Identification of malaria transmission and epidemic hotspots in the western Kenya highlands: its application to malaria epidemic prediction
© Wanjala et al; licensee BioMed Central Ltd. 2011
Received: 9 February 2011
Accepted: 19 May 2011
Published: 19 May 2011
Malaria in the western Kenya highlands is characterized by unstable and high transmission variability which results in epidemics during periods of suitable climatic conditions. The sensitivity of a site to malaria epidemics depends on the level of immunity of the human population. This study examined how terrain in the highlands affects exposure and sensitivity of a site to malaria.
The study was conducted in five sites in the western Kenya highlands, two U-shaped valleys (Iguhu, Emutete), two V-shaped valleys (Marani, Fort-Ternan) and one plateau (Shikondi) for 16 months among 6-15 years old children. Exposure to malaria was tested using circum-sporozoite protein (CSP) and merozoite surface protein (MSP) immunochromatographic antibody tests; malaria infections were tested by microscopic examination of thick and thin smears, the children's homes were georeferenced using a global positioning system. Paired t-test was used to compare the mean prevalence rates of the sites, K-function was use to determine if the clustering of malaria infections was significant.
Results and Discussion
The mean antibody prevalence was 22.6% in Iguhu, 24% in Emutete, 11.5% in Shikondi, 8.3% in Fort-Ternan and 9.3% in Marani. The mean malaria infection prevalence was 23.3% in Iguhu, 21.9% in Emutete, 4.7% in Shikondi, 2.9% in Fort-Ternan and 2.4% in Marani. There was a significant difference in the antibodies and malaria infection prevalence between the two valley systems, and between the two valley systems and the plateau (P < 0.05). There was no significant difference in the antibodies and malaria infection prevalence in the two U-shaped valleys (Iguhu and Emutete) and in the V-shaped valleys (Marani and Fort Ternan) (P > 0.05). There was 8.5- fold and a 2-fold greater parasite and antibody prevalence respectively, in the U-shaped compared to the V-shaped valleys. The plateau antibody and parasite prevalence was similar to that of the V-shaped valleys. There was clustering of malaria antibodies and infections around flat areas in the U-shaped valleys, the infections were randomly distributed in the V-shaped valleys and less clustered at the plateau.
This study showed that the V-shaped ecosystems have very low malaria prevalence and few individuals with an immune response to two major malaria antigens and they can be considered as epidemic hotspots. These populations are at higher risk of severe forms of malaria during hyper-transmission seasons. The plateau ecosystem has a similar infection and immune response to the V-shaped ecosystems. The U-shaped ecosystems are transmission hotspots.
Methods and materials
Sample size calculation
Because malaria prevalence in the study area was not well known, the sample size calculation was carried out assuming the prevalence was 50%. The sample size was calculated to achieve a 95% confidence and precision level of 5%.
The sample size required was 384 children from all the study sites, i.e. 77 children per site. Assuming a truancy of 10%, a total sample size of 425 children was obtained from a population of 600 children (120 from each site) who had given consent to participate in the study. The homes of the 600 children were geo-referenced, mapped, and tested for clustering. The human population density in all the study sites is very similar and so equal numbers of children were allocated to each site. Of the final sample size of 425 children, each site was allocated 85 children.
Cohorts of 170 children in the two U-shaped valleys, 170 children in the two V-shaped valleys and 85 children at the plateau aged 6-15 years were recruited for monthly Plasmodium falciparum surveys. The cohorts were followed monthly for 16 months (from May 2009- August 2010). Before the children were allowed to participate in the study, consent was obtained from their parents or guardians. Children aged between 6-15 years residing in the study sites and with no reported chronic illness, except malaria, were allowed to participate in the study. Children who were found having fever at the time of sampling were taken to the nearest government clinic for treatment.
