Background

In response to the coronavirus disease 2019 (COVID-19) pandemic that led to a substantial epidemiological burden, countries worldwide adopted considerably different intervention approaches. In particular, the strictness of border control policies varied considerably by region and country, and ranged from total border closure and quarantine of all incoming travelers to only a brief screening of travelers [1,2,3]. As population-wide vaccination progressed, interference with the freedom of movement gradually ceased or was occasionally coupled with the request for vaccine certificates [4,5,6]. However, the remaining border control measures were maintained throughout the pandemic until at least early 2023, particularly in countries in the Western Pacific region [7].

Screening travelers at international borders is a less intrusive border control measure than travel bans, regardless of the means of transportation (i.e., air, sea, or land) [8]. Screening is divided into exit screening, which occurs at the origin, and entry screening, which occurs at the destination. In Japan and many other countries, exit screening for COVID-19 required passengers to submit a certificate for a negative real-time reverse transcription-polymerase chain reaction (RT-PCR) or antigen test result carried out within 72 h before the departure time [9, 10]. At the beginning of the pandemic, entry screening primarily consisted of thermal scanning [11]. However, entry screening was soon substantially strengthened by combining thermal scanning with symptom checks on arrival and RT-PCR or antigen testing, either immediately on arrival or within a few days after arrival [12, 13].

Although many countries imposed mandatory exit screening, little is known about the quantitative performance of such screening. In a limited number of studies, researchers used epidemiological or simulation-based approaches to evaluate the effectiveness of exit screening [11, 14,15,16,17,18]. The results of a simulation study indicated that the population-level transmission risk from importation decreased when anterior nasal polymerase chain reaction (PCR) testing occurred within 3 days before departure [14]. Similarly, another study indicated that a 10–29% decrease in transmission risk was observed when testing 3 days before travel was implemented [15], whereas testing on the day of travel resulted in a more significant reduction of 44–72%. From March 20 to September 3, 2022, pre-departure testing was associated with a 52% reduction in post-arrival severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) positivity in the United States [16], and the strategy of conducting rapid antigen testing immediately before departure was shown to be associated with reducing the proportion positive by 37.4–46.7% on arrival [17]. A United Kingdom (UK) study showed that pre-flight testing reduced the number of infectious travelers entering the community, with a relative risk of 0.69 compared with the risk of no testing either before departure or after arrival [18]. However, few studies inferred effectiveness from actual exit screening data; the shortage of such empirical evidence is attributable to the difficulty in connecting exit screening at the origin to the identification of infections at the destination.

In this context, Japan is one of the countries that implemented several border control measures, primarily at three designated international airports, that is, Tokyo-Narita, Tokyo-Haneda, and Kansai, which were then gradually expanded to other airports. From March 2020 to December 2022, all foreign visitors were required to enter the country at one of these airports, and all incoming passengers, including Japanese citizens, were subject to testing (mostly using real-time RT-PCR and antigen testing from late July 2020) as part of entry screening on arrival. Moreover, to fly to Japan, all foreign visitors were required to submit a negative test certificate as part of exit screening. Japanese nationals were exempt from exit screening until late 2020; however, given the high positivity rate during entry screening, submitting a certificate for a negative test result also became essential in January 2021 for Japanese nationals [9]. Such an abrupt policy change offered the opportunity to explicitly estimate the effectiveness of exit screening. In contrast to some nations that experienced a prolonged (i.e., for several consecutive weeks) absence of visitors, a consistent influx of travelers from the UK entered Japan during this period [19]. Moreover, a country-wide prevalence survey that was jointly conducted by the Office of National Statistics (ONS) in the UK and university researchers from April 2020 yielded a gold standard estimate of the prevalence of infection in the UK throughout the pandemic [20, 21]. The ONS infection survey results can be assumed to mirror the risk of infection among UK residents, including those flying to Japan over time.

We evaluated the effectiveness of the exit screening policy imposed by the Japanese government on passengers arriving from other countries by analyzing the airport screening data of passengers entering Japan from the UK between October 2020 and April 2022.

