Effect of climatic variability on malaria trends in Baringo County, Kenya

Malaria is a global health problem that causes an estimated 438,000 deaths annually; 88% of which occur in the sub-Saharan Africa [1]. Seventy-five percent of the malaria clinical episodes worldwide occur in Africa with a corresponding high public health burden [2]. Up to 35.4 million disability adjusted life years (DALYs) are lost in the sub-Saharan Africa region alone due to malaria mortality and morbidity [3].

In Kenya, malaria is among the leading causes of morbidity and mortality and is responsible for almost half of all outpatient attendance and 20% of all admissions to health facilities [4]. Pregnant women and children under five years old are most vulnerable to malaria infections [5] with an estimated 170 million working days being lost to malaria in Kenya each year [6].

The high burden of malaria in Kenya and the larger sub-Saharan Africa region may be associated with a number of factors among them climatic and environmental [7]. Given that malaria is vector-transmitted, with a complex life cycle in both the mosquito and human, transmission and patterns of malaria infection are dependent on both environmental and climatic factors [8]. A study by Githeko et al. [9] showed that inter-annual and inter-decadal climate variability influences the epidemiology of vector-borne diseases directly, while temperature and rainfall have long been known to influence seasonal and inter-annual variability of malaria [10].

The effect of temperature on the life history traits of mosquitoes and malaria transmission has been reported. Temperature can affect the development time of mosquito larvae, the probability of mosquito survival and the development time of malaria parasite (Plasmodium falciparum) in infected mosquitoes either positively or negatively [11]. A rise in temperature to a certain threshold can accelerate the metabolic rate of vectors, increase egg production and increase frequency of blood meals, while temperatures below or above these thresholds can be detrimental to mosquitoes and parasite development [12]. Several mechanistic models concur that effect of temperature on malaria transmission is non-linear, limited to temperatures between 16 and 34 °C with a peak at 25 °C [13‐15]. The non-linear temperature sensitivities throughout the mosquito life cycle have a large impact on the adult population dynamics and, therefore, on the mosquitoes’ ability to act effectively as malaria vectors.

Rainfall influences vector longevity indirectly by creating wet conditions that favour vector breeding. This in turn influences the geographical range and seasonal variability of disease vectors [16]. The relationship between malaria incidence and rainfall is non-linear, implying that an increase in precipitation would not necessarily increase malaria cases [17]. Moderate rainfall has a positive effect on mosquito abundance, while intense precipitation can wash away mosquito breeding sites, and therefore reduce malaria transmission shortly following heavy rains [18]. The Normalized Difference Vegetation Index (NDVI) is a spectral measure of amount, relative greenness, phenological characteristics and productivity of vegetation [19]. It is defined as the difference between the visible (RED) and near-infrared (NIR) bands over their sum, (NIR−RED)/(NIR + RED). It is a robust indicator of vegetation condition which allows valid comparisons of seasonal and inter-annual variations in vegetation growth and activity [20]. In the study area, the seasonal NDVI variations are linked to rainfall. NDVI values range between −1 to +1 An NDVI value of zero means no green vegetation and close to +1 (0.8–0.9) indicates the highest possible density of green leaves. NDVI can be used as a surrogate for precipitation based on their close correlation [21]. The capability of NDVI time-series to monitor and predict vector-borne diseases depends on the correlation between disease incidence, vegetation greenness and precipitation [22].

The nature of vector biological processes and the degree to which the vectors depend on environmental and climatic factors makes malaria transmission somewhat region specific [23]. In Kenya, there are four epidemiological zones whose diversity in malaria transmission and risk is determined by altitude, rainfall patterns and temperature. The zones include: the endemic Lake Victoria and coastal regions, epidemic-prone Western highlands, seasonal transmission arid and semi-arid areas, and low risk central highlands [24]. Parts of Baringo County being semi-arid experience seasonal malaria transmission [25], while the presence of numerous seasonal and permanent water bodies provides suitable breeding microhabitats for malaria vectors at certain times of the year.

This study was thus necessitated by the lack of information on the interactions of climatic and environmental factors, and their role in driving the transmission and prevalence of malaria across the different ecological zones of Baringo County, which has been hampering the planning of intervention strategies against malaria. The study modelled the effect of climatic variations on the prevalence and long-term trend of malaria so as to identify the seasonal climatic drivers of malaria transmission in the different ecological zones of Baringo County.

