Spread of airborne plant diseases from a propagule source is classically assessed by fitting a gradient curve to aggregated data coming from field experiments. But, aggregating data decreases information about processes involved in disease spread. To overcome this problem, individual count data can be collected; it was done in the case of short-distance spread of wheat brown rust. However, for such data, the gradient curve is a limited model since heterogeneity of hosts is ignored and, consequently, overdispersion occurs. So, we propose a parametric frailty model in which the frailties represent propensities of hosts to be infected. The model is used to assess dispersal of propagules and heterogeneity of hosts.