Methods

Brief description of the border control policy in Japan

From September 1, 2020, (Table 1) to enter Japan, foreign nationals were required to undergo exit screening, which involved presenting a certificate of a negative RT-PCR or quantitative antigen test result obtained within 72 h before departure. The sample collection methods were nasopharyngeal swabs, saliva, nasopharyngeal and oropharyngeal swabs, or nasal swabs. Japanese nationals were subjected to the same requirement from January 13, 2021 (Fig. 1A). Japanese nationals were additionally mandated to undergo exit screening given the emergence of the Alpha variant (B.1.1.7), which had higher transmissibility than the previous variant [13]. Entry screening of arriving passengers was conducted continuously from May 16, 2020 [22]. In addition to the screening policy, changes to entry and residence permit rules were made that corresponded with the dynamically changing number and infection risk among passengers entering Japan. For instance, new entry was suspended for an extended period of time during the pre-vaccination period (Table 1).

Fig. 1
figure 1

Entry screening data for English and Japanese passengers at the Japanese border overlaid with the estimated prevalence of SARS-CoV-2 in the United Kingdom from the Office of National Statistics infection survey, 2020–2022. (A) The orange line represents weekly positivity based on real-time reverse transcription-polymerase chain reaction testing of all incoming English passengers entering Japan from the UK. The blue line represents the positivity of Japanese passengers arriving from the UK, whereas the red line indicates the weekly estimated prevalence of coronavirus disease 2019 from the ONS infection survey. Whiskers denote the 95% CI computed using the Wilson score method for each observation point, with corresponding colors for clarity. Travel policies in Japan are indicated on the chart. The three-color bar at the bottom represents the level of dominance of the Alpha, Delta, and Omicron variants of the severe acute respiratory syndrome coronavirus in the UK. (B) Weekly COVID-19 vaccination coverage in Japan for the second and third (booster) doses. Light green indicates the second-dose coverage from May 2021 and dark green represents the third-dose coverage, which began in December 2021. (C) Weekly COVID-19 vaccination coverage in the UK for the second dose and the third (booster) dose. Light green reflects the second-dose coverage from January 2021 and dark green represents the third-dose coverage, which began in September 2021

Full size image

Once the rollout of primary series vaccination became widespread, Japanese government launched a vaccination passport policy. Whether a passenger had a vaccination passport did not influence the individual’s entry to Japan, but did help to relax the duration of quarantine.

Table 1 Airport screening policy and border control measures from September 2020 to March 2022 in Japan
Full size table

Epidemiological dataset

Japan allowed international arrivals and departures mostly via three international airports—Narita, Haneda, and Kansai—during most of the course of the COVID-19 pandemic (i.e., from 2020 to 2022). All passengers arriving in Japan regardless of nationality were subjected to RT-PCR testing or antigen testing on disembarkation and before immigration inquiry. During testing, all passengers were asked to immediately provide sputum in a tube container and self-collected saliva samples were sent to a testing center located in each international airport. All passengers were asked to remain in a waiting space in the airport until their test results were available. Approximately 1–2 h or more were required to obtain results, depending on the volume of passengers arriving. Travelers who tested negative were allowed to proceed to the immigration gate. If positive, travelers were guided to a quarantine station (i.e., hotel facilities or hospitals) to self-isolate. In Japan, movement restrictions were enforced on passengers seated on the same plane within two rows in front of and two rows behind the positive passenger, regardless of the laboratory testing results.

In this study, we used the aforementioned entry screening results in Japan that were reported by the Ministry of Health, Labour and Welfare for the period from October 4, 2020 to April 23, 2022 [19]. The total number of tests and number of positive tests were recorded weekly by nationality (i.e., for Japanese and other passengers). However, in the entry screening results for flights from the UK to Japan, detailed information regarding flight connections and the specific nationality of each traveler beyond Japanese nationals was unavailable. Therefore, we had to impose an assumption that “other” travelers mostly included people who held a UK passport.

In addition to examining airport screening data, we investigated the prevalence of SARS-CoV-2 infection in the UK. The estimated prevalence data were retrieved from the Coronavirus (COVID-19) Infection Survey, which is openly available on the ONS website [20]. Weekly prevalence in England was used to represent the proportion positive among both Japanese and English visitors flying to Japan because the dataset for Northern Ireland and Scotland was available only as aggregated biweekly data in October 2020. Information on vaccination coverage for Japan and the UK was derived from the Japanese vaccination record system [23] and the UK Health Security Agency [24], respectively.

