Optimal Staged Reopening Schedule Based on ICU Capacity: A Model-Informed Strategy

Abstract

Reopening plans and strategies remain to be a top priority issue, post a pandemic wave, for economic recovery with a relatively safe community. We use a transmission dynamics model, parameterized through model fitting to cumulative incidence data during different social distancing escalation phases, to identify the optimal timing of reopening based on social-distancing de-escalation in a population. We use the Province of Ontario, Canada as a case study. The optimization is subject to the constraint that a future COVID-19 outbreak will not lead to the need of acute and intensive care unit beds beyond their local public health capacity. We minimize the cost, the total number of contacts lost until we reach a certain targeted date. We illustrate this optimization process by considering a particular de-escalation strategy that simply ‘reverses’ the escalation steps. We consider several scenarios depending on the number of de-escalation phases (characterized by an increase in the daily number of contacts), and different lengths of the period between a date when the cost is evaluated until a targeted data when the constraint is to be removed (for example, the anticipated date when a mass COVID-19 vaccination or an effective treatment becomes available) so that normal contacts prior to COVID-19 pandemic can be resumed. In the case of Ontario, we conclude that resuming 80% of pre-pandemic activity level should not happen 2 months before the vaccine/drug treatment becomes available. We also show that improving contact tracing and diagnosis capacity has a significant effect on the reopening date, whereas increasing ICU beds capacity has only a minor effect.