Systemic immune inflammation index and peripheral blood carbon dioxide concentration at admission predict poor prognosis in patients with severe traumatic brain injury.

Background: Recent studies have shown that systemic inflammation responses and hyperventilation are associated with poor outcomes in patients with severe traumatic brain injury (TBI). The aim of this retrospective study was to investigate the relationships between the systemic immune inflammation index (SII = platelet × neutrophil/lymphocyte) and peripheral blood CO concentration at admission with the Glasgow Outcome Score (GOS) at 6 months after discharge in patients with severe TBI.

Methods: We retrospectively analyzed the clinical data for 1266 patients with severe TBI at three large medical centers from January 2016 to December 2021, and recorded the GOS 6 months after discharge. The receiver operating characteristic (ROC) curve was used to determine the best cutoff values for SII, CO, neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR), and lymphocyte to monocyte ratio (LMR), and chi-square tests were used to evaluate the relationships among SII, CO and the basic clinical characteristics of patients with TBI. Multivariate logistic regression analysis was used to determine the independent prognostic factors for GOS in patients with severe TBI. Finally, ROC curve, nomogram, calibration curve and decision curve analyses were used to evaluate the value of SII and coSII-CO2 in predicting the prognosis of patients with severe TBI. And we used the multifactor regression analysis method to build the CRASH model and the IMPACT model. The CRASH model included age, GCS score (GCS, Glasgow Coma Scale) and Pupillary reflex to light: one, both, none. The IMPACT model includes age, motor score and Pupillary reflex to light: one, both, none.

Results: The ROC curves indicated that the best cutoff values of SII, CO, PLR, NLR and LMR were 2651.43×10, 22.15mmol/L, 190.98×10, 9.66×10 and 1.5×10, respectively. The GOS at 6 months after discharge of patients with high SII and low CO were significantly poorer than those with low SII and high CO. Multivariate logistic regression analysis revealed that age, systolic blood pressure (SBP), pupil size, subarachnoid hemorrhage (SAH), SII, PLR, serum potassium concentration [K], serum calcium concentration [Ca], international normalized ratio (INR), C-reactive protein (CRP) and co-systemic immune inflammation index combined with carbon dioxide (coSII-CO) (P < 0.001) were independent prognostic factors for GOS in patients with severe TBI. In the training group, the C-index was 0.837 with SII and 0.860 with coSII-CO. In the external validation group, the C-index was 0.907 with SII and 0.916 with coSII-CO. Decision curve analysis confirmed a superior net clinical benefit with coSII-CO rather than SII in most cases. Furthermore, the calibration curve for the probability of GOS 6 months after discharge showed better agreement with the observed results when based on the coSII-CO rather than the SII nomogram. According to machine learning, coSII-CO ranked first in importance and was followed by pupil size, then SII.

Conclusions: SII and CO have better predictive performance than NLR, PLR and LMR. SII and CO can be used as new, accurate and objective clinical predictors, and coSII-CO, based on combining SII with CO, can be used to improve the accuracy of GOS prediction in patients with TBI 6 months after discharge.