Introduction Only a minority of heavy drinkers develop alcoholic liver disease (ALD). A previous cohort study suggested an association between cigarette smoking and risk of alcoholic cirrhosis.1 Smoking has also been linked to disease severity in primary biliary cirrhosis.2 However, there is a negative association between smoking and other chronic inflammatory diseases such as PSC3 and ulcerative colitis.4

Methods To compare smoking behaviour between heavy drinkers with and without liver disease we recruited two cohorts of heavy alcohol drinkers (>60 units/wk (men) and >40 units/wk (women) for ≥5 years). Cases (n=284, 202 men (71%), age (mean±SD) 47±10 years) had decompensated ALD, defined as Child's Grade B or C. Controls (n=199, 157 men (79%), age 48±9 years) had no clinical evidence of liver disease, persistently normal serum bilirubin, albumin and prothombin time, and ultrasound showing normal or echo-bright liver texture and no other abnormality. We collected data on smoking duration and the average number of cigarettes/day at different phases in life. Subjects were classified as current, past and never smokers. Comparison between groups was by Student t, χ² and Mann–Whitney U tests. Results were adjusted for age and gender using logistic linear regression analysis (*).

Results All Subjects smoked cigarettes including 11(3.9%) cases and 4(2%) controls who also smoked rollups and cigars. There were significantly more “never smokers” among cases than controls (n=94 (33%) vs. 40(20%), p=0.002). However, the percentage of current smokers was not different (n=153 cases (54%) vs. n=109 controls (55%)). Case-control Comparisons are shown in Abstract 050. The differences were also significant when comparison was limited to:(1) Cases with Child's grade C (n=123) vs. Controls (p=0.001–0.015) and (2) Cases vs Heath care seeking controls (n=137) (p=0.001–0.008).

Conclusion These results show a significant negative association between smoking and ALD. This may be related to differences in the liver enzyme CYP2A6 responsible for nicotine metabolism, or confounding environmental factors associated with smoking behaviour. The data are also consistent with a protective effect of smoking against ALD in heavy drinkers.