A case-control analysis was conducted on baseline cross-sectional data from the community-based Guangzhou Biobank Cohort Study (2003-2008), using standard 12-lead resting electrocardiograms. By comparing 191 new ECG-LVH cases with 4311 controls, excessive drinking (>210?g/week) showed excess risks for ECG-LVH (odds ratio [OR]?=?1.90, 95 % confidence interval [CI]?=?1.12-3.24), after adjusting for education, income, occupation, physical activity, smoking, body mass index, fasting glucose, triglyceride, total cholesterol, high-density lipoprotein cholesterol, BP, and antihypertensive medication. Mediation analysis showed a significant mediating effect of BP on the association between excessive drinking and ECG-LVH: systolic (31 % ) and diastolic (16 % ). After multivariate adjustment, no significant association was found between occasional drinking (<once/week: OR?=?1.20, 95 % CI?=?0.80-1.80) and moderate drinking (¡Ýonce/week to ¡Ü210?g/week: OR?=?0.88, 95 % CI?=?0.53-1.47) with increased/decreased risk of ECG-LVH.
Alcohol consumption at >210?g/week in Chinese men is an independent risk factor for ECG-LVH. Low power prevented us from examining whether drinking at <once/week to ¡Ü210?g/week is associated with decreased/increased risk of ECG-LVH. Elevated BP partially mediates between alcohol and ECG-LVH. A Mendelian randomization approach with a large sample size is warranted to determine the relations among alcohol consumption, BP, and LVH.