Background
Lymphatic filariasis (LF) is one of the neglected tropical diseases (NTDs), which presents chronic disabling and disfiguring pathologies with occasional painful attacks on affected persons [
35]. LF is a mosquito-borne infection caused by filarial nematodes:
Wuchereria bancrofti,
Brugia timori, and
B. malayi [
34]. These worms produce larvae i.e., microfilariae (
mf) - transmitted by mosquitoes in endemic areas; thus, reducing
mf levels is significant towards LF eradication [
18]. It is estimated that over 1.4 million individuals are at risk of infection in 83 endemic countries [
7]. Currently, the mainstay eradication strategies, which include Mass Drug Administration (MDA) and vector control, have significantly interrupted LF transmission in many previously endemic settings [
13]. While these achievements are commendable, there is the need to adopt novel approaches, especially in foci, where LF transmission is ongoing despite several years of implementing these control strategies.
In Ghana, studies on LF have shown differences in disease prevalence and multiplicity of symptoms in two geographically distinct regions, i.e., the northern and southern parts [
23]. The northern regions of the country exhibit higher prevalence compared to the southern regions, but the middle forest belt is relatively free from the infection ([
23]: [
14]). Elsewhere, a study has revealed some level of genetic variability in parasite strains in the two endemic areas [
9]. Furthermore, Pi-Bansa et al. [
32] identified a vector of very high vectorial capacity specific to the coastal areas i.e., southern Ghana. These variations in the two regions could be due to different climatological, land cover, and socioeconomic risk factors.
De Souza et al. [
10] reported that ecological and climatic variables such as elevations greater than 200 m, mean daily precipitation between 2.6–3.8 mm, and mean daily temperature range between 24.5–26.0 °C, influence the distribution of
Anopheles gambiae, one of the vectors for LF transmission in Ghana. At the global scale, another study used climatic and environmental variables in a boosted regression tree (BRT) model to map the transmission limits of LF [
6], confirming the influence of geo-environmental risk factors on vector population and vectorial capacity [
17,
11].
While these studies present very useful findings, their spatial scale of analysis obscures some micro-level risk factors [
28], which may be important for designing disease control strategies, especially in hotspots zones. According to Williams et al. [
40], the spatial scale for analysis should include the known environmental or geographic limits of the species under study for quality model predictions. In the West African sub-region, different geographical zones have been documented [
11]. The south is characterized by wetlands, while the north is characterized by drylands and sub-Sahelian climate [
32]. In Ghana, the northern and the southern regions, although both highly endemic for
W. bancrofti infections have distinct geographic characteristics (i.e., land cover and climate). This distinction is likely to influence vector proliferation and transmission potential differently.
Therefore, to facilitate LF elimination in these two highly endemic areas, a local understanding of the environmental, climatic, and socioeconomic factors that drive transmission is required to review existing control programmes. In line with this, the present study sought to map the environmental niches of LF and examine the behaviour of the diverse risk factors that drive transmission in Ghana’s northern and southern zones. Since the prevalence of LF is denoted by the presence or absence of microfilaria (mf) cases, data on mf survey from sentinel and spot-check sites across Ghana were stratified into the northern zone (NZ) and southern zone (SZ). These were used to run Species Distribution Models (SDMs), while climate, socioeconomic and land cover variables were used as covariates. The analysis was then performed over the entire country (countrywide (CW)) for comparison.
The remainder of the manuscripts is organized as follows. First, we evaluated and selected the range of risk variables influencing the occurrence of LF in the study zones. Second, six SDMs were selected for the mapping of environments suitable for LF transmission. Third, we described and compared the response curves of observed covariates on the probability of LF occurrence in the NZ and SZ study zones.
Discussion
Despite several years of mass drug administration and vector control measures against human lymphatic filariasis in Ghana, some areas continue to serve as hotspots for its transmission. LF occurrence in Ghana appears to vary from one location to another. In Ghana, two major endemic zones are known for LF, i.e., northern and southern zones. The mid-section has only a few cases of
mf believed to have been imported from the north [
23]. Understanding the differences in risk factors, i.e. environmental, climatic and socioeconomic covariates that drive
mf transmission in the two study zones, is key to providing appropriate elimination strategies.
