Applying epidemiologic concepts and principles to explore factors associated with error detection by radiotherapy quality assurance processes
MetadataShow full item record
Background: Incident Learning System (ILS) data are used by radiotherapy (RT) quality assurance (QA) programs for the purpose of improving quality and safety of RT delivery for patients. The objectives of this thesis were: 1) to develop an analytic framework for ILS data analysis that can be used to statistically explore factors that are associated with error detection throughout reported incidents; 2) to apply the analytic framework to identify and assess factors associated with error detection by RT QA processes using a real-world ILS dataset; and 3) to apply the framework in a pre-post study to explore the effect of various changes made to RT delivery on error detection by RT QA processes. Methods: We developed an analytic framework for ILS data analysis that combined relevant incident theory with core concepts and principles of quantitative research that are common to epidemiology. The analyzes utilizing the analytic framework were conducted on an ILS dataset from the Ottawa Hospital Cancer Centre (TOHCC), Ottawa, Canada. The TOHCC ILS dataset contained 4620 characterized incidents reported at the centre from 2007-2017. Epidemiologic methods were used to identify and assess factors that were associated with error detection by specific QA processes throughout the reported incidents. Results: The analytic framework was successfully piloted on the TOHCC dataset. The RT treatment technique, treatment intent and the error domain of origin were identified as factors that were associated with successful error detection by certain RT QA processes. Changes that were made to RT delivery process at TOHCC also appeared to have both positive and negative effects on error detection by certain RT QA processes. Conclusions: We developed and piloted an analytic framework that can be used to statistically analyze ILS data to explore factors associated with error detection by RT QA processes. We have shown that with careful interpretation, this analytic approach can be used to identify factors that may contribute to error detection by redundancy implemented throughout the RT workflow sequence.