In addition to underreporting actual alcohol consumption in epidemiologic studies, another limitation is the lack of information about drinking pattern. In some epidemiologic studies (Allen et al., 2009; Hamajima et al., 2002) women report consumption on a per week basis, which is converted into drinks / day by dividing by 7. While this is seemingly a simple and straightforward calculation, it raises a major interpretational issue. According to this approach, a woman who drinks 3–4 drinks on Friday and Saturday night and no alcohol during the rest of the week could accurately report drinking 7 drinks / week. After dividing by 7, she would be classified as drinking one drink / day, and her data combined with those of women who do in fact drink only one drink / day. Clearly, drinking multiple drinks in the same sitting will result in higher blood alcohol levels than from a single drink, which can result in qualitatively different metabolic consequences, such as the induction of CYP2E1 and formation of free radicals (see below). In support of this concept, a recent study (Chen
