Young Ae Lim, Chorong Park, Jin Kim
Following implementation, sATR and ATR detection rates peaked at 1.26% and 0.53%, but gradually declined over time, with further reductions during periods of healthcare system strain. EANS-generated alerts accounted for 94.4% of sATR reports (PPV 58.1%), whereas clinician-initiated reports demonstrated higher specificity (PPV 92.5%). In the focused analysis, only 1.5% of alerts were reported (SNR 1:78, one notification per five units transfused), yet 86.2% of reported alerts were confirmed ATRs after clinical triage. Nearly two-thirds of alerts were attributable to multiple vital sign entries or documentation errors. Body temperature-containing and multi-parameter alerts showed markedly higher rates of potential missed ATR than isolated triggers.
BACKGROUND: Automated surveillance systems such as the electronic automatic notification system (EANS) have been introduced to improve recognition of adverse transfusion reactions (ATRs), but their long-term sustainability and operational limitations remain incompletely defined.
STUDY DESIGN AND METHODS: We conducted a 6-year retrospective study (2019-2025) evaluating EANS performance at a tertiary hospital, analyzing quarterly suspected ATR (sATR) and confirmed ATR detection rates. Detection sources, positive predictive value (PPV), and reporting pathways were evaluated. During a focused 87-day period, all alerts were reviewed to assess signal-to-noise ratio (SNR) and vital sign-specific alert patterns.
RESULTS: Following implementation, sATR and ATR detection rates peaked at 1.26% and 0.53%, but gradually declined over time, with further reductions during periods of healthcare system strain. EANS-generated alerts accounted for 94.4% of sATR reports (PPV 58.1%), whereas clinician-initiated reports demonstrated higher specificity (PPV 92.5%). In the focused analysis, only 1.5% of alerts were reported (SNR 1:78, one notification per five units transfused), yet 86.2% of reported alerts were confirmed ATRs after clinical triage. Nearly two-thirds of alerts were attributable to multiple vital sign entries or documentation errors. Body temperature-containing and multi-parameter alerts showed markedly higher rates of potential missed ATR than isolated triggers.
DISCUSSION: Automated vital sign-based surveillance expands hemovigilance coverage but generates considerable alert burden potentially placing the system at risk of reduced reporting performance during prolonged use and periods of healthcare system strain. Continuous system optimization and vital sign specific alert prioritization strategies may be necessary to improve performance of automated hemovigilance systems.