A Comparison of Incident Learning in the Academic and Community Radiation Oncology Setting

Abstract

Purpose/Objective(s): Incident reporting systems are now frequently used in academic centers to identify errors that can occur in a radiation oncology setting. It is important to understand how the patterns of such events depend on environment of care. A comparison of incidents reported in an electronic Variance Reporting System (VRS) from both academic and community centers since 2011 was performed. Materials/Methods: A total of 1057 incidents were reported into the VRS from 3 different locations in a large hospital system, including both academic and community centers, from February 2011 to November 2013. All incidents were scored for severity using the previously validated French Nuclear Safety Authority (ASN) 5-point scoring system. These incidents were further categorized by error type into four groups: (1) human only, (2) human-software interface, (3) human-human interface, and (4) software/ hardware only. The incidents were also categorized by the reporter and by the origin of the incident. These groups were analyzed using descriptive statistical analysis including chi-squared tests. Results: A total of 955 incidents were reported from two academic centers and 102 incidents were reported from the community setting. 31 incidents, 28 at academic centers and 3 in the community setting, were considered to potentially have negative clinical consequences with a severity score of 2. There were no incidents that surpassed a severity score of 2. The majority of incidents (428 incidents, 44.8%) occurring in academic centers were due to human only error; in contrast the majority (46 incidents, 45.0%) of incidents reported in the community setting were due to human/software interface error (p Z 0.034). The majority of incidents that occurred at academic centers were reported by dosimetrists (43.5%). 463 (48%) incidents occurred in the treatment planning stages, 258 (27%) occurred in the imaging stage of treatment delivery and 76 (8%) occurred between treatment review and verification. The majority of incidents that occurred at community centers were reported by therapists (56.8%). Thirty-eight (37%) incidents occurred in the treatment planning stages, 20 (20%) occurred in the initial patient assessment and 12 (12%) occurred in the imaging stage for treatment delivery. Conclusions: The majority of incidents at academic centers were linked directly to human error and potentially related to hand-offs. In contrast, the majority of incidents in the community setting was due to errors with the human/software interface and might be due to lack of education or experience with software. This raises the question of whether process deviations may differ depending on the environment of care. Identifying such factors may lead to improved workflow design in an effort to reduce incidents. Author Disclosure: S. Moningi: None. K. Smith: None. E.C. Ford: None. T.L. DeWeese: None. S.A. Terezakis: None.