It is a good analysis when…: Quality criteria for a reactive analysis following an adverse event

Although radiation oncology relies heavily on adverse event reporting to improve patient safety, the quality of retrospective analyses remains heterogeneous, often leading to ineffective action plans (retraining, simple reminders to remain vigilant). This article aims to propose criteria defining a high-quality reactive analysis based on the human, social, and organizational factors that determine its effectiveness, regardless of the investigation methodology used (ALARM, ORION, etc.).

Conducting an effective analysis requires, above all, a non-punitive "just culture" and impartial leadership by an expert trained in cognitive biases. The analysis process is broken down into three key steps: an objective and immediate gathering of facts by a neutral person; a multidisciplinary analysis meeting focused on systemic failures rather than individual blame, dedicating most of its time to developing robust corrective actions; and finally, monitoring the deployment. Furthermore, evaluating the effectiveness of these actions cannot be reduced to the simple absence of recurrence, particularly for rare events, and requires the implementation of proactive approaches such as targeted audits or unannounced "stress tests".

Ultimately, the value of an analysis depends more on the rigor and fairness with which it is conducted than on the specific tool selected. To sustainably improve safety, it is essential to move beyond technological performance alone and embed collective learning and patient safety within the organization's culture and identity. Direct integration of patients into these analytical processes represents the next challenge in advancing shared systemic resilience.