|
Objective To investigate the impact of the major risk factors for postoperative complications of gastric cancer, the establishment of the major risk factors for the logistic regression model, and evaluation model in predicting postoperative complications sensitivity, specificity, and accuracy. This study is a retrospective study, A case - control study was conducted a retrospective survey of 650 patients with gastric cancer preoperative status, surgical approach, tumor pathological conditions common clinical observation indicators. First surveyed 60 clinical indicators univariate analysis, continuous variables using independent sample t-test (Independent-Sample t Test), two categorical variables and disorder variables using the chi-square test (Chi-squareTest), ordinal variables nonparametric test (Nonparametric Test NPar),. 16 variables significant then take a multitude of factors Logistic regression analysis using SPSS11.0 software for statistical analysis, P = 0.05 significant boundaries to establish Logstic regression equation: P = ExpΣB i < / sub> X i / 1 ExpΣB i X i , calculated the relative risk of various factors: OR = Exp (B), and to evaluate The model predicts postoperative complications, sensitivity, specificity, and accuracy. The results of various clinical indicators Univariate analysis showed that patient age, tumor size, intraoperative blood loss, operative time, Child-Pugh score, white blood cell count, albumin level, preoperative complications of portal hypertension, total gastrectomy, sweeping No. 16a lymph nodes, cleaning No.16b lymph nodes, combined with pancreatic body and tail and splenectomy, heart disease, chronic obstructive pulmonary disease, is closely related to the surgeon, physician surgical cases, lymph node dissection range 18 indicators and postoperative complications of gastric cancer. Postoperative complications as the dependent variable (0 = no, 1 = yes), these 18 single-factor analysis of indicators of significant input Backward Act binary multivariate logistic regression analysis, the results showed that a total of eight factors enter Logistic regression equation, according to the role the strength as follows: No.16b preoperative complications (OR = 2.933), and pancreatic body and tail and splenectomy (OR = 2.723), the No.16a group (OR = 2.674), lymph node dissection, lymph node cleaning (OR = 2.457), total gastrectomy (OR = 1.948), liver function Child-Pugh score (OR = 1.573), intraoperative blood loss (OR = 1.003), physician surgical cases (0R = 0.254), get Logistic regression prediction model: P = 1/1 ExpΣ (4.327-1.076X 1 -1.002X 2 -0.984X 3 -0.899X < sub> 4 -0.667X 5 -0.453X 6 -0.003X 7 1.369X 8 ). Probability value of 0.5 as a junction point, based on the predictive value of the table for comparison with actual data, the results show, this probability model to determine the accuracy of the postoperative complications of gastric cancer was 80.46% (523/650), and a sensitivity of 79.82% ( 87/109), a specificity of 80.59% (436/541). Conclusion Preoperative comorbidities, and pancreatic of splenectomy, 16a group lymph node dissection, 16b
|