Optimal policies for mitigating pandemic costs

TL;DR

This paper applies optimal control theory to develop top-down policies for reducing pandemic costs, balancing health and socioeconomic factors, and finds that targeted, adaptive strategies outperform broad lockdowns.

Contribution

It introduces an optimal control framework for pandemic policy design based on an age-structured disease model, providing robust, adaptable strategies that minimize costs.

Findings

Optimal policies vary from targeted social distancing to partial lockdowns.

Long-term broad lockdowns are rarely optimal and are highly unstable.

Combining top-down policies with behavioral changes enhances mitigation.

Abstract

Several non-pharmaceutical interventions have been proposed to control the spread of the COVID-19 pandemic. On the large scale, these empirical solutions, often associated with extended and complete lockdowns, attempt to minimize the costs associated with mortality, economic losses and social factors, while being subject to constraints such as finite hospital capacity. Here we pose the question of how to mitigate pandemic costs subject to constraints by adopting the language of optimal control theory. This allows us to determine top-down policies for the nature and dynamics of social contact rates given an age-structured model for the dynamics of the disease. Depending on the relative weights allocated to life and socioeconomic losses, we see that the optimal strategies range from long-term social-distancing only for the most vulnerable, to partial lockdown to ensure not over-running hospitals, to alternating-shifts with significant reduction in life and/or socioeconomic losses. Crucially, commonly used strategies that involve long periods of broad lockdown are almost never optimal, as they are highly unstable to reopening and entail high socioeconomic costs. Using parameter estimates from data available for Germany and the USA, we quantify these policies and use sensitivity analysis in the relevant model parameters and initial conditions to determine the range of robustness of our policies. Finally we also discuss how bottom-up behavioral changes can also change the dynamics of the pandemic and show how this in tandem with top-down control policies can mitigate pandemic costs even more effectively.