Prevalence of malaria parasites in adults and its determinants in malaria endemic area of Kisumu County, Kenya

The sample frame is a sub-county in Kenya, in an area endemic for malaria, namely Maseno division within Kisumu County, Western Kenya which has a population of 70,805 (Kombewa HDSS 2012). Females constitute 53% of the population. The mean household number is four people per household with a population density of about 374 people/km². The population is largely young with a mean age of 23 years. Among the population sampled, those aged 0–14 years constitutes 46%, with those aged 15–64 years and 65+ years constituting 49 and 5%, respectively.

The population is primarily black African, and the languages spoken are Dholuo (predominant ethnic group), Kiswahili and English. The area is largely rural, with most residents living in villages, which are a loose conglomeration of family compounds near a garden plot and grazing land. The majority of the houses are mud-walled with either grass thatched or corrugated iron-sheet roofs. Water is sourced mainly from community wells, local streams and the lake for those living on the shores of Lake Victoria. Most water sources are not chlorinated. Subsistence farming, animal husbandry and fishing are the main economic activities in the area.

Malaria is holoendemic in this area, and transmission occurs throughout the year. The ‘long rainy season’ from late March to May produces intense transmission from April to August. The ‘short rainy season’ from October to December produces another, somewhat less intense, transmission season from November to January. This area was used for an earlier study on prevalence of common mental disorder and psychosis, and their risk factors in 2004 [11, 12].

Figure 1 shows the study site, Maseno area.

The study sample was selected from Maseno division which is sub-divided into four locations, 17 sub-locations and 184 enumeration areas (villages) based on mapping work done earlier by the Kombewa Health and Demographic Surveillance System (Kombewa HDSS) run by the KEMRI/Walter Reed Project. The Kombewa HDSS is a longitudinal population registration system set up to monitor the evolving health and demographic problems of the study population in Kombewa and Maseno divisions [13]. Some villages with less than 50 households were merged together to create new enumeration areas. A total sample size of 1,010 adults was needed for the survey and, since the sample frame included 170 enumeration areas, this necessitated the random selection of seven households in each enumeration area from which to select one adult to give a reasonable safety margin of a sample of 1,190 in order to allow for an estimated 15% refusal rate. Village maps were used to assign households and guide the research assistants during the survey. Using the Kish Grid Method [14], one individual was selected from each of the sampled household. The demographics and reasons for the refusal were recorded in notebooks by the research assistants. A total of 1,190 households were visited.

Meetings were held with the community leaders to explain the purpose of the study and answer questions. The participants in the survey were approached in their own homes for voluntary written and witnessed informed consent, and then capillary blood samples were collected for preparation of thick and thin blood films (reported here) and immunity (which will be reported in accompanying papers). A structured epidemiological interview assessment accompanied the blood tests, which included detailed socio-demographic and housing variables (reported here) and additional sections on psychosocial variables (which will be reported in accompanying papers).

Field staff training was undertaken in November 2012 for 5 days in the KEMRI field centre in Kisumu. Team supervisors and research assistants were trained on the rationale and methodology of the interviews, use of personal digital assistants (PDAs) to collect the interview data, and the methodology for blood samples. Prior to the field work, the communities via their community leaders received information about the study and an opportunity to discuss the project. The field work commenced on December 2012, and continued intermittently until June 2013. The research assistants were driven to the study sites, and were closely supervised by field supervisors. The questionnaires were programmed on to the PDAs and tested before the actual field work. Socio-demographic data were captured on the PDAs. They were then transferred at the end of each day to a personal computer for merging, and transmission to the statistician for analysis.

Venipuncture blood samples of about 5 mL were collected aseptically into EDTA tubes from the enrolled participants before any treatment interventions. Blood smears were immediately prepared, dried and packed carefully in the field, and sent to KEMRI at Kisian, Kisumu, where staining and malaria parasite determination was performed by well trained laboratory technologists. Briefly, thick and thin peripheral blood smears were stained with Giemsa reagent and the prevalence of parasitaemia was determined by counting the asexual parasites against 300 leukocytes assuming a count of 8,000 leukocytes per microlitre of blood.

