Risk identification and intervention of cold-chain pharmaceutical delivery in internet hospitals based on the HFMEA framework

Objective Cold-chain pharmaceutical delivery is an essential service for internet hospitals, yet it faces multiple latent safety risks under open-system third-party logistics. This study aimed to identify potential failure modes within a defined workflow scope: from pharmacy dispensing to patient signature confirmation, excluding upstream manufacturing logistics and post-receipt patient home storage, implement targeted interventions based on the Healthcare Failure Mode and Effect Analysis (HFMEA) framework, and explore potential risk-reduction outcomes associated with HFMEA-guided quality improvement. Methods A multidisciplinary eight-member expert panel followed the VA-NCPS HFMEA guidance. Severity (S) and Occurrence (O) were scored on a 1–4-point scale to calculate hazard scores (S × O); detectability was assessed qualitatively via a decision-tree model rather than incorporated into quantitative scoring. After one-year's implementation of corrective interventions, hazard scores were re-assessed. Objective monthly operational indicators were analyzed using the Mann–Kendall trend test. Results In total, 27 failure modes and 17 root causes were identified across dispensing, packing, delivery, and signature stages. The aggregated hazard score decreased from 243 to 73 after intervention, and the number of high-priority failure modes dropped from 25 to 12. Non-parametric Mann-Kendall trend analysis found statistically significant upward trends in both monthly pharmaceutical dispensing volume and cold-chain delivery order volume ( Z = 3.771, P < 0.001; Z = 3.634, P < 0.001, respectively), showing operational throughput expanded continuously over the observation period. Despite expanding operational throughput, both absolute event counts and exposure-adjusted safety rates exhibited statistically significant downward trends: monthly dispensing error count ( Z = −4.046, P < 0.001) and dispensing error rate ( Z = −4.046, P < 0.001); monthly abnormal delivery order count ( Z = −3.652, P < 0.001) and abnormal delivery rate ( Z = −3.771, P < 0.001). Conclusion This VA-NCPS-aligned HFMEA approach identified multi-link risks, with declines in expert-assessed hazard scores and favourable temporal changes in operational safety metrics observed within this single-centre before-after assessment. The standardized risk-control strategies provide practical references for internet hospital pharmaceutical quality management.