Impact of housing condition on indoor-biting and indoor-resting Anopheles arabiensis density in a highland area, central Ethiopia

Background Exposure of individuals to malaria infection may depend on their housing conditions as houses serve as biting and resting places of vectors. This study describes the association of housing conditions with densities of indoor-biting and indoor-resting Anopheles arabiensis in Hobe, Dirama and Wurib villages of a highland area in central Ethiopia. Methods Data on housing conditions, including presence of house apertures, number of occupants and number and the type of domestic animal tethered inside, were collected. Indoor-biting mosquitoes were sampled using Centers for Disease Control (CDC) light traps and indoor-resting mosquitoes sampled with pyrethrum spray catches (PSCs) monthly for two years (July 2008 to June 2010). Female anophelines were identified to species and processed. Univariate and general linear estimating equation allowing for repeated measures were used to assess the contribution of housing conditions for indoor-biting and indoor-resting An. arabiensis. Results About 96% (4,597/4,788) of anophelines were caught inside residential houses. Nine anopheline species were identified, among which An. arabiensis was most prevalent (2,489; 52%). Vectors entering houses were higher in those situated at low (β = 4.475; 95% CI = 3.475-5.476; p <0.001; β = strength of the association) and medium (β = 2.850; 95% CI = 1.975-3.724; p <0.001) altitudes compared to high altitude, and where houses have no windows (β = -0.570; 95% CI = -1.047-0.094; p = 0.019) compared with those that have. Numbers of indoor-resting vectors were higher in those situated at low (β = 6.100; 95% CI = 4.571-7.629; p <0.001) and medium (β = 4.411; 95% CI = 2.284-6.537; p <0.001) altitudes compared to high altitudes, and where houses had open eaves (β =1.201; 95% CI = 0.704-1.698; p <0.001) compared with those that had closed eaves. Conclusion Housing conditions such as presence of open eaves, absence of window, location at low and mid altitudes, were strong predictors of indoor exposure to An. arabiensis bite in a highland area of south-central Ethiopia.

Conclusion: Housing conditions such as presence of open eaves, absence of window, location at low and mid altitudes, were strong predictors of indoor exposure to An. arabiensis bite in a highland area of south-central Ethiopia.

Background
Malaria affects 68% of the Ethiopian population [1]. Although control efforts brought reduction in malariarelated mortality compared to the previous years [1,2], the disease is still among the top causes of morbidity in the country [3,4]. It is seasonal in most areas below 1,500 m altitude and unstable in areas above 1,500 m [1].
Transmission of the disease depends largely on local topography, climate and land use. It may also be influenced by housing conditions [5][6][7]. Conditions, such as nearby irrigated land, earth roof, tethering livestock inside, window presence, open eaves, absence of separate kitchen and presence of a single sleeping room, were associated with high incidence of child malaria in northern Ethiopia [8]. In Burkina Faso, children living in mud-roofed houses were at a higher risk of Plasmodium falciparum infection compared to those in iron sheetroofed houses [9]. In The Gambia, eaves were the main routes of Anopheles gambiae and Mansonia spp. entry [10,11]. Houses with a grass roof were associated with increased malaria risk in Mozambique [12].
The association of poorly constructed houses with high malaria infection risk may result from their suitability to indoor abundance of vectors [9,10,13]. Houses are the principal site where malaria vectors bite and rest [10,11,14], hence improved housing may reduce indoor occurrence and the risk of malaria transmission in Ethiopia. However, housing conditions and their impact on indoor abundance of vectors may vary with respect to geography, socio-economy and individual household factors. This study was undertaken to assess the contribution housing conditions make to indoor-biting and indoor-resting Anopheles arabiensis in a highland area of central Ethiopia.

