233 - Implementation of a Radiation Oncology Incident Learning System in a Limited Resource Setting: Risk Analysis and Safety Culture

Purpose/Objective(s): Few reports address the use of incident learning systems (ILS) in low- and middle-income countries (LMICs), where quality improvement (QI) efforts face unique challenges. We hypothesize that a web-based ILS can support QI needs in an LMIC clinic, and that a commitment to continuous QI can overcome perceived barriers to ILS uptake.

Materials/Methods: A preliminary survey assessing safety culture and perceived barriers to ILS use was distributed to all oncologists, physicists, therapists, and nurses at a specialized radiation oncology clinic in sub-Saharan Africa. An ILS was implemented as a single-page web form based on the International Atomic Energy Agency Safety In Radiation Oncology (IAEA-SAFRON) taxonomy. The ILS was made available to all staff members in the clinic through a QR code. After seven months of ILS use, ILS reports were analyzed by type, clinic area, clinical role, and origin. The quality control quantification (QCQ) method was used to determine the QA intervention most likely to prevent or detect each error. 23 representative failure modes were abstracted from the ILS reports; an interdisciplinary team of clinicians used failure modes and effects analysis (FMEA) methodology to rank failure modes by risk.

Results: 16 responses to the preliminary survey indicated that “concern on the part of the reporter about their reputation or the effects of reporting a colleague” was a primary perceived barrier to incident reporting. In spite of perceived barriers, the ILS was actively throughout the seven-month accrual period, collecting eighty-four reports. Most were classified as near-misses (32, 38%), followed by operational inefficiencies (20, 24%), unsafe conditions (16, 19%), incidents (11, 13%), and minor or unavoidable events (5, 6%). Most reports were made by medical physicists (50 events, 59.5%), but therapists made the initial discovery of many events (31, 36.9%), some of which were ultimately reported by other users. Although most reports were entered from the LINAC control rooms (41, 48.8%), treatment planning was the most common origin of errors (28, 33.3%). The most frequently identified QA intervention that would have prevented errors was a comprehensive physics plan check (29 reports). Risk priority number (RPN) of failure modes ranged from 19 to 720. The highest-RPN failure mode was associated with a treatment delivery error related to the oncology information system (OIS). Treatment planning and hand calculation were associated with the most failure modes (N = 12, 52%; RPN range = [35, 720]).

Conclusion: ILS reports indicate that this clinic’s QI needs center on treatment planning and technical plan review, as well as OIS implementation. To our knowledge, this study is one of few to explore the use of ILS in the LMIC setting, and illustrates how a commitment to QI can overcome perceived barriers to ILS use.