key: cord-0803659-zfnma54i
authors: Hakim, Avi J.; Victory, Kerton R.; Chevinsky, Jennifer R.; Hast, Marisa A.; Weikum, Damian; Kazazian, Lilit; Mirza, Sara; Bhatkoti, Roma; Schmitz, Michelle M.; Lynch, Michael; Marston, Barbara J.
title: Mitigation Policies, Community Mobility, and COVID-19 Case Counts in Australia, Japan, Hong Kong, and Singapore
date: 2021-02-12
journal: Public Health
DOI: 10.1016/j.puhe.2021.02.001
sha: 2021e98db6c7628cedb329a9e46e0f608e4a8f6b
doc_id: 803659
cord_uid: zfnma54i

Objectives To characterize the timing and trends of select mitigation policies, changes in community mobility, and COVID-19 epidemiology in Australia, Japan, Hong Kong, and Singapore. Study design and methods: Prospective abstraction of publicly available mitigation policies obtained from media reports and government websites. Data analyzed include seven kinds of mitigation policies (mass gathering restrictions, international travel restrictions, passenger screening, traveler isolation/quarantine, school closures, business closures, and domestic movement restrictions) implemented between January 1 and April 26, 2020, changes in selected measures of community mobility assessed by Google Community Mobility Reports data, and COVID-19 epidemiology in Australia, Japan, Hong Kong, and Singapore. Results During the study period, community mobility decreased in Australia, Japan, and Singapore; there was little change in Hong Kong. The largest declines in mobility were seen in places that enforced mitigation policies. Across settings, transit-associated mobility declined the most and workplace-associated mobility the least. Singapore experienced an increase in cases despite the presence of stay-at-home orders, as migrant workers living in dormitories faced challenges to safely quarantine. Conclusions Public policies may have different impacts on mobility and SARS-CoV-2 transmission. When enacting mitigation policies, decision makers should consider the possible impact of enforcement measures, the influence on transmission of factors other than movement restrictions, and the differential impact of mitigation policies on sub-populations.

The rapid spread of SARS-CoV-2, the virus that causes coronavirus disease 2019 (COVID-19), left countries with little recourse but to adopt layered policies over time to mitigate the virus' impact while researchers investigate therapeutics and vaccines. In the absence of medical interventions, a range of nonpharmaceutical interventions (NPIs) have been employed, including international travel restrictions, traveler screening at ports of entry, border closures, domestic movement restrictions, school and business closures, physical distancing, and hand hygiene 1, 2 . Public policies implemented to support these and other mitigation measures aimed to prevent importation of the virus, limit mobility, and decrease opportunities for virus transmission. The use of international travel restrictions and wholesale limitations on domestic movement is unprecedented and either not mentioned in guidance or presented as an extreme policy option to be taken with great consideration 3, 4 .

Countries have taken myriad approaches to the use of NPIs to address the threat of COVID-19. With little data about SARS-CoV-2, policy makers responded as they might to pandemic influenza 5 . Though international travel-related restrictions have been shown to delay but not prevent the spread of influenza once seeded in a country, many countries imposed and maintained travel restrictions despite having local transmission of SARS-CoV-2 [6] [7] [8] . Other countries have taken few, if any, steps to mitigate the spread of the virus 9, 10 . In an effort to limit mobility, and eventually social mixing and disease exposure, many countries closed schools and non-essential businesses 11, 12 . Particularly in countries with community transmission, these policies were often followed by stay-at-home orders to limit exposure even further 13 . Such a layered approach in the application of mitigation policies draws on lessons learned from the 1918 influenza pandemic, as well as the SARS outbreak and H1N1 pandemic [14] [15] [16] .

The effect of NPIs in the context of COVID-19 is still being investigated. Studies have shown that mitigation policies can slow the spread of COVID-19, though the extent of the effect differs by setting 17, 18 . For example, mobility has been shown to decline in some settings after implementation of mitigation policies. In addition, each policy may come with a human cost and a cost to the economy, now and in the future. It is estimated that global gross domestic product will contract by 5.2% in 2020 and it may take time to recover lost jobs and time in school 19, 20 . School closures and stay-at-home orders may further inflict a cost on children and adults through increased interpersonal violence and decreased access to food [21] [22] [23] . Governments are weighing their ability to stem disease transmission with their implications for the economy 19, 24 . 20 .

