Abstract: Objective To establish a nomograph model for predicting the risk of dyslipidemia in obese adolescents by using existing anthropometric indicators, so as to provide methodological reference for early screening of dyslipidemia in obese adolescents. Methods A total of 421 obese adolescents aged 13 to 16 were recruited from 2020 to 2021 to measure anthropometric indices (height, weight, chest, waist, hip, thigh and calf circumference) and 13 anthropometric indices were calculated, including body mass index(BMI), waist-to-hip ratio (WHR), waist-to-height ratio (WHtR), conicity index(CI), abdominal volume index (AVI), body adiposity index (BAI), cl-nica universidad de Navarra-body adiposity estimator (CUN-BAE), a body shape index (ABSI), body roundness index (BRI), triponderal mass index(TMI), waist divided by height 0.5(WHT.5R), relative fat mass (RFM) and BMI multiply by the square root of WC(BMIWC).Blood lipids of children were tested, using the "Reference Standards for Dyslipidemia in Children and Adolescents Aged 6 to 18 in China" to diagnose dyslipidemia.LASSO regression was used to screen the characteristic variables of dyslipidemia in obese adolescents, and a nomograph model with multi-indices was established.Hosmer-lemeshow goodness of fit test and bootstrap method (1 000 times) were used for model verification.The prediction ability of nomograph model was evaluated by the receiver operating characteristic(ROC) curve. Results The chest circumference, thigh circumference and ABSI were screened by LASSO regression to establish the nomograph model of obese boys.The Hosmer-lemeshow goodness of fit P value was 0.575, the mean absolute error of the calibration curve was 0.023, and the ROC-AUC was 0.62.BMI, TMI and ABSI were selected to establish the nomogram model of obese girls.The corresponding Hosmer-lemeshow goodness of fit P value was 0.422, the mean absolute error of the calibration curve was 0.023, and the ROC-AUC was 0.70. Conclusion The nomogram model of multi-indices combination established by LASSO regression can be used to predict dyslipidemia in obese adolescents.

Key words: obese, adolescent, dyslipidemia, anthropometric indices, nomogram

摘要： 目的 采用现有的人体测量学指标构建预测肥胖青少年血脂异常风险的列线图模型,为早期筛查肥胖青少年血脂异常提供方法学参考。方法 2020—2021年共招募421名13~16岁肥胖青少年,测量身高,体重及胸、腰、臀、大腿和小腿等围度,计算13个人体测量学衍生指标,包括体重指数(BMI)、腰臀比(WHR)、腰高比(WHtR)、锥形指数(CI)、腹部容积指数(AVI)、身体肥胖指数(BAI)、纳瓦拉大学体脂评估器(CUN-BAE)、身体形态指数(ABSI)、身体圆周指数(BRI)、三维体重指数(TMI)、腰高0.5比(WHT.5R)、相对脂肪量(RFM)、BMI和腰围平方根(BMIWC);检测血脂四项,采用《我国6~18岁儿童青少年血脂异常的参考标准》诊断血脂异常。统计分析采用LASSO回归筛选肥胖青少年血脂异常的特征变量并建立多指标联合的列线图模型;Hosmer-Lemeshow拟合优度和Bootstrap重抽样法(1 000次)进行模型内验证;受试者工作特征曲线评估列线图模型的预测能力。结果 LASSO回归筛选出胸围、大腿围和ABSI建立肥胖男生的列线图模型,Hosmer-Lemeshow拟合优度P值为0.575,校准曲线的平均绝对误差为0.023,ROC-AUC为0.62;BMI、TMI和ABSI建立肥胖女生的列线图模型,Hosmer-Lemeshow拟合优度P值为0.422,校准曲线的平均绝对误差为0.023,ROC-AUC为0.70。结论 采用LASSO回归建立多指标联合的列线图模型可用于预测肥胖青少年血脂异常。

关键词: 肥胖, 青少年, 血脂异常, 人体测量学指标, 列线图