In this prospective cohort study of women, free of reported CVD at study entry, the relationship between light-to-moderate alcohol intake and SCD was U-shaped; women with alcohol intake of 5.0-14.9 g or approximately ½-1 drinks per day had a 36% lower risk of SCD compared to abstainers, and there was no evidence that the benefit of alcohol was restricted to one type of beverage. Higher intakes of alcohol (>30 g or 2 drinks/day) were not associated with increased risk of SCD; however, the number of SCD among women who consumed >30 g/day was limited. In contrast, the association between alcohol and non-sudden cardiac events was more linear with significantly lower relative risks at higher levels of alcohol intake. This inverse association between alcohol and non-sudden CHD risk was remarkably consistent with results observed in the prior report from this cohort, with only 4 years of follow-up(13
The U-shaped association for SCD observed in this prospective cohort of women is consistent with the results in another prospective study of men. In the Physicians’ Health Study, men who consumed light-to-moderate amounts of alcohol (2-6 drinks/week) had up to an 80% lower risk of SCD compared to nondrinkers, while men with higher intake (≥2 drinks/day) had no significant increase in risk(9
). Similar to the present study, associations for non-sudden CHD events differed, and higher levels of alcohol intake associated with lower risks of non-fatal MI and non-sudden CHD death. However, despite these consistencies, several earlier prospective cohort studies did not observe a reduction in SCD risk associated with moderate alcohol intake despite observing reductions in other CHD endpoints(6
). In these early studies, the number of SCD cases was relatively small, between 58 and 117 cases, which may have limited the ability to detect significant reductions in risk. Furthermore, tests for non-linear associations were not conducted, and moderate alcohol intake included intakes as high as 40 g/day (3-4 drinks per day) in some of these studies(7
In contrast to other prospective studies(5
), we found no significant increase in SCD risk at any level of alcohol intake; however, the range of alcohol intake among these female nurses was truncated, with few women (4%) reporting moderate-to-heavy consumption (>30 g or >2 drinks/day), and even fewer (<1%) reporting more extreme intake (>50 g/day). Therefore, we could not assess the hazards of heavy alcohol consumption on SCD risk. Among men without pre-existing CHD, heavy drinking was associated with greater risk of sudden death(6
). We did not observe an increased risk among women with pre-existing CHD, although power to detect an association was limited. Alternatively, while CHD underlies the majority of SCD events in men, the etiology of SCD among women is more likely to be from other causes(21
), and therefore the interaction between alcohol and pre-existing disease may not hold among women.
In experimental studies, alcohol exerts proarrhythmic effects, including inhibition of cardiac sodium channel gating, increased ventricular repolarization, increased heart rate, decreased heart rate variability and increased sympathetic activity(22
). However, alcohol also has beneficial effects on pathways involved in atherosclerosis progression, including plaque rupture(26
), and hemostatic factors(28
), through which it may lower risk of coronary disease and SCD. Additionally, moderate alcohol intake was positively associated with blood concentrations of marine n-3 fatty acids(29
), which have anti-arrhythmic properties and are strong predictors of SCD risk(30
). The differential dose-response of alcohol with sudden compared to non-sudden cardiac events suggests potentially different biological effects of alcohol at various doses. Alcohol intake at low-to-moderate levels may lower risk of both coronary disease and SCD through processes related to atherosclerosis, while at higher doses, the proarrhythmic effects of alcohol may nullify any potential benefits for SCD.
This study has several additional limitations that warrant discussion. Although this is the largest prospective study to examine the relationship between alcohol and SCD in women, we were unable to assess the role of drinking patterns, and specifically binge drinking, on risk of SCD. Binge drinking, but not usual alcohol intake, was associated with increased risk of SCD in men(7
), but data among women are scarce. Drinking frequency modifies the association between heavy alcohol intake and risk of coronary disease. In a recent meta-analysis, heavy alcohol drinkers who drank more frequently had a lower risk of CHD, while irregular and binge drinkers had an elevated risk(31
). However, the prevalence and importance of binge drinking may be greater in men than in women(32
). Therefore, more research is needed to understand the role of drinking patterns in relation to SCD and to evaluate potential effect modification of this relationship by gender.
These women may consume alcohol in a more healthful pattern. In survey data from the general US population, 6% of women age 20 or older reported consuming ≥2 drinks/day on average(33
), and between 2.3% and 5.0% of women 45 and older report drinking ≥7 drinks/week on average(34
). In this population, more women consume ≥1 drink/day (11%), but fewer women consume >2 drinks/day (4%). Although we attempted to control for confounding by healthy lifestyle, residual confounding remains possible. Furthermore, the homogeneity of this cohort of US female registered nurses may limit the generalizability of these findings. However, this homogeneity also minimizes variation in socioeconomic status and other potential confounders. We also had limited power to detect plausible interactions by incident CHD or to explore potential differing associations by beverage type, particularly for subcategories such as red versus white wine where sparse data precluded meaningful analyses. Finally, drug therapy evolved drastically throughout this study. For example, women were enrolled during in an era of Type I antiarrhythmic drugs and had a much lower prevalence of CVD medications, like ACE inhibitors or beta-blockers. Unfortunately, we do not have comprehensive data on drug therapies, in particular during the early follow-up years, and could not control for these factors adequately.
An important strength of these data is the repeated assessment of alcohol intake, which allowed us to explore relatively short-term effects of alcohol, which may be the most biologically relevant period for arrhythmias and SCD. We separated the former drinkers from abstainers using the updated alcohol data, minimizing any potential “sick quitter” bias to the extent possible. However, when we included former drinkers in the referent non-drinking category, the results were similar. Additionally, the categories of alcohol intake in previous studies included a relatively wide range of intake (i.e. 2-6 drinks/day). The details on alcohol intake in the FFQ used in this analysis allowed for the calculation of alcohol in grams per day, which enhanced the precision of our exposure.