Table 2 Interventions and main findings of the 18 cluster randomised controlled included in the systematic review

Camargo (2002) Brazil [36]

Sep 2000 – Jun 2001

1% temephos applied to HH water containers 3 monthly.

Community removal of “removable” water containers.

17,994

Community removal of “removable” water containers.

37,955

BI, CI measured monthly for 10 m in 300 HH randomly selected in both intervention and control clusters

BI and CI slightly lower in control cluster than intervention cluster at most time points. Both BI and CI were related to rainfall.

Not clear why temephos was not effective.

Kroeger (2006) Mexico and Venezuela [37]

MA

Mexico

Oct 2002 - Nov 2003

Venezuela Jan –Nov 2003

Mexico

Window curtains with lambdacyhalothrin and pyriproxyfen chips in water. (HH took out chips; not considered further.)

Venezuela

Window curtains and water jar covers treated with longlasting deltamethrin

1116

Control clusters received no interventions

1108

HI, CI, BI, PPI measured at baseline, 4w, 4 m & 12 m (Mexico) 9 m (Venezuela).

Adult dengue IgM serology at baseline & 8 m in Venezuela in approx 650 HH

At last measurement in Mexico and Venezuela, no significant difference in entomological indices between intervention and control clusters; but indices in all clusters significantly lower than at baseline. No such fall in nearby “external control” sites.

Levels of dengue IgM lower in intervention clusters than control clusters (p = 0.06)

The fall in control sites was due to spill-over effect. Insecticide treated curtains can reduce dengue vector levels and potentially dengue transmission

Lenhart (2008) Haiti [38]

MA

Jul 2003 –

Jan 2004

Insecticide (permethrin) treated bednets (ITNs) supplied to households

495

No treatment for 5 months; received ITNs after 6 months

522

BI, HI, CI, PPI measured at baseline, 1 m, 5 m and 12 m.

IgM dengue serology measured at 12 m

At 1 m, all indices fell in the ITN sites. By 5 m, indices had also fallen in control sites and were now lower than in ITN sites. Control HH near to ITN sites had lower indices.

At 12 m all sites had significantly lower indices and fewer IgM positives than at baseline.

Lack of difference between ITN and control sites due to spill-over. ITNs can reduce vector indices and potentially dengue transmission.

Ocampo (2009) Colombia [39]

MA

Apr 2004 – Jul 2005

In 3 intervention clusters used a. Lethal Ovitraps with deltamethrin (LO), b. Bacillus thuringiensis israelensis briquettes (Bti) and c. LO + Bti

Initial education of HH about dengue and vector breeding and bi-weekly visits of research team

240

Initial education of HH about dengue and vector breeding and bi-weekly visits of research team

80

HI, PHI, adult index measured at baseline and twice monthly for 4 m in 10 HH for each intervention cluster (total 30 HH) and 10 control HH

No difference between intervention clusters and the control cluster in any indices.

Control cluster indices not different from indices from all clusters measured in previous year, but those in all three intervention clusters combined were lower than those in the previous year.

Lack of difference between intervention and control clusters suggests initial education and repeated visits were as effective as the interventions. Small sample size was an issue.

Rizzo (2012) Guatemala [40]

Mar 2009 -Oct 2010

Intervention 1

Window & door nets treated with deltamethrin and water container covers treated with deltamethrin (wrong size so not used).

Govt programme treated water with 1% temephos in 3 intervention and 3 control clusters.

Intervention 2

After 17 months, nets replaced as needed and “productive” containers treated with temephos or discarded.

Govt programme continued as above

970

Govt programme treated water with 1% temephos in 3 intervention and 3 control clusters

865

At baseline, 6w after first intervention and 6w after second intervention measured total pupae, PPI, HI, BI and CI.

6w after first intervention, indices higher in all clusters than baseline. Total pupae and PPI increased more in control clusters but difference not significant.

6w after second intervention, indices were lower overall. Total pupae reduced significantly more in intervention clusters, borderline difference for PPI. HI reduced significantly more in intervention clusters, borderline difference for BI and no significant difference for CI.

The combination of insecticide treated curtains and targeting productive container types (with temephos and discarding containers) can reduce the dengue vector population.

Vanlerberghe (2013)

Thailand [41]

MA

Oct 2007 -Sep 2008

Window and door nets treated with long-lasting deltamethrin formulation.

Max 5 nets/HH. Insecticide supposed to last two years.

