Skip to main content
. 2016 Sep 15;11(9):e0162930. doi: 10.1371/journal.pone.0162930

Table 2. Association of alcohol (10g ethanol) with fasting serum glucose using Mendelian Randomization analysis.

Observational multivariable regression analysis* X-Y Mendelian randomization analysis
Selection Number β SE F-statistic G-X P for endogenous test β SE
Total Total* 4,367 0.49 0.11 280.94 0.0344 1.57 0.43
Excluding heavy users 1) 4,306 0.70 0.14 327.10 0.0773 1.75 0.48
Excluding heavy users 2) 4,059 0.54 0.20 301.85 0.0518 2.11 0.71
Excluding elderly 3) 4,262 0.51 0.11 275.49 0.0280 1.64 0.43
Men Total* 2,993 0.52 0.12 302.62 0.0009 1.78 0.41
Excluding heavy users1) 2,932 0.76 0.16 370.80 0.0026 2.01 0.47
Excluding heavy users2) 2,699 0.63 0.23 371.60 0.0004 2.51 0.67
Excluding elderly3) 2,936 0.54 0.12 296.18 < .0001 1.78 0.41
Women Total* 1,374 0.32 0.05 44.30 0.8844 -0.21 2.02
Excluding heavy users1) 1,374 0.32 0.05 44.30 0.8844 -0.21 2.02
Excluding heavy users2) 1,360 0.24 0.03 55.12 0.7514 -0.30 2.71
Excluding elderly3) 1,326 0.32 0.05 43.76 0.6245 0.53 1.98

*Adjusted for age, sex, smoking status, exercise, and family history of diabetes

1) Participants with alcohol intake <100 g/day

2) participants with alcohol intake <50 g/day

3) Participants aged 65 years or older

Tests of endogeneity is Wu-Hausman test, Ho: variables are exogenous