Analysis of predictors of rabies-positive biting animals in Cambodia using spatio-temporal Bayesian regression modelling.

Abstract

Cambodia is endemic for rabies, a fatal zoonotic viral disease transmitted through dog bites. The Institut Pasteur du Cambodge through its Rabies Prevention Center is the main institution in charge of rabies prevention and surveillance in the country. Its main tool for prevention is post-exposure prophylaxis (PEP) for bite victims. Allocation of specific PEP regimen is done based on the assessment of the severity of the wound and the information collected by IPC doctors from patients regarding the attack's characteristics and the attacking animal's health status. Furthermore, a small proportion of patients bring animals for testing, 60% of which were tested positive for rabies. Using the data collected from patient interviews from 2000 to 2016, we used a Bayesian spatio-temporal regression model to identify predictors for two outcomes: i) a patient bringing an animal for testing and ii) an animal testing positive for rabies. The ultimate aim of the analysis was to provide information that could help with allocation of PEP resources. Notably non-owned animals, a large number of bite victims, and unprovoked attacks were all predictive of a positive test. A suspected rabies status assigned by doctor based on animal symptom description was also highly predictive of a rabies test, with 94.6% of tested animals that were assigned as sick being positive for rabies. Furthermore, we identified three Provinces of Cambodia with higher odds of positive tests: Kampong Cham, Kandal and Kampong Thom. This information could help allocate limited PEP resources, though this study showed IPC already a strong protocol to identify patients exposed to a rabies suspect dog.