A risk assessment approach to meticillin-resistant Staphylococcus aureus (MRSA) in pet dogs.
PhD thesis, University of Glasgow.
Full text available as:
A risk assessment approach was used to: a) define the risk of acquisition of meticillin-resistant Staphylococcus aureus (MRSA) in pet dogs over a defined time period, b) attempt to identify important priority data gaps for future research efforts in this area and c) comment on the usefulness of risk assessment in this regard, in a data sparse area and in a field in which it has not previously been used.
A conceptual model was defined that identified the potential pathways for acquisition of MRSA in a pet dog over any given 24 hours. A qualitative risk assessment, using categorical qualitative estimators and combining estimates using a matrix approach was undertaken. It was found that this approach was unsatisfactory for the specification of this biological model that encompassed a non-modular, non-sequential form, characterised by non-mutually exclusive pathways of acquisition. Modification of the categorical descriptors enabled a relative, rather than absolute assessment of risk and resulted in the conclusion that both veterinary and non-veterinary routes were potentially important for the acquisition of MRSA in pet dogs and that family members and staff were likely to be the most important sources of the organism in the community and veterinary clinic environments respectively. Given the limitations encountered, a quantitative risk assessment was pursued.
Data gaps that were defined within the qualitative risk assessment were addressed through dedicated data-collection studies and an expert opinion elicitation exercise. The studies found a lower veterinary environmental and staff prevalence than had previously been reported, corroborated prior estimates of dog-dog and dog-human interactions and justified the inclusion of dog foods as environmental sources of MRSA rather than independent and important sources in their own right. The expert opinion elicitation exercise used a modified technique to obtain numerous estimates relating to prevalence and transmission of MRSA. However, it was found that experts lacked confidence in estimation of transmission variables in particular, and the resulting distributions for these variables demonstrated divergence between experts and resulted in wide and poorly-informative combined distributions.
These results were utilised, along with published and unpublished data, to parameterise a second order stochastic simulation-based quantitative risk assessment model. The model produced a biologically plausible outcome and allowed the application of sensitivity analyses with the intention of identifying areas of putative importance for future research efforts. The implementation of logistic regression analyses directly to the input/output relationship within the simulation model represented a novel application of a variance-based sensitivity analysis technique in the area of veterinary medicine, and was implemented with and without the consideration of interaction terms. In addition, one-at-a-time (OAT) and Plackett-Burman (P-B) analyses were also completed. The results of the sensitivity analyses were complicated and ambiguous. While family members and the environment were identified as potentially important independent and non-independent sources of MRSA, respectively, it was not possible to discount, or defensibly rank the importance of other sources.
In conclusion, it was found that, despite the application of well researched and previously utilised methods, marked limitations were encountered in the use of risk assessment to address a biologically complex phenomenon that is characterised by sparse data such as this.
Actions (login required)