Four-Way Decomposition of Effect of Alcohol Consumption and Body Mass Index on Lipid Profile

Both obesity and alcohol consumption are strongly associated with dyslipidemia; however, it remains unclear whether their joint effect on lipid profiles is through mediation, interaction, or a combination of the two. In total, 9849 subjects were selected from the 2009 panel of China Health and Nutrition Survey (CHNS). A four-way decomposition method was used to validate the pathways of drinking and body mass index (BMI) on lipids (total cholesterol, TC; triglyceride, TG; low-density lipoprotein cholesterol, LDL-C; high-density lipoprotein cholesterol, HDL-C; apolipoprotein A, APO-A; and apolipoprotein B, APO-B). According to four-way decomposition, the total effects of drinking on lipids were found to be statistically significant, except for LDL-C. The components due to reference interaction were 0.63, 0.48, 0.60, -0.39, -0.30, and 0.20 for TC, TG, LDL-C, HDL-C, APO-A and APO-B, respectively ( < 0.05 for all). The effect size of pure indirect effect and mediated interaction were 0.001~0.006 ( > 0.05 for all). Further, linear regression models were used to examine the effect of BMI on lipid profiles in drinkers and non-drinkers. The associations of BMI and lipids were higher in all drinkers than in non-drinkers (0.069 versus 0.048 for TC, 0.079 versus 0.059 for TG, 0.057 versus 0.037 for LDL-C, -0.045 versus -0.029 for HDL-C, -0.024 versus -0.011 for APO-A and 0.026 versus 0.019 for APO-B, interaction