The Framingham results on alcohol and breast cancer.
The Framingham results on alcohol and breast cancer.
复制标题
弗雷明汉关于酒精和乳腺癌的结果。
DOI：
10.1093/oxfordjournals.aje.a009773
复制
发表时间：
1999
影响因子：
5
通讯作者：
Longnecker,MP
中科院分区：
文献类型：
--
作者：
Longnecker,MP
In a study presented in this issue of the Journal, Zhang et al.(1) suggest that the effect of light alcoholic beverage consumption on the risk of breast cancer remains controversial. In 1994, I showed that the risk associated with, on average, one alcoholic drink daily (-13 g of ethanol) was about 10 percent greater than that in nondrinkers in a combined analysis of the results of 38 studies (2). A relation of this size is likely to remain controversial because, in this instance, it is beyond the resolution of epidemiologic methods, as the possibility of confounding or other bias seems difficult to exclude. Furthermore, with a true relative risk of that size, by chance alone some studies will appear unsupportive of an association and others supportive of a stronger association. The risk of breast cancer increases with alcoholic beverage consumption of up to 40 g of ethanol per day in a monotonic doseresponse fashion consistently in data from multiple studies (2, 3) or in data from a single, especially large study (4), when analyzed by different methods (categorical (2—4), nonparametric (3), and quadratic (2, 3)). Zhang et al. in this issue of the Journal (1) found that the multivariate-adjusted relative risk of breast cancer associated with consumption of, on average, 5—< 15 g of ethanol per day was 0.7 (95 percent confidence interval 0.5-1.1) compared with that of nondrinkers. As with the earlier combined analysis, a recent pooled analysis of data from six prospective studies on alcohol and breast cancer gave rather different results (3). In the pooled analysis, among women consuming 5—< 15 g of ethanol per day, the multivariate-adjusted relative risk was 1.06 (95 percent confidence interval 0.96-1.17). The pooled analysis (3) was based on observations for 4,335 cases, compared with 287 cases in the analysis by Zhang et al. A test of the difference between the relative risks for the 5-< 15 g of
