Background
Globally, human populations and settlements are undergoing rapid expansion, concurrent with speedy urbanization [
1]. In the 1950s, less than one-third of global population lived in urban areas, but this had increased to 54% by 2014 and will likely exceed two-thirds by 2050 [
1]. Most of these changes are expected to occur in developing countries in Africa and Asia [
1,
2]. In Tanzania, about 33% of people currently live in urban areas, compared to just 20% in 2000 [
3,
4].
Malaria burden has significantly declined worldwide over the past two decades, largely because of deliberate control efforts like scale-up of insecticide-treated nets (ITNs) and improved treatments [
5] as well as due to socio-economic developments [
6,
7]. Urbanization is one of the major aspects of development that significantly impacts malaria transmission [
7‐
9]. Processes of urbanization reduce malaria transmission primarily because urban environments lack suitable
Anopheles breeding habitats [
8‐
10]. Additionally, urban settings generally have better access of health services and improved housing as well as greater education and awareness than rural areas [
7,
9]. In a 2013 analysis, Tatem et al. concluded that ongoing urbanization in endemic regions would cause further declines in malaria transmission, and that these declines would be synergized by the deliberate scale-up of core malaria control tools [
7].
Ifakara town, in the Kilombero river valley in the mostly agricultural Morogoro region, is among the fast growing towns in Tanzania. The town is surrounded by rain forests, wetlands and vast savannah lands, making it an attractive setting for both farmers and pastoralists. It is also a key trading hub in the region, mostly for farm and animal products. Towns such as Ifakara that appear in the middle of low-lying tropical floodplains present unique scenarios for assessing spatial and temporal variations of malaria transmission intensities [i.e. entomological inoculation rates (EIR), measured as number of infectious bites/person/year (ib/p/yr)]. Historical measures of EIR in Ifakara date back to 1965, when the town still had just one-fifth of its current population, and when Freyvogel and Kihaule conducted a small survey of
Anopheles mosquitoes in the area [
11]. Though parasite assessment techniques at the time were still fairly insensitive compared to options available today, Freyvogel and Kihaule reported monthly “apparent inoculation rates” for both
Anopheles gambiae and
Anopheles funestus, the two predominant malaria vectors at the time [
11]. The “apparent inoculation rate” represented how frequently a single person was likely to get infected, and was estimated based on factors such as proportion of mosquitoes with sporozoites in their salivary glands upon dissection, gonotrophic cycles of the mosquitoes, and estimated daily biting rates (computed as ratio of total indoor mosquito catches and number of sleepers in collection room). In later years, there were multiple EIR assessments in the surrounding villages, including which reported 329 ib/p/yr from early 1990s in Namawala, a rural community 30 km away from the Ifakara town [
12]. Because of the differences in assessment methods, it is difficult to directly compare these historical estimates against measures obtained after 1990s onwards.
In 2003, Drakeley et al. published findings of another entomological survey that assessed malaria transmission intensities in Ifakara, which at that time was still a peri-urban area surrounded by very high transmission villages in the Kilombero valley [
13]. In the years preceding this report, the team had conducted fortnightly mosquito collections using CDC-light traps inside 32 randomly selected houses in Ifakara area. They used two different methods to estimate the P
fEIR. First was the standard method, which involved calculations of the product of sporozoites rate (SR) and human-biting rate (BR), and the second method involved calculation of the ratio of sporozoites-positive mosquitoes to the number of nightly catches. Drakeley et al. estimated an EIR of 31 ib/p/yr using the standard method and 29 ib/p/yr using the alternative method [
13].
This study aimed to assess malaria transmission intensities in Ifakara town and its adjacent wards, more than a decade after the assessment by Drakeley and colleagues. This new study employs similar mosquito sampling techniques, but with addition of indoor/outdoor comparisons of transmission risk, and an inclusion of an odour-bated trap to sample outdoor host-seeking vectors. Similar methods were also used for calculating and reporting P
fEIR as used by Drakeley et al. [
13]. The primary objective was to assess the extent to which malaria transmission in the study area had changed over the past 15 years, considering the rapid growth in population [
2,
14], urbanization trends [
15] and the sustained use of effective malaria treatment as well as prevention measures such as ITNs [
16‐
19].
Discussion
Africa is the world’s most rapidly urbanizing continent [
1,
2]. Currently about 40% of the African population resides in urban settings, but that proportion is expected to reach 56% by 2050 [
1,
2]. Malaria, like most other infectious diseases, is negatively impacted by urbanization [
7‐
10], since urban settings are characterized by better access to health care services, better housing and reduced availability of larval habitats [
8‐
10]. In Tanzania, the urban population is increasing at 5.4% per annum. Tanzanian urban population was estimated at 5.7% in 1967 [
15], but it had increased to 22.6% in 2002 [
4], 29.1% in 2012 [
51] and is estimated to be 33% in 2018 [
3].
