During the 5-year study period, there were significant shifts in the epidemiology of malaria in Roraima state. From 2016 to 2018, there was an increase in imported malaria cases from Venezuela and Guyana. The number of imported cases peaked in 2018, with 4,478 cases reported as originating in Venezuela and 610 cases from Guyana. However, this trend reversed in 2019 and even more so in 2020, when the border to Venezuela closed to mitigate spread of COVID-19. Similar to another study in Roraima from 2016–2018, Plasmodium falciparum cases had double the odds of being imported (7). These findings demonstrate that humanitarian events within and outside Brazil, such as the crisis in Venezuela and the global COVID-19 pandemic can shift migration patterns and livelihoods activities quickly. Surveillance programs can capture the corresponding changes in malaria dynamics that occur as a result of shifts in migration and livelihoods, but resources and systems are needed so that control efforts can respond nimbly to rapid epidemiological shifts and changes in transmission dynamics. Since the border reopened with Venezuela in July 2021, there has been an influx of Venezuelan migrants again through Pacaraima, and given the available evidence, there will be an increase in malaria, especially Plasmodium falciparum (28). Hence, innovative active surveillance approaches are needed to identify asymptomatic infections and artemisinin-resistant infections (29, 30)
Second, there were Plasmodium species shifts. The share of Plasmodium falciparum cases increased over time, peaking in 2020 at 18.50%, and was strongly associated with imported cases from originating from Venezuela and Guyana (27.08% of imported cases were Plasmodium falciparum, compared to 9.00% of autochthonous cases). However, there were increases in the proportions of Plasmodium falciparum among indigenous people, especially in 2020.
Thus, during the initial years of the analysis (2016–2017), the majority of Plasmodium falciparum malaria cases came from outside Brazil, primarily from Venezuela, where access to diagnostics and treatment was low. However, we hypothesise that at that time, the diagnostic network and health system in Roraima could support the increased burden from Venezuela, avoiding epidemics (at the time, there was increased national financial and human resources support to Roraima and increased state-level support to Pacaraima and other municipalities). But, in 2019 and 2020, the increase in Plasmodium falciparum cases occurred in indigenous areas, where gold mining was taking place. In the region, both indigenous reserves and illegal gold mining areas are characterised by lack of access to malaria diagnostics and treatment, allowing Plasmodium falciparum malaria to spread more widely. Further, there evidence of an association between Plasmodium falciparum malaria and gold mining, observed in neighbouring French Guiana (31). This shift occurred despite Brazil’s efforts to reduce Plasmodium falciparum malaria.
Contrary to our results, which show little difference in time from symptoms to treatment between Plasmodium falciparum and Plasmodium vivax, Plasmodium falciparum is usually related to a delay in diagnosis. The time required for gametocyte maturation differs between the Plasmodium species, and a Plasmodium falciparum gametocyte takes 8 to 10 days for maturation, whilst Plasmodium vivax gametocytes require 48 hours for development and disappears from circulation within 3 days of sexual maturation (32). Therefore, we would have expected to more delays in treatment among Plasmodium falciparum cases.
Third, our analysis showed a strong link between mining and imported malaria, with miners nearly nine times more likely to be imported cases. This is aligned with other evidence from Roraima from 2016 to 2018 showing that mining was a strong correlate with cross-border malaria (7). Over the study five-year period, the primary occupation of imported cases was in mining, but the proportion of autochthonous cases reporting their primary occupation as mining increased nine-fold from 2018 to 2020. This is compatible with findings in other mining regions of Brazil, such as Tapajós, where transmission in mining areas increased 17.8% in the first part of 2020 (33). By species, whilst Plasmodium falciparum cases reported working predominantly in mining (47.3% overall), only 15.99% of Plasmodium vivax cases said they were miners. In the Guyana shield, which includes Roraima, mining is a known risk factor for malaria, and especially for falciparum malaria (31). In Guyana, 94% of reported malaria cases occurred in major gold mining regions, while in Venezuela, gold mining accounted for ~ 60% of countrywide cases, demonstrating the potential for small, isolated, malaria-dense populations where there are minimal resources to stop outbreaks, reversing progress toward elimination (9, 31). Evidence from Venezuela has shown that these illegal mining areas not only sustain transmission, but also can restore it after interventions have reduced malaria locally or even achieved local elimination in other areas, putting these areas at the crux of elimination efforts (9, 34).
Most informal and illegal mining operations in Roraima are located on indigenous Yanomami reserves. Between 2017 and 2019, gold mining destroyed 25,315 acres of land across three indigenous territories—Munduruku, Yanomami, and Kayapó—located in Brazil’s Legal Amazon (BLA). As of 2020, there were over 20,000 miners reported on indigenous Yanomami reserves, mostly in Roraima state (35). As illegal mining on indigenous lands surged over time, there were increases in number of malaria cases among indigenous people, and in the proportion of cases reporting timber/fishing as their primary occupation, from 6.88–21.44% between 2016 and 2020, the primary occupation of most indigenous people in the area.
