Considering that Brazil has a goal of eliminating malaria cases by the year 2035 [11] new malaria cases and recurrences were analysed together in the epidemiological profile, even though the database recurrences number are probably underestimated. From the perspective of elimination, recurrence cases must be considered in the time series analysis of the profile, because an untreated recurrence case not treated may become an outbreak source, and it needs to receive malaria treatment. As for the predictive analysis, only new malaria cases were used because 79% of the cases notified in the extra-Amazon region were imported new infections from the Amazon region—as in the study of Machado and collaborators [18]. It is important to say that CSV is underestimated in Sinan because imported cases could move back to their regions and so it is notified in a different information system or another country.
The results highlight that the occurrence of malaria cases in men, as it is in the Amazon region, is higher than in women. Additionally, the cases are made up of mostly adults (20 to 59 years) and have a similar distribution between people of white and brown color/race. The distribution of malaria cases by age group, for the same period, differs from the profile of cases reported in the Amazon region, where the concentration of cases in people under 20 years of age is approximately 46.3% of the total cases. In the extra-Amazon region, the portion of people under 20 years of age does not exceed 9%. As for the distribution of cases by race/colour, the profile is also different because in the Amazon region about 60.8% of cases are brown, 17.6% indigenous and only 6.2% are white (data updated on 06–09-2021) [19, 20].
Considering that most cases reported in the extra-Amazon region are imported cases and that travelling to endemic areas is a known risk factor for contracting malaria [21], this difference in the profile may be related to socio-economic issues such as years of schooling, economic activity, and income. This indicates that the occurrence of malaria cases in the extra-Amazon and Amazon regions of Brazil relates to social inequalities.
The MoH traditionally presents different strategies for both regions, however, it is suggested that specific internal control goals for the country be worked out for the extra-Amazon region [11, 12, 19].
The notification of autochthonous cases in the extra-Amazon region is a warning to epidemiological surveillance centers as it represents the establishment of community transmission, which is a risk for the reintroduction of malaria in these non-endemic, but still susceptible regions [22]. It is important to highlight that approximately 27.5% of the extra-Amazon cases were notified in the year 2018, which was an atypical year for malaria control in Brazil, as the country was showing constant reductions in malaria rates until 2017. In 2016 Brazil had reported approximately 129,000 cases and in 2017 reported over 194,000 [11].
When analysing the PI and the notification place in the extra-Amazon region according to their professional activity, it became clear that a large number of imported malaria cases are from people with professional activities related to travelling or tourism, with the mining activities also standing out. There is a portion of cases that were classified as indeterminate autochthony which are patients with professional activity also related to travelling [23, 24]. Agriculture activities stand out as the most frequent activity, being the most frequent between autochthonous cases. This may be related to the area that they work or live in, considering that these activities may be agriculture for self-sustaining. The high number of cases with undetermined PI also stood out. It should be noted that epidemiological information is as important as the response to possible outbreaks. It is advisable that specific studies be carried out to determine job-related malaria characteristics in Brazil, given that over 38% of economic activities are blank, ignored, or defined as “Other”.
The data has shown two autochthonous cases of malaria by Plasmodium ovale in Brazil, which is a parasitic species with no previous records of community transmission in the country [3, 22, 25]. These notifications should be reviewed by the state and municipal health departments to better understand these cases and certify whether or not there was community transmission. These notifications may be system-filling or diagnostic errors because they were isolated cases, one in 2011 infected in the state of Rondônia and reported in Rio de Janeiro and another in 2014 infected and reported in Espírito Santo.
It is important to highlight that 4.4% of the records for the period had no defined autochthony. Therefore, epidemiological investigation in the extra-Amazon region needs improvement.
As long as the Amazon region presents a high incidence of autochthonous malaria cases the extra-Amazon case notifications will not decrease. It is necessary to understand that the occurrence of autochthonous and imported cases in the extra-Amazon region follows a seasonal pattern and a trend. Those components’ behaviour is related to the malaria epidemiological behavior in the Amazon region [18, 23].
It is critical that specific actions and policies for preventing deaths be directed to the extra-Amazon region, mainly in the states of São Paulo, Minas Gerais, and Goiás, which are the ones with the highest accumulated number of deaths in the last 10 years. Those states may have the highest number of deaths because there have the largest fluxes of travellers and people exposed to infections. Therefore, they are the states with highest number of cases and the number of deaths and fatality rate reflects failures in the identification and opportune treatment of cases. Policies must be directed at identifying suspected cases, testing them and providing proper care. International studies highlight some common risk factors for the occurrence of deaths, such as cases of non-autochthonous malaria, or that come from endemic regions, or that experience therapeutic failures or treatment delay [26, 27].
As shown by Duarte and collaborators [7] it is necessary to implement control and prevention strategies tailored to the realities of each geographic grouping. In addition, it is necessary to assess the quality of death notifications data, as the highest frequency of deaths was for ICD-10 B54 which is "Malaria not specified". The literature [22], as well as results from this study, suggests a high frequency of deaths from P. falciparum in the extra-Amazon region. Linkages techniques should be used in official health information systems for better specification of the species related to malaria deaths.
According to the Brazilian Health Ministry online data, the predictions for the 2021 initial 3 months were close to the value registered in Sinan for malaria cases in the extra-Amazon region (predicted: 108; observed: 113) [20]. Thus, the developed prediction allows better planning of specific control strategies for the containment of malaria cases, especially when considering the seasonality component of the disease [23, 28, 29]. The trend towards a reduction in cases in the extra-Amazon region can be interpreted as a result of the success in disease control and prevention actions carried out on the national and local levels. The model used in this study can be replicated on a state level.
The decrease in 2020 may relate to travel restrictions in the face of the COVID-19 pandemic, thus reducing visits to endemic areas in the North region of the country. This behaviour was previously noted in notifiable diseases in Australia and Taiwan, and this may also have occurred in Brazil [30, 31].
As for the limitations of this study, there is the fact that secondary data were used; the database had some missing information, which is shown by the results tabulations values that do not match each other. This difference is because some notifications do not have all the information on their investigation form filled out. In addition, defining autochthony of death is not possible due to SIM databank limitations. Linkage techniques may be an interesting strategy to present additional and more accurate information. Lastly, occupational activities in the SIM database may not be precise; therefore, further investigation is advised.