Trends in Interregional Travel to Shopping Malls and Restaurants Before and After Differential COVID-19 Restrictions in the Greater Toronto Area

In the fall of 2020, the government of Ontario, Canada, adopted a 5-tier, regional framework of public health measures for the COVID-19 pandemic in its 34 public health regions.¹ The goal of nonpharmaceutical interventions was to suppress transmission by reducing contact rates, which can be indirectly assessed using mobility data. Five of the 6 most populous health regions in Ontario are located in the Greater Toronto Area: Toronto (3.0 million), Peel (1.5 million), York (1.2 million), Durham (0.7 million), and Halton (0.6 million). The urban core of Toronto and Peel is a perpetual hotspot for COVID-19² and remains highly interconnected with the peripheral regions of York, Durham, and Halton.

Toronto and Peel were the first regions in Ontario to enter the highest restriction tier (ie, lockdown) during the second wave of COVID-19. On November 23, 2020, Toronto and Peel closed restaurants to in-person dining and limited nonessential businesses, including shopping malls, to curbside pickup. York entered lockdown on December 14, 2020, followed by the rest of the province, including Durham and Halton, on December 26, 2020. In this cohort study, we examine whether the implementation of differentially timed restrictions in a highly interconnected metropolitan area was associated with increased interregional travel, potentially driving further transmission of SARS-CoV-2.

This cohort study received ethical approval from the University of Toronto research ethics board through the Ontario COVID-19 Modeling Consensus Table. Informed consent was waived because data were anonymous, and the study posed minimal risk. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

We used anonymized mobile device data from Veraset representing 154 089 unique devices (3.4% of the population) to analyze patterns of travel by residents of regions in the urban core (Toronto and Peel) to shopping malls and restaurants in peripheral regions in the week before the November 23 lockdown compared with the week after the lockdown (eFigure in the Supplement). Restaurants and shopping malls are both important settings for transmission risk.^3,4 A device’s home region for a given month was identified as where it spent most of its time during that month. The proportion of devices in the data set that visited malls or restaurants was multiplied by the population of the region (2019 estimates)⁵ to estimate the actual number of visitors. We also measured visits by residents of Toronto and Peel to shopping malls in York relative to a baseline calculated for each day of the week from January 1 to February 5, 2020. Neighborhood sociodemographic characteristics of devices captured in the Veraset sample are contrasted with the general population of Toronto and Peel in the eTable in the Supplement.

One-sided P values were calculated using the bayesian posterior distribution of a structural time series fit to the preintervention daily data. Statistical significance was set at P < .05, and data analysis was performed between January 2021 to June 2021 using the statistical package R version 4.0.2 (R Foundation for Statistical Computing).

Residents of Toronto and Peel took fewer trips to shopping malls and restaurants in the week following lockdown (shopping malls: Toronto, −15.3% [95% CI, −28.5 to −5.4]; Peel, −18.2% [95% CI, −30.0 to −4.7]; restaurants: Toronto, −16.9% [95% CI, −28.8 to 0.0]; Peel, −20.2% [95% CI, −32.1 to −6.7]) (Table). During the same time, there was a significant increase in trips to shopping malls in peripheral regions by residents of the regions in lockdown (Toronto: +40.7% [95% CI, 27.0 to 56.6]; Peel: +65.5% [95% CI, 54.2 to 81.7]); however, visits to peripheral regions were still well below historical means (Figure). Visits to restaurants in peripheral regions did not decrease (Toronto: +6.3% [95% CI, −8.0 to 23.6]; Peel: +11.8% [95% CI, −6.0 to 20.9]).

Lockdowns in the urban core were associated with reduced overall visits to shopping malls and restaurants by residents but were not associated with decreased travel to these businesses in peripheral regions, where restrictions permitted indoor dining and shopping for nonessential businesses. We observed a large increase in visits to shopping malls in the peripheral regions by residents of the urban center in the week following the lockdown. These heterogeneous restrictions may lead to unintended consequences, undermining lockdowns in the urban core and driving residents from zones of higher transmission to zones of lower transmission. While our sample was limited to a fraction of the population, neighborhood sociodemographic characteristics were similar to the general population. Regional nonpharmaceutical intervention frameworks could avoid these consequences by implementing restrictions spanning both the core and periphery of urban areas or using interregional travel restrictions. These concerns are likely generalizable to other major metropolitan areas, which often comprise interconnected but administratively independent regions.⁶

Accepted for Publication: June 25, 2021.

Published: August 31, 2021. doi:10.1001/jamanetworkopen.2021.23139

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Soucy JPR et al. JAMA Network Open.

Corresponding Author: Jean-Paul R. Soucy, MSc, Division of Epidemiology, Dalla Lana School of Public Health, 155 College St, Rm 500, Toronto, ON M5T 3M7, Canada (jeanpaul.soucy@mail.utoronto.ca).

Author Contributions: Dr Ghasemi had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Mr Soucy and Dr Ghasemi provided equal contribution.

Concept and design: Soucy, Ghasemi, Sturrock, Berry, Buchan, Brown.

Acquisition, analysis, or interpretation of data: Soucy, Ghasemi, Buchan, MacFadden, Brown.

Drafting of the manuscript: Soucy, Ghasemi, Brown.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Ghasemi, Brown.

Obtained funding: Brown.

Administrative, technical, or material support: Sturrock, Berry, MacFadden, Brown.

Supervision: Brown.

Conflict of Interest Disclosures: Mr Soucy reported receiving personal fees from Public Health Agency of Canada outside the submitted work. Ms Sturrock reported receiving personal fees from The Regional Municipality of York outside the submitted work. No other disclosures were reported.

Additional Contributions: We thank Nick Daneman, MD, MSc, (Sunnybrook Hospital, Toronto, Ontario, Canada) and Nicholas Gibb, MSc (Public Health Agency of Canada, Ottawa, Ontario, Canada) for their contributions to this analysis and manuscript. These individuals received no compensation.