基于降秩回归的血脂相关膳食模式与糖尿病的关系

目的  探寻血脂相关膳食模式与糖尿病的关联，为糖尿病的膳食干预提供科学依据。方法  选取4 759名来自伊犁州霍城县的农村居民作为研究对象，通过问卷调查、体格检查及实验室检测收集研究对象的相关资料。通过降秩回归法提取血脂相关膳食模式；利用Pearson相关分析评估血脂相关膳食模式评分及各食物组摄入量与血脂的相关性；并通过logistic回归分析模型探寻血脂相关膳食模式与糖尿病患病风险之间的关联。结果  血脂相关膳食模式主要以小麦、畜肉和油炸面食摄入较多，而杂粮、蔬菜和水果等摄入较少为特征。血脂相关膳食模式与TC (r=0.296, P < 0.001)、LDL-C(r=0.225, P < 0.001)及TG(r=0.332, P < 0.001)呈正相关，而与HDL-C(r=-0.237, P < 0.001)呈负相关。logistic回归分析模型分析结果显示，调整混杂因素后，血脂相关膳食模式评分Q4组比Q1组的糖尿病患病风险增加99%(OR=1.99; 95% CI: 1.47~2.68; P < 0.001)，且随着膳食模式评分增高，糖尿病患病风险具有升高的线性趋势(P_趋势 < 0.001)。结论  血脂相关膳食模式评分与糖尿病患病风险呈正相关，通过膳食来优化血脂水平可能有助于糖尿病的早期预防。

Association between blood lipids-related dietary patterns derived by reduced rank regression and diabetes

  Objective  To explore the association between lipid-related dietary patterns and diabetes, thus providing a scientific basis for dietary intervention in diabetes.  Methods  A total of 4 759 rural residents from Huocheng County in Yili were selected as the study subjects, and relevant data were collected through questionnaires, physical examinations and laboratory tests. Lipid-related dietary patterns were extracted by reduced-rank regression. The correlations between lipid-related dietary pattern scores and intake of each food group and lipid were assessed by Pearson correlation analysis. The relationship between lipid-related dietary pattern and diabetes risk was analyzed by logistic regression model.  Results  Lipid-related dietary pattern was characterized by higher intakes of wheat, meat and fried pasta, and lower intake of mixed grains, vegetables and fruits. Lipid-related dietary pattern was positively correlated with total cholesterol (r=0.296, P < 0.001), low density lipoprotein cholesterol (r=0.225, P < 0.001) and triglyceride (r=0.332, P < 0.001), and negatively correlated with high density lipoprotein cholesterol (r=-0.237, P < 0.001). Logistic regression analysis showed that after adjusting for confounding factors, the risk of diabetes increased by 99% (OR=1.99; 95% CI: 1.47-2.68; P < 0.001) in the lipid-related dietary pattern score Q4 group compared with Q1 group. And there was a linear trend of significantly higher risk of diabetes with increasing dietary pattern scores (P_trend < 0.001).  Conclusions  Lipid-related dietary pattern scores are positively associated with diabetes risk, and optimizing lipid levels through diet may contribute to early prevention of diabetes.