key: cord-0785681-xsd5263d authors: Hernandez Mejia, G.; Hernandez-Vargas, E. A. title: When is SARS-CoV-2 in your shopping list? date: 2020-06-14 journal: nan DOI: 10.1101/2020.06.11.20128850 sha: 6166f34d37b237974f413b6dc5a9b85f82fe0d3e doc_id: 785681 cord_uid: xsd5263d The pandemic of coronavirus disease 2019 (COVID-19) has caused, by May 24th 2020, more than 5.3 million confirmed cases worldwide. The necessity of keeping open and accessible public commercial establishments such as supermarkets or pharmacies increases during the pandemic provided that distancing rules and crowd control are satisfied. Herein, using agent-based models, we explore the potential spread of the novel SARS-CoV-2 considering the case of a small size supermarket. For diverse distancing rules and number of simultaneous users (customers), we question flexible and limited movement policies, guiding the flow and interactions of users in place. Results indicate that a guided, limited in movement and well-organized policy combined with a distance rule of at least 1 m between users and a small number of them (15) may aid in the mitigation of potential new contagions in more than 90% compared to the usual policy of flexible movement with more users (30) which may reach up to 64% of mitigation of potential new infections under the same distancing conditions. This study may guide novel strategies for the mitigation of the current COVID-19 pandemic, at any stage, and prevention of future outbreaks of SARS-CoV-2 or related viruses. Introduction small-to-medium size supermarket. Also, we test flexible and limited movement policies that guide the flow and interactions of users inside the supermarket. Results indicate that a guided, limited in movement and well-organized policy may significantly 35 collaborate into the mitigation of potential new contagions of SARS-CoV-2 in public commercial establishments. Commercial establishment layout 38 We consider a general layout of a public commercial establishment, in this case, a small-to-medium size supermarket structure. 39 The considered dimensions of the establishment may also be suitable for pharmacies, discount markets, and convenience stores. Figure 1 . Supermarket layout and spread model. A) Corridors in the layout (paths) are highlighted with dot-colored (marks) lines, showing the routes that agents can follow, also, the identifier number of each corridor is shown. The layout represents a small to medium size supermarket of 30 × 16 m. In this case example, the distance between marks is 50 cm (distancing rule). Since this is a general layout, the labels of the areas of the supermarket can be modified for a more suitable representation. B) Supermarket layout with arrows indicating the directions that each path can handle, some of them are bidirectional, some others unidirectional. C) Agents are divided into four subsets, susceptible (green), potential newly infected agents (orange), workers of the supermarket (cashiers, blue), and the infected agent (red). All agents, except the cashiers, can freely move all around the supermarket using and standing over the marks of the paths, therefore, the distance between agents is given by the separation of the marks of the paths. The layout shows the initial allocation of all agents considering that an infected agent enters the supermarket and moves around it following the paths. D) The infected agent is now in the checkout area after some movements in the layout, four potential newly infected agents have been produced. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 14, 2020. . Figure 1 depicts the considered structure, accounting for one entrance access and one exit. Corridors and shelves are 41 distributed within the supermarket in such a manner that the users can freely move through them, avoiding bottlenecks. Figure 42 1-A shows the corridors marked with dot-colored lines and an identifier number. The marks stand for the distancing rule of an 43 agent when moving, although it is also allowed that more than one agent can stand in the same mark as they freely move. All 44 paths are designed following this metric to map all the supermarket corridors. Figure 1 -B shows the directions that agents can 45 follow on each corridor highlighting the routes available when moving on the layout. Whenever an agent reaches the end of 46 a corridor and can take more than one corridor to continue, it selects the next one based on a probability of selection (50%) with a binomial distribution. We consider three checkout stations operating with one supermarket worker (cashier) per station. As shown in Figure 1 , the labels on the shelves indicate a general conception of the supermarket areas, these can naturally be 49 named differently. The model is rendered in a Cartesian layout where each agent is represented through its xand y-coordinates 50 that change as the agent moves through the supermarket paths. Spread model 52 The model considers the uninfected (U) and infected (I) populations, represented by agents U j and I i , respectively. We consider 53 a potential contagion from an infectious