Immunoassays for exposure to malaria
The SCIMEDX (SCIMEDX Corporation, Deville, New Jersey, USA) rapid diagnostic antibody test kit contains a membrane strip, which is pre-coated with recombinant malaria P. falciparum capture antigens (MSP, CSP) on the test band P. falciparum region and with recombinant malaria P.vivax capture antigens (MSP, CSP) on the test band P.vivax region. The recombinant malaria Plasmodium vivax (P.v)./Plasmodium falciparum (p.f). antigen (MSP, CSP) -colloidal gold conjugate and serum sample moves along the membrane chromatographically to the test region (P.f, or P.v.) and forms a visible line as antibody-antigen gold particle complex forms with 91.3% sensitivity and 98.5% specificity. The control line is used for procedural control. Blood samples were collected by a standard finger prick and drawn into capillary tubes, then centrifuged at 1,800 revolutions per minutes for 5 minutes to obtain the serum. After the test kit had been brought to room temperature, a drop of serum was added in to the sample window and allowed to soak in, and then two drops of the diluent were added into the sample window. The positive results were read after 10 minutes and the negative results confirmed after 20 minutes as indicated by the kit manufacturer.
Microscopy for the presence of the malaria parasites
Blood samples were collected by the standard finger-prick method and thick and thin smears prepared on labeled slides. The smears were allowed to air dry then fixed in methanol and stained in 4% Giemsa for 30 min. The stained smears were then examined using the magnification of × 1,000 (oil immersion) to identify and count the parasite species. Random checks were carried out on the slide counts by independent microscopists to ensure quality control. Parasite density was scored against 300 leukocytes in positive slides. Parasite densities were converted to number of parasites per microliter of blood, assuming a leukocyte count of 8,000 cells/μl .
Spatial distribution maps
The homes of the participants were geo-referenced using a global positioning system (GPS) unit. A database was created with participants' names, ages, GPS location, and malaria parasite infection results. Participants' names were coded for confidentiality. ArcView 3.3 (Environmental Systems Research Institute, Redlands, CA, USA) was used to create spatial-distribution maps of infected and uninfected participants.
Mean monthly rainfall, maximum and minimum temperature for the study sites was obtained from the Kenya Department of Meteorology.
Site specific prevalence rates were determined by expressing positive blood smears as a percentage of the total blood smears examined. Only positive slides were considered when the geometric mean parasite density was calculated for each site. The paired t-test was used to determine differences of the prevalence of antibody and malaria parasite infections among the sites. To determine spatial distribution patterns of malaria infections in the study area, ArcView 3.3 (Environmental Systems Research Institute, Redlands, CA, USA) was used to create Spatial-distribution maps of infected and uninfected participants for the 5 sites. Plasmodium falciparum infections were tested for clustering by household with the global spatial statistic, the K function, weighted by parasite density [9–11] by using Point Pattern Analysis software (San Diego State University, San Diego, CA, USA). The global weighted K function, L(d), examines the spatial dependence of malaria infection by household over a wide range of scales of pattern . The observed L(d) function values were tested against the null hypothesis that the spatial distribution of all infected residents in the study area was random, by using 1,000 Monte Carlo iterations. The global spatial cluster analysis was conducted separately for the five sites.
Scientific and ethical clearance was given by Kenya Medical Research Institute. Inclusion criteria were provision of informed consent, age >5 years at recruitment, and no reported chronic or acute illness, except malaria.
Weather data from the meteorological stations near the study sites
Summary of water data from meteorological stations near the study sites
Kakamega vs Kisii
Kakamega vs Kericho
Kericho Vs Kisii
Total annual rainfall (mm)
p > 0.05
p > 0.05
p > 0.05
Mean annual maximum temp °C
p < 0.05
p < 0.05
p < 0.05
Mean annual minimum temp °C
p < 0.05
p < 0.05
p < 0.05
Mean annual temperature °C
p > 0.05
p < 0.05
p < 0.05
Thus rainfall and mean annual temperatures were not significantly different between Kakamega and Kisii districts.
