Malaria transmission dynamics at a site in northern Ghana proposed for testing malaria vaccines


Maxwell Appawu (corresponding author), Samuel Dadzi, Kwadwo Koram, William Rogers and Francis Nkrumah, Noguchi Memorial Institute for Medical Research, Legon, Accra, Ghana. Fax +233-21-502 182, E-mail:,,,,
Seth Owusu-Agyei, Victor Asoala and Francis Anto, Navrongo Health Research Center, Navrongo, Ghana. E-mail:,,
Stephen L. Hoffman, Sanaria, 305 Argosy Dr., Rockville, MD, USA. E-mail:
David J. Fryauff, Naval Medical Research Center, Silver Spring, MD, USA. E-mail:


We studied the malaria transmission dynamics in Kassena Nankana district (KND), a site in northern Ghana proposed for testing malaria vaccines. Intensive mosquito sampling for 1 year using human landing catches in three micro-ecological sites (irrigated, lowland and rocky highland) yielded 18 228 mosquitoes. Anopheles gambiae s.l. and Anopheles funestus constituted 94.3% of the total collection with 76.8% captured from the irrigated communities. Other species collected but in relatively few numbers were Anopheles pharoensis (5.4%) and Anopheles rufipes (0.3%). Molecular analysis of 728 An. gambiae.s.l. identified Anopheles gambiae s.s. as the most dominant sibling species (97.7%) of the An. gambiae complex from the three ecological sites. Biting rates of the vectors (36.7 bites per man per night) were significantly higher (P < 0.05) in the irrigated area than in the non-irrigated lowland (5.2) and rocky highlands (5.9). Plasmodium falciparum sporozoite rates of 7.2% (295/4075) and 7.1% (269/3773) were estimated for An. gambiae s.s. and An. funestus, respectively. Transmission was highly seasonal, and the heaviest transmission occurred from June to October. The intensity of transmission was higher for people in the irrigated communities than the non-irrigated ones. An overall annual entomological inoculation rate (EIR) of 418 infective bites was estimated in KND. There were micro-ecological variations in the EIRs, with values of 228 infective bites in the rocky highlands, 360 in the lowlands and 630 in the irrigated area. Approximately 60% of malaria transmission in KND occurred indoors during the second half of the night, peaking at daybreak between 04.00 and 06.00 hours. Vaccine trials could be conducted in this district, with timing dependent on the seasonal patterns and intensity of transmission taking into consideration the micro-geographical differences and vaccine trial objectives.


Malaria continues to be a major scourge of mankind in the developing world, and is estimated to cause 3 000 deaths each day, 90% of which occur in Africa, south of the Sahara (Nchinda 1988, WHO 2002). Integrated approaches, against both the parasites and vectors, are necessary for effective malaria control. Nonetheless, the spread of drug resistant parasites and insecticide resistant mosquitoes has caused much interest in the use of vaccines to control malaria. However, the application of focused and timely interventions such as vaccines requires detailed information on the local epidemiology of malaria and the distribution of risk groups. An ideal indicator of malaria risk is the entomological inoculation rate (EIR), an index that relates both the human-biting activity of the Anopheles vectors and the risk to humans of malaria infection. Risk of exposure of human to infectious bites of vectors in Africa is not uniform, and variations in the abundance and dispersal of mosquitoes occur in a given area, and over time. These variations impact on the level of malaria transmission (Smith et al. 1995). Evaluation of vaccines under conditions of natural transmission requires that vaccine-testing sites be identified and characterized.

In preparation for such field trials, baseline data on EIRs are required for selecting study sites, determining the timing of trials and calculating the volunteer sample sizes needed to evaluate vaccine efficacy. Also, we need precise information and close follow-up of variations in malaria transmission in an area (McElroy et al. 1994) to enable correct interpretations of malaria parameters such as parasitaemia, morbidity, and associated immune responses in relation to efficacy of vaccines.

Kassena Nankana district (KND) in northern Ghana has been identified for development as a vaccine field trial site. Malaria is responsible for high infant and child mortality there, accounting for 23% of child deaths in 1991–1992 (Binka et al. 1994). Significant seasonal variations in malaria morbidity and mortality have been recorded in KND, with higher rates occurring during the wet seasons (Binka et al. 1994). Although entomological sampling has identified Anopheles gambiae s.s. as the dominant member of the Anopheles gambiae complex in the KND (Appawu et al. 1994), there is no published information on dynamics of malaria transmission in the district.

