Moderate Drinking and Reduced Risk of Heart Disease

Arthur L. Klatsky, M.D.

Although heavier drinkers are at increased risk for some heart diseases, moderate drinkers are at lower risk for the most common form of heart disease, (CAD) than are either heavier drinkers or abstainers. This association has been demonstrated in large-scale epidemiological studies from many countries. Abstainers may share traits potentially related to CAD risk, such as psychological characteristics, dietary habits, and physical exercise patterns. However, evidence supports a direct protective effect of , even after data have been adjusted for the presence of these factors. The alcohol-CAD relationship is also independent of the hypothetically increased risk status among abstainers who stopped drinking for medical reasons. All alcoholic beverages protect against CAD, although some additional protection may be attributable to personal traits or drinking patterns among people who share some beverage preferences or to nonalcohol ingredients in specific beverages. Alcohol’s protective effect may result from favorable alterations in blood chemistry and the prevention of clot formation in arteries that deliver blood to the heart muscle. Because CAD accounts for a large proportion of total mortality, the risk of death from all causes is slightly lower among moderate drinkers than among abstainers, but heavier drinkers are at considerably higher total mortality risk. KEY WORDS: moderate AOD use; heavy AOD use; coronary artery disorder; risk factors; protective factors; AODR (alcohol and other drug related) disorder; AODR mortality; alcoholic ; hypertensive disorder; cardiac ; stroke; alcoholic beverage; public health; AOD use frequency; literature review

iseases of the heart and blood disorders (i.e., cardiomyopathy), high truth, the relationships between drink- vessels (i.e., cardiovascular blood pressure (i.e., hypertension), brain ing and CVD are both complex and Ddiseases [CVDs]) are a major damage from ruptured blood vessels interconnected (Klatsky 1995a). Discus- cause of illness and disability in the (i.e., hemorrhagic stroke), and heart sions of alcohol’s effects on CVD must United States. Heart disease is the rhythm irregularities (i.e., ). clearly differentiate between different leading cause and stroke is the third Lighter drinking is related to lower levels of drinking as well as the specific leading cause of death among adult risk of coronary artery disease (CAD) type of CVD being considered. Americans. Together, these two condi- and of ischemic stroke, which is char- This article briefly reviews the tions account for more than 40 percent acterized by blockage of blood vessels effects of heavier drinking on certain of all deaths annually (Dufour 1996). that supply the brain.1 Researchers have studied extensively Sweeping generalizations circulated ARTHUR L. KLATSKY, M.D., is a senior the role of alcohol in CVD, especially in the popular media have perpetuated consultant in at the Division in relation to drinking level. Heavier public misconceptions about the effects of Cardiology, Department of Medicine, drinking (see the following section, of moderate drinking on CVD. In and the Division of Research, Kaiser “Definitions of Moderate Drinking”) Permanente Medical Center, Oakland, is related to higher risk of heart muscle 1See glossary, p. 23, for terms not defined in the text. California.

Vol. 23, No. 1, 1999 15 CVDs and considers at greater length in drinking patterns. Subjects in popu- guishable either clinically or patholog- the cardiovascular effects of moderate lation studies are generally requested to ically. Thus, alcohol cardiomyopathy drinking. The article concentrates on describe their average total alcohol con- is generally diagnosed when dilated CAD for two major reasons. First, CAD sumption over a given time period (e.g., cardiomyopathy of unknown origin is is the most common type of CVD and 1 week or 1 month). Thus, a person who encountered in a patient with a his- therefore dominates epidemiological habitually consumes 2 drinks each even- tory of long-term heavy drinking. statistics when CVDs are studied as a ing might report the same average weekly The proportion of cardiomyopathy group. Second, alcohol has been reputed consumption as a person who hypothet- attributable to alcohol is unknown. to have a protective effect against CAD. ically consumes 14 drinks within a Although substantial circumstan- The discussion of CAD suggests possi- few hours every Saturday night. The tial evidence implicates alcohol as a ble mechanisms to account for this cause of , the protective effect, including the poten- absence of specific diagnostic tests tial role of beverage type. Finally, the seriously impedes epidemiological article discusses implications of the study. The most convincing circum- alcohol-CVD relationship in terms of Moderate drinking stantial evidence for alcoholic cardio- total mortality and overall public health. does not appear is the extensive data, in animals and humans, of nonspecific to increase the cardiac abnormalities related to alco- Definitions of Moderate hol. One key report (Urbano-Marquez Drinking risk of alcoholic et al. 1989) showed a clear relationship cardiomyopathy. in alcoholics between lifetime alcohol Definitions of moderate drinking vary consumption and both structural and widely (see the article by Dufour, pp. functional abnormalities of heart and 5–14). This article defines moderate skeletal muscle. The amounts of alco- drinking as the consumption of fewer hol necessary to produce such evidence than three standard drinks per day. A health risks posed by such widely were large, equivalent to eight or more standard drink is equivalent to approx- varying patterns of consumption may drinks per day over a period of 20 imately 12 ounces (oz) of beer, 5 oz of differ substantially. years. Thus, moderate drinking does wine, or 1.5 oz of distilled spirits, each not appear to increase the risk of alco- of which contains approximately 12 holic cardiomyopathy. grams (0.5 oz) of alcohol. Effects of Heavy Drinking Both clinical observations and exper- Problems inherent to epidemiological imental data (Ballester et al. 1997) sug- studies of alcohol and CVD include gest that can individual susceptibility and the catego- Alcoholic Cardiomyopathy improve with abstinence, which is there- rization of drinking levels. Individual fore a cornerstone of therapy. factors that influence interpretation of Clinicians and researchers have long rec- study results include sex, age, dietary ognized that alcohol consumption can Hypertension habits, cigarette smoking, the consump- directly damage heart muscle cells inde- tion of caffeinated beverages, and various pendently of any other cardiovascular Although an association between heavy psychosocial factors that are difficult to effect (Klatsky 1995a). Breathlessness alcohol consumption and hypertension characterize and measure. and fatigue may be early signs of such was recognized as early as 1915 (Lian The characterization of drinking lev- heart muscle disease (i.e., cardiomyopa- 1915), the epidemiological significance els is complicated by the use of subjects’ thy). Complications may develop as the of this finding was largely ignored for self-reported estimates. In particular, disease progresses, including heart fail- the next 60 years. Since the mid- heavier drinkers may underestimate ure, embolism, and arrhythmias, possi- 1970s, alcohol has joined obesity and their alcohol consumption. If some bly resulting in sudden death. salt intake as an acknowledged factor proportion of heavy drinkers report Although can be potentially contributing to the risk for lighter drinking, then the “lighter” caused by viral infection, exposure to hypertension. Two types of epidemio- drinking group will include some toxic substances and, possibly, genetic logical studies provide evidence for this people who really drink more, and a factors, most cases are considered to relationship. Cross-sectional studies condition related only to heavy drink- be of unknown cause. The type of seek a correlation between reported ing (e.g., alcoholic cardiomyopathy) heart muscle damage produced by alcohol use and blood pressure; pro- may erroneously appear to be related alcohol is called dilated cardiomyopa- spective studies follow subjects over to moderate alcohol consumption thy, because one or more heart cham- time to document the development of (Klatsky 1994). bers are abnormally distended with increased blood pressure in relation to Another potential source of error is blood. Alcoholic and nonalcoholic reported alcohol use. Of the more than failure to consider individual differences cardiomyopathy are not readily distin- 50 cross-sectional and 10 prospective

