Association of the C-reactive protein–to–albumin ratio (CAR) with clinical outcomes in chronic kidney disease patients undergoing surgery: a retrospective cohort analysis of the INSPIRE database

To evaluate the relationship between C‑reactive protein–albumin ratio (CAR) and clinical outcomes among perioperative patients with chronic kidney disease (CKD). This study retrospectively analyzed CKD patients who underwent surgery using data from the INSPIRE database, a single-center registry of perioperative clinical, laboratory, and outcome data from adult surgical patients at Seoul National University Hospital. Patients were stratified into quartiles on the basis of their CAR upon admission. Associations between CAR and outcomes were evaluated using adjusted Cox proportional hazards models (in‑hospital and 30‑day mortality), multivariable logistic regression (ICU admission), and a generalized linear model (length of hospitalization). The cohort comprised 2338 CKD patients, with an average age of 60.7 ± 14.8 years, and 65.8% were males. After controlling for possible confounding factors, an elevated log2-CAR was independently associated with in-hospital mortality (HR = 1.08, 95% CI 1.02–1.15; p = 0.009), with the highest quartile (Q4) showing a 1.99-fold higher risk than Q1 (p < 0.001). The secondary outcomes demonstrated similar trends: a higher CAR correlated with increased 30-day mortality (adjusted HR = 1.1, 95% CI 1.02–1.18; p = 0.012), ICU admission (adjusted OR = 1.05, 95% CI 1.01–1.09; p = 0.021) and a prolonged hospital stay (adjusted exp(β) = 1.06; 95% CI 1.05–1.08; p < 0.001). Subgroup and sensitivity analyses yielded consistent findings. An elevated CAR is positively associated with increased risks of in‑hospital death, 30-day mortality, ICU admission, and longer hospital stays in CKD patients who have undergone surgery. Our findings support the use of the preoperative CAR, a readily available, cost-effective marker, to inform risk stratification and guide perioperative care in patients with CKD. Not applicable.