The prevalence of malaria parasites was analysed in relation to a variety of socio-demographic, housing and economic variables. Predictors of parasitaemia were also examined using STATA [15] to calculate unadjusted and adjusted odds ratios. Households have been categorized into different socio-economic levels using an index of household assets, constructed applying the principal component analysis (PCA) procedure, as a proxy indicator for socio-economic status. In coming up with the asset quintiles, type of house, roofing and walling material, source of water, toilet facility and land have been used [16, 17].

Ethical approval was granted by the Kings College London and Kenya Medical Research Institute Ethics Review committees (PNM/11/12-54 and SSC 2374, respectively). Written informed consent was sought from participants to take part in the study, be interviewed and to give blood, and for the data to be analysed and published. Permission was obtained to conduct the study in households in Maseno area, which is part of the KEMRI/WRP Kombewa HDSS.

This study is the first published survey of the adult prevalence of malaria in Kenya. The survey is of a general household population living in a malaria endemic area, and found that the adult prevalence of malaria in Maseno division is 28% over a 6-month period, ranging from 40 to 60% in December, January and February to around 10% between March and June. We enquired in detail about aspects of housing and assets as well as socio-demographic factors, but the only significant socio-demographic risk factor that we were able to find in this sample size was gender in both univariate and multivariate analyses. Women were 40% more likely to have malaria than men (multivariate analysis). There are at least two plausible contributions to the higher prevalence in women. In the study area, the HIV prevalence is higher in women so this may contribute to the higher prevalence of malaria that we found in women, because HIV is known to cause loss of immunity and thus adults who have HIV are more likely to have malaria than those who do not have HIV [18, 19]. In addition, pregnant women specifically primigravidae and secondigravidae, lose the acquired semi-immunity of adulthood, and are more prone to malaria than other adults so this may also contribute to the higher prevalence found in women [20].

The finding of an adult prevalence of malaria of 28% in an area endemic for malaria is double the 14% microscopy prevalence found in a recent household survey of a malaria endemic area in Mozambique [2]. A 22-year longitudinal study in a village in Senegal found that parasite prevalence declined in adults from 58% in 1990 to 0.3% in 2012 following systematic eradication and treatment interventions. [21].

Previous studies have found that in malaria endemic zones, the risks of malaria attacks declines with age between childhood and adulthood, indicating that protective immunity is acquired with increasing age up to adulthood [22]. This study did not find a change of parasite prevalence across the adult age range, suggesting that there is no further increase in protective immunity, with increasing adult age. Previous studies on general population prevalence of malaria have not been analysed by gender. Mayor et al. [2] did not disaggregate their data on adult malaria by gender, and the national Kenya malaria survey 2010 does not disaggregate childhood parasite prevalence by gender [9]. It is of interest that the malaria prevalence was marginally lower in adults in the highest wealth quartile, similar to the findings of the Kenya malaria survey 2010 in children.

Despite earlier vigorous calls for adult malaria surveys [4] in order to better monitor the transmission of malaria, adult household surveys of malaria are scarce, and it has been strongly argued that epidemiological studies of malaria in Africa are essential for eradication strategies [7]. This study has demonstrated the relatively high prevalence of malaria parasites in adults in an endemic area, ranging from 60% at the height of the malaria season to around 10% in the other months. The mortality of children from malaria remains high in endemic areas, and so there is considerable policy and practice attention to factors reducing transmission, such as use of bed nets, indoor residual spraying, testing of all persons with fever, and treating confirmed cases with artemether–lumefantrine. Despite these efforts, such a high adult prevalence of parasites emphasises the relevance of the adult malaria parasite reservoir on the transmission of disease in both children and adults [23] and it supports the need for further consideration and research evaluation of treatment of asymptomatics, including approaches such as intermittent screening and treatment (IST) as part of an integrated malaria control programme, if malaria elimination is to be achieved [24].

This study represents a contribution to the epidemiology of malaria in adults in an endemic area of Kenya. The probability sample of households and adults was drawn from a demographic surveillance site.

The field work, which had been planned to last 3 months, had to be extended to 6 months to cover interruptions during the 2013 national election, due to concerns about potential election violence. Participating women were not asked if they were pregnant, and this is likely to be a major factor in the increased prevalence of malaria in women compared to men. We did not examine parasite prevalence by polymerase chain reaction (PCR), which generally finds a higher prevalence than microscopy alone.