16 Alcohol Research & Health Moderate Drinking and Heart Disease

population studies, almost all have abstention or marked reduction of or more drinks per day as among peo- demonstrated a link between alcohol alcohol consumption (Klatsky 1995b). ple consuming less than one drink per and hypertension in nonhospitalized The association between alcohol con- day. Certain arrhythmias are extremely persons in a number of countries sumption and blood pressure appears common in older persons, and only a (Klatsky 1995b). Studies differ about to be independent of salt intake, small proportion of arrhythmias result whether a threshold of alcohol con- physical activity, and psychosocial from heavy alcohol consumption. sumption exists below which the stress. No elevations of blood pressure hypertensive effect does not occur. The have been reported as a result from Hemorrhagic Stroke preponderance of evidence suggests alcohol withdrawal (Klatsky 1995b). that clinically significant hypertension These results suggest that alcohol Alcohol-stroke relationships are com- is related only to heavier drinking. plays a contributory role in the devel- plex (Van Gign et al. 1993). Several Two Kaiser Permanente studies opment of hypertension, at least in reports suggest that alcohol consump- are among the largest of the cross- the short term. However, research has tion, especially at heavier drinking sectional population surveys. failed to provide a convincing biologi- levels, is associated with higher risk of The first of these studies (Klatsky cal explanation for alcohol’s putative stroke. Some studies examined only et al. 1977) showed slightly lower causal effect. Studies of the long-term drinking sprees; others did not differ- blood pressures among women who health effects of sustained hypertension entiate between hemorrhagic and consumed no more than two drinks (e.g., CAD, stroke, and congestive ischemic strokes. Several studies have per day; the study found no such rela- ) are difficult to interpret suggested that alcohol is related only tionship among moderately drinking because of alcohol’s ability to influence to hemorrhagic stroke; others have men, although blood pressures were these same conditions directly. Estimates shown drinkers to be at higher risk for highest among heavier drinkers (three of the proportion of all hypertension hemorrhagic stroke but at lower risk for or more drinks per day) in both sexes. attributable to alcohol consumption ischemic stroke. Elevated blood pres- These results were independent of range from 5 to 30 percent. Because sure associated with heavier drinking other potentially contributory factors hypertension affects an estimated 50 may contribute to the relationship such as age, sex, race, smoking, coffee million Americans, even the lowest between heavy alcohol consumption intake, reported past heavy drinking, estimate suggests that heavy drinking and hemorrhagic stroke, and alcohol’s education, obesity, and habitual salt may be the most common cause of inhibitory effects on the mechanism use. Hypertension (e.g., blood pres- potentially reversible hypertension in of blood clotting may contribute to sure greater than 160/95 mm Hg developed societies (Fifth Report 1993). increased risk at both lighter and heav- [millimeters of mercury], the units of ier drinking levels. Hemorrhagic strokes measure on a blood pressure reading) Cardiac Arrhythmias carry a high risk of death (as much as occurred among twice as many people 50 percent in several reports) and per- who reported consuming six or more An association of heavier alcohol con- manent disability. No clear data exist drinks per day as among light drinkers sumption with disturbances of normal that afford an estimate of the pro- or abstainers. heart rhythm has been suspected for portion of such deaths that might be The second Kaiser Permanente study decades. Ettinger and colleagues (1978), attributable to alcohol. (Klatsky et al. 1986a) demonstrated describing a series of patients hospital- similar results following adjustment ized for episodes of arrhythmia follow- for the influence of age, obesity, smok- ing heavy weekend or holiday drinking Effects of Moderate ing, consumption of coffee or tea, and sprees, referred to this phenomenon Drinking certain biochemical factors measured as the “.” by blood chemistry analysis. Former Symptoms of acute alcohol-induced drinkers did not have higher blood arrhythmia include and Ischemic Stroke pressure than lifelong abstainers, and difficulty breathing. Onset usually subjects’ reported alcohol consumption follows a large meal accompanied by Although hemorrhagic strokes are asso- during the week before examination heavy drinking. Symptoms usually ciated with higher mortality, ischemic suggested rapid reversal of alcohol- abate on abstinence, with or without strokes are a common cause of serious associated hypertension with abstinence. other specific treatment. disability, especially among the elderly Several clinical experiments have A study of approximately 4,000 (see glossary, p. 23). The relationship been undertaken on alcohol’s short- subjects investigated the occurrence of of alcohol to ischemic stroke is com- term blood pressure effects (i.e., effects various arrhythmias, including rapid, plex, because underlying heart disease that develop within a period of a few irregular, and premature contractions of almost any type predisposes to days to several weeks). Consumption of the upper chambers of the heart. embolism, as does atherosclerotic dis- of three to eight alcoholic drinks per The relative risk for these conditions ease in the blood vessels within, or day was associated with increased (Cohen et al. 1988) was at least twice leading to, the brain. Although further blood pressure, which decreased upon as high among people consuming six study is needed, the preponderance of

