Identification of Dietary Patterns Using Two Statistical Approaches and Their Association With Breast Cancer Risk: A Case-Control Study in Southern France

详细信息	查看全文 | 推荐本文 |

• 作者：FaÏ ; za Bessaoud ; PhD^a ; ^{faiza.bessaoud@orange.fr} ; Brigitte Tretarre ; MD^a ; Jean-Pierre Daurè ; s ; MD ; PhD^a ; Mariette Gerber ; MD ; PhD^b
• 关键词：Dietary Patterns ; Breast Cancer ; Case&ndash ; Control Study ; Principal Component Analysis ; Variable Clustering
• 刊名：Annals of Epidemiology
• 出版年：2012
• 期刊代码：15_10472797
• 类别：med
• 出版时间：July, 2012
• 卷：22
• 期：7
• 页码：499-510
• 文件大小：288 K

摘要

| Figures/TablesFigures/Tables | ReferencesReferences

Purpose

The aim of this case-control study was to compare two different statistical methods in the identification of dietary patterns by use of principal component analysis (PCA) and variable clustering (VC) and to examine their association with the risk of breast cancer (BC).

Methods

A dose-response association was then performed by the use of an adaptation of free knot spline function in logistic models.

Results

A 鈥淲estern鈥?pattern was revealed by PCA and VC and was then shown to be associated with a nonsignificant increase of BC risk. Only PCA identified a 鈥渕eat/alcohol鈥?pattern. Above the spline threshold, BC risk increased significantly (OR 鈮?s vs. < s = 2.56, 95%CI 1.54-4.27). When we used PCA, a 鈥淢editerranean鈥?pattern was identified, but no association between BC risk and this pattern was shown. VC split the 鈥淢editerranean鈥?dietary pattern in two: 鈥渞aw vegetables and olive oil鈥?and 鈥渇ish.鈥?Above the spline threshold, the 鈥渇ish鈥?pattern tended to protect against BC risk (OR 鈮?s vs. < s = 0.77, 95%CI 0.58-1.01), whereas an excess of raw vegetables and olive oil increased BC risk (OR 1 se聽= 1.22, 95%CI = 1.06-1.32).

Conclusions

Some results from the PCA and the VC methods were similar, whereas others were different but gave complementary results.