The time of implementation and relaxation of non-pharmaceutical interventions (NPIs) such as border closures, national lockdowns, quarantines, and social distancing protocols during the COVID-19 pandemic has been a polarising topic of debate globally. Decision-makers are having a tough time grasping how best to balance the surge in infections and the socio-economic costs of sustaining NPIs and related restrictions.
What is the effective reproduction number (R0)?
The effective reproduction number (R0), displayed on COVID-19-related dashboards, is one of the key early-warning parameters that inform NPI policy. Theoretically, R < 1 means that the epidemic is waning, and R > 1 means it is growing; hence, an escalation from R < 1 to R > 1 is a warning of the resurgence of infection.
However, in practice, precisely identifying this transition during low incidence periods is not easy. The lack of enough data during the lull between epidemic waves cripples standard inference approaches and blurs the early-warning signals. Reliable estimates of infection are critical for decision-making and improving the chances of elimination of second waves of infection.
A novel framework for de-noising inter-wave data and accurately assessing R0
Researchers from the Imperial College London and the University of Oxford recently proposed a new framework for de-noising inter-wave data, showing how timely introduction of NPIs in New Zealand achieved local epidemic elimination and avoided dangerous resurgence of infection. Their study is published on the preprint server medRxiv*.
The researchers presented an early-warning system for accurately assessing R and the possibility of elimination. This framework circumvents previously discussed problems and highlights the varied roles of imported and local cases in causing resurgence. It also underscores the importance of timely implementation and relaxation of NPIs.
Local transmission dynamics of COVID-19 in New Zealand. The top panel shows the local cases by date reported (red) and the additional cases due to introductions or imports (grey). Vertical lines provide key policy change-times and alert levels in response to these caseloads. The bottom panel presents effective reproduction number (R) estimates from EpiFilter (red with 95% confidence bands – these rigorously extract more information from incidence curves than several standard approaches) and corresponding probabilities (in %) of epidemic elimination (Z) – defined as the probability of no future local cases (blue). Both analytics account for the difference between local and imported cases. Transmission is largely driven by repeated imports with mostly subcritical local spread following timely interventions. Not only was national lockdown impactful, but elimination could be declared with 95% (99%) confidence on June 5. It was actually declared on June 9 . Recurring importations after this point eventually seeded a new epidemic that was decisively averted by timely measures in August. This resurgence may have presented more risk than the initial wave in March.
New Zealand’s timely application of NPIs helped achieve local epidemic elimination and avoided resurgence
New Zealand’s COVID-19 dynamics present a remarkable example of how swift and decisive intervention policy can impact resurgence and was used by the researchers as a representative but generalizable use-case for their new framework.
The first recorded local transmission of SARS-CoV-2 in New Zealand was in mid-March 2020, and in less than two weeks, a rapid 4-level alert system was initiated to inform NPI deployment. This led to a nation-wide lockdown (alert level 4) in New Zealand on March 26. During this lockdown period, surveillance and testing were increased, and the interval between symptom-onset and case notification was reduced to less than two days. As the epidemic started to die down, many of the NPIs were relaxed by May 14 (alert level 2). When no new cases were reported for a long period, and the epidemic was declared as eliminated on June 9 (alert level 1).
However, when transmission was again detected in early August, NPIs were promptly re-introduced to avoid a second wave (alert levels 2-3). De-escalation to alert level 1 followed on October 7, the last date analyzed in this study.
While previous studies have explained how New Zealand’s NPI policy facilitated the control and elimination of the epidemic, the researchers computed new transmission and risk indicators that closely align key transmission details with policy action-points, showing how timely NPIs helped eliminate and averted the resurgence of the epidemic. They introduced two informative early-warning analytics – the local R number and the Z number, which measure the community transmission and confidence in local elimination, respectively.
Locally relevant strategies vital to NPI-related policymaking
A thorough understanding of the transmission forces driving the epidemic spread is crucial to the design and timely implementation of NPIs. The authors argue that locally relevant strategies based on the specific dynamics of an area are vital to NPI-related policymaking. Their novel early-warning (R, Z) framework supports this argument, especially in the critical lull period between epidemic waves when available transmission data is limited.
According to the authors, epidemic growth can often be mischaracterized when differences in local and imported cases are ignored. Existing local R inference options do not fully exploit the information contained in the incidence data. Hence, they struggle to perform when data is scarce and are forced to rely on their prior model assumptions.
“While our framework provided rigorous underpinning and insight into New Zealand’s national response, it can also be applied at regional or district levels, both in real-time and retrospectively, to extract fine-scale insights.”
medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.