventions when serological indicators were used as outcome; in that case paediatric cohorts seems to be favoured because seroconversion was easier to diagnose. In contrast, the effectiveness of an intervention performed in a context of effective routine vector control with low mosquito infestation level and focusing on disease outcome (i.e. dengue cases) is more difficult to prove. This difficulty was mentioned in several of the studies included.^23,44,74

In several studies routine dengue vector control interventions were commonly conducted in the control arms. In general, when routine vector control interventions succeed in maintaining low levels of mosquito infestations, it is more difficult to prove the gain or effect of additional interventions, as illustrated by Toledo et al. (2017, 2015).^44,55 Suppressing mosquito populations through routine vector control is useful and necessary.^{19,20,22,23,60,66,67,69} Additional efforts, however, are needed to reduce the number of disease cases and to avoid outbreaks. Therefore, communicating individual risk-reducing behaviour is a crucial prerequisite and not a competing approach to centralized approaches. With the worldwide improvement of routine dengue vector control in the last decades, additional gain in term of mosquito and dengue reduction becomes more difficult to obtain. This may also explain why more recent and complex interventions did not outperform earlier simpler ones. The effectiveness of background vector control, however, can easily be overlooked when investigating additional interventions.^44,75 Relaxing the vigilance either because of public service disruption or because of overconfidence in a new intervention, often leads to an increase of cases.^54,65

The success of an intervention depends not only on its efficacy but on its implementation within the community. Among others, Bowman et al.⁷⁷ showed that routine interventions involving the community yielded significantly better results. Several studies^78,79 demonstrated that vertically structured centralized interventions combined with community mobilization and empowerment improved both short-term success and long-term sustainability of vector-control approaches. In this systematic review, local communities were involved in vector control activities targeting the house as unit of allocation in 29.5% of the studies. It was implemented mainly in interventions against the aquatic stages of mosquitoes (61.1%), specifically in source reduction and when using aquatic predators, and against both aquatic stages and adult mosquitoes (27.8%). Community involvement played only a minor role (11.1%) in interventions against adult mosquitoes only (Fig. 4). This may represent a lost opportunity for individual households, and therefore the community in general, which might still rely on traditional vector control methods focusing on immature stages, as advocated for decades by health authorities. Promoting ownership and responsibility of the community in general and also for interventions against adult mosquitoes may help to overcome the effectiveness and sustainability gap. Moreover, the success of an approach also depends on the ease of obtaining and handling vector control materials by individuals in the community. In particular, technical equipment such as specific mosquito traps or insecticides in formulations not readily available on the market make it difficult to implement routinely and to maintain interventions. Interventions such as community-organized removal of breeding sites or propagation and deployment of aquatic predators are less dependent on external actors. This highlights an ongoing need for further technical innovation and logistical availability of technical supply designed to serve residents’ needs.

We searched the scientific literature for all studies looking into the different aspects of making the house with its residents safer with respect to Aedes-borne or dengue infections. A similar approach was taken for malaria by Furnival-Adams et al., in 2021.⁸⁰ Interestingly, for dengue, there is not a single study that fulfilled the inclusion criteria and reported on house structure specifically, aspects of the immediate surroundings, gutters and eaves, construction material or differences of constructions in urban and rural areas. One publication on piped water interventions and three on space spraying interventions have been excluded from our systematic review because they required community-wide implementation and could not be assigned to a single household as the unit of allocation. Interventions such as waste management, water storage, and impregnated nets/lids emerged as elements of house modification. It is striking that in comparison to work on malaria, there are no research studies available on house structure issues. While this may be interpreted as a research gap, it may also reflect the behaviour of the day-biting Aedes mosquito. Because the Aedes mosquito follows the movement of residents, the focus of intervention may shift from structural feature to barriers hindering the mosquito entry during daily activities, as reflected in the use of impregnated nets/curtains considered as “soft construction measures”.

