RATIONALE & OBJECTIVE: Systematic evaluation of the prognosis from sepsis-associated acute kidney disease (SA-AKD) using real-world data is limited. This study aimed to use data algorithms on the electronic health records to trace the SA-AKD trajectory from acute kidney injury (AKI) to chronic kidney disease (CKD). STUDY DESIGN: A retrospective cohort study. SETTING & PARTICIPANTS: Adult inpatients with first sepsis episode surviving 90 days after AKD in a quaternary referral medical center. EXPOSURE: We defined SA-AKD as having sustained ≥1.5-fold increased serum creatinine levels or initiating kidney replacement therapy after the SA-AKI, and we classified SA-AKD into recovery, relapse, and persistent SA-AKD subgroups. OUTCOMES: All-cause mortality, kidney replacement therapy (KRT), ANALYTICAL APPROACH: A multivariable Cox proportional hazards models. RESULTS: Of 24,038 eligible inpatients with sepsis, 42.2% had SA-AKI, and 17.6% progressed to SA-AKD (43.6% recovery, 8.3% relapse, 32.2% persistent, and 15.9% unclassified). Compared with the recovery subgroup, the 1-year mortality risk for the relapse, persistent, and unclassified SA-AKD subgroups were 1.57 (adjusted hazard ratios [aHRs]
95% CI, 1.22-2.01), 1.36 (1.13-1.63), and 0.65 (0.48-0.89), respectively. Risks of KRT initiation were 3.27 (2.14-4.98), 6.01 (4.41-8.19), and 0.98 (0.55-1.74), respectively, and corresponding aHRs for LIMITATIONS: Selection bias and information bias could be present because of limiting population to sepsis survivors and because of no standardized follow-up protocol for kidney function. CONCLUSIONS: SA-AKD without recovery is associated with increased and long-term risks of KRT initiation, mortality, and increased risk of