More information is needed to describe the effectiveness of mitigation policies in diverse settings. To this end, we characterized the timing and trends of select mitigation policies, changes in community mobility, and COVID-19 epidemiology in Australia, Japan, Hong Kong, and Singapore. These locations were among the first in the world to report COVID-19 cases, are close in proximity, utilized different policy approaches, and all saw their epidemics peak during the study period.

Seven kinds of policy measures to support SARS-CoV-2 mitigation-mass gathering restrictions, international travel restrictions, passenger screening, traveler isolation/quarantine, school closures, business closures, and domestic movement restrictions-implemented by government authorities between January 1 and April 26, 2020, were abstracted from media reports, and government and United Nations websites. These data were examined in conjunction with daily new case counts. Daily new case counts starting from January 22, 2020, were extracted from the COVID-19 dashboard maintained by the Center for Systems Science and Engineering (CSSE) at the Johns Hopkins University 25 . A three-day moving average of the daily case count was calculated to account for changes in testing and reporting.

Mobility data beginning February 15, 2020, were obtained from Google's Community Mobility Reports; these provide country and territory-level information on the percentage change in visits to or time spent in six location types compared to each country's baseline or median level of activity on that day of the week from January 3 to February 6, 2020 26 . The baseline period was determined by Google Community Mobility Reports and cannot be modified by users. Visits to and time spent in locations were tracked among smartphone users who turned on their phone's Location History setting 26 . Three mobility measures were selected for analysis: visits to retail and recreation, transit stations, and workplaces. Other measures-parks, residential, and groceries and pharmacies-were not included as we would expect visits to parks to increase as other venues closed as well as more time spent in residential areas. Because groceries and pharmacies provide essential services, these were also excluded.

Four countries or territories were identified for analysis to reflect different mitigation strategies and epidemics: Australia, Japan, Hong Kong, and Singapore. These locations were chosen due to the relatively early date of their first detected COVID-19 case, and because differences in mitigation policy responses and the ensuing epidemiologic patterns may inform application of these policies elsewhere. Additionally, as neighbors in the East Asia Pacific Region, governments and residents may be influenced by events unfolding in their neighbors.

In order to visually represent the effects that mitigation measures might have on community mobility and on the incidence of COVID-19 mitigation over time, mobility, and COVID-19 epidemiology (MME) graphics were developed for each country 27 .

The protocol was reviewed according to the Centers for Disease Control and Prevention's (CDC) human research protection procedures and was determined not to be research.

Australia, Japan, Hong Kong, and Singapore all reported their first COVID-19 cases in January 2020 25 . By April 26, 2020, the end of the study period, Australia reported 6,714 cases and 83 deaths, with the number of cases reported per day peaking at 500 on March 28 28 . Japan reported 13,441 cases and 372 deaths. Its epidemic peaked on April 17, with 1,161 cases reported 28 . Hong Kong reported 1,037 cases and four deaths, with the epidemic peaking with 80 cases reported on March 29 28 . Singapore reported 13,624 cases and 12 deaths, with the number of cases reported per day peaking at 1,426 on April 20 28 .

Australia enacted its first non-travel-related mitigation policy, a ban on gatherings of more than 500 people, on March 15, 2020, approximately two weeks after identification of its first case (Figure 1) . A succession of other policies over the next two weeks reduced the size of permitted gatherings, banned entry of non-residents, extended the summer school holiday, and recommended that the elderly, chronically ill, and indigenous people remain at home (Table 1) . On April 5, the states of New South Wales, Queensland, Victoria, and Tasmania, which together account for 80% of Australia's population and 83% of national COVID-19 cases on that date, instituted mandatory stay-at-home orders 29, 30 . The largest declines in the targeted mobility outcomes occurred on April 10, a national holiday, and ranged from 76% to 82% below baseline. Australia's mitigation policies were generally requirements rather than recommendations and included penalties for non-compliance. The rapid application of increasingly restrictive mitigation policies was followed by a gradual decline in mobility, with the largest change occurring in public transit-associated mobility (median 45% decrease).

Japan implemented its first international travel restriction on January 21, 2020 ( Figure 2 ). Mass gatherings were restricted on February 25, but a limit on the size of the gathering was not set (Table 1) .

On February 27, Prime Minister Shinzo Abe asked primary and secondary schools to close starting March 2; a stay-at-home request was made on March 26 for residents of Tokyo, and was extended nationally on April 16. These policies were non-mandatory requests. Community mobility in Japan began declining in late February, then remained stable for one month before beginning a slow and consistent decline around the time of Tokyo's requested stay-at-home order. The largest declines in the targeted mobility outcomes occurred on April 18 and ranged from 26% to 60% below baseline. Though Japan experienced a decrease in daily cases, relative to the implementation of mitigation policies the decline occurred later than Australia's.

