石油工人肾功能异常危险因素分析及风险预测研究  

Risk factor analysis and risk prediction research of renal dysfunction in petroleum workers

作　　者：苏艳[1] 库雅军 凌鑫[1] 王慧[1] 王海忠 彭艳军 张红霞 SU Yan;KU Yajun;LING Xin;WANG Hui;WANG Haizhong;PENG Yanjun;ZHANG Hongxia(Occupational Disease Prevention and Control Institute of North China Petroleum Administration,Cangzhou 062550,China)

机构地区：[1]华北石油管理局职业病防治所职业卫生科,河北省沧州市062550

出　　处：《中国煤炭工业医学杂志》2024年第6期626-635,共10页Chinese Journal of Coal Industry Medicine

基　　金：科技部国家重点研发计划(编号:2016YFC0900605)

摘　　要：目的 探索导致石油工人肾功能异常的危险因素,建立石油工人肾功能异常的预测模型。方法 本研究采用队列研究的方法,纳入中国北方某石油公司油田工人2 292名作为研究对象。2017年9月通过问卷调查、体格检查、实验室检查收集了性别、年龄、文化程度、人均月收入、体质量指数(BMI)、血压、血糖、血脂、尿酸、吸烟情况、饮酒情况、饮食评分、高温暴露、噪声暴露、生产性粉尘暴露、苯系物暴露、倒班情况等基线资料,分别于2019年4月、2020年4月和2021年1月进行3次随访调查。根据肾小球滤过率判断肾功能异常,采用Cox分析石油工人肾功能异常的影响因素。选用logistic回归、随机森林和XG Boost分别建立石油工人肾功能异常风险预测模型,比较区分度、校准度和临床适用度等指标以筛选最优模型。结果 2 292名研究对象中,男性工人1 520人,女性工人772人,年龄分布M(P25,P75)为42(37,46)岁;其中肾功能异常的人数为562人,四年间的累积发病率为24.52%,发病密度为87.44千人年。Cox分析结果显示,性别(HR:3.017,95%CI:2.378~3.829)、年龄(HR:1.500,95%CI:1.236~1.820)、人均月收入(HR:0.546,95%CI:0.459~0.649)、血糖异常(HR:1.424,95%CI:1.113~1.822)、吸烟(HR:1.499,95%CI:1.200~1.872)、饮酒(HR:1.434,95%CI:1.164~1.767)、高温(HR:1.330,95%CI:1.099~1.609)、噪声(HR:1.425,95%CI:1.181~1.719)、苯系物暴露(HR:1.703,95%CI:1.418~2.045)和倒班(HR:1.060,95%CI:0.840~1.340)为石油工人肾功能异常的危险因素;模型验证集结果显示,随机森林模型的准确率、灵敏度、特异度、曲线下面积(AUC)、Brier分数、Log损失和校准截距表现最好,分别为80.11%、30.95%、96.73%、0.822、0.132、0.428和0.059,优于其它两种模型。结合三种模型受试者工作特征(ROC)曲线、校准曲线以及临床决策曲线,本研究表明随机森林模型具有最好的预测性能。结论 石油工人肾功能异常不仅受一�Objective To explore the risk factors that lead to abnormal renal function in petroleum workers and establish a predictive model for abnormal renal function in petroleum workers.Methods This study used a cohort study method and included 2292 petroleum workers from a petroleum company in northern China as the research subjects.In September 2017,gender,age,education level,per capita monthly income,BMI,blood pressure,blood glucose,blood lipids,uric acid,smoking status,alcohol consumption,diet score,high temperature exposure,noise exposure,productive dust exposure,benzene exposure,shift work status,etc.were collected as baseline data through questionnaire surveys,physical examinations,and laboratory tests.Three follow-up surveys were conducted in April 2019,April 2020,and January 2021.The abnormal renal function was determined based on the glomerular filtration rate,and the influencing factors of abnormal renal function in petroleum workers were analyzed using Cox regression.Logistic regression,random forest,and XG Boost were used to establish risk prediction models for abnormal renal function in petroleum workers,and indicators such as discrimination,calibration,and clinical applicability were compared to select the optimal model.Results Among the 2292 subjects,there were 1520 male workers and 772 female workers.The age distribution M(P 25,P 75)was 42(37,46)years old.Among them,the number of people with abnormal renal function was 562,with a cumulative incidence rate of 24.52%over four years and an incidence density of 87.44 thousand person-years.The Cox analysis results showed that gender(HR:3.017,95%CI:2.378-3.829),age(HR:1.500,95%CI:1.236-1.820),per capita monthly income(HR:0.546,95%CI:0.459-0.649),abnormal blood glucose(HR:1.424,95%CI:1.113-1.822),smoking(HR:1.499,95%CI:1.200-1.872),alcohol consumption(HR:1.434,95%CI:1.164-1.767),high temperature(HR:1.330,95%CI:1.099-1.609),noise(HR:1.425,95%CI:1.181-1.719),benzene exposure(HR:1.703,95%CI:1.418-2.045)and shift work(HR:1.060,95%CI:0.840-1.340)were risk factors fo

关 键 词：石油工人 肾功能异常 风险预测 随机森林 

分 类 号：R692[医药卫生—泌尿科学]