Author + information
- Eric S. Leifer, PhD∗ ()
- ↵∗Address for correspondence:
Dr. Eric Leifer, Division of Cardiovascular Sciences, National Heart, Lung, and Blood Institute, 6701 Rockledge Drive, MSC 7913, Bethesda, Maryland 20892-7913.
The ill effects of excessive alcohol consumption are well understood, but the risks and benefits of moderate alcohol consumption (MAC) remain an active area of research and debate. In this issue of JACC: Heart Failure, Di Castelnuovo et al. (1) contribute a well-done observational study of 22,824 heart failure (HF)-free and 22,628 atrial fibrillation (AF)-free individuals from central and southern Italy. Notably, Di Castelnuovo et al. (1) obtained a significant association between MAC and a reduction in incident HF (p = 0.049), but no association with incident AF (p = 0.42). They defined MAC as 2 to 4 drinks per day, where a drink contains 12 g of ethanol. This study had 943 incident HF cases and 554 incident AF cases in a cohort that was followed for a median of 8.2 years. Given the large numbers of individuals and events and the careful statistical methods used, this study is an important contribution to research on the relationship between alcohol consumption and cardiovascular outcomes. Nevertheless, given harmful associations between MAC and other health outcomes reported in other studies, including various cancers, this study does not provide enough evidence to encourage nondrinkers to begin drinking solely to prevent HF.
Understanding the relationship between MAC and health outcomes is important because the World Health Organization estimates that nearly 39% of the world’s population age 15 years or older had consumed alcohol during 2013 (2). The World Health Organization also estimates that in 2010 worldwide alcohol consumption among individuals 15 years of age and older was 13.5 g of pure alcohol per person per day. Generally, alcohol consumption is more prevalent with higher per capita consumption in wealthier countries.
Roerecke and Rehm (3) provide an excellent review of the beneficial and harmful associations between MAC and various health outcomes in epidemiological studies. In summary, they report beneficial associations between MAC and ischemic heart disease (IHD), ischemic stroke, and type 2 diabetes. They also report harmful associations between MAC and hemorrhagic stroke and various cancers. With respect to the beneficial association between MAC and IHD, they refer to their own meta-analysis of 44 observational studies involving 38,627 IHD events among 957,684 subjects (4). Similar to the finding for HF by Di Castelnuovo et al. (1), the analysis by Roercke and Rehm (4) showed a J-shaped relationship between average alcohol intake (g/day) and IHD morbidity and mortality, suggesting a beneficial association with MAC. They found that an average of 2 to 3 drinks per day for men and a maximum of 1 drink per day for women corresponded to an IHD mortality relative risk (RR) versus that of abstainers of 0.78 (95% confidence interval [CI]: 0.63 to 0.97) in men and 0.84 (95% CI: 0.74 to 0.96) in women. For IHD morbidity, they found RRs of 0.74 (95% CI: 0.53 to 1.02) in men and 0.54 (95% CI: 0.45 to 0.65) in women. The physiological basis for these associations is partially provided in the meta-analysis of 44 studies of the relationship between alcohol consumption and biological markers associated with the risk of coronary heart disease by Brien et al. (5). They found that alcohol consumption was significantly associated with increased levels of high-density lipoprotein cholesterol, apolipoprotein A1, and adiponectin. Alcohol was associated with decreased fibrinogen levels and did not affect triglycerides.
With respect to the harmful association between alcohol consumption and cancer, Roerecke and Rehm (3) describe a strictly increasing relationship between the amount of alcohol consumed and the risk for cancers of the oral cavity, pharynx, larynx, esophagus, liver, colorectum, and female breast. They suggest that this could be due to ethanol metabolism in the human body through acetaldehyde. The seriousness of the cancer risk of alcohol consumption was underscored in a recent European cohort study of more than 300,000 participants. That study found that among men and women, 10% (95% CI: 7% to 13%) and 3% (95% CI: 1% to 5%) of total cancer incidence was attributable to former and current alcohol consumption (6). Roerecke and Rehm (3) concluded that “attributable fractions are higher for alcohol attributable cancers” and “there is clear and substantial evidence for a detrimental association of any alcohol consumption and several cancer sites” (3).
