Preventive measures focused on the urban-rural interface protect rural food-producing communities from SARS-CoV-2

Abstract Introduction: Rural food-producing communities are fundamental for the development of economic activities associated with sustainability and food security. However, despite the importance of rurality in Colombia, preventive strategies continue to be implemented homogeneously, without considering the dynamics of SARS-CoV-2 in rural food-producing communities. Objective: To model real areas in Colombia involving rural and urban populations that have intrinsic SARS-CoV-2 transmission dynamics. Characterize rural-urban interactions by means of a parameter that provides different scenarios and allows us to identify interactions capable of preventing SARS-CoV-2 transmission in rural food-producing communities. Materials and methods: The dynamics of SARS-CoV-2 infection was modeled in five case studies (Boyacá, Caquetá, Cundinamarca, Santander and Sucre) considering urban and rural areas and their interaction (connectivity) in the urban-rural interface. For this purpose, an epidemiological compartmental model considering a classification of individuals according to their economic activity and their epidemiological status was assessed. Results: Preventive measures focused on the urban-rural interface impact the number of deaths in rural areas. Hence, it is possible to assume that the dynamics of the disease in rural areas depend on the constant interaction with infected individuals from urban areas, which occurs due to the food production dynamics in the urban-rural interface. Conclusions: Preventive measures should focus on places of high transmissibility and risk for rural communities, such as the urban-rural interface. This work highlights the importance of national heterogeneous preventive measures and the protection of rural communities from the social and economic impacts of SARS-CoV-2.

In 2020, the World Health Organization (WHO) declared the severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) outbreak, originally reported on December 31, 2019, in Wuhan of the Hubei province in China, as a public health emergency (1). Until today, the preventive measures against SARS-CoV-2 have focused globally in the homogeneous implementation of mitigation measures such as the interruption of non-essential services for prolonged periods, generating social and economic costs (2), or suppression measures such as social distancing that includes the restriction of mobility and the establishment of remote work. On March 17, 2020, the Colombian government issued Decree 417 of 2020, declaring the State of Economic, Social and Ecological Emergency throughout the national territory (3). However, differences (heterogeneities) between urban and rural areas were not considered and, therefore, the rural population had difficulties adapting to mobility restriction measures and implementing remote work, suffering negative economic consequences as a result (2).
Colombia is one of the Latin American countries with the largest rural population (4). According to the Departamento Administrativo Nacional de Estadística (DANE), the estimate of inhabitants in rural areas in the country is 22.9% of the total estimated population of 48,258,494 (5). Although Colombia, like most countries in the world, has undergone an important urbanization process, only 0.3% of the entire Colombian territory corresponds to urban areas (6) and 53% of the population is concentrated in rural territories or in the urban-rural interface (7). Although it has heterogeneous conditions, and despite the more than 130,000 deaths reported associated to the SARS-CoV-2 pandemic, and more than six million reported cases (8). preventive measures continue to be implemented homogeneously, without considering the virus dynamics in rural food-producing communities and in the urban-rural interface.
Based on the impossibility of implementing suppression (i.e., mobility restriction and remote work) or mitigation (i.e., interruption of activities) measures in rural food-producing communities, this work aims to model real areas in Colombia that have an intrinsic SARS-CoV-2 transmission dynamics in these populations. We characterize rural-urban interaction by means of different scenarios that allows us to identify interactions capable of preventing SARS-CoV-2 transmission in rural food-producing communities.

Study area
Colombia has a population of approximately 50 million people, distributed in 32 administrative units called departments, with Bogotá, D.C., as the capital district. In this work, five departments were considered: Boyacá, Caquetá and Cundinamarca (dairy producer), Santander (poultry producer) and Sucre (fish producer) (table 1). In each department, the main urban center was considered, as well as the main neighboring rural municipalities associated with each production system (figure 1).

Data
The data were obtained from the public reports of mortality due to SARS-CoV-2 reported by the Instituto Nacional de Salud of Colombia (8)

Model
To understand the dynamics of SARS-CoV-2 infection in selected rural food-producing communities, an epidemiological model was considered, classifying individuals according to their economic activity (rural -R: development of activities exclusively in the rural area), urban -U: development of activities exclusively in the urban area) or urban-rural -UR: development of activities in the urban-rural interface) and their epidemiological status (susceptible -S: at risk of developing the disease); exposed -E: infected but not infectious); infectious -I: capable of transmitting the disease; recovered -R: recovered from the disease; or deceased -F: death due to the disease (9).
According to this classification, susceptible individuals residing in rural areas could be exposed to SARS-CoV-2 infection by developing activities at the urban-rural interface through contact with infected individuals residing in urban areas. Likewise, these individuals residing in urban areas could be exposed to the infection through contact with other infected individuals who also reside in urban areas or when carrying out activities in the urban-rural interface. Due to the low contact rate reported in rural areas (2), the model does not consider transmission of SARS-CoV-2 among people residing in rural areas. As shown in table 2, this model can be described by twelve different transitions (reactions) from the ten different states and transitions (figure 2).