Statistical estimation of exit screening effectiveness

The epidemiological dataset contained weekly records of administered RT-PCR tests and the number of confirmed SARS-CoV-2-positive cases on arrival. These records allowed the calculation of the positivity rates for both English and Japanese passengers traveling from the UK. To derive the prevalence of SARS-CoV-2 infection in the UK, we divided the weekly ONS estimates of the average number of individuals who tested positive for SARS-CoV-2 in England by the total population of England estimated by the ONS for the year 2021. As the number of healthy samples in the ONS survey was maintained in the UK, we assumed that the testing frequency in England remained sufficiently high over the course of time. We regarded the observed traveler population at airports as if it were a random sample from the general population; hence, travel took place statistically independent of infection. This allowed us to impose the essential assumption that the observed positivity rate in Japan’s airports was proportional to the prevalence of SARS-CoV-2 infection in the UK, x(t) in week t, where t represents the number of weeks that elapsed from October 4, 2020 (thus, Week 13 contained December 27, 2020, when Japanese travelers from the UK were mandated for exit screening). Moreover, we introduced a constant scaling factor \(\:\alpha\:\), which corresponds to the proportionality of airport positivity in Japan to the ONS prevalence data, and another relative positivity parameter σ for Japanese travelers compared with English travelers. We modeled positivity p(t) in week t among Japanese travelers arriving from the UK as

$$p(t) = \left\{ {\begin{array}{*{20}{l}}{\alpha \sigma x(t),}&{(t\leq12;4Oct - 26Dec2020)} \\ {(1 - \in )\alpha \sigma x(t),}&{(13\leq t\leq30;27Dec2020 - 1May2021)} \\ {(1 - \in )\alpha \sigma \delta (1 - \beta {y_{j2}}(t))x(t),}&{(31\leq t\leq61;2May - 4Dec2021)} \\ \begin{gathered}{\text{ }}(1 - \in )\alpha \sigma \delta \omega \hfill \\\quad \left\{ {1 - {\beta _3}[{y_{j2}}(t) + {y_{j3}}(t)(1 - {\beta _3})]} \right\} \hfill \\\quad x(t), \hfill \\ \end{gathered} &{(62\leq t\leq81;5Dec2021 - 23Apr2022)} \end{array}} \right.{\text{ }}$$
(1)

where \(\in\) represents the effectiveness of the exit screening. Hence, we multiplied the relative reduction (\(\:(1-\in)\:\)by the observed positivity from Week 13. We assumed that the proportion positive on entry was elevated by a constant factor \(\:\delta\:\) when the Delta variant (B.1.617.2) began to spread in the UK in early May 2021 (Week 31). At approximately the same time, Japan started its vaccination campaign for the second dose. \(\:{y}_{j2}\left(t\right)\) and \(\:{y}_{j3}\left(t\right)\) represent the time-dependent fractions of the second-dose and third-dose (booster) vaccination coverage, respectively, among the Japanese, as shown in Fig. 1B, and \(\:\beta\:\) represents the effect of vaccination on positivity among travelers, which is not identical to the biological concept of “vaccine efficacy” or the epidemiological term “vaccine effectiveness,” which are unique causal impacts. Our vaccine effect represents the vaccine-induced relative changes in positivity among travelers, which is prone to various biases (e.g., vaccinated people are more likely to be mobile and have a greater number of contacts than unvaccinated people). We measured the vaccination effect during the Delta variant dominant period by multiplying \(\:\beta\:\) by the time-dependent coverage of the second dose. The Omicron variant (B.1.1.529) began to spread in December 2021 (Week 62) [25] and passengers became more likely to be positive than before on entry. We described the elevated relative proportion by a constant factor \(\:\omega\:\). As the vaccination effect is considered to have been altered by entry into the Omicron variant dominant period, we introduced another parameter \(\:{\beta\:}_{3}\) and multiplied it by the third-dose (booster) vaccination coverage. We modeled positivity b(t) among English passengers arriving at the airport from the UK as