Prevalence of P. falciparum malaria infections
Mean parasite density in the five study sites
P. falciparum gametocyte prevalence in the five sites
In Marani, no gametocytes were detected throughout the study period, while in Fort Ternan the mean prevalence was 0.8%, Shikondi 1%, Iguhu 2.8% and Emutete 3.3%.
The prevalence of gametocytes in the five sites
Prevalence of CSP-MSP antibodies in the five study sites
Spatial distribution of malaria cases
The majority of children with parasite positive blood smears in the U-shaped valleys were found infected on 4-6 occasions during 12 months while those in the V-shaped valleys and the plateau had 1-2 infections per year. Thus the children in the U-shaped valleys were infected for a longer period. (Figures 7 and 8)
Temporal variation of P. falciparum malaria prevalence in the highlands of western Kenya is a function of meteorological variables (rainfall, temperature and humidity) while the spatial variation is a function of the terrain characteristics. Furthermore these two drivers of malaria transmission have a direct impact on the development of an immune response to malaria infections. In this study we examine how the terrain characteristics are associated with prevalence levels of asexual and sexual stages of P. falciparum malaria spatial distribution of infections and the immune response. Knowledge gained has contributed towards differentiation of malaria epidemic and transmission hotspots which will further lead to a better understanding of the evolution of malaria epidemics.
Rainfall was similar among all the sites and the mean annual temperatures were not significantly different between Kakamega (Iguhu) and Kisii (Marani) which have U and V-shaped valleys respectively. Despite having similar weather, the two sites had very different P. falciparum malaria parasite prevalences, immune response and spatial distribution of malaria infections. The U-shaped valleys system had 8.5-fold more malaria infections and 2.6-fold greater prevalence of antibodies compared to the V- shaped valleys systems. This observation has implications in terms of predicting the outcome of weather-driven malaria epidemics and their prevention and control. In the highlands, most severe malaria cases during an epidemic come from human populations that have not been regularly exposed to malaria infections . Physically the highlands of western Kenya are not a homogeneous ecosystem and have been shown to have a heterogeneous eco-epidemiology . Thus while the climate may be similar in many places malaria infections also respond to terrain characteristics such as the shapes of the valleys which determine the availability and stability of vector breeding habitats and subsequently the level of malaria transmission and immunity. The outcome of a malaria epidemic is a direct function of the level of transmission and immunity.
The early malaria epidemic prediction model developed by Githeko and Ndegwa  detects temperature and rainfall anomalies that are supportive of increased vector populations and rapid sporogonic parasite development. It has been known that certain areas in the highlands are more prone to epidemics than others, but causes of this phenomenon were not clearly understood. In the western Kenya highlands the former Kisii, Kericho and Nandi districts have had a long history of severe malaria epidemics. We recognize that these districts have V-shaped valleys. In Kakamega district unpublished data (Ouna & Githeko unpublished) indicated a sharp increase in severe malaria cases at the plateau village of Shikondi during the 1997/8 El Niño driven epidemics. Malaria prevalence in this village is low (4.7%) and mean proportion of children with antibodies in this study was 11.5%, leaving a large proportion of the population with no detectable immune response.
During late 2009, El Niño type of rains started in September and continued to December. An increase in the proportion of children with antibodies increased by 24.2% in the U-shaped valleys system and this was associated with a 13.4% decrease in malaria prevalence in December. In the V-shaped valleys system an increase of 20.8% in the prevalence of antibodies was associated with only 1.7% decrease in malaria prevalence in November. These observations suggest that the immune response in the U-shaped valley systems may be more robust than that of the V-shaped valley systems. Besides a higher prevalence in malaria infections in the U-shaped valleys system, children were infected 3-fold more frequently than in the V-shaped valley system which may contribute to a better development of an immune response to malaria. Our results also showed a high positive correlation between site-specific mean antibodies and site specific mean parasite prevalence rates; this indicates that the site-specific malaria infection prevalence is predictive of the site-specific prevalence of CSP and MSP antibodies. While this relationship was strong at the population level, the relationship at the individual level was much less obvious and could have been affected by the sensitivity of kit, antigenic polymorphism and delayed immune response. Other studies have shown a log-linear relationship between exposure and antibodies prevalence predicted by mathematical models assuming exposure-dependent immunity . The log-linear relationship between transmission intensity and prevalence of both infections and enlarged spleens thus support the existence of exposure-dependent acquired immunity .
