摘要
目的建立非酒精性脂肪肝(NAFLD)的发生风险预测模型,为NAFLD的预防及发生提供管理策略。方法选取2015年1月至2018年7月大连医科大学附属第二医院健康管理中心年度体检数据库中18~59岁、至少有2次连续体检记录、基线未发生NAFLD且无重要指标缺失者的数据,观察结局为NAFLD。收集基本信息、体格检查、实验室检查和腹部超声检查资料,将所有研究对象随机分为建模组和验证组。采用SPSS 23.0进行χ^(2)检验、t检验、秩和检验、单因素Cox回归分析。利用建模组资料进行多因素Cox回归分析选取预测指标,用RStudio软件绘制线图,构建NAFLD发生风险预测模型。通过一致性指数(C指数)和校正曲线对建模组和验证组模型的预测效果进行验证。结果本研究共纳入2377名研究对象,其中建模组1585人,验证组792人。本研究共有467人发生NAFLD(累积发病率为19.6%),平均随访时间为(27.06±8.02)个月。其中,建模组NAFLD发病人数为310人(发病率为19.6%),验证组NAFLD发病人数为157人(累积发病率为19.8%)。多因素Cox回归分析结果显示,高密度脂蛋白胆固醇水平(HR=0.334,95%CI:0.209~0.534)为NAFLD发病的独立保护因素,而体质指数(HR=1.220,95%CI:1.172~1.271)、甘油三酯(HR=1.114,95%CI:1.052~1.180)、低密度脂蛋白胆固醇(HR=1.252,95%CI:1.054~1.487)、丙氨酸氨基转移酶(HR=1.013,95%CI:1.005~1.021)、血尿酸(HR=1.003,95%CI:1.001~1.004)为NAFLD发病的独立危险因素(P<0.05,P<0.01)。利用上述影响因素成功构建NAFLD发生风险预测模型。建模组和验证组的C指数分别为0.789(95%CI:0.766~0.812)、0.777(95%CI:0.742~0.812),校正曲线显示模型预测结果与实际观察结果吻合良好。结论本研究构建的NAFLD发生风险预测模型可以准确地预测NAFLD的发生概率,为早期识别NAFLD高危人群提供新思路。
Objective To establish the risk prediction model for nonalcoholic fatty liver disease(NAFLD),and to provide the management strategies for the prevention of NAFLD.Methods The residents(18-59 years old)with two times physical examination records,baseline without NAFLD and with important indexes were selected from the annual physical examination database during January 2015 to July 2018 in the health management center in the second affiliated hospital of Dalian medical university as the subjects to observe the NAFLD outcome.The data of basic information,physical examination,laboratory test and abdominal ultrasonography were collected,the subjects were randomly divided into establishing model group and validation group.Theχ^(2) test,t test,rank-sum test and single factor Cox regress were used to analyze the data,the used software was SPSS 23.0;the data of establishing model group were used in multi-factors Cox regress for selecting the prediction indexes;the RStudio statistical software was used to draw the nomogram for establishing the risk prediction model of NAFLD.The prediction efficiency of two groups was verified with consistency index(C index)and calibration curve.Results In this study,2377 subjects were included,including 1585 cases in the model group and 792 cases in the validation group.A total of 467 subjects developed NAFLD(morbidity:19.6%),average duration of following up was(27.06±8.02)months.The number of NAFLD in the model group was 310 cases(morbidity:19.6%),and the number of NAFLD in the validation group was 157 cases(morbidity:19.8%).The multi-factors COX regress showed that HDL-C(HR=0.334,95%CI:0.209-0.534)was the independent protective factor of NAFLD;while body mass index(HR=1.220,95%CI:1.172-1.271),triglyceride(HR=1.114,95%CI:1.052-1.180),LDL-C(HR=1.252,95%CI:1.054-1.487),alanine aminotransferase(HR=1.013,95%CI:1.005-1.021)and serum uric acid(HR=1.003,95%CI:1.001-1.004)were the independent risk factors(P<0.05 or P<0.01).Above influencing factors were used to establish the risk prediction mode
作者
李楠
王雪莹
郭佳桐
冷松
LI Nan;WANG Xue-ying;GUO Jia-tong;LENG Song(Health Management Center,the Second Affiliated Hospital of Dalian Medical University,Dalian.Liaoning Province,116023,China)
出处
《中国慢性病预防与控制》
CAS
CSCD
北大核心
2021年第3期167-171,共5页
Chinese Journal of Prevention and Control of Chronic Diseases
关键词
非酒精性脂肪肝
血尿酸
列线图
COX回归模型
Nonalcoholic fatty liver disease
Serum uric acid
Nomogram
Cox regression model
