The SYF outbreak that occurred in Brazil from 2016 onwards reached states in the southeast, northeast, and south. It became the most important outbreak in recent decades due to the large number of cases and deaths in humans and NHPs, reaching areas with high population densities and low vaccination coverage5,12,19,20. In SP, this process started in a region with high vaccination coverage and progressed to areas with low coverage and/or without vaccination recommendations, with high population densities11,12. The first region with these characteristics to be affected by the outbreak was Campinas12 and the period resulting from this advance was considered the first wave. In the second half of 2017, SYF advanced to the east, south, southeast, and northeast regions of SP, characterizing the second wave.
Hill et al.4 analyzed only the SYF epizootics that occurred in SP between 2016 and 2018 and divided this process into three phases. Although this makes more sense for the epizootic disease analysis, considering the outbreak process in two waves made it possible to adequately capture the spatiotemporal process of SYF diffusion and to characterize it by contagion. Moreover, this characterization allowed a discussion on the relevance of the vaccine strategy adopted by SES-SP, especially during the second wave. The Ministry of Health, in view of the YF outbreak in areas without recommendation and/or low vaccination coverage, advocated the expansion of vaccination based on calculations of affected area/expanded area, with the municipality as smallest unit17. Until mid-2017, this strategy was an adequate response. However, it became demonstrably insufficient when the virus reached populous areas with low coverage and/or without a vaccination recommendation, both in SP12 and in other southeastern Brazilian states21.
In SP, the advancement of YF within a period of less than six months from areas with high vaccination coverage to populous areas with low coverage or without vaccination recommendations triggered a public health crisis that generated panic in the population. At that time, there were insufficient time and vaccine doses to serve the entire population at risk21. Hence, SES-SP identified the priority vaccination areas based on the analysis of virus circulation in NHPs at forest fragments, which comprise functional ecological corridors for viral dispersion, and were therefore demarcated and used to select areas where populations were most exposed to YF risk. Instead of considering the entire municipal territory with the same risk level, the new strategy sought to identify priority intra-municipal areas for vaccination, considering the speed of dispersion16. Moreover, this strategy, plus the use of single doses1 and fractional doses22, allowed reasonable equation, in space–time, of the demand for vaccines in a large population contingent with an insufficient supply of the immunobiological product. It should be noted that this strategy had already been recommended by the WHO in outbreaks within the African continent in 2015/201621,23.
Our results demonstrated that the YF virus dispersion in SP was caused by the outbreak process of territorial spread by contagion; therefore, the mosquito vectors and NHPs could act as a route of viral amplification and further transmission on epizootic waves. In this situation, there was a cadence in the spatiotemporal pattern of viral dispersion through contiguous and nearby areas. This also shows the appropriateness of the vaccination strategy adopted by SES-SP, which allowed the population to receive the vaccine at least two months before the establishment of on-site transmission risk15. A spatiotemporal process of sequential spreading was observed, wherein municipalities located at shorter distances from the areas with YF virus transmission were more likely to be affected first7,8. Although our analyses are limited by the fact that we used the municipality as a spatial study unit, this spreading pattern can also be observed at the intra-municipal level.
The speed of the SYF dispersion that we obtained for the first and second waves, disregarding the RDSs of Registro and Itapeva, were similar to those reported by Hill et al.4. Through phytogeographic analysis of YF genomes in NHPs, they reported spreading speeds of around 1.0 km per day. Notably, they took into account epizootics that occurred in SP up to February 2018, and the occurrences of human cases and epizootics in the RDS of Registro and Itapeva were recorded from February 2018 onwards. The differences in the speed of viral dispersion between these two RDSs and the rest of SP during the second wave may be related to the greater vegetation cover and forest preservation of these areas, which can cause a dilution effect, as already demonstrated for other vectorborne diseases24. The spread of the SYF outbreak by contagion in most of SP, with a speed of approximately 1.0 km per day, opened a spatiotemporal window of opportunity for the vaccine to arrive before the virus15, avoiding or minimizing the occurrence of human cases and deaths. Since 2019, this outbreak has advanced to the southern region of the country and has reached the states of Paraná and Santa Catarina, and is still ongoing. This generated an emergency situation similar to that in SP, and the same vaccination strategy adopted in SP has been applied in this region16,18,20.