The high gametocytes prevalence rates in the U-shaped valleys compared with the V- shaped valleys indicate that the reservoir of malaria infections in the U-shaped valleys is higher than in the V-shaped valleys and the plateau. While the population living in the U-shaped valley maintains a large reservoir of infectious gametocytes, the people living in the V-shaped valley comprise a high proportion of susceptible individuals. Under permissive climatic conditions the infectious vector population could increase, leading to higher rates of malaria prevalence. Earlier studies indicated that the mean gametocyte prevalence in Kericho district was 1.8% , 2.8% in Kakamega district . In contrast gametocyte prevalence of 39.1% was reported in the holoendemic Kisumu District during the rainy season  and a rate of 6.6% in the mesoendemic Suba district, Western Kenya . In our study the mean gametocyte density was significantly different between the V-shaped, the U-shaped valleys, and the plateau. The low prevalence of gametocytes in the V-shaped valleys with no gametocytes observed in Kisii indicates that there is a weak transmission system in the V-shaped valley.
Spatial analysis of the malaria antibodies and infections indicated that there was a significant positive clustering of malaria infections in the flat bottomed U-shaped valleys resulting from the high availability and stability of vector breeding habitats. In contrast malaria infections were randomly distributed in the V-shaped valleys, and this can be explained by the fact that these valley systems are characterized by fast running rivers at the valley bottoms, and steep slopes that provide good drainage in the area and so there are few vector breeding habitats in these ecosystems. We also observed less clustering in the plateau though it was not statistically significant. Earlier studies in the same area indicated that malaria was mesoendemic at the plateau at 26.7% prevalence . Our results are consistent with studies at Iguhu which indicated that Anopheline larval habitats were generally clustered near the streams and rivers . Similar studies have shown that the risk of malaria is strongly associated with distance from breeding sites . Lower altitude within a highland area has been described in several studies as a risk factor for malaria [21, 22]. Vector densities have been shown to cluster in low-lying flat areas , and reclaimed swamps [20, 23].
Ongoing studies in the same areas indicate that the V-shaped valley ecosystems require anomalously high temperatures and rainfall over an extended period for epidemics to occur. These epidemics are defined by the numbers of people infected, severity of the disease and mortality. In Kericho district Western Kenya out of 254 malaria deaths 31% were due to cerebral malaria, 37% severe anemia and 32% due to malaria with gastroenteritis or pneumonia . It is important to know which populations in the highlands are at high risk of severe disease during epidemics. Such areas require early and effective interventions to reduce high morbidity and mortality. Severe forms of malaria require hospitalization and treatment with quinine or blood transfusion and the demand on medical services can be tremendous.
As a strategy for epidemic prevention special attention should be focused on the V-shaped valley and plateau ecosystems as they constitute epidemic hotspots. It is likely that malaria could be eliminated in the current epidemic hotspots however the current transmission hotspots may convert to epidemic hotspots as the current malaria control efforts reduce transmission.
This study has provided evidence that the V-shaped ecosystems have very low malaria prevalence and few individuals with an immune response to two major malaria antigens. This ecosystem can be regarded as an epidemic hotspot. These populations are a higher risk of severe forms of malaria during periods that support hyper-transmission of the parasites. When predicting malaria epidemics this factor should be taken into consideration. Plateaus with water bodies may also behave like the V-shaped valleys. It is critical that ecosystems sensitivities are taken into account in predicting and controlling malaria.
Authors thank the International Development Research Centre, Canada (IDRC) and DFID (UK) for funding this project (Project ID: 104707-001) through the Climate Change Adaptation in Africa (CCAA) program.
