Outcome	Number of studies	Overview of effect by study	Comparison used in each study	Effect direction per study (positive ▲; negative ▼; no change/mixed effects/conflicting findings ◀▶)
Symptom/exposure‐based screening at borders
Outcome category: cases avoided due to the measure
Number or proportion of cases exported	1 modelling study	Wells 2020: Assuming that only 35.7% of symptomatic individuals are detected, the number of cases exported per day from China would reduce by 82% (95% CI 72 to 95) resulting from screening measures put in place across the world, compared with no screening measures.	Measure versus no measure	Positive (▲)
Outcome category: shift in epidemic development
Time to outbreak	4 modelling studies	Clifford 2020a: Entry and exit screening, alone or combined, and measures to increase awareness and encourage appropriate responses would delay an outbreak in a hypothetical population. Assuming a sensitivity of 86%, if introduced at the beginning of an outbreak when very few infected individuals arrive, the measures would delay the outbreak by several days (ranging from 1 to 8 days). If introduced later, when more infected individuals arrive, the measures would do little to delay the outbreak (ranging from less than 1 to 1 day).	Measure versus no measure	Positive (▲)
		Mandal 2020: With Rt = 2.0, entry screening of symptomatic individuals would lead to a delay in reaching 1000 cases (2.7‐day delay, from 45 to 47.7 days) in a hypothetical population compared to no screening. If screening could detect 50% and 90% of asymptomatics the delay would increase to 7.4 and 20 days, respectively. With higher community transmission (Rt = 4.0) these values of sensitivity are all lower.	Measure versus no measure	Positive (▲)
		Nuckchady 2020: Assuming one infected person entered Mauritius per day, entry or exit screening with a sensitivity of 64% would delay an outbreak by 9.7 days, and screening with a sensitivity of 100% by 20 days.	More versus less stringent measure	Positive (▲)
		Wilson 2020: Under the assumption of one flight per day (7.1% of normal travel volume) in a hypothetical disease‐free area (modelled on New Zealand), exit screening alone with 50% sensitivity, would delay an outbreak by 0.5 years, from 1.7 years (95% CI 0.04 to 6.09) to 2.2 years (95% CI 0.6 to 8.11) compared with no screening.	Measure versus no measure	Positive (▲)
Risk of outbreak	1 modelling study	Nuckchady 2020: Assuming one infected person entered Mauritius per 100 days, entry screening with 100% sensitivity would reduce the probability of an outbreak within 3 months to 10% and screening with 50% sensitivity would reduce the probability to 48%.	More versus less stringent measure	Positive (▲)
Outcome category: cases detected due to the measure
Number or proportion of cases detected	4 modelling studies	Bays 2020: Entry screening of all arriving travellers would detect 0.8% of infected travellers in a hypothetical population in a limited exposure scenario (i.e. short‐term stay in country of departure and short flight) and 12% of cases in a higher‐exposure scenario (i.e. longer‐term stay in country of departure and long flight). The effectiveness of entry screening would thus be influenced by the time window in which the exposure may have occurred (i.e. longer windows of exposure mean a higher likelihood that incubation may have occurred prior to departure) as well as the duration of the flight (i.e. longer flights increase the likelihood that symptoms develop during the flight and can thus be detected through screening).	Measure versus no measure	Positive (▲)
		Gostic 2020: With 25% of cases assumed to be subclinical, combined entry and exit screening using thermal scanners and self‐reporting of exposure would detect 27% (95% CI 10 to 47) of cases in a hypothetical population. On their own, exit and entry screening would detect 17% (95% CI 3 to 33) and 20% (95% CI 7 to 40) of cases, respectively. As the proportion of subclinical cases increases the proportion of cases detected goes down; conversely, as the proportion of subclinical cases decreases, the proportion of cases detected goes up.	Measure versus no measure	Positive (▲)
		Quilty 2020: Assuming a sensitivity of 86% for thermal scanner‐based screening and 17% of asymptomatic cases being undetectable, entry and exit screening combined and entry screening alone both detected 53% (95% CI 35 to 72) of cases in a hypothetical population; exit screening alone was comparatively less effective, detecting 44% (95% CI 33 to 56) of cases.	Measure versus no measure	Positive (▲)
		Taylor 2020: Entry screening of all incoming travellers in the UK would lead to the detection of 0.8% (95% CI 0.2 to 1.6) of cases when using thermal imaging scanners and 1.1% (955 CI: 0.4 to 2.1) of cases when using health checks. The proportion of cases detected would be lower when compared with the self‐isolation of all incoming travellers (51.3% for 7 days of self‐isolation; 78% for 14 days of self‐isolation).	Measure versus no measure	Positive (▲)
Test‐based screening at borders
Outcome category: cases avoided due to the measure
Proportion of secondary cases	1 modelling study	Dickens 2020: Compared with no measure targeting incoming travellers in a hypothetical population, testing all incoming travellers upon arrival, followed by the isolation of test‐positives and requiring a negative test at the end of isolation would lead to a reduction in secondary cases of 88% (95% CI 87 to 89) for a 7‐day isolation period and 92% (95% CI 92 to 93) for a 14‐day isolation period.	Measure versus no measure	Positive (▲)
Proportion of imported cases	1 modelling study	Dickens 2020: Compared with no measure targeting incoming travellers in a hypothetical population, testing all incoming travellers upon arrival, followed by the isolation of test‐positives and requiring a negative test at the end of isolation would lead to a reduction of 90% of imported cases for a 7‐day isolation period and 92% for a 14‐day isolation period. Testing all incoming travellers and refusing entry to test positives led to a reduction of 77%.	Measure versus no measure	Positive (▲)
Outcome category: shift in epidemic development
No contributing study
Outcome category: cases detected due to the measure
Days at risk of transmission	2 modelling studies	Clifford 2020b: Requiring a single PCR test upon arrival to the UK from EU countries would have led to 2.0 days at risk of transmission (95% CI 0 to 10.8). This is shorter than the days at risk of transmission for symptom/exposure‐based entry screening alone (2.1 days at risk (95% CI 0 to 11.2)). Requiring an additional pre‐flight test would slightly improve the effect of the PCR test upon arrival.	Measure versus alternative measure	Positive (▲)
		Russell WA 2020: Requiring all incoming travellers to test upon arrival in a hypothetical population would have led to 2.3 days at risk of transmission (95% CI 2.1 to 2.6). This is shorter than the days at risk of transmission for no measure at entry (2.6 days at risk, (95% CI 2.3 to 2.9)).	Measure versus no measure	Positive (▲)
Probability of releasing an infected individual into the community	2 modelling studies (Clifford 2020b, Steyn 2020)	Clifford 2020b: Requiring a single PCR test upon arrival to the UK from EU countries would reduce the risk of releasing an infected individual into the community compared with symptom/exposure‐based entry screening alone (RR: 0.55, 95% CI 0.28 to 0.83). Requiring an additional pre‐flight test would slightly improve the effectiveness of the PCR test upon arrival.	Measure versus alternative measure	Positive (▲)
		Steyn 2020: The probabilities of releasing an infected individual as a result of testing at departure and upon arrival in New Zealand were 48%, 50%, and 53% for scenarios assuming no, moderate, and high risk of transmission while travelling, respectively. These were higher compared with the probability of releasing an infected individual following a 14‐day quarantine of all incoming travellers.	Measure versus no measure	Positive (▲)