Russell Marcia, Chu Bong Chul, Banerjee Aniruddha, Fan Amy Z, Trevisan Maurizio, Dorn Joan M, Gruenewald Paul
Prevention Research Center, Pacific Institute for Research, Evaluation, 1995 University Avenue, Berkeley, CA 94704 USA.
Alcohol Clin Exp Res. 2009 Feb;33(2):324-31. doi: 10.1111/j.1530-0277.2008.00836.x. Epub 2008 Nov 19.
The relation of alcohol intake to cardiovascular health is complex, involving both protective and harmful effects, depending on the amount and pattern of consumption. Interpretation of data available on the nature of these relations is limited by lack of well-specified, mathematical models relating drinking patterns to alcohol-related consequences. Here we present such a model and apply it to data on myocardial infarction (MI).
The dose-response model derived assumes: (1) each instance of alcohol use has an effect that either increases or decreases the likelihood of an alcohol-related consequence, and (2) greater quantities of alcohol consumed on any drinking day add linearly to these increases or decreases in risk. Risk was reduced algebraically to a function of drinking frequency and dosage (volume minus frequency, a measure of the extent to which drinkers have more than 1 drink on days when they drink). In addition to estimating the joint impact of frequency and dosage, the model provides a method for calculating the point at which risk related to alcohol consumption is equal to background risk from other causes. A bootstrapped logistic regression based on the dose-response model was conducted using data from a case-control study to obtain the predicted probability of MI associated with current drinking patterns, controlling for covariates.
MI risk decreased with increasing frequency of drinking, but increased as drinking dosage increased. Rates of increasing MI risk associated with drinking dosage were twice as high among women as they were among men. Relative to controls, lower MI risk was associated with consuming < 4.55 drinks per drinking day for men (95% CI: 2.77 to 7.18) and < 3.08 drinks per drinking day for women (95% CI: 1.35 to 5.16), increasing after these cross-over points were exceeded.
Use of a well-specified mathematical dose-response model provided precise estimates for the first time of how drinking frequency and dosage each contribute linearly to the overall impact of a given drinking pattern on MI risk in men and women.
酒精摄入与心血管健康的关系较为复杂，其具有保护和有害两种作用，这取决于饮酒量和饮酒模式。由于缺乏将饮酒模式与酒精相关后果联系起来的明确数学模型，关于这些关系本质的现有数据解读受到限制。在此，我们提出这样一个模型并将其应用于心肌梗死（MI）数据。
所推导的剂量反应模型假定：（1）每次饮酒行为都有一个增加或降低酒精相关后果发生可能性的效应；（2）在任何饮酒日摄入更多的酒精会使这些风险的增加或降低呈线性累加。风险通过代数方法简化为饮酒频率和饮酒量（饮酒量减去饮酒频率，这是衡量饮酒者在饮酒日饮酒超过1杯的程度的指标）的函数。除了估计频率和饮酒量的联合影响外，该模型还提供了一种计算与酒精消费相关的风险等于其他原因导致的背景风险的临界点的方法。基于该剂量反应模型，利用一项病例对照研究的数据进行了自抽样逻辑回归，以获得与当前饮酒模式相关的心肌梗死预测概率，并对协变量进行控制。
心肌梗死风险随饮酒频率增加而降低，但随饮酒量增加而升高。饮酒量增加导致心肌梗死风险增加的速率在女性中是男性的两倍。相对于对照组，男性每日饮酒量<4.55杯（95%可信区间：2.77至7.18）和女性每日饮酒量<3.08杯（95%可信区间：1.35至5.16）与较低的心肌梗死风险相关，超过这些交叉点后风险增加。
使用一个明确的数学剂量反应模型首次精确估计了饮酒频率和饮酒量如何分别线性地对特定饮酒模式对男性和女性心肌梗死风险的总体影响产生作用。