agent to have a probability of spread (P Spread ) of 50% with a binomial distribution. The spread is also governed by the "physical" distance (Euclidean distance, EUC(I i ,U j )) between the infectious agent I i and 55 the uninfected one U j . A potential contagion must therefore first satisfy a minimum distance threshold (M D ) between agents 56 and a positive outcome of the binomial probability. The simulation initiates with a fixed population size N = U + I and, each 57 potential newly infected agent is taken from the U-population and appended to the I-population. A case example with an initial allocation of 15 uninfected users, three cashiers and one infected user is shown in Figure 1 -C. This presents a case in which the infected agent (user) enters the supermarket and starts moving using the paths following the 60 directions of Figure 1 -B, all other agents also move and can leave the supermarket. Those agents that leave the supermarket, 61 infected or uninfected, are replaced by uninfected agents. After some movements inside the supermarket, the infected agent is 62 on the checkout area and has had potential contagion with at least four other uninfected agents, as shown in Figure 1 -D. A 63 case example with an initial allocation of 15 uninfected users, three cashiers and one infected user is shown in Figure 1 -C. This presents a case in which the infected agent (user) enters the supermarket and starts moving using the paths following the directions of Figure 1 -B, all other agents also move and can leave the supermarket. Those agents that leave the supermarket, 66 infected or uninfected, are replaced by uninfected agents. After some movements inside the supermarket, the infected agent is 67 on the checkout area and has had potential contagion with at least four other uninfected agents, as shown in Figure 1 -D. Finally, 68 the simulation steps are as follow: all agents move from the initial allocation to the immediate next one (agent steps) according 69 to the distancing rule and the corridor direction, the minimum distance threshold and contagion probability are checked (all 70 uninfected agents respect to the infectious), potential newly infected are generated if applicable, and all agents move again. The simulation stops when the infectious agent reaches a maximum number of agent steps or leaves the supermarket. The 72 maximum number of steps also allows representing the spent time in the supermarket this can go from 15 to 40 minutes. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 14, 2020. . https://doi.org/10.1101/2020.06.11.20128850 doi: medRxiv preprint Flexible movement policy 75 We aim to test the impact of different distancing rules and population sizes on the generation of potential newly infected. 76 Therefore, we test 5 distancing rules, 30 cm, 50 cm, 1 m, 1.5 m, and 2 m. Besides, we test populations of 15, 20, 30, and 77 50 uninfected agents. In all cases, we account for 3 cashiers and one infected agent. The minimum distance threshold is 1.5 78 m. We perform 1000 simulations per case registering the number of potential newly infected cases as well as the percentage 79 of repetition (frequency) of each case for all the distancing rules and population sizes. We identify this test as the flexible The distancing rule may weakly affect the total of potential newly infected, however, the smaller uninfected populations may 94 add to the mitigation of potential contagions. In Figure 2 -F, the total number of potential contagions present slight differences 95 between distancing rules for populations of up to 20 agents, both cases with around 10 thousand (10 K) total cases. In the case 96 of 30 agents, the total of potential contagions remains around 20 K in all distancing rules, and, for 50 agents, the total reaches 97 almost 40 K of potential newly infected. The greater difference in the number of potential newly infected agents relies on 98 keeping the population inside the supermarket to be less than 30 users, preferably between 15 and 20 users. Figure 1 -A, however, the access to corridor 14, for instance, is fully restricted. There are some corridors in 104 which a user can enter but not exit to the same corridor, for instance, corridor 7 can be accessed from corridor 3 but can not 105 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 14, 2020. . https://doi.org/10.1101/2020.06.11.20128850 doi: medRxiv preprint return to corridor 3 or 6. If corridor 7 is accessed from corridors 8 or 9, the user must return to use one of these corridors. These The general behavior depicts that accounting for the same initial users' allocation shown in Figure 1 -C, users interact with a 108 limited number of other users, mostly with those who entered the supermarket. A case example is depicted in Figure 3 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 14, 2020. Figure 3 . Limited movement policy layout. A) The corridors conserve the same identification number as in Figure 1 -A as well as the dotted lines indicating the distancing rule of the paths. The corridors also show the new movement policy highlighted