Comparison of malaria incidence rates and socioeconomic-environmental factors between the states of Acre and Rondônia: a spatio-temporal modelling study

Human malaria emerged from the tropical forest of Africa, propagated globally and became a tropical and subtropical disease in the second half of last century [1‐3]. Six species of Plasmodium parasites can cause disease in humans: Plasmodium falciparum, Plasmodium vivax, Plasmodium malariae, Plasmodium ovale curtisi, Plasmodium ovale wallikeri and Plasmodium knowlesi [1, 4‐7]. Recently, Plasmodium simium emerged as another potential species to infect humans [8]. The malaria transmission cycle includes Plasmodium spp., anopheline and human components [9]. In 2017, 91 countries reported a total of 219 million cases of malaria, with 435,000 deaths [10]. Worldwide, P. falciparum malaria is more prevalent than P. vivax malaria. Plasmodium falciparum malaria accounted for 99.7% of the cases in areas across sub-Saharan Africa [10]. In the Americas, P. vivax malaria occurs more frequently than P. falciparum malaria [11‐13], with 723,000 (74%) infections reported in 2017 [10]. Whereas P. falciparum causes higher levels of morbidity and mortality than P. vivax [14, 15], the latter is gaining attention as a major hurdle in the era of malaria elimination [16, 17]. A reason for this can be that current malaria commodities, including the available anti-malarial drugs, are not very effective against P. vivax, leading to a high proportion of P. vivax asymptomatic reservoirs that can infect anopheline vectors [13, 16], further propagating the parasites in environments where competent mosquito vectors occur.

The global malaria incidence rate has declined from 76 to 59 cases per 1000 population at risk from 2010 to 2017 [10]. However, the rate of decrease has either slowed or reversed in some regions since 2015 [10]. In the Americas (in 2017), malaria incidence has been increasing since 2013, mainly because of the Bolivarian Republic of Venezuela, Brazil and Nicaragua [10, 13]. Between 2016 and 2017, malaria incidence increased approximately 100% in Nicaragua and Venezuela. In 2017, Venezuela accounted for 53% of reported cases, followed by Brazil (22%) [10]. Malaria distribution is spatially clustered, with hotspots of transmission in Choco (in Colombia), Loreto (in Peru) and Bolivar (in Venezuela) [10, 18]. In Brazil, approximately 45% of reported cases are from 15 municipalities in the states of Acre and Amazonas [18].

In Brazil, malaria decreased by 65% from 2010 (384,655) to 2016 (133,591). However, the disease increased by 63% between 2016 and 2017 (217,928) in comparison to 2015 [10]. Most malaria cases occur in the Amazon River Basin. In 2017, 193,000 cases occurred in the Amazonian Region (99.95%), which were mostly P. vivax malaria (174,000; ~ 90%). Consequently, the majority of the studies have focused on hotspots of malaria transmission in areas across the Brazilian Amazon (e.g., [19‐29]). Malaria transmission has been associated with several scenarios: (1) legal and illegal mining with high human exposure to mosquito bites, human movement and extensive environmental changes [16]; (2) expansion of agricultural frontiers, leading to deforestation, land-use changes and human encroachment in forested areas [30]; (3) discontinuity of malaria control programmes in poorly accessed remote areas [21]; and (4) ecological factors, which can drastically increase vector abundance, such as fish ponds in rural areas and towns [16, 25, 31]. These aforementioned transmission settings can represent transmission hotspots, and they were employed to construct a flexible model for predicting malaria emergence in similar scenarios [28, 32].

Frontier malaria is a concept offering an explanation for the trajectory of malaria incidence with deforestation and was applied for predicting the emergence of malaria in the Brazilian Amazon region [28]. This concept model predicts high malaria transmission risk in the first years of a human settlement in the Amazon forest. The main mechanisms are (1) a high number of immunologically naïve immigrants intermixed with asymptomatic human reservoirs, (2) a high contact rate between the main malarial vector and human hosts, and (3) a precarious socio-environmental matrix [28]. After 10 years of colonization and development in the settlement, the frontier malaria concept predicts a steep malaria decline rate. The mechanisms for the decline are related to overall improvements in the settlement with economic gains from agriculture, ranching and urban development [28].

