Gaston Symielle A, Kendrick Karla N, Ogbenna Bethany T, Sandler Dale P, Stanford Fatima Cody, Jackson Chandra L
Epidemiology Branch, National Institute of Environmental Health Sciences, National Institutes of Health, Research Triangle Park, NC, USA.
Winchester Hospital Weight Management Center, Beth Israel Lahey Health, Woburn, MA, USA.
Int J Obes (Lond). 2025 Apr 14. doi: 10.1038/s41366-025-01784-8.
Underreporting of weight and overreporting of height is consistently shown among women, thereby reducing accuracy of estimation of body mass index-and thus obesity-in epidemiologic studies that rely on self-reported data. Additionally, misreporting has been shown to differ by socioeconomic status and race and ethnicity, which can result in differential misclassification and bias that can obfuscate associations with obesity across groups in multiethnic and socioeconomically varying populations. Therefore, we sought to assess agreement between self-reported and objectively measured weight, height, and derived body mass index (BMI) across levels of educational attainment within racial and ethnic groups in a population-based cohort of US women.
Among self-identified White, Black, and Latina women enrolled in the Sister Study (2003-2009), we assessed mean differences in self-reported vs. objectively measured weight, height, and derived BMI. Using adjusted linear and multinomial logistic regression, we compared measurement error among participants reporting some college/vocational school or ≥college vs. ≤high school. We assessed BMI agreement using Bland-Altman plots and weighted kappa (k) statistics.
Among 18,638 participants (age: mean ± standard deviation = 56 ± 9.0 years), 84% identified as White, 10% Black, and 5% Latina. Approximately half (49%) attained a college education. Weight and height were generally underreported. Higher underreporting of weight among participants with ≥college vs. ≤high school was of larger magnitude among Black and Latina vs. White participants. Bland-Altman results revealed that agreement in continuous BMI was good among White participants but generally fair among Black and Latina participants. Categorical BMI agreement was consistently high with minor variation by race and ethnicity and educational attainment (weighted k range = 0.92-0.93).
Despite higher measurement error in weight among Black and Latina participants with ≥college education, self-reported and objectively measured BMI categories showed high agreement across groups. Results support the utility of self-reported data that reliably estimate BMI category across socioeconomic, racial, and ethnic groups in this cohort.
在女性中一直存在体重报告不足和身高报告过度的情况,这降低了在依赖自我报告数据的流行病学研究中体重指数(BMI)估计的准确性,进而影响肥胖症的估计。此外,报告错误已被证明因社会经济地位、种族和民族而异,这可能导致差异错误分类和偏差,从而混淆多民族和社会经济状况各异人群中不同群体与肥胖症之间的关联。因此,我们试图在美国女性的一个基于人群的队列中,评估种族和民族群体中不同教育程度水平下自我报告的体重、身高以及推导的BMI与客观测量值之间的一致性。
在参与姐妹研究(2003 - 2009年)的自我认定为白人、黑人及拉丁裔女性中,我们评估了自我报告的体重、身高及推导的BMI与客观测量值之间的平均差异。使用调整后的线性和多项逻辑回归,我们比较了报告上过一些大学/职业学校或大学及以上学历的参与者与高中学历及以下参与者之间的测量误差。我们使用Bland - Altman图和加权kappa(κ)统计量评估BMI的一致性。
在18638名参与者(年龄:均值±标准差 = 56±9.0岁)中,84%为白人,10%为黑人,5%为拉丁裔。约一半(49%)拥有大学学历。体重和身高普遍报告不足。大学及以上学历参与者相比高中学历及以下参与者体重报告不足的情况在黑人和拉丁裔参与者中比在白人参与者中更为严重。Bland - Altman结果显示,连续BMI的一致性在白人参与者中良好,但在黑人和拉丁裔参与者中总体一般。分类BMI的一致性始终很高,种族、民族和教育程度的差异较小(加权κ范围 = 0.92 - 0.93)。
尽管大学及以上学历的黑人和拉丁裔参与者体重测量误差较大,但自我报告的和客观测量的BMI类别在各群体间显示出高度一致性。结果支持了在该队列中自我报告数据在可靠估计社会经济、种族和民族群体BMI类别方面的效用。