Downscaling predictions for the owned-dog population size in Brazilian cities

• Grant number: 18/14279-0
• Support Opportunities: Regular Research Grants
• Start date: May 01, 2019
• End date: April 30, 2020
• Field of knowledge: Agronomical Sciences - Veterinary Medicine - Preventive Veterinary Medicine
• Principal Investigator: Oswaldo Santos Baquero
• Grantee: Oswaldo Santos Baquero
• Host Institution: Faculdade de Medicina Veterinária e Zootecnia (FMVZ). Universidade de São Paulo (USP). São Paulo , SP, Brazil

Abstract

Health diagnostics at the population level, epidemiological surveillance, and evaluation of the effectiveness of interventions, depend on quantitative measures such as the size and distribution of populations of interest. The lack of reliable estimates of dog population size has resulted in the extrapolation of the human/dog ratio (HDR) estimated at a given location, to estimate population sizes in other locations. However, small differences in the HDR imply differences that limit or prevent reasonable comparisons. For example, in a human population with 10,000 individuals, the dog population would be 3,333 if the HDR was 3/1, and 2,500 if the HDR was 4/1. If 1,800 dogs were vaccinated, rabies vaccine coverage would be 72% (above recommended) according to the second HDR, and 54% (below recommended) according to the first HDR. The Brazilian National Health Survey of 2013 (PNS) included a question that allows estimation of the dog population in all Brazilian states, but not in all cities. These estimates of the dog population size in the Brazilian states could be used as a predictive target for models including census variables aggregated by state. Subsequently, these models could be fed by census variables aggregated at the city level, to predict the dog population size in all Brazilian cities. This procedure known as downscaling predictions, would allow the prediction of the dog population size in all Brazilian cities, based on data from the PNS and the national census. The quality of the predictions could be evaluated directly and indirectly, and if satisfactory, could support programs of epidemiological surveillance and population management. (AU)