A mathematical model was developed to address the possible dynamical trajectories of malaria with land-use change in frontier regions [33]. This work is a theoretical generalization of the frontier malaria concept, with a mathematical model coupling land-use change, malaria transmission and economic development. Most of the plausible parameter space led to numerical simulations with malaria population dynamics showing an initial increase in malaria incidence followed by a decrease in this incidence afterwards [33]. The initial state of high malaria risk in early stages of land-use change is driven by environmental conditions. Malaria risk decreases over time because these environmental conditions interact with the socioeconomic factors that tend to reduce risk on slower and longer timescales [33]. The tension between environmental and socioeconomic forces supports the pattern of the rise and fall of malaria population dynamics under land transformation (Fig. 1).

Fragmented landscapes with approximately 30–70% forest cover have the forest fringe effect, which maximizes the abundance of the main malarial vector (Anopheles darlingi) in the Amazon [24]. Malaria transmission can be sustained in any given landscape with the forest fringe effect [34]. Landscapes near natural conservation units (e.g., federal forests and indigenous reserves) are generally represented by settlements with intermediate forest cover and present a high risk of malaria incidence [35]. The variation in malaria incidence associated with changes in forest cover (100–0%) can be depicted by a convex curve [36].

Herein, a test of the unimodal (i.e., convex curve) relationship between malaria incidence and forest cover on a large scale (Amazonian states) is proposed. The importance of this work is the necessity of depicting the big picture and overall knowledge of malaria transmission for tailoring interventions. Determinants of the disease in two Amazonian states (Rondônia and Acre) that share a common historical root and started colonization at the same time in the 1900s were addressed. The dissimilarity between the two is that Rondônia represents a deforested Amazonian state, while Acre represents a forest-conserved Amazonian state. In addition, the state of Rondônia was the epicentre of the malaria burden in the 1980s to 1990s, but this state has recently seen a strong decrease in its incidence rate [23, 28]. In contrast, in the state of Acre, transmission is stable with some areas defined as hotspots of malaria in Brazil [25, 29]. The hypothesis is that this difference in the transmission level is related to the following: (1) most of the area of Rondônia previously covered by forest has been deforested [30]; and (2) the state of Acre, which has larger areas of preserved forest, is under anthropogenic changes in the natural environment, and forest fragmentation is increasing in some regions [30]. The specific aims are as follows: (1) to analyse the spatio-temporal distribution of the incidence rates and compare them between the states of Acre and Rondônia (in the western Brazilian Amazon); and (2) to address potential determinants of the disease.

The results of this study showed that the correlation between accumulated deforestation and malaria incidence can be discordant, showing either a positive or a negative statistical association. In Rondônia, the accumulated deforestation was three times higher than in Acre, and consequently, the trend in malaria incidence declined with increased deforestation. In contrast, the correlation was positive and statistically significant in Acre. Mechanistically, this pattern can be related to the frontier malaria concept [28] and the extension of this concept model by Baeza et al. [33], but it is also related to other works that state the importance of forest cover in malaria incidence in Amazon [24, 35, 36].

In the late 1970s, 2% of the state of Rondônia was deforested. Deforestation was intensified during the 1980s–1990s, affecting larger areas because of intensive migration. Malaria increased at very high rates during that time [16, 17]. However, starting in the late 1990s, Rondônia has gone through a turning point in its economic growth [40]. Mid-sized cities, which were merely a flow trail of natural resources to the urban centres of the capital (Porto Velho) or to southern Brazil in the 1980s, emerged as a central nerve in the production chain due to urban growth in the 2000s [40]. The five most important local hubs in Rondônia (Ji-Parana, Ariquemes, Vilhena, Cacoal, and Rolim de Moura) underwent population increases of 15–43% from 2000 to 2010 [40]. Capital investments that come to these urban centres in exchange for the region’s rich reserves of natural resources remain in the form of economic growth, rising socioeconomic indictors and public investments [40]. In addition, north-western Rondônia, which includes the capital (Porto Velho) and its adjacent municipalities (Fig. 5c), is considered a logging zone and a traditional wood transportation route in Brazil [30]. The fall of malaria observed in Rondônia can be related to both (1) socioeconomic factors that surpassed environmental forces on malaria transmission [28, 33] and (2) the loss of available habitats for the malarial vector due to deforestation [36].

