中国台湾35-74岁健检人群高尿酸血症5年发病风险预测模型

Prediction model for the risk of hyperuricemia within a 5 year follow-up among adults aged 35-74 in Taiwan

目的构建中国台湾35-74岁健检人群高尿酸血症5年发病风险预测模型。方法选择随访刚满5年且基线无高尿酸血症的健康体检人员21 190人,分为建模队列和验证队列,采用多因素逐步回归logistic分析构建模型,以ROC曲线下面积（AUC）评价拟合优度。结果研究人群5年高尿酸血症发病率为12.64%。纳入预测模型变量有性别、舒张压、血尿酸、总胆固醇、LDL-C、甘油三酯、教育,其OR值（95%CI）分别为0.397（0.334-0.472）、1.021（1.015-1.027）、1.018（1.016-1.020）、0.763（0.651-0.894）、1.416（1.261-1.591）、1.304（1.194-1.424）和0.877（0.830-0.927）。预测模型ROC曲线下面积（AUC）为0.766（95%CI,0.753-0.779）,验证队列为0.764（0.752-0.776）。结论由研究数据库建立的高尿酸血症风险预测模型有较高验证效度,对于该健康体检人群高尿酸血症预测具有实用、可行的特点,对评估发病风险有一定应用价值。

Objective To develop aprediction model to assess the risk of hyperuricemia within a 5year followup among adults aged 35-74years in Taiwan.Methods A cohort of 21190adults who had no hyperuricemia on baseline and had finished a 5year follow-up were selected into the study.They were divided into two cohorts：modeling and validating cohorts.Multivariate logistic regression was used to analyze the model and ROC curve′s AUC was used to test the goodness-of-fit.Results The incidence of hyperuricemia within a 5-year follow up was 12.64%.Variables adopted in the prediction model were gender,diastolic pressure,blood uric acid,total cholesterol,LDL-c,triglyceride and education with the OR（95%CI）of 0.397（0.334-0.472）,1.021（1.015-1.027）,1.018（1.016-1.020）,0.763（0.651,0.894）,1.416（1.26-1.591）,1.304（1.194-1.424）and 0.877（0.830-0.927）.The area under the ROC curve（AUC）was 0.766（95%CI,0.753-0.779）in modeling cohort and was 0.764（95%CI,0.752-0.776）in validating cohort.Conclusions The prediction model for forecasting the risk of hyperuricemia established in the study is able to effectively predict hyperuricemia among healthy adults.