Analysis Of Risk Factors Of Lung Cancer And Pulmonary Nodules And Establish Of Risk Assessment Model In Hebei Province

Part one Distribution characteristics and risk factors of pulmonary nodules in LDCT screening in Hebei ProvinceObjective: The pulmonary nodules detected by LDCT screening in Hebei Province were described and the related risk factors were analyzed in order to provide basic data for prevention and treatment of lung cancer.Methods: The Fourth Hospital of Hebei Medical University used LDCT for screening lung cancer in high-risk populations in high-risk areas of Hebei Province from January 2014 to December 2017.At the baseline assessment,participants completed questionnaires and physical examinations,included LDCT scan.The data of all participants' questionnaires and the first LDCT scans were collected for statistical analysis.Logistic regression analysis was used to calculate the odd ratios and 95% confidence intervals(95% CI)for each factor.The population attributable fraction(PAF)was also calculated.Results:1.Distribution of pulmonary nodule type and location: A total of 8929 participants were included in this study and 2786(31.20%)had at least one pulmonary nodule.In pulmonary nodule type,the most pulmonary nodules were solid nodules(<5mm)had 1644 case(17.29%);70(0.78%)and 228(2.55%)were part-solid nodules and GGNs,respectively.The number of suspicious malignancy cases was 111(1.24%).Solitary pulmonary nodules were most common in the upper right lobe(21.68%);the proportion of multiple pulmonary nodules accounted for 30.58% of the total number.2.Analysis of risk factors of pulmonary nodules: Multivariate analysis found that age,gender,smoking,SHS exposure at home,using mosquito repellent,cooking oil fume,mental status,vegetables and fruits,bean and whole grains food,pickled foods,fried foods,hot food,taste,alcohol drinking,drinking tea,occupational exposure,history of lung disease and family history of cancer were associated with pulmonary nodules.Compared with less than 40 years old,people who over 70 year old was 3.826 times in the incidence of pulmonary nodules(PAF = 43.31%);the OR of cooking oil fume was 1.585(95% CI: 1.287,1.952).The PAF of smoking and SHS exposure at home were 13.07% and 4.07%,respectively;3.60% and 4.70% of pulmonary nodules attributed to history of lung disease and family history of cancer.However,vegetables(OR=0.763;95%CI: 0.650,0.894)and fruits(OR=0.779;95%CI: 0.679,0.895)consumption ? 7 times/week maybe could decreased the incidence of pulmonary nodules;Beans,whole grains food and drinking tea(OR=0.783;95%CI: 0.705,0.869)also could reduce the incidence of pulmonary nodules.Part two Establishment of risk assessment model for pulmonary nodule occurrence and developmentObjective: By establishing a population-based prospective cohort study,the influencing factors of pulmonary nodules occurrence and development were explored.Methods: The Fourth Hospital of Hebei Medical University established a prospective study cohort for the screening of lung cancer in January 2014.LDCT was used for lung cancer screening in high-risk populations in high-risk areas of Hebei Province.We followed up according to the Fleischner Society 2017 Guidelines.The follow-up deadline was December 2017.In this section,we collected information from participants who had undergone more than two LDCT examinations.The study used flexible parametric survival model to estimate baseline hazards and hazard ratios of pulmonary nodule progression.Results:1.Changes in risk of pulmonary nodules: A total of 3479 participants participated in at least two LDCT examinations.The number of cases that the first scan was positive and the last scan also positive was 58(1.70%)cases;2673(86.00%)cases were negative at the first and last scan.Compared with patients who had negative results at the first scan,patients with?1 indeterminate pulmonary nodules(OR=12.260;95%CI: 9.631,15.606)and ? 