Introduction : Drug-induced QT interval prolongation is a well-recognized safety concern in hospitalized patients. Clinical decision support systems (CDSS) generate alerts to identify high-risk drug combinations, but their impact on clinical management remains uncertain.

Methods : This study evaluated clinical responses from prescribing clinicians to QT prolongation alerts, focusing on electrocardiogram (ECG) monitoring and treatment modifications. We conducted a 10-month prospective study in two hospital wards (acute geriatrics and orthopedics) at Geneva University Hospitals. The CDSS embedded within the electronic health record used a pharmacodynamic risk scoring system (Riskbase) that assigns scores to individual drugs based on pharmacological and clinical evidence and generates alerts when cumulative risk exceeds a predefined threshold. Clinical interventions were assessed within 7 days following high-risk QT alerts.

Results : A total of 154 patients were included (123 in geriatrics and 31 in orthopedics). Intervention rates ranged from 60.2% to 91.9% in geriatrics and from 45.2% to 80.6% in orthopedics, depending on the time window. However, these interventions were almost exclusively driven by ECG monitoring. Treatment modifications were rare, occurring in only 2.0% of patients. Most alerts were associated with combinations of low- or moderate-risk drugs, particularly antipsychotics, antidepressants, and antiemetics, some of which may have alternatives with lower QT-prolongation potential and may be considered depending on the clinical context. In addition, medications prescribed on an as-needed basis contributed substantially to QT risk scores, despite limited recent administration.

Discussion : In clinical practice, QT prolongation alerts are associated with frequent monitoring but rarely lead to therapeutic changes. These findings highlight a gap between risk identification and pharmacological risk mitigation and suggest opportunities to improve medication safety through more targeted prescribing and optimization of alert systems.