Study site
The study was carried out in Dares Salaam where a new Urban Malaria Control Programme has recently been re-initiated and a variety of associated studies on malaria vectors have been carried out [4–6, 12–15]. The area experiences modest malaria transmission intensity with an entomologic inoculation rate (EIR) [16, 17] of approximately one infectious bite per person per year [4, 6]. The main malaria vectors are members of the An. gambiae complex that predominantly feed out-doors [5].
Dar es Salaam is a coastal city in Tanzania (6'46' S Latitude and 39'14 E Longitude) with approximately 2.7 million inhabitants living in an administrative region which covers a total area of 1400 km2 [18]. Dar es Salaam has two rainy seasons: the short rains from late October to early December and the long rains from March to June. The climate is warm and tropical, with temperatures averaging 27°C (80°F) and rainfall varying from 750 to 1,400 mm per year. During the dry season temperatures often exceed 35°C.
The UMCP covers 15 wards of urban Dar es Salaam, encompassing a surface area of 55 km2 with a total population of 609,514 people [4]. It operates primarily at the grassroots level through street health committees, using a community-based system originally developed by one of the three municipal councils (Ilala) that comprise the administrative region of Dar es Salaam. Presently, the program operates in five wards in each of the three municipalities (Ilala, Kinondoni and Temeke) as a community-based pilot-scale program [14]. In 2004, the UMCP recruited and provided preliminary training to teams of Community Owned Resource Persons (CORPs) who performed weekly surveys of mosquito breeding habitats [12]. Operational larviciding in three selected wards with Bacillus thuringiensis var israelensis commenced in 2006 [4, 6]. Currently, the UMCP implements four major activities in all the three municipalities: namely larval control, larval surveillance, adult mosquito surveillance and household parasitological surveys [4–6, 14]. This study was based within the sampling frame and reporting system of the routine adult mosquito surveillance programme, which conducts monthly sampling of mosquitoes by HLC at 268 location distributed across the 15 wards of the UMCP [5, 6].
Experimental design and selection of the sampling site for the three traps
The study was carried out in 12 wards in the study area of the Urban Malaria Control Program. One neighbourhood (mtaa in Kiswahili) in each ward was randomly selected for this study. In each selected neighbourhood, four HLC sites already existed in four Ten Cell Units (TCUs) which were deliberately chosen to be well distributed across the neighbourhood and as close to potential breeding sites as possible. For each pre-existing HLC site, a nearby (100–300 meters away) house was selected arbitrarily for both application of the ITT-B and the SRB. Therefore, in each neighbourhood, eight houses from different TCUs were used for the three sampling methods: four houses for the HLC and four houses for the ITT-B and the SRB, totaling 48 houses for the HLC and 48 for the combined ITT-B and SRB methods, respectively. Concomitant sampling with ITT-B and SRB began in the first enrolled wards in December 2007 and the last of the 12 wards had begun by March 2008. Data and mosquitoes collected up to end of June 2008 were included in this analysis, spanning a period of between seven and four months for each of the 12 wards.
Field mosquito collection and processing
Routine human landing catch was conducted outdoors once a week in each neighbourhood by one catcher working from 18.00 to 06.00 hours for a period of 45 minutes every hour, allowing the catcher to have a 15 minutes break. To minimize the possibility of data fabrication by the catchers, they were obliged to record the approximate number of each relevant mosquito taxon in their catches for each hour as they finished them. Moreover, spot checks were conducted inconsistently, unpredictably and at arbitrary times of the night by a team of 4 supervisors. The mosquitoes caught were collected by the project vehicle the following morning and taken to the laboratory for further processing.
A protocol for sampling malaria vectors and other mosquitoes using the ITT-B and SRB was developed to enable community members to trap, record and submit samples of malaria vectors without any night-time supervision and only occasional contact with program staff. This protocol was used to evaluate the sensitivity of the ITT-B and the SRB relative to that of carefully controlled HLC as follows: Prior to the supply of materials for ITT-B and SRB experiments, demonstrations were provided to train the community-based staff on correct use of the two traps. The operators were supplied with all the necessary materials that allowed them to continuously collect mosquitoes for a period of one week while recording them on a form they were provided with. Mosquitoes trapped in the ITT-B were carefully aspirated using hand-held aspirators and placed into paper cups, once in the middle of each night (00.00–01.00) and then early in the morning the next day (05.00–06.00). Operators were allowed to choose, at their own discretion, which nights of every week they slept in the traps and what time they entered and left the trap, under the condition that they recorded these dates and times. While still in paper cups, the mosquitoes were suffocated with a small ball of cotton wool soaked in petroleum ether. The dead mosquitoes were then transferred into smaller silica gel-filled containers for storage and preservation with a label indicating the ward, mtaa, site and day of collection.
