Patterns and seasonality of malaria transmission in the forest-savannah transitional zones of Ghana

Kintampo North and South districts lie within the forest-savannah transitional ecological belt in the Brong Ahafo region of Ghana. They cover an area of approximately 7,162 km² with a resident population of approximately 140,000 living in approximately 22,000 houses in 156 villages. Communities in Kintampo are predominantly agricultural, engaging in farming of maize and yam and raising some livestock. The area has a mean monthly temperature range of 18°C to 38°C and a rainfall averaging 1,250 mm per annum, which falls mainly between April and October. The Kintampo Health Research Centre implements a health and demographic surveillance system (KHDSS), which tracks the resident population dynamics every six months [17].

Sixteen communities or clusters were randomly selected for inclusion in this study following stratification reflecting the mixed micro-ecology (based mainly on the vegetation and water bodies) of the two districts. Communities were geo-located using a simple hand-held GPS receiver (GARMIN series) and integrated into a Geographic Information System (GIS) for analysis (ArcGIS 9.2 version). Every house within clusters or communities was given an equal chance of participation. Selection of houses to receive light traps was done without replacement until all houses within a particular community had been considered.

One of the meteorological stations linked to the Ghana National Meteorological Department is located in Kintampo town and serves the two districts. Daily collection of data (including rainfall data) was performed for the entire period of the survey.

Mosquito collection was performed over a two-year period (November 2003 to November 2005), the first year alongside a parasitological survey and the second year alone. The Centre for Disease Control (CDC) light traps were used to collect mosquitoes in rooms of randomly selected households. In the first year, mosquito trapping was undertaken the night before a parasitological survey in a particular cluster. Additional traps were then set weekly in rooms of study subjects who met selection criteria in both cohort studies [18] ensuring that at least one trapping took place in each month in each of the 16 selected communities and throughout the whole of the first year of the study. Verbal consent was sought from household heads and occupants of each room in which a trap was set and untreated bed nets provided to be used during the night that the trap was set in the room. Traps were hung approximately 1.5 m above the floor at the foot of the bed/mat of the index person. In the second year of mosquito collection (December 2004 to November 2005), the number of CDC light traps was reduced from four per community per night to two per community per night of trapping due to funding limitation. All other processes were maintained as during the preceding year.

Anopheline vectors were morphologically identified into species using identification keys [19], stored in 1.5 ml micro-centrifuge tubes enclosed in zip lock plastic bags with silica gel. A maximum of 10 mosquitoes of the same species from the same compound was put in a tube. Non-Anopheles species were discarded after recording numbers caught.

Heads and thoraces of the two major vectors of malaria, An. gambiae and An. funestus, were checked for the presence of circumsporozoite (CS) antigen of P. falciparum using the sandwich enzyme-linked immunosorbent assay (ELISA), as described [20]. Presence of CSP in the mosquitoes was read at 405 nm wavelength using a micro plate ELISA reader. A cut-off of 0.2 nm absorbance after subtraction of the average value from seven negative test mosquitoes was considered positive. Heads and thoraces of male Anopheles vectors were used as the negative test controls. All positive mosquitoes were retested to confirm positivity. Results were analysed with the aid of Pool Screen^® computer software. EIR was calculated by multiplying the proportion of positive tested vector species by their Man Biting Rate (MBR), which is estimated as the geometric mean of the number of vectors caught in a light trap.

A total of sixty-four sub-samples were randomly selected to explore the most predominant specie within An. gambiae complex. The samples were selected from the month of February (to represent dry season) and August (to represent wet season) in 2005 catches of Anopheles and tested by Polymerase Chain Reaction (PCR) as per protocol [21]. Enzyme digestion to differentiate the molecular forms of An. gambiae s.s was performed as per protocol [22] on samples that were first PCR successful for sibling species analysis. The presence of kdr alleles conferring knock-down resistance in West Africa was assessed as described [23] on samples that were PCR successful for molecular form differentiation analysis. Results were analyzed proportionally and by Hardy-Weinberg test statistic.

In year 1 (November 2003 - November 2004), 664 CDC light traps caught 19,771 mosquitoes; 51.5% of these were Culicines, 1.0% Aedes, 35.0% An. funestus, 10.6% An. gambiae and 1.9% Anopheles rufipes. In year 2 (December 2004-November 2005), 355 CDC traps set captured 3,571 mosquitoes; 44.1% of these were Culicines, 4.8% Aedes, 37.9% An.funestus, 11.6% An.gambiae and 1.6% An. rufipes (Table 1). Monthly abundance of Anopheles vectors caught varied in the two collection periods. Anopheles funestus was caught most frequently in November 2003, September 2004 and October 2005. Anopheles gambiae was also caught frequently in these months but in lower numbers than An. funestus (Figure 1). Abundance of both vectors correlated with monthly rainfall patterns (Figure 2). The annual rainfall in Kintampo for three years consecutively (2003-2005) was steady and ranged between 1,200 mm and 1,400 mm. The highest rainfalls were recorded in the months of April, July and September in the first year and in May, September and October in the second year (Figure 2).

The all-age group prevalence of Plasmodium falciparum ranged between 36% and 57%. The prevalence was highest during the rainy seasons (May-October) but high throughout the year (Figure 3).

The distances of communities the study was carried out ranged between eight kilometres and 50 kilometres from the meteorological station centrally located in the two districts. The farthest community to the extreme south (Ajina), is 46 km and that to the extreme north (Kawampe), is 50 km (Figure 4).

The two main malaria vectors were An. gambiae s.s and An. funestus. In year one, 8,418 Anopheline samples were assayed by CS-ELISA; 6,542 were An.s funestus and 1,876 were An. gambiae. Plasmodium falciparum CS antigen positivity was 1.5% in An. funestus and 4.7% for An. gambiae. The annual EIR, including both vector species, was 269 infective bites per person per year (ib/p/y) with variations in community level EIRs. A total of 1,794 Anopheline samples were tested by CS-ELISA in year two; 1,054 were Anopheles funestus and 740 were An. gambiae. Plasmodium falciparum CS antigen positivity was 3.7% in An. funestus compared to 1.2% for An. gambiae. The annual EIR was 231ib/p/y with An. funestus (141ib/p/y) contributing more infective bites than An. gambiae (90ib/p/y), the opposite pattern from that observed in the previous year. Sporozoite rates (SRs) were generally higher in year 2 than in year 1. Sporozoite rates were highest in communities in the north east and in the southern parts of the districts (Table 2).

Fifty-five out of the 64 selected An. gambiae s.l. were successfully tested by polymerase chain reaction (PCR) to determine their sub-species. All were An. gambiae s.s; no An. arabiensis were detected in either survey period. Sixteen of 22 An. gambiae s.s. mosquitoes collected during the wet season (August 2005) were S molecular form, two M molecular forms and one a hybrid (S/M). Three samples failed PCR analysis. A similar pattern was seen during the dry season (February 2005) with 11 of 16 An. gambiae s.s. mosquitoes tested being S molecular form, two M molecular form and three (S/M) hybrids (Table 3). Ten An. gambiae s.s mosquitoes, five collected in the rainy season and five collected in the dry season were tested for kdr mutations (Table 3). Six were kdr^RR, two kdr^R/S and two had the susceptible kdr^ss genetotypes.