铁路职业人群行为生活方式、心理因素与糖尿病的网络分析

目的 本研究基于西南职业人群队列，采用网络分析方法厘清铁路职业人群行为生活方式、心理因素与糖尿病的关系。方法 本研究采用系统随机抽样，获得四川省、贵州省、重庆市50多个站点的成都局铁路职工作为参与者，采用电子问卷调查收集基本人口学特征、行为生活方式、心理因素等变量，结合体检指标获得参与者空腹血糖资料。采用混合图模型（mixed graph model, MGM）构建网络，以探索各特征因素与糖尿病疾病的联系。结果 研究共纳入25 198名参与者，其中1 264名（5.02%）诊断为糖尿病。构建的网络中，从基本特征看，年龄（边权：0.46）、性别（边权：0.37）、文化程度（边权：0.05）与糖尿病在网络中有关联；从行为生活方式看，吸烟（边权：0.07）、饮酒（边权：0.06）与糖尿病在网络中有关联。年龄是网络节点中度中心性（2.00）、接近中心性（0.005 3）、中介中心性（44）最高的节点，从行为生活方式和心理因素看，抑郁（DEP）的度中心性最高，为1.20，其次是吸烟的度中心性，为0.88。网络稳健性检验可见整体较为稳定，大部分边权之间存在差异性（P<0.005）。结论 研究...

Network analysis of lifestyle,psychological factors,and diabetes among railway occupational population in southwestern China

Objective To clarify the relationship between behavioral lifestyle, psychological factors, and diabetes among railway occupational population based on Southwest occupational cohort through a network analysis. Methods In this study, systematic random sampling was used to recruit railway workers at more than 50 stations of the Chengdu Bureau in Sichuan Province, Chongqing, and Guizhou Province. Electronic questionnaires were used to collect information including basic demographic characteristics, lifestyle, and psychological factors, in addition to physical examination indicators to obtain fasting blood glucose information from participants. A mixed graph model (MGM) was used to construct a network to explore the association between each factor and diabetic disease. Results A total of 25 198 participants were included in the study, of whom 1 264 (5.02%) were diagnosed with diabetes mellitus. In the constructed network, age (edge weight: 0.46), gender (edge weight: 0.37), and education level (edge weight: 0.05) were associated with diabetes in the network in terms of basic characteristics. Smoking (edge weight: 0.07) and alcohol consumption (edge weight: 0.06) were associated with diabetes in the network in terms of lifestyle. Age had the highest strength centrality (2.00), closeness centrality (0.0053), and mediation centrality (44) among all the nodes in the network. Depression (DEP) had the highest strength centrality (1.20), followed by smoking (0.88) in terms of lifestyle and psychological factors. Network robustness tests showed overall stability of the network, with variability between most of the edge weights (P<0.005). Conclusion Smoking, alcohol consumption, and depression are modifiable risk factors for diabetes. Older, men, and those who are illiterate are key populations prone to diabetes and early intervention should be targeted at risk factors and key populations to prevent the onset and progression of diabetes.