Is the effect of wine on health confounded by diet?

Abstract

To the Editor: A light to moderate intake of wine may have both cardioprotective and cancer inhibitory effects. 1 We have found that those who had a daily intake of 3–5 glasses of wine had a 50% lower risk of dying from all causes compared to non-drinkers of wine, while beer, and spirits drinking had a much smaller beneficial effect, if any. 1 However, diet may confound the relation. A high intake of fruit, vegetables, and fish and a low intake of saturated fat may reduce the risk of coronary heart disease morbidity and mortality. 2 There may also be protective effects of fruit and vegetables intake on risk of cancer. 3 Preference of wine compared to preference of beer or spirits may be associated with a more frequent intake of fruit, fish, vegetables, salads and olive oil. 4 Others have suggested that wine is associated with a presumably healthy diet, but none have taken dietary factors sufficiently into account in prospective studies of the health effects of wine. We have quantitatively assessed in a theoretical sensitivity analysis 5 whether diet is a plausible confounder of the relation between wine intake and mortality by applying the method to previously reported data from the Copenhagen City Heart Study. 1 In the present analysis, the unadjusted odds ratio for the 50 deaths among the 257 exposed (ie, those who had a daily intake of wine) and the 780 deaths among the 2,553 unexposed (non-wine drinkers) was estimated as 0.6. 1 The odds ratios for the relation between wine intake and mortality, adjusted for a hypothetical confounder, have been calculated for various scenarios (Table). It appears that even a very strong confounder (odds ratio = 0.3 or 0.1) would have to exhibit a very uneven distribution among wine drinkers and non-wine drinkers to fully explain the findings noted above. Table 1: Odds Ratios for Death among Wine Drinkers (1–5 Glasses of Wine per Day) Compared with Non-Wine Drinkers, Adjusted for Different Distributions of Confounding Factors of Three Different Strengths, Copenhagen City Heart StudyAlthough both errors in dietary assessment and intraindividual variability of the diet can lead to underestimation of dietary differences and effects, it seems very unlikely that the previously reported lower mortality among wine drinkers compared to non-wine drinkers can be explained fully by dietary factors. The finding that wine drinkers more frequently than beer and spirits drinkers may have a “healthier” diet 4 should, however, be taken into account in interpretations of the prospective studies of effects of both type of alcohol and diet. Until now, there is no clear evidence that there is a large protective effect of the healthy diet on morbidity or mortality. Further, the studies suggesting any such effect have not controlled their results for intake of alcohol or type of alcohol. The simple analysis presented here only takes a single confounder into account. It is evident that a conglomerate of several confounders, as well as possible effect modifiers (such as several dietary factors combined with, for example, a high level of physical activity), may play a role. Morten Grønbæk Thorkild I.A. Sørensen