This paper is concerned with a model for the spread of an epidemic
in a closed, homogeneously mixed population in which new infections occur at
rate $\beta(z)xy/(x + y)$, where x, y and z denote,
respectively, the numbers of susceptible, infective and removed individuals.
Thus the infection mechanism depends upon the number of removals to date,
reflecting behavior change in response to the progress of the epidemic. For a
deterministic version of the model, a recurrent solution is obtained when
$\beta(z)$ is piecewise constant. Equations for the total size distribution of
the stochastic model are derived. Stochastic comparison results are obtained
using a coupling method. Strong convergence of a sequence of epidemics to an
unusual birth-and-death process is exhibited, and the behavior of the limiting
birth-and-death process is considered. An epidemic model featuring sudden
behavior change is studied as an example, and a stochastic threshold result
analagous to that of Whittle is derived.