Analysis of clinical outcome risk factors in glioma

Abstract

In this paper, survival curves and Cox regression models were used to form an integrated disease risk factor model. Stepwise Cox regression models using the AIC criterion and the BIC criterion were included in the comparative models, penalties were imposed on the regression parameters for variable selection, and important covariates were selected to analyze censored data so as to analyze the dependence of dichotomous outcome events and survival time on the covariates. Applying the model of this paper to investigate postoperative infection and seizure risk in glioma, the results showed that 104 strains of pathogenic bacteria were detected. Gram-positive bacteria were the most prevalent infection-causing organisms with 67 strains, with Streptococcus grass-green and Staphylococcus epidermidis being the most common. There were 34 strains (32.69%) of gram-negative bacteria, and Acinetobacter baumannii was the most common one. 3 strains (2.88%) of fungi were present. Cox regression analysis was used to screen for five independent risk factors for surgical site infections in patients undergoing glioma surgery: neutrophil ratio, lymphocyte ratio, surgical grade (III), duration of antibiotic use, and postoperative day three temperature. The presence or absence of preoperative epilepsy has an impact on the prognosis of glioma patients, with fluctuating seizures being the most common postoperative condition in patients with grade II-III gliomas, which often recur after 12 months of seizure-free epilepsy. To sum up, the findings of this report are highly instructive in determining the prognosis for glioma treatment.

Keywords

• survival curves
• cox regression model
• postoperative infection
• seizures
• risk factor analysis
• 97p10