Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.401644
Title: The quantitative epidemiology of canine neoplastic disease : risk factor identification using diagnostic histopathology data
Author: Richards, Heather G.
ISNI:       0000 0001 3516 5516
Awarding Body: University of Glasgow
Current Institution: University of Glasgow
Date of Award: 2003
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Research was undertaken to investigate the risk factors for neoplasia in two canine biopsy populations, the first originating fi-om the Canine Infectious Diseases Research Unit (CIDRU) diagnostic histopathology service, based at the University of Glasgow Veterinary School (GUVS), and the second from the diagnostic histopathology service operated by a commercial organisation. Both provide histopathology reports to veterinary practitioners located throughout the United Kingdom and occasionally overseas. The studies were undertaken to determine the feasibility of using these data sources for meaningful epidemiological analyses of host-related risk factors for canine neoplasia. The analytical scope was expanded to explore the effect of submitting practice as a risk factor for canine neoplasia. Finally, spatial and spatio-temporal epidemiological techniques were applied to the CIDRU data to explore the significance of its geographical origin upon an outcome of neoplasia in a canine biopsy. Records pertaining to canine biopsy submissions were extracted from both histopathology databases. The records were subjected to a hierarchical and iterative data cleaning process, which focused upon the main host-related variables of age, gender and breed of dog, when available, and biopsy site of origin. This procedure highlighted a number of important quality assurance issues in both datasets. The coding system used in the CIDRU database was found to be adequate for assisting data preparation, although there were issues relating to the lack of integral data checks and the use of free text input. The extensive use of free text input for the commercial dataset limited the amount of data content that could be prepared fi-om this database for subsequent analysis. Following establishment of data integrity, case-control studies of the cleaned datasets were performed. Multivariable logistic regression was used to assess the effect of the host- related risk factors of age, gender and breed of dog, when available, and site of biopsy, on the outcome of neoplasia in a biopsy submitted to the histopathology service. Similar results were produced in analyses of both datasets. The grouping of data by submitting veterinary practice was considered to cause violation of the assumption of independence for individual biopsies because of unknown practice-related factors associated with biopsy submission. Following the application of inclusion criteria to the data, a practice variable was entered into the host related multivariable models first as a fixed-effect term, then as a random- effect term. The introduction of a variable for practice verified that group effects due to practice were significant in the data from both histopathology services. Spatial and space-time analyses were conducted on the CIDRU dataset using spatial and space-time scan statistics. Graphical display of the results with a Geographical Information System (GIS) illustrated a trend for clusters with low risk of neoplasia diagnosis in biopsies submitted from the north of the UK compared to high risk clusters located in the south of the country. A number of individual practices caused significant subclusters within the main clusters, leading to the proposal of conducting practice-based research to investigate practice-related factors that influence tissue biopsy submission. It was concluded that the histopathology databases provided data suitable for the epidemiological analysis of host-related risk factors for canine neoplasia. The findings of the studies suggest that future research should focus upon identification of factors within individual practices which directly influence the procurement of tissue for histopathological analysis, to accurately ascertain their effect upon the diagnosis of neoplasia in canine biopsies. Practice-based research may also provide insight into differences in geographical occurrence of canine neoplasia, which may lead to the generation of hypotheses regarding possible environmental factors contributing to the aetiology of canine neoplastic disease.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.401644  DOI: Not available
Share: