Abstract: Background　The global prevalence of diabetes mellitus is escalating, posing a substantial threat to public health. While associations between individual sleep behaviors and diabetes have been well-established, the impact of multidimensional sleep patterns and their underlying mechanisms remains to be elucidated. Objective　To investigate the association between a composite healthy sleep score and the prevalence of diabetes, and to assess the mediating effects of the triglyceride-glucose（TyG） index and its obesity-related derivatives（TyG-BMI and TyG-WC）. Method　This cross-sectional study analyzed data from 36,319 participants who underwent health examinations and completed surveys at the Health Management Center of the 11th People's Hospital of Guangzhou between January 4 and December 30, 2022. A weighted sleep score was constructed based on sleep duration, quality, and timing. Multivariable logistic regression, subgroup analysis, and interaction tests were used to estimate the association between sleep score and diabetes. Structural equation modeling was employed to examine the mediating effects of TyG, TyG-BMI, and TyG-WC. Results　Multivariable logistic regression analysis showed that after adjusting for confounding factors, participants in the low-risk sleep group had a lower risk of diabetes compared to those in the high-risk group（OR=0.88, 95%CI=0.80-0.97）. Trend tests further confirmed a linear decrease in diabetes prevalence as the sleep risk score decreased（Ptrend <0.001）. Subgroup analysis revealed that this protective association was more pronounced in individuals aged ≥ 60 years（OR=0.78, 95%CI=0.66-0.93）, and an interaction was observed between age and sleep patterns （Pinteraction=0.019）. Mediation analysis indicated that the association between sleep score and diabetes was mediated by TyG-WC (indirect effect: -0.013) and TyG-BMI (indirect effect: -0.019). Conclusions　A healthier sleep pattern—comprising normal sleep duration（7-9 h/d）, optimal sleep timing（21：00-23：00）, and good sleep quality—is associated with a reduced prevalence of diabetes. This association is particularly pronounced among individuals aged ≥ 60 years and is partially mediated by insulin resistance and obesity-related metabolic markers.

Key words: Diabetes mellitus, type 2, Sleep, Triglyceride-glucose index, Risk factors, Mediation analysis

摘要： 背景　糖尿病全球发病率持续上升，严重影响公共健康。虽然睡眠时长，睡眠质量等单一的睡眠行为与糖尿病的关联已获证实，但睡眠作为多维度行为模式其整体效应与糖尿病关联及机制尚待阐明。目的　探讨健康睡眠模式与糖尿病的关联，并检验甘油三酯葡萄糖指数（TyG）及其相关指标（TyG-BMI，TyG-WC）在其中的中介作用。方法　本研究为一项横断面研究，纳入了 2022-01-04—2022-12-30 在广州市第十一人民医院健康管理中心体检并完成横断面调查的 36 319 名参与者。研究构建了一项包含睡眠时长、质量和入睡时间的加权睡眠评分。采用多因素Logistic 回归、亚组分析及交互检验评估睡眠评分与糖尿病的关联，并以结构方程模型检验 TyG、TyG-BMI 和 TyG-WC的中介作用。结果　多因素 Logistic 回归分析显示，调整混杂因素后，与高风险睡眠模式组相比，低风险组糖尿病患病风险降低（OR=0.88，95%CI=0.80~0.97，P=0.009）。趋势性检验进一步证实，随着睡眠风险降低，糖尿病患病风险呈线性下降趋势（P趋势<0.001）。亚组分析显示，在年龄≥ 60 岁人群中，低风险睡眠模式的糖尿病患病风险降低（OR=0.78，95%CI=0.66~0.93），年龄在睡眠模式分组与糖尿病患病风险关系中存在交互作用（P 交互 =0.019）。中介分析显示，TyG-WC（间接效应值 -0.013）和 TyG-BMI（间接效应值 -0.019）在睡眠评分与糖尿病风险之间发挥中介作用。结论　正常睡眠时长（7~9 h）、正常入睡时间（21：00~23：00）及良好睡眠质量的更健康睡眠模式与糖尿病风险降低相关，此关联在≥ 60 岁人群中尤为显著，并部分由胰岛素抵抗及肥胖相关代谢指标所介导。

关键词: 糖尿病, 睡眠, 甘油三酯 - 葡萄糖指数, 危险因素, 中介分析