Dependence Structure and Epidemic Outcomes in Heterogeneous SIR Models

TL;DR

This paper analyzes how the dependence between susceptibility and infectivity traits in a heterogeneous SIR model influences epidemic size and outbreak probability, revealing that positive dependence can increase severity.

Contribution

It introduces a joint distribution framework for susceptibility and infectivity, showing how dependence affects epidemic outcomes and providing bounds based on concordance order.

Findings

Positive dependence increases epidemic severity.

Variance alone does not determine epidemic size.

Dependence structure crucially influences outbreak probability.

Abstract

We study a well mixed SIR epidemic model with heterogeneous susceptibility and infectivity, allowing for an arbitrary joint distribution of these traits. Using an exact final size formulation and a branching process approximation for early epidemic dynamics, we show that both the final epidemic size and the probability of a major outbreak are monotone with respect to the concordance order of the joint susceptibility infectivity distribution. In particular, among all couplings with fixed marginal trait distributions, comonotonic dependence maximizes epidemic severity, yielding sharp distribution free upper bounds. A key implication is that epidemic outcomes cannot be ordered by susceptibility heterogeneity alone: while increasing susceptibility variance reduces the final size under independence, positive dependence between susceptibility and infectivity can locally increase epidemic size for any basic reproduction number exceeding one. We further show that neither susceptibility variance nor the sign of the covariance suffices to determine epidemic severity under dependence. These findings offer new insights into epidemic risk assessment under limited information.