Study area and housing conditions
A longitudinal study on the relationship between housing conditions and number of indoor-biting as well as indoor-resting An. arabiensis was undertaken in Hobe, Dirama and Wurib villages of south-central Ethiopia once a month for two years (July 2008 to June 2010). The same villages and houses were used for related studies [15,16].
Most of the houses were constructed with mudplastered wooden walls and grass roofs. They did not have ceilings or separate kitchen. A single living house is used for sleeping, keeping all household belongings, cooking and dining, keeping warm by burning wood and also for tethering domestic animals at night (Figure 1). Data on housing conditions, including presence of house apertures, number of occupants that slept the previous night and number and type of domestic animals tethered indoor the previous night were recorded, while undertaking mosquito sampling, once per month. In addition, the location of each house where mosquitoes were sampled was categorized into either low altitude (Hobe), mid altitude (Dirama) or high altitude (Wurib). The study period was categorized into either dry or wet. Wet were months with average rainfall of greater than 1 mm. They include May, June, July, August, September and October. The number of occupants and domestic animals (cattle, sheep, goat, horse, donkey, and chicken) was recorded by interviewing the head of household or the next elder occupant. House apertures, such as door (unfit or fit), window (absent or present), open eaves (absent or present), hole on wall (absent or present), and hole on roof (absent or present) were recorded by direct observation. All the houses (except one) had unfit doors; therefore the variable door fitness was excluded from the analysis.

Mosquito sampling
Mosquito sampling was undertaken using Centers for Disease Control (CDC) light trap, pyrethrum spray collection (PSC) and artificial pit shelter (APS) [17] from Hobe, Dirama and Wurib villages. CDC light trap-based collection was made for two consecutive nights inside 20 houses resulting in 40 tap nights per month per village. PSC was made in ten randomly selected houses where no CDC light trap catches was undertaken. Five APSs constructed in shaded areas were used for outdoorresting mosquito collection in each village. CDC light trap catches were used to collect mosquitoes that attempted to bite humans inside houses during night hours. PSC was used to collect mosquitoes that rest indoors during daylight hours. All female anopheline catches were identified to species, counted and processed, while culicines were discarded after counting. The detailed method is described elsewhere [16].

Statistics
Indoor-and outdoor-sampled mosquitoes were depicted in a frequency table. Association of each housing condition with the number of either indoor-biting or indoor-resting An. arabiensis catches was assessed independently using univariate analysis from which the mean number of An. arabiensis catches, including 95% confidence interval (CI) for the mean and significant level was calculated. In the univariate analysis, an independent variable with p value less than 0.1 was considered as a potential predictor and was re-analysed using generalized estimating equation (GEE) multivariate analyses for repeated measures. The dependent variable, number of An. arabiensis, fitted to a negative binomial distribution with a log link function [18]. Variables with p values <0.05 in the GEE were considered as strong predictors. Data were analysed using PASW Statistics version 18 (SPSS Inc, Chicago, IL, USA).

Ethics
The study was ethically cleared by the Ethical Committee of the Faculty of Medicine, Addis Ababa University and The National Health Research Ethics Review Committee (NERC) of Ethiopia with reference number RDHE/48-85/ 2009. All anopheline collections were undertaken following verbal consent of households.
Housing conditions that predict indoor-biting and indoor-resting An. arabiensis are presented in Table 3. The number of An. arabiensis that bite inside houses located at the low altitude village (Hobe) was 4.475 (95% CI = 3.475-5.476; p <0.001) times relative to the number in the high altitude village. Similarly, the number in the mid altitude village was 2.850 (95% CI = 1.975-3.724; p <0.001) times relative to the high altitude. Houses with window had 57% lower number of indoor-biting An. arabiensis (β = -0.570; 95% CI = -1.047-0.094; p = 0.019) relative to those with no window. Similarly, house location at the low or mid altitude village relative to the high altitude and presence of open eaves relative to no open eaves were strong predictors of indoor-resting An. arabiensis.
The mean number of indoor-biting An. arabiensis characterized by feeding status, blood meal source and Plasmodium sporozoite infection status with respect to housing condition is presented in Table 4. Houses located in the low altitude village were observed to have significantly highest mean number of fresh fed (2.58), half gravid (0.89), gravid (0.72), unfed (0.75) and bovine fed (1.31) An. arabiensis caught by CDC light trap. Houses with no window had higher mean number of fresh fed, unfed, bovine fed, human fed and human and cattle mixed blood fed An. arabiensis and the differences were significant.
The mean numbers of indoor-resting (caught by PSC) fresh fed, half gravid, gravid, bovine fed, human fed, and human and bovine mixed blood fed An. arabiensis were significantly higher in houses having open eaves than in those with no open eaves and also in houses located at either the low or mid altitude village than in the high altitude village (Table 5).

