体重指数与体脂率指标评价肥胖:基于诊断试验的比较研究

Relationship Between Body Mass lndex and Percent Body Fat in the Diagnosis of Obesity:Based on Diagnostic Tests

目的:通过诊断试验评价体重指数(BMI)诊断肥胖的价值,并比较BMI与体脂率两种指标对肥胖相关疾病高血压、动脉硬化的预测价值,为肥胖的评价研究及指标应用的适应性提供参考。方法:采用分层整群方法,以3149名江苏省社区自然人群作为研究对象进行回顾性研究,同时测量BMI和体脂率。体脂率采用WHO和ASBP(美国减肥专科医学会)两种标准评价,筛查高血压和动脉硬化作为相关疾病。以体脂率定义的肥胖为效标,采用诊断试验的受试者工作特征(ROC)曲线评估BMI的诊断价值,并分析和比较BMI与体脂率两种指标对肥胖相关疾病的预测价值。结果:(1)BMI对肥胖(以体脂率定义)的诊断价值:女性的ROC曲线下面积为0.949(WHO标准)、0.906(ASBP标准),高于男性的0.864(WHO、ASBP两标准相同);不同年龄组中,20~39岁青年人群的曲线下面积最高;上述非参数检验P<0.01。(2)根据ROC曲线,与原切点相比,BMI调整切点为男26 kg/m^2、女25 kg/m^2(体脂率WHO标准)或男26 kg/m^2、女23 kg/m^2(体脂率ASBP标准),预测肥胖的特异度从90%~99%下降到76%~87%,但灵敏度从17%~43%大幅度增加到78%~89%,总体精确程度大幅增加;调整切点后,肥胖检出率的一致性检验Kappa系数男性从0.475提高到0.537,女性从0.115提高到0.655。当体脂率为WHO标准时,BMI预测肥胖的灵敏度、特异度均比ASBP标准更高。(3)BMI预测高血压、动脉硬化的ROC曲线下面积分别为男性0.688(95%CI:0.656~0.720)、0.613(95%CI:0.586~0.642),女性0.745(95%CI:0.708~0.782)、0.692(95%CI:0.659~0.726);体脂率预测高血压、动脉硬化的ROC曲线下面积分别为男性0.687(95%CI:0.655~0.718)、0.635(95%CI:0.608~0.663),女性0.723(95%CI:0.681~0.764)、0.683(95%CI:0.648~0.718);上述P<0.01。(4)男性体脂率对动脉硬化的预测价值(曲线下面积)高于BMI(u=2.05,P<0.05),女性无差异(u=0.75,P>0.05);男、女体脂率和BMI对高血压的预测价值均无差异(u=0.92、1.26,P>0.05)。结论:(1)大样本研究时,BMI对肥胖(体脂率评价)有较高的诊断价值,尤其是女性、青年人群;BMI指标具备可替代性,但需要考虑切点的调整。(2)BMI和体脂率两种指标均可有效地预测人群高血压、动脉硬化风险,在女性和青年人中应用价值更大。(3)参照中国肥胖问题工作组的BMI标准,体脂率WHO标准比ASBP标准更适合中国人群评价肥胖。

Objective To evaluate and compare the value of predicting diseases such as hypertension and arteriosclerosis of the body mass index（BMI） and percent body fat（PBF）,so as to provide the ba-sis for evaluation of obesity. Methods Totally 3149 common residents in Jiangsu province were sur-veyed using the method of stratified and cluster sampling. Effective data included height,weight,PBF,blood pressure and brachial-ankle pulse wave velocity（ba PWV）. Obesity was defined by percent bodyfat with the evaluation of WHO and ASBP. Areas under ROC curves（AUC）of BMI and PBF were esti-mated by the nonparametric test and then the two diagnostic tests were compared by predictive valueof related diseases. Results（1）Predictive value of BMI to diagnose obesity（defined by PBF）：AUC arerespectively 0.949（for WHO standard）、0.906（for ASBP standard）in women and 0.864 in men. In agegroup,the 20-39 years have the highest AUC of BMI to predict obesity. P〈0.01 for above nonparamet-ric tests.（2）According to ROC curves we got the adjusted cut-off points of BMI are respectively 26 kg/m-2 in men and 25 kg/m-2 in women for WHO standard or 26 kg/m-2 in men and 23 kg/m-2 in womenfor ASBP standard,while the specificity of predicting obesity decreased from 90% - 99% to 76% -87%,but the sensitivity significant increased from 17% - 43% to 78% - 89%;After adjusting the cut-off points of BMI,the value of kappa of prevalence of obesity increased from 0.475 to 0.537 in menand 0.115 to 0.655 in women.（3）To predict hypertension,areas of BMI were 0.688（95%CI：0.656-0.720）in men and 0.745（95%CI：0.708-0.782）in women,similarly,of PBF were 0.687（95%CI：0.655-0.718）inmen and 0.723（95%CI：0.681-0.764）in women;To predict arteriosclerosis there showed highly consisten-cy that the areas of BMI were 0.613（0.586-0.641）in men and 0.692（95%CI：0.659-0.726）in women,meanwhile,for PBF they were 0.635（95%CI：0.608-0.663）in men and 0.683（95%CI：0.648-0.718）inwomen（P〈0.01）.（4）Paired test of the two areas under ROC curves showed that PBF had higher diag-nostic value than BMI in men to predict arteriosclerosis（u=2.05,P〈0.05）,however,no statistical differ-ence in women（u=0.75,P〈0.05）and in predicting hypertension（u=0.75 to men and u=1.26 to women,P〈0.05）. Conclusions Using BMI and PBF can all predict hypertension and arteriosclerosis effectively,thus,obesity can be evaluated suitably by both BMI and PBF in large-scale population study,especial-ly in women and youth. Relatively,WHO standard of PBF is more suitable for Chinese population toevaluate obesity than ASBP standard.