论文部分内容阅读
目的分析影响体检人群臂-踝脉搏波传导速度(ba PWV)测量值的因素,构建ba PWV的logit预测模型并对其进行评价。方法对本院2010—2014年检测的4 159例体检者的体检资料进行回顾性分析,纳入考察的体检指标包括:性别、年龄、收缩压、舒张压、脉搏、空腹血糖、三酰甘油、胆固醇、丙氨酸转氨酶、谷氨酰转肽酶、肌酐、高密度脂蛋白、低密度脂蛋白、尿酸。采用logistic回归分析方法探究ba PWV的影响因素,建立ba PWV的logit预测模型并对其进行评价。结果单因素分析结果显示,除高密度脂蛋白外,其余检测指标在ba PWV正常组与异常组之间的差异均具有统计学意义(P<0.05)。多因素分析得到的logit模型为:logit(p)=-17.888+0.001×尿酸+0.004×丙氨酸转氨酶+0.105×空腹血糖+0.023×脉搏+0.032×舒张压+0.061×收缩压+0.092×年龄+0.411×性别;logit模型对体检人群ba PWV的预测正确率为79.6%,ROC曲线下面积为0.869(95%CI:0.859~0.879)。结论性别、年龄、收缩压、舒张压、脉搏、空腹血糖、丙氨酸转氨酶、尿酸值等均可能影响ba PWV,logit模型可作为研究此类问题的较好模型。
Objective To analyze the factors influencing the measurement of ba PWV in the physical examination population and construct the logistic regression model of ba PWV. Methods The physical examination data of 4 159 cases from 2010 to 2014 in our hospital were retrospectively analyzed. The physical examination indexes included in the survey included gender, age, systolic blood pressure, diastolic blood pressure, pulse, fasting blood glucose, triglyceride, cholesterol , Alanine aminotransferase, glutamyl transpeptidase, creatinine, high density lipoprotein, low density lipoprotein, uric acid. Logistic regression analysis was used to explore the influencing factors of ba PWV, and the log PWI prediction model of ba PWV was established and evaluated. Results The results of univariate analysis showed that except for high-density lipoprotein, the differences of other indexes between the normal group and the abnormal group of ba-PWV were statistically significant (P <0.05). Logit (m) = - 17.888 + 0.001 × uric acid + 0.004 × alanine aminotransferase + 0.105 × fasting glucose + 0.023 × pulse + 0.032 × diastolic pressure + 0.061 × systolic pressure + 0.092 × age + 0.411 × sex. The logistic regression model showed that the correct rate of ba PWV was 79.6% and the area under the ROC curve was 0.869 (95% CI: 0.859-0.879). Conclusions Gender, age, systolic blood pressure, diastolic blood pressure, pulse, fasting blood glucose, alanine aminotransferase, and uric acid value all may affect ba PWV. Logit model can be used as a good model to study such problems.