Modeling epidemics caused by respiratory syncytial virus (RSV)

https://doi.org/10.1016/S0025-5564(01)00066-9Get rights and content

Abstract

Respiratory syncytial virus (RSV) is the most common cause of acute lower respiratory tract infection in children. In this paper we use models of RSV transmission to interpret the pattern of seasonal epidemics of RSV disease observed in different countries, and to estimate epidemic and eradication thresholds for RSV infection. We compare the standard SIRS model with a more realistic model of RSV transmission in which individuals acquire immunity gradually after repeated exposure to infection. The models are fitted to series of monthly hospital case reports of RSV disease from developed and developing countries. The models can explain many of the observed patterns: regular yearly outbreaks in some countries, and in other countries cycles of alternating larger and smaller annual epidemics, with shifted maxima in alternate years. Previously these patterns have been attributed to the transmission of different strains of RSV. In some countries the timing of epidemics is not consistent with increased social contact among school children during term time being the major driving mechanism. Climatic factors appear to be more important. Qualitatively different models gave equally good fits to the data series, but estimates of the transmission parameter were different by a factor of 4. Estimates of the basic reproduction number (R0) ranged from 1.2 to 2.1 with the SIRS model, and from 5.4 to 7.1 with the model with gradual acquisition of partial immunity.

Introduction

Respiratory syncytial virus (RSV) is the most common cause of acute lower respiratory tract infection in children. Clinical features of RSV infection range from those of a common cold to bronchiolitis and severe pneumonia. Cases of RSV disease occur in seasonal epidemics. The majority of hospitalized cases are aged under 6 months, but the infection can also be a cause of mortality in the elderly. In this paper we develop a model of RSV transmission, and apply it to the interpretation of series of monthly case reports of RSV disease from developed and developing countries. The mechanisms underlying the different patterns of seasonal epidemics observed in different countries are not well understood; our aim in this paper is to examine the effect of population dynamics of transmission on the interpretation of the role of climatic and social factors in driving these epidemics. Furthermore, RSV vaccines are under development; if suitable models can be developed they can be used to evaluate the likely impact of vaccination programmes.

In Section 2 we outline the properties of RSV. In Section 3 we describe extensions to the SIRS model which includes features specific to RSV transmission. Suitable values for some of the model parameters could be obtained from results given in the medical literature, but some parameters, especially the mean b0 of the transmission parameter β and its seasonal amplitude, had to be estimated by fitting the models to monthly case reports of RSV disease (Section 4). Separately we investigated the effects of different assumptions about the duration of immunity on model behavior (Section 5), and the influence of stochastic variation in some parameter values. In Section 6 we give the results of several simulations when the seasonal variation of the weather is correlated to the annual variation of the transmission parameter. Threshold conditions for our models are given in Section 7. Some final discussions are given in Section 8.

Section snippets

Properties of RSV

RSV is the most common cause of acute lower respiratory tract infection in children worldwide [1], [2], [3]. A striking feature of RSV infection is its seasonality: in temperate climates, most of the RSV-associated disease episodes occur in the cold season, whereas in the tropics, most appears to occur in the wet season [3]. RSV appears to be a predominantly human pathogen: it has been found in some animals, but these are not believed to play a role in the transmission to humans [4]. The

Models

In our models the population is divided into distinct groups (of susceptible, infected, or immune individuals) and we use deterministic continuous transitions between the states. So the overall setting is the one treated in-depth in the literature, see, e.g., [27]. We assume homogeneous mixing in the population and do not model spatial spread of epidemics. For RSV almost simultaneous outbreaks of RSV epidemics with different strains of the virus have been reported throughout a whole area (see,

Parameter estimates

For the simulations and parameter estimation we used Matlab. The functions describing the vector fields and their Jacobians were generated from a representation of the dynamical system in the computer algebra system Maple by a tool written by Goller [35] under the supervision of the first author. The integrators used were the standard Matlab functions ode23t and ode15s.