Vol. 23, No. 1, 1999 17 evidence suggests that moderate drink- studies have concentrated on objective, CAD and overall CVD mortality risk ing is associated with lower risk of readily defined events, such as CAD- than lifelong abstainers, but the dif- ischemic strokes (Van Gign et al. 1993). related hospitalization or death. ference disappeared when adjusted for Most studies show that heavier other traits. Among current drinkers, Coronary Artery Disease drinkers are at similar or lower risk of lighter drinkers had the lowest risk for hospitalization for CAD than are both total CVD deaths and CAD The most common heart disease cate- moderate drinkers. Several population deaths, yielding U-shaped mortality gory is CAD, which causes about 60 studies show a progressive inverse curves with lowest risk at one to two percent of CVD deaths and 25 percent relationship between CAD-related drinks per day and at three to five of all deaths (Dufour 1996; Klatsky et deaths and the amount of alcohol drinks per day, respectively. Mortality al. 1990a, 1992). A common mani- consumption; that is, the rate of CAD curves were not influenced by differ- festation of CAD is pectoris, deaths declines steadily with increas- ences in patients’ baseline CAD risk characterized by chest pain or discom- ing alcohol intake. Data from other or by the occurrence of pre-existing fort, often of a vague or nonspecific studies, however, indicate that lighter CAD. Results of the study also con- nature. The symptoms of angina were drinkers are at lower risk for CAD firmed the observation that alcohol’s described more than two centuries ago deaths than are either heavier drinkers CVD effects depend on the specific by Heberden (1786), who also noted or abstainers. This type of association condition (see table 1). that “wine and spiritous liquors . . . can be depicted graphically by a U- The Nurses Health Study (Stampfer afford considerable relief.” Alcohol was shaped curve, with the lowest risk et al. 1988), a large prospective study widely presumed to alleviate angina among people who consume moderate of women free of baseline CAD, by dilating the coronary blood vessels. amounts of alcohol, specifically one to showed a progressive inverse relation- Electrocardiographic measurements three drinks per day. ship of alcohol use to major CAD, taken from patients during exercise, Some controversy persists about independent of any previous reduction however, suggest that alcohol does not the apparent higher CAD risk among in alcohol consumption or nutrient improve coronary blood flow or increase abstainers. For example, a much pub- intake. For women who reported daily oxygen delivery to heart muscle. Thus, licized hypothesis (Shaper et al. 1988) alcohol intake of approximately two the perceived beneficial effects of suggests that the group of abstainers drinks per day, their relative risk2 of alcohol consumption may be subjec- studied may have included persons CAD was 0.4. The overall net benefi- tive, perhaps resulting from alcohol’s with pre-existing alcohol-related cial effects of moderate alcohol in sedating or anesthetic effects. Rather health problems who were already at these women was present only among than being beneficial, such effects may high CAD risk (i.e., “sick quitters”). persons at above-average CAD risk dangerously mislead the patient by However, studies that differentiate (basically, age 50 or older); in other masking symptoms of serious heart between lifelong abstainers and past women adverse effects of alcohol dom- disease (Klatsky 1994). drinkers suggest that both of these inated or canceled the benefit (Fuchs Early in this century, some research- types of nondrinkers are at higher risk et al. 1995). ers reported that increasing levels of of CAD than are current drinkers Other large prospective studies also alcohol consumption were associated (Renaud et al. 1993; Maclure 1993; confirm the lower CAD risk of drinkers, with decreased prevalence of atheros- Klatsky 1994). In addition, a prospec- independent of confounders or the clerotic disease, including CAD. One tive Kaiser Permanente study of alco- occurrence of other diseases. The explanation offered was that prema- hol habits and CAD hospitalizations American Cancer Society Study, a 9- ture deaths related to heavy alcohol (Klatsky et al. 1986b) showed that year prospective mortality study of consumption precluded the gradual former drinkers and infrequent 490,000 men and women, showed a development of chronic age-related drinkers (i.e., those who consumed 30- to 40-percent lower total CVD changes that lead to CAD. Research less than one drink per month) were mortality among both men and women since 1974 (Klatsky 1994) has confirmed at similar CAD risk as lifelong who drink one or two drinks per day this relationship in various countries, abstainers. All other subjects had a (Thun et al. 1997). In the Health several racial groups, and both sexes lower CAD risk independent of a Professional Followup Study of 51,529 for both fatal and nonfatal CAD. number of potential indirect explana- men (Rimm et al. 1991) who were well tions (i.e., confounders), including controlled for dietary habits, newly Epidemiological Evidence. beverage choice (see the section, “The diagnosed CAD was inversely related Epidemiological investigations into the Role of Beverage Choice,” p. 19) and to increasing alcohol intake. A study relationship between alcohol consump- baseline CAD risk (i.e., pre-existing of both men and women, the Auckland tion and CAD include international CAD risk as evaluated at the patient’s Heart Study (Jackson et al. 1991), was comparisons, time-trend analyses, case initial examination). control studies, and longitudinal pop- In a prospective study of total CVD 2Relative risk is defined as the CAD death risk for ulation studies. Because the symptoms mortality (Klatsky et al. 1990a), for- each drinking category compared with the CAD of angina are subjective and variable, mer drinkers had higher age-adjusted risk of lifelong abstainers.