$$b(t) = \left\{ {\begin{array}{*{20}{l}}{(1 - \in )\alpha x(t),}&{(t\leq14;4Oct2020 - 9Jan2021)} \\ {(1 - \in )\alpha (1 - \beta {y_{e2}}(t))x(t),}&{(15\leq t\leq30;10{\mkern 1mu} Jan - 1May2021)} \\ {(1 - \in )\alpha \delta (1 - \beta {y_{e2}}(t))x(t),}&{(31\leq t\leq52;2May - 2Oct2021)} \\ \begin{gathered}(1 - \in )\alpha \delta \left\{ {1 - \beta [{y_{e2}}(t) + {y_{e3}}(t)} \right. \hfill \\\left. {\quad\quad\quad\quad\quad(1 - \beta )]} \right\}x(t), \hfill \\ \end{gathered} &{(53\leq t\leq61;3Oct - 4Dec2021)} \\ \begin{gathered}(1 - \in )\alpha \delta \omega \tau \left\{ {1 - {\beta _3}[{y_{e2}}(t) + {y_{e3}}(t)} \right. \hfill \\\quad\quad\quad\quad\quad\quad\left. {(1 - {\beta _3})]} \right\}x(t), \hfill \\ \end{gathered} &{(62\leq t\leq81;5Dec2021 - 23Apr2022)} \end{array}} \right.$$
(2)

Exit screening among non-Japanese travelers began on September 1, 2020, which was earlier than the obtainable data period. Given that exit testing occurred in the same country (the UK), we also represented the effectiveness of exit screening among English passengers by \(\:\epsilon\:\). The time-dependent second-dose vaccination coverage \(\:{y}_{e2}\left(t\right)\) in the UK began to increase from the week of January 10, 2021, as shown in Fig. 1C, which reflects the national vaccination program that began in December 2020, approximately 3–4 months earlier than in Japan. The third-dose vaccination rollout in the UK \(\:{y}_{e3}\left(t\right)\:\)started in September 2021 (Fig. 1C), approximately 2–3 months earlier than in Japan [26,27,28]. Equations (1) and (2) did not account for the leaky or partial effect of vaccination. Additionally, although the vaccination effect of our model does not purely measure the causal impact of vaccination, the formulation of beta with the third dose is in line with the all-or-nothing assumption. When the Omicron variant became widespread, the UK government decided not to impose population-wide restrictions to suppress transmission [29], and on February 21, 2022, the UK government explicitly announced the relaxation of its suppression policy [30]. Rapidly growing risk was recognized in Japan, and the Japanese government decided to prohibit new entry again from December 2021 (new entry was temporarily allowed for a few months prior to the closure). Because of this change, the characteristics of foreign passengers arriving in Japan during the Omicron variant period may have changed, whereas Japanese passengers remained unchanged, and we assumed that the change was captured by a constant factor 𝜏. During the pandemic, Japanese passengers were primarily business travelers who had brief stays in the UK. Thus, we assumed that vaccination coverage in Japan captured coverage among Japanese travelers.

We assumed that a positive entry screening result was a binomial sampling process for each week and used the maximum likelihood method to obtain parameter estimates that governed Eqs. (1) and (2). We considered the likelihood that used Eq. (1) and the likelihood that used Eq. (2) to be conditionally independent, and obtained the total log-likelihood as the summation of these likelihoods. We determined the 95% confidence intervals (CIs) of the parameters using the profile likelihood and calculated the 95% CI of the passenger positivity rate using the Wilson score method. We calculated the 95% CIs of the model prediction using the bootstrap method.

The final models are presented as Eqs. (1) and (2). We compared Akaike’s information criterion with those of other models to identify the best-fit explanatory model. Specifically, we modeled prevalence in an alternative manner by considering the impact of the vaccination passport policy on both Japanese and English travelers (i.e., we multiplied the risk by an additional constant parameter \(\:c\) during week \(\:53\le\:t\le\:61\), i.e., October 3 to December 4, 2021).

Results

Figure 1A shows COVID-19 positivity over time among English and Japanese passengers overlaid with data for the prevalence of SARS-CoV-2 infection in England that we obtained from the ONS infection survey data. From October 4, 2020 to April 23, 2022, 36,484 Japanese nationals and 14,298 English nationals who departed from the UK underwent entry screening. Of these, 449 Japanese and 114 English nationals tested positive on entry screening. The overall positivity was 1.23% (95% CI: 1.12, 1.35) among Japanese nationals and 0.80% (95% CI: 0.66, 0.96) among English nationals.

Exit screening became mandatory for Japanese travelers on December 27, 2020. The average proportion positive among Japanese passengers during the month before mandatory screening was 0.67% (95% CI: 0.45, 0.98), whereas the proportion decreased to 0.49% (95% CI: 0.21, 1.15) in the first month of exit screening, which yielded a crude absolute risk reduction of 0.18% points and risk ratio of 0.74 (95% CI: 0.29, 1.92). The data we obtained only covered the period after English nationals underwent exit screening, which made it impossible to estimate a similar risk ratio among English nationals.

