Analyzing Correlative Factors Of Hematoma Expansion In Intracerebral Hemorrhage And Developing A Prediction Score

Objective:1.To analyze the characters of patients with spontaneous intracerebral hemorrhage(ICH), and identify the correlative factors of hematoma expansion.2.To develop a practical prediction model and score for hematoma expansionfollowing ICH by multivariate logistic regression analysis.Materials and Methods:1.We prospectively registered the ICH cases admitted to the Neurology departmentin the Norman Bethune1st Hospital of Jilin University during December2012toDecember2013, and collected the demographic characteristics, clinical data, laboratorydata, CT characteristics and so on.2.The follow-up CT were taken24hours after onset or when the condition ofpatients got worse. The study used (V2-V1)/V1≥33%or V2-V1≥12.5ml as the definitionof hematoma expansion.3.We used SPSS17.0software for statistical analysis: By univariate analysis, weidentified the significantly correlative factors of hematoma expansion (p<0.05),measurement data with the Gaussian distribution using t test, measurement data withoutthe Gaussian distribution using Mann-Whitney U test, and enumeration data using2test. By multivariable logistic regression analyses, we found the independent predictors,and developed the prediction model. Value the variables on the basis of the odds ratio(OR). According to the the score of each case, we drawn the receiver operatorcharacteristic curve (ROC curve). And we assessed the predictive strength based on thearea under the ROC curve (AUC). We regarded the score when Youden’s index comes tobe the biggest as the best predictive value. Results:1.The study included308patients with ICH, and46cases (14.9%) had hematomaexpansion.2.By univariate analysis, there were significant differences (p＜0.05) in drinking,NIHSS score at admission, GCS score at admission, systolic blood pressure (SBP) atadmission, pulse pressure (PP) at admission, the time from onset to baseline CT, CTfeatures (bleeding position, baseline hematoma volume, hematoma shape irregularity,whether the hematoma breaking into ventricle), laboratory data (percentage ofneutrophils, serum creatinine, fasting blood-glucose) between patients with hematomaexpansion and patients without hematoma expansion.3.By multivariable logistic regression analyses, we found five independentpredictors of hematoma expansion: drinking (OR=2.543,95%CI1.239-5.220, p=0.011),NIHSS score at admission≥9(OR=2.477,95%CI,1.210-5.071, p=0.013), SBP atadmission≥180mmHg (OR=2.001,95%CI1.010-3.964, p=0.047), the time from onset tobaseline CT≤6h (OR=2.660,95%CI1.102-6.421, p=0.030), hematoma shape irregularity(OR=3.171,95%CI1.545-6.509, p=0.002).4.The prediction model:Logit (p)=0.933×A+0.907×B+0.694×C+0.978×D+1.154×E-4.207A, B, C, D, E respectively represents drinking, NIHSS score at admission, SBP atadmission, the time from onset to baseline CT, hematoma shape irregularity, for1or2.5.The prediction score(11-point):2points for drinking,2points for NIHSS score atadmission≥9,2points for SBP at admission≥180mmHg,2points for the time from onsetto baseline CT≤6h,3points for hematoma shape irregularity. According to the the scoreof each case,we drawn the ROC curve. The AUC is0.757with95%CI0.681-0.832,AUC standard error is0.038, and there has significant difference compared with thereference line (p=0.000). The best predictive cutting point is6.5, with a sensitivity of65.2%, a specificity of74%, a positive predictive value of30.5%, a negative predictivevalue of81.0%, and a total coincidence rate of82.5%. The prediction score≥7is closelyrelated to high risk of hematoma expansion (OR=6.125, p＜0.001). Conclusion:1.The incidence of hematoma expansion was14.9%.2.The research found five independent predictors of hematoma expansion as follow:drinking, NIHSS score at admission≥9, SBP at admission≥180mmHg, the time fromonset to baseline CT≤6h, hematoma shape irregularity.3.The research developed a practical11-point prediction score for hematomaenlargement, which can provide reference for the clinical identification of patients withhighest risk of hematoma expansion. The prediction score≥7is closely related to highrisk of hematoma expansion.