In this issue of the ISNCC Academic Express column, we highlight research led by Jianfei Xie and colleagues from China, published in the Journal of Nursing Management. The study applies machine learning to predict nursing interruption during medication administration, a key patient safety challenge in cancer care.
By identifying major workplace and personal risk factors and developing practical prediction tools, the research supports proactive risk management and safer medication practices. This study provides valuable insights for oncology nursing practice and contributes to improving the quality and safety of patient-centred cancer care.
Why We Chose This Research Topic?
Oncology nurses handle high-alert medications under immense cognitive load. As the final safeguard before patient contact, nurses intercept up to 86% of potential errors. However, 99% report being interrupted during medication tasks, with each interruption increasing error risk by 13%.
Despite these high stakes, current clinical management relies heavily on retrospective reporting. To bridge this gap, we developed a machine learning prediction model to shift from reactive monitoring to proactive risk stratification, empowering nurses and safeguarding cancer care.
What We Discovered?
Surveying 4,758 clinical nurses across 12 Chinese tertiary hospitals, we found that 52.1% experienced at least one NIMA event during their latest shift. Using machine learning algorithms, we identified 18 key predictors of NIMA. The Logistic Regression model performed best, achieving an AUC of 0.748.
Interestingly, the most significant predictors were interpersonal demands, specifically, the needs of doctors, patients, and patients' families. Furthermore, nurses' personal factors, such as high resignation intention and lower general self-efficacy, significantly increased interruption risks. To bridge the gap between complex algorithms and clinical utility, we translated our findings into an 18-variable nomogram and a simplified 5-variable interactive web-based calculator for rapid bedside assessment.
Implications for Cancer Care
Our findings offer highly actionable insights for oncology nursing:
1.Proactive Risk Management: Using our 5-variable web calculator, managers can conduct rapid bedside assessments to identify nurses at high risk of NIMA. This enables a crucial shift from reactive incident reporting to proactive risk management, allowing for targeted supervision and optimized staffing before errors occur.
2.Managing Interpersonal Demands: Since demands from doctors, patients, and families are primary drivers of NIMA, oncology wards must implement structural interventions. Establishing “no-interruption zones” during chemotherapy preparation and streamlining interdisciplinary communication can effectively shield nurses from non-urgent requests, ultimately protecting their cognitive focus.
3.Supporting Nurses’ Well-being: Oncology nursing carries a profound emotional burden. The strong link between turnover intention, self-efficacy, and NIMA proves that patient safety is deeply intertwined with psychological health. By providing targeted psychosocial support and addressing oncology-specific burnout, managers can foster professional confidence, reduce medication errors, and retain invaluable staff.
By leveraging data-driven tools, we can create safer work environments, protect our oncology nurses from cognitive overload, and ultimately ensure the highest standard of care for our patients.
Reference:
Dong, X., et al. Identifying Nurses at Risk of Nursing Interruptions During Medication Administration Using Machine Learning: A Multicenter Cross-Sectional Study. Journal of Nursing Management, 2026, 4433675.
Jianfei Xie (Corresponding Author)
