Spatial dynamics of dengue fever spreading for the coexistence of two serotypes with an application to the city of São Paulo, Brazil

https://doi.org/10.1016/j.cmpb.2022.106758Get rights and content

Highlights

  • A model based on cellular automata and ordinary differential equations is proposed for dengue.

  • The spatial dynamics of the disease is studied with the coexistence of two serotypes.

  • A methodology for a spatial basic reproduction (R0s) number is proposed.

  • The model is applied for a neighbourhood in São Paulo, Brazil.

  • R0s was assertive in showing areas with a higher disease incidence.

Abstract

Background and objective:Dengue fever is a disease in which individuals’ spatial distribution and Aedes aegypti mosquitoes breeding places are important factors for the disease dynamics. Typically urban, dengue is a problem for least developed countries due to the ineffectiveness in controlling the vector and disorderly urbanization processes. The result is a composition of urban sanitation problems and areas with high demographic densities and intense flows of people. This paper explores the spatial distribution of vector breeding places to evaluate introducing a new dengue serotype to a population at equilibrium for a pre-existing serotype. The paper’s objective is to analyze the spatial dynamics of dengue using variations of the basic reproduction number.

Methods:A model based on probabilistic cellular automata is proposed to permitting the necessary flexibility to consider some spatial distributions of vector breeding places. Then, ordinary differential equations are used as a mean-field approach of the model, and the basic reproduction number (R0) is derived considering the next-generation matrix method. A spatial approach for R0 is also proposed, and the model is tested in a neighbourhood from the city of São Paulo, Brazil, to examine the potential risks of vector breeding cells distribution.

Results:The results indicated that the more spread out these places, the higher are the values of R0. When the model is applied to a neighbourhood in São Paulo, residential areas may boost the infections and must be under public vigilance to combat vector breeding sites.

Conclusions:Considering the mean-field approximation of the cellular automata model by ordinary differential equations, the basic reproduction number derived returned an estimative of the disease dynamics in the population. However, the spatial basic reproduction number was more assertive in showing areas with a higher disease incidence. Moreover, the model could be easily adapted to be used in real maps enabling simulations closer to real problems.

Introduction

Dengue is responsible for 96 million symptomatic cases and an estimated 40,000 death per year globally; also, around 3.9 billion people in 129 countries lives in risk regions for dengue [6], [44], [46]. Despite the efforts to combat and control the disease, dengue is one of the oldest diseases that humankind has contact [6], [26], [45], [53].

Having the vector for transmitting the disease as the Aedes aegypti female mosquito, the dengue fever has a peculiar problem: sequential infections may evolve to dengue shock syndrome, increasing disease severity due to a process called antibody-dependent enhancement [12], [21], [22], [40]. This is a big concern for a disease with five antigenically distinct viruses: DENV-1, DENV-2, DENV-3, DENV-4 and DENV-5 [1], [20], given that an individual has life-long immunity to the cured case serotype and a temporary low cross-immunity of 3 to 9 months for other serotypes [1], [32], [47], [61]. In Brazil, there is a concern that the co-circulation of multiple dengue serotypes, called hyperendemicity, is becoming standard for an increasing number of cities and metropolitan regions [11], [17], [59]. This is a problem since the 1990 decade, and it was confirmed by a few studies which report the presence of individuals with antibodies to three serotypes in northeastern Brazil in 2005 [4], [8], [24], [41], and the four serotypes with simultaneous circulation in Manaus [3], [4], [50].

The Aedes aegypti females mosquitoes breed in clean still water and go in search of human blood, reaching distances of eight hundred meters in six days [27]. Therefore, it is a disease with spreading dynamics dependent on the spatial distribution of humans, vectors and places of standing water. For instance, the work reported in Freitas et al. [18] showed that cases of dengue, zika and chikungunya were not detected at the same location and time in Rio de Janeiro in 2015/2016, with the reasons raised by the authors related to the spatial dynamics of the diseases: competition between viruses for breeding resources, and change in behaviour of susceptible human population as a response to increasing cases in the area. Such cases clusters were also studied in Bhoomiboonchoo et al. [5], where the depletion of cases occurred after spatial and temporal targeted interventions. Finally, other studies considered the landscape and seasonality to describe and predict dengue cases [31], [49], [62].