Our study aimed at characterizing the site for vaccine trials by identifying the Anopheline populations of the KND, measuring their respective Plasmodium transmission rates, and determining the ecological, temporal and seasonal variations in species composition, density, biting behavior and the intensity of malaria transmission.

Materials and methods

Study areas

Kassena Nankana District (KND) in northern Ghana lies between latitude 10 °30′ and 11 °00′N, longtitude 1 °00′ and 1 °30′W and covers about 1 674 sq km of Sahelian savannah with a population of 1 40 000. Most people live in multi-family compounds of dispersed settlements, which form the basis of the address system used in the Navrongo Demographic Surveillance System (NDSS), and are separated from one another by agricultural land. Virtually all inhabitants engage in subsistence farming of millet, groundnut and livestock. The average annual rainfall is 850 mm, almost all of which occurs in the wet months of July–September, with the rest of the year being relatively dry. A large reservoir (Tono) in the middle of the district provides water throughout the year mainly for irrigation. The reservoir spans over 1860 ha with maximum storage capacity of 93 × 106 m3, serving 32 km of main canals. An open irrigation system floods the fields during the dry seasons. There are also many small water impoundments scattered throughout the district that store water in the wet season for dry season use. We divided the district into three micro-ecological zones: (1) Irrigated areas, including areas with large water reservoirs such as Tono and other small water impoundments schemes that lie to the south of the district; (2) Lowland areas at the northeastern end of the district which are well drained with most streams flooding their banks during the wet seasons; and (3) the rocky, highland areas at the west end of the district which are mostly dry with some small dug-out dams to provide water for livestock.

Mosquito collections

Mosquitoes were collected weekly in randomly selected compounds within each of the three micro-ecological zones between June 2001 and May 2002. All night human landing catches were made inside and outside dwelling rooms by two teams of four collectors. Mosquitoes biting the collectors were detected using a flashlight, aspirated and placed in paper cups covered by mesh screen (WHO 1975). Collections were made for 50 min each hour from 18.00 to 06.00 hours. The collectors rotated between indoors and outdoors after each period of collection, to compensate for differences in individual attraction or repulsion for mosquitoes. The mosquito collectors were offered prompt and immediate access to treatment in case of any infection, however, none of the mosquito collectors reported ill with malaria during the study. Ethical and scientific review bodies at the National Institutes of Health (NIAID), Noguchi Memorial Institute for Medical Research (NMIMR), and Navrongo Health Research Center (NHRC) reviewed and approved the research plan and informed consent process that was applied to the mosquito collectors.

Anopheline identification and processing

Captured mosquitoes were anaesthetized, sorted, identified morphologically to species (Gillies & de Meillon 1968) and counted. All Anopheles were stored in 1.5-ml Eppendorf snap vials with desiccants for further laboratory analysis. Anopheles heads and thoraces were tested for the presence of circumsporozoite antigens of Plasmodium falciparum (PfCSP) using enzyme-linked immunosorbent assay (ELISA) (Burkot et al. 1984; Wirtz et al. 1987). Parity was determined by inspection of ovarian tracheoles in a sample of the unfed anophelines captured at each site (Detinova 1962).

Molecular identification of malaria vectors

Members of the Anopheles gambiae complex were identified using the PCR technique described by Scott et al. (1993) with minor modifications. This involved homogenizing legs of Anopheles gambiae s.l. in 15 μl of DEPC-water and centrifuging at 16 000 g for 5 min to extract the DNA. PCR was then performed on the DNA template using An. gambiae s.l. species diagnostic primers in a total reaction mix of 25 μl. Anopheles gambiae s.s. specimens were further analysed by Restriction Fragment Length Polymorphism (RFLP) which involved the digestion of amplified PCR products with restriction enzymes Tru9I and HhaI. The digest was visualized on 2% agarose gel to determine the Savanna (S) and Mopti (M) population forms (Favia et al. 1997).

Indices of malaria parasite transmission and statistical analysis

The human biting rates were calculated as the number of Anopheles biting per person per night. The parous rates were determined as the proportion of Anopheles found to be parous at a given time and place, and the sporozoite rates were inferred from the proportion of human-biting anophelines that tested positive for PfCSP by ELISA. The EIRs were calculated as the product of the human-biting rate at a given time interval and place and the sporozoite rate. Tests of significance between proportions and means were carried out using chi-squared and t-test analysis, respectively.


