Alcohol Consumption and Risk for Congestive Heart Failure in the Framingham Heart Study

Regular, heavy consumption of alcohol is associated with subclinical impairment of left ventricular function (1-3) and occasionally results in overt cardiomyopathy (4). This may be a consequence of direct toxic effects of alcohol or its metabolites (5); coexisting malnutrition (6); associated hypertension (7, 8); increased ventricular mass (9) or, rarely, toxic additives to alcoholic beverages (10). Conversely, moderate alcohol consumption appears to be protective against coronary heart disease (11-14). Myocardial infarction is an important risk factor for congestive heart failure (15-18). By preventing coronary heart disease, moderate alcohol consumption may indirectly protect against congestive heart failure secondary to myocardial infarction. Thus, the relation of alcohol consumption to the risk for congestive heart failure is probably complex, reflecting the interplay of its coronary protective effects and its myocardial toxic effects. Little is known about the effect of alcohol consumption on the incidence of congestive heart failure in community-based populations. We therefore sought to determine the relation between alcohol consumption and risk for congestive heart failure in participants in the Framingham Heart Study. Methods Study Sample The Framingham Heart Study is a prospective epidemiologic cohort study established in 1948 to evaluate potential risk factors for coronary heart disease. The original cohort consisted of 5209 residents of Framingham, Massachusetts, 28 to 62 years of age at entry, who have undergone follow-up evaluations every 2 years. In 1971, 5124 additional participants (the offspring of the original participants and their spouses) were enrolled into the Framingham Offspring Study. These participants undergo follow-up evaluations every 4 years. The study design and entry criteria for both cohorts are described elsewhere (19-22). The study sample was drawn from members of the original cohort who attended examination 12, 15, 17, 20, or 22 (1971 to 1994) and members of the offspring cohort who attended examination 2, 4, or 5 (1979 to 1995). Each of these examinations served as the baseline for subsequent follow-up intervals. Participants with complete information on alcohol consumption at a baseline examination and an examination 6 to 10 years earlier (to identify former drinkers) were eligible for the analysis. All examinations and procedures were approved by the institutional review board of Boston University School of Medicine, and all participants gave informed consent. Ascertainment of Alcohol Consumption At examinations 12 through 15 and 17 through 22 of the original cohort and at all examinations of the offspring cohort, participants were asked about the number of 1.5-oz cocktails, 12-oz glasses (or cans) of beer, and 4-oz glasses of wine they consumed in 1 week. Alcohol intake (g/wk) among participants who consumed at least 1 drink/wk was computed by using the following equation: 28.35 + (0.57 + the number of cocktails per week + 0.44 + the number of beers per week + 0.40 + the number of glasses of wine per week) (23). For each participant, the number of drinks consumed per week was calculated by assuming that each drink contains 13 g of alcohol (24). Men who consumed 15 drinks/wk or more and women who consumed 8 drinks/wk or more were considered to be heavy drinkers in accordance with definitions established by the National Institute of Alcohol Abuse and Alcoholism (25). Participants who reported consuming less than 1 drink/wk at the index examination were further separated into two groups on the basis of reported alcohol consumption at an examination 6 to 10 years before the index examination. Participants who reported consumption of less than 1 drink/wk at the index examination but more than 1 drink/wk at the previous examination were classified as former drinkers. Participants who reported consumption of less than 1 drink/wk at both examinations were classified as nondrinkers and served as the reference group. Baseline Measurements Medical histories and physical examinations were performed for each participant at every clinic visit. Systolic and diastolic blood pressure were measured twice in the left arm of each participant. The average of the two readings was used for each blood pressure variable. The diagnosis of hypertension was based on a systolic blood pressure of 140 mm Hg or greater, a diastolic blood pressure of 90 mm Hg or greater, or current use of antihypertensive drugs (26). Total and high-density lipoprotein (HDL) cholesterol levels were measured at each examination except examination 17 of the original cohort. For participants attending examination 17, total and HDL cholesterol levels measured at examination 15 were used. Body mass index was calculated as weight in kilograms divided by the square of the height in meters. Participants were separated according to smoking status into nonsmokers, former smokers, and