C-Reactive Protein-to-Albumin Ratio and Hospital-Recorded Mortality in Patients with ICD-10-Coded Acute Kidney Injury: A Retrospective Cohort Study

Background: The prognostic value of the C-reactive protein-to-albumin ratio (CAR) relative to its components in acute kidney injury (AKI) is uncertain. Methods: We analyzed 7181 patient-date encounters (5744 patients; 2016–2025) identified using ICD-10 codes N17.x after exclusion of coded chronic kidney disease; AKI could not be verified against KDIGO criteria. The primary outcome was hospital-recorded death within 30 days, with 180-day mortality exploratory. Age- and sex-adjusted associations and discrimination of CAR, C-reactive protein (CRP), and albumin were evaluated in a common complete-marker sample, with sensitivity and patient-level bootstrap analyses. Results: CAR was available in 6603 encounters; hospital-recorded 30-day mortality was 9.6%. Each doubling of CAR was associated with higher odds of recorded death (adjusted odds ratio 1.25, 95% confidence interval 1.20–1.29). Areas under the receiver operating characteristic curve were 0.722 for albumin, 0.659 for CAR, and 0.642 for CRP. Separately modeled components outperformed CAR in adjusted models (0.737 versus 0.688; bootstrap 95% confidence interval for the difference, 0.034–0.065). CAR performance was attenuated in hospitalized-only analyses. Conclusions: CAR was associated with hospital-recorded mortality in code-identified AKI but did not outperform albumin. External validation is required; out-of-system deaths may have been missed. CAR should not be used as a stand-alone clinical decision tool.