Introduction

'Topical repellents like N,N-Diethyl-meta-toluamide (DEET ) or para-menthane-3,8-diol (PMD) are excellent tools to protect individuals against mosquito bites especially in areas where vectors usually bite early in the evening [1-4].These compounds interfere with the mosquito’s olfactory system preventing them from identifying their hosts [5] and obtaining a blood-meal [6]. However, repellents do not kill the host-seeking mosquitoes, they simply reduce man-vector contact. In order to be effective, repellents must be applied every evening and regular compliance is essential.

Repellents may play an important role in public health, as they are one of the few tools that can protect against mosquitoes that feed outside sleeping hours, during which time long-lasting insecticidal nets (LLINs) are available and effective. To date, vector control programs in sub-Saharan Africa have focused on the implementation of strictly intra-domiciliary interventions like LLINs and indoor residual spraying (IRS). These interventions focus on reducing human-vector contact indoors and have contributed to a malaria decline in various regions of sub-Saharan Africa [7]. However, the use of LLINS cause selective pressure upon mosquito populations, because those interventions target only indoor and late-evening biting mosquitoes, resulting in an increased relative abundance of early-evening biting disease vectors that are not killed by these interventions [8]. In addition, urbanization and and the greater access to electricity gives many people a reason to stay awake later, increasing the availability of humans for early feeding and vector borne disease transmission [9].

In southern Tanzania, despite high bed net coverage, malaria transmission persists. It is estimated that around 20% of transmission occurs before individuals retire to bed, therefore topical repellents may have a benefit if used as an additional vector control tool to LLINs in this setting [10]. However, there are several problems related to implementation of topical repellents: assuring compliance and reaching complete coverage. If a proportion of the population at risk does not use repellents there is a chance that mosquitoes will be diverted to non-repellent users [11]. In this case people who are not using repellents as a result of poor awareness or poverty will be more exposed to mosquito bites and consequently to vector borne diseases. Diversion has been observed in households with incomplete bed net coverage, and in some scenarios mosquitoes may feed on animals, which would enhance protective efficacy [12]. In the Gambia a study using permethrin-impregnated bed-nets observed that despite mosquitoes being deterred from entering experimental huts by the volatile chemicals, no increase in mosquito numbers were observed in neighbouring huts [13]. It is therefore crucial to investigate neighbouring effects and evaluate if mosquito diversion occurs between households in a community of repellent users and non-users to ensure health equity if interventions with a repellent mode of action are to be considered as public health interventions. Therefore, the present study was devised to measure if diversion occurs from households that use repellents to those that do not within a community with 80% topical repellent coverage.

Methods

Mosquito collection

Weekly mosquito collections were performed on three consecutive weekdays where each household was sampled once. Typically, each village was visited on one day each week. Host seeking mosquito densities from the enrolled households were estimated through the number of resting mosquitoes collected from early morning indoor and outdoor aspirations using aspirating devices [14]. Outdoor resting mosquitoes were sampled from artificial resting places immediately next to each household and from underneath the thatch roof of the outdoor kitchen area of each household (Figure 1). The artificial resting shelters, installed by the research team, consisted of a large dark barrel covered with vegetation (diameter 0.65m and depth 0.75m) and a regular car tyre placed directly outside each household. Collections were done indoors and outdoors to maximize the number of collected resting mosquitoes per household. After collection, mosquitoes were sorted by genera and members of the Anopheles gambiae complex were identified using PCR (polymerase chain reaction) [15].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Mosquito collection sites and some households of the Mbingu area, Kilombero Valley, Tanzania.

A – Household made with thatch wall and thatch roof with a blue barrel and a car tyre serving as artificial resting places. B – Household made with brick walls and thatch roof, the artificial resting places are not visible because they were deployed behind the household. C – Household made of mud walls and thatch roof with a blue barrel and a car tyre serving as artificial resting places. D – Kibanda, outdoor kitchen area.

https://doi.org/10.1371/journal.pone.0084875.g001

Results

Of the 49 households enrolled in the study, we collected mosquitoes from 44 households at each of the 10 weekly visit.. Of the remaining households, 3 households were missing one measurement and two other households were missing 2 and 4 measurements, respectively. Overall, 481 household-days of mosquito data were available for analysis. A total number of 2650 resting mosquitoes were collected, of which 105 were anophelines (83% were Anopheles gambiae s.s. and the remaining Anopheles arabiensis). Most mosquitoes were found resting indoors (47.6%), followed by underneath the thatch roof of the outdoor cooking areas (kibanda) (20.4%), from inside the large barrel (17.9%) and from inside the car tyre (14.1%) (Table 3). Most of female mosquitoes collected were unfed (Table 4). Of the 49 households in the study, 33% had mud walls and 61% had burned mud brick walls; 73% were thatched with grass and 27% had corrugated steel roofs (Table 1). A few households (61%) owned chickens; no households owned cattle. There were considerable differences in housing structure between the three villages with Matete having poorer housing, a factor that might influence mosquito densities inside houses. However, there was no evidence that roof type or wall types were associated with the outcome. Other sources of variance including day of week, number of occupants, previous week’s treatment and collector were shown to be not significant in the model and were therefore excluded from the model. The final model included the pre-specified fixed effects for village and week of the study. Regular spot checks indicated that all project participants complied with the repellent usage.