Our findings (Fig. 2) showed important differences in human cases and epizootics that occurred in SP during the second wave, reflecting the degree to which municipalities have adopted the vaccination strategy advocated by the SES-SP. If this is true, what happened in Mairiporã and Atibaia (37% of the total human cases between 2016 and 2019) could also have occurred in municipalities such as Jundiaí, Bragança Paulista, Itapecerica da Serra, Pinhalzinho, Louveira, and São Paulo. If the YF vaccination strategy had not been adapted for the emergency situation during the second wave, we could have had a worse outcome than that observed. These possible scenarios could be the subject of future studies.
At the end of 2016 and 2017, the detection of YF epizootics in NHPs anticipated the notification of human cases by two and three months, respectively. This result is expected, since the seasonality observed among NHPs in Brazil differs from that observed in human cases. In primates, circulation is generally detected in September, whereas in humans, circulation is usually observed in December, with the detection of cases among non-immunized people and those exposed to the virus17,25. This also highlighted the importance of the NHP epizootic surveillance strategy, aimed at the early detection of the circulation of the YF virus while still in the enzootic cycle4,26. However, despite the heavy investment of SES-SP in making municipalities sensitive to detection of NHP mortality, the detection of epizootic diseases is still marked by a strong reporting bias26.
One of the causes for the anticipation of epizootics in relation to human cases can be explained by seasonality of the precipitation in SP during the year. From the middle to the end of autumn, depending on the year, the total rain decreased and registered monthly values that were increasingly smaller. This becomes reversed in the beginning of spring, when an increase in the precipitated volume begins to be registered throughout SP27,28. To increase YF transmission, mosquito vectors need to be found in large quantities, and one of the determining factors for the proliferation of mosquitoes is the level of precipitation, as this allows the accumulation of water in reservoirs and the hollows of trees29.
Another important factor is the rise in temperatures from the end of winter and the beginning of spring. Temperature increases accelerate the time for larval development of the vectors29 and reduces the extrinsic incubation period of the virus30. Precipitation and temperature directly influence the mosquito’s life cycle and viral replication31, hence their increase is an optimal scenario for the proliferation of YF vectors and for the increased occurrence of epizootic diseases in SP29. By contrast, the increase in the probability of detecting human SYF cases in December may be related to the greater degree of exposure among unvaccinated people due, among other issues, to tourism, and to the fact that this occurs simultaneously with the transmission of the YF virus sustained by NHPs.
The occurrence of SYF outbreaks in regions with high population density and without adequate vaccine coverage represents a risk of YF reintroduction in urban areas. Even with the vaccination campaigns carried out so far, a large part of the Brazilian population has not yet been immunized5. UYF has been absent in Brazil since 1942, and in human cases that have occurred so far, there has been no epidemiological link with a possible urban cycle20 or involvement of its main urban vector, Ae. aegypti, in viral transmission3. However, this risk is increased by this vector’s presence in almost all Brazilian municipalities32. Another source of concern for the re-urbanization of YF in Brazil is the presence of Ae. albopictus. This mosquito reportedly transmits the YF virus in the laboratory setting and has already been found naturally infected by this virus in the city of Minas Gerais20,33. Present in both urban areas and the rural and forest areas of the country, this mosquito could be a link between the sylvatic and urban forms of the disease3,34. Efforts must be made to prevent the occurrence of UYF epidemics, as could constitute a major public health issue. Among the measures that can be adopted, the most urgent are investments in the production of vaccines, vaccination of the entire Brazilian population, and the development of effective measures to control Ae. aegypti1,21,35.
This study has several limitations, such as the use of secondary surveillance data, which are subject to both notification and underreporting errors. Examples of these problems are the need to eliminate three municipalities from wave modeling and the NHP epizootic underreport. The unavailability of the exact probable site of infection for human cases and epizootics, as well as their occurrence dates, obliged us to consider the centroids of the municipalities and the months of the year. These limitations provide a partial view of the outbreak and did not allow us to investigate, for example, the characteristics of the places where the cases and epizootics occurred. Another limitation was that vaccination coverage was based only on vaccination data for children under 5 years of age.
However, this study has strengths that contribute to its internal and external validity. Among them are the use of both information about sylvatic human cases, as well as epizootics, to investigate the outbreak process. Another strong point was the use of kriging geostatistics to assess the spread of SYF. This is a spatiotemporal process, and the use of kriging allowed us to consider the autocorrelation of the phenomenon in space–time and numerically represent it throughout the SP area.
Source: Ecology - nature.com
