Table 4. Confounder adjusted multivariable and instrumental variable associations of drinking versus not drinking alcohol with biomarkers for liver function.
Mean difference in each outcome comparing drinkers non-drinkers (95% CI) | |||||
ALT (%)N = 58,265 | γ-GT (%)N = 58,270 | ALP (%)N = 58,271 | Bilirubin (%)N = 58,271 | Prothrombin (%)N = 57,012 | |
Multivariable | 4.2 (2.9, 5.4) | 9.5 (8.0, 11.0) | –6.1 (–6.9, −5.4) | 3.4 (2.3, 4.4) | 0.5 (0.0, 1.1) |
Instrumental variable | 28.1 (–19.0, 75.2) | 57.8 (2.3, 113.3) | 53.6 (26.1, 81.0) | 0.4 (–39.5, 40.3) | –14.7 (–34.8, 5.4) |
Pdifference instrumentalvariable vs. multivariablea | 0.44 | 0.11 | <0.0001 | 0.61 | 0.22 |
CI: confidence interval; ALT: alanine aminotransferase; γ-GT: γ-glutamyl-transferase; ALP: alkaline phosphatase; Prothrombin: Prothrombin action.
In the multivariable analysis all results are adjusted for age, gender, smoking, physical activity, education and income.
In the instrumental variable analysis the control function method was used with ADH1B and ADH1C used jointly as categorical (indicator) instrumental variables. The first stage F-statistic for all instrumental variable analyses = 21.
Test of null hypothesis that there is no difference in association of alcohol with each outcome between the confounder adjusted multivariable association (row 1) and the instrumental variable association using the control function (row 2); p-value obtained from the bootstrap distribution.