Prompt case finding is essential to prevent a SARS-CoV-2 outbreak in a remote Aboriginal community. A high transmission propensity, due to interconnected and often crowded households, means that in an unmitigated scenario the majority of the community would be rapidly infected. By the time early cases are identified, active infections in the community may be up to ten-fold higher. We assume only half of all infected patients will self-present to health services for testing, due to absent or minimal symptoms, fear, or stigma. This may be an overestimate, but evidence that pre-symptomatic transmission may contribute > 40% of SARS-CoV-2 transmission exists [12, 19]. This non-presenting proportion may not be detected using a passive case finding approach, although a high prevalence of other co-morbidities may result in non-COVID related presentations resulting in ‘co-incidental’ case detection. Higher non-presenting proportions would lead to poorer mitigation in all scenarios, and vice versa (see Appendix).
Of the contact tracing strategies, quarantining extended household members (residents of all dwellings used by the case) is the most effective strategy for constraining the initial outbreak, reducing peak prevalence from 60–70% to ~ 10% (Fig. 3). However, large numbers of people must be quarantined for extended periods and infections may resurge when community mixing resumes, with overall community attack rates exceeding 80% (Table 3). Clearance testing modestly reduces this attack rate to 65%. Lockdown of all non-quarantined households for 14 days, concurrent with this quarantine strategy, results in the greatest likelihood of definitive outbreak control. Peak prevalence of the initial outbreak is less than 5%, and the overall attack rate less than 10%. Clearance testing from lockdown further improves control, preventing subsequent waves of infection due to undetected infections being released (Fig. 4): overall infections are constrained to < 5% with clearance testing, versus > 80% without. This strategy also requires fewer tests due to prompt suppression, fewer person-days in quarantine, and remains effective with delays of up to 6 days (Fig. 5). Larger communities benefit most from lockdown, with the effect dampened in smaller communities (100–500) by the large proportion already in quarantine. Compliance with lockdown must be at least 80–90%, or epidemic control will be lost.
Our findings are consistent with recent guidelines for a ‘contain and test’ strategy developed by Central Australian health organisations [8], which acknowledge that symptom-based case identification will be insufficient, and endorse active case finding and lockdown with multiple rounds of voluntary testing. Analyses of SARS-CoV-2 outbreaks overseas also support the effectiveness of lockdowns. In the Italian town of Vo, researchers concluded that a 14-day lockdown reduced transmissibility of infections (including asymptomatic) by 82–98% [20]. Lockdowns in Wuhan contributed to a significant decrease in spread [21], and an analysis of French data suggested that over 80% of potential COVID-19 deaths were averted by their lockdowns [22]. Recent modelling from the UK, examining the impact of delays with testing and contact tracing, suggests that if cumulative delays exceed 3 days for these processes, control of an outbreak is unlikely [23].
The participatory process employed between this study’s investigators, the IAG, and other public health end users throughout, have allowed for direct feedback of our findings and incorporation into IAG guidelines [9], and collaborative development of plain-language messaging for health providers and community members. Prompt case finding and a rapid public health response will be critical for effective control, with access to decentralised point-of-care testing (e.g. GeneXpert) facilitating this. Local planning and preparation should occur in advance, and must involve community members to ensure cultural appropriateness, local support and community control. Early patient presentation should be encouraged, and testing, contact tracing and isolation/quarantine guidelines and facilities clarified. The extensive public health response required to achieve best outcomes necessitates prior preparedness planning to ensure that the significant logistical and human resources support needed can be rapidly mobilised. Throughout an outbreak, community trust must be preserved in order to maximise compliance; in particular, the historical context and consequent sensitivities regarding enforced lockdowns in remote Aboriginal communities must be kept foremost in mind in the design and implementation of such strategies.
Limitations
Our model is informed by simplifying assumptions derived from observational data regarding population structure and mixing. Other ‘real world’ mixing opportunities (e.g. schools and workplaces) have not been explicitly included. Assumptions regarding transmission dynamics are derived from non-Aboriginal populations, but where possible we have erred on the side of caution. The high R0 to which the model is calibrated is based on early estimates from Wuhan and amplified to reflect the propensity for intense transmission in remote households. We assume perfect sensitivity and specificity of testing throughout the infectious period. Morbidity and mortality outcomes have not been estimated in this model, or the anticipated demand on health resources (testing requirements aside).
We assume that cases in isolation and contacts in quarantine will have no contact with others (i.e. will not transmit SARS-CoV-2). This may not be possible to achieve, but by representing this ideal we assess the maximum effectiveness of these measures and demonstrate the added value of lockdown.