Assessing the probability of acquisition of meticillin-resistant Staphylococcus aureus (MRSA) in a dog using a nested stochastic simulation model and logistic regression sensitivity analysis

TY - JOUR

T1 - Assessing the probability of acquisition of meticillin-resistant Staphylococcus aureus (MRSA) in a dog using a nested stochastic simulation model and logistic regression sensitivity analysis

AU - Heller, J

AU - Innocent, G T

AU - Denwood, Matthew

AU - Reid, S W J

AU - Kelly, L

AU - Mellor, D J

N1 - Copyright © 2010 Elsevier B.V. All rights reserved.

PY - 2011/5/1

Y1 - 2011/5/1

N2 - Meticillin-resistant Staphylococcus aureus (MRSA) is an important nosocomial and community-acquired pathogen with zoonotic potential. The relationship between MRSA in humans and companion animals is poorly understood. This study presents a quantitative exposure assessment, based on expert opinion and published data, in the form of a second order stochastic simulation model with accompanying logistic regression sensitivity analysis that aims to define the most important factors for MRSA acquisition in dogs. The simulation model was parameterised using expert opinion estimates, along with published and unpublished data. The outcome of the model was biologically plausible and found to be dominated by uncertainty over variability. The sensitivity analysis, in the form of four separate logistic regression models, found that both veterinary and non-veterinary routes of acquisition of MRSA are likely to be relevant for dogs. The effects of exposure to, and probability of, transmission of MRSA from the home environment were ranked as the most influential predictors in all sensitivity analyses, although it is unlikely that this environmental source of MRSA is independent of alternative sources of MRSA (human and/or animal). Exposure to and transmission from MRSA positive family members were also found to be influential for acquisition of MRSA in pet dogs, along with veterinary clinic attendance and, while exposure to and transmission from the veterinary clinic environment was also found to be influential, it was difficult to differentiate between the importance of independent sources of MRSA within the veterinary clinic. The implementation of logistic regression analyses directly to the input/output relationship within the simulation model presented in this paper represents the application of a variance based sensitivity analysis technique in the area of veterinary medicine and is a useful means of ranking the relative importance of input variables.

AB - Meticillin-resistant Staphylococcus aureus (MRSA) is an important nosocomial and community-acquired pathogen with zoonotic potential. The relationship between MRSA in humans and companion animals is poorly understood. This study presents a quantitative exposure assessment, based on expert opinion and published data, in the form of a second order stochastic simulation model with accompanying logistic regression sensitivity analysis that aims to define the most important factors for MRSA acquisition in dogs. The simulation model was parameterised using expert opinion estimates, along with published and unpublished data. The outcome of the model was biologically plausible and found to be dominated by uncertainty over variability. The sensitivity analysis, in the form of four separate logistic regression models, found that both veterinary and non-veterinary routes of acquisition of MRSA are likely to be relevant for dogs. The effects of exposure to, and probability of, transmission of MRSA from the home environment were ranked as the most influential predictors in all sensitivity analyses, although it is unlikely that this environmental source of MRSA is independent of alternative sources of MRSA (human and/or animal). Exposure to and transmission from MRSA positive family members were also found to be influential for acquisition of MRSA in pet dogs, along with veterinary clinic attendance and, while exposure to and transmission from the veterinary clinic environment was also found to be influential, it was difficult to differentiate between the importance of independent sources of MRSA within the veterinary clinic. The implementation of logistic regression analyses directly to the input/output relationship within the simulation model presented in this paper represents the application of a variance based sensitivity analysis technique in the area of veterinary medicine and is a useful means of ranking the relative importance of input variables.

KW - Animals

KW - Computer Simulation

KW - Dog Diseases

KW - Dogs

KW - Environmental Microbiology

KW - Female

KW - Logistic Models

KW - Male

KW - Methicillin-Resistant Staphylococcus aureus

KW - Risk Assessment

KW - Staphylococcal Infections

KW - Stochastic Processes

KW - Zoonoses

U2 - 10.1016/j.prevetmed.2010.10.007

DO - 10.1016/j.prevetmed.2010.10.007

M3 - Journal article

C2 - 21277032

SN - 0167-5877

VL - 99

SP - 211

EP - 224

JO - Preventive Veterinary Medicine

JF - Preventive Veterinary Medicine

IS - 2-4

ER -