current smokers (those who smoked cigarettes regularly within 1 year of the index examination). The number of cigarettes smoked per day was recorded for current smokers. Diabetes was defined as a nonfasting blood glucose level of 11.1 mmol/L or greater ( 200 mg/dL), a fasting blood glucose level of 7.8 mmol/L or greater ( 140 mg/dL), or use of insulin or an oral hypoglycemic agent (27). Diagnostic criteria for coronary heart disease (including myocardial infarction, coronary insufficiency, and angina pectoris) are described elsewhere (22). Valvular heart disease was defined as the presence of a systolic murmur louder than II/VI or any diastolic murmur. Electrocardiographic left ventricular hypertrophy was diagnosed if a participant had voltage criteria for left ventricular hypertrophy accompanied by lateral repolarization changes (28). Outcome Measurements The primary outcome of interest was incident congestive heart failure. Participants were monitored for the development of congestive heart failure and other cardiovascular events by using routine periodic clinic examinations and surveillance procedures. Information about such events was obtained through medical history, physical examination, and hospitalization records and by communication with personal physicians. All suspected new events were reviewed by a panel of three experienced investigators who evaluated all pertinent medical and hospital records and pathology reports. Table 1 shows criteria for diagnosis of congestive heart failure. Congestive heart failure was diagnosed if at least two major criteria or one major and two minor criteria were present. Minor criteria were acceptable only if they could not be attributed to another medical condition (such as pulmonary hypertension, chronic lung disease, cirrhosis, ascites, or the nephrotic syndrome) (29). Table 1. Criteria for Congestive Heart Failure The criteria used to identify participants with congestive heart failure in the Framingham Heart Study compare favorably to other clinically based criteria used to identify persons with congestive heart failure (30) and left ventricular systolic dysfunction (31). Mosterd and colleagues (30) reported that the Framingham Heart Study criteria were 100% sensitive and 78% specific for identifying persons with definite congestive heart failure as determined by clinical assessment by a cardiologist. Statistical Analysis Separate analyses were performed for men and women. Because men and women differed markedly in reported alcohol intake, level of alcohol consumption was categorized differently for each sex. Baseline risk factors were computed for each level of alcohol consumption. Age-adjusted and multivariable-adjusted hazard ratios for congestive heart failure were calculated by using Cox proportional-hazards regression models (32), with pooled follow-up periods and reported alcohol intake at original cohort examinations 12, 15, 17, 20, and 22 and offspring cohort examinations 2, 4, and 5. Nondrinkers served as the reference group. The following covariates were included in the multivariable-adjusted models: age, smoking (cigarettes/d), body mass index, diabetes, valvular heart disease, HDL cholesterol level, and hypertension. The maximum duration of follow-up was 5 years. For participants attending consecutive examinations that were fewer than 5 years apart (original cohort examinations 15 and 17, and 20 and 22 and offspring examinations 4 and 5), follow-up was censored at the time of the repeated examination (approximately 4 years later). Participants who attended more than one index examination could contribute to multiple observations periods. We used the AndersonGill model for multiple failure times (33) to account for the potential for clustering of observations by participant. Because the effect of alcohol consumption on development of congestive heart failure may be mediated through the effects of alcohol on hypertension or HDL cholesterol level, we performed secondary analyses using multivariable-adjusted models, with and without hypertension status and HDL cholesterol as covariates. To determine the association between alcohol consumption and risk for congestive heart failure without interceding myocardial infarction, we calculated age-adjusted and multivariable-adjusted hazard ratios, censoring participants at the time of myocardial infarction that occurred during follow-up. To adjust for differences in medical therapy across alcohol consumption categories, models were constructed that adjusted for medications taken at the index examination (aspirin, -blockers, diuretics, calcium-channel blockers, and angiotensin-converting enzyme inhibitors) in addition to the covariates included in the model used for the primary analysis. For the final model, the proportional hazards assumption was tested and found to be appropriate. Analyses for length-selection biases showed that from the earliest ascertainment of alcohol consumption in the Framingham Study until the period of the current primary analysi