Creatinine and urine-output criteria and subsequent kidney progression in critically ill adults with cirrhosis: a two-database 24-hour landmark study with alternative weight definitions

Abstract Background Serum creatinine and urine output can identify different manifestations of acute kidney injury (AKI) in cirrhosis. We examined their joint association with subsequent kidney progression and the influence of the body-weight definition used to index urine output. Methods We studied adults with cirrhosis in MIMIC-IV and eICU-CRD who were alive and in the intensive care unit (ICU) at 24 h, had assessable kidney data, and had neither KDIGO stage 2/3 AKI nor kidney replacement therapy (KRT) during the first 24 h. Exposure was the number of KDIGO creatinine and duration-based urine-output criteria met during that period (0, 1, or 2). The outcome was first stage 2/3 AKI or KRT during the following 48 h (ICU hours 24–72). Database-specific Fine–Gray models accounted for ICU death and live discharge. Ideal body weight defined the primary analysis; actual-weight analyses reassessed both staging and eligibility. Results MIMIC-IV included 999 patients with 251 progression events; eICU-CRD included 674 patients from 129 hospitals with 83 events. Cumulative incidence with 0, 1, and 2 positive criteria was 19.0%, 32.7%, and 46.6% in MIMIC-IV and 10.2%, 20.0%, and 50.0% in eICU-CRD; the last estimate involved only 10 patients and five events. Adjusted summary subdistribution hazard ratios per additional criterion were 1.72 (95% CI 1.42–2.08) and 2.28 (1.49–3.48). After actual-weight eligibility reassessment, 763 and 616 patients remained, with estimates of 1.45 (1.19–1.76) and 2.82 (1.92–4.15). In MIMIC-IV, interval-specific estimates were 1.90 during ICU hours 24–48 and 1.39 during hours 48–72; the interval contrast was uncertain ( P = 0.135). Urine-output progression identified most first events in both databases. Conclusions Joint review of creatinine and urine output identified groups with different short-term progression incidence in these selected 24-h survivors. Classification, eligibility, and association magnitude depended on weight indexing and kidney-data ascertainment, which should be considered when interpreting the two criteria together.