Our study analyzed 831 non-dialysis chronic kidney disease (CKD) patients admitted from January 2007 to January 2023, using 665 for creating and 166 for validating the CKD-Infectious Inflammation (CKD-INF) score. This score distinguishes inflammation types using logistic regression outcomes and aligns with the International Sepsis Definition Conference criteria. The study’s participants had an average age of 76.2 ± 11.2 years, with 37% female and over half diagnosed with diabetes. The resulting score was as follows: \(\begin{gathered} {\text{CKD}} \text{-} {\text{INF}} = 1.8*{\text{CRP}}\left( {{\text{mg}}/{\text{L}}} \right) + 3.3*{\text{NLR}} + 7.9*{\text{WBC}}\left( {{\text{K}}/{{\upmu \text{L}}}} \right) \hfill \\ \quad \quad \quad \quad \quad \quad\;\;\; + 172.1*{\text{Fever}} \ge 38\,{^\circ }{\text{C}}\left( {{\text{no}} = 0,{\text{yes}} = 1} \right), \hfill \\ \end{gathered}\) where NLR refers to the neutrophil-to-lymphocyte ratio, CRP to C-reactive protein, and WBC indicates white blood cell count. CKD-INF exhibited a higher AUC of 0.85 (with a 95% CI of 0.82–0.87), higher sensitivity (75.3%) and specificity (82%) compared to its components including CRP. The CKD-INF score showed higher weighted κ values (0.57) for infection prediction at hospital admission than its individual components. In both training and validation groups, the CKD-INF score outperformed CRP as a predictor of infection in univariate and multivariate analyses. The CKD-INF score, derived from routine laboratory tests, can be an effective tool for predicting infections in CKD patients, potentially aiding in their early management.