In Hong Kong, a policy instituted on January 29, 2020 called for non-essential civil servants to work from home, and sports, cultural, and leisure facilities to close (Figure 3 ). Primary and secondary schools were closed on February 3. After daily case counts had remained below 15 for more than one-month, civil servants returned to their offices on March 2, followed by the reopening of sports, cultural, and leisure facilities (Table 1) . Mobility remained below baseline and relatively stable, declining to a low of 26% to 60% below baseline, as case counts increased and mitigation policies were reapplied. The number of daily cases peaked on March 29, the same day that gatherings of more than four people were banned. Bars were closed five days later. Daily case counts declined to less than 10 per day by mid-April.

Singapore restricted mass gatherings of over 250 people on March 13, 2020. On March 26, the ban was applied to gatherings of >10 people and some businesses were closed (Figure 4 ). The enactment of these policies was followed by declines in mobility outcomes from 10% to 40% below baseline. During the week of April 7, gatherings of any size were banned, non-essential businesses were closed, schools and universities were closed, and a national stay-at-home order was issued (Table 1) . Mobility outcomes fell sharply by 63% to 83% below baseline. However, Singapore experienced an increase in daily cases with a peak of 1,426 cases on April 20. Many of these cases were in migrant workers living in dormitories housing upwards of 20 people per room 31 .

Our findings reveal that mitigation policies were followed by decreases in the mobility outcomes of interest. Across settings, transit-associated mobility declined the most and workplace-associated mobility the least. Between January 1 and April 26, 2020, Australia, Japan, Hong Kong, and Singapore took varied approaches to implementing similar policies to mitigate the spread of COVID-19. Policies were implemented at different times, including after the peak in daily cases, and in Hong Kong, some policies were lifted and reapplied in responses to fluctuations in daily reported cases. The later epidemic peaks in Japan and Singapore, approximately three weeks after the peaks in Australia and Hong Kong, may reflect the consequences of different mitigation strategies.

Changes in mobility and case counts following the implementation of mitigation policies were uneven, possibly reflecting differences in enforcement, differential ability of specific sub-populations to adhere to the policies, and the impact of changes in practices by individuals, business, and civic organizations in response to media coverage and personal risk assessment. For example, greater reductions in mobility in Australia compared to Japan might reflect the differential enactment and enforcement of policies in these two countries, with Australia enacting mandatory restrictions, while Japan made recommendations. Singapore's experience of business closures and stay-at-home orders highlights that mitigation policies may not be equally effective for all groups within a population. Memories of the 2003 Severe Acute Respiratory Syndrome outbreak may have aroused people in Hong Kong and Singapore to take preemptive measures and limit mobility prior to the implementation of mitigation policies.

Mitigation policies can be enhanced through careful consideration to avoid inadvertently increasing risk of SARS-CoV-2 transmission among sub-populations. They can also be tailored for sub-populations and resources can be provided to enhance the ability of sub-populations to adhere to policies. 32 Migrants may similarly benefit from special policies to protect them from COVID-19 and increase their access to health services during the pandemic 33 .

Our findings suggest an association between the implementation of select mitigation policies and subsequent changes in relevant mobility outcomes. These findings are limited by our inability to attribute changes in mobility to a single mitigation policy. Furthermore, mobility measures may not fully reflect changes, for example in physical distancing, that impact the degree of contact between people. Further research is necessary to assess the association between mobility and SARS-CoV-2 transmission, and to determine the types and degree of changes in mobility needed to achieve a certain level of reduction in SARS-CoV-2 transmission. The application of other mitigation policies, including personal measures such as increased handwashing and the wearing of masks, or engineering controls, such as barriers between cashiers and customers or the use of markers on the ground to indicate where people in a queue should stand, may further modify the effect of mobility by making interactions safer.

Our analysis is also limited by restricting it to locations that experienced a peak in the epidemic, and by our use of national mobility data and case counts which may obscure sub-national variations. Potential variation in the implementation of sub-national policies that may have influenced mobility patterns similarly limits a complete assessment of the role of mitigation policies on mobility. Although mobile phone use is high in Australia, Japan, Hong Kong, and Singapore, it is not universal, and specific subpopulations may not be accurately reflected in Google Community Mobility data. Younger and bettereducated persons in Australia and Japan are most likely to own a smartphone 34 . The same is likely true in Hong Kong and Singapore where mobile phone penetration is over 100% 35, 36 . Moreover, the number of people and geographic distribution of people represented by these data is unknown. Data from media sources may inadvertently include errors.

