Population Susceptibility Variation and Its Effect on Contagion Dynamics

Christopher Rose, Andrew J. Medford, C. Franklin Goldsmith, Tejs Vegge, Joshua Weitz, Andrew A. PetersonarXiv 2020 · OpenReview

TL;DR

This paper introduces a mathematical framework to model how variations in individual susceptibility to a disease can dramatically alter contagion dynamics and herd immunity thresholds, moving beyond the simplified SIR model. The authors demonstrate that ignoring this variation leads to overestimating herd immunity and suggest that understanding and measuring susceptibility distributions is crucial for effective pandemic response.

What Problem Does This Paper Address?

The standard SIR (Susceptible-Infected-Recovered) model, a cornerstone of epidemiological modeling, often assumes a uniform susceptibility to infection across the entire population. However, in reality, individuals possess varying degrees of susceptibility due to genetic, environmental, or behavioral factors. This paper addresses the critical gap left by the SIR model by investigating how this inherent variation in susceptibility influences the spread of a contagion and, consequently, the effectiveness of public health interventions like vaccination or inoculation. The authors aim to quantify this impact and understand how the contagion process itself can sculpt the susceptibility distribution over time.

Methods

The authors develop a new mathematical framework to explicitly incorporate population susceptibility variation. They introduce a probability density function, $f_\epsilon(t)(\epsilon)$, to describe the distribution of susceptibility $\epsilon$ within the susceptible population at time $t$. This function evolves according to a differential equation derived from the standard SIR model, but modified to account for varying susceptibility rates. The core of their method involves:

  • Differential Equation for Susceptibility Distribution: They derive a differential equation for $f_\epsilon(t)(\epsilon)$ that shows how the distribution changes as more susceptible individuals are infected. This equation highlights that individuals with higher susceptibility are preferentially removed, thus altering the remaining population's susceptibility profile.
  • General Solution: They derive a general closed-form solution for $f_\epsilon(t)(\epsilon)$ under certain conditions, showing how the initial susceptibility distribution $g_0(\epsilon)$ transforms over time.
  • Order Parameter K: To quantify the relationship between the average susceptibility and the number of susceptible individuals ($x_s$), they introduce an "order parameter" $K$. This parameter helps understand how the contagion state influences the average susceptibility and vice-versa.
  • Herd Immunity Threshold Calculation: They develop a method to calculate the herd immunity threshold that explicitly accounts for the initial susceptibility distribution, moving beyond the simplified calculations of the standard SIR model.

Datasets

Not discussed in the paper. The paper focuses on theoretical modeling and mathematical derivations, using hypothetical initial susceptibility distributions (e.g., uniform, Gamma, Pareto) for illustrative examples rather than real-world datasets.

Key Insights and Ideas

  • Ignoring Variation is Misleading: The most striking insight is that assuming uniform susceptibility always leads to an overestimation of the herd immunity threshold. This means we might underestimate the true proportion of the population that needs to be immune to control an epidemic.
  • Contagion Sculpts Susceptibility: The process of contagion itself actively shapes the susceptibility distribution. Individuals with higher susceptibility are removed faster, leading to a population that becomes, on average, less susceptible over time.
  • Gamma Distributions as "Eigenfunctions": Gamma distributions exhibit a special property: if the initial susceptibility distribution is Gamma, it remains Gamma throughout the contagion process, with the order parameter changing. This suggests a potential attractor behavior for the susceptibility distribution.
  • Targeted Interventions are Key: The paper strongly suggests that interventions should be targeted towards individuals with the highest susceptibility. Zeroing out the susceptibility of the most susceptible individuals is shown to be the most effective strategy for maximizing the herd immunity threshold.
  • Order Parameter K as a Dynamic Measure: The introduction of the order parameter $K$ provides a novel way to characterize the dynamic relationship between the overall contagion state ($x_s$) and the average susceptibility ($\bar{\epsilon}$), revealing that this relationship is not static and can change over time.

Key Results

  • Closed-form expressions for herd immunity thresholds: The authors derive formulas for herd immunity thresholds that depend on the initial susceptibility distribution, offering a more nuanced understanding than traditional models.
  • Demonstration of overestimation: The paper quantitatively shows how ignoring susceptibility variation leads to overestimating the herd immunity threshold, with the difference being potentially dramatic.
  • Examples of susceptibility evolution: Through figures and derivations, the paper illustrates how different initial susceptibility distributions (e.g., uniform, sinusoidal, linear) evolve into more concentrated distributions, often approximating Gamma distributions, as contagion progresses.
  • Identification of optimal intervention strategy: The analysis indicates that the most effective intervention is to inoculate (or otherwise reduce the susceptibility of) the fraction of the population with the highest susceptibility.

Limitations and Future Work

  • Static Susceptibility Assumption: The current model assumes static individual susceptibility, meaning an individual's susceptibility doesn't change over time unless they are removed from the susceptible pool. Future work could explore dynamic susceptibility.
  • Measurement Challenges: While the paper discusses potential proxies for susceptibility (e.g., mobility, contact intensity), it acknowledges the difficulty and expense of directly measuring susceptibility. Developing robust and real-time susceptibility measurement methods remains a significant challenge.
  • Complex Initial Distributions: The analysis is most straightforward for distributions that can be approximated by power series. More complex initial distributions, especially those with singularities or non-compact support, require more intricate mathematical treatment.
  • Model Extensions: The authors suggest that their framework could be extended to incorporate other factors like varying transmission rates or recovery rates, further enhancing the realism of contagion models. They also propose that real-time measurement of susceptibility could be a critical determinant of future pandemic policy.

Summary updated

Abstract

Susceptibility governs the dynamics of contagion. The classical SIR model is one of the simplest compartmental models of contagion spread, assuming a single shared susceptibility level. However, variation in susceptibility over a population can fundamentally affect the dynamics of contagion and thus the ultimate outcome of a pandemic. We develop mathematical machinery which explicitly considers susceptibility variation, illuminates how the susceptibility distribution is sculpted by contagion, and thence how such variation affects the SIR differential questions that govern contagion. Our methods allow us to derive closed form expressions for herd immunity thresholds as a function of initial susceptibility distributions and suggests an intuitively satisfying approach to inoculation when only a fraction of the population is accessible to such intervention. Of particular interest, if we assume static susceptibility of individuals in the susceptible pool, ignoring susceptibility diversity {\em always} results in overestimation of the herd immunity threshold and that difference can be dramatic. Therefore, we should develop robust measures of susceptibility variation as part of public health strategies for handling pandemics.
BibTeX
@misc{rose2020population,
  author = {Christopher Rose and Andrew J. Medford and C. Franklin Goldsmith and Tejs Vegge and Joshua Weitz and Andrew A. Peterson},
  title = {{Population Susceptibility Variation and Its Effect on Contagion Dynamics}},
  year = {2020},
  eprint = {2009.09150},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2009.09150},
}
Figure