Study On Risk Factors Related To Postoperative Recurrence And Establishment Of Predictive Model Of Chronic Subdural Hematoma

Objective: According to the analysis of clinical and radiological information of patients with chronic subdural hematoma(CSDH),we found out the risk factors related to postoperative recurrence of CSDH.Based on these risk factors,a nomogram model was established to predict the recurrence of CSDH and validated.Methods: 514 patients were retrospectively analyzed,they were treated in Qinghai Provincial People’s Hospital between January 2010 and November 2017,and 231 patients were finally included in this study.We analyzed the clinical and radiological data of patients who met the inclusion criteria,screened out the risk factors related to the recurrence of CSDH by single factor analysis,and found out the significant related risk factors by multi-factor analysis.We used these factors to establish a nomogram model and validate it.Results: The total rate of recurrence after initial surgery for CSDH was 14.29%(33/231)during the following three months.The study found that larger preoperative hematoma volume(> 121 ml),postoperative residual cavity volume(> 72 ml),hiper-density and mixed-density hematoma(including laminar type,separated type,grading type and trabecular type)and older age(> 65 years old)were significantly related to recurrence(P < 0.05).We used 50% recurrence rate as the classification cut-off,with the corresponding points of 252 to validate the nomogram model.The accuracy of predicting the recurrence of CSDH calculated by the binary logistic regression model was 91.70%.The sensitivity and specificity of the nomogram were 87.88% and 84.85%,respectively.Conclusion: The study found that preoperative hematoma volume,postoperative residual cavity volume,computed tomography(CT)scan imaging type and age were significantly related to recurrence,and developed a nomogram model to predict recurrence.This nomogram model had a high precision to predict the recurrence of CSDH,and need more external and prospective validation in the future.We expect this model could be used in different neurosurgical problems as well.