Routine government vector control.

2032 (80–110 hh/cluster)

Routine government vector control including temephos available to HH and deltamethrin spraying if case of dengue detected.

660 (10 hh/cluster)

BI, HI, CI and PPI measured at baseline, 6 m & 18 m.

All HH in control clusters; random half of HH in intervention clusters

At 6 m, BI was significantly lower in intervention clusters. HI, CI and PPI were also lower in intervention clusters. At 18 m, BI was no longer lower in intervention clusters, and nor were HI, CI and PPI. At 6 m, 71% of HH used the nets, but only 33% used them at 18 m.

Insecticide treated window and door nets can reduce vector breeding. The effect is coverage dependent.

Quintero (2015) Colombia [42]

MA

Jul 2013 – Mar 2014

Immediate: Window and door nets treated with deltamethrin in all 10 clusters.

After 8 m in 4 clusters: Water container covers treated with deltamethrin.

Routine government vector control activities continued.

922

Routine government vector control activities: temephos in water containers, health education, occasional malathion space spraying.

891

Measured at baseline, 9w after first intervention, and 4-6w after second intervention: CI, HI, BI, PPI

At first follow up indices fell more (cf baseline) in intervention clusters than control clusters; I-C difference significant for BI only. PPI increased in intervention clusters.

At second follow up all indices including PPI decreased more in intervention clusters; significant by t-test but not by non-parametric test.

The intervention package can reduce dengue vector density. Needs behaviour change for sustained effect.

Che-Mendoza (2015)

Mexico [43]

Mar 2011 –Oct 2013

Door and window screens treated with alpha-cypermethrin. After 14 m, productive containers also treated with spinosad every 2 m.

Routine government vector control continued.

1000

Routine government vector control: temephos in water containers, space spraying with chloropyrifos and propoxur.

1000

Measured BI, CI, HI & PPI at baseline and at 5, 12, 18 and 24 m Also measured adult mosquitoes.

Only adult mosquitoes less in intervention HH after the treated screens. At 18 m (after treatment of productive containers), BI, CI, HI and PPI significantly lower in the intervention clusters. At 24 m, only PPI significantly lower.

Insecticide treated screens & treatment of productive containers with spinosad can reduce vector breeding for up to 24 m

Kittayapong (2012)

Thailand [44]

MA

May –Nov 2010

Community mobilisation meetings and recruitment of ecohealth volunteers. Either copepods or Bacillus thuringiensis israelensis toxin (Bti) to HH water containers, plus screen net covers for containers. Education about dengue vector by ecohealth volunteers

441

No intervention

448

HI, CI, BI & PPI measured at baseline, 2 m, 4 m and 6 m.

Vector indices lower in all clusters than at baseline. No significant difference between intervention and control clusters in HI, CI, BI. PPI was significantly lower in intervention than control clusters at 2 m, 4 m & 6 m

It was feasible to implement the intervention in urban and peri-urban settings. Reduced the vector density (as judged by PPI)

Espinoza-Gomez (2002) Mexico [45]

Sep 1998 –Apr 1999

a. Education. House visits by university students, educational materials (eg calendars), group meetings with video + sociodrama (47 HH)

b. Chemical. ULV spraying malathion & temephos to water containers (46 HH)

c. Education & chemical (49 HH)

142

d. Control. No intervention (45 HH)

45

Baseline and 6 m. Measured BI, CI, HI and positive containers/HH (C+/H).

Reported on C+/H only.

Reduced from baseline to 6 m in Education only cluster, but not in Chemical or Control clusters. In the Education & chemical cluster, reduction from baseline was less marked.

Education intervention was effective but Chemical intervention was not. The Chemical intervention reduced the effect of the Education intervention, perhaps by false sense of security.

Vanlerberghe (2009)

Cuba [46]

MA

Jan 2005 –Feb 2006

>Stakeholder discussions, steering committee

>Community working groups, action plans

>Coordination between community and services

>Harmonisation with local vector control plan.

Government routine vector control programme continued.

8422

Government routine vector control programme: House inspections, temephos to water containers, space spraying with cypermethrin or cloripyriphos, health education, fines for law infringements

10,748

HI, BI and PPI measured at several points between baseline and end at 15 m

The HI, BI and PPI were not different between intervention and control clusters at baseline. At 15 m, HI, BI and PPI were all significantly lower in intervention clusters compared with control clusters.

