Genki Plaza Medical Center for Health Care, Tokyo, Japan.
Division of Cardiovascular Medicine, Department of Medicine, Jichi Medical University School of Medicine, Tochigi, Japan.
J Clin Hypertens (Greenwich). 2018 May;20(5):880-890. doi: 10.1111/jch.13270. Epub 2018 Mar 31.
No integrated risk assessment tools that include lifestyle factors and uric acid have been developed. In accordance with the Industrial Safety and Health Law in Japan, a follow-up examination of 63 495 normotensive individuals (mean age 42.8 years) who underwent a health checkup in 2010 was conducted every year for 5 years. The primary endpoint was new-onset hypertension (systolic blood pressure [SBP]/diastolic blood pressure [DBP] ≥ 140/90 mm Hg and/or the initiation of antihypertensive medications with self-reported hypertension). During the mean 3.4 years of follow-up, 7402 participants (11.7%) developed hypertension. The prediction model included age, sex, body mass index (BMI), SBP, DBP, low-density lipoprotein cholesterol, uric acid, proteinuria, current smoking, alcohol intake, eating rate, DBP by age, and BMI by age at baseline and was created by using Cox proportional hazards models to calculate 3-year absolute risks. The derivation analysis confirmed that the model performed well both with respect to discrimination and calibration (n = 63 495; C-statistic = 0.885, 95% confidence interval [CI], 0.865-0.903; χ statistic = 13.6, degree of freedom [df] = 7). In the external validation analysis, moreover, the model performed well both in its discrimination and calibration characteristics (n = 14 168; C-statistic = 0.846; 95%CI, 0.775-0.905; χ statistic = 8.7, df = 7). Adding LDL cholesterol, uric acid, proteinuria, alcohol intake, eating rate, and BMI by age to the base model yielded a significantly higher C-statistic, net reclassification improvement (NRI), and integrated discrimination improvement, especially NRI (NRI = 0.127, 95%CI = 0.100-0.152; NRI = 0.108, 95%CI = 0.102-0.117). In conclusion, a highly precise model with good performance was developed for predicting incident hypertension using the new parameters of eating rate, uric acid, proteinuria, and BMI by age.
尚未开发出包含生活方式因素和尿酸的综合风险评估工具。根据日本《工业安全与健康法》，对 2010 年接受健康检查的 63495 名血压正常个体（平均年龄 42.8 岁）进行了为期 5 年的每年一次的随访检查。主要终点是新发高血压（收缩压[SBP]/舒张压[DBP]≥140/90mmHg 和/或报告高血压开始使用降压药物）。在平均 3.4 年的随访期间，有 7402 名参与者（11.7%）发生高血压。预测模型包括年龄、性别、体重指数（BMI）、SBP、DBP、低密度脂蛋白胆固醇、尿酸、蛋白尿、当前吸烟、饮酒、进食速度、按年龄划分的 DBP 和按年龄划分的 BMI，并使用 Cox 比例风险模型计算 3 年绝对风险来创建。推导分析证实，该模型在区分度和校准度方面表现良好（n=63495；C 统计量=0.885，95%置信区间[CI]，0.865-0.903；χ 统计量=13.6，自由度[df]=7）。此外，在外部验证分析中，该模型在区分度和校准度特征方面表现良好（n=14168；C 统计量=0.846；95%CI，0.775-0.905；χ 统计量=8.7，df=7）。将 LDL 胆固醇、尿酸、蛋白尿、饮酒、进食速度和按年龄划分的 BMI 添加到基本模型中，可显著提高 C 统计量、净重新分类改善（NRI）和综合区分度改善，尤其是 NRI（NRI=0.127，95%CI=0.100-0.152；NRI=0.108，95%CI=0.102-0.117）。总之，使用新的进食速度、尿酸、蛋白尿和按年龄划分的 BMI 参数，开发了一种预测高血压事件发生的高度精确且性能良好的模型。
