Dawson D A
Division of Biometry and Epidemiology, National Institute on Alcohol Abuse and Alcoholism, National Institutes of Health, Bethesda, Maryland 20892-7003, USA.
Alcohol Clin Exp Res. 2000 Dec;24(12):1820-9.
This study compared four sets of US low-risk drinking guidelines (two interpretations of the US Dietary Guidelines and two variations of the NIAAA physicians' guidelines) in terms of adherence and how well they predicted five different alcohol-related outcomes.
Using data from a nationally representative sample of 17,542 US adults 21 years of age and over who drank 12 or more drinks in the past year, this study assessed the sensitivity, specificity, overall accuracy, positive and negative predictive values, and odds ratios of the various drinking guidelines (specifically, of having exceeded them with different degrees of frequency) as predictors of alcohol dependence, impaired driving, liver disease, peptic ulcer, and hypertension.
The proportions of past-year regular drinkers exceeding the four sets of guidelines varied from 20.9%, whose average intake exceeded the weekly limits, to between 21.0% and 42.7% who exceeded the daily guidelines at least once a week, and to between 69.2% and 94.2% who ever exceeded the daily limits in the year preceding the interview. Sensitivity and odds ratios were highest for the ever exceeding the Dietary Guidelines daily limits, intermediate for ever exceeding the two variations based on the NIAAA physicians' guidelines, and lowest for exceeding the Dietary Guidelines interpreted as weekly limits. The opposite pattern was observed for specificity and overall predictive accuracy. When frequently exceeding the daily limits was considered, their sensitivity declined but their specificity and positive predictive value increased.
If sensitivity and specificity are deemed equally important, the NIAAA physicians' guidelines incorporating both daily and weekly limits seem to do the best job of balancing these dimensions in the prediction of a variety of alcohol-related outcomes.
本研究比较了四套美国低风险饮酒指南（美国膳食指南的两种解读以及美国国立酒精滥用与酒精中毒研究所（NIAAA）医生指南的两种变体）在依从性方面以及它们对五种不同酒精相关结果的预测能力。
利用来自全国代表性样本的17542名21岁及以上美国成年人的数据，这些人在过去一年中饮用了12杯或更多饮料，本研究评估了各种饮酒指南（具体而言，不同频率超过这些指南的情况）作为酒精依赖、酒后驾车、肝病、消化性溃疡和高血压预测指标的敏感性、特异性、总体准确性、阳性和阴性预测值以及比值比。
过去一年中超过这四套指南的经常饮酒者比例各不相同，从平均摄入量超过每周限量的20.9%，到每周至少一次超过每日指南的21.0%至42.7%，再到访谈前一年中曾超过每日限量的69.2%至94.2%。对于每日超过膳食指南限量的情况，敏感性和比值比最高；对于超过基于NIAAA医生指南的两种变体的情况，敏感性和比值比处于中等水平；而对于超过解读为每周限量的膳食指南的情况，敏感性和比值比最低。在特异性和总体预测准确性方面则观察到相反的模式。当考虑频繁超过每日限量时，它们的敏感性下降，但特异性和阳性预测值增加。
如果敏感性和特异性被认为同样重要，纳入每日和每周限量的NIAAA医生指南似乎在预测各种酒精相关结果时，在平衡这些方面做得最好。
