Table 6:

Crude and adjusted hazard ratios (HRs) for incident atrial fibrillation, by binge-drinking status^*

Alcohol consumption*	% (no.) of patients with incident atrial fibrillation	HR (95% CI)
		Unadjusted analysis	Model 1†	Model 2‡	Model 3§	Model 4¶
Low	6.3 (1179/18 775)	1.00 (ref)	1.00 (ref)	1.00 (ref)	1.00 (ref)	1.00 (ref)
Moderate or high
No binge drinking	7.8 (814/10 454)	1.22 (1.12–1.33)	1.14 (1.03–1.25)	1.15 (1.05–1.26)	1.13 (1.03–1.24)	1.13 (1.03–1.24)
Binge drinking	8.3 (100/1 204)	1.31 (1.07–1.61)	1.34 (1.09–1.65)	1.38 (1.12–1.69)	1.36 (1.10–1.67)	1.35 (1.10–1.67)
p value for trend		< 0.001	< 0.001	< 0.001	< 0.001	< 0.001

Note: CI = confidence interval, ref = reference category.

^*Binge drinking = > 5 drinks/d at any one time or a calculated average intake of > 5 drinks/d; non-binge drinking = ≥ 1 drink/wk but not meeting the criterion for binge drinking; low = < 1 drink/wk; moderate = 1–14 drinks/wk (about 2 drinks/d) for women and 1–21 drinks/wk (about 3 drinks/d) for men; high = > 14 drinks/wk (> 2 drinks/d) for women and > 21 drinks/wk (> 3 drinks/d) for men.

^†Model 1: Adjusted for age, sex, body mass index and region (Europe or Americas v. Australia/Asia).

^‡Model 2: Adjusted for covariates in model 1, plus medical history of coronary artery disease, stroke or transient ischemic attack, hypertension, diabetes, chronic renal disease, and sleep apnea.

^§Model 3: Adjusted for covariates in model 2, plus smoking status, education (1–8 yr, 9–12 yr or college/trade v. none), physical activity (2–4 times/wk or ≥ 5 times/wk v. < 1 time/wk) and stress.

^¶Model 4: Adjusted for covariates in model 3, plus use of statin and treatment allocation (ramipril, telmisartan or both v. placebo) in the trials.