Modeling the Comparative Impact of Individual Quarantine vs. Active Monitoring of Contacts for the Mitigation of COVID-19

Corey M Peak; Rebecca Kahn; Yonatan H Grad; Lauren M Childs; Ruoran Li; Marc Lipsitch; Caroline O Buckee

all contacts are traced and under active monitoring, since symptom onset is the event that triggers, with some delay, isolation. Mass quarantine is expected to result in prompt isolation upon symptom onset of any truly infected individuals, but the impact of this strategy on COVID-19 will depend heavily on whether presymptomatic exposure within the group is decreased or increased by the approach to confinement. That is, mass quarantine may reduce or increase the number of uninfected contacts exposed to presymptomatic infectiousness of those who do go on to develop the disease. In serial interval scenario 1, where a mean of 20% of transmission is expected to occur before symptom onset, the positive effect of prompt isolation can be offset by an increase in presymptomatic transmission in a confined space. Mass quarantines can also result in unintended consequences that can exacerbate transmission of COVID-19 such as avoidance of contact tracers and inaccurate recall, or other infectious diseases more broadly, such as a reduction in healthcare worker support, availability of supplies, or high-density settings. 20 The impact of travel restrictions on human mobility, a necessary first step in the causal chain to outbreak containment, is difficult to measure, but the impact has been documented in Sierra Leone during the 2014-2016 epidemic of Ebola Virus Disease. 21 The conflicting conclusions from our two scenarios, driven largely by the differences in the extent of presymptomatic transmission, highlight the urgent need for more data to clarify key epidemiological parameters of COVID-19, particularly the serial interval and the extent of presymptomatic transmission, in order to inform response efforts. These highly influential parameters warrant further study to improve data-driven policy-making. In addition, our choice to consider shifts of the latent period relative to the infectious period implicitly assumes a similar shape to the underlying distributions, albeit with different means. As the amount of asymptomatic transmission will depend not only on the average timing of latent period relative to the incubation period, but also on the standard deviation of these distributions, more data on their true shapes is urgently needed.