Using roaming behaviours of dogs to estimate contact rates: the predicted effect on rabies spread

Abstract

Domestic dogs display complex roaming behaviours, which need to be captured to more realistically model the spread of rabies. We have previously shown that roaming behaviours of domestic dogs can be categorised as stay-at-home, roamer and explorer in the Northern Peninsular Area (NPA), Queensland, Australia. These roaming behaviours are likely to cause heterogeneous contact rates that influence the speed or pattern of rabies spread in a dog population. The aim of this study was to define contact spatial kernels using the overlap of individual dog utilisation distributions to describe the daily probability of contact between pairs of dogs exhibiting these three a priori roaming behaviours. We further aimed to determine if the kernels lead to different predicted rabies outbreaks (outbreak duration and number of rabid dogs) by incorporating the spatial kernels into a previously developed rabies spread model for the NPA. Spatial kernels created with both dogs in a pair being explorers or one dog explorer and one dog roamer (who roamed away from their residence) produced short but large outbreaks compared with spatial kernels with at least one stay-at-home dog. Outputs from this model incorporating heterogeneous contacts demonstrate how roaming behaviours influence disease spread in domestic dog populations.

Modelling targeted rabies vaccination strategies for a domestic dog population with heterogeneous roaming patterns

“…In the NPA, three roaming patterns in the domestic dog population have been recently described– explorer dogs who frequently roam away from their owner’s residence, roamer dogs who mainly remain around their owner’s residence but roam away sometimes, and stay-at-home dogs who spend most (if not all) of their time around their owner’s residence, roaming to only the nearest neighbours [5]. These roaming patterns have been shown to create heterogeneous contact probabilities between individuals in the population and have been described by six spatial contact kernels based on all possible combinations of categories between a pair of dogs [6]; two explorer dogs (EE kernel), an explorer dog and a roamer dog (ER kernel), an explorer dog and a stay-at-home dog (ES kernel), two roamer dogs (RR kernel), a stay-at-home dog and a roamer dog (SR kernel) and two stay-at-home dogs (SS kernel). Previously, these kernels were individually incorporated into a rabies-spread model for the NPA developed by Dürr and Ward [3].…”

“…Previously, these kernels were individually incorporated into a rabies-spread model for the NPA developed by Dürr and Ward [3]. This was to study in isolation the differences in the contact probabilities produced by the spatial kernels and to quantify the effects of the different contact probabilities on predicted disease spread [3, 6]. However, heterogeneous contacts within the population can greatly affect epidemic spread and subsequent model predictions [7].…”

“…However, centre point vaccination alone was insufficient to reach 70% coverage and more expensive in remote pastoralist communities of Tanzania [18]. The three roaming categories of dogs in the NPA described by Hudson et al [6], provides an opportunity to use the NPA rabies model developed by Dürr and Ward [3] to evaluate targeted vaccination strategies. Targeting explorer and roamer dogs for vaccination in preference to stay-at-home dogs could produce the same or better results as the recommended overall 70% vaccination coverage potentially with fewer resources such as vaccine numbers.…”

“…Targeting explorer and roamer dogs for vaccination in preference to stay-at-home dogs could produce the same or better results as the recommended overall 70% vaccination coverage potentially with fewer resources such as vaccine numbers. This is because explorer and roamer dogs travel longer distances and subsequently come into contact with more dogs, producing a higher probability of transmitting rabies compared to stay-at-home dogs [6].…”

“…The objectives of this study were to incorporate the six spatial contact kernels for each of the roaming categories and their combinations created by Hudson et al [6] into the rabies-spread model for the NPA developed by Dürr and Ward [3], to more accurately represent the heterogeneous contact structure of the dog population. With this updated model, we then explore reactive vaccination strategies that target the individual dog categories within the NPA population at different coverage levels, and compare model predictions to the recommended 70% population vaccination coverage.…”