Spontaneous behavioural changes in response to epidemics

https://doi.org/10.1016/j.jtbi.2009.04.029Get rights and content

Abstract

We study how spontaneous reduction in the number of contacts could develop, as a defensive response, during an epidemic and affect the course of infection events. A model is proposed which couples an SIR model with selection of behaviours driven by imitation dynamics. Therefore, infection transmission and population behaviour become dynamical variables that influence each other. In particular, time scales of behavioural changes and epidemic transmission can be different. We provide a full qualitative characterization of the solutions when the dynamics of behavioural changes is either much faster or much slower than that of epidemic transmission. The model accounts for multiple outbreaks occurring within the same epidemic episode. Moreover, the model can explain “asymmetric waves”, i.e., infection waves whose rising and decaying phases differ in slope. Finally, we prove that introduction of behavioural dynamics results in the reduction of the final attack rate.

Introduction

The epidemic dynamics depends on the complex interplay between the characteristics of the pathogens’ transmissibility and the structure and behaviour of the host population. Spontaneous change of behaviour in response to epidemics (Ferguson, 2007), possibly related to risk perception (Bagnoli et al., 2007, Risau-Gusman and Zanette, 2008, Shaw and Schwartz, 2008), has been recently proposed as a relevant factor in the comprehension of infection dynamics. While the merits and influence of such phenomena are still debated (D’Onofrio et al., 2007, Moneim, 2007), experience from the 1918–1919 pandemic indicates that a better understanding of behavioural patterns is crucial to improve model realism and enhance the effectiveness of containment/mitigation policies (Bootsma and Ferguson, 2007).

Human behaviour is driven by evaluation of prospective outcomes deriving from alternative decisions and cost-benefit considerations. Past experience, response to the action of others and changes in exogenous conditions all contribute to the balance, to which game theory provides a rich and natural modelling framework (von Neumann and Morgenstern, 1947, Hofbauer and Sigmund, 1998). It is not surprising, therefore, that looking at behaviours through the lens of game theory has recently attracted the attention of the epidemiology community, for example, when modelling the evolution over time of voluntary vaccination uptakes (Bauch and Earn, 2004, Bauch, 2005).

In this paper we model a fairly general situation in which a population of individuals is subject to an epidemic outbreak developing according to an SIR model, but in which contact rates depend on the behavioural patterns adopted across the population. More specifically, all susceptible individuals can conform to either one or the other of two different behaviours, ba and bn, respectively, corresponding to an “altered” and a “normal” behavioural pattern. The first gives the individuals an advantage in terms of reduced risk of infection, yet at some extra cost. For example, avoidance of crowded environments reduces the risk of infection, but also entails disadvantages deriving from greater isolation. Individuals adopting the second (bn) are exposed to a normal risk of infection, but are spared the extra cost associated with ba. Individuals may choose to switch between ba and bn at any time, depending on cost-benefit assessments based on the perception of risk.

The resulting model consists in the coupling of two dynamical systems, one describing the epidemic transmission and the other describing the behavioural changes. In principle, there is no reason for the two phenomena to evolve at the same speed. It is therefore crucial to study the model allowing for different time scales, embodied in different time units.

We give a full description of the model when the dynamics of the behavioural changes are “fast” with respect to the epidemic transmission. In particular, we provide sufficient conditions on the parameters for generating sequences of epidemic waves. Moreover, we show that the model is able to account for “asymmetric waves”, i.e., infection waves whose rising and decaying phases differ in slope. However, similar patterns can also be observed when the time scales of the two dynamics are comparable. When the dynamic of behavioural changes is “slow”, the model basically reduces to a classical SIR.

The model's dynamics gives rise to patterns that are morphologically compatible with multiple outbreaks and the same-wave asymmetric slopes recently reported for the Spanish influenza of the 1918–1919 (Chowell et al., 2006b, Chowell et al., 2006a; Ferguson et al., 2006, Mills et al., 2004). For these phenomena (trivially incompatible with the classical SIR model) a variety of alternative explanations have in fact been advanced: military demobilization at the end of the First World War (Ferguson et al., 2006), genetic variation of the influenza virus (Castillo-Chavez et al., 1989, Andreasen et al., 1997, Boni et al., 2004), exogenous time changes in transmission rates, such as seasonal forcing (Colizza et al., 2006, Colizza et al., 2007). Other explanations have been proposed invoking coinfection scenarios (May and Nowak, 1995, Adler and Losada, 2002, Edwards et al., 2004, Merler et al., 2008).