Resting boxes were installed nearby (10–20 m) the ITT-B in each neighbourhood. The boxes were emptied between 06.00 and 08.00 in the morning of each working day using hand-held aspirators. Since experiments with ITT-B and SRB ran concurrently, suffocation, preservation and submission to the laboratory was accomplished in exactly the same way and at the same time as those from the ITT-B.
Laboratory processing and analysis
All the mosquitoes collected in the field by HLC were taken to the laboratory and killed by suffocation with chloroform. For mosquitoes caught by ITT-B and SRB, this process was completed in the field by the trap operators who submitted their samples for identification and laboratory processing after a one week period of sampling. In the laboratory, all mosquitoes were identified morphologically using taxonomic keys [19] according to sex as males or females, morphologically as An. gambiae s.l., Anopheles funestus, Anopheles coustani, Cx. species, or Aedes species while the abdominal status was scored as gravid/semi-gravid, fed or unfed for all the An. gambiae s.l. and for a manageable proportion of Cx. species. All An. gambiae s.l. caught by the three trapping methods were subsequently desiccated over silica gel and kept at room temperature until they were further processed.
A wing or a leg of every An. gambiae s.l. mosquito caught was analyzed by PCR to identify its exact species within the An. gambiae complex [20]. An enzyme-linked immunosorbent assay (ELISA) using a monoclonalantibody that recognizes a repetitive epitope on the circumsporozoiteprotein of Plasmodium falciparum was used to assessmalaria sporozoite infection status in each individual An. gambiae s.l. [21].
Data handling and analysis
All data handling and analysis was conducted with Microsoft Excel® 2007 and SPSS® 15.0. The only mosquito taxa considered for analysis were An. gambiae s.l. and Cx. species because these were the only ones for which sufficient numbers were collected throughout the study period.
To allow direct comparison with HLC conducted in the same area and in the same week, data was first aggregated by station and week, giving a total of 48 mean catches for matching station-week combinations over a period of 30 weeks. Prior to this analysis step, the numbers in each catch (x) were normalized by transforming to log10 [x+1] [3]. The relationship between catches by ITT-B or SRB and that of the HLC, in the same week and station, was initially assessed using simple Pearson's linear correlation method. Regression using generalized estimating equations was used to test for density dependence of the relative sampling efficiency of the ITT-B and SRB methods relative to the sum of the ITT-B and the HLC. On several occasions, the three traps recorded zero values for An. gambiae s.l. mosquitoes even after aggregation by station-week so no logical comparison could be made and these data were discarded. Since divisions by zero gives infinite values, data for several week-site observations were sorted by the sum of the catches for the traps (alternative plus the reference) and then aggregated by this sum with the mean of each of the two catches as the summary variables. The mean catch of the alternative collection methods divided by the mean catch of the reference method was treated as the dependent variable with a log link function and a gamma distribution for An. gambiae s.l. and a normal distribution for Cx. species. The sum of the alternative and the reference methods was treated as a continuous independent variable in the model.
To test for consistent variations in species composition, sporozoite prevalence and abdominal condition of the mosquitoes sampled by the different traps, binary logistic regression in SPSS was used. Each outcome was treated as a binary variable with trap design as an independent categorical factor in the model. The results of abdominal status and sibling species identity were expressed as binary outcomes: fed (partially or fully-fed) versus non-fed (gravid or unfed) and An. gambiae s.s. versus An. arabiensis, respectively, as described previously [8]. Although sporozoite infection status was determined in the laboratory and the dependence of sporozoite prevalence upon trap type was tested for using a similar statistical approach, the number of mosquitoes caught was not sufficient to enable meaningful conclusions to be reached regarding this relatively rare fraction of the vector population.
Ethical consideration and informed consent
Informed consent was obtained from all the participants, namely the household owners and the mosquito catchers. Moreover, thick and thin blood smears were taken from all the participants whenever they complained of fever to examine the presence of malaria parasites. When found positive, they were treated with Coartem® (artemether-lumefantrine).