1 positive pulmonary nodule(OR=28.629;95% CI: 18.403,44.539)were more likely to have non-negative results in the last scan.2.Pulmonary nodule occurrence and development related risk factors: A total of 571 participants had pulmonary nodule enlargement or new pulmonary nodules.By establishing the flexible parametric survival model,we found that smoking(HR=1.334;95%CI: 1.051,1.693),kitchen fume exposure(more: HR=1.826;95%CI: 1.197,2.785),fried food(HR=1.709;95%CI: 1.202 2.430),occupational exposure(HR=1.304;95%CI: 1.076,1.579),history of lung disease(HR=1.269;95%CI: 1.019,1.580)and family history of cancer(HR=1.134;95%CI: 1.030,1.180)may increase the risk of pulmonary nodule progression.Regular exercise(HR=0.828;95%CI: 0.695,0.987),fresh fruit intake and tea consumption(HR=0.512;95%CI: 0.428,0.612)may reduce the risk of pulmonary nodule progression.3.Flexible parametric survival model assessment: By comparing the two models,we found that the influencing factors of the two methods were basically the same,but the factors of the family history of cancer were included in flexible parametric survival model.The covariate selected by the two models was ranked from small to large base on ? value,the differences were mainly concentrated on the factors that had a relatively little impact on pulmonary nodule occurrence and development.Through comparison of the standard errors of the covariates of the two models,the most of standard error of the covariates in flexible parametric survival model was smaller than that of the Cox regression model,which indicates that flexible parametric survival model in parameter evaluation is better than the Cox regression model.Part three A model to identify high risk population in Hebei ProvinceObjective: Established a screening predict model of screening high-risk populations of lung cancer suitable for use in the region to concentrate high-risk populations,increase screening positive rates,and reduce input costs,providing a basis for population screening for lung cancer screening.Methods: The Fourth Hospital of Hebei Medical University used LDCT for screening lung cancer in high-risk populations in high-risk areas of Hebei Province from January 2014 to December 2017.We collected the data of all participants' questionnaires,LDCT scan results and pathological examination for statistical analysis.Logistic regression analysis was used to calculate the OR and 95% CI of each factor;using the ROC curve to evaluate the models;divided the participants into 10 evenly distributed risk categories to predict their risk.Results:1.Lung cancer risk factors: A total of 8929 participants were included in this part,and 73 patients were diagnosed with lung cancer.Stratified by gender,it was found that there were some differences.In males,Smoking(OR=3.173;95%CI: 1.324,7.606),home decoration(OR=9.896;95%CI: 1.299,75.424),fried food(OR=3.568;95%CI: 1.048,12.151)could increase risk of lung cancer.In women,SHS exposure(OR=1.998;95%CI: 1.027,3.889),kitchen fumes exposure(OR=3.163;95%CI: 1.143,8.856),family history of cancer(OR=3.197;95%CI: 1.652,6.185)would increase the risk of lung cancer,and regular exercise can reduce the risk of lung cancer.2.Comparison of predictive models and effectiveness of lung cancer screening: Compared with whole population model with gender stratified model,in man: AUCfull model=0.906(95% CI: 0.854,0.958),AUCmale model=0.900(95% CI: 0.841,0.959),P=0.569,the difference was not statistically significant;in women: AUCfull model = 0.850(95% CI: 0.785,0.915),AUCfemale model = 0.856(95% CI: 0.789-0.922),P = 0.667,and the difference was also not statistically significant.Using the whole population to predict the risk of lung cancer,the AUC was 0.884(95% CI: 0.838,0.920),which was better than the simple predictive model(AUCsimple model=0.680;95%CI: 0.620,0.740,P<0.001).In this study,123 subjects needed to screen to detect one lung cancer.If the predictive model was used to screen high-risk individuals,screening 98 people could find a case of lung cancer.The effect of lung cancer screen can be improved by 30.89%.Conclusions:1.Age,smoking,exposure to SHS,dietary factors,occupational exposures,disease history and family history of cancer can affect the occurrence and development of pulmonary nodules and lung cancer.Therefore,we propose to form a health habit of life,regular physical exercise,quit smoking,intake fresh fruits and vegetables,and reduce the consumption of unhealthy food;for those who have basic diseases such as cardiovascular diseases and lung disease,should active treatment to prevent the occurrence and development of pulmonary nodules and lung cancer.2.Compared with the classical Cox regression model,the risk assessment model established by the flexible parametric survival model is true and reliable,and its parameter estimation is better than the Cox regression model.3.Compared with the simple model,all population models can better predict the risk of lung cancer.Using the established lung cancer risk assessment model,the lung cancer detection rate can be increased by 44.71%.