The accepted interpretation of the J-shaped curve relating alcohol intake to cardiovascular events or mortality is that the lowest point on the curve (light-to-moderate drinking) represents optimum exposure to alcohol, and the increased risk in nondrinkers or heavy drinkers reflects the consequence of suboptimal exposure. This means that unlike tobacco, for which there is no safe level of consumption (Figure 1), the nadir of J-shaped curves for alcohol (Figure 2) is proposed as a healthy range for the general population (1, 2). Numerous studies have used J-shaped curves to describe the relationship between alcohol use and total mortality. In a meta-analysis of 34 prospective studies in men and women, including more than 1 million subjects and more than 94,500 deaths (2), a clear J-shaped relationship between alcohol intake and mortality was observed. After an initial reduction in mortality, as alcohol intake increased, the curve reached a nadir and then showed an increase in mortality at higher doses (Figure 2). The lowest mortality risk (À19%) was observed at 6 g/day of alcohol or approximately one-half of a drink per day, but lower mortality compared with no alcohol consumption was observed with up to 4 drinks/day in men and 2 drinks/day in women. Higher doses of alcohol were associated with increased mortality (2). The basic J-shaped curve persisted after adjustment for many relevant confounding variables. However, prominence is often given to the hypothesis that the well documented J-curve relationship of lower risk of mortality with light-to-moderate alcohol intake might be due to a misclassification of drinking patterns. The higher risk among abstainers might have been due to the inclusion of high-risk subjects who had become abstainers (3). In 2010, an analysis of 9 nationally representative samples of US adults showed that light-to-moderate alcohol consumption was inversely associated with cardiovascular mortality (4). Compared with lifetime abstainers, a significant protection of 31% and 38% among light and moderate drinkers was found, respectively. The magnitude of lower risk was similar in subgroups of sex, age, or baseline health status. There was no simple relation of drinking pattern with risk, but risk was consistently higher among those who consumed> 3 compared with 2 drinks/day (4).

In this issue of the Journal, Xi et al.(5) analyzed the amount of alcohol intake and risk of all-cause, cardiovascular, and cancer deaths in 333,247 American adults from 13 waves of US National Health Interview data. After a median follow-up of 8.2 years, 34,754 participants died from all causes (including 8,947 cardiovascular disease [CVD] deaths and 8,427 cancer deaths). Compared with lifetime abstainers, light-tomoderate alcohol consumers were at significantly reduced risk of all-cause (light: À21%; moderate: À22%) and CVD mortality (light: À26%; moderate: À29%), respectively. In contrast, significantly increased risks of all-cause (11%) and cancer mortality (27%) were found in adults with heavy alcohol consumption compared with those who never consumed. The data by Xi et al.(5)