Finally, and regardless of the relative speeds of dynamics, we show that the fraction of susceptible individuals at the end of the epidemic is always larger than that of a classical SIR model in which all individuals adopt the normal behaviour (bn) with the same parameters.

Section snippets

The model

Our model consists of the coupling of two mutually influencing phenomena: (a) the epidemic transitions and (b) the behavioural changes in the population of susceptible individuals.

As for the epidemic transitions, whose time unit is t, our model is based on an SIR scheme.1 We consider that susceptible individuals may adopt two mutually exclusive behaviours, bn (“normal”) and ba (“altered”).

Study of dynamics

System (5) admits a continuum of equilibria, namely (S,0,1-S,x) withx=1-2μ+1+4μ22and S[0,1]. We consider in detail only the case of an epidemic of a novel pathogen type, so that we are interested only at the equilibrium with S=1, i.e., (1,0,0,x). This is unstable when βnx+βa(1-x)>γ. Thus, we assume that the initial values for system (5) are the following: (S(0),I(0),R(0),x(0))=(1-I0,I0,0,x) with I0 close to 0. Note that 1/2<x<1 and x1 when μ0. Moreover, this equilibrium is stable

Discussion

When studying the spread of epidemics, behaviour and contact patterns are typically considered “background” for the infection—i.e., they are not themselves variables of the dynamics. It is interesting, however, to address cases for which the population behaviour cannot be merely considered as an independent (though time-varying) parameter, but it is better modelled as a variable whose evolution influences, and is influenced by, the dynamics of the infection.

With the introduction of an explicit

Acknowledgements

We are grateful to Vittoria Colizza, Mimmo Iannelli and Giuseppe Jurman for reading earlier versions of the paper and providing useful suggestions. We are also grateful to two anonymous reviewers for their suggestions that certainly improved this manuscript. We thank the European Union FP7 FLUMODCONT project for research funding.

References (31)

  • C.T. Bauch et al.

    Vaccination and the theory of games

    Proceedings of the National Academy of Sciences

    (2004)
  • M.C.J. Bootsma et al.

    The effect of public health measures on the 1918 influenza pandemic in U.S. cities

    Proceedings of the National Academy of Sciences

    (2007)
  • C. Castillo-Chavez et al.

    Epidemiological models with age structure, proportionate mixing, and cross-immunity

    Journal of Mathematical Biology

    (1989)
  • M. Ciofi degli Atti et al.

    Mitigation measures for pandemic influenza in Italy: an individual based model considering different scenarios

    PLoS ONE

    (2008)
  • V. Colizza et al.

    The role of the airline transportation network in the prediction and predictability of global epidemics

    Proceedings of the National Academy of Sciences

    (2006)
  • Cited by (142)

    • Waves of infection emerging from coupled social and epidemiological dynamics

      2023, Journal of Theoretical Biology
      Citation Excerpt :

      Many have discussed individual-based models including explicit consideration of the space and network structures of individuals (Wang et al., 2015). The analyses were performed numerically (except for a few; e.g. Poletti et al., 2009). In the present study, we considered a model with only two variables.

    • Multilayer networks with higher-order interaction reveal the impact of collective behavior on epidemic dynamics

      2022, Chaos, Solitons and Fractals
      Citation Excerpt :

      The interplay between the collective behavioral response of the population and the contagion dynamics has a significant bearing on the epidemic evolution. Game-theoretic models explicitly account for behavioral adaptation and the connection with epidemic spreading [6], mostly with a separation of time scales between the spreading dynamics and behavioral response [7–10]. For instance, behavioral changes only occur at the beginning of each time period or happen at a much lower frequency.

    View all citing articles on Scopus
    View full text