Event
Reaction Parameter Exposure of a susceptible individual living in an urban area by contact with infectious individuals in an urban area.
Exposure of a susceptible individual living in an urban area by contact with infectious individuals in the rural area at the urban-rural interface.
Su → Eu ϵ 2 Exposure of a susceptible individual residing in a rural area by contact with infectious individuals in the urban area at the urban-rural interface.
Sr → Er ϵ 3 Infection of an exposed individual living in an urban area. Eu → Iu β 1 Infection of an exposed individual residing in a rural area.
Er → Ir β 2 Recovery of an infectious individual living in an urban area.
Iu → Ru ɣ 1 Recovery of infectious individual residing in rural area.
Ir → Rr ɣ 2 Death of an infectious individual living in an urban area.
Iu → Fu ð 1 Death of an infectious individual residing in a rural area.
Ir → Fr ð 2 Table 2. Events, reactions, and parameters of the SEIRF epidemiological model

Model fitting and scenario simulations
The epidemiological model was fitted to the number of deaths reported by the INS through the adjustment of the model parameters using a maximum likelihood function (9)(10)(11). For this adjustment, the number of initial susceptible individuals in each urban or rural municipality was considered according to the population size reported by the DANE (5). In this work, we used the maximum likelihood optimizer through the function "mle2" of the package "bbmle" of the computational language R (Bolker, 2020) to estimate this method for the model parameter θ of each department.
In order to evaluate the impact of preventive measures focused on the urban-rural interface on rural areas, four scenarios associated with variations in the contact rate were considered (12,13) (ϵ = 70%, ϵ = 40%, ϵ = 20%, ϵ = 0%) between the individuals on this interface. Purple and green dots in figure 3 describe the cumulative number of deaths reported in urban and rural food-producing communities, respectively, while the lines indicate the fit of the model to the actual data. There is a clear difference in the epidemiological profile of the pandemic in the different municipalities. Specifically, the first epidemiological week, defined as the initial week in which the first infections were detected, it is different for each department, as can be observed in the x-axis of figure 3. Within the  Su: Urban susceptibles; Sr: Rural susceptibles; Eu: Urban exposed; Er: Rural exposed; Iu: Urban infected; Ir: Rural infected; Fu: Urban deaths; Fr: Rural deaths; Ru: Recovered urban; Recovered rural departments of Sucre and Caquetá, a difference in the first epidemiological week can also be observed for urban and rural populations. However, in Cundinamarca and Boyacá, departments with a high interaction between urban and rural communities, deaths occur simultaneously in rural and urban populations.

Results
There is also a marked difference between the curves describing the accumulated deaths in urban and rural areas in Caquetá, Sucre, and Santander. However, in the departments of Cundinamarca and Boyacá, there is a very similar accumulated number of deaths for rural and urban areas. This difference in the case study departments is mainly because the coefficient that simulates the interaction in the urban-rural interface is markedly higher in Cundinamarca and Boyacá (table supplementary 1).
The importance of the interaction in the urban-rural interface becomes explicit when we consider its variation throughout four scenarios, namely: ϵ = 70%, ϵ = 40%, ϵ = 20%, ϵ = 0%. The modulation of the parameter ϵ counts for the implementation of different preventive measures focused on the urbanrural interface. As observed in figure 3, the preventive measures focused on the urban-rural interface impact the number of deaths in rural areas of all the considered departments, so it can be assumed that the dynamics of the disease in rural areas depend on connectivity (contact rate) at the urban-rural interface. Additionally, as shown in the case of Sucre, preventive measures focused on the urban-rural interface can delay the epidemic peak in urban areas.

Discussion
Our findings confirm that information about human mobility and connectivity should be a starting point for modeling important dynamic processes in human and animal epidemiology, population ecology, biology, and evolution (14)(15)(16)(17). The urban-rural interface is essential for the development of economic activities associated with sustainability and food security (18), so it is essential to establish strategic and differentiated preventive measures for these communities. Preventive measures focused on the urban-rural interface may include: mandatory and appropriate use of personal protection items such as face masks (12,19,20); implementation of transportation systems/ schedules to avoid crowds, including movement restrictions for all family members (20-23); trade products exclusively in open places that social distancing and low contact rates with implementation of sinks and cleaning items that allow constant handwashing (24,25). These recommendations focused on the urban-rural interface will potentially impact the dynamics of the infection, reducing the number of deaths in rural communities.
In the context of a global health emergency, the contributions of epidemiological models are essential to expand the knowledge regarding transmission dynamics, identify patterns and individuals with greater susceptibility, and propose strategic measures that minimize the inevitable adverse health, economic, and social effects (26,27). The findings of the proposed model warn that highly connected areas, such as the urban-rural interface, are vulnerable to infectious outbreaks and can hardly be adapted to the preventive measures proposed worldwide. Measures focused on reducing connectivity or guaranteeing safe interactions in the urban-rural interface can be an effective mitigation strategy to avoid the geographical spread of diseases to rural areas.
Considering the little importance given to the interaction between the rural population, the implementation of preventive measures focused on rural areas is not of fundamental significance in terms of reducing the transmission of SARS-CoV-2. Therefore, people associated exclusively with food production systems do not need to be subjected to the preventive measures implemented by the national government, such as mobility restrictions. Therefore, these areas are exposed to SARS-CoV-2 when trading their products and conducting activities at the urban-rural interface. Thus, in order to reduce exposure to the virus infection, and prevent itstransmission to rural communities in Colombia, measures should focus on places with high transmissibility and risk for rural communities, such as the urban-rural interface.