18 Alcohol Research & Health ModerateModerate Drinking Drinking and and Heart Heart Disease designed to study the hypothesis that fatty deposits from within large blood Some data suggest that alcohol persons at high CAD risk were likely to vessels. In the absence of severe liver affects various aspects of blood clot- become nondrinkers; the analysis showed impairment, alcohol ingestion raises ting (Renaud et al. 1993; Klatsky that moderate drinkers had lower CAD HDL levels.3 Subtypes of HDL exist, 1994; Ridker et al. 1994). Such an risk than both lifelong abstainers and and some data suggest that the HDL2 action of alcohol could partially former drinkers. subtype, which is less related to alcohol account for the lower CAD risk at Reduced risk of CAD presents at consumption, may be more protective very light drinking levels (e.g., several various ages, although its impact on than other HDL subtypes. More recent drinks per week) seen in several of the total mortality in Kaiser Permanente studies, however, suggest that both epidemiologic studies, but this pro- studies was clearest in older age brackets, HDL2 and HDL3 are protective. Bio- tective mechanism is less established and the adverse effects of alcohol were chemical pathways for the HDL effect than the HDL cholesterol pathway. greater among younger persons. Among of alcohol are poorly understood persons older than age 60, overt or (Gaziano et al. 1993; Renaud et al. The Role of Beverage Choice. latent CAD may play a role in risk of 1993; Klatsky 1994). In the early 19th century, a perceptive death from causes other than CAD Four studies have examined the Irish physician with a great interest (Klatsky et al. 1992). hypothesis that alcohol consumption in angina pectoris described what has protects against CAD by increasing Possible Protective Mechanisms. HDL cholesterol levels (Criqui et al. come to be known as the “French Possible mechanisms exist by which alco- 1987; Suh et al. 1992; Gaziano et al. Paradox” (Black 1819). Noting the hol consumption might protect against 1993; Marques-Vidal et al. 1996). high prevalence of angina in Ireland CAD (Renaud et al. 1993; Klatsky Findings suggest that higher HDL compared with that in France, he 1994). The most well-established levels in drinkers mediate approxi- attributed the difference to “the French protective effect of alcohol consump- mately one-half of the lower CAD habits and modes of living, coinciding tion is the increased concentration of risk. One report (Gaziano et al. 1993) with the benignity of their climate and a natural component of human blood suggests that both HDL2 and HDL3 the peculiar character of their moral called high-density lipoprotein (HDL) are involved in this protective effect. affections.” Using observational evi- cholesterol. This substance, often dence arguably less circumstantial, researchers are still striving 180 years referred to in the popular media as 3Conversely, low-density lipoprotein (LDL) choles- the “good cholesterol,” may protect terol is a blood constituent that contributes to the later to define those favorable “habits against CAD by helping to remove deposition of fatty deposits within the coronary arteries. and modes of living.”

Table 1 Relative Risk of Death of Various Cardiovascular Conditions and Cirrhosis by Ex-Drinkers’ Former Level of Alcohol Use

Relative Risk for Each Drinking Category (drinks per day) Condition (number of deaths) Ex-drinkers <1/mo <1/day but >1/mo 1Ð2/day 3Ð5/day 6+/day

All coronary artery disease (CAD) (600) 1.0 0.9 0.8a 0.7a 0.7a 0.8 Acute (284) 1.0 0.7 0.8 0.6a 0.5a 0.6 Other CAD (316) 0.9 1.0 0.7 0.8 0.7 1.0 Stroke (138) 1.0 0.8 0.8 0.8 0.7 1.4 Hemorrhagic (41) 1.4 1.5 1.6 1.8 1.3 4.7 Ischemic (34) 0.9 0.5 0.5 0.3 0.4 —b Nonspecific (63) 1.1 0.7 0.9 1.0 1.0 1.2 Hypertension (64) 2.8 2.4 1.9 1.3 2.2 2.1 Cardiomyopathy (24) 3.4 8.5a 4.0 5.6 2.4 8.0 Syndromes* (82) 0.6 0.6 0.5 0.4a 0.6 1.0 Arterial** (41) —b 1.1 1.6 0.4a 1.7 —b Cirrhosis (42) 10.8a 1.4 1.0 4.3 8.1a 22.0a

*Includes “symptomatic heart disease” (n = 32); disorders of heart rhythm (n = 22); and ill-defined heart disease (n = 28). **Includes arteriosclerosis (n = 15); aneurysms (n = 23); peripheral vascular disease (n = 2); and arterial embolism and thrombosis (n = 1). aSignificantly different from 1.0. bInsufficient cases for estimate. NOTE: Relative risk is defined as the CAD death risk for each drinking category compared with the CAD risk among lifelong abstainers. This comparison is expressed as a ratio, using the CAD risk among abstainers as a reference, set at 1.0. For example, people who consume one to two drinks per day are almost twice as likely to die from hemorrhagic stroke than are abstainers (i.e., relative risk = 1.8), but they are approximately one-half as likely to die from acute myocardial infarction than are abstainers (i.e., relative risk = 0.6). Adjustments have been made for the influence of gender, age, race, smoking, education, and coffee consumption. SOURCE: Adapted from Klatsky et al. 1990a.