The optimal parameter fits were obtained in most cases by non-linear least squares using functions provided in the `Matlab

Results of further simulations

In fitting the models we held the parameters μ,ν,σ,γ, and ξ at fixed values. For some of the parameters, especially γ, present knowledge is only sufficient to determine a certain interval for its values, so we performed additional simulations in order to estimate the sensitivity of the simulations to changes in this parameter (Fig. 5) and to changes in b0 and b1 (Fig. 6, Fig. 7).

As initial values for the simulations each group was given the same size and we used 100 years of simulated time as a

Correlating the transmission parameter with weather data

In the work of several authors, e.g., [3], [38], attempts to correlate weather conditions with the epidemics of RSV have been made. This work is motivated by the fact that cyclic changes in the social behavior of school children (the beginning of school terms) do not seem to be linked to RSV epidemics [22]. For example, in Gambia, outbreaks occurred before the beginning of the school year [37]. It is possible that other social factors such as indoor crowding during the rainy season might play a

Thresholds for the models

The SIRS model with simple vital dynamics (as in Section 3.2, but with the transmission parameter β(t) assumed to be a constant, β) has a unique equilibrium if the basic reproduction rate β/(μ+ν) exceeds one [41]. If the birth and recovery rates μ and ν take the values given in Table 2, then β must exceed 36 for the infection to persist in the community. This value is about half the value estimated from fitting the model to the case reports series (Table 2).

It can be shown (by solving the

Discussions

We found that RSV epidemics can be modeled surprisingly well by relatively simple ODE models. These models, which can reproduce the reported regular annual outbreaks of epidemics seen in many places, can also reproduce the alternating cycles of large and smaller annual epidemics observed in some countries, with a change in the value of a single parameter (the relative seasonal amplitude of the transmission parameter). Fitted values of the relative amplitude were greater in places with

Acknowledgements

We are grateful to K. Dietz for helpful remarks on a draft of this paper.

References (44)

  • R.K. Ditchburn et al.

    Respiratory syncytial virus in hospital cross-infection

    Br. Med. J.

    (1971)
  • H.M. Kravetz et al.

    Respiratory syncytial virus III. Production of illness and clinical observations in adult volunteers

    JAMA

    (1961)
  • M.M. Ogilvie et al.

    Maternal antibody and respiratory syncytial virus infection in infancy

    J. Med. Virol.

    (1981)
  • F.W. Henderson et al.

    Respiratory-syncytial-virus infections, reinfections and immunity. A prospective, longitudinal study in young children

    New Engl. J. Med.

    (1979)
  • A.S. Monto et al.

    The Tecumseh study of respiratory illness. 3. Incidence and periodicity of respiratory syncytial virus and Mycoplasma pneumoniae infections

    Am. J. Epidemiol.

    (1971)
  • C.B. Hall et al.

    Immunity to and frequency of reinfection with respiratory syncytial virus

    J. Infect. Dis.

    (1991)
  • C.B. Hall et al.

    Respiratory syncytial virus infections within families

    New Engl. J. Med.

    (1976)
  • W.P. Glezen et al.

    Risk of primary infection and reinfection with respiratory syncytial virus

    Am. J. Dis. Child.

    (1986)
  • R.B. Belshe et al.

    Impact of viral respiratory diseases on infants and young children in a rural and urban area of southern West Virginia

    Am. J. Epidemiol.

    (1983)
  • C.B. Hall et al.

    Possible transmission by fomites of respiratory syncytial virus

    J. Infect. Dis.

    (1980)
  • M.A. Mufson et al.

    Subgroup characteristics of respiratory syncytial virus strains recovered from children with two consecutive infections

    J. Clin. Microbiol.

    (1987)
  • P.M. Muelenaer et al.

    Group-specific serum antibody responses in children with primary and recurrent respiratory syncytial virus infections

    J. Infect. Dis.

    (1991)
  • Cited by (148)

    View all citing articles on Scopus
    1

    Partially supported by Deutsche Forschungsgemeinschaft under grants We 1945/1-1 and Ku 966/6-1. Major parts of research done while working at the Symbolic Computation Group, Department of Computer Science, University of Tübingen, Germany.

    View full text