Effect of internationally imported cases on internal spread of COVID-19: a mathematical modelling study

Abstract

Background: Countries have restricted international arrivals to delay the spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). These measures carry a high economic and social cost, and might have little effect on COVID-19 epidemics if there are many more cases resulting from local transmission compared with imported cases. Our study aims to investigate the extent to which imported cases contribute to local transmission under different epidemic conditions.

Methods: To inform decisions about international travel restrictions, we calculated the ratio of expected COVID-19 cases from international travel (assuming no travel restrictions) to expected cases arising from internal spread, expressed as a proportion, on an average day in May and September, 2020, in each country. COVID-19 prevalence and incidence were estimated using a modelling framework that adjusts reported cases for under-ascertainment and asymptomatic infections. We considered different travel scenarios for May and September, 2020: an upper bound with estimated travel volumes at the same levels as May and September, 2019, and a lower bound with estimated travel volumes adjusted downwards according to expected reductions in May and September, 2020. Results were interpreted in the context of local epidemic growth rates.

Findings: In May, 2020, imported cases are likely to have accounted for a high proportion of total incidence in many countries, contributing more than 10% of total incidence in 102 (95% credible interval 63-129) of 136 countries when assuming no reduction in travel volumes (ie, with 2019 travel volumes) and in 74 countries (33-114) when assuming estimated 2020 travel volumes. Imported cases in September, 2020, would have accounted for no more than 10% of total incidence in 106 (50-140) of 162 countries and less than 1% in 21 countries (4-71) when assuming no reductions in travel volumes. With estimated 2020 travel volumes, imported cases in September, 2020, accounted for no more than 10% of total incidence in 125 countries (65-162) and less than 1% in 44 countries (8-97). Of these 44 countries, 22 (2-61) had epidemic growth rates far from the tipping point of exponential growth, making them the least likely to benefit from travel restrictions.

Interpretation: Countries can expect travellers infected with SARS-CoV-2 to arrive in the absence of travel restrictions. Although such restrictions probably contribute to epidemic control in many countries, in others, imported cases are likely to contribute little to local COVID-19 epidemics. Stringent travel restrictions might have little impact on epidemic dynamics except in countries with low COVID-19 incidence and large numbers of arrivals from other countries, or where epidemics are close to tipping points for exponential growth. Countries should consider local COVID-19 incidence, local epidemic growth, and travel volumes before implementing such restrictions.

Funding: Wellcome Trust, UK Foreign, Commonwealth & Development Office, European Commission, National Institute for Health Research, Medical Research Council, and Bill & Melinda Gates Foundation.

Figure 1

Our methodology: the modelling procedure and our policy recommendations A descriptive schematic of the modelling process used to arrive at the overall risk ratios is shown (A), with a flow-chart outlining the necessary conditions that determine our policy recommendations (B). R_t=reproduction number.

Figure 2

Risk rating by country, in the absence of international travel restrictions, in each of the four scenarios about international travellers in May and September, 2020 (A) Travel assumed to be at the same levels as in May, 2019. (B) Travel estimated to be at the same levels as in May, 2020, by downscaling the May, 2019, levels by flight path-specific reduction factors calculated by using the ratio of the OpenSky datasets for May, 2019, and May, 2020. (C) Travel assumed to be at the same levels as in September, 2019. (D) Travel estimated to be at the same levels as in September, 2020, by downscaling the September, 2019, levels by flight path-specific reduction factors calculated using the ratio of the OpenSky datasets for September, 2019, and September, 2020. See appendix for the full list of results (p 22) and for the lower and upper 95% credible intervals of our results (pp 17–18).

Figure 3

Percentage reduction in passenger numbers required for countries to bring proportion of total incidence due to imported cases to less than 1% Figure shows 118 countries in scenario D where imported cases account for at least 1% of total incidence. Error bars represent the lower and upper 95% credible intervals of our expected number of imported cases estimates. Countries are grouped by UN region. *Geographically part of Asia.