Effect of different resumption strategies to flatten the potential COVID-19 outbreaks amid society reopens: a modeling study

The effect of the COVID-19 outbreak has led policymakers around the world to attempt transmission control. However, lockdown and shutdown interventions have caused new social problems and designating policy resumption for infection control when reopening society remains a crucial issue. We investigated the effects of different resumption strategies on COVID-19 transmission using a modeling study setting. We employed a susceptible-exposed-infectious-removed model to simulate COVID-19 outbreaks under five reopening strategies based on the business resumption progress in China. The effect of each strategy was evaluated using the peak values of the epidemic curves with respect to confirmed active cases and cumulative cases. We found that a hierarchy-based reopen strategy performed best when current epidemic prevention measures were maintained save for lockdown, reducing the peak number of active cases and cumulative cases by 50% and 44%, respectively. However, the modeled effect of each strategy decreased when the current intervention was lifted somewhat. Additional attention should be given to regions with significant numbers of migrants, as the potential risk of COVID-19 outbreaks amid society reopening is intrinsically high. Business resumption strategies have the potential to eliminate COVID-19 outbreaks amid society reopening without special control measures. The proposed resumption strategies focused mainly on decreasing the number of imported exposure cases, guaranteeing medical support for epidemic control, or decreasing active cases.


Introduction
. Although China has used a series of non-pharmaceutical strategies (including 23 the Wuhan travel ban on January 23, 2020 [4-6]) to limit the epidemic, the high transmission 24 ability of the coronavirus has led to secondary outbreaks worldwide [7][8][9][10]. Governments and the 25 general public have expressed significant concern regarding these outbreaks and, at the end of May 26 2020, over six million confirmed cases and at last 300 thousand deaths have occurred due to this 27 disease worldwide [11]. 28 Contrasting the explicit life-threatening nature of COVID-19, non-pharmaceutical measures 29 to curb the pandemic have also indirectly influenced people's lives. China has kept its confirmed 30 active COVID-19 cases at a mild order of magnitude, becoming the best example of pandemic 31 control [12]. The rapid increase of confirmed cases in many other countries has driven their 32 governments to set various strict pandemic prevention measures [13][14][15][16]. Such policies (e.g., 33 lockdown and shutdown) have urged hundreds of millions of people to stay at home, creating new 34 social problems (e.g., food shortage, increased unemployment, economic downturn) [17][18][19][20][21]. 35 Given such circumstances, policymakers must cope with the tradeoff between strict epidemic 36 preventions and social operation to address COVID-19 transmission and possible future 37 outbreaks. 38 As COVID-19 infection has been eased with lockdown measures, some societies have 39 suspended operations. China has reported a year-on-year fall in economic output for the first time 40 since 1992, when the National Bureau of Statistics started releasing GDP growth numbers 41 quarterly [22]. China's service industry reported a 5.2 percent fall in value-added due to lockdown 42 measures [23]. Despite this, the Chinese economy has recovered well with factories reopening and 43 some workers returning to cities for work after the extended break caused by . In 44 addition, many other countries have also unveiled plans to reopen their societies, such as the US 45 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 26, 2020. . https://doi.org/10.1101/2020.06.25.20140418 doi: medRxiv preprint affairs. The resumption progress of the first theme was reported on February 3, 2020, with China's 90 domestic mask producers resuming 60 percent of production capacity as the epidemic outbreak 91 was developing. The other themes reported subsequently according to outbreak circumstances. For 92 instance, civil affairs resumed when the epidemic had been somewhat curbed. The resumption 93 ratio is characterized by circle size, and the resumption extent is depicted by circle envelopes 94 (Source: see appendix). 95

96
The most straightforward strategy where the whole working population returns to the 97 workplace together, and epidemic transmission is associated with no further disturbances. 98

99
In this strategy, only people living in low-and-moderate-risk regions are allowed to return to 100 work. Such regions are defined as somewhere the COVID-19 incidence possesses a controlled 101 level. In contrast to DR, the ratio of exposed cases to migrants decreased. 102

103
A more cautious approach would be to have people return to work in batches instead of all at 104 once during the initial stage. We simplified this strategy by allowing one quarter of total migrants 105 to return to work with a second quarter returning after three weeks. We did not distinguish the 106 occupations of those resuming work. 107 TR 108 Considering epidemic intervention, essential material support and services for epidemic 109 control usually reopen first. After that, migrants who returned to work are assumed to have a 110 reduced COVID-19 incidence because of sufficient medical material and services support. After 111 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 26, 2020.  CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 26, 2020. . https://doi.org/10.1101/2020.06.25.20140418 doi: medRxiv preprint intervention is characterized by ‫ݓ‬ . As epidemic prevention measures are always positive to 130 epidemic control, the model is not shaped by any control measures. Those returning to work are 131 deemed asymptomatic due to related measures and are isolated from residents, so the only 132 imported exposed cases come from returned workers, and the susceptible population remained 133 limited to health workers. Therefore, epidemic evolution depended on resuming patterns of 134 socioeconomic activities with a particular transmission model. 135 We ran 500 simulations for each strategy, and imported the exposed cases from a Poisson 136 distribution with the expectation that exposed cases corresponded with the defined ratio for each 137 stage. The model was calibrated using epidemic data from Beijing (cumulative cases, recoveries, 138 and deaths) with a fixed ܴ ൌ

