WEB-BASED RADIOTHERAPY INCIDENT ANALYSIS SYSTEM

Abstract

Objective: A web-based radiotherapy incident analysis system was developed and tested for safer radiotherapy implementation. Radiation dose incidence reporting is a known evaluation method for learning from the errors that occur during the radiotherapy procedure. Established as a safety-critical, non-punitive, just-culture system, the waterfall model was employed in the construction process of the system to identify and learn from incidents, non - conformance, and near-misses in radiotherapy settings. Method: The theoretical framework of the thesis was based on the Systems-Theoretic Accident Model and Processes (STAMP). The system algorithms were designed to identify sixty-two (62) radiotherapy errors. Results: The results of system implementation require patient test data that was selected based on the PRISMA 2009 method. Records identified through the radiotherapy manual database were 4479 patient data set. The system reported 1215 treatment sessional errors which are equivalent to 219 patient errors when analyzed with simple descriptive statistics. Incident data were identified directly by the system, in terms of incident level, form and patient incident, year dependent, site-specific, primary site incident, and treatment status. Frequency of error types were 10% incidents, 85% non-conformance and 5% near-misses. Patient error types identified 58.447% incidents, 13.699% non-conformance and 27.854% near-misses. Conclusion and Recommendation: Treatment status gave an overview of the quality of clinical decisions and implementation in the management of the patient. In future iterations, error tagging and solution recommendation parts with supervised machine learning algorithms would be made available to show the types of errors captured and chances of mitigating risks in terms of percentages for incident learning.