key: cord-0733048-kek9i9zd
authors: Ho, Faith; Tsang, Tim K.; Gao, Huizhi; Xiao, Jingyi; Lau, Eric H.Y.; Wong, Jessica Y.; Wu, Peng; Leung, Gabriel M.; Cowling, Benjamin J.
title: Restaurant-Based Measures to Control Community Transmission of COVID-19, Hong Kong
date: 2022-03-03
journal: Emerg Infect Dis
DOI: 10.3201/eid2803.211015
sha: d768e2f639fec4838adce45adcdfed7ad3f0b1df
doc_id: 733048
cord_uid: kek9i9zd

Controlling transmission in restaurants is an important component of public health and social measures for coronavirus disease. We examined the effects of restaurant measures in Hong Kong. Our findings indicate that shortening operating hours did not have an effect on time-varying effective reproduction number when capacity was already reduced.

. A series of community epidemics have occurred, the largest of which have been the third wave in June-October 2020, which had 3,978 cases, and the fourth wave in November 2020-March 2021, which had 6,048 cases. To suppress local transmission of COVID-19, the government implemented a combination of public health and social measures (PHSMs): bar closures, restaurant capacity restrictions and opening hour restrictions, bans on live music performances and dancing, and work-from-home advisories (2). Ongoing assessment of the effect of these measures on transmission can guide evidence-based policy. One type of location in which COVID-19 transmission is known to occur is restaurants (3) . Earlier studies have evaluated the impact of PHSMs, including restrictions on large group gatherings (4-6), but the specific effect of restaurant measures was not studied. Here we focus on the effect of restaurant measures on transmission in Hong Kong.

We collected details and time of implementation of each intervention of all the PHSMs applied during the third and fourth waves from the official reports of the Hong Kong government (7) (Appendix Table 1 , https://wwwnc.cdc.gov/EID/ article/28/3/21-1015-App1.pdf). In wave 3, a ban on dine-in service after 6:00 pm was in force during July 15-August 27, 2020 ( Figure, panel A) . Other PHSMs were implemented on the same day and kept in place for longer. Wave 4 was initiated by multiple superspreading events in a network of dancing venues. A ban on dine-in service after 6:00 pm was implemented on December 10, 2020, which was a week to a month later than the implementation of other PHSMs ( Figure, panel B) . Hence, we could disentangle the effect of shortened dine-in hours from other measures. No other PHSMs were implemented before the study period.

To determine the effect of the ban on dine-in services after 6:00 pm, we applied a previous approach to estimate time-varying reproduction number (R t ) (8, 9) . Then, we fitted LASSO regression models to log(R t ) to assess the effect of the ban on dine-in services after 6:00 pm on R t , accounting for the effect from other PHSMs (10). We allowed for a 7-day lag between implementation of a measure and its effect on incidence, to account for the incubation period. In both waves, we grouped the PHSMs other than ban on dine-in services after 6:00 pm into a single variable to indicate the period when >3 of these other PHSMs were in place.

We estimated that the ban on dine-in services after 6:00 pm did not reduce R t in both waves, but other PHSMs were associated with substantial reductions in R t . In wave 3, R t rose rapidly to 4.5 on June 27, 2020, but ≈1 week after measures were applied it was <1.0 (Appendix Figure, panel A) . Implementation of >3 other PHSMs was associated with a 53% (95% CI 44%-59%) decrease in R t (Table) .

In wave 4, R t increased to 3.1 on November 16, 2020, and then decreased to ≈1.0 after PHSMs began (Appendix Figure, panel B) . Implementation of >3 other PHSMs was associated with a 40% (95% CI 28%-47%) decrease in R t . Another model that excluded basic civil service arrangement in other PHSMs showed that a ban on dine-in service beginning at 6:00 pm did not have an effect (Table) . We performed sensitivity analysis to remove the effect of superspreading in wave 3 by changing the start date to July 1, 2020; we found the ban on dine-in service from 6:00 pm did not have an effect (Appendix Table 2 ).

Our analysis suggested that the PHSMs were critical for suppressing the third and fourth waves of COVID-19 in Hong Kong. However, we found that a ban on dine-in hours after 6:00 pm might not have had an effect in both waves when capacity was already reduced. A complete closure of restaurants in Hong Kong would have considerable social impact because dining out is very common. We 

Ban on dine-in service after 6:00 PM 0 >3 other PHSMs, excluding basic civil service arrangement hypothesize that encouraging restaurants to extend dine-in hours, but with capacity restrictions to reduce crowding, could be a reasonable approach to reduce transmission. A limitation of our analysis is that we cannot distinguish the effect of some PHSMs because they began simultaneously. We cannot rule out that a ban on dine-in service after 6:00 pm might have an effect if it began earlier than other PHSMs or in regions with high incidences. In addition, changes in R t are a consequence of individual behavioral changes such as avoiding crowded areas; increasing incidence and implementation of multiple PHSMs could raise the public's perception of risk. Determining the effectiveness of alternative PHSMs would provide evidencebased guidance on control strategies. 

The Centre for Health Protection (CHP) of the Department of Health (DH) of Hong Kong. CHP investigates 13 additional confirmed cases of COVID-19

The Government of the Hong Kong Special Administrative Region. Government further tightens social distancing measures

COVID-19 outbreak associated with air conditioning in restaurant

Imperial College COVID-19 Response Team. Estimating the effects of non-pharmaceutical interventions on COVID-19 in Europe

Physical distancing interventions and incidence of coronavirus disease 2019: natural experiment in 149 countries

Inferring the effectiveness of government interventions against COVID-19

The Government of the Hong Kong Special Administrative Region. Press releases

A new framework and software to estimate time-varying reproduction numbers during epidemics

Accounting for imported cases in estimating the time-varying reproductive number of coronavirus disease 2019 in Hong Kong

Regularization paths for generalized linear models via coordinate descent

Ms. Ho is a research postgraduate student at the School of Public Health, University of Hong Kong. Her research interest is the transmission and control of emerging infections.