Given this background, it is useful to review the key findings of the study by Di Castelnuovo et al. (1) that MAC was associated with a lower incidence of HF (p = 0.049). What does this significant p value correspond to? It corresponds to the multivariable hazard ratio (HR) curve (solid curve) in Figure 1A (see Di Castelnuovo et al. ), which plots the HR of HF incidence as a function of alcohol intake (g/day), using zero g/day as the reference category. If the amount of alcohol intake had no relationship with HF incidence, then the true HR curve would be the horizontal line HR = 1. As can be seen in the figure, the HR curve decreases to a nadir HR of 0.80 at approximately 20 g/day, corresponding to a 20% reduction in HF risk, and then very slightly increases to an HR of 0.85 at 100 g/day. The significant p value of 0.049 is obtained because the upper 95% confidence band (i.e., upper dotted curve [Fig. 1A]) is below the HR = 1 line between 0 and 50 g/day. This provides statistical evidence that the HR curve is statistically different from the HR = 1 line between 0 and 50 g/day.
It is important to note that the HR curve was obtained by statistically adjusting for several potential confounders which are listed in the caption of Figure 1. This adjustment is a proper thing to do because these data are observational (i.e., subjects were not randomized to consume alcohol or abstain). Interestingly, without recognition of this adjustment, a reader of this paper could be confused by the HF event rates in Table 2 (Di Castelnuovo et al. ). There the event rates for the categories 12.1 to 24, 24.1 to 48, and >48 g/day are all higher than the Never Drinkers category (4.4% to 4.7% vs. 3.7%, respectively). These higher event rates are reflected in the unadjusted HRs for these categories of 1.12 to 1.23 in the harmful direction (Table 2, Di Castelnuovo et al. ). The reason for the discrepancy between the unadjusted and multivariable HRs in Table 2 (Di Castelnuovo et al. ), which are 0.80 to 0.87, is due to the substantial differences in baseline characteristics between the alcohol categories seen in Table 1 and as would be expected in a nonrandomized study. Indeed, consider the baseline characteristics of the Never Drinkers versus those in the 24.1 to 48 g/day category. The HF incidence rate for the Never Drinker category is 3.7% compared to 4.7% for the 24.1 to 48 g/day category. However, the Never Drinker category was predominantly female and younger than subjects in the 24.1 to 48 g/day category (19.6% female vs. 76.1% male; 54 ± 12 vs. 57 ± 12 years of age, respectively). Moreover, compared to the 24.1 to 48 g/day category, the Never Drinker category had lower percentages of smokers (20% vs. 26%, respectively), hypertension (51% vs. 62%, respectively), diabetes (8.3% vs. 11.9%, respectively), and cardiovascular disease (CVD; 4.6% vs. 7.1%, respectively). Thus, the HF risk factor profile was worse for the 24.1 to 48 g/day category than for the Never Drinker category, so it is conceivable that these risk factors were driving the higher event rate in the 24.1 to 48 g/day category. That the multivariable HR was 0.80 for the 24.1 to 48 g/day category versus the Never Drinker category suggests that alcohol consumption may counteract some of the risk factors’ harm.
However, as with virtually any observational study, we cannot be certain of the risk or benefit of alcohol consumption despite the best attempts at statistical adjustment. To obtain a definitive answer on risk versus benefit, a well-conducted randomized clinical trial is needed. Fortunately, the MACH15 (Moderate Alcohol and Cardiovascular Health Trial; NCT03169530) is anticipated to begin recruitment in the fall of 2017. MACH15 will randomize approximately 7,800 participants worldwide, 50 years of age and older, with advanced CVD risk to consume approximately 15 g/day of alcohol or to abstain. The average follow-up will be 6 years. The primary outcome measurement will be a composite CVD endpoint, but several other outcomes will be collected including incident diabetes, cancer, HF, AF, and hemorrhagic stroke. Primary funding will be provided by the U.S. National Institute on Alcohol Abuse and Alcoholism. Although estimated study completion date is not until 2026, the results of this unprecedented randomized controlled trial of alcohol consumption should expand our understanding of the risks and benefits of MAC.
The author thanks Drs. James Troendle and Lauren Kunz for their careful review of this article.
↵∗ Editorials published in JACC: Heart Failure reflect the views of the authors and do not necessarily represent the views of JACC: Heart Failure or the American College of Cardiology.
The views expressed in this article are those of the author and do not necessarily represent the views of the National Heart, Lung, and Blood Institute; the National Institutes of Health; or the U.S. Department of Health and Human Services. Dr. Leifer has reported that he has no relationships relevant to the contents of this paper to disclose.
- Di Castelnuovo A.,
- Costanzo S.,
- Bonaccio M.,
- et al.
- ↵World Health Organization. Global Status Report on Alcohol and Health. 2014 edition. Geneva. Available at: http://apps.who.int/iris/bitstream/10665/112736/1/9789240692763_eng.pdf. Accessed September 19, 2017.
- Brien S.E.,
- Ronksley P.E.,
- Turner B.J.,
- et al.
- Schütze M.,
- Boeing H.,
- Pischon T.,
- et al.