Malaria is one of the most serious health problems in Kenya. According to the World Health Organization (WHO), in 2002, there were over 124,000 reported cases of malaria nationwide in a population reported at just over 32 million 1. Moreover, cases are likely grossly underreported 2. According to one study, the mortality from malaria in one area of the Lake Victoria Basin was 20.8 per 1000 person-years 3. The malaria burden is increased by high levels of resistance to antimalarial drugs 456789 and by frequent self-administration of antimalarials 1011121314151617.

Between June and September 2004, researchers from the Kenya Medical Research Institute (KEMRI) and the United States Army Medical Research Unit – Kenya (USAMRU-K) carried out a community-based malaria parasitaemia and arboviral serologic survey of residents of three ecologically and culturally diverse Kenyan districts: Samburu, Malindi, and Busia, to assess potential socioeconomic risk factors for exposure to these pathogens. Extensive demographic, socioeconomic, health history, and recent antimalarial use data were collected along with blood for malaria smear and arboviral antibody testing. The malaria parasitaemia rates found in these three districts are reported here. Additionally, this study examines individual risk factors for parasitaemia, such as occupation, literacy, recent illness, health care utilization, and use of antimalarials.

Overall, this study found an association between parasitaemia and district of residence, literacy, and a history of diarrhoeal illness in the previous three months.

An interesting finding of this survey is that a history of diarrhoeal illness is a significant predictor for parasitaemia in Kenyan adults. The literature on the association of diarrhoea with malaria is somewhat conflicting and generally refers to the relationship between diarrhoea and acute malaria, while this survey correlates a history of diarrhoea in the past three months with a current parasitaemia. While diarrhoea is an often-cited symptom of malaria 222324, there are several studies in the literature which demonstrate a negative correlation or lack of association between diarrhoea and malaria, both in non-immune travelers and in residents of endemic areas 25262728. It is important to ascertain whether malaria is presenting as diarrhoeal illnesses in these populations to avoid missing the diagnosis of malaria and to avoid unnecessary antibacterial therapy for diarrhoea.

This positive relationship between a history of diarrhoea in the past three months and a positive malaria smear reached only marginal statistical significance in the aggregate, but was consistent in the individual districts, making type I error an unlikely explanation for this association. There are certainly many possible explanations for this association. It is of considerable interest that self-reported diarrhoea was associated with smear-positivity, whereas self-reported malaria was not. In these semi-immune populations, it is quite possible for diarrhoea to be the predominant symptom of malaria. As a diarrhoeal illnesses may not be recognized as malaria, treatment for malaria may not be sought or provided. While this difference in obtaining antimalarial therapy could explain a higher risk of parasitaemia among those with self-reported diarrhoea than self-reported malaria, this study does not support this as the explanation. While 30% of those with a history of malaria report taking an antimalarial in the previous month, 29% of those with a history of diarrhoea also report doing so. This difference is not statistically significant (p = 1.000). However, both of these groups were more likely to have taken an antimalarial than those with other recent reported illnesses, of whom 14% took antimalarials (RO = 2.62, 95% CI 1.84 – 3.75) or those with no reported illness, of whom 6% took antimalarials (RO = 6.52, 95% CI 4.06 – 10.48). Of those with a history of either diarrhoea or malaria, those with a positive malaria smear were no more or less likely to have taken an antimalarial in the past month than those with a negative smear (p = 0.531, p = 0.708). There was no difference in the antimalarials taken by those with diarrhoea and those with malaria (p = 0.869). Further study of this association of diarrhea with parasitaemia would be useful in providing guidance for evaluation of diarrhoea in these populations and perhaps would prevent inappropriate antibiotic therapies from being administered.

Literacy was another significant risk factor for parasitaemia in this study, even after controlling for potential confounding effects. This was a consistent finding in all three districts, although not reaching statistical significance for individual districts. This association cannot be attributed to occupation, recent antimalarial treatment, or other known factors. Literacy can directly impact health-seeking behaviors and also serves as a marker for other socioeconomic differences between individuals. It is plausible that differences in health-seeking behaviors or other socioeconomic factors would affect rates of malaria infection. However, literacy would have been anticipated to be protective against, not a risk factor for, infection as literacy and education are believed to be associated with more appropriate use of medical care 29303132333435. There is some evidence, though, that higher educational level does not correlate with appropriate judgment of illness severity 36. This observation of higher parasitemia rates among literate subjects warrants further study.

A clear finding of this study is that district of residence is the factor most strongly associated with differences in parasitaemia rates at the time of enrollment. Specifically, compared to those enrolled in Samburu, the subjects enrolled from Busia and, less so, Malindi, had higher odds of parasitaemia, even after adjustment for known potential confounders. The difference in ecologies of a semi-arid, a coastal, and a Lake Victoria basin district could well result in different disease transmission rates, particularly for vector-borne diseases 373839. District differences in public health infrastructure, sanitation, and cultural practices could also account for differences in disease occurrence and health perceptions.

There was no statistically significant association between history of antimalarial use in the past four weeks and the presence of parasitaemia on enrollment. This study did not have sufficient power to detect an association between specific antimalarial use and parasitaemia, even if such an association actually existed. Given an overall rate of 12.2% positivity, to find a 5% difference in malaria positivity with a power of 80% would require 826 subjects with a history of antimalarial use, a much higher number than the 187 enrolled.

There are several further limitations of this study. First is the use of self-reported data, particularly in the area of recent health histories and antimalarial usage. As noted above, the low number of patients with a recent history of antimalarial use limits the study's ability to investigate antimalarial resistance. Also, as previously noted, the enumeration unit was the village, not the individual subject.

While district of residence was the variable most highly predictive of parasitaemia among Kenyan adults surveyed, both a recent history of diarrhoeal illness and literacy were at least marginally statistically significant predictors. No other demographic or occupational variable, including age and sex, was associated with parasitaemia in these populations. Further study is required to determine the nature of these associations.

RLC served as overall Principal Investigator on the study, involved in all aspects of protocol design and execution, data analysis, and manuscript preparation. TP and FO designed the sociodemographic questionnaire, supervised field workers, and assisted in manuscript preparation. VOO designed the laboratory system and supervised the reading of malaria slides. NA conceived the study, supervised field and data entry personnel, and provided significant contribution to the manuscript. All authors read and approved the final manuscript.