Abstract:
Objective To establish an individualized warning model for predicting death in early stage in patients with severe traumatic brain injury (sTBI) based on emergency laboratory indicators.
Methods Clinical materials of 249 sTBI patients treated through emergency pathway in Northern Jiangsu People′s Hospital from January 2020 to June 2022 were retrospectively analyzed. Based on the emergency laboratory indicators of the patients, the least absolute shrinkage and selection operator (LASSO) algorithm was used to screen the maximum correlation features of death in early stage of sTBI, and a binary Logistic regression model was furtherly used to establish an individualized emergency indicator line nomogram. The clinical values of the emergency laboratory indicators and the established line nomogram were evaluated by decision curve analysis.
Results The predictive values of the emergency laboratory indicators retained by the LASSO algorithm were as follows: the area under the curve (AUC) for base excess was 0.711, AUC for hemoglobin was 0.718, AUC for prothrombin time was 0.754, AUC for activated partial thromboplastin time was 0.804, AUC for fibrinogen was 0.656, AUC for D-dimer was 0.804, and AUC for transfusion of concentrated red blood cells was 0.796. The predictive ability of the nomogram established by the above seven laboratory indicators was higher (AUC of 0.975).
Conclusion The traditional emergency laboratory indicators can be used for early prediction of death in sTBI patients, and the accuracy of predicting death in early stage in sTBI patients by the nomogram established with the seven emergency laboratory indicators is high.
下载: