Dog bites in humans in the city of Jaboticabal, SP, Brazil from 2014 to 2016

Author(s)

Mirelle Andréa de Carvalho Picinato

• Document type: Doctoral Thesis
• Press: Jaboticabal. 2017-03-02.
• Institution: Universidade Estadual Paulista (Unesp). Faculdade de Ciências Agrárias e Veterinárias. Jaboticabal
• Defense date: 2017-01-20
• Advisor: Antonio Sergio Ferraudo; Adolorata Aparecida Bianco Carvalho

Abstract

The aim of this study was to characterize dog bites in humans in the city of Jaboticabal, SP, Brazil, from 2014 to 2016. A geographic dataset was built using the answers obtained from a semi-structured questionnaire applied to the local residents. Those information allowed to identify the pattern of the attacking dogs, general information of the residents, owners and victims. The dynamic of the dog bites process was evaluated by the means of descriptive statistics, multivariate exploratory analysis, logistic regression, neural networks analysis and geoprocessing tools. The spatial structure of the dog bites was evaluated by the means of dispersion and aggregation indexes and normal distributed kernel functions. The interpolation of the inverse of the distance showed the neighbourhoods where dog bites occurred, thus describing the spatial distribution of dog bites. The Local and Global Moran indexes and the Bayesian empirical rates estimated the risk of occurrence of dog bites in each sector. From a total of 301 questionnaires 70% of dog bites were caused by male dogs and 71% non-castrated; 68% were cross-breed, 22% had background of aggressions; 48% go out on a leash and 27% go out freely, without human companionship; 54% of dogs react aggressively during playing time or removing of objects; 48% reacts aggressively when their owner get threatened or give more attention to children. Considering the owners, 81% described theirs dogs as obedient and 74% had no knowledge about zoonosis. The statistical analysis by sectors (83) considering the three dimensions (victims and occurrences, animals characteristics and animal behaviour that could lead to aggression) allowed to see the dog bites pattern and the associated factors: age (adults and elderly), gender of the attacking dogs, number of animals per sector, medium-sized animals, locked-up in the presence of strangers, elevated aggressiveness due to dominant behaviour and lack of obedience. The neural networks had a better performance compared to the logistic regression to classify the sectors according to the probability of occurrence of dog bites. This may be occurred due to the nonlinear behaviour of the data. The spatial analysis highlighted the sectors of elevated social vulnerability as the ones with higher chance of dog bite occurrence. These sectors are located on the peripheral zone of the city and are associated with elevated risk of dog bites probably due to the elevated number of errant dogs that causes epidemiologic and sanitary issues like accidents, diseases, spreading of litter and faeces. (AU)

• FAPESP's process: 13/13784-0 - DISTRIBUTION MODELS AND SPATIAL ANALYSIS OF RISK OF DOG AND CAT AGGRESSIONS ON HUMANS
• Grantee: Mirelle Andréa de Carvalho Picinato
• Support Opportunities: Scholarships in Brazil - Doctorate