Our findings show that non-indigenous people, which include most miners, are more likely to report getting malaria outside Brazil, whilst indigenous people more often acquire malaria locally, with indigenous people 62% less likely to be imported cases. Further, between 2018 and 2020, malaria cases (especially of Plasmodium falciparum) increased notably in young indigenous children; over the five-year period, the median age of cases among indigenous people was 12.69 years compared to 30.57 years in non-indigenous cases. This is consistent with findings a study focused on factors associated with malaria in indigenous populations in Amazonas state (19). Another study in Amapá on the border with French Guiana showed that indigenous children had a disproportionate burden of malaria when compared to non-indigenous children and among indigenous populations in Venezuela (20, 36). The increase in malaria cases among indigenous children, especially of Plasmodium falciparum, begs further investigation into effective approaches to optimize prevention for this population. Malaria could be transmitted to children at night while they are sleeping, so prevention measures can focus on use of long-lasting insecticide treated nets (LLINs), if culturally acceptable, and house spraying when indigenous people live in villages and wood houses, whilst ensuring mining sites that contain pools of stagnant water are not in close proximity to indigenous settlements. But, as observed in SIVEP data, children’s occupation reflects that malaria could be transmitted to them while accompanying parents in activities such fishing and timber. More research is needed on delivery and sustainability of prevention measures in indigenous communities, determining when and where malaria is likely to be transmitted to indigenous people, especially children.
Overall, among all cases, it is concerning that there were almost six times more cases in 2020 compared to 2019 in children under five years of age. This could lead to an increase number of deaths in this age groups. Although malaria deaths in Brazil have historically been more frequent in adults. During the period from 2015 to 2019, there were 204 recorded deaths from malaria in all of Brazil, with 113 in the Amazon region.
Indigenous individuals are getting more timely treatment for both Plasmodium falciparum and Plasmodium vivax malaria, but timeliness to treatment lags for non-indigenous populations, many of whom are miners. However, gold miners and indigenous populations often live as neighbours in remote areas with less access to prompt diagnosis and treatment (37). Thus, this finding requires further investigation as it runs counter to what is expected, given that indigenous people usually live further from health facilities and therefore must wait longer periods to commence treatment. The indigenous health facilities sometimes use different malaria surveillance forms, so this finding may in part be explained by errors in classification on the forms.
In some municipalities, malaria cases are being diagnosed and notified in areas far from the municipality of infection, mainly in Boa Vista, the largest city in Roraima. Malaria cases among non-indigenous people (who are predominantly miners) in Alto Alegre municipality are often notified in Boa Vista, with the reported as place of infection as in Alto Alegre or outside Brazil, mostly in Venezuela. In contrast, malaria cases among indigenous populations in Alto Alegre are being diagnosed and reported in Alto Alegre. One hypothesis for this finding is that because gold mining in Alto Alegre is illegal and in Yanomami territory, miners may incorrectly report infections as coming from Venezuela. They also may not seek care in Alto Alegre as most clinics there are Indigenous Health Posts. Due to the stigma associated with mining and tensions between miners and indigenous communities, miners travel back to Boa Vista or larger towns to seek treatment, do not disclose their true occupation, or do not report cases through the national health system at all, opting to purchase treatments through the informal markets.
Areas where informal mining and indigenous populations coexist are often remote, posing challenges to control and elimination efforts that require innovative solutions (38). Both are also socioeconomic risk factors. Therefore, innovative surveillance and treatment approaches are needed to promptly identify and treat cases in these populations (29). Prevention measures such as LLINs and diagnosis and treatment kits can be made available closer to mining sites. Additionally, careful attention must be given to the design and material of LLINs and other vector control interventions, which play a significant role in user preferences and appear to drive net use (39). In French Guiana, Malakit targets gold miners working illegally with free malaria self-diagnosis and self-treatment kits for Plasmodium falciparum (40). Further, in light of the shift epidemiology of malaria species composition, both prevention and treatment strategies need to use nuanced approaches to combat Plasmodium falciparum and Plasmodium vivax differently.
Our findings show that malaria remains a serious threat in Roraima state, especially among high-risk populations, such as miners, migrants, and indigenous people. The transition from malaria control to elimination requires understanding of risk factors among high-risk populations in order to tailor interventions appropriately(41). In 2016, health ministers from across the Region adopted developed and the Plan of Action for Malaria Elimination 2016–2020 (42). One of its strategic goals was “to further improve surveillance systems with early detection of cases and outbreaks and advocate collection of malaria data (by case, including information on age, sex, ethnicity, and other variables that facilitate appropriate analysis of disparities and inequalities between populations).” Malaria transmission in Roraima is dynamic and its changing epidemiological profile should be monitored closely through robust surveillance systems and strategies tailored to specific population risk groups, considering the local context and changes in each group’s occupational tendencies and residence status (43). Furthermore, a strong early detection and rapid response system will facilitate the detection of increased malaria cases and epidemic outbreaks at the community and sub-community levels, border regions and mining sites.
This study had several limitations. Education could not be included in the final multivariable model investigating risk factors because children’s educational status was missing or incomplete, since they were still in the process of completing their education. Also, following SIVEP guidance, the occupation of children should be classified according to the category closely related to possible exposure to malaria in the last 15 days. The child’s occupation is therefore usually classified as that of their primary caregiver; however, many times the “other” category is selected, overrepresenting this category in the analysis. This is surveillance data and only data on positive cases were available. Information on negative cases would have lent to more robust analysis of risk factors.