Strength and limitations of the study

This systematic review's strength is that it extensively searched the scientific literature with over 14,000 titles screened. Publication bias remains as an issue, since negative results are less reported. However, this bias has been addressed with a very broad search strategy, and manual searches. A limitation of this study was the difficulty of separating the effects of multi-intervention designs and sometimes of disentangling community application from household interventions. For example, indoor residual spaying (IRS) interventions were not included even though new targeted IRS approaches could be performed by house inhabitants, whereas standard IRS is currently delivered mainly by public officials because of health and environmental risks. This also relates to supportive measures such as the application of household insecticide aerosols.81, 82, 83 Another limitation resides in generally judging an intervention success on the basis of statistical significance because it is strongly dependent on the background level of mosquito infestation, the quality of routine vector management, and the primary outcome variable (entomological vs. epidemiological).

Conclusions and recommendations

No basic research on housing structure, modification or urbanization was found to be eligible for this systematic review, either because the study did not focused on an individual house but on a whole community or because there was no control element in the design. One of the most critical issues in evaluating interventions in a trial setting is certainly their successful transition into a sustained routine intervention. There is an arsenal of effective interventions with proven impact on vector indices and also, to a lesser extent, on disease. Based on the results of this systematic review, approaches targeting the aquatic stages of the vectors appear to be particularly promising in this regard. Compared to measures targeting the adult vectors, these are not only more effective, but also remain functional after longer follow-up observation. However, several studies showed that combined interventions against both immature and adult mosquito stages improved outcome when implemented thoroughly and regularly. The key elements of success include: the willingness and commitment of the community to consistently and regularly perform the required interventions; acceptability, affordability, ease of handling and sustainable delivery of adequate vector-control material; as well as vertical and motivational support. Therefore, the choice of a specific vector control package needs to consider all these context-specific aspects. For example: to favour multipronged approaches in the context of community sensitization and involvement in order to increase the chance of successful interventions; or to sustain routine vector management with an integrated panel of tested context-specific approaches.

When properly implemented, classical approaches applied at the household level, such as many presented here, contribute to the reduction of Aedes infestation and dengue cases but do not ensure full mosquito control. In the future, alternative approaches or technical innovations, if effective and sustainable, may help to further depress dengue incidence. Community commitment and motivation along with routine vector control are likely to continue playing an important role. Basic research on the role of the housing structure itself to ease the burden on communities is lacking and may offer new opportunities for dengue control.

Based on the results of this systematic review, we recommend further research on the following:

• i.Outcome measures should not be limited to entomological indicators but should include epidemiological indicators (such as serological conversion or disease cases) where possible;
• ii.Background dengue endemicity needs to be reflected in the study protocol;
• iii.The effectiveness of intervention should be measured for at least 12 months and preferably longer, and the means of maintaining it over time (e.g. three years) to ensure sustainability should be documented;
• iv.There needs to be further technical innovation and logistical availability of technical supply serving and acknowledging residents' needs;
• v.Community-level research on housing structure and house modification focusing on the reduction of Aedes vector and its transmitted diseases is still lacking; here considering vector adaptation and biology might be relevant.

Contributors

Conceptualization: CAMQ, VRL, OH, RV, PD, SRR.

Methodology: CAMQ, VRL, OH, PD, SRR.

Validation: CAMQ, VRL, OH, RV, PD, SRR.

Formal analysis: CAMQ, VRL, OH, PD, SRR.

Investigation: CAMQ, VRL, OH, PD, SRR.

Resources: OH, VRL, SRR.

Data curation: CAMQ, VRL, SRR.

Writing – original draft preparation: VRL, PD, SRR.

Writing – review & editing: CAMQ, VRL, OH, RV, PD, SRR.

Visualization: CAMQ, VRL, OH, PD, SRR.

Supervision: VRL, OH, VR, SRR.

Funding acquisition: OH, SRR.

All authors read and approved the final version of the manuscript.

All authors accessed and verified the underlying data.

Data sharing statement

All data are available in their respective articles.

Declaration of interests

We declare no competing interests.