C-reactive protein/albumin ratio as a predictor of mortality in critically ill children: A single-center, prospective, cohort study
M. Sreya Nair, Vinitha Prasad, C. JayakumarAbstract
Background:
Critically ill children need early risk stratification to guide intervention and intensive care unit (ICU) resource use. Pediatric risk of mortality (PRISM) III is reliable but requires multiple variables. The C-reactive protein (CRP)/albumin ratio is simple and prognostic in adult critical care, but evidence in pediatrics is limited. This study evaluated the CRP/albumin ratio as a predictor of pediatric ICU (PICU) mortality and compared it with the PRISM III score.
Subjects and Methods:
This single-center, prospective, cohort study included children aged 2 months to 18 years admitted to the PICU. CRP, albumin, and PRISM III scores were measured within 24 h of admission. Patients were followed until discharge or death. Receiver operating characteristic analysis was used to assess performance, with cutoffs determined by the Youden Index. Multivariable logistic regression adjusted for age, sex, diagnosis, invasive ventilation, inotropic support, CRP/Albumin ratio, and PRISM III score.
Results:
Among 256 children, 38 died (14.8%). The CRP/albumin ratio showed good discrimination for mortality prediction, with an AUC of 0.770 (95% confidence interval [CI]: 0.694–0.847). At a cutoff of ≥7.56, sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 84.2%, 62.8%, 28.3%, 95.8%, and 66.0%, respectively. PRISM III had an AUC of 0.839 (95% CI: 0.768–0.911). In the sepsis subgroup, CRP/Albumin showed excellent performance, with an AUC of 0.917. On multivariable analysis, PRISM III, invasive ventilation, and inotropic support were independently associated with mortality; CRP/albumin was not.
Conclusions:
The CRP/albumin ratio showed good discriminative ability for mortality prediction in critically ill children but requires external validation before routine clinical use.