Discussion
Most Anopheles mosquito species in Hobe, Dirama and Wurib villages of central Ethiopia occur inside residential houses. Houses having open eaves, no window, and located at either low or mid altitude village were associated with higher risk of malaria. The indoor occurrence of anophelines in these highland villages could be attributed to several factors among which appropriate indoor microclimate is one [19,20]. The tradition of cooking, sleeping and tethering livestock inside residential houses could contribute to the indoor occurrence of mosquitoes by increasing indoor temperature and providing access to blood meal sources. This in turn contributes to the survival and increased malaria transmission potential of the vectors in the area. Indoor-resting mosquitoes of East Africa are estimated to transmit malaria between 0.3 and 22.5 days earlier than those of outdoor-resting mosquitoes [19]. This study reveals that An. arabiensis and An. pharoensis, which are malaria vectors in the area [16] and the remaining seven anopheline species, exhibit endophilic behaviour indicating the need to construct mosquito proof houses.
Densities of both indoor-biting and indoor-resting An. arabiensis were highest in the low altitude village and decreased with increasing altitude. Similarly, densities of both immature and adult stages of the vector were observed to decrease significantly with increasing altitude in the area during the period [15,16] and so was the risk of acquiring P. falciparum and Plasmodium vivax malaria [16,21,22]. Density of vectors generally decreases with increasing altitude in highland areas [23].
In this study, houses with open eaves were strongly associated with indoor-resting An. arabiensis relative to the houses with no such opening. Eaves could enhance An. arabiensis entry to houses and its blood meal sources (human and cattle) which stay indoor during night hours [16] and then rest in the house until oviposition. Houses with open eaves and no ceilings were observed with higher number of An. gambiae than those with closed eaves and ceilings [10]. Open eaves were associated with increased risk of An. gambiae s.l. and Culex pipiens s.l. entry in The Gambia [11,24]. Anopheles gambiae s.s., An. arabiensis, Mansonia africana and Ma. uniformis were noted to prefer eaves as the main entry points in Tanzania [25]. The high density of An. arabiensis inside houses with open eaves could result from the upward-flying behaviour of the mosquito when encountering wall surfaces and entering houses through these holes having been attracted by microclimatic conditions and odours of humans and cattle coming from the houses [10,11,19,20,26].
This study indicates the need to construct houses with closed eaves, roof and ceilings in Hobe, Dirama and Wurib villages of central Ethiopia in order to minimize indoor-resting An. arabiensis, which is the most prevalent and major malaria vector in the area [15,16]. House ceilings made of plywood, synthetic-netting, insecticidetreated synthetic-netting, and plastic insect screen, all installed below open eaves and mud-closed eaves, reduced entry of An. gambiae into experimental huts in Gambia [10]. Closing eaves resulted in a three-fold reduction in An. gambiae s.l. caught indoors [11]. Eaves screening reduced density of indoor An. gambiae s.l., Ma. africana and Ma. uniformis significantly in southern Tanzania [25]. Screening houses fully and also equipping them with screened ceilings can reduce indoor exposure to An. arabiensis bites as noticed in The Gambia [27] and Kenya [28]. In addition, constructing houses with iron-sheet roof instead of thatched roof may reduce malaria infection risk in south-central Ethiopia as reported from Burkina Faso [9]. The number of An. arabiensis that attempted to bite indoors at night was 57% lower in houses with windows than in those with no window. The presence of windows might have increased aeration inside houses, which could reduce indoor temperature. Low indoor temperature in these highland villages could deter the indoor-biting mosquitoes at night. In The Gambia [11], windows and doors were found less important for An. gambiae s.l. entry into houses but were the main entry routes of culicines.
Anopheles arabiensis, which is the principal malaria vector in Hobe, Dirama and Wurib villages in particular [16] and in Ethiopia in general, was prevalent inside houses located in the low altitude village and in the mid altitude village. Houses with open eaves were also observed to have high density of indoor-resting An. arabiensis. Better designed houses and house screens, together with existing malaria control programmes, may help to reduce indoor-biting as well as indoor-resting An. arabiensis and hence transmission of the disease significantly.