Severity Evaluation And Prognosis Prediction Of Traumatic Brain Injury With Serum Uric Acid

Objective To investigate the value of serum uric acid in severity evaluation and prognosis prediction for traumatic brain injury.Methods A total of 330 patients of traumatic brain injury(TBI) were enrolled. Patients of TBI with GCS 3-14 included, while patients of age<14, pregnant, a history of autoimmune disorder or gout, no available data, excluded. Clinical data such as: Gender, Age, GCS, Pupil reaction, etc. were collected. The outcome of TBI was described by mortality and Glasgow outcome scale(GOS). The levels of serum uric acid and blood glucose were determined from the vein blood on the morning of the 2nd day after admission. The data were analyzed by logistic regression analysis, ROC curve, calibration curve and discriminant analysis. Patients were divided into two groups: survival group(GOS 2-5) and death group(GOS 1) in the logistic regression analysis, while three groups: good(GOS 4-5), poor(GOS 2-3) and dead(GOS 1) in the discriminant analysis.Results Spearman correlation analyses indicated that the level of serum uric acid was negatively correlated with GCS(r =- 0.2701, P = 0.000) and GOS(r =- 0.2512, P = 0.000). Age, GCS, pupil reaction and serum uric acid were determined as independent predictors of TBI mortality by logistic analysis. The adjusted OR of uric acidwas1.0048,(95%CI:1.0019-1.0076, P = 0.001); The area under the ROC curve: AUC = 0.718,(95%CI:0.666 to 0.766), the optimum cut-off value determined by the Youden index was 304 μmol/L(sensitivity: 60.24%, specificity: 78.14%, correctly classified: 73.64%). The R2 value of the core model(Age + GCS + Pupil reaction) was 0.4764, which could be increased to 0.5105(7.2% increased) after serum uric acid added into the model. The performance of serum uric acid was close to blood glucose for mortality prediction in discrimination(C statistics: 0.718 vs. 0.781) and calibration(P values of Hosmer and Lemeshow test: 0.151 vs. 0.337), respectively. Strong predictors included, the Fisher discriminant functions were: F1 =- 0.763- 0.432* Sex + 0.024* Age- 0.156*GCS + 0.962* Pupil reaction + 0.002* Uric acid + 0.072* Blood glucose,F2 =- 4.902 + 0.646* Sex- 0.004* Age + 0.314*GCS + 1.791* Pupil reaction + 0.004* Uric acid- 0.050* Blood glucose. 82.4% of original grouped cases correctly classified and 82.1% of cross-validated grouped cases correctly classified.Conclusions Serum uric acid in early stage of TBI is an independent predictor of mortality, and significantly correlated with severity. Serum uric acid has a close performance to blood glucose in discrimination and calibration for mortality prediction. More precise prediction model for TBI could be established based on discriminant analysis with strong predictors.