Logistic Regression Equation Construction and Predictive Value of Influencing Factors of Postoperative Hypoproteinemia in Gastric Cancer Patients
Objective To explore the influencing factors of postoperative hypoproteinemia in patients with gastric cancer through Logistic regression,and establish regression equation to analyze its predictive value.Methods Retrospective analysis was conducted to collect clinical medical records of 102 gastric cancer patients who underwent radical gastrectomy.The incidence of postoperative hypoproteinemia was evaluated according to the albumin(ALB)level detected within 7 days after operation,and the patients were divided into the occurrence group(n=32)and the non occurrence group(n=70)according to the evaluation re-sults.The baseline data,clinical data and laboratory indexes of the two groups were compared.The influencing factors were ana-lyzed by Logistic regression,and the Logistic regression equation was constructed.The predictive value was discussed by receiver operating curve(ROC).Results The tumor diameter,the proportion of patients with diabetes and the total amount of crystalloid and colloid transfused during operation in the occurrence group were significantly higher than those in the non occurrence group(P＜0.05).Logistic regression analysis showed that tumor diameter,total amount of crystalloid and colloid infused during operation,and diabetes were the influencing factors of postoperative hypoproteinemia in gastric cancer patients(OR＞1,P＜0.05).The a-bove influencing factors were brought into the regression equation to obtain Logist(P)=-27.349+1.001 × tumor diameter+0.015 × total intraoperative infusion of crystals and colloids+1.300 × diabetes.The ROC curve of Logist(P)was drawn for the regression model,and the area under the curve(AUC)value was 0.838,the best cutoff value was 0.584,the sensitivity was 0.813,and the specificity was 0.771.Conclusion Tumor diameter,the total amount of crystalloid and colloid transfused during operation,and diabetes are the influencing factors of postoperative hypoproteinemia in patients with gastric cancer.The prediction model based on the above indicators has good predictive value.
NETL
NSTL
万方数据