This current study focused on one of the fast-growing towns in Tanzania (i.e. Ifakara), and where demographic, socio-economic and epidemiological transitions are strongly evident. Ifakara town, in the Kilombero river valley is surrounded by rain forests, wetlands, savannah lands making it an attractive setting for both farmers and pastoralists. More importantly, as a town in the middle of an expansive low-lying flood-plain, only eight degrees south of the Equator, the area presents a unique epidemiological profile with very high malaria transmission in surrounding areas.
Though there have been several entomological surveys in the villages across the Kilombero valley over the past three decades [
12,
52‐
54], there have been limited investigations within the Ifakara town centre itself, or in its closely adjacent wards. The most recent examination of malaria transmission in the town area was completed more than 15 years ago by Drakeley and colleagues [
13]. Over the years, since the study by Drakeley et al. in early 2000s, the area has experienced significant increases in population concurrent with urbanization, due to in-migration by small business owners, farmers and pastoralists (Mayor of the Ifakara Town Council—Personal Communication). This current study presents the first reassessment of transmission in this community in nearly two decades.
A year-long entomological surveillance of malaria vectors was conducted in the five administrative wards making up the Ifakara town, an area with a total population estimated at 70,000 people. The magnitude and biting patterns of main malaria vector species, the levels of ongoing transmission, as well as the key factors driving the transmission in the town were identified. A general observation was that densities of malaria vectors within the area was found to be much lower compared the vector densities usually observed in nearby rural villages on both sides of the Kilombero river [
21,
22]. Overall P
fEIR was estimated to be 0.102 ib/p/yr, which was entirely driven by
An. funestus and only detectable by HLC. This was more than 99% decrease from the P
fEIR reported by Drakeley et al. in 2003 from the same area. The P
fEIR estimates are also much lower than those recently observed by Kaindoa et al. in the neighbouring rural villages of Ulanga district (P
fEIR = 18.45), about 20 km south of Ifakara town [
21]. Some of the factors that could be attributed to this dramatic decline could be urbanization [
7,
8,
10], the universal bed net coverage campaigns that started in 2004 [
17,
55], improved housing [
56] as well as improved diagnosis and treatment [
16]. More than 90% of the houses in this study area now have brick walls and metal roofs, more than 50% had screened windows, and all (100%) had at least one bed net (Finda et al., unpublished). Unfortunately, this study was limited by the lack of aerial maps for the study area from 15 years ago, which would have enabled more detailed comparison of housing developments in the area. Nonetheless, these improvements, coupled with the ongoing large-scale use of long-lasting insecticide-treated nets (LLINs) and other malaria control interventions, as well as the changes in potential mosquito breeding habitats, it can be expected that malaria transmission should plummet even further.
The ease of detecting
Plasmodium circumsporozoite proteins in mosquitoes using simple ELISA assays and the simplicity of calculating entomological inoculation rates (P
fEIR) has made this approach fairly popular for assessing malaria endemicity and transmission intensity [
44,
57]. It requires only simple measures of mosquito biting rates and sporozoites positivity rates, and is therefore a more direct method of measuring transmission intensity compared to other measures such as parasite incidence or prevalence and has been used widely for assessing impact of vector control programs in Africa [
57,
58]. This method estimates the number of bites by infectious mosquitoes per person per unit time. It is the product of the “human biting rate”—the number of bites per person per day by vector mosquitoes—and the fraction of vector mosquitoes that are infectious [
44]. However, despite its widespread use, there have also been concerns that P
fEIR estimates may be inconsistent, and that there can be large variabilities in P
fEIR estimates from the same area [
57,
58]. Such variations are heightened by the differences in methods of measurement [
57,
58] and the lack of clear relationship between P
fEIR estimates and malaria parasite prevalence of incidence rates in localities with low transmission intensities [
59,
60]. Besides, at very low P
fEIR ranges, available methods are imprecise and there are very slim chances of getting mosquitoes at a precise time when innoculation takes place [
60]. Therefore, the very low P
fEIR estimates observed here do not necessarily mean absense of local transmission, but rather the lack of effective measurement methods.
In this study, only the human landing catches method was able to detect sporozoites positive mosquitoes. On the contrary, neither the CDC light traps nor the Suna
® traps caught any infected mosquito. This suggests the need for much more sensitive tools and approaches for measuring human exposure to malaria parasites in conditions such as this, where parasite densities in mosquitoes have become too low. Nevertheless, it has also been argued that if similar methods are used to consistently estimate transmission over time, such as this study has done, and if adequate stratification is performed, then interventions that drive malaria transmission to P
fEIR < 1, could be effective in achieving local elimination [
61]. In this survey, the overall P
fEIR of 0.102 ib/p/yr was indeed far below the threshold of 1 ib/p/yr, beyond which sustained local efforts could lead to complete disruption of local malaria transmission.
