Prognostic associations of urine output in AKI and dialysis vintage in ESKD at continuous kidney replacement therapy initiation in ICU patients

Background Prognostic assessment at continuous kidney replacement therapy (CKRT) initiation remains challenging in ICU practice. CKRT recipients are heterogeneous; most have acute kidney injury (AKI), whereas a clinically important subset has end-stage kidney disease (ESKD). We evaluated whether pre-CKRT urine output in AKI and dialysis vintage in ESKD were associated with short-term prognosis after CKRT initiation. Methods We conducted a single-center retrospective cohort study of consecutive adults who started CKRT in a combined medical–surgical ICU between 2012 and 2021. Patients were classified at CKRT initiation as ESKD, oliguric AKI, or non-oliguric AKI. ESKD was defined as maintenance hemodialysis or peritoneal dialysis before ICU admission. Among patients with AKI, oliguric and non-oliguric AKI were defined as pre-CKRT urine output <0.5 and ≥0.5 mL/kg/h, respectively. The primary outcome was time to all-cause death within 90 days after CKRT initiation. We used Cox proportional hazards models adjusted for prespecified covariates measured at CKRT initiation. Within ESKD, we modeled dialysis vintage as a continuous variable using Cox models adjusted for APACHE II score and sepsis. Results Among 560 patients, 66 had ESKD, 305 had oliguric AKI, and 189 had non-oliguric AKI. Documented death within 90 days occurred in 236 patients (42.1%). Using ESKD as the reference, oliguric AKI was not associated with a different hazard of death within 90 days (adjusted hazard ratio [HR], 0.86; 95% confidence interval [CI], 0.58–1.29), whereas non-oliguric AKI was associated with a lower hazard (adjusted HR, 0.63; 95% CI 0.41–0.99). Among patients with ESKD, longer dialysis vintage was associated with a higher hazard of death within 90 days (adjusted HR per 1-year increase, 1.07; 95% CI 1.02–1.11). Conclusions In ICU patients starting CKRT, pre-CKRT urine output in AKI and dialysis vintage in ESKD were associated with short-term prognosis. Non-oliguric AKI was associated with a lower adjusted hazard of death than ESKD, and longer dialysis vintage was associated with a higher hazard of death within ESKD. These findings require external validation in contemporary multicenter cohorts.