Anopheline species abundance

A total of 1144 man-nights (380 for lowland; 382 for rocky highland and 382 for irrigated) of sampling yielded 18 228 anopheline mosquitoes from the three micro-ecological sites during the year. Anopheles gambiae s.l. and Anopheles funestus constituted 94.3% (17192/18228) of the total anophelines captured. Other species collected but in relatively few numbers were Anopheles pharoensis (5.4%, 979/18228) and Anopheles rufipes (0.3%, 57/18228). Significant numbers of anophelines were captured from compounds in the irrigated area, (14007/18228; 76.8%) compared with the lowland (1973/18228; 10.8%) and highland (2,248/18228; 12.3%) areas.

A total of 584 man-nights of collections made inside houses, yielded 10 755 anophelines, of which An. gambiae s.l. and An. funestus constituted 95.9% (10310/10755) of the captures. Out of this indoor biting collections, An. gambiae s.l. and An. funestus represented 94.9% (8103/8535), 99.3%(1038/1045) and 99.5%(1169/1175) in the irrigated, rocky highland and in the lowland areas, respectively.

On the other hand, 560 man-nights of outdoor collections produced 7473 anophelines, of which the two species accounted for 92.1% (6882/7473). This represented 89.8% (4916/5472), 98.8% (1189/1203) and 97.4% (777/798) of the total captures in the irrigated, rocky highland and the lowland areas, respectively. A total of 728 anophelines from the An. gambiae complex were processed by PCR. An. gambiae s.s accounted for the majority (97.7%) of the biting mosquitoes collected with An. arabiensis present in fewer numbers (2.3%). PCR also identified M (78.7%) and S (21.3%) population forms of the An. gambiae s.s. in the study site.

Man-biting rates

Anopheles gambiae and An. funestus were biting the inhabitants throughout the year in the irrigated area with overall biting rate of 36.7 bites per man per night (B/M/N) compared with 5.2 B/M/N and 5.9 in the lowland and rocky highland, respectively. In the wet season, biting rates of 43, 16 and 14 B/M/N were estimated for the irrigated, lowland and rocky highland areas, respectively. These rates declined significantly to 23.3 in the irrigated, 0.9 in the lowland and 1.7 in the rocky highland areas during the dry season. The monthly biting rates of An. gambiae s.l. in the three ecological areas peaked in September 2001, a month after the peak of rainfall in August. In the rocky highland and lowland areas, its biting rates declined drastically during the dry months of December 2001–May 2002. Meanwhile, during the same period in the irrigated area, the biting increased steadily and peaked in February and March 2002. The biting rates of An. funestus in the three ecological sites appeared less affected by rainfall, with rates peaking in June and November at a period when the density of An. gambiae s.l. in the irrigated area was low. Significantly higher numbers of vectors (P < 0.005) in both the irrigated and lowland areas were biting indoors than outdoors (44.6 B/M/N vs. 28.7.3 B/M/N in irrigated; 6.2 B/M/N vs. 4.2 B/M/N in lowland) P < 0.005. However, in the rocky highland, more vectors, mostly An. funestus were biting outdoors (6.3 B/M/N) than indoors (5.5 B/M/N) especially during the peak of rainfall in August 2001.

The biting of An. gambiae s.l. peaked early at 22.00 and 24.00 hours and the biting was sustained till daybreak. However, the biting rate of An. funestus rose gradually through the night and peaked at 04.00–05.00 hours, and continued till 06.00 hours when most people were out of bed. Most of the vectors (67%) that bit during the second half of the night were parous.

Sporozoite rates

Plasmodium falciparum sporozoite rates detected by ELISA in An. gambiae s.l. and An. funestus from the three ecological areas were 7.2% (295/4075) and 7.1% (269/3773), respectively and these were not significantly different (P > 0.05). The overall sporozoite rate of both species was 7.2% (564/7848). None of the An. arabiensis (N = 8) and An. pharoensis (N = 256) tested was found infective. Sporozoite rate of 4.7% (264/5654) for both An. gambiae s.s. and An. funestus in the irrigated area was significantly lower, than in lowland (19.0%, 153/805) and rocky highland (10.6%, 147/1389) areas. The sporozoite rate of both An. gambiae s.s. and An. funestus being 8.5% (541/6390) in KND during the wet season was significantly higher (P < 0.05) than the dry season, 1.6% (23/1458).