Figure 2 illustrates the predicted and observed positivity on entry screening using Eqs. (1) and (2). Japanese and English travelers had a similar time-related pattern of positivity. The ONS data indicated that airport screening results among English travelers demonstrated prevalence that was 0.38 times (95% CI: 0.20, 0.71) the ONS prevalence; and positivity among Japanese travelers was 1.43 times (95% CI: 0.84, 2.49) that among English travelers. However, when the Delta variant emerged, overall positivity among both Japanese and English was 3.46 times (95% CI: 1.72, 7.50) that during the pre-Delta variant period. Positivity during the Omicron variant period was 1.46 (95% CI: 0.93, 2.30) times that during the pre-Delta variant period. When the Omicron variant was widespread and new foreign entrants were refused entry into Japan again, positivity among English passengers became 1.93 times (95% CI: 1.07, 3.55) that among Japanese travelers. The estimated effectiveness of exit screening was 59.3% (95% CI: 19.6, 81.3) for the entire study period. The results showed that vaccination contributed to reducing infections. The estimated coefficient for the vaccine effect, which we multiplied by vaccination coverage, was 1.00 (95% CI: 0.69, 1.16) before the circulation of the Omicron variant. We considered that the primary series did not significantly contribute to lowering risk among passengers arriving in Japan. When Omicron emerged and was widespread, the estimated coefficient was 0.72 (95% CI: 0.52, 0.87). The effect of the vaccine was more efficient during the Delta variant period than the Omicron variant period.

Fig. 2
figure 2

Comparison of observed and predicted positivity of entry screening in Japan, 2020–2022. A. The line shows the predicted proportion of positive tests among Japanese passengers overlaid with dots representing the observed entry screening positivity over time. The shaded light green area represents the 95% CI (bootstrap method) of predicted positivity. B. Similarly, the line illustrates the predicted proportion of positive tests among English passengers overlaid with observed entry screening positivity over time. The shaded light yellow area represents the 95% CI (bootstrap method) of predicted positivity

Full size image

Table 2 compares the results of the main model and another model using Akaike’s information criterion. The best-fit model is described above. Model No. 2 indicated that the vaccination passport policy did not have a significant risk-reducing effect on positivity and the estimated coefficient was 1.18 (95% CI: 0.53, 2.48). As indicated by the best-fit model (Model No. 1), relying solely on the vaccination effect during the Delta and Omicron variant periods, along with exit screening and changes associated with the variant, was sufficient for describing the time-varying proportion positive among passengers arriving in Japan. Supplementary Table S1 shows the parameter estimates.

Table 2 Comparison of models for positivity on entry screening among travelers to Japan
Full size table

Model 1 is the baseline model described as Eqs. (1) and (2). Parameter c represents the effectiveness of the vaccination passport policy (i.e., an additional parameter was multiplied for week 53 ≤ t ≤ 61).

Discussion

In this study, we evaluated the effectiveness of the exit screening policy in the UK using entry screening data in Japan. We demonstrated a consistent time-dependent pattern between the entry screening data and the ONS data from the infection survey conducted in the UK. Exit screening was underway when the prevalence of the ONS data considerably surpassed that observed in entry screening, which indicates a substantial risk reduction from exit screening. Given the inconsistent time patterns, we used a statistical model to describe prevalence at entry screening that accounted for the estimated 59.3% (95% CI: 19.6, 81.3) effectiveness of exit screening. Our model estimates captured the observed positivity pattern from October 4, 2020 to April 23, 2022, which encompassed the period in which the Alpha, Delta, and Omicron variants were dominant.

This is not the first study in which researchers quantified the effectiveness of exit screening for COVID-19; however, it is the first in which researchers evaluated the effectiveness of such screening by focusing on the UK and exploiting the long-term dynamics of COVID-19 in both the origin country (using ONS infection survey data in the UK) and the destination country (using universal testing data gathered on entry to Japan). Using a simplistic model, we showed that exit screening contributed to reducing passenger positivity by 59.3%. Despite this, the Delta and Omicron variants elevated prevalence among travelers by 3.46 times and 1.46 times that during the pre-Delta variant period. The latter result may reflect the well-known capacity of the Omicron variant to escape existing immunity in the population [31,32,33] and its short mean generation time of 2–3 days [34]. Clearly, we ignored the small time gap for the replacement of the variant and our model remained very simplistic, with instant replacement. Our model still successfully demonstrated that exit screening within 72 h before departure during the dominance of the Omicron variant was not very effective in reducing infection among travelers.