The lowest densities of resting mosquitoes were found in households within villages where everybody had been given a repellent lotion (Table 4). These households had less than half of the mosquitoes that would normally be found in households in the no coverage scenario (Median=3; IQR=[2-6]; IRR=0.39; 95%CI=[0.25 - 0.6]; p<0.001). Only 8% fewer mosquitoes were found resting in repellent-using households in an incomplete coverage scenario compared to households in a no-coverage scenario (Median=2; IQR=[0-5]; IRR=0.92; 95%CI=[0.72 - 1.17]; p=0.485). On the other hand, placebo households in an incomplete coverage scenario that were surrounded by repellent-using households, had more than 4 times more resting mosquitoes compared to a non-coverage scenario where nobody in a village uses repellent (Median=8.5; IQR=[2-20]; IRR=4.17; 95%CI=[3.07 - 5.65]; p<0.001). Diversion to animals is unlikely to have occurred as none of the households possessed domestic animals besides chickens.

Discussion

The WHO recommends topical repellents as the first line of protection against outdoor biting vectors [16]. Similarly, the U.S. Centres for Disease Control and Prevention (CDC) has acknowledged that ‘Repellents are an important tool to assist people in protecting themselves from mosquito- borne diseases’ [17]. Although repellents are popular amongst travellers and are commonly used in some urban areas' [9,18], the local population in rural Tanzania does not commonly use them. This may be because the local population has limited knowledge of repellents; these products are generally unavailable in small local shops in rural areas and are typically not affordable to most of the local population who live on subsistence incomes. Unlike bed-nets and anti-malarial drugs, topical repellents have never been included in subsidization programs in Africa. Countries like South Africa and Thailand recommend their use and some non-governmental organisations (NGOs) provide repellents to populations, including PSI in South America and the Global Fund in the Mekong region of Thailand (Michael MacDonald, personal communication).

The results from this study provide evidence that mosquito diversion from repellent users to non-users occurs under an incomplete coverage scenario. Households of non-repellent users within villages where 80% of the village uses repellents have over four times more mosquitoes resting in their households. We also observed that households that used repellent in an incomplete coverage scenario maintained high numbers of host seeking mosquitoes in their households despite compliance with the intervention. It is unlikely that mosquitoes will feed on these household members since they are protected with mosquito repellent. This is indicative that complete coverage of topical repellents offers a community effect by reducing the number of mosquitoes in all human dwellings more markedly than if only incomplete coverage is reached. Hence complete coverage of topical repellents would not only benefit those who otherwise would not be covered by avoiding diversion of mosquitoes but also individuals who usually comply with topical repellents. Even if topical repellents are made available to all, it is unlikely that complete coverage is attainable since topical repellents require regular compliance. As a result, mosquito diversion is likely to occur between those who comply with the intervention and those that do not. In case of malaria control, complete coverage is the recommended strategy for implementing topical repellents, however, if a large proportion of the population regularly complies with topical repellent usage, it is likely that the malaria parasite load in the community will be reduced through reduction in man-vector contact as has been observed on a community scale with untreated bed nets that act only as a personal protection tool [19]. Our study did not recover many blood fed mosquitoes or malaria vectors due to a drought that occurred during the study. Therefore, it is not clear if malaria vectors are being diverted and feeding on unprotected individuals. We plan future studies that will analyse the blood meal from fed mosquitoes found resting indoors to ascertain whether mosquitoes have fed from humans. Such analyses would help to determine whether diversion results in increased risk of bites to unprotected humans.

Other repellent interventions such as spatial repellents using volatized pyrethroids may offer a better solution than repellents since they prevent mosquitoes from feeding and disrupt their host-seeking behaviour [20]. Such interventions may not necessarily divert mosquitoes to other households but disable the host seeking behaviour of mosquito vectors resulting in a more effective reduction of man vector contact than observed with topical repellents. Currently further research is underway to measure diversion in a community using incomplete coverage of volatized pyrethroids.

Acknowledgments

We would like to thank the people of villages Matete, Upper and Lower Sanje for their cooperation and support to the project. We also thank SC Johnson for providing the project with the repellent and placebo lotions.