Sporozoite rates of individual vectors in all the study areas during the wet season were not significantly different (P > 0.05): An. gambiae s.s. (8.7%, 292/3363) and An. funestus (8.2%, 249/3027) However, during the dry season, (P = 0.05) sporozoite rate were significantly higher at 2.7%(20/746) for An. funestus than An gambiae s.s., (0.4%, 3/712). More (76.5%) of the infective mosquitoes (An. gambiae s.s. and An. funestus) were biting after midnight during the wet season than the dry season with biting peaks occurring at dawn between 04.00–06.00 hours (Figure 1). There was also a steady rise in sporozoite infective bites from An. funestus during the night till daybreak, whereas bites from An. gambiae s.s started to decline after 03.00 hours (Figure 2). Infective An. gambiae s.s. were biting throughout the night, whilst a significantly higher proportion of infective An. funestus, (72.2%) were biting after midnight.

Figure 1.

Distribution of sporozoite infective bites of Anopheles gambiae and Anopheles funestus during wet and dry seasons in Kassena Nankana district.

Figure 2.

Distribution of sporozoite infective bites of Anopheles gambiae and Anopheles funestus by hour of the night in Kassena Nankana district.

Entomological inoculation rates (EIR)

An overall annual EIR of 418 infective bites per person was estimated for the KND during the period of the study. There were variations in the EIR across the micro-ecological sites and seasons. The annual EIR in the irrigated area, 630 was about 2–3 times that of lowland, 360 and rocky highland, 228. The daily EIRs during the wet season were 2.4, 3.2 and 1.5 infective B/M/N in the irrigated, lowland and rocky highland areas, respectively. This reduced about 6–32 times during the dry season to 0.4 in the irrigated area, and 0.1 in the lowland with virtually no transmission recorded during the dry season in the rocky highland areas. Table 1 shows the EIR of An. gambiae s.s. and An. funestus biting indoors and outdoors of houses. The indoor EIRs per night in the irrigated and lowland areas were about twice that of outdoors (2.3 vs. 1.2 in irrigated areas, 1.2 vs. 0.7 in lowlands), whilst in the rocky highland areas, the indoor and outdoor EIRs, (0.6 vs. 0.8) were almost the same.

Table 1.  Comparison of entomological inoculation rates (EIR) inside and outside human dwellings in the three ecological areas in Kassena Nankana district
Infective bites per man per nightInfective bites per man per yearInfective bites per man per nightInfective bites per man per yearInfective bites per man per nightInfective bites per man per year

Figure 3a–d shows the monthly changes in daily EIR of An. gambiae s.s. and An. funestus in the three ecological areas in relation to rainfall. EIR by An. gambiae s.s increased steadily from June 2001 when the rains started, peaked in September and declined into the dry months of October and November 2001. In the irrigated and lowland areas, An. gambiae s.s. was mainly responsible for malaria transmission with An. funestus playing a relatively minor role. In the rocky highland area, both An. gambiae s.s. and An. funestus were playing almost equal roles in transmission although at relatively lower rates.

Figure 3.

a–d Monthly changes in entomological inoculation rates (EIR) for Anopheles gambiae and Anopheles funestus in three ecological areas, in relation to rainfall in Kassena Nankana district. (a) Irrigated area, (b) lowland, (c) rocky highland area and (d) total rainfall.


This study demonstrates both micro-geographical and seasonal variations in malaria transmission in KND. The study shows that transmission was generally high in KND and that individuals living in the district could be exposed to 418 infective bites in a year. In the irrigated area of the district, mosquito vectors were in constant contact with inhabitants throughout the year because of the availability of water from the irrigation system even during the dry season. Thus transmission intensity was much higher compared with the other sites without irrigation. Sixty percent (60%) of infective sporozoite inoculations in the irrigated and lowland areas occurred indoors whereas similar proportions of sporozoite inoculations in the rocky highland areas occurred outdoors. Children living in KND could be potentially exposed to >2000 sporozoite inoculations by the time they are 6 years old.

Malaria transmission in KND occurred mostly during the wet season and fell to very low levels during the long dry season. An. gambiae s.s. constituted more than 90% of the sibling species of the An. gambiae complex in KND. Mopti population forms were dominant in permanent waters in the irrigated area, whereas savannah forms were dominant in temporary shallow waters of lowland and rocky highland areas. This finding is similar to observations in some parts of Burkina Faso where the M form was common in the centre of rice fields, but at the edges of irrigated areas, the S form was more dominant (Robert et al. 1989). An. arabiensis in the KND has no vectorial role in contrast to other areas in the West African sub-region, including Mali, where An. arabiensis could exhibit considerable higher vectorial roles than An. gambiae s.s. (Toure et al. 1996).

