| Background:Trauma is an instantaneous high-energy injury,and stress response,ischemia,hypoxia,and inflammatory response are the main pathophysiological changes in the early post-traumatic body.Among all types of trauma,the incidence of Traumatic brain injuries(TBI)ranks second only to limb injuries,but the case fatality rate ranks first,especially severe head injury,so far there is still a 30% case fatality rate,which has become a major problem affecting human survival and quality of life,and the impact on the survival and quality of life of all human beings far exceeds other diseases.Objective:Independent risk factors related to the 6-month neurological prognosis of TBI at different time points after admission were screened,the best cut-off value of each risk factor was calculated,and the prediction model of 6-month neurological prognosis was constructed according to the independent risk factors at different time points,and a flow chart was made to guide the treatment process of patients.Methods:TBI patients treated in the Department of Neurosurgery of the Second Affiliated Hospital of Air Force Military Medical University from January 2017 to December 2019 were collected.The basic information of the patient’s age,gender,injury mechanism,extracranial injury and other basic information was collected,and the patient’s C-reactive protein CRP,red blood cell distribution width RDW,procalcitonin PCT,sodium ion,chloride ion,potassium ion,blood glucose,blood glucose-potassium ion ratio,physical examination,etc.were collected.Computed tomography on the first admission and days 1,3,and 5 of admission.The patients were divided into two groups according to the Glasgow outcome scale GOS: 69 cases(4-5 points)in the good prognosis group and 106 cases(1-3 points)in the poor prognosis group.Univariate and multivariate logistic regression analysis were used to screen for independent risk factors related to the patient’s 6-month neurological prognosis,and prediction models at different time points were established respectively,the best cut-off values of each independent risk factor were calculated,and the False-positive rate FPR and False negative rate FNR corresponding to the best cut-off values of the corresponding indicators at different time points were obtained.Finally,a risk assessment flowchart for poor prognosis was produced.At the same time,the nomogram was drawn,and the Receiver operating curve ROC and Hosmer-Lemeshow goodness-of-fit test were used to verify and evaluate the prediction model.Results:1.The independent risk factors on the first day of admission were CT cerebral edema score,GCS score,and pupil condition.Among them,on the first day of admission,when the CT cerebral edema grade ≥ 3,the FPR with poor prognosis was0.232 and FNR was 0.245;when the GCS score was ≤ 6 points,the FPR with poor prognosis was 0.145 and FNR was 0.151;when there was unilateral mydriasis and dull or absent light reflex,the FPR with poor prognosis was 0.043 and FNR was 0.443;the sensitivity and specificity of the predictive model were 0.887,0.928,the prognostic FPR was 0.072,the FNR was 0.113,and the area under the ROC curve was 0.947.2.The independent risk factors on the 3rd day of admission were CT cerebral edema score,GCS score,and pupil condition,when the CT cerebral edema grade ≥ 2,the prognostic FPR was 0.478 and the FNR was 0.085;when the GCS score was ≤ 7 points,the prognostic FPR was 0.29 and the FNR was 0.075;when unilateral pupil dilation,slow or absent light reflex appeared,The FPR with poor prognosis was 0.014 and FNR of 0.358,the sensitivity and specificity of the prediction model were 0.821 and 0.957,and the FPR with poor prognosis was 0.043,FNR was 0.179,and the area under the ROC curve was 0.944.3.The independent risk factors on the 5th day of admission were RDW,blood glucose-potassium ratio,CT cerebral edema score,GCS score,and pupil condition,when the RDW ≥ 13.75%,Its poor prognostic FPR was 0.435 and FNR of 0.123;when the blood glucose-potassium ratio was ≤ 1.61,its poor prognostic FPR was 0.507 and FNR was 0.189;when CT cerebral edema grade ≥ 2,its poor prognostic FPR was 0.507 and FNR was 0.104;when the GCS score was ≤ 8 points,the prognostic FPR was 0.319 and FNR was 0.047;when unilateral pupil dilation,dull or absent light reflex appeared,The FPR with poor prognosis was 0.087 and FNR of 0.33,the sensitivity and specificity of the predictive model were 0.906 and 0.913,the FPR with poor prognosis was 0.087,the FNR was 0.094,and the area under the ROC curve was 0.95.4.Select the indicators with low false positive rate(low misdiagnosis rate)and low false negative rate(low missed diagnosis rate),first screen out as many patients with a poor prognosis as possible through the indicators with low missed diagnosis rate,and then further screen and evaluate the screened patients through the indicators with low misdiagnosis rate,so as to achieve the effect of low missed diagnosis rate and misdiagnosis rate,according to this principle,respectively,make the risk assessment flow chart of patients with adverse prognosis on the 1st,3rd and 5th days.Conclusions:1.The independent risk factors for poor prognosis of TBI patients on the 1st,3rd and 5th days after admission were screened,and after in-depth analysis,the change law of the optimal cut-off value of each independent risk factor at different time points was found,and the FPR,FNR and Area under receiver operating characteristic curve(AUC)of the prediction model at different time points were better than the single risk factor overall.2.By making risk assessment flow charts and nomogram of poor prognosis of patients on days 1,3 and 5,high-risk patients can be screened out and the prognosis of patients can be dynamically assessed and predicted. |
PDF Full Text Request