|
|
|
 |
|
|
Kimball A W et al Nonlinear modeling of alcohol consumption for analysis of beveragetypeeffects and beverage preference effects. AM J EPIDEMIOL 1992;135:2287-92. "In this analysis, wine emerged as the type of beverage least likely to be related to increased blood pressure." Complex statistical models were developed to study nonlinear health outcomes related to types of alcohol consumption in order to differentiate between effects of beverage preference versus the actual beverage type effect. The data originated from the 1982 Maryland Hypertension Survey, a cross-sectional population-based collection of home-measured blood pressures taken by trained technicians. An alcohol consumption questionnaire defined drinks per day as the number of twelve ounce beers, four to five ounce glasses of wine, and 1.5 ounce servings of liquor. Final analysis of 4,458 men and women, aged eighteen years or more, was restricted to those who consumed three drinks/day or less. Comparison is between the three drinking groups and not against abstainers since beverage preference cannot be measured among abstainers. Covariables found to be significant at p < 0.01 were added into the models in the following order: age, body mass index, sex, hypertensive medication, age × sex, marital status, smoking and exercise. There was no systolic blood pressure difference between wine, beer and spirits for beverage preference. However, there was a beverage type difference between wine and the other two beverages. "The beer and spirits curves show an increase of about 5.5 mm Hg at three drinks/day, whereas the wine curve decreases about 1.5 mm Hg and stabilizes at that level of consumption." The p-values describing the differences between equations for drinkers of only one type of alcoholic beverage are as follows: wine vs spirits p = 0.04; wine vs beer p = 0.07. The authors conclude with a plea to differentiate wine, beer and spirits in future studies: "Because the models proposed in this study and in the previous report permit the simultaneous evaluation of both beverage type effects and beverage preference effects, and because it has been demonstrated that the models can detect such effects, it seems appropriate that these effects should be examined in all studies relating health outcomes to alcohol consumption. Studies which do not account for these effects when they exist may produce biased estimates of the relations between alcohol consumption and health outcomes."
Nevill A M et al Modelling the associations of BMI, physical activity and dietwith arterial blood pressure: some results from the Allied Dunbar National FitnessSurvey. ANN HUMAN BIOL 1997;24:229-47. Wine drinkers had lower mean systolic blood pressure. The authors believe the Allied Dunbar National Fitness Survey "is the most comprehensive study into the activity and fitness levels of the English population ever undertaken." Almost three thousand adult subjects were randomly chosen and extensively interviewed about their health, lifestyle and physical activity and then given a physical examination including measurement of blood pressure. Because the blood pressures were found to be "non-normally" distributed, complicated multiplicative models were necessary to adjust for differences in known associations. They confirmed such associations as body mass index, age, sex, vigorous physical activity, and more consumption of fresh fruit, rice or pasta. The log-linear regression model, having adjusted for bias, resulted in the following conclusions regarding alcohol consumption:
The greatest blood pressure lowering effect was seen at approximately 17 units of wine per week. Thereafter, a further increase in wine consumption (greater than 17 units per week) was associated with a rise in systolic blood pressure.
|
|
|
|
[Home] [About] [Health/Wine] [Calendar] [Membership] [Links] [Photo Gallery] [Newsletter] |
|
|
|
|
|
copyright 2004/Disclaimer |
|||
|
Site created and maintained by Ed Bierman, MLS |
|||
