Alcohol and cirrhosis: dose–response or threshold effect?