Inflammatory Biomarkers and Delirium Severity in Critically Ill Adults: A Prospective Single-Center Cohort Study

Background: This study examined associations between C-reactive protein (CRP), interleukin-6 (IL-6), ordinal procalcitonin (PCT) categories, and delirium severity in critically ill adults. Secondary exploratory analyses evaluated whether baseline CRP and IL-6 discriminated subsequent incident delirium and remained associated with this outcome after adjustment. Methods: This prospective observational cohort included 267 adults treated in a general intensive care unit. Delirium severity was assessed using CAM-ICU-7 at 12 h after admission and on days 2, 4, 7, and 14. CRP and IL-6 were analyzed as continuous variables, whereas PCT was analyzed as a three-level ordinal variable across all five assessment times. Associations with delirium severity and intensive care unit length of stay were evaluated using Spearman rank correlations. Among patients without delirium at 12 h, bootstrap receiver operating characteristic analysis and Firth-penalized logistic regression adjusted for age and sex were performed as secondary exploratory analyses. PCT was not included in these prospective analyses because only ordinal PCT categories were available. Results: Delirium was recorded at least once in 38 of 267 patients (14.2%). Thirteen patients had delirium at 12 h, and 25 additional patients subsequently developed incident delirium. Concurrent correlations with CAM-ICU-7 severity ranged from 0.271 to 0.898 for CRP and from 0.312 to 0.735 for IL-6. Concurrent correlations for ordinal PCT categories were 0.304, 0.666, 0.711, 0.720, and 0.329 at the five respective assessment times. All concurrent associations were positive and statistically significant at p < 0.001. PCT categories also showed weak positive correlations with intensive care unit length of stay, with coefficients ranging from 0.151 to 0.281. Among 254 patients without delirium at 12 h, baseline CRP showed an AUC of 0.969 (95% bootstrap CI, 0.923 to 0.997), compared with an AUC of 0.653 (95% bootstrap CI, 0.569 to 0.750) for IL-6. Both baseline CRP concentrations and incident delirium were concentrated at floor values in most patients, and this discrimination should be read accordingly as a largely binary separation between floored and elevated CRP rather than a graded biomarker-severity gradient. In the joint Firth model, CRP remained associated with incident delirium, with an adjusted odds ratio of 10.75 per 5 mg/L increase, 95% confidence interval 4.50 to 33.18, p < 0.001. IL-6 was not associated with incident delirium after adjustment for CRP, age, and sex, with an adjusted odds ratio of 0.81 per 5 pg/mL increase, 95% confidence interval 0.42 to 2.04, p = 0.510. Conclusions: CRP, IL-6, and ordinal PCT categories were positively associated with concurrent delirium severity, although no consistent biomarker hierarchy was observed across assessment times. In secondary exploratory analyses, baseline CRP showed high apparent within-sample discrimination for subsequent incident delirium and remained associated with the outcome after adjustment for IL-6, age, and sex. However, the cutoff, discrimination, and effect estimates were derived from a single-center cohort with only 25 incident events. Model calibration, optimism-corrected performance, net clinical benefit, and clinical utility were not evaluated. These findings should not be interpreted as establishing a diagnostic test or clinical prediction tool and require validation in independent external cohorts before any clinical application.