2421 - Clinical Drivers of Recurrent Failure Modes in Radiation Oncology Identified through a Retrospective Mini-FMEA CQI Framework
Presenter(s)
V. Chaswal1, R. P. Tolakanahalli2, A. Hart3, Y. Weiss4, L. LeGrand1, U. Ramos1, M. L. Chisem2, N. McAllister5, C. Tohtz5, V. Mishra2, L. Persaud1, C. Golden6, R. R. Fernandez5, C. Frau1, M. A. Valladares1, M. D. Chuong2, A. Wroe7, and A. Gutierrez4; 1Miami Cancer Institute, Baptist Health South Florida, Miami, FL, 2Department of Radiation Oncology, Miami Cancer Institute, Baptist Health South Florida, Miami, FL, 3Baptist Health South Florida, Coral Gables, FL, United States, 4Department of Oncological Sciences, Herbert Wertheim College of Medicine, Florida International University, Miami, FL, 5Miami Cancer institute, Baptist Health South Florida, Miami, FL, 6Miami Cancer Institute, Miami, FL, United States, 7Miami Cancer Institute, Miami, FL
Purpose/Objective(s): While RO-ILS enables event reporting, few studies demonstrate how structured continuous quality improvement (CQI) frameworks translate reported events into clinically actionable system insights. We evaluated nine months of CQI data generated using an in-house, retrospective mini–failure modes and effects analysis (mini-FMEA) framework to identify recurrent failure patterns and their clinical drivers.
Materials/Methods: A departmental monthly CQI process applied a retrospective mini-FMEA (focused on the near-vicinity of occurrence in workflow) to institutional RO-ILS submissions. Each event was analyzed to define a narrative (primary) failure mode (FM) and associated causal FMs. Risk Priority Numbers (RPNs) were assigned using standardized severity, occurrence, and detectability criteria by a multidisciplinary radiation oncology panel. FMs were categorized into core domains: process control, verification control, treatment planning, documentation, patient setup, and human factors. Recurrence, risk exposure, and downstream impact (medical event, replan, resimulation, treatment delay) were assessed. Corrective actions were classified by intervention type.
Results: Across nine months, 31,034 treatments (>300,000 processes) yielded 252 FMs (0.08%): 64 primary and 188 causal. Most FMs arose from process control vulnerabilities (45%) and verification control deviations (19%), followed by treatment planning (8%) and documentation (8%); human-factor–attributed deviations were uncommon (5%). Verification failures included missed checklist steps and incomplete imaging time-outs or setup evaluations. Patient setup deviations were infrequent but associated with high RPNs, including collision-risk and immobilization verification lapses leading to safety or therapeutic incidents. Corrective actions were primarily process improvements (58%) and education (18%), with engineered forcing functions (hard stops) rare (<2%). Risk-weighted analysis showed low-frequency upstream failures carried disproportionately high clinical and operational risk.
Conclusion: A structured retrospective mini-FMEA applied to RO-ILS data identifies clinically meaningful system vulnerabilities beyond isolated events. Failures were driven predominantly by workflow and verification gaps rather than operator error. The low FM rate (0.08% of >300,000 processes) reflects a robust baseline program, with CQI serving as a refinement layer that exposes residual high-impact vulnerabilities, particularly low-frequency upstream failures with disproportionate clinical and operational risk, supporting targeted system redesign including enforced verification steps and simulation-readiness constraints.