Vol. 23, No. 1, 1999 19 Over the past two decades, inter- table 2). CAD risk did not differ sig- 1999; Klatsky 1999). Thus, it appears national comparison studies (St. Leger nificantly among subjects who drank that the healthier lifestyle of wine et al. 1979; Rimm et al. 1996) have red wine, white wine, both red and drinkers is not limited to those in the documented a lower prevalence of white wine, or other types of wine. United States. CAD in wine-drinking countries than These researchers concluded that (1) Although the effects of alcoholic in beer- or liquor-drinking countries. all alcoholic beverage types protect beverage choice on CAD risk remain These findings are complemented by against CAD and (2) some additional controversial, most researchers con- the presence in wine of potentially protection may be attributable to per- cluded that available evidence does beneficial nonalcoholic substances.4 sonal traits (e.g., race and gender) or not support a major role for beverage These substances appear to fall into drinking patterns (e.g., number of choice (Rimm et al. 1996) and that three categories. The first group con- drinks per day) that may characterize no compelling health-related data sists of chemicals that can inhibit blood groups of people who share certain exist that preclude personal preference clotting. A second group consists of beverage preferences. Similarly, earlier as a guide to one’s choice of beverage. chemicals that can relax the walls of studies had suggested that the blood vessels, potentially dilating con- decreased risk of death from CAD Does Moderate Drinking Decrease stricted coronary arteries. Finally, cer- among drinkers who prefer wine CAD Risk? The validity of the inverse tain chemicals called polyphenols can might be related to group differences alcohol-CAD relation is supported by interfere with the metabolic process in dietary habits, physical exercise, the following evidence discussed in by which LDL cholesterol promotes and the use of antioxidant supple- this section: fat deposition within blood vessels. ments (Klatsky and Armstrong 1993; The mechanism of this effect appears Klatsky et al. 1990b). • Independence of results from subjects’ to be based on the polyphenols’ antiox- The hypothesis that wine drinkers CAD risk at baseline examination idant properties. However, although are at lower CAD risk than persons diets high in antioxidants seem protec- drinking spirits or beer is supported • Absence of higher CAD risk tive against CAD, it should be noted by the Copenhagen City Heart Study among subjects who had reduced that prospective clinical trials of (Grønbæk et al. 1995). Risk of CAD their alcohol consumption for antioxidant supplements remain showed less of a decrease among beer medical reasons inconclusive (Klatsky et al. 1997). drinkers, and it increased among drink- Prospective population studies pro- ers of spirits compared with nondrinkers. vide no consensus about the role of In addition, the wine drinkers showed • Evidence that the higher unadjusted beverage choice in alcohol-related CAD a lower risk of non-CVD mortality at CAD risk of former drinkers is risk. Those studies with relevant data levels of wine consumption up to an attributable to confounding factors show that statistically significant inverse average of three to five drinks per day. associated with past alcohol use relationships to CAD for beer, liquor, Interestingly, wine drinkers who par- and wine were generally accompanied ticipated in this study reported a more • Absence of an association between by inverse, nonsignificant relationships healthy diet than did drinkers of CAD risk and maximal past alcohol for other alcoholic beverage types beer or spirits (Tjønneland et al. consumption among former drinkers (Renaud et al. 1993; Klatsky 1995a; Rimm et al. 1996). A Kaiser Permanente study deter- mined the CAD risk per drink per Table 2 Effect of Beverage Choice on Relative Risk for Coronary Artery day associated with each type of alco- Disease (CAD) holic beverage among 3,931 patients hospitalized for CAD (Klatsky et al. Relative Risk for CAD 1997). Results, not controlled for total Group Wine Liquor Beer alcohol intake, showed inverse rela- tionships between daily CAD risk and Both sexes 0.8a 0.9a 0.7a all alcoholic beverage types. When Men 0.9 0.9 0.7a results were adjusted to account for Women 0.7a 0.9 0.7 total alcohol consumption, only beer use by male subjects remained signifi- aRR significantly different from 1.0. NOTE: Relative risk (RR) is defined as the CAD death risk for each drinking category compared with the CAD cantly related to daily CAD risk (see risk among lifelong abstainers, adjusted for total alcohol consumption. For example, among both sexes, although all beverage types afford some protection against CAD (i.e., RR is less than 1.0 in each case), RR is lowest among beer drinkers and highest among those who consume liquor. This comparison is expressed as a 4These different types of protective substances ratio, using the CAD risk among abstainers as a reference, set at 1.0. Results are adjusted for age, sex, race, smoking, education, marital status, and obesity. have been reported mostly in red wine, although SOURCE: Adapted from Klatsky et al. 1997. low concentrations of polyphenols exist in certain white wines and beer.

20 Alcohol Research & Health Moderate Drinking and Heart Disease

• Absence of a relationship between exercise patterns. As previously Conclusion and Public infrequent alcohol consumption stated, however, no substantial evi- Health Considerations (i.e., less than one drink per month) dence exists indicating that such a and CAD risk trait occurs in persons of both sexes, This brief survey documents the evi- various countries, and multiple dence for the disparity of alcohol’s • Similar reduction of CAD risk racial groups. A causal, protective effects on risk for different CVDs. regardless of specific beverage effect of alcohol is a simpler and Table 3 summarizes the relationships, preference. more plausible explanation for the with emphasis on the disparity between observed association. the overall favorable relations of lighter An appropriately designed, drinking and the overall unfavorable prospective-controlled clinical trial relations of heavier drinking. could definitively rule out indirect Relationship of Drinking Nonetheless, public health advice explanations for an inverse alcohol- to Total Mortality cannot be formulated easily. A more CAD association. If many medical plausible option is for individual health conditions appeared to be related As stated earlier, the relationship of practitioners, using personal knowledge inversely to lighter drinking, the moderate alcohol consumption to of their clients, to privately advise those alcohol-CAD relationship would be overall mortality in population studies of their patients who are concerned less convincing; however, the relative is dominated by the lower CVD risk about the probable health effects of specificity of the association is strong of moderate drinkers. Moderate drinking (Friedman and Klatsky 1993). evidence of alcohol’s direct protective drinkers are at slightly lower risk than In general, heavier drinkers (i.e., effect. CAD stands almost alone in abstainers and at considerably lower men who consume three or more its inverse relationship to lighter drink- risk than heavier drinkers. Because of drinks per day and women who con- ing. Thus, nondrinkers’ increased increased non-CVD mortality among sume two or more drinks per day) are CAD risk most likely does not heavier drinkers (e.g., from cancer, at increased risk of major social, per- represent a general predilection for accidents, and cirrhosis), the overall sonal, and adverse health conse- serious illness. alcohol-mortality risk curve is more quences and should abstain from The possibility remains that life- or less J-shaped, rather than U- alcohol or reduce their level of con- long abstainers differ from drinkers shaped, with heavier drinkers at high- sumption. In addition, because most in other traits potentially related to est risk, lighter drinkers at lowest risk, nondrinkers have important personal CAD risk, such as psychological char- and abstainers at intermediate risk or health reasons for abstention, indis- acteristics, dietary habits, or physical (Klatsky 1995a; Klatsky et al. 1992). criminate advice to drink for health

