In our study, a metapopulation SEIR model with inspected migration was applied to project the case number in the presence or absence of complete border closure. Complete border closure to travelers can reduce the cumulative COVID-19 case number in Hong Kong by 13.99% and mortality by 13.98%. The results suggested that even in the presence of established local transmission, travel restriction remained an effective measure to reduce the cumulative cases in the recipient region. In order to prevent overloading the isolation facilities in Hong Kong, apart from complete border closure, implementation of other effective public health measures to keep R0 below 1.6 would also be required.
Countries or cities with a high population density and aged population including Hong Kong is at risk of severe outbreak of emerging infectious diseases such as COVID-19 . As the disease is spreading rapidly in multiple continents, many countries implemented border restrictions towards regions with severe outbreak in order to reduce local case number and mortality [20,21,22]. This is particularly important for developing countries with inadequate medical resources to tackle massive local outbreak. However, the WHO advised against utilizing travel restriction as an infection control measure . Furthermore, it is particularly challenging to implement border restriction in certain regions due to political, social and economic reasons. To date, there is inadequate scientific data to support border restriction as a public health measure to limit the scale of local outbreak in the presence of an established local transmission. Using Hong Kong and mainland China as an example, we quantitatively illustrated border restriction is effective in reducing cumulative caseload, mortality and healthcare facility occupancy with a metapopulation SEIR model with inspected migration. It was projected that complete border closure would result in meaningful reduction of cumulative cases (4079 cases at R0 of 2.2), mortality (57 deaths at 1.4% in-patient mortality) and a delay in isolation facility overload in Hong Kong.
It is important to emphasize that in our projection, border closure alone is insufficient to prevent healthcare overload, as measured by isolation facilities occupancy. Effective and targeted public health intervention to slow local transmission and reduce local R0 is needed. It can be in the form of universal usage of surgical mask, contact tracing and quarantine system, or social distancing policies such as school cessation [23, 24]. The outbreak on Princess Diamond Cruise clearly illustrated the limitation in outbreak control by border restriction solely with no public health intervention [25, 26]. In early 2020, a number of passengers on Princess Diamond Cruise were found to have COVID-19. Despite there was no further import of COVID-19 cases onto the cruise after the immediate quarantine, there was still rapid rise in the number of COVID-19 cases on the cruise. It was believed that insufficient on-board personal protective equipment and inadequate social distancing were the causes of the unfortunate event. Of the 3711 individuals on the cruise, 624 of 3011 tested passengers were diagnosed with COVID-19 (16.7%) . Unfortunately, implementation of strict public health measures may not be feasible to combat COVID-19 in many regions. For instance, social distancing may not be feasible due to environmental, economic, cultural or religion reasons. There may be a shortage of trained personnel and facilities for performing contact tracing and quarantine .
In the past 1 year, multiple regions had exponential rise in COVID-19 cases which caused extreme stress to their local health care system [11, 27]. In Wuhan, which was the epicenter of the COVID-19 outbreak in China, severe shortage in isolation facilities has necessitated the urgent construction of multiple temporary hospitals . COVID-19 related mortality in regions with severe outbreak tend to be higher due to relative shortage of medical resources outweigh demand [28, 29]. Advanced life support facilities such as intensive care unit, ventilators, extracorporeal membrane oxygenation (ECMO) machines and anti-viral medications are essential in severe COVID-19 cases but their availability is limited [28, 29]. In addition, COVID-19 also severely hinders other non-COVID-19 related medical services. In Hong Kong, although the total confirmed COVID-19 cases are less than the available isolation facilities at the moment, a significant proportion of other less urgent medical services include elective investigations and surgeries have been suspended to reserve resources for COVID-19 . In less resourceful regions, the effect may even be more pronounced. Although morbidity and mortality caused by such service suspension are not included in the official COVID-19 statistics, the effects cannot be overlooked. Furthermore, uncontrolled local epidemic can cause outbreaks in other regions with close ties . The damage brought by a severe local outbreak of COVID-19 is unbearable. Therefore, it is paramount for governments around the world to prevent or limit scale of local outbreak. As suggested by our projection, border restriction against regions with severe outbreak could reduce local caseload, mortality and isolation facilities occupancy. Furthermore, aggressive and efficient public health measures to reduce local R0 is necessary .
The study finding have important implication on policy making. While the COVID-19 pandemic has not ended and many countries made progress in COVID-19 vaccination, how to fine-tune the border restriction should be based on scientific decision. This can be done by using this metapopulation SEIR model with inspected migration to estimate the risks of loosening the border restrictions between different countries and areas. This could inform a risk-based re-opening of the border for resumption of social and economic activities. Incorporating this metapopulation SEIR model with inspected migration into public health policy making would allow timely and scientific decision on border restriction measure, which is an area of ongoing debate nowadays, when there is an urge for opening the borders to facilitate the resumption of economic activities.
Strength of the model
The spread of infectious disease is closely related to the migration of population between regions [13, 18]. Conventional single-patch SEIR models are not suitable for such analysis. A metapopulation SEIR model with inspected migration was specifically applieded for this purpose. In addition to COVID-19, the developed model can be used to perform projection for other emerging infectious diseases in the future . Furthermore, parameters such as effectiveness of custom inspection were included to improve accuracy of projection. The presented model is also suitable for further analysis of other emerging infectious diseases.
Firstly, interaction was assumed to be well-mixed within patch. The spatial effect in disease transmission within each patch is not directly addressed in the model, which can have a non-trivial effect on the dynamic of infectious disease . Secondly, the proposed model is deterministic in nature which ignores the randomness in migration and in the interactions among people; a stochastic model would be more realistic especially early in the disease [35, 36]. Thirdly, key parameters such as rate of spread are still unclear so we assumed a parametric form of the rate of spread with reference to 2003-SARS . In general, parameter calibration can be performed by some criteria , for example, minimizing residuals sum of square between the historical and fitted infected cases. Meanwhile, missing information, such as travel history across regions, leads to crucial statistical uncertainty. A stochastic metapopulation migration model to explore the corresponding statistical properties with data would be a fruitful direction in the future . While the above shortcomings may be the expected tradeoff between computation time and model simplicity , it will show the core message that border restriction reduces cumulative case, mortality and delay healthcare system exhaustion. Lastly, economic impact is beyond the scope of this study . While full border closure can have a negative impact on the economy, one cannot ignore the negative economic impact from an otherwise preventable major outbreak.