Mäkinen Ville-Petteri, Zhao Siyu, Ihanus Andrei, Tynkkynen Tuulia, Ala-Korpela Mika
Systems Epidemiology, Research Unit of Population Health, Faculty of Medicine, University of Oulu, Oulu, Finland.
Biocenter Oulu, University of Oulu, Oulu, Finland.
Int J Obes (Lond). 2025 Sep 19. doi: 10.1038/s41366-025-01895-2.
BACKGROUND/OBJECTIVES: Tracking excess adiposity at population scale is essential for managing the obesity pandemic in human populations. New formulas based on weight, height, waist and hip measurements have been suggested as better alternatives to the classic body mass index and waist-hip ratio, but the lack of systematic benchmarking on how these formulas reflect adiposity, metabolic dysfunction and clinical sequelae causes confusion on how to best monitor the health of populations.
SUBJECTS/METHODS: Participants from the Northern Finland Birth Cohort 1966 were included based on data availability at the 46-year visit (2511 women and 1908 men). Cross-sectional sex-adjusted Spearman correlations with clinical biomarkers and serum and urine NMR metabolomics were calculated for body mass index (BMI), waist-hip ratio (WHR), waist-height ratio (WHER), abdominal volume index, body adiposity index, body roundness index, body shape index, conicity index and impedance-based body fat. UK biobank participants were selected based on available data at initial visit (244,947 women and 205,949 men). Prevalent and incident cases of type 2 diabetes, hypertension, liver disease and heart disease were ascertained through register linkage. Prevalent cases were predicted from adiposity measures by age- and sex-adjusted logistic regression and incident cases by age- and sex-adjusted Cox regression.
Adiposity measures were highly collinear and exhibited low biomolecular specificity. BMI and WHR together captured almost all body shape information related to cardiometabolic diseases. For instance, the c-statistic of the BMI & WHR model for diabetes (0.8012; CI95: 0.7963, 0.8061) was near the theoretical maximum of 0.8047. Diabetes was also predicted by WHER (0.7951; CI95: 0.7903, 0.8000). Other adiposity measures showed equal or worse prediction accuracy. This pattern repeated across multiple disease diagnoses.
We did not observe sufficient benefits from the more recent body adiposity formulas over body mass index, waist-hip or waist-height ratio to warrant their widespread application in cardiometabolic epidemiology.
背景/目的：在人群层面追踪过多的肥胖情况对于应对全球肥胖流行至关重要。基于体重、身高、腰围和臀围测量的新公式被认为是经典体重指数和腰臀比的更好替代方法，但缺乏关于这些公式如何反映肥胖、代谢功能障碍和临床后遗症的系统基准测试，这使得人们在如何最好地监测人群健康方面感到困惑。
受试者/方法：根据1966年芬兰北部出生队列46岁随访时的数据可用性纳入参与者（2511名女性和1908名男性）。计算了体重指数（BMI）、腰臀比（WHR）、腰高比（WHER）、腹部容积指数、身体肥胖指数、身体圆润指数、身体形状指数、锥度指数和基于阻抗的体脂与临床生物标志物以及血清和尿液核磁共振代谢组学的横断面性别调整Spearman相关性。根据英国生物银行参与者首次就诊时的可用数据进行选择（244,947名女性和205,949名男性）。通过登记链接确定2型糖尿病、高血压、肝病和心脏病的现患病例和新发病例。通过年龄和性别调整的逻辑回归从肥胖测量指标预测现患病例，通过年龄和性别调整的Cox回归预测新发病例。
肥胖测量指标高度共线，且生物分子特异性较低。BMI和WHR共同捕捉了几乎所有与心血管代谢疾病相关的身体形状信息。例如，BMI&WHR模型对糖尿病的c统计量（0.8012；95%置信区间：0.7963，0.8061）接近理论最大值0.8047。WHER对糖尿病的预测效果也较好（0.7951；95%置信区间：0.7903，0.8000）。其他肥胖测量指标的预测准确性相同或更差。这种模式在多种疾病诊断中反复出现。
我们没有观察到相较于体重指数、腰臀比或腰高比，最新的身体肥胖公式有足够的优势，因此不建议在心血管代谢流行病学中广泛应用。
