Table 2.
Effects of alcohol use on depression – OLS and 2SLS estimation results.
| (1) All participants | (2) Excluding heavy drinkers | (3) Excluding former drinkers | |
|---|---|---|---|
| N = 476 | N = 425 | N = 400 | |
| (a.) Multivariable linear regressions | |||
| (1) Key explanatory variable: drinking or not | |||
| b (95% CI) | −0.068 (−0.126, −0.010) | −0.082 (−0.148, −0.016) | −0.065 (−0.129, −0.001) |
| p | 0.022 | 0.015 | 0.046 |
| (2) Key explanatory variable: drinking times during the past 30 days | |||
| b (95% CI) | −0.005 (−0.010, −0.001) | −0.006 (−0.011, −0.001) | −0.006 (−0.012, −0.001) |
| p | 0.017 | 0.012 | 0.019 |
| (3) Key explanatory variable: weekly ethanol consumption (g) | |||
| b (95% CI) | −0.000 (−0.000, 0.000) | −0.000 (−0.000, 0.000) | −0.001 (−0.001, 0.000) |
| p | 0.191 | 0.19 | 0.092 |
| (b.) Mendelian randomization | |||
| (1) Key explanatory variable: drinking or not | |||
| b (95% CI) | −0.127 (−0.253, −0.001) | −0.136 (−0.267, −0.004) | −0.149 (−0.295, −0.003) |
| p | 0.048 | 0.043 | 0.045 |
| Cragg-Donald F statistics of weak instrument tests (p) | 44.491 (0.000) | 55.917 (0.000) | 34.221 (0.000) |
| Sargan statistics of overidentification tests (p) | 1.837 (0.175) | 1.036 (0.309) | 2.061 (0.151) |
| (2) Key explanatory variable: drinking times during the past 30 days | |||
| b (95% CI) | −0.012 (−0.023, −0.001) | −0.012 (−0.024, −0.001) | −0.015 (−0.030, −0.000) |
| p | 0.040 | 0.036 | 0.047 |
| Cragg-Donald F statistics of weak instrument tests (p) | 28.664 (0.000) | 36.403 (0.000) | 19.967 (0.000) |
| Sargan statistics of overidentification tests (p) | 1.605 (0.205) | 1.595 (0.207) | 2.190 (0.139) |
| (3) Key explanatory variable: weekly ethanol consumption (g) | |||
| b (95% CI) | −0.001 (−0.002, −0.000) | −0.001 (−0.002, −0.000) | −0.000 (−0.001, −0.000) |
| p | 0.039 | 0.027 | 0.044 |
| Cragg-Donald F statistics of weak instrument tests (p) | 18.113 (0.000) | 20.818 (0.000) | 20.386 (0.000) |
| Sargan statistics of overidentification tests (p) | 1.240 (0.265) | 2.559 (0.110) | 2.092 (0.148) |
Depression was defined as CESD-10 score ≥ 12 (Cheng and Chan, 2005). Abbreviations: 95% CI represents 95% confidence interval. All models were adjusted for age, gender, education, income, smoking, and province fixed effects. MR results were additionally adjusted for the number of drinking parents and individual genetic ancestral compositions of Northern Han, Southern Han, Mongolian, and Japanese. The First-stage Cragg-Donald F statistics (with values of p) and Sargan statistics (with values of p) are test statistics of the weak instrument and overidentification tests, respectively, which indicate that genetic instruments of ALDH2 rs671 and ADH1B rs1229984 used in MR satisfied with the relevance assumption and exclusion restriction.