Using dynamic stochastic modelling to estimate population risk factors in infectious disease: the example of FIV in 15 cat populations.

BACKGROUND:In natural cat populations, Feline Immunodeficiency Virus (FIV) is transmitted through bites between individuals. Factors such as the density of cats within the population or the sex-ratio can have potentially strong effects on the frequency of fight between individuals and hence appear a...

Full description

Bibliographic Details
Main Authors: David Fouchet, Guillaume Leblanc, Frank Sauvage, Micheline Guiserix, Hervé Poulet, Dominique Pontier
Format: Article
Language:English
Published: Public Library of Science (PLoS) 2009-10-01
Series:PLoS ONE
Online Access:http://europepmc.org/articles/PMC2770169?pdf=render
Description
Summary:BACKGROUND:In natural cat populations, Feline Immunodeficiency Virus (FIV) is transmitted through bites between individuals. Factors such as the density of cats within the population or the sex-ratio can have potentially strong effects on the frequency of fight between individuals and hence appear as important population risk factors for FIV. METHODOLOGY/PRINCIPAL FINDINGS:To study such population risk factors, we present data on FIV prevalence in 15 cat populations in northeastern France. We investigate five key social factors of cat populations; the density of cats, the sex-ratio, the number of males and the mean age of males and females within the population. We overcome the problem of dependence in the infective status data using sexually-structured dynamic stochastic models. Only the age of males and females had an effect (p = 0.043 and p = 0.02, respectively) on the male-to-female transmission rate. Due to multiple tests, it is even likely that these effects are, in reality, not significant. Finally we show that, in our study area, the data can be explained by a very simple model that does not invoke any risk factor. CONCLUSION:Our conclusion is that, in host-parasite systems in general, fluctuations due to stochasticity in the transmission process are naturally very large and may alone explain a larger part of the variability in observed disease prevalence between populations than previously expected. Finally, we determined confidence intervals for the simple model parameters that can be used to further aid in management of the disease.
ISSN:1932-6203