Combining COVID-19 Vaccination with Social Distancing Measures

TL;DR

This paper develops an optimal control model combining COVID-19 vaccination and social distancing, considering demographic factors, seasonality, and fatigue, to inform effective policy strategies.

Contribution

It introduces a novel optimal control framework that integrates vaccination prioritization, social distancing, seasonality, and demographic heterogeneity in a simple SIR model.

Findings

Vaccination priority depends on healthcare demand and mobility.

Seasonal social distancing policies are optimal.

Vaccination and social distancing policies are sensitive to demographic and parameter variations.

Abstract

We analyze an optimal control version of a simple SIR epidemiology model that includes a partially specified vaccination policy and takes into account fatigue from protracted application of social distancing measures. The model assumes demographic (age related) categories, and is otherwise homogeneous. Maximum capacity for vaccination is exogenous and the authorities select its allocation to the categories. They can also adopt measures to uniformly diminish the contact rate between infected and susceptible individuals at an economic cost. Total or partial immunity is modeled, while the contact rate is allowed to exhibit seasonality. We apply optimal control methods to minimize a cost index while guaranteeing that health system capacity is not exceeded. A reasonable parameter selection leads to policies specifying that vaccination priority should be given to the category with a higher demand for the limited health care resources regardless of cost variations among categories. Vaccination priority reverses if some demographic category has significantly higher mobility while for some parameter values vaccination alternates among categories. Optimal social distancing policies exhibit seasonality, and hence reducing susceptibles below the average spontaneous disease extinction level does not necessarily lead to a repeal of social distancing.