Therefore, the objective of this paper is to explore the spatial distributions of breeding places for mosquitoes in a model based on probabilistic cellular automata [47] for the co-circulation of two strains of the dengue virus. The basic reproduction number (R0) is derived from the ordinary differential equations (ODE) proposed as a mean-field approximation of the cellular automata. Moreover, we propose a method for estimating the spatial basic reproduction number (R0S) to understand the real impact of this spatial configuration on the disease outbreak. The following points summarize the novelty of the paper: consider two dengue serotypes for numerical spatial analysis, given that many regions in Brazil face such a problem [3], [11], [59]; explore spatial distributions of mosquitoes breeding places; obtain the basic reproduction number for a model which takes into account the severity of a second infection of dengue; propose a spatial numerical basic reproduction number based on similar processes of spatial analysis; use the model for an accurate population distribution to estimate the impact of mosquitoes breeding places in residences and public spaces.

The basic reproduction number is “an epidemiologic metric used to describe the contagiousness or transmissibility of infectious agents” [13]. It represents the number of secondary infections from a single infected individual in a population fully susceptible to the disease. When ordinary differential equations (ODE) are used for compartmental models, R0 is also a bifurcation parameter, where its value comes from the stability analysis of the system [28], [33].

Regarding vector-borne diseases, R0 can be interpreted as the number of secondary disease cases in humans from a single infected human individual in a population of humans and vectors completely susceptible [7]; which is a metric most often used due to the difficulty of estimating biological and environmental mosquitoes information [13], [28], [38], [43], [58]. The value of R0 for dengue depends on many variables, such as population density, social organization, and weather seasonality [13], with global warming being a concern as it may increase the R0 of dengue in temperate regions [34].

The next-generation matrix can be used to derive the basic reproduction number. The matrix is formed by the infectious subsystem of a compartmental model linearised about the disease-free equilibrium state. It represents the expected number of secondary infections caused by a single infected individual for a population entirely susceptible for each combination of row and columns of the matrix. This method was first described in Diekmann et al. [15] and further elaborated in Van Den Driessche and Watmough [56]. Here, the basic reproduction number is calculated from the ordinary differential equations using the next-generation matrix [60]. The ODE parameters are obtained from the probabilistic cellular automata simulations, in which multiple breeding places distributions are used.

Heterogeneous spatial distributions of human and vector populations require alternatives for modelling vector-borne diseases dynamics. Releases of female Aedes aegypti and Aedes albopictus to estimate their flight range is an example of an experimental process to evaluate the spatial infection spreading in Rio de Janeiro, Brazil [27]. In the same city, patches of the map representing neighbourhoods were used to detect spatio-temporal clustering of dengue, chikungunya and Zika, assessing the cases of the disease per patch [18]. Patches of maps have also been used for space-time scan statistical analyses of dengue in China [62], and to mapping the basic reproduction number for vector-borne diseases in Netherlands [25]. Still, in the Hartemink et al. study, the basic reproduction number is calculated for 107 spots in the map, representing the local situation in these regions. Mathematical models demonstrate that the spatial density of the winged female Aedes aegypti and the aquatic form of mosquitoes are distance-dependent to breeding sites [5], [16]. Therefore, studies consider the distance to a point in the map as a variable to determine epidemiological parameters, such as the basic reproduction number [39], mosquito eggs count in a map [49], and the probability of being bitten by a tick in forests [57].

A water supply crisis took place in the southeast Brazil region in 2014–2015. It was partly responsible for a peak in annual dengue cases since households started to store water in inappropriate compartments, which helped the multiplication of mosquito breeding sites. Only in the city of São Paulo, more than 100,000 cases have been reported in 2015 [55]. This is one of the reasons to apply the developed model and methodology for a neighbourhood in São Paulo. Two scenarios are tested to explore the dynamics of the spatial distributions of vector breeding sites. The first considers that these breeding sites are on residential areas, and the second considers public squares, parks and schools.

This paper is organized as follows: the next section contains the methodology and description of the models, the results of the simulations are in Section 3, and a final discussion is presented in Section 4.

Section snippets

Methods

The human population is based on a probabilistic cellular automaton with cells representing individuals in one of the disease’s states. The lattice is square with side n and a total human population of NH=n×n, and the neighbourhood, i.e., the neighbour cells that individuals visit per time step consists of C visits inside a Moore radius r per time step. The neighbourhood is different at each time step, with the C visits being randomly chosen according to the following procedure: the individual

Results

For the simulations, we consider a lattice with side n=200, and NH=200×200=40,000 human individuals. The vector constant population consists of NV=72,000 vector individuals initially distributed over the 3600 breeding cells. Further details will be explained later. The parameters for PCA simulations are Pc1=0.7, Pd1=0.3, Pc2=0.5, Pd2=0.4, PdV=0.2, and PB=0.1. The simulation runs for ts=200 time steps starting with the initial conditions S1,2(0)=0.99, I1(0)=0.01, and I2(0)=I1/2(0)=I2/1(0)=R(0)=0

