Abrupt transition of the efficient vaccination strategy in a population with heterogeneous fatality rates

TL;DR

This study uses a metapopulation SIRD model to analyze vaccination strategies for COVID-19, revealing a sudden transition in optimal strategy depending on vaccine supply and contagion rate, with implications for real-world epidemics.

Contribution

It identifies a discontinuous transition in the optimal vaccination strategy and demonstrates its relevance to real-world diseases like COVID-19 and tuberculosis.

Findings

Fatality-based strategy is more effective at high contagion and low vaccine supply.

Contact-based strategy outperforms fatality-based when vaccine supply is high.

A discontinuous transition in optimal strategy occurs, similar to hysteresis.

Abstract

An insufficient supply of effective SARS-CoV-2 vaccine in most countries demands an effective vaccination strategy to minimize the damage caused by the disease. Currently, many countries vaccinate their population in descending order of age (i.e. descending order of fatality rate) to minimize the deaths caused by the disease; however, the effectiveness of this strategy needs to be quantitatively assessed. We employ the susceptible-infected-recovered-dead (SIRD) model to investigate various vaccination strategies. We constructed a metapopulation model with heterogeneous contact and fatality rates and investigated the effectiveness of vaccination strategies to reduce epidemic mortality. We found that the fatality-based strategy, which is currently employed in many countries, is more effective when the contagion rate is high and vaccine supply is low, but the contact-based method outperforms the fatality-based strategy when there is a sufficiently high supply of the vaccine. We identified a discontinuous transition of the optimal vaccination strategy and path-dependency analogous to hysteresis. This transition and path-dependency imply that combining the fatality-based and contact-based strategies is ineffective in reducing the number of deaths. Furthermore, we demonstrate that such phenomena occur in real-world epidemic diseases, such as tuberculosis and COVID-19. We also show that the conclusions of this research are valid even when the complex epidemic stages, efficacy of the vaccine, and reinfection are considered.