Table 3 Relationships of Alcohol Consumption to Cardiovascular Diseases (CVDs)

Amount of Alcohol Drinking* Condition Moderate Heavy Comment Dilated cardiomyopathy No relationship Probably causal Unknown cofactors in some cases may influence association

Hypertension (HTN) Little/no relationship Probably causal Mechanism unknown

Coronary artery disease Protective Possibly protective Dominates all epidemiologic (CAD) data on CVD

Arrhythmia Probably none Probably causal Other alcohol-related CVDs may in some cases influence susceptibility

Hemorrhagic stroke May increase risk Increased risk Via HTN and inhibition of blood clotting

Ischemic stroke Protective Possibly protective Complex interactions with other CVDs and risks

*Moderate drinking = consumption of fewer than three standard drinks per day; heavy drinking = consumption of three or more standard drinks per day. A standard drink = approximately 12 ounces of beer, 5 ounces of wine, or 1.5 ounces of distilled spirits, each of which contains approximately 12 grams (0.5 ounce) of alcohol.

Vol. 23, No. 1, 1999 21 should not be given. Finally, the decreased risk of myocardial infarction. New England MACLURE, M. Demonstration of deductive meta- Journal of Medicine 329(25):1829–1834, 1993. analysis: intake and risk of myocardial majority of the population consume infarction. Epidemiologic Reviews 15:328–351, 1993. only moderate amounts of alcohol and GRØNBÆK, M.; DEIS, A.; SORENSON, J.A.; BECKER, are therefore, as a group, at lowest U.; SCHNOHR, P.; JENSEN, G. Mortality associ- MARQUES-VIDAL, P.; DUCIMETIÈRE, P.; EVANS, A.; ated with moderate intake of wine, beer, or spir- CAMBOU, J.-P.; AND ARVEILER, D. Alcohol con- total mortality risk. Persons at higher- its. British Medical Journal 310:1165–9, 1995. sumption and myocardial infarction: A case-control study in France and Northern Ireland. American than-average CAD risk who drink HEBERDEN, W. Some account of a disorder of the moderately should not be advised to breast. Medical Transactions of the Royal College of Journal of Epidemiology 143:1089–1093, 1996. reduce their alcohol consumption Physicians (London) 2:59–67, 1786. RENAUD, S.; CRIQUI, M.H.; FARCHI, G.; AND VEENSTRA, J. Alcohol drinking and coronary unless they are at special risk for disor- JACKSON, R.; SCRAGG, R.; AND BEAGLEHOLE, R. ders induced or aggravated by alcohol Alcohol consumption and risk of coronary heart heart disease. In: Verschuren, P.M., ed. Health disease. British Medical Journal 303(6796):211– Issues Related to Alcohol Consumption. Washington, (e.g., liver disease or ). 216, 1991. DC: ILSI Press, 1993. pp. 81–124.