Positivity among passengers never exceeded the prevalence of infection in the ONS data over time. This finding may reflect the lower overall risk propensity of travelers compared with the general population. Kucharski et al. [35] showed that if testing was conducted both at departure and arrival, passengers arriving with infections represented only a subset of individuals who may have been infected and attempted to travel. In our study, we used a scaling factor α to adjust for the difference in the overall gap between the ONS prevalence data and positivity in Japan. Despite a lowered positivity for individual travelers, overall positivity was kept high as the number of travelers increased, as was the case during the period in which the Omicron variant was dominant.

Exit screening alone was the only factor in our model that explained a consistent reduction in the proportion positive throughout the study. Before our study, published findings explored the optimal timing for implementing RT-PCR or antigen testing as part of exit screening. According to a modeling study by Kiang et al. [14], the most effective approach for reducing onward transmission is to test within 3 days before departure, followed by a post-quarantine test on arrival. In another modeling study, Johansson et al. [15] showed that testing a day before arrival resulted in a 17–35% risk reduction, whereas testing 3 days before arrival yielded a 5–13% reduction in the risk of introduction. Using real-world data, Clifford et al. [18] found that testing a day before travel is more effective in reducing post-arrival transmission compared with testing 4–7 days before travel. Similarly, Chad et al. [17] observed that delays of up to 72 h in obtaining RT-PCR test results led to a high probability (39.2%) of onward transmission, with a modest 4.5% reduction in expected post-arrival transmission. Bart et al. [16] showed that post-arrival test results were 52% less likely to be positive when pre-departure COVID-19 testing occurred within a day. In this study, the recommended timing for testing was 72 h before the departure time; this policy was associated with a 59.3% reduced post-arrival proportion positive among passengers from the UK. The effectiveness was diminished with the Omicron variant and when society was about to be reopened in the UK [36,37,38]. However, if the time of testing was closer to departure (e.g., 24 h before departure), the published evidence suggests that the positive impact of exit screening may have been observed from 2022.

As a side finding, the vaccination passport policy—the confirmation of vaccination status that facilitated international business and travel—had a limited impact. Despite the ongoing debate on the ethical and legal issues regarding this subject [39, 40], our analysis suggests that the impact of this policy was relatively restricted. Although vaccination passports can have an additional impact (e.g., enhancing vaccine acceptance [41, 42]), we focused our evaluation to the context of the proportion positive among travelers and examined the effectiveness of the passport policy in reducing the positivity during international travel. Given that the policy facilitated increased international travel, even among infected individuals, it is plausible that the estimated effectiveness in positivity reduction is small or absent. Future studies that evaluate this policy require a set of evaluations that rely on multidisciplinary views and approaches.

Five technical limitations of the study should be acknowledged. First, we did not consider the duration of travel, purpose of travel, flight connections, and passport types of Japanese and English passengers, potentially overlooking critical insights into travel patterns and risks associated with their travel. Second, we did not collect the vaccination history on an individual basis (i.e., we did not capture the vaccination history of negative test entry); thus, we were unable to fully address the link between vaccination and the risk of infection. Third, we solely relied on time-dependent patterns of positivity and ignored the individual heterogeneity of the risk of illness onset (e.g., age, obesity, and underlying comorbidities). Fourth, in addition to analyzing positivity, Japan conducted genome sequencing using samples that were detected at the airport. However, the sequencing results (e.g., in cases infected with variants of concern) were not tied with positivity data at the individual level. Fifth, we acknowledge that the age distributions of infections in the UK may differ from that of travelers, and our assumption of proportionality of entry screening data to ONS prevalence data may be biased. However, even though the overall positivity may be biased, we assumed that the proportionality of time-dependent changes between airport screening data and ONS data held over the course of time, which allowed an estimation of the exit screening impact.

Conclusions

Despite the aforementioned limitations, we believe that we successfully estimated the effectiveness of exit screening in the UK using a simplistic statistical model. Although the proportion positive later increased considerably and precluded preventing the introduction of imported cases, submitting a certificate of negative test result decidedly contributed to reducing positivity among travelers.