生物电阻抗法和双能X线吸收法测量成人体脂率的最小乘积回归偏差分析和结果校正  

Agreement analysis and result calibration by ordinary least product regression between bioelectrical impedance analysis and dual energy X-ray absorptiometry

作　　者：宋晓昀 王惠君 苏畅 欧阳一非 吕志梅 魏楠 李岩 马玉霞[2] 李青霞[2] 张兵 Song Xiaoyun;Wang Huijun;Su Chang;Ouyang Yifei;Lyu Zhimei;Wei Nan;Li Yan;Ma Yuxia;Li Qingxia;Zhang Bing(National Institute for Nutrition and Health,Chinese Center for Disease Control and Prevention,Beijing 100050,China;School of Public Health,Hebei Medical University,Shijiazhuang 050017,China)

机构地区：[1]中国疾病预防控制中心营养与健康所,北京100050 [2]河北医科大学公共卫生学院,石家庄050017

出　　处：《卫生研究》2021年第3期426-431,共6页Journal of Hygiene Research

基　　金：国家重点研发计划(No.2019YFC1605100);国家自然科学基金(No.81573155);国家财政项目(No.131031107000160004)。

摘　　要：目的基于最小乘积回归法(ordinary least product regression, OLP)分析生物电阻抗(bioelectrical impedance analysis, BIA)法和双能X线吸收(dual energy X-ray absorptiometry, DXA)法测量中国成年人体脂率的一致性,并对BIA法结果进行校正。方法招募18~65岁成年志愿者,分别采用BIA法(Tanita BC-601F仪器)和DXA法(Hologic Discovery QDR仪器)测量受试者的总体脂率和躯干体脂率,采用OLP比较2种方法测量结果的一致性,采用Bootstrap方法对校正方程进行内部验证,采用OLP分析校正后2种方法测量结果的一致性。结果校正前,男、女总体脂率和躯干体脂率的OLP回归线截距的95%CI均不包括0,斜率的95%CI均不包括1.00。校正后,男、女总体脂率和躯干体脂率的OLP回归线截距的95%CI均包括0,斜率的95%CI均包括1.00。结论 BIA法和DXA法测量的男、女总体脂率和躯干体脂率均存在固定偏差和比例偏差,使用BIA法测量体脂率需注意校正。OBJECTIVE To analyze agreement of body fat percentage measured by bioelectrical impedance analysis(BIA) and dual energy X-ray absorptiometry(DXA) in Chinese adults, and to calibrate BIA measurement by ordinary least product regression(OLP). METHODS Volunteers aged 18-65 were recruited to measure total body fat percentage and trunk fat percentage by BIA(Tanita BC-601 F) and DXA(Hologic Discovery QDR), respectively. Agreement between the two methods was compared by OLP. Bootstrap method was used to do the internal validation of the calibration equation. The calibration effect was verified by the agreement analysis of the calibrated BIA result and the DXA result. RESULTS Before calibration, all 95%CIs of the intercepts of the OLP regression lines did not include 0, and all 95%CIs of the slopes of the OLP regression lines did not include 1.00. After calibration, all 95% CIs of the intercepts of the OLP regression lines included 0, and all 95% CIs of the slopes included 1.00. CONCLUSION There are fixed bias and proportion bias between BIA and DXA in measuring total body fat percentage and trunk fat percentage in Chinese men and women. Calibration should be considered when body fat percentage is measured by BIA.

关 键 词：体脂率 生物电阻抗 双能X线吸收 最小乘积回归 校正方程 

分 类 号：R151[医药卫生—营养与食品卫生学] R151.4[医药卫生—公共卫生与预防医学]