by oval, elbow, unidirectional and bidirectional arrows all over the supermarket layout. Some corridors like the number 8 and 9, for instance, remain in a bidirectional policy while others like 4 and 7 now present an oval arrow policy. The last indicates that these corridors can be, for example, accessed from corridor 3 but no return can be made to this corridor. Also, corridors 4 and 7 can be accessed from corridors 8 and 9, and users must return to use one of these corridors. Lastly, corridor 14 is no longer available. B) The last steps of a case example simulation accounting for 15 susceptible agents and the limited movement policy, the initial allocation of agents is similar to the one in Figure 1 -C. In this case, the newly infected agents remain closer to the infectious agent, in the checkout area, since agents follow the guidance rules. We further analyze the impact of the tested schemes on reducing the viral spread through the percentage of mitigation 135 of potential contagions, as reported in Table 1 . In this approach, we consider that a population of 50 users is to enter the 136 supermarket, however, we explore different cases using the complete population or subsets of it to enter the supermarket at a 137 time. Accounting for the simulations results in Figure 5 , we compute the mitigation benefit as the percentage of users that naturally with the 2 m rule, the mitigation using the limited movement approach reaches up to 87.5% for the same distancing 153 rule. In this direction, whenever up to 15 users are allowed to share the establishment, the mitigation of potential new contagions 154 benefits from both movement approaches and distancing rules, as may be expected. However, clearer differences can be found 155 when comparing movement policies. In the common approach which is represented by the flexible policy, the percentage of 156 mitigation reports at least 82% and reaches 87.5% for 2 m distancing, while the limited movement policy already reaches 88% Table 1 . Percentage of mitigation of potential contagions. The percentages are presented according to the distancing rules and number of simultaneous users of the commercial establishment. A population size of 50 potential users is considered to evaluate how the number of simultaneous users, distancing rules, and movement policies impact the percentage of mitigation, which is the portion of users that remain susceptible after a simulation case. The percentages of mitigation are presented for the policies of flexible movement and limited movement. besides, we explore two motion policies, the flexible and limited movement policies, whose results comparison are shown interventions, which may produce a second wave of infections 18 , a phenomenon that has been seen in influenza pandemics 19 . Finally, these schemes can be followed when still not a vaccine is available and during the first-in-human trials 20, 21 , but also be 187 combined with vaccination strategies, once feasible. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 14, 2020. . https://doi.org/10.1101/2020.06.11.20128850 doi: medRxiv preprint The approaches herein developed, results and interpretations look to guide policy-makers, merchant authorities, public Social distancing strategies for curbing the COVID-19 epidemic Isolation, quarantine, social distancing and community containment: pivotal role for 202 old-style public health measures in the novel coronavirus (2019-ncov) outbreak Interventions to mitigate early spread of SARS-CoV-2 in Singapore: a modelling study Estimating the number of infections and the impact of non-pharmaceutical interventions on 206 covid-19 in 11 european countries How will country-based mitigation measures 208 influence the course of the COVID-19 epidemic? Evaluating the effectiveness of social distancing interventions against COVID-19. medRxiv What is the evidence for social distancing during global pandemics? a 211 rapid summary of current knowledge Nonpharmaceutical measures for pandemic influenza in nonhealthcare settings-social distancing 213 measures Scientific and ethical basis for social-distancing interventions against COVID-19. The Lancet Epidemiological investigation and intergenerational clinical characteristics of 24 COVID-19 patients 217 associated with supermarket cluster Agent-based modeling in public health: current applications and future directions Agent-based simulation for weekend-extension strategies to mitigate influenza outbreaks Uncovering antibody cross-reaction dynamics in influenza a infections Population simulations of COVID-19 outbreaks provide tools for risk assessment and continuity 225 planning Adoption and impact of non-pharmaceutical interventions for COVID-19 Epidemic preparedness in urban settings: new challenges and opportunities Agent-based social simulation for a checkout layout design 229 of a specific supermarket Beware of the second wave of COVID-19 A perspective on multiple waves of influenza pandemics SARS-CoV-2 vaccines: status report & et. all. Safety, tolerability, and immunogenicity of a recombinant adenovirus type-5 vectored