The model in a real map

The model is adapted to a real map for evaluating the impact of breeding cells over space. Therefore, we used a region from the east part of the São Paulo city called Vila Curuça, where there are some required characteristics on the map, such as residential neighbourhood, approximately 10,000 residents per square kilometre, and public areas consisting of squares, parks and children’s education schools. Therefore, the map and the layers for each feature of the map were downloaded from [42] using

Discussion

This paper proposed a model based on probabilistic cellular automata to study the dengue fever dynamics on a few spatial distributions of breeding places. Ordinary differential equations were used to derive the basic reproduction number from the next-generation matrix method. Simulations were set so that the population was in the steady state for one strain of the disease when another strain was added to the population. Five spatial distributions of breeding places were tested, and the cases

Statements of ethical approval

Ethics approval was not required for this study.

Funding

PHTS is supported by grants #440025/2020-6, #307194/2019-1 and #402874/2016-1 of Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq). FMMP is supported by scholarship grant #88887.505459/2020-0 of Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES).

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

PHTS and FMPP would like to show gratitude to Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) and São Paulo Research Foundation (FAPESP) due to the current and past funded projects.

References (63)

  • QGIS Development Team

    QGIS Geographic Information System

    (2009)
  • B.B. Salgado, F.C. de Jesus Maués, R.L. Pereira, J.O. Chiang, M.N. de Oliveira Freitas, M.S. Ferreira, L.C. Martins,...
  • W.R. Shaw et al.

    Vector biology meets disease control: using basic research to fight vector-borne diseases

    Nat. Microbiol.

    (2019)
  • P.M. São Paulo, Dengue - srie histrica de 2007 a 2021 - dve - covisa, 2021,...
  • P. Van Den Driessche et al.

    Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmission

    Math. Biosci.

    (2002)
  • H.M. Yang

    The basic reproduction number obtained from jacobian and next generation matrices a case study of dengue transmission modelling

    BioSystems

    (2014)
  • H.M. Yang et al.

    Assessing the effects of vector control on dengue transmission

    Appl. Math. Comput.

    (2008)
  • M. Aguiar et al.

    Epidemiology of dengue fever: a model with temporary cross-immunity and possible secondary infection shows bifurcations and chaotic behaviour in wide parameter regions

    Math. Model. Nat. Phenom.

    (2008)
  • M. Aguiar et al.

    Mathematical models of dengue fever epidemiology: multi-strain dynamics, immunological aspects associated to disease severity and vaccines

    Commun. Biomath. Sci.

    (2017)
  • M.d.S. Bastos et al.

    Simultaneous circulation of all four dengue serotypes in Manaus, State of Amazonas, Brazil in 2011

    Rev. Soc. Bras. Med. Trop.

    (2012)
  • J.M.T. Bezerra et al.

    Entry of dengue virus serotypes and their geographic distribution in Brazilian federative units: a systematic review

    Rev. Bras. Epidemiol.

    (2021)
  • P. Bhoomiboonchoo et al.

    The spatial dynamics of dengue virus in Kamphaeng Phet, Thailand

    PLoS Negl. Trop. Dis.

    (2014)
  • O.J. Brady et al.

    The global expansion of dengue: how Aedes aegyptimosquitoes enabled the first pandemic arbovirus

    Annu. Rev. Entomol.

    (2020)
  • P.M. Castanha et al.

    Force of infection of dengue serotypes in a population-based study in the northeast of Brazil

    Epidemiol. Infect.

    (2013)
  • A.Y. Chang et al.

    Combining google earth and GIS mapping technologies in a dengue surveillance system for developing countries

    Int. J. Health Geogr.

    (2009)
  • T.E. Colombo et al.

    Co-infection of dengue virus by serotypes 1 and 4 in patient from medium sized city from Brazil

    Rev. Inst. Med. Trop. Sao Paulo

    (2013)
  • W. Dejnirattisai et al.

    Cross-reacting antibodies enhance dengue virus infection in humans

    Science

    (2010)
  • P.L. Delamater et al.

    Complexity of the basic reproduction number (R0)

    Emerg. Infect. Dis.

    (2019)
  • O. Diekmann et al.

    The construction of next-generation matrices for compartmental epidemic models

    J. R. Soc. Interface

    (2010)
  • O. Diekmann et al.

    On the definition and the computation of the basic reproduction ratio R0 in models for infectious diseases in heterogeneous populations

    J. Math. Biol.

    (1990)
  • J. Fang et al.

    Spatial dynamics of a dengue transmission model in time-space periodic environment

    J. Differ. Equ.

    (2020)
  • Cited by (0)

    View full text