Biting from the malaria vectors (An. gambiae and An. funestus) started early in the evening and continued till daybreak (04.00 and 06.00 hours). This pattern is different from the coastal areas of Ghana, where virtually no biting occurred during the early hours of the morning (Appawu et al. 2001). An interesting finding of this study, which has not been reported in many parts of the sub-region, is the distribution of sporozoite-laden bites during the night, which indicated that malaria transmission in the study sites occurred mostly from late evening and peaked at daybreak. This has important epidemiological implications because this is an area where insecticide treated bednets (ITN) have been introduced and are been used by the people, but most of the inhabitants were usually awake during the period of 05.00 and 06.00 hours and out of their bed-nets, thereby increasing their exposure to infective inoculations from the vectors. This finding may explain why ITN use in the area was associated with reduced prevalence of parasitaemia among children but not in adults over 15 years (Koram et al. 2003). This is because unlike children who will be in bed during the peak time of biting of mosquitoes and benefit from the protective effect of the ITN, adults who are likely to be awake at that time preparing for farming and doing other domestic activities would be exposed to the infective bites. The finding of our study also may explain why although ITN usage was higher in the irrigated site than in areas outside the influence of irrigation, prevalence of parasitaemia was significantly higher among children living in irrigated areas.

Irrigation provides ideal breeding sites for the principal vectors of malaria in Africa (Carnevale et al. 1999) and we saw, irrigated fields generated large numbers of mosquitoes but smaller proportions were infective than in non-irrigated communities. There is no simple association between irrigated fields and the degree of exposure to malaria parasites (EIR) as measured using classical entomological methods (Ijumba et al. 2002). Thus, the intensity of transmission or EIR in irrigated communities can either appear higher, similar or less than in neighbouring communities outside the irrigation schemes (Coosemans 1985; Robert et al. 1985; Ijumba 1997; Ijumba & Lindsay 2001). The other finding in this study, which is different from others from Burkina Faso and other parts of the continent (Boudin et al. 1992; Ijumba 1997) indicates that exposure to infective bites of vectors, was higher for people in the irrigated communities than the non-irrigated ones. This was corroborated by the results of prevalent studies carried out concurrently at KND showing that prevalence of parasitemia was significantly higher (P = 0.001) among children living in irrigated areas than those residing outside the influence of irrigation (Koram et al. 2003).

Many inhabitants, particularly in the irrigated area, try to protect themselves and their children from the high biting nuisance by using impregnated bed-nets. However, the fact that most sporozoite inoculations occur at the early hours of the morning when many people were awake with their bed-nets folded puts them more at risk of getting infected. This may explain the higher transmission rates in the irrigated area.

Vaccine trials could be conducted in this district, with timing dependent on the seasonal patterns of transmission and taking into consideration the micro-geographical differences and vaccine trial objectives. In trials where short-term, high-intensity exposure would be required, trials should be planned to coincide with peak levels of transmission. The predictable sharp rise in transmission during and immediately following the short rainy season in KND indicates that optimal periods for vaccination of volunteers would be before the rain begins, to achieve peak antibody responses coinciding with increasing transmission.


Sincere appreciation is extended to all members of the Navrongo Health Research Center and Noguchi Memorial Institute for Medical Research for their administrative, logistical and technical assistance and to Drs Lee Hall, Maureen Power, Abdollah Naficy and Tonu Wali of the NIAID for guidance and support. Special thanks go to members of the NHRC entomology team and to the people of the KND for their cooperation and understanding. Ghanaian committees for the protection of human subjects reviewed and approved the procedures followed in this research. The views of the authors expressed herein do not purport to reflect those of the Ghanaian Ministry of Health or the National Institute of Allergy and Infectious Diseases (NIAID). This work was made possible by National Institute of Allergy and Infectious Diseases (NIAID) contract NO1 AI95363 (Malaria: Clinical Research & Trial Preparation Sites) to the Noguchi Memorial Institute for Medical Research (NMIMR) the Navrongo Health Research Centre (NHRC) and the Naval Medical Research Centre (NMRC).