Development And Validation Of A Prediction Model For The Risk Of Neurological Deterioration In Moderate Traumatic Brain Injury

Objectives1.This study aimed to analyze the clinical characteristics and risk factors for early neurological deterioration(ND)in moderate traumatic brain injury(mTBI)and to develop and validate a prediction model for mTBI-ND based on a clinical cohort.To provide a basis for early clinical diagnosis and intervention in ND.Provide a rationale for early clinical diagnosis and intervention.2.Another purpose of this study was to explore the indications for intracranial pressure(ICP)monitoring in mTBI patients with(Glasgow Coma Score,GCS)9-11 points and its impact on the incidence of early ND and long-term neurological outcomes in mTBI patients.Methods1.The subjects of part 1 were patients with mTBI who met the inclusion and exclusion criteria from January 2015 to December 2020 at Tangdu Hospital of Air Force Military Medical University and Yan’an People’s Hospital.Baseline information on patients’demographic and clinical characteristics at admission was collected.With ND as the outcome event within seven days,variables were screened by univariate,multifactorial logistic regression,and least absolute shrinkage and selection operator(Lasso)regression,combined with a clinical background,literature review,stability,and accessibility of variables to identify candidate variables.The final clinical prediction model was established by multivariate logistic regression,and the model was presented using Nomogram and website.The model was validated internally by C-Index and a calibration plot,and its clinical utility was evaluated by clinical decision curve(DCA)and impact curve(CIC).Finally,the model’s predictive performance is again validated using an external validation cohort.2.The population of part 2 were patients with mTBI with GCS scores of 9-11 who met the inclusion and exclusion criteria from January 2015 to December 2021 in the Neurological Intensive Care Unit(N-ICU)Tangdu Hospital,Air Force Military Medical University.Information on patients’demographics,clinical characteristics at admission,and treatment records was collected.Practice differences in ICP management interventions were assessed using the ICP Treatment Intensity Level(TIL)score.Mortality at six months,Glasgow Outcome Scale Expanded(GOS-E)scores,and ND incidence were used as outcomes.Univariate,multivariable logistic regression and subgroup analyses explored indications for ICP monitoring and the impact on outcomes.Results1.ND occurred in 135 of 479 patients with mTBI(28.2%)in part 1.The presence of a history of hypertension,Marshall CT classification,a company of subdural hemorrhage(SDH),location of contusion,injury severity score(ISS)>13,D-dimer level>11.4 mg/L,GCS≤10,and blood platelets(PLT)≤152×10~9/L were independent risk factors for the development of early ND in patients with mTBI.The logistic model was developed with the above predictors and presented as a Nomogram and a website.Internal validation using the Bootstrap method showed that C-Index was 0.881(95%CI,0.849-0.909).External validation in another cohort showed an area under the receiver operating characteristic curve of 0.827(95%CI,0.763-0.880).The calibration curve,DCA and CIC showed that the model had good discrimination,calibration,and clinical utility.2.For patients with GCS 9-11 mTBI,GCS≤10,midline shift≥2.5mm,and SDH were independent risk factors for ICP monitoring.GCS score,midline shift,ICP monitoring((Odds ratio,OR)=0.815(95%CI,0.712-0.933),Marshall CT classification,SDH,and location of brain contusion were associated with poor 6-month neurological outcome.Adjunctive treatment with ICP monitoring was negatively related to the occurrence of early ND in mTBI patients(χ2=11.375,P=0.010),as was the intensity of treatment for intracranial pressure management(TIL score),which was negatively associated with the incidence of early ND and poor 6-month neurological outcome.mTBI patients with GCS scores of 9-11 had a mean ICP>15.8 mm Hg within 72 hours post-injury,or the frequency of peak ICP>15 mm Hg was more than 25.5 times or the frequency of peak ICP>20mm Hg more than 6 times were associated with poor 6-month neurological outcome.Conclusions1.The mTBI-ND prediction model was constructed by hypertension,Marshall CT classification,SDH,location of contusion,ISS>13,D-dimer>11.4 mg/L,GCS≤10,and PLT≤152×10~9/L,had good discrimination,calibration,and clinical applicability.This predictive model was the first tool that provided a new method for early risk assessment of ND and gains time for early intervention in ND.However,it still needs to be validated in multicenter,prospective,and large-sample studies.2.mTBI patients with GCS≤10,midline shift≥2.5mm,and the presence of SDH are recommended to accept ICP monitoring.ICP-monitored assistant therapy is negatively associated with the incidence of early ND and 6-month poor neurological prognosis in mTBI patients with a GCS score of 9-11.Early abnormal ICP features were also associated with poor neurological outcomes at 6 months.The findings support ICP monitoring in severe mTBI patients,which may reduce the incidence of ND and improve the long-term neurological prognosis by upgrading the treatment level.This study provides new evidence for managing ICP in patients with mTBI.