The Kakuma Refugee Camp is located in an arid environment that, without an exogenous source of water for mosquito larva breeding sites, is unlikely to sustain even intermittent malaria transmission. However, this study describes persistent, malaria attack rates among the refugee population living in the camp and among host nationals living near the camp, with prevalence suggestive of a hyperendemic malaria epidemiology. A subsequent entomological survey, conducted in the dry season, revealed adult anopheline mosquitoes resting inside homes in the camp, when none would be expected given the climatic conditions. Also surprisingly, indoor-resting adult anopheline mosquito densities were nearly as high during the dry season as they were during the wet season. Larval surveys revealed that the vast majority of larvae were found in man-made, tap-stand pits and their drainage systems during both the rainy and dry seasons, indicating that these pits supported mosquito production and, therefore, malaria transmission. They were most likely responsible for most of the malaria-related morbidity and mortality in the rainy season and all malaria in the dry season. Unfortunately, these pits served very little purpose other than as mosquito breeding sites, since the kitchen gardens they were meant to support were rarely implemented by the refugee population (only three such gardens were observed during these surveys).
Human malaria in the refugee camp was entirely due to P. falciparum
infection albeit one slide was read as mixed infection with P. vivax
and one was positive for P. malariae
. P. vivax
is typically not found in Kenya, but is a common infection in Somalia and Ethiopia, including the arid south-west region of Ethiopia bordering Kenya [5
]. Overall results suggest a low level of malaria infection in the camp at the time of sampling (late August 2005), with an estimated attack rate of 0.5 cases per day per 1,000 of camp population. Attack rates varied by age of presenting patient, and were relatively low in the youngest age category (children under two years of age), highest in children and adolescents from two to 17 years old and lowest in the adult age group (age 18 years and older). Young children and pregnant mothers in the camps were provided ITNs, which likely protected these groups from infection. Older children and adolescents have been noted in other populations to be the group least likely to sleep protected by an ITN [19
]. There was no difference in infection prevalence between the different nationalities. The timing of the investigation was too late to gauge the true picture of the epidemic, to capture the epidemic curve, or to estimate the number of deaths due to malaria; thus, malaria data reported here likely reflect the post-epidemic transmission pattern more typical of the camp's endemicity, when the rains had subsided and the dry season had already commenced, and larval stages of vectors were supported solely by man-made habitats. In January 2007, the established early warning system in the camp reported another increase in clinical malaria cases in the camp (more than 2,500 cases per week) from a usual average of less than 4000 cases per month which subsided by the end of the month. This increase was linked to heavy rains and some flooding in the camp in November and December 2006. Hospital records indicated that there were 8 deaths (2 children and 6 adults) out of 4,800 cases though patients with malaria presented with a wide range of symptoms other than fever including running nose, cough and rash. These results emphasize endemic transmission in the camp with cyclical peaks and not a prevalence of malaria due merely to importation of cases as refugees arrived from other endemic areas.
The rainfall patterns observed from the nearby Lodwar meteorological station confirm a seasonal and modest annual precipitation, but do not indicate that the malaria epidemic in Kakuma refugee camp was due to excessive rainfall in 2005. Rainfall was more constrained seasonally and possibly more intense in that year, however, it did not fall outside of the 95% confidence interval for an eight-year average (Figure ).
Entomological studies showed that transmission of P. falciparum
in Kakuma refugee camp was entirely due to A. arabiensis
as it was the only species of malaria vector found in the area. The lack of livestock animals in the camp likely directed most blood feeding of a relatively zoophilic vector to humans and facilitated parasite transmission. Bovine blood meal sources for vectors were unavailable in the refugee camp because residents were not allowed to keep animals in the camps due to lack of space and the risk of raids by cattle rustlers. The rate of malaria parasite infection in the vectors of 3% in the wet season is consistent with rates quantified elsewhere in A. arabiensis
populations during the rainy months of the dry regions of northern, sub-Saharan Africa [21
]. Indoor-resting densities were modest but probably reflected true population densities as outdoor resting habitat was minimal, due to sparse vegetation and high outdoor temperatures. Even male A. arabiensis
were caught indoors, suggesting that the indoor environment was a favourable resting habitat for these mosquitoes; otherwise, adults of both sexes of this species commonly rest outdoors [22
The vector populations were maintained in the camp by a constructed water delivery and catchment system, consisting of a series of tap-stands connected by piping to bore holes, with cemented and soil-lined pits provided to catch spill-over water. These pits had the well-intentioned function of providing irrigation water for kitchen gardens, but such gardens were few in evidence. The most abundant larval habitat was those small bodies of water associated with tap-stand pits, suggesting that the process of pit construction and maintenance underlies the man-made nature of malaria transmission in the camp, and underscores a fundamental conflict between water use and transmission of human malaria. The fact that higher larval densities were found in the much fewer rain-fed puddles and tire tracks is likely a consequence of the rapid concentration of these small environments due to evaporation of water from them, and not to any inherent property of those habitats making them better breeding sites. The similar indoor densities of A. arabiensis
males and females from the dry (February) and wet (June) samples supports this conclusion; otherwise, rainfall would encourage larger vector populations in June than was actually observed. All available types of larval habitats within the camp were colonized by Anopheles
vectors and all parts of the camps had at least one house with adult Anopheles
mosquitoes. However there was aggregation of productive larval habitats in the dry season and houses with highest vector densities in both dry and wet season. This conclusion is supported by the variance to mean ratios of the sampling data [23
], which were all much greater than 1, indicating extreme aggregation (i.e., non-random distribution) of larval stages amongst sampled habitats and adults amongst sampled houses. Camp zones and residential sectors with the highest vector densities also had the highest malaria parasite attack rate, suggesting co-aggregation of human exposure to infectious bites and spatial distribution of infected humans. Further analysis of larger data set with adequate spatial information on patients and vectors and habitats within the camp and the surrounding environment where such constructed water supplies systems do not exist would be required to fully understand these relationships.
The number of malaria vectors inside houses at the different sites was positively correlated with the number of habitats in the site. Due to the nature of the sources of these vectors, larval control either by larvicide application or by source reduction or both could be a useful and easily implemented tool for control of malaria in the camp. It could be targeted and monitored readily because of the distinct nature of the habitats. For instance, all pits that were not used for their intended purposes could be filled with soil or modified with a drain outlet to prevent water from accumulating in them. If this recommendation were not feasible, tap-stand monitors should be posted at every tap to manage the immediate environment of the tap-stand by draining pits twice weekly to interrupt the development of larvae, removing any drainage channels emanating from the tap-stand pits, and reporting the presence of any mosquito larvae to the camp authority for targeted larviciding.
Harvesting waste water from tap-stands makes practical sense given common water shortages in the area. If in use and emptied regularly, these tap-stand pits likely would not have had sufficiently stable water sources for the mosquitoes to complete their immature life stages. Because these tap-stand pits were numerous and remained filled with water at all times, there was a year round presence of vector breeding sites throughout the camp leading to a year round production of malaria vectors and the resultant phenomenon of malaria endemicity in an area with otherwise no malaria or at worst seasonal malaria. These findings underline the relevance of monitoring environmental impact of interventions; an old lesson worth heeding.