Association of observationally and genetically determined BMI and plasma adiponectin concentration. Observational analyses used multiple linear regression multivariable adjusted for age, sex, smoking status, cumulative tobacco consumption, alcohol, education, income, leisure time physical activity, systolic blood pressure, plasma cholesterol and triglycerides, and baseline diabetes. One-sample Mendelian randomization analyses based on the Copenhagen General Population Study (CGPS) used instrumental variable analyses with two-stage least-squares regression and unweighted genetic scores adjusted for age and sex. Two-sample Mendelian randomization analyses based on GIANT and ADIPOGen consortia used instrumental variable analyses with inverse-variance weighted regression. R² = variance in exposure explained by the SNPs. BMI = body mass index. SNP = single nucleotide polymorphisms. NA = not applicable.