Going forward, improved measurements are therefore a critical component of the malaria elimination agenda [
62]. It will thus be essential to deploy improved measurement methods that can assess both the burden and the transmission of malaria pathogens in such situations so as to support further efforts for elimination. It would also be important to couple such an entomological survey with a malaria parasite prevalence study, using methods capable of detecting low-level parasitaemia so as to more comprehensively understand the residual malaria epidemiology in the area.
Within the Ifakara town, the highest density of malaria vectors was observed in Katindiuka and Mlabani (Fig.
6) wards, which were also the most rural of the five wards. Households in Katindiuka and Mlabani wards were surrounded largely by rice paddies and water ponds (Fig.
5), and nearly all the houses in these wards did not have electricity. Brick-making activities were fairly common in these areas in the dry season which resulted in a lot of pits with standing water in the rainy season, which may have provided adequate breeding habitats for mosquitoes. Although nearly all of the households surveyed were made with bricks and metal roof, majority of the households in Mlabani and Katindiuka lacked electricity, hence they were in the dark through most of the night, which may also have provided a suitable environment for host seeking mosquitoes [
63].
In the study done by Drakeley et al, 91.5% of all
An. gambiae s.l. were
An. arabiensis while only 8.5% were
An. gambiae s.s., none of which were sporozoites-positive. However, several studies done since the Drakeley et al. study have documented absence of
An. gambiae s.s. from the Kilombero river valley, mainly attributable to the use of LLINs [
22,
52,
54,
64]. While the most abundant malaria vector species was found to be
An. arabiensis, it was
An. funestus that was found to drive all of the transmission in the town. Only 400
An. funestus mosquitoes were collected compared to 7795
An. arabiensis, yet the only
Plasmodium infected mosquito was an
An. funestus. Though only a single infected
An. funestus might be considered too few to conclude on the dominance of the species, evidence from neighboring villages suggest this is most likely the case. Similar patterns of dominance were indeed observed by Kaindoa et al. [
21], who showed
An. funestus to carry more than 80% of all the malaria transmission in neighboring villages in the Kilombero Valley. All blood-fed mosquitoes were found to contain human blood, and this can be explained by the absence of big livestock such as cows and goats in the urban settings, hence humans were the most available source of blood meals for the vectors. This could suggest that household-based interventions could still be effective in targeting malaria vectors in the Ifakara town. However, the actual blood-feeding proportions were relatively higher for
An. funestus (i.e. 9% (36 out of 400) were blood-fed) compared to
An. arabiensis for which 0.05% (38 out of 7795) were blood fed. This further demonstrates the importance of
An. funestus.
Relatively high parity rates were observed; over three quarters of all the mosquitoes dissected were found to be parous, a fact that also emphasizes the need for interventions that prevent man–vector contact to limit local transmission. On the basis of parity rates and human blood index, both An. funestus and An. arabiensis can be considered important vectors in the area. However, based on actual contribution to PfEIR estimates, An. funestus is likely mediating most of the on-going residual transmission in both the wider valley and in Ifakara itself. In this study, An. funestus densities were also observed to be higher in the dry season, which is the time of the year that people are most relaxed with regards to protection against mosquitoes.
There was evidence of phenotypic resistance in
An. arabiensis to permethrin, deltamethrin, lambda cyhalothrin, pirimiphos methyl and DDT. However, the same mosquito populations from the study villages were found 100% susceptible to bendiocarb and malathion. The most surprising aspect of this specific finding was that the organophosphate, pirimiphos methyl has not previously been used in the area for vector control. It is however widely used in agriculture (Matowo et al., pers.comm.), which could be the source of this resistance pressure exhibited here. Nonetheless, the resistance profile of the mosquitoes collected here is worrying as it signals that there is a very narrow set of insecticide options now available for malaria vector control. One minor limitation with this aspect of the study was that the resistance assays were conducted only in two of the five wards, primarily due to availability of
Anopheles larvae at the time of the tests. These findings can however still be considered fairly representative of the whole study area, and concur with other studies that have been done in surrounding villages in the valley [
21,
50].
Finally, the major decreases of malaria transmission seen are indeed impressive but should not be taken as a sign of impending local elimination. Instead, it should be interpreted as evidence that local elimination is possible given multisectoral approaches that combine house improvement to other technologies such as LLINs and effective case management, with proper diagnostics and medicines. The authors propose a parasitological survey to enable assessment of actual malaria cases prevalence and incidence rates, and also to examine in greater detail the actual rates of importation. Households should be encouraged to continue use of their long-lasting insecticide treated nets and to visit health facilities urgently whenever they experience any fevers. Given the significance of house improvement, these efforts too, should be encouraged to further reduce exposure to malaria and other mosquito-borne infections.
Authors’ contributions
MFF was involved in study design, data collection, entry and analysis, interpretation of the results and drafting of the manuscript. FOO was involved in study design supervision and manuscript revision. AJL was involved in study design, data collection and preparing maps for the study sites. HSN and JKS were involved in the statistical analyses for the study data. NSM was involved in conducting the resistance tests, and EWK generated land use maps Ifakara town between 2000 and 2016. All authors read and approved the final manuscript.