Table 3.
Control measures used and their estimated efficacy in preventing SARS-CoV-2 transmission.
| Control measure | Value/s used | Comment |
|---|---|---|
| Pre-flight testing using saliva (PCR) of travellers in the 24 h before departure | 62.3% sensitivity | For test sensitivity we used a meta-analysis that gave a sensitivity for saliva testing (PCR) at 62.3% (95%CI: 54.5%–69.6%). This was less than for nasopharyngeal aspirate/swab and throat swab (73.3%, 95%CI: 68.1%–78.0%); and for sputum (97.2%, 95%CI: 90.3%–99.7%)21. We note however, that the sensitivity we used here for saliva testing may be an underestimate given other work showing equivalence between nasopharyngeal and saliva based PCR tests22* |
| Facility-based quarantine in NZ for travellers (current practice as per March 2021 in NZ) | 7, 14 and 21 days | We ran the simulations for three different lengths of quarantine, including 14 days as used in NZ, and longer as in some other settings (e.g., 21 days in China). We assumed a high quality quarantine process where there was no cross-infection within the quarantine facility to facility workers or to other travellers in quarantine. But we assumed no additional PCR testing within quarantine. In reality, NZ combines the 14-day quarantine process with PCR testing (nasopharyngeal swab) upon arrival (“day 1”) (travellers from the majority of countries), and days 3 and 12 (all travellers). This process further helps reduce the risk by allowing for infectious individuals to be put into isolation and so reduce the risk of infecting others who are also in quarantine. But these additional benefits from testing are probably outweighed by the limitations in NZ’s processes that have resulted in various failures of the quarantine/isolation facility system that utilises converted hotels23 |
| Testing instead of facility-based quarantine: PCR test for SARS-CoV-2 at various times | The time course of sensitivity values from Kucirka et al. was used |
We used the results of a study24 which fitted a Bayesian hierarchical logistic regression model for test sensitivity. This meant for example, at day 4 after infection, 67% of test results were false negatives. This decreased to 20% on day 8 and then increased after this e.g., up to 66% on day 21. For cases who already recovered before their PCR test, we use the final value reported by Kucirka et al. (i.e., 34% sensitivity). In the days after arrival and before the next PCR test, we assume that people act normally and so can potentially spread infection to the NZ public (albeit with mask use when with other people as per the details below). For more details, see Supplementary Information In the absence of relevant data, we had to assume test result independence i.e., a false negative for a test was not correlated with a false negative for a later test. If both results were negative, we assumed no further follow-up. We considered a wide range of different timing options for PCR tests after arrival in NZ (see the “Results”) |
| Contact tracing if (i) a scheduled PCR test is positive or (ii) if people develop symptoms and seek medical attention (see below) | 80% of infected contacts are traced and isolated within 48 h | We used performance data for the cluster of cases in Auckland in August 2020 where the official estimate was 80% of contacts contacted within 48 h (as reported by the Prime Minister)25. We divided this into 60% within the first 24 h and 20% in the next 24 h. Of note is that variable performance for contact tracing has been reported for NZ at other times in August 2020, with 86% of contacts traced in 48 h at one point26* |
| The proportion of infected travellers who when they develop any symptoms seek medical attention (i.e., they are in the 60% who will ever develop symptoms) | 50% (self-reporting occurs on average 1 day after symptom onset) |
We assumed that this proportion is somewhat higher than that for the general community (see below) on the assumption that these travellers would be provided information on the flight and on arrival in NZ on the critical importance of seeking medical attention if they develop any symptoms. They would also be told that such medical attention would be provided free of charge. We assumed PCR confirmation of self-reported symptoms and if a positive test, then we assumed case isolation and potentially triggering contact tracing Of note is that routinely in NZ, 39.5% of people with “fever and cough” symptoms seek medical attention, as reported by the NZ Flutracking surveillance system27. This is very similar to international estimates for people with influenza who seeking medical attention at 40% e.g., as used in other modelling6 |
| Quarantine of traced contacts | 1 day after detection of index cases | Traced contacts are assumed to be effectively quarantined with no further spread of infection |
| Mandatory mask use by incoming travellers up to the time of their final PCR test in the NZ community | 66% transmission reduction | We used the results of the most recent meta-analysis we could identify which involved 29 studies on infection with SARS-CoV-2, SARS, or MERS28. It reported that “type N-95 masks (corresponding approximately to FFP-2), surgical masks, or similar multilayer cotton masks can greatly reduce the infection risk for the wearers (RR 0.34 [0.26; 0.45]).”* |
*See Supplementary Information for consideration of uncertainty and probabilistic sensitivity analysis.