Abstract
<title>Abstract</title> <p>Background Acute kidney injury (AKI) in critically ill obstetric patients may reflect systemic decompensation, but evidence based on early and reproducible serum creatinine phenotypes across critical care databases remains limited. We evaluated the association between early serum creatinine-based AKI (Scr-AKI) and short-term adverse maternal outcomes among obstetric intensive care unit (ICU) patients. Methods We performed a multicentre retrospective cohort study using MIMIC-IV v3.1 (2008–2022) and the eICU Collaborative Research Database v2.0 (2014–2015). Female ICU patients aged 15–55 years with evidence of pregnancy or an obstetric diagnosis and sufficient serum creatinine data were included, retaining the first eligible ICU or unit stay. Scr-AKI was defined within 48 hours after ICU admission using Kidney Disease: Improving Global Outcomes serum creatinine criteria. The primary outcome was a composite adverse maternal outcome excluding renal replacement therapy (RRT), comprising in-hospital death, invasive mechanical ventilation, vasopressor use, and prolonged ICU length of stay. Multivariable logistic regression estimated adjusted associations, with database-stratified and sensitivity analyses. Results The primary cohort included 712 patients, 387 from MIMIC-IV and 325 from eICU. Scr-AKI occurred in 121 patients (17.0%), and 381 experienced the composite outcome. The outcome was more frequent with Scr-AKI than without Scr-AKI (75.2% vs 49.1%). In the pooled adjusted model, Scr-AKI was associated with the composite outcome (odds ratio 3.36, 95% confidence interval 2.14–5.29). Estimates were directionally consistent in MIMIC-IV and eICU, with no statistically significant AKI-by-database interaction. Results were stable after excluding early RRT and on risk-ratio and risk-difference scales. Urine-output-related analyses were limited by lower urine-output observability in eICU. Conclusions Among critically ill obstetric ICU patients, early Scr-AKI was associated with a concurrent short-term maternal severity and resource-use composite excluding RRT. Mortality components were sparse and should be interpreted descriptively. Scr-AKI may be a simple marker of concurrent maternal critical illness severity, but causal and predictive interpretation requires prospective, temporally anchored evaluation.</p>