Mathematical epidemiology helps formalize and quantify epidemiological processes such as disease outbreak, epidemic recurrence, disease invasion and disease persistence in landscapes with well defined levels of heterogeneity. It is within the context of a mathematical framework that theoreticians address "what if" questions that are central to the development of public health policy or that aid the understanding of disease evolution. Two kinds of modeling frameworks have been proposed in the context of spatial epidemics. They are reaction diffusion and spatial contact models. The first class of models assumes that individuals diffuse at random while the second assumes that individuals "disperse" via distribution kernels. This dissertation attempts to tease out some of the reasons behind the theoretical predictions generated by both classes of models in the context of disease dynamics.Spatial models quantify disease spread in terms of epidemiological parameters (infection and recovery rates) that influence the speed of disease propagation-traveling epidemic fronts. A recurrent assumption behind both type of models is uniformity in disease propagation. Such an assumption while unrealistic facilitates the mathematical analysis. In this dissertation the assumption of uniform mixing (homogeneity) is relaxed, spatial heterogeneity in the transmission process is allowed.A novel reaction diffusion model is introduced and used to identify necessary and sufficient conditions for the aggregation of individuals that may result in response to the introduction of a communicable disease. The methodology and techniques used in the analysis of this model, which exhibits diffusive instability, include Turing theory, which as far as I know, has not been used in this context.The mathematical results are obtained in a framework that ignores demographic change. Extensions incorporate demographic factors making the mathematical analysis beyond my means Hence I explore the impact of demography numerically. The analytical results together with the numerical simulations suggest that the incorporation of different diffusion coefficients for at least two distinct infectious classes are not only necessary for diffusive instability but also serve to reduce the average number of contacts between infected and susceptible individuals which lead to reductions in prevalence.