Climatic factors are considered important in the spatial and temporal distribution of vector borne diseases as they determine vector distribution, and influence inter-annual variability, epidemics and long-term trends [15]. There is a strong discernible link between malaria outbreaks, temperature [13] and rainfall [38]. In the current study, malaria cases generally increased in highland and mid-altitude zones but decreased in the riverine and lowland zones from the year 2011 onwards during the study period (Fig. 2). Two malaria peak seasons were identified in the lowland zone while three malaria peak seasons were identified in the other zones, largely following climatic seasons in the study area. Statistically significant differences in monthly malaria peaks was recorded in the highlands and mid-altitude zones suggesting seasonal malaria transmission. However there was no statistical significance in malaria peaks in the lowland and riverine zones, suggesting that malaria transmission in these two zones is perennial rather than seasonal as previously thought [24].

The mean monthly rainfall for this study showed positive significant correlation with malaria cases at 2 months lag across all zones while the mean maximum temperature showed positive significant correlation with malaria cases in two zones, the highlands and the riverine zones. Previous studies examining the link between climate and malaria established lagged associations between climate variables (temperature and rainfall) and malaria cases over time periods ranging from weeks to months [10, 18, 39‐42]. These studies attributed the lags to the creation of mosquito breeding habitats, the time required by mosquitoes to develop to adulthood, acquire and transmit malarial infection, and for symptoms to arise in the human host as the most probable cause.

According to Confalonieri et al. [43], periods of unusually high rainfall, altered humidity or warmer temperatures can result in modified distribution and duration of malaria, as well as increased transmission; even in areas where control is strong. Consistent with the current findings, Small et al. [44] cited precipitation and temperature as key drivers of malaria case variations across Africa, while acknowledging the complexities of some climatic factors.

The difference in environmental relationship to malaria cases across the zones is attributed to variations in environmental factors between the zones. The mid-altitude zone has no rivers or water bodies and rainfall is therefore the only source of surface water that serves as breeding points for malaria vectors. Although the other zones have permanent water bodies in the form of lakes, rivers, swamps, dams and water pans, rainfall still contributes to malaria cases through creation of additional seasonal breeding sites for malaria vectors.

Consistent with our study findings, Chaves et al. [38] cited increased microhabitats resulting from relative humidity caused by moderate rainfall. These conditions increase the longevity of adult mosquitoes by prolonging vector life span. Paaijmans et al. [45] highlighted the complex interrelationship between precipitation and vectors, noting that drought may eliminate mosquito habitats, while floods could create isolated pools suitable for vector breeding. The relatively low annual rainfall in the mid-altitude zone and the general absence of permanent water bodies contributed to the observed low but varying numbers of recorded malaria cases; possibly due to the varying climatic conditions. All in all, there is a general consensus that rainfall can influence malaria transmission either positively by creating suitable habitats or negatively by flushing breeding sites depending on its intensity [46, 47].

Temperature plays a key role in malaria transmission by influencing vector and parasite life cycles. Studies have highlighted the biological amplification nature of temperature on mosquitoes [48]. This study showed that the mean maximum temperatures within the four zones varied. While the mean maximum temperature significantly influenced malaria cases at lag 0 in the riverine zone and lag 1 in the highlands, it was non-significant in the mid-altitude and lowland zones. The difference in the contribution of maximum temperature to malaria cases between zones is attributed to the differences in prevailing temperatures in the four zones. Being colder, temperature was probably the limiting factor in malaria vector development in the highland and riverine zones; hence a rise in the maximum temperature increased vector and parasite development rates [40]. Since temperature influences the development and survival rates of both vectors and parasites, malaria transmission rates tend to increase with increasing temperature but up to a given threshold [49].

Craig et al. [14] put the optimal temperatures for malaria transmission at between 22 and 32 °C, while Bi et al. [50] reported temperatures of between 20 and 30 °C as being optimal for Anopheles survival and that temperatures below 16 °C and above 30 °C have a negative impact on mosquitoes survival. Chikodzi [51] noted that temperatures above 32 °C can cause high vector population turnover, with thermal death for mosquitoes expected to occur around 41–42 °C.

Vegetation index often acts as a surrogate for precipitation and surface temperatures and has been correlated to vector borne diseases [22]. In this study, vegetation cover followed a positive trend with the amount of precipitation received. In this study, EVI did not play any significant role in malaria transmission across the four zones.

This study established seasonality in malaria transmission over the study period (2004–2014) in the highland and mid-altitude zone. Malaria transmission in the lowland and riverine zone was shown to be perennial. Peak malaria cases followed increased rainfall with a time lag of 2 months across the study area and increased maximum temperatures with a time lag of 0 and 1 months in the riverine and highland zones respectively. The observed time lags between peak malaria cases and climatic variables are particularly important in forecasting malaria outbreak using local weather data. Therefore, monitoring rainfall and temperature trends and early recognition of anomalies in weather patterns can provide a fairly accurate forecast of transmission risk within Baringo County, and hence inform timely action including vector control measures.