Passengers are screened on both departure (exit screening) and entry (arrival screening). We assume that screening is unable to detect infected individuals while they are asymptomatic. That is, during the first two days after being infected. By screening sensitivity we mean the probability that border screening will identify an arriving traveler who is symptomatic. In our calculations we varied the sensitivity of screening from 0 to 1 (perfect symptomatic screening). At the departure border, the probability that a symptomatic individual is detected is sD, which is zero when there is no border control or isolation of cases in place in the source region. Upon arrival, the probability that a symptomatic individual is detected is sA.
The results are such that it suffices to illustrate them for the two scenarios sD=sA=0 (zero, or completely ineffective, screening) and sD=sA=1 (100% sensitive screening for symptomatic travelers). Symptomatic border screening, even of a very high sensitivity on departure and arrival, has a negligible effect upon the delay until a local epidemic gathers momentum. For example, for a scenario of R0=2.5 and 100 travelers per day attempting to depart the source region, the effect of implementing 100% sensitive screening on both departure and arrival only increases the median delay from 27 to 28 days (Figure 3.4).

The effect of perfect symptomatic border screening (100% sensitivity) on the distribution of the time delay until an epidemic reaches 20 concurrent cases in Australia following identification in the source country.  Calculations assume a peaked infectious

    Figure 3.4 The effect of perfect symptomatic border screening (100% sensitivity) on the distribution of the time delay until an epidemic reaches 20 concurrent cases in Australia following identification in the source country (assumed to occur when there are 10 concurrent cases in infected areas).

In order to find a setting where screening does appreciably increase the delay we have looked at very low values of R0. However, even under the rather benign scenario of R0=1.2 and 100 travelers/day attempting to depart the infected source country, symptomatic screening only increases the delay to an epidemic gathering momentum from 140 to 148 days (Figure 3.5).



    Figure 3.5 The effect of perfect symptomatic border screening (100% sensitivity) on the distribution of the time delay until an epidemic reaches 20 concurrent cases in Australia following identification in the source country. Calculations assume a peaked infectiousness function and no other interventions

The failure of symptomatic border screening, even of perfect sensitivity, to reliably prevent epidemic initiation arises from the two day incubation period acting in concert with short travel times. For example, assuming R0=1.5, the probability of a randomly selected infected traveler disembarking in Australia undetected after a 12 hour journey is 0.29 if screening is 100% sensitive. Increasing R in the source country increases the probability of escaping detection, as it is more likely that an infected traveler will have been recently infected and be asymptomatic. For example, with R0 = 3.5 (and other conditions as above) the probability of an infected traveler evading screening increases to 0.48.

The curves in Figures 3.1, 3.4 and 3.5 illustrate that the delay decreases substantially as R increases, particularly in the range 1 < R < 2. Specifically, the median delays in the absence of screening, assuming 100 travelers per day from the infected source region, are given by

R
1.2
1.5
2.5
3.5
Median delay (days)
140
65
27
19

This highlights again the benefit of seeking to make R as low as possible in the source region, and to quantify the transmission parameters of the newly emerged viral strain as early as possible.

Document download

This publication is available as a downloadable document.

Using Mathematical Models to Assess Responses to an Outbreak of an Emerged Viral Respiratory Disease(PDF 873 KB)