The Effect of Reference Group Classification and Change in Alcohol Consumption on the Association Between Alcohol Consumption and Cardiovascular Disease.

ABSTRACT

BACKGROUND: Many studies suggest that mild alcohol consumption can help avert cardiovascular disease (CVD). This study investigated the association between alcohol consumption and CVD incidence, and assessed whether this differed by reference group classification. As alcohol consumption amounts may change over time, the results of simple and time-dependent analyses were compared. METHODS: Data were from a community-based cohort study on 40- to 69-year-old Koreans recruited in 2001 to 2002. A total of 8,330 participants were followed up for 10 years and classed as nondrinkers (0 g/d), drinker group 1 (<15 g/d), and drinker group 2 (≥15 g/d). The risk of CVD, including myocardial infarction and coronary artery disease, was compared among groups using simple and time-dependent Cox analysis. Occasional drinkers (<2.5 g/d), nondrinkers, and lifetime abstainers were used as comparison reference groups. RESULTS: Simple Cox analysis indicated that drinker group 1 exhibited a significantly lower risk of myocardial infarction (hazard ratio [HR] 0.44, 95% confidence interval [CI] 0.21, 0.92) and coronary artery disease (HR 0.61, 95% CI 0.4, 0.94) than nondrinkers. Time-dependent analysis based on alcohol consumption change showed that the beneficial effects of drinker group 1 were significant only for myocardial infarction, not for coronary artery disease. The benefits did not change significantly when either nondrinkers or lifetime abstainers were the reference group. However, when occasional drinkers were included in the reference group, the benefits of drinker group 1 were not significant for myocardial infarction (HR 0.65, 95% CI 0.29, 1.45) or coronary artery disease (HR 0.71, 95% CI 0.42, 1.19). Occasional drinkers and drinkers had more similar sociodemographic characteristics than did nondrinkers and drinkers. CONCLUSIONS: Further studies on alcohol consumption and its effects on health must use repeated measurement to define drinking status, as simple and time-dependent analyses can show different alcohol consumption risks. These findings do not indicate a beneficial effect of drinking <15 g/d when occasional drinkers and nondrinkers are included in the reference group.