The neighborhoods with the highest propensity for a severe COVID–19 epidemic were Aldeota, Cais do Porto, Centro, Edson Queiroz, Vicente Pinzon, Jose de Alencar, Presidente Kennedy, Papicu, Vila Velha, Antonio Bezerra, and Cambeba. The main finding of this study indicates higher levels of propensity to the COVID–19 epidemic, in areas with different socioeconomic profiles, including a group of very poor neighborhoods on the western border of the city (Vila Velha), a set of neighborhoods characterized by a large number of subnormal agglomerates in the Cais do Porto region (Vicente Pizon), and neighborhoods in the oldest central area of the city, where low-income areas exist despite the overall wealth in the area (Aldeota and the adjacent Edson Queiroz).
Indeed, the propensity for a severe COVID–19 epidemic in the neighborhoods of Fortaleza is very heterogeneous and reflects not only the population’s mobility in the urban space, but also the dynamics of transmission of a disease that is influenced by the living situation of a population in a city. Recently, an article indicating that the heterogeneity of the distribution of the incidence of COVID–19 is determined by socio-economic factors was published on the ABC American communication network. This article states that in New York City, a ‘stark contrast’ in COVID–19 infection rates can be observed, based on education and ethnicity [9].
The seven sociodemographic indicators assessed separately had spatial distributions with relevant heterogeneity. Inequality expressed by income (data not shown) had a distribution with an expression of less inequality in the periphery of the municipality and greater inequality in neighborhoods with better economic conditions, such as those in the eastern and coastal zones. Almost in a complementary way, the distribution of the proportion of unemployment showed higher rates in neighborhoods located in the outskirts of the municipality.
A similar situation was observed for indicators of household agglomeration (more than two people per bedroom) and households without access to water or sanitation. The inequality of these distributions indicates that these phenomena are correlated and probably express the evolution of the urban space occupation process in Fortaleza.
Very few studies have assessed the spread of the COVID–19 epidemic, and so far, no articles have been published that appreciat the influence of specific population factors linked to people’s mobility and to predict the occurrence of severe outbreaks in areas within cities. Spatial analysis was used by Kang et al. [10] to understand the epidemic spread of COVID–19. While the authors described the spatiotemporal pattern and evaluated the spatial association of the early stages of the COVID–19 epidemic in mainland China from January 16 to February 6, 2020, they sought only to identify the occurrence of spatial autocorrelation measured by Moran’s I for the various periods studied.
Fan et al. [11] studied the epidemiology of the Novel COVID–19 in Gansu Province, China. They concluded that different from findings from Wuhan Province, the spatial distribution pattern analysis indicated hot spots and spatial outliers in Gansu Province. To detect the spatial distribution pattern of COVID–19 cases at county levels during the study periods, they used local indicators of spatial association to evaluate the relationship between a given location and the surrounding spatial units by local Moran’s I (LISA).
Giuliani et al. [12] studied the spatiotemporal spread of COVID–19 in Italy. They sought to model and predict the number of COVID–19 infections, drawing out the effects of its spatial diffusion. They argue that “forecasts about where and when the disease will occur may be of great usefulness for public decision-makers, as they give the time to intervene on the local public health systems”. However, the authors did not consider the population heterogeneities and their influence in predicting the epidemic in the studied regions.
The study that used the methodological approach most similar to our study was conducted by Pluchino et al. [13]. They proposed a data-driven framework for assessing the epidemic risk of a geographical area (in a predictive way), and to identify high-risk areas within a country. They constructed a risk index combining three different features: (1) the disease hazard, (2) the infection exposure of the area, and (3) its vulnerability. However, vulnerability was considered based on the local data regarding air pollution, mobility, winter temperature, housing concentration, health care density, population size, and age.
Public transport is presented as a definite spatial trend, with trips mainly concentrated in the central and western regions of the city of Fortaleza, which are directly related to the provision of public transport and radio-concentric bus lines. However, as a part of the assumptions in this approved study, these dimensions should be considered while studying the transmission of SARS-CoV–2, in particular using samples of other respiratory-based infectious diseases [14, 15].
Our study aims to contribute to mathematical modelling studies in a complementary manner, in order to predict the dynamics of the COVID–19 epidemic in Brazil. Complementary methodological approaches are required to broaden the understanding of the epidemic and its possible determinants. Indeed, many mathematical models were used to estimate the epidemic curve of the COVID–19 outbreak in Brazilian cities. Rocha-Filho et al. [16] used a variant of the SEIR (Susceptible, Exposed, Infectious, Recovered) classical model, including hospitalized variables (SEIHR model) and an age-stratified structure to analyze the expected time evolution during the onset of the epidemic in the metropolitan area of São Paulo.
One of the main limitations of this study is that the prediction inherent to the methodological approach does not specify the time at which the severe epidemic will most intensely occur in the neighborhoods of Fortaleza. The simplicity of the approach used in our study, which does not exhaust all potential factors that influence the epidemic, is also an advantage over other methods.