In this study, the evaluation of model performances revealed that RF and GBM algorithms performed better for all three zonal
mf datasets. The two models in the current study showed that the AUC of success rate ranged from 0.95 to 0.98 and 0.91 to 0.95 for RF and GBM, respectively. This may be attributed to the fact that RF and GBM are better able to handle large covariates [
37], as provided in this study.
The probability of
mf occurrence is influenced by different combinations of variables in northern and southern Ghana. In the north, the occurrence of
mf was influenced by low values of annual precipitation but decreased with high values above 1000 mm. The precipitation variable behaved differently in southern Ghana, with precipitation of the driest quarter sustaining LF transmission. In Ghana, heavy rainfall from April to June usually results in flooding in the northern region [
27]. In the south, high rainfall patterns and low elevation, particularly along the coast, may result in surface water run-offs. These occurrences may sweep away breeding habitats reducing the survival of LF vector and subsequent transmission. However, rain availability especially in the coastal areas during the driest period of the year from late December to March, can create pockets of stagnant water bodies to sustain mosquito breeding, therefore increasing LF transmission. This implies that, whereas rainfall is needed for vector breeding, excessive rainfall could potentially result in flooding and sweeping away breeding sites [
12]. Findings from this study are consistent with previous studies by Abiodun et al. [
1], Cano et al. [
6] and Eneanya et al. [
17].
Similarly, the occurrence of
mf declined with high land surface temperature during the day. This is consistent with adult mosquito survival and larval development, which suggests that both adult and larvae are unable to survive at high temperatures [
25]. The response curve shows that minimum day land surface temperature values between 23 °C to 24.5 °C (GBM and RF, respectively) may increase the mortality rate of either larvae or adult mosquitoes in the north. Comparatively, vector survival is supported in the south beyond these temperature ranges until temperatures between 27 °C to 29 °C. This may result from thick vegetation cover or tree canopy in southern Ghana, which may create suitable conditions likely to sustain vector survival even at high temperatures. It was also observed that the probability of
mf occurrence decreased with increasing terrain slope in both areas. This finding confirms with studies by Eneanya et al. [
17]. What accounts for such observation is that steeper surfaces could lead to faster surface water run-off, thus decreasing water collection in pockets and eventually reducing breeding sites for vectors associated with LF transmission. In addition to poverty, areas of poor housing support transmission in the north.
Distance to stable night light was an important covariate for both northern and southern Ghana as well as at the countrywide scale. The response curve in the north shows that suitable areas for mf occurrence were generally rural and poor communities. In the south, some communities located in peri-urban to urban communities had a high probability of mf occurrence. This is true because some coastal communities located in peri-urban areas in the Western region have high mf prevalence.
Although some areas of the two the study zones (north and south) were suitable for mf occurrence, there exists slight differences in the suite of risk factors. This implies that any effort or strategy intended to eliminate the disease should consider unique conditions prevailing at a relatively fine spatial scale. The major limitation of the study was that the models did not consider important demographic risk factors at the community or individual level that are likely to improve the predictions. Finally, a larger sample size could lead to more precise predictions.
Conclusion
This study has demonstrated that different variable combinations influence the occurrence of lymphatic filariasis in northern and southern Ghana. For both zones, the disease is highly prevalent in poor rural communities in low lying areas. In northern Ghana, areas suitable for transmission are relatively warm, low lying rural communities with poor housing, especially those characterized by mud houses. Besides, mean annual precipitation between 900 mm to 1000 mm provides a conducive environment for LF transmission. Similarly, rural, poor, low-lying, and most coastal communities in the south present a suitable environment for LF transmission. However, some peri-urban areas along the coast were also observed to be suitable areas. Generally, the infection is efficiently transmitted in warm lowland communities within 2 km of inland water bodies such as mangroves, lagoons, and rivers in the south. Moreover, rainfall within the relatively warm part of the year was identified as an important risk factor as it may contribute to the formation of stagnant water bodies suitable for mosquito breeding. Interventions such as improvements in housing and sanitary conditions may reduce LF transmission in endemic areas. The findings of the present study can be utilized by policymakers in advancing evidence-based strategies to eliminate LF.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.