Abstract

Mass gathering events have been identified as high-risk environments for community transmission of coronavirus disease 2019 (COVID-19). Empirical estimates of their direct and spill-over effects however remain challenging to identify. In this study, we propose the use of a novel synthetic control framework to obtain causal estimates for direct and spill-over impacts of these events. The Sabah state elections in Malaysia were used as an example for our proposed methodology and we investigate the event’s spatial and temporal impacts on COVID-19 transmission. Results indicate an estimated (i) 70.0% of COVID-19 case counts within Sabah post-state election were attributable to the election’s direct effect; (ii) 64.4% of COVID-19 cases in the rest of Malaysia post-state election were attributable to the election’s spill-over effects. Sensitivity analysis was further conducted by examining epidemiological pre-trends, surveillance efforts, varying synthetic control matching characteristics and spill-over specifications. We demonstrate that our estimates are not due to pre-existing epidemiological trends, surveillance efforts, and/or preventive policies. These estimates highlight the potential of mass gatherings in one region to spill-over into an outbreak of national scale. Relaxations of mass gathering restrictions must therefore be carefully considered, even in the context of low community transmission and enforcement of safe distancing guidelines.

Highlights

  • Mass gathering events have been identified as high-risk environments for community transmission of coronavirus disease 2019 (COVID-19)

  • Results indicate an estimated (i) 70.0% of COVID-19 case counts within Sabah post-state election were attributable to the election’s direct effect; (ii) 64.4% of COVID-19 cases in the rest of Malaysia post-state election were attributable to the election’s spill-over effects

  • We demonstrate that our estimates are not due to pre-existing epidemiological trends, surveillance efforts, and/or preventive policies. These estimates highlight the potential of mass gatherings in one region to spill-over into an outbreak of national scale

Read more

Summary

Introduction

As of 10 November 2020, there have been over 60 million reported cases and over a million fatalities due to coronavirus disease 2019 (COVID-19) worldwide. [1] In the absence of a vaccine, non-pharmaceutical interventions such as lockdown orders, limits to large gatherings, and travel restrictions have been employed to curb the onward spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). [2] these measures have been largely effective in suppressing disease transmission, [3] their inevitable gradual relaxation has led to resurgences of COVID-19 across many regions, including North America, Europe, and Asia. [4,5,6] A key driver of this resurgence is large in-person gatherings, such as religious services and public rallies, where physical distancing between individuals is virtually unavoidable. [4,5,6] A key driver of this resurgence is large in-person gatherings, such as religious services and public rallies, where physical distancing between individuals is virtually unavoidable. These gatherings have been identified as high-risk environments for community transmission, [7] the direct and spill-over effects of these events on COVID-19 case counts have not been explicitly quantified and their broader implications are not known. Surges in COVID-19 case counts were reported in almost all major cities nationwide after a state election was held in Sabah, a large East Malaysian state in northern Borneo that is home to over 3.9 million people [8]. Airlines increased flight frequencies to ferry politicians and voters between Peninsular Malaysia and Sabah. [10] Politicians outside Sabah state were documented travelling into and within Sabah with campaign workers for physical rallies, with a total of 257 rallies or gatherings approved. [11] On the election day itself, over 1.1 million voters turned out to vote physically at designated polling stations (see Table A in S1 Text for summary). [12]

Methods
Results
Discussion
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call