In Hong Kong, non-essential government workers were told to work from home beginning January 29. Entertainment venues were closed the same day, and on February 3, schools were closed. Any changes in mobility resulting from these policies and occurring through February 6 would have been part of the baseline. As such, it is possible that Hong Kong experienced a greater change in mobility than we have reported due to Google's baseline period.

Changes in mobility may differ by locality based on the number of cases reported. In addition, comparison of mitigation and mobility to trends in case counts did not account for the effects of case isolation, contact tracing, and quarantine of contacts. Finally, there are many reasons that case counts may be incomplete, including related to testing policies and testing capacity.

Our findings highlight seven kinds of mitigation policies governments may enact and the trajectories that epidemics may take. Monitoring mitigation, mobility, and epidemiology graphs (Figures 1-4 ) may help inform countries considering applying or lifting mitigation policies to address COVID-19 37 . MME graphs can provide users with a sense of how mitigation policies are affecting the COVID-19 pandemic. This can inform decision makers as they apply or reapply mitigation policies as part of the response to the COVID-19 pandemic. However, implementation of mitigation policies, changes in community mobility, and daily COVID-19 case counts varied; differences may be explained in part by differences in enforcement, changes in prevention practices that are not related to imposed restrictions, and differential ability to adhere to policies by different sub-populations. Policy makers can consider these also issued for elderly, chronically ill, and indigenous populations) April 5**: mandatory April 16: recommendation *Japan's limits on mass gatherings, business closures, and stay-at-home orders are recommendations, not mandatory orders. **Regional ***Bold font signifies the lifting of policies NR = none reported J o u r n a l P r e -p r o o f 

COVID-19) Mitigation, Mobility, and Epidemiology in Hong Kong

Non-pharmaceutical interventions for pandemic influenza, national and community measures

Effect of non-pharmaceutical interventions to contain COVID-19 in China

Community Mitigation Guidelines to Prevent Pandemic Influenza -United States

Influenza Risk Management: A WHO guide to inform & harmonize national & international pandemic preparedness and response

Strategies for mitigating an influenza pandemic

The effect of travel restrictions on the spread of the 2019 novel coronavirus (COVID-19) outbreak

Nonpharmaceutical Measures for Pandemic Influenza in Nonhealthcare Settings-International Travel-Related Measures

COVID-19 travel restrictions and the International Health Regulations

Love in the time of COVID-19: negligence in the Nicaraguan response. The Lancet Global health

Covid-19: Brazil now has third highest number of cases behind US and Russia

The role of population heterogeneity and human mobility in the spread of pandemic influenza

Human mobility and the spatial transmission of influenza in the United States

Adoption and impact of non-pharmaceutical interventions for COVID-19

Public health interventions and epidemic intensity during the 1918 influenza pandemic

Nonpharmaceutical interventions implemented by US cities during the 1918-1919 influenza pandemic

The effect of large-scale anticontagion policies on the COVID-19 pandemic

Estimating the effects of non-pharmaceutical interventions on COVID-19 in Europe

It could take three years for the US economy to recover from COVID-19

Parenting in a time of COVID-19

Increased Risk for Family Violence During the COVID-19

2020. 24. UN. A UN framework for the immediate socio-economic response to COVID

Google COVID-19 Community Mobility Reports

Guide to Create a Mitigation, Mobility, and Epidemiology Graph

Timing of Community Mitigation and Changes in Reported COVID-19 and Community Mobility -Four U.S. Metropolitan Areas

Australian Bureau of Statistics. 3101.0 -Australian Demographic Statistics

?o pendocument&tabname=Summary&prodno=3101.0&issue=Sep%202019&num=&view=. 31. Beech H. Singapore Seemed to Have Coronavirus Under Control, Until Cases Doubled. The New York Times

Australian Government Department of Health. Coronavirus (COVID-19) advice for Aboriginal and Torres Strait Islander peoples and remote communities

Authority; GoSIMD. Mobile Penetration Rate

Introduction to mitigation, mobility, and epidemiology graphs

 Tables and Figures:  Table 1 . Public policies to support mitigation of SARS-CoV-2 transmission and dates of implementation-Australia, Japan, Hong Kong, and Singapore, January 21