. 2
, where the daily detection rate ‫ݓ‬ reflected the epidemic 139 prevention measures. For generalization, we assumed the imported exposed cases were 140 proportional to the population of returned workers with a basic ratio of 0.1%, wherein the 141 population of migrant residents evaluated the population of returned workers. The ratio for low-142 and moderate-risk regions was set to 70 percent of the basic ratio, and the restarting of essential 143 material support and services for the epidemic control effect could further reduce COVID-19 144 incidence by 20%. The parameters of the model are given in Table 1. 145 The daily infection detection rate 0.5762 (95% CI, 0.2655 to 0.8870) Calibrated by epidemic data in Beijing . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 26, 2020.

150
Overall, as the reproductive number increased, the risk of the virus spreading increased under 151 all five proposed business resumption strategies. DR represented natural outbreaks leading to most 152 severe outbreak with the corresponding transmission being set as the baseline. Among the other 153 four resumption strategies, HR was the most effective way to reduce COVID-19 incidence for 154 reproductive numbers 1.4 and 2.2 (see Fig 2). Given a sizeable reproductive number of 3.9, the 155 business resumption strategy had little effect. 156 Confirmed active cases described how many infections were hospitalized, indicating strain on 157 local medical services. The whole four strategies (RR, TR, OR and HR) performed well given a 158 low reproductive number, while their effects decreased as reproductive numbers increased. 159 Cumulative cases represented infected subjects including imported exposed cases. Given 160 consistent epidemic prevention measures, OR could not decrease the cumulative cases (Fig 2). Our 161 simulation set February 20, 2020, as the day for reopening society. With a reproductive number of 162 1.4, DR and RR controlled transmission within one month (by early April, 2020), while OR, TR 163 and HR required four months (up to late June) to control the epidemic. Given a reproductive 164 number of 2.2, each strategy required roughly nine months (up to late November of 2020) to 165 control the epidemic. A reproductive number of 3.9 saw each strategy needing about five months 166 (until early July, 2020) to control the epidemic. While lifting the control measures (ܴ ൌ 3 . 9 ) 167 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 26, 2020.  Four business resumption strategies led to reduced COVID-19 incidence with differing effects, 177 as shown in Fig 3. In the simulation, RR eased the epidemic by decreasing the total imported 178 exposed cases according to regional COVID-19 incidence (i.e., 70% of the DR strategy). OR eased 179 the epidemic by reducing daily infections and, consequently, the total number of imported exposed 180 cases remained identical to DR. TR functioned equivalently to OR as it also asked people to return 181 to work in batches by occupation. As this strategy called epidemic prevention occupations to 182 return first, it was assumed that regional COVID-19 incidence would decrease accordingly. 183 Therefore, TR imported 75% of exposed cases compared to DR and, to this end, HR combined the 184 above three strategies to grant it the lowest number of imported exposed cases (52.5% of DR . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 26, 2020.  For each reproductive number, the heights presented the ratio of peak numbers between each 198 strategy and DR, with respect to confirmed active, cumulative and imported exposed cases. The 199 RR, OR, TR, and HR had 70%, 100%, 75% and 52.5%, respectively, of imported exposed cases  We found that OR, RR, TR and HR strategies could weaken an epidemic in terms of confirmed 235 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 26, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 26, 2020. . https://doi.org/10.1101/2020.06.25.20140418 doi: medRxiv preprint secondary COVID-19 outbreak. To characterize each strategic effect, we simply assumed one 260 exposed case per ten thousand returned people. Only exposed cases were imported due to strict 261 travel restrictions and control measures in place to prevent those with symptoms going to work. In Our study indicates that societal reopening strategies can support epidemic control without 274 further special control measures. As returning to work is an urgent need for several countries, this 275 study could provide valuable insights into business resumption plans. Our model could also be 276 calibrated according to national situations to test proposed plans based on local situation. However, 277 migrants of different occupations are commonly placed in isolated workplaces, while our 278 simulation simply placed them together. Transmission abilities may also vary with occupation, 279 changing model parameters beyond our study's limits. 280 281 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 26, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 26, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 26, 2020.