KLATSKY, A.L. Epidemiology of coronary heart RIDKER, P.M.; VAUGHAN, D.E.; STAMPFER, M.J.; disease—Influence of alcohol. Alcoholism: GLYNN, R.J.; AND HENNEKENS, C.H. Association References Clinical and Experimental Research 18(1):88–96, of moderate alcohol consumption and plasma 1994. concentration of endogenous tissue-type plas- BALLESTER, M.; MARTI, V.; CARRIO, I.; OBRA- minogen activator. Journal of the American DOR, D.; MOYA, C.; PONS-LLADO, G.; BERNA, KLATSKY, A.L. Cardiovascular effects of alcohol. Medical Association 272(12):929–933, 1994. L.; LAMICH, R.; AYMAT, M.R.; BARBANOJ, M.; Scientific American Science and Medicine 2(2):28– IMM IOVANNUCCI ILLETT GUARDIA, J.; CARRERAS, F.; UDINA, C.; AUGE, 37, 1995a. R , E.B.; G , E.L.; W , W.C.; COLDITZ, G.A.; ASCHERIO, A.; ROSNER, B.; AND J.M.; MARRUGAT, J.; PERMANYER, G.; AND CARALPS- LATSKY K , A.L. Blood pressure and alcohol intake. STAMPFER, M.J. Prospective study of alcohol con- RIERA, J.M. Spectrum of alcohol-induced damage In: Laragh, J.H., and Brenner, B.M., eds. detected by Indium-111-labeled monoclonal anti- sumption and risk of coronary heart disease in Hypertension: Pathophysiology, Diagnosis and men. Lancet 388(8765):464–468, 1991. bodies. Journal of the American College of Management. 2d ed. New York: Raven Press, Cardiology 29(1):160–167, 1997. 1995b. pp. 2649–2667. RIMM, E.; KLATSKY, A.L.; GROBBEE, D.; AND STAMPFER, M.J. Review of moderate alcohol con- BLACK, S. Clinical and Pathological Reports. New KLATSKY A.L. Is it the drink or the drinker? sumption and reduced risk of coronary heart dis- York: Alex Wilkinson, 1819. pp. 1–47. Circumstantial evidence only raises a probability. (editorial). American Journal of Clinical Nutrition ease: Is the effect due to beer, wine, or spirits? COHEN, E.J.; KLATSKY, A.L.; AND ARMSTRONG, British Medical Journal 312:731–736: 1996. M.A. Alcohol use and supraventricular arrhythmia. 69:2–3, 1999. SHAPER, A.G.; WANNAMETHEE, G.; AND WALKER, American Journal of Cardiology 62(13):971–973, 1988. KLATSKY, A.L., and Armstrong, M.A. Alcoholic beverage choice and risk of coronary artery disease M. Alcohol and mortality in British men: Explaining CRIQUI, M.H.; COWAN, L.D.; TYROLER, H.A.; the U-shaped curve. Lancet 2(8623):1267–1273, 1988. BANGDIWALA, S.; HEISS, G.; WALLACE, R.B.; AND mortality: Do red wine drinkers fare best? American COHN, R. Lipoproteins as mediators for the effects Journal of Cardiology 71(5):467–469, 1993. STAMPFER, M.J.; COLDITZ, G.A.; WILLETT, of alcohol consumption and cigarette smoking on KLATSKY, A.L.; FRIEDMAN, G.D.; SIEGELAUB, W.C.; SPEIZER, F.E.; AND HENNEKENS, C.H. cardiovascular mortality: Results from the Lipid A.B.; AND GERARD, M.J. Alcohol consumption Prospective study of moderate alcohol consump- Research Clinics Followup Study. American and blood pressure. New England Journal of tion and the risk of coronary disease and stroke in Journal of Epidemiology 126(4):629–637, 1987. Medicine 296(21):1194–2000, 1977 women. New England Journal of Medicine 319(5):267–273, 1988. DUFOUR, M. Introduction: Epidemiological alcohol KLATSKY, A.L.; FRIEDMAN, G.D.; AND ARMSTRONG, studies. In: Zakhari, S., and Wassef, M., eds. Alcohol M.A. The relationship between alcoholic beverage ST. LEGER, A.S.; COCHRANE, A.L.; AND MOORE, and the Cardiovascular System. National Institute on use and other traits to blood pressure: A new F. Factors associated with cardiac mortality in devel- and Alcoholism Research Monograph Kaiser Permanente study. Circulation 73(4):628– oped countries with particular reference to con- No. 31. NIH Pub. No. 96–4133. Bethesda, MD: 636, 1986a. sumption of wine. Lancet 1(8124):1017–1020, 1979. the Institute, 1996. pp. 3–14. KLATSKY, A.L.; ARMSTRONG, M.A.; AND FRIEDMAN, SUH, I.; SHATEN, J.; CUTLER, J.A.; AND KULLER, ETTINGER, P.O.; WU, C.F.; DE LA CRUZ, C.; G.D. Relations of alcoholic beverage use to subse- K.H. Alcohol use and mortality from coronary WEISSE, A.B.; AHMED, S.S.; AND REGAN, T.J. quent coronary artery disease hospitalizations. heart disease: The role of high-density-lipoprotein Arrhythmias and the “holiday heart”: Alcohol- American Journal of Cardiology 58(9):710–714, 1986b. cholesterol. Annals of Internal Medicine associated cardiac rhythm disorders. American 116:881–887, 1992. Heart Journal 95:555–562, 1978. KLATSKY, A.L.; ARMSTRONG, M.A.; AND FRIEDMAN, G.D. Risk of cardiovascular mortality THUN, M.J.; PETO, R.; LOPEZ, A.D.; MONACO, Fifth Report of the Joint National Committee on in alcohol drinkers, exdrinkers, and non-drinkers. J.H.; HENLEY, S.J.; HEATH, C.W., JR.; AND DOLL, R. Detection, Evaluation, and Treatment of High American Journal of Cardiology 66(17):1237– Alcohol consumption and mortality among mid- Blood Pressure. Archives of Internal Medicine 1242, 1990a. 153(2):158–183, 1993. dle-aged and elderly U.S. adults. New England KLATSKY, A.L.; ARMSTRONG, M.A.; AND KIPP, H. Journal of Medicine 337(24):1705–1714, 1997. FRIEDMAN, G.D., AND KLATSKY, A.L. Is alcohol Correlates of alcoholic beverage preference: Traits TJØNNELAND, A.; GRØNBÆK, M.; STRIPP, C.; AND good for your health? New England Journal of of persons who choose wine, liquor or beer. British Medicine 329(25):1882–1883, 1993. OVERVAD, K. Wine intake and diet in a random Journal of Addiction 85(10):1279–1289, 1990b. sample of 48763 Danish men and women. American UCHS TAMPFER OLDITZ F , C.S.; S , M.J.; C , G.A.; KLATSKY, A.L.; ARMSTRONG, M.A.; AND FRIED- Journal of Clinical Nutrition 69(1):49–54, 1999. IOVANNUCCI ANSON AWACHI G , E.L.; M , J.E.; K , I.; MAN , G.D. Alcohol and mortality. Annals of URBANO-MARQUEZ, A.; ESTRUCH, R.; NAVARRO- HUNTER, D.J.; HANKINSON, S.E.; HENNEKENS, Internal Medicine 117(8):646–654, 1992. C.H.; AND ROSNER, B. Alcohol consumption and LOPEZ, F.; GRAU, J.M.; AND RUBIN, E. The effects mortality among women. New England Journal of KLATSKY, A.L.; ARMSTRONG, M.A.; AND FRIED- of alcoholism on skeletal and . New Medicine 332(19):1245–1250, 1995. MAN, G.D. Red wine, white wine, liquor, beer, England Journal of Medicine 320(7):409–415, 1989. and risk for coronary artery disease hospitalization. VAN GIGN, J.; STAMPFER, M.J.; WOLFE, C.; AND GAZIANO, J.M.; BURING, J.E.; BRESLOW, J.L.; American Journal of Cardiology 80:416–420, 1997. GOLDHABER, S.Z.; ROSNER, B.; VANDENBURGH, ALGRA, A. The association between alcohol con- M.; WILLETT, W.; AND HENNEKENS, C.H. LIAN, C. L’alcoholisme cause d’hypertension sumption and stroke. In: Verschuren, P.M., ed. Moderate alcohol intake, increased levels of high arterielle. Bulletin de l’Academie de Medicine Health Issues Related to Alcohol Consumption. density lipoprotein and its subfractions, and (Paris) 74:525–528, 1915. Washington, DC: ILSI Press, 1993. pp. 43–80.

