Electronic Alerts for Acute Kidney Injury: Retrospective Comparison of KDIGO- and AACC-Based Criteria for In-Hospital...

Electronic alert systems (e-alerts) based on blood creatinine changes are used to detect acute kidney injury (AKI) in hospitalized patients. While most e-alerts follow the KDIGO (Kidney Disease: Improving Global Outcomes) definition for AKI, the Association for Diagnostics & Laboratory Medicine (ADLM, formerly AACC) has proposed alternative AACC criteria. This study directly compares e-alerts using both AKI definitions in relation to in-hospital outcomes.

E-alerts based on KDIGO and AACC criteria were retrospectively assessed over 1 year in hospitalized patients at a tertiary care hospital. Adverse outcomes included in-hospital death, initiation of dialysis, and renal function decline at hospital discharge according to acute kidney disease criteria.

Among 14 719 patients, 1304 had adverse outcomes. KDIGO-based e-alerts were fired in 2598 patients of whom 2554 also met AACC criteria, 1138 were detected by AACC-based e-alert alone. AACC-based alerts were more sensitive (81.4% vs 74.0%) but less specific (80.4% vs 87.8%) for the composite adverse outcome. KDIGO but not AACC-based e-alerts showed reduced sensitivity in female patients and patients without chronic kidney disease. AACC-based alerts triggered earlier than KDIGO-based e-alerts in 22% of dual-alert cases. This subgroup had a higher rate of adverse outcomes (42.7%) than those with e-alerts firing simultaneously (35.1%).

AACC-based e-alerts improve early AKI detection and reduce bias, potentially enhancing risk stratification. However, they also increase alert burden. These findings provide a basis for future prospective comparative studies.