We also extend our gratitude to Lucy Atieli, Edna ototo, Judith Otete, David Cheryot, Nicolas Juma, Jared Rioba and Eliud Simiyu for their technical assistance. We wish to thank Prof Laban Ogallo for providing the project with meteorological data, to Dr Yaw Afrane and Eliningaya Kweka for their contribution in this work.
This paper has been published with the permission of the Director, Kenya Medical Research Institute, Kisumu.
- Hay SI, Simba M, Busolo M, Noor AM, Guyatt HL, Ochola SA, Snow RW: Defining and detecting malaria epidemics in the highlands of Western Kenya. Emerg Infect Dis. 2002, 8: 555-562.PubMed CentralView ArticlePubMedGoogle Scholar
- Githeko AK, Ndegwa W: Predicting malaria epidemics using climate data in Kenyan highlands: a tool for decision makers. Global Change and Human Health. 2001, 2: 54-63. 10.1023/A:1011943131643.View ArticleGoogle Scholar
- Noland GS, Hendel-Paterson B, Min XM, Moormann AM, Vulule JM, Narum DL, Lanar DE, Kazura JW, John CC: Low prevalence of antibodies to preerythrocytic but not blood-stage Plasmodium falciparum antigens in an area of unstable malaria transmission compared to prevalence in an area of stable malaria transmission. Infect Immun. 2008, 76: 5721-5728. 10.1128/IAI.00591-08.PubMed CentralView ArticlePubMedGoogle Scholar
- Afrane YA, Lawson BW, Githeko AK, Yan G: Effects of microclimatic changes caused by land use and land cover on duration of gonotrophic cycles of Anopheles gambiae (Diptera: Culicidae) in Western Kenya highlands. J Med Entomol. 2005, 42: 974-980. 10.1603/0022-2585(2005)042[0974:EOMCCB]2.0.CO;2.View ArticlePubMedGoogle Scholar
- Ndenga B, Githeko A, Omukunda E, Munyekenye G, Atieli H, Wamai P, Mbogo C, Minakawa N, Zhou G, Yan G: Population dynamics of malaria vectors in Western Kenya highlands. J Med Entomol. 2006, 43: 200-206. 10.1603/0022-2585(2006)043[0200:PDOMVI]2.0.CO;2.View ArticlePubMedGoogle Scholar
- Himeidan YE, Zhou G, Yakob L, Afrane Y, Munga S, Atieli H, El-Rayah el A, Githeko AK, Yan G: Habitat stability and occurrences of malaria vector larvae in Western Kenya highlands. Malar J. 2009, 8: 234-10.1186/1475-2875-8-234.PubMed CentralView ArticlePubMedGoogle Scholar
- Himeidan YE, Malik EM, Adam E: Epidemiological and seasonal pattern of malaria in irrigated areas of Eastern Sudan. Am J Infect Dis. 2005, 1: 75-78.View ArticleGoogle Scholar
- Slutsker L, Taylor TE, Wirima JJ, Steketee RW: In-hospital morbidity and mortality due to malaria-associated severe anaemia in two areas of Malawi with different patterns of malaria infection. Trans R Soc Trop Med Hyg. 1994, 88: 548-551. 10.1016/0035-9203(94)90157-0.View ArticlePubMedGoogle Scholar
- Getis A, Franklin J: Second-order neighborhood analysis of mapped point patterns. Ecology. 1987, 68: 473-477. 10.2307/1938452.View ArticleGoogle Scholar
- Haase P: Spatial pattern analysis in ecology based on Ripley's K function: introduction and methods of edge correction. J Veg Sci. 1995, 6: 575-582. 10.2307/3236356.View ArticleGoogle Scholar
- Ripley B: The second-order analysis of stationary point processes. J Appl Probab. 1976, 13: 255-266. 10.2307/3212829.View ArticleGoogle Scholar
