Clinical Trial Service Unit & Epidemiological Studies Unit (CTSU), Nuffield Department of Population Health, University of Oxford, Oxford, UK.
Health Data Research UK Oxford, University of Oxford, Oxford, UK.
Int J Epidemiol. 2024 Aug 14;53(5). doi: 10.1093/ije/dyae125.
Higher body mass index (BMI) is associated with higher incidence of cardiovascular and some non-cardiovascular diseases (CVDs/non-CVDs). However, uncertainty remains about its associations with mortality, particularly at lower BMI levels.
The prospective China Kadoorie Biobank recruited >512 000 adults aged 30-79 years in 2004-08 and genotyped a random subset of 76 000 participants. In conventional and Mendelian randomization (MR) analyses, Cox regression yielded adjusted hazard ratios (HRs) associating measured and genetically predicted BMI levels with incident risks of major vascular events (MVEs; conventional/MR 68 431/23 621), ischaemic heart disease (IHD; 50 698/12 177), ischaemic stroke (IS; 42 427/11 897) and intracerebral haemorrhage (ICH; 7644/4712), and with mortality risks of CVD (15 427/6781), non-CVD (26 915/4355) and all causes (42 342/6784), recorded during ∼12 years of follow-up.
Overall, the mean BMI was 23.8 (standard deviation: 3.2) kg/m2 and 13% had BMIs of <20 kg/m2. Measured and genetically predicted BMI showed positive log-linear associations with MVE, IHD and IS, but a shallower positive association with ICH in conventional analyses. Adjusted HRs per 5 kg/m2 higher genetically predicted BMI were 1.50 (95% CI 1.41-1.58), 1.49 (1.38-1.61), 1.42 (1.31-1.54) and 1.64 (1.58-1.69) for MVE, IHD, IS and ICH, respectively. These were stronger than associations in conventional analyses [1.21 (1.20-1.23), 1.28 (1.26-1.29), 1.31 (1.29-1.33) and 1.14 (1.10-1.18), respectively]. At BMIs of ≥20 kg/m2, there were stronger positive log-linear associations of BMI with CVD, non-CVD and all-cause mortality in MR than in conventional analyses.
Among relatively lean Chinese adults, higher genetically predicted BMI was associated with higher risks of incident CVDs. Excess mortality risks at lower BMI in conventional analyses are likely not causal and may reflect residual reverse causality.
较高的体重指数（BMI）与心血管疾病和一些非心血管疾病（CVD/非 CVD）的发病率增加有关。然而，其与死亡率的关系仍存在不确定性，尤其是在较低的 BMI 水平下。
前瞻性的中国慢性病前瞻性研究（CKB）于 2004-2008 年招募了 512000 多名年龄在 30-79 岁的成年人，并对其中 76000 名参与者进行了随机亚组的基因分型。在常规和孟德尔随机化（MR）分析中，Cox 回归得出了与主要血管事件（MVEs；常规/MR 68431/23621）、缺血性心脏病（IHD；50698/12177）、缺血性卒中（IS；42427/11897）和颅内出血（ICH；7644/4712）发生率相关的、经测量和经遗传预测的 BMI 水平的调整后的风险比（HRs），以及与 CVD（15427/6781）、非 CVD（26915/4355）和所有原因（42342/6784）死亡率风险相关的 HRs，这些数据在大约 12 年的随访期间记录。
总体而言，平均 BMI 为 23.8（标准差：3.2）kg/m2，13%的人的 BMI 低于 20 kg/m2。在常规分析中，经测量和经遗传预测的 BMI 与 MVEs、IHD 和 IS 呈正对数线性关系，但与 ICH 的正相关关系较浅。与每增加 5kg/m2的经遗传预测的 BMI 相关的 HRs 分别为 1.50（95%CI 1.41-1.58）、1.49（1.38-1.61）、1.42（1.31-1.54）和 1.64（1.58-1.69），分别用于 MVEs、IHD、IS 和 ICH。这些 HRs 强于常规分析中的关联[1.21（1.20-1.23）、1.28（1.26-1.29）、1.31（1.29-1.33）和 1.14（1.10-1.18），分别]。在 BMI≥20kg/m2的人群中，MR 中 BMI 与 CVD、非 CVD 和全因死亡率的正对数线性关联比常规分析更强。
在相对偏瘦的中国成年人中，较高的经遗传预测的 BMI 与 CVD 发病风险的增加有关。常规分析中较低 BMI 时的过度死亡风险可能不是因果关系的，可能反映了残余的反向因果关系。