22 Alcohol Research & Health Moderate Drinking and Heart Disease

GLOSSARY Aneurysm: A dilated portion of an artery, probably caused by Embolism: Blood clots or fatty substances that travel through a structural weakness in the arterial wall and commonly the bloodstream. Emboli that lodge in blood vessels of the associated with atherosclerosis. Although some aneurysms brain can cause ischemic stroke. are without symptoms, they may promote the formation Hemorrhagic stroke: A condition in which a ruptured artery of emboli and are subject to sudden rupture. interrupts the brain’s blood supply, depriving the affected Angina pectoris: A manifestation of coronary artery disease area of oxygen. The pressure of accumulating blood may also (CAD), characterized by poorly defined chest discomfort. cause tissue swelling and damage. Symptoms may include Arrhythmia: Irregular heartbeat. sudden severe headache, vomiting and, ultimately, coma. Atherosclerosis: Thickening and hardening of the coronary Hypertension: High blood pressure, frequently defined as arteries, caused largely by the deposition of fatty substances. pressure in excess of 140/90. The significance of a blood pressure measurement may depend on age, emotional Cardiomyopathy: Any disease of heart muscle, often charac- state, or other factors. terized by enlargement and flabbiness of the heart muscle, leading to fatigability, , and progressing : Local oxygen deprivation within a tissue, generally to congestive heart failure. resulting from blockage of a blood vessel. Congestive heart failure: Inadequacy of heart function, often Ischemic stroke: Brain damage caused by blockage of an characterized by shortness of breath, fluid accumulation in artery supplying the brain. Symptoms of sensory, intellec- tissues, and complications including arrhythmia, embolism, tual, or motor impairment may appear suddenly, often and sudden death. resulting in death or permanent damage. Coronary artery disease (CAD): Narrowing of the coronary Myocardial infarction: A “heart attack”; a manifestation of arteries resulting from atherosclerosis, which promotes the CAD in which oxygen deprivation leads to death of heart formation of blood clots that may obstruct the flow of muscle cells in the affected region, followed by scar forma- blood in the arteries, depriving heart muscle tissue of its tion that may continue to impair heart function. The con- oxygen supply. The major manifestations of CAD are dition often leads to death, instantly or later, sometimes angina, myocardial ischemia, and sudden death. from another heart attack, or from complications such as arrhythmias, heart failure, or embolism. Dilated cardiomyopathy: Any cardiomyopathy in which one or more chambers of the heart are abnormally distended Peripheral vascular disease: Any disorder of the circulatory with blood. This category of conditions includes alcoholic system that does not directly involve the heart or coronary cardiomyopathy. arteries. The term is often taken to exclude strokes as well.

What can employers do to address alcohol abuse in the workplace? he answer to this and other questions can be found in Alcohol Alert, the quarterly bulletin published by the National Institute on Alcohol Abuse and Alcoholism. Alcohol Alert provides timely information on Talcohol research and treatment. Each issue addresses a specific topic in alcohol research and summarizes critical findings in a brief, four-page, easy-to-read format. Our most recent issue—Alcohol and the Workplace (No. 44)—discusses recent research on the effects of employee drinking, workplace alcohol policies, and workplace-based programs for preventing and reducing alcohol-related problems. Forthcoming issues include the following: • Alcohol and the Cardiovascular System (No. 45), which summarizes recent research on the association between alcohol and the risk for coronary artery disease.

For a free subscription to Alcohol Alert, write to: National Institute on Alcohol Abuse and Alcoholism, Attn.: Alcohol Alert, Publications Distribution Center, P.O. Box 10686, Rockville, MD 20849Ð0686. Fax: (202) 842Ð0418. Alcohol Alert is available on the World Wide Web at http://www.niaaa.nih.gov

Vol. 23, No. 1, 1999 23 Are the latest NIAAA Research Monographs in your collection?

Each monograph presents research by noted scientists, reviews research progress, and offers a glimpse of future research in key areas. Scientists, clinicians, and oth- ers with an interest in alcohol research will find these volumes a welcome addi- tion to their library.

Women and Alcohol: Issues for Prevention Research (No. 32) presents research on alcohol use and prevention among women. Topics include • Alcohol use across the life span • Alcohol use in the workplace • Alcohol-related birth defects • Parenting interventions for preventing children’s alcohol and other drug use • Influence of genetics, sexuality, and violent victimization on alcohol use.

Alcohol Problems and Aging (No. 33) reviews research on alcohol’s effects on the aging process and on the social, eco- nomic, and health status of older Americans. Topics include • Biological mechanisms underlying alcohol’s effects on the elderly • How alcohol affects cognition, sleep, and driving • Medical consequences of heavy drinking by the elderly • Life-context factors and late-life drinking behavior • Treatment and prevention of alcohol problems in the elderly.

Supplies are limited. To order your free copies, contact

National Institute on Alcohol Abuse and Alcoholism Publications Distribution Center, P.O. Box 10686 Rockville, MD 20849–0686 You may fax your order to (202) 842–0418 or order through NIAAA’s World Wide Web site (www.niaaa.nih.gov).

24 Alcohol Research & Health