- Githeko AK, Ayisi JM, Odada PK, Atieli FK, Ndenga BA, Githure JI, Yan G: Topography and malaria transmission heterogeneity in western Kenya highlands: prospects for focal vector control. Malar J. 2006, 5: 107-10.1186/1475-2875-5-107.PubMed CentralView ArticlePubMedGoogle Scholar
- Zhou G, Minakawa N, Githeko AK, Yan G: Association between climate variability and malaria epidemics in the East African highlands. Proc Natl Acad Sci USA. 2004, 101: 2375-2380. 10.1073/pnas.0308714100.PubMed CentralView ArticlePubMedGoogle Scholar
- Gupta S, Snow RW, Donnelly CA, Marsh K, Newbold C: Immunity to non-cerebral severe malaria is acquired after one or two infections. Nat Med. 1999, 5: 340-343. 10.1038/6560.View ArticlePubMedGoogle Scholar
- Bodker R, Msangeni HA, Kisinza W, Lindsay SW: Relationship between the intensity of exposure to malaria parasites and infection in the Usambara Mountains, Tanzania. Am J Trop Med Hyg. 2006, 74: 716-723.PubMedGoogle Scholar
- Ayisi JM, Githeko AK, Owaga ML, Ekisa WS, Anyona DB, Obala AA, Oloo AJ: A survey of malaria endemicity in Kericho District, Kenya. Proceedings of the 12th Annual Medical Scientific Conference of KEMRI/KERTI,. Nairobi. 1991, 242-246.Google Scholar
- Munyekenye OG, Githeko AK, Zhou G, Mushinzimana E, Minakawa N, Yan G: Plasmodium falciparum spatial analysis, Western Kenya highlands. Emerg Infect Dis. 2005, 11: 1571-1577.PubMed CentralView ArticlePubMedGoogle Scholar
- Githeko AK, Brandling-Bennett AD, Beier M, Atieli F, Owaga M, Collins FH: The reservoir of Plasmodium falciparum malaria in a holoendemic area of Western Kenya. Trans R Soc Trop Med Hyg. 1992, 86: 355-358. 10.1016/0035-9203(92)90216-Y.View ArticlePubMedGoogle Scholar
- Gouagna LC, Okech BA, Kabiru EW, Killeen GF, Obare P, Ombonya S, Bier JC, Knols BG, Githure JI, Yan G: Infectivity of Plasmodium falciparum gametocytes in patients attending rural health centres in Western Kenya. East Afr Med J. 2003, 80: 627-634.PubMedGoogle Scholar
- Minakawa N, Sonye G, Mogi M, Yan G: Habitat characteristics of Anopheles gambiae s.s. larvae in a Kenyan highland. Med Vet Entomol. 2004, 18: 301-305. 10.1111/j.0269-283X.2004.00503.x.View ArticlePubMedGoogle Scholar
- Brooker S, Clarke S, Njagi JK, Polack S, Mugo B, Estambale B, Muchiri E, Magnussen P, Cox J: Spatial clustering of malaria and associated risk factors during an epidemic in a highland area of Western Kenya. Trop Med Int Health. 2004, 9: 757-766. 10.1111/j.1365-3156.2004.01272.x.View ArticlePubMedGoogle Scholar
- Drakeley CJ, Carneiro I, Reyburn H, Malima R, Lusingu JP, Cox J, Theander TG, Nkya WM, Lemnge MM, Riley EM: Altitude-dependent and -independent variations in Plasmodium falciparum prevalence in northeastern Tanzania. J Infect Dis. 2005, 191: 1589-1598. 10.1086/429669.View ArticlePubMedGoogle Scholar
- Lindblade KA, Walker ED, Onapa AW, Katungu J, Wilson ML: Land use change alters malaria transmission parameters by modifying temperature in a highland area of Uganda. Trop Med Int Health. 2000, 5: 263-274. 10.1046/j.1365-3156.2000.00551.x.View ArticlePubMedGoogle Scholar
- Malakooti MA, Biomndo K, Shanks GD: Reemergence of epidemic malaria in the highlands of Western Kenya. Emerg Infect Dis. 1998, 4: 671-676. 10.3201/eid0404.980422.PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.