Work

Risk factors for acute kidney injury in patients with traumatic brain injury: a single-centre retrospective cohort study with an exploratory external validation in MIMIC-IV

Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Background Acute kidney injury (AKI) is a frequent serious complication of traumatic brain injury (TBI) in the ICU, yet its modifiable risk factors are incompletely characterised. We aimed to identify independent AKI risk factors after TBI and test their reproducibility in an external cohort. Methods We retrospectively included 255 TBI patients admitted to the neurosurgical ICU of the Affiliated Hospital of Nantong University (March 2021–August 2024). An external cohort of 120 ICU patients with TBI was drawn from MIMIC-IV. AKI was defined by KDIGO criteria. Independent risk factors were identified by univariable and multivariable logistic regression (odds ratios per 1-SD increase) and their reproducibility was assessed in the external cohort; model discrimination was assessed in both cohorts. Results AKI occurred in 47.1% (120/255) of the local cohort and 24.2% (29/120) of the external cohort. In multivariable analysis, severe TBI (GCS ≤ 8; aOR 9.42, 95% CI 3.74–23.69), loop-diuretic use within 48 h (aOR 3.93, 95% CI 1.74–8.87) and elevated CK-MB (aOR 1.78, 95% CI 1.30–2.45) were positively associated with AKI, whereas greater first-day fluid balance (aOR 0.60, 95% CI 0.44–0.81) and mannitol use within 48 h (aOR 0.37, 95% CI 0.16–0.87) were inversely associated (all P < 0.05); neither 72-h magnesium nadir nor pupillary abnormality was significant. These associations were not replicated externally (all P > 0.05, including Firth penalisation); external patients were far less severely injured (10.0% vs 81.6% severe TBI [GCS ≤ 8]). An exploratory secondary prediction-model analysis likewise showed no external discrimination (AUC 0.531, 95% CI 0.399–0.662; see Supplementary Material), supporting non-replication rather than refutation. Conclusions In the local cohort, AKI after TBI was associated with severe TBI, loop-diuretic use and myocardial injury markers, with several fluid-management factors inversely associated. Neither the risk-factor associations nor the prediction model generalised externally, indicating centre specificity; multicentre prospective validation is warranted.
Institution: Nantong University
Language: English
Subfield: Nephrology
Field: Medicine
Open Access status: green
OpenAlex ID: W7236437106
Cites: 15
Cited by:
Related to: 10