Alcohol intake and high density lipoprotein-cholesterol: comparison of food frequency questionnaire and diet record in a general population sample.

Abstract

BACKGROUND Moderate alcohol consumption is known to affect several cardiovascular risk factors, including high density lipoprotein-cholesterol (HDL-C). Accurate measurement of alcohol intake in epidemiological studies is crucial for valid estimation of alcohol-related effects. METHODS We compared two widely used alcohol-intake assessment methods, a food frequency questionnaire (FFQ) and a 7-day diet record (DR) regarding the association between alcohol intake and HDL-C in the general population. RESULTS In a representative sample (842 men, 1103 women) of the population of former West Germany (VERA study), 75.6% of the participants reported drinking alcoholic beverages in the FFQ, whereas the percentage was 84.9% according to the DR. The median daily alcohol intake of consumers was 10.3 and 12.1 g, as assessed by the FFQ and the DR, respectively. There was moderate agreement between reported consumption levels ranked by quintiles according to FFQ and DR (kappa = 0.49). With both approaches alcohol intake showed a strong positive association with HDL-C in linear regression models. The estimated increases in HDL-C (mg dL(-1)) per gram alcohol per day from the FFQ and DR were 0.15 [95% confidence interval (CI): 0.12-0.18] and 0.14 (95% CI: 0.11-0.17), respectively. When participants with different classification of the amount of alcohol consumed by FFQ and DR were excluded from the analysis, the association between alcohol intake and HDL-C was more pronounced. DISCUSSION Our analyses suggest that both FFQ and DR lead to similar results regarding the alcohol-HDL-C association in the general population, despite major differences in the recorded prevalence of drinking and amount of alcohol consumed. Both methods may tend to under-estimate the true alcohol-HDL-C association.