Economic development in Acre is historically dependent on forest conservation for rubber exploitation and other extractivist activities, as well as fish farming [41]. Fish farming is not associated with deforestation [42] but can increase the risk of malaria [25, 31]. Cruzeiro do Sul was a former rubber town on the Jurua River and is now a local hub of economic growth and public investment in the westernmost area of Acre [40]. Additionally, Cruzeiro do Sul is also considered a local hub for the new frontier of logging zones [30]. The rise of malaria in the Jurua Valley Region may be related to environmental factors that tend to increase malaria risk in the early stages of colonization and to the lack of or still-incipient socioeconomic forces that tend to reduce malaria risk in the long term [33].

Parallel with the use of the frontier malaria concept [28] to predict malaria emergence in the Amazon is the debate regarding the association between deforestation in newly colonized sites and malaria emergence [43, 44]. The generality of the relationship between deforestation and malaria emergence was challenged [35] because the authors found higher malaria incidence in human settlements near priority areas for nature conservation. The controversy between the deforestation-malaria hypothesis [35] stimulated intensive debates [45, 46]. An alternative was proposed: deforestation may benefit or be harmful to the malarial vector population, depending on the pattern and proportion of forest cover [24].

The proposed unimodal relationship between forest cover and malaria emergence indicates that 30% to 70% of the remaining forest cover represents a landscape scenario that can encompass the ecological and environmental conditions that can favour peak transmission of malaria [24, 36, 47]. This risky scenario can occur either in newly colonized or old settlements [34]. For instance, the landscapes shown in Fig. 7 started colonization in the 1970s [34] and currently have high levels of transmission, with an estimated malaria incidence of 45–100 cases per day and a P. vivax reproduction number of 3.3–16.8 [48].

The satellite imagery composite shows Cruzeiro do Sul and Mancio Lima divided by a natural barrier: the hydrographic basin of the Moa River (Fig. 7). The configuration of the land use land cover shown in Fig. 7 can support an increase in malaria incidence [35] because of the availability of larval habitats for the malarial vector [24]. In addition, Lana et al. [49] identified improvements in socioeconomic factors in the landscape GUA (Fig. 7) at the same time as a high risk of malaria transmission due to (1) the abundance of malarial vectors and (2) the mobility of people in this urban centre of Mancio Lima. The pattern depicted in Fig. 7 seems supported by the frontier malaria concept [28] and Baeza et al. [33], thus representing the increasing phase of malaria population dynamics. Malaria decline may occur later in this real scenario (Fig. 7), when socioeconomic development can reduce transmission risk and accumulated deforestation can decrease larval habitat availability for the mosquito vectors.

The main malarial vector in the Amazon is Nyssorhynchus darlingi, formerly known as Anopheles darlingi [30, 48]. Foster et al. [50] built a globally based phylogeny of Anophelinae and concluded that Neotropical subgenera (including Nyssorhynchus) can be elevated to the genus level. In frontier malaria, Ny. darlingi is abundant, and its contact rate with humans is high [48]. On the one hand, other anopheline species known to be malarial vectors are not well adapted as Ny. darlingi in the anthropogenic matrix [51]. On the other hand, anopheline diversity continues to be underestimated in frontier malaria, with several species thought to be unknown [52]. Additionally, in specific scenarios, other species (e.g., Nyssorhynchus albitarsis sensu lato) can emerge as the primary vectors [53, 54].

A proposition for future research is herein made. The best study design for testing a temporal phenomenon as frontier malaria is a long-term prospective study. In the 1970s, a long-term prospective study was conceived for testing ecological theories (e.g., island biogeography) in the Amazon: the Forest Fragments Project (http://​pdbff.​inpa.​gov.​br/​), e.g., [55]. Considering malaria elimination as a global target [56], the timing might be optimal for a bold proposal, such as a long-term prospective study on land transformation and its impact on socioeconomic and environmental determinants of malaria transmission.

Spatial and temporal variations in malaria incidence were not assessed by a statistical autoregressive model that considers time and space [57].

Landscape modification caused by accumulated deforestation is an important driver of malaria population dynamics in Amazonia. In the initial phase of human settlement development, accumulated deforestation transforms a landscape with high forest cover into a landscape with intermediate levels of forest cover, increasing the odds of malaria emergence. In a later phase of development, when forest cover is reduced to low levels and its capacity to sustain malarial vectors’ larval habitats is decreased, the on-going accumulated deforestation only decreases the risk of malaria transmission.

The westernmost area of the state of Acre currently has stable malaria foci because it represents an initial phase of development, whereas the north-western area of the state of Rondônia, which had been considered the main hub for malaria in the 1980s and 1990s, is now seeing its malaria burden decline, which thus represents the later phase of development.