Shifts in Acute Kidney Injury Trends Post-Alert System Implementation: A Retrospective Observational Cohort Study

Rationale & Objective: Acute kidney injury (AKI), a complex syndrome involving multiple organ systems, requires timely intervention to prevent adverse outcomes. Despite the promise of AKI alert systems in improving early detection and management, their long-term effectiveness remains unevaluated. This study aimed to investigate the sustained impact of an AKI alert system over a decade of implementation. Study Design: Retrospective observational cohort study. Setting & Participants: Conducted at a Korean tertiary hospital, analyzing 13,301 adult patients with AKI, from 246,650 admissions (2013-2023). Exposure: Introduction of an automated AKI alert system (implemented June 2014). Outcomes: Primary outcomes included 14-day AKI recovery, stage progression, dialysis onset, and mortality. Process outcomes comprised early nephrology consultation, overlooked AKI cases, and discontinuation of nephrotoxic medications. Long-term outcome was assessed as ≥35% estimated glomerular filtration rate (eGFR) decline within 3 years. Analytical Approach: Outcomes were compared across pre-alert (January 2013-May 2014), alert implementation (June 2014-December 2019), and pandemic (January 2020-December 2023) phases, using adjusted multivariable logistic analyses calculating unadjusted and covariate-adjusted odds ratios (ORs). Interrupted time series analysis evaluated temporal trends. Results: AKI recovery rates increased substantially in both alert implementation and pandemic phases (adjusted OR 2.40 [95% confidence interval (CI) 2.16-2.66] and 2.19 [1.94-2.47], respectively). Early nephrology consultation improved significantly (adjusted OR 3.33 [2.81-3.95] and 2.65 [2.20-3.19]), and overlooked AKI cases decreased (adjusted OR 0.49 [0.43-0.55] and 0.43 [0.36-0.51]). Long-term renal outcomes showed lower risk of significant eGFR decline during the implementation phase (adjusted OR 0.86 [0.76-0.99]). Interrupted time series analysis showed immediate improvements in recovery rates, early consultation rates, and overlooked AKI cases following alert implementation. Limitations: Single-center, retrospective design limits generalizability. Patient characteristic changes may confound results. Conclusions: The AKI alert system achieved sustained improvements in AKI recovery, early nephrology consultation, and AKI monitoring, over a decade, demonstrating long-term effectiveness despite evolving patient populations and health care environments.