A community based environmental management strategy on top of routine programme reduced dengue vector indices.

Arunachalam (2012)

India [47]

MA

Jun 2009 – Dec 2010

>Stakeholder consultation meetings

>Involvement of women self-help groups

>Mobilisation of schools, teachers & schoolchildren

>Communities distributed locally-made container covers and educational materials.

Routine government control services.

1000

Routine government control services only. Some of the trial educational materials

1000

CI, BI, HI and PPI measured at baseline, 5 m and 10 m

At 10 m there were significant reductions in the HI, BI, CI and PPI in the intervention vs control clusters.

A community-based approach involving multiple stakeholders to implement control actions reduced dengue vector indices.

Abeyewickreme (2012),

Sri Lanka [48]

Feb 2009 – Feb 2010

>Building partnerships of local stakeholders

>Household solid waste management promoted by HH volunteers

>Promoting composting of biodegradable waste

>Improvement of local gov rubbish collection

803

Local government services

790

Measured PPI, HI, CI and BI at baseline, 3 m, 9 m and 15 m

No significant differences between intervention and control clusters for HI, CI. BI significantly lower at 15 m. PPI significantly reduced in intervention clusters.

Household and community involvement helped reduce solid waste containers which are major site of dengue breeding.

Castro (2012) Cuba [49]

Oct 2004 –Dec 2007

Participatory strategy:

> Organisation and management structures  > entomological risk surveillance

>capacity building at local & intermediate level

>community work in vector control, led by community working groups (CWGs) who visited HH, planned actions

Government routine vector control programme continued

389

Government routine vector control programme: House inspections, temephos to water containers, space spraying with cypermethrin or cloripyriphos, health education, fines for law infringements

390

BI measured monthly from government surveillance figures before and during intervention from mid 2005 to Dec 2007.

Over the intervention period, the BI remained significantly lower in the intervention clusters than in the control clusters; the difference was bigger after the CWGs began their activities.

The empowerment strategy increased community involvement and added effectiveness to routine vector control.

Caprara (2015) Brazil [50]

Jun 2012 – May 2013

>Community workshops

>Mobilising elders and schoolchildren for solid waste management

>Government workers encouraged covering water containers

>Educational materials

1689

Routine government vector control programme.

1580

HI, CI, BI, PPI measured at baseline and 6 m

All indices significantly lower in the intervention clusters at 6 m.

Social participation and environmental management is feasible and significantly reduced vector indices.

Mitchell-Foster (2015)

Ecuador [51]

Nov 2012 – Nov 2013

An integrated intervention strategy (IIS)

>Elementary school education programme

>Clean Patio Safe Container programme with community volunteer activators

993

Government control programme:

>Initially temephos and space spraying with insecticide

>Midway, changed to biolarvicide (Bti) and HH education for source reduction

993

HI, BI and PPI measured at baseline and 12 m

PPI was significantly reduced in intervention clusters vs the control clusters (now with Bti) but only when clusters without full implementation were excluded.

Complicated by change in government programme midway through trial period. Need to explore integration of biolarvicide with the IIS approach.

Basso (2015) Uruguay [52]

MA

Nov 2012 – Apr 2013

Campaign with community members & health institutions for removal of water containers around households (bags with containers collected). Engagement of community opinion makers, leaflets, & press conference.

1000

Routine removal of the containers by services

1000

BI, CI, HI, PPI & PHI measured at baseline and 5 m (1 m after intervention)

The increase in indices from dry to wet season was less in the intervention communities but the difference was not statistically significant.

Low vector densities meant sample size did not have sufficient power to detect differences as significant.

Andersson (2015) Nicaragua and Mexico [53]

MA

Jul 2010 - Feb 2013

Community discussions of baseline evidence on vector breeding sites & infection in children. Community groups planned actions: HH visits by community brigades, school activities, & community clean-up activities and events.

Government control programme continued.

9529

Government dengue control programme: temephos in HH water containers & peridomestic space spraying.

9309

HI, CI, BI, PPI & IgM dengue saliva serology measured at baseline, 12 m, and 15 m (Mexico) 17 m (Nicaragua)

All vector indices significantly lower in intervention than control clusters in follow up survey.

Dengue infection rates in children aged 3–9 years (paired saliva samples) and self-reported dengue cases significantly lower in intervention than control sites.

Evidence based community mobilization effective for dengue vector control. Tailored implementation for individual sites gives local customization & strong community engagement.