Environment Factors and Risk Assessment Model of Gastric Cancer
|Keywords||Gastric Cancer Meta analysis Multi-factor dimensionality reductionmodel analysis Logistic regression analysis|
Background and AimsGastric cancer is one of the common malignant tumors in the world. Its low survival rate is serious harm to human health. In China, the mortality of gastric cancer is highest in gastrointestinal tumors, especially in developing countries. Etiology of gastric cancer is more complex, a large number of studies have been considered at home and abroad, the occurrence of gastric cancer is the result of the role of genetic factors and environmental factors. The aim of this study is to explore the risk factors of gastric cancer and set a screening risk assessment model.The model is widely used in the high-risk groups, which is not only to detect high-risk groups earlier,but also can take some interventions to improve their lives.Through the prediction of the model, people with higher gastric cancer risk can examine for medical treatment as soon as possible. And the prognosis and quality of the patients’life can be improved.ObjectiveThe purpose of this study is to investigate the risk factors of gastric cancer, establish a gastric cancer risk assessment model and propose prevention measures and recommendations, the sensitivity and specificity of the model were evaluated in order to provide some guideline for population screening, disease prevention and control.MethodsAll case-control studies on association of gastric cancer and risk factors were obtained. Literatures were collected comprehensively. Used REV Man4.2. to perform all statistic test and calculated pooled OR value(with95%CI) of GC.Using Stata8.0to perform heterogeneity test.A case-control study was conducted in Han population from Henan province, and249hospital cases and271normal controls were matching to cases by age and gender as objects. The questionnaire survey included personal habits, family history, past history and other information. Then multi-factor dimensionality reduction analysis was chose to filter out the best model, which contains different risk factors.Based on the results of the meta-analysis and mul-factor dimensionality reduction analysis, a logistic regression prediction model had been created.Diagnosis ed the muli-co-linearity of the independent variables. The area under the ROC curve had been evaluated. And we assessed the goodness fit of the model by residual analysis.Results1Meta analysis resultsCollecting24documents, in which14met the inclusion criteria. Risk factors odds ratio (OR) and95%CI were:gastric cancer family history:OR=2.73(1.76,4.25), history of gastrointestinal disease:OR=2.97(1.60,5.52), smoking:OR=1.85(1.39,2.47), drinking:OR=1.60(0.98,2.61).2Risk assessment model resultsSelecting the fifth-order model including gastric cancer family history, smoking, history of gastrointestinal disease, history of drinking consumption and HP infection, which was the optimal model as MDR model. The training set and test set were0.7226and0.7038, the cross-consistency was10/10.Smoking and drinking were multi-co-linearity.The total of three factors into the logistic regression, the correlation from strong to weak were as follows:hot food:(OR=2.97),95%CI is (1.56,5.67); gastric cancer family history:(OR=2.67),95%CI is(1.42,5.00); HP infection:(OR=1.74),95%CI is (1.11,2.73).The logistic regression model was P=1/1+Exp∑(19.81+21.29X4+1.09X6+0.55X8). The area of the ROC curve was0.72(0.67,0.76), sensitivity was62.2%, specificity was94.4%.Goodness fit of the model was:Deviance=11.34,P=0.18.Conclusion1Meta analysisFamily history of gastric cancer, history of gastrointestinal disease, smoking are risk factors for gastric cancer, there was no relationship between consumption and risk of gastric cancer.2Risk prediction modelThree factors were chose into the model, they were family history of gastric cancer, HP infection and hot food. The sensitivity of the model was62.2%, and the specificity of the model was94.4%,the area of the ROC curve was0.72. Prompting the model can provide a reference for high-risk population screening.