Elsevier

Ageing Research Reviews  衰老研究评论

Volume 106, April 2025, 102696
第 106 卷，2025 年 4 月，102696
Ageing Research Reviews

Review article  综述
Association of frailty and pre-frailty with all-cause and cardiovascular mortality in diabetes: Three prospective cohorts and a meta-analysis
糖尿病中衰弱和预衰弱与全因及心血管死亡率的关系：三项前瞻性队列研究和荟萃分析

https://doi.org/10.1016/j.arr.2025.102696Get rights and content  获取权利和内容
Full text access  全文访问

Highlights  要点

  • •
    The prevalence of frailty in diabetes (aged 40–90) was estimated to be 10–12 %.
    糖尿病（40-90 岁）中的衰弱患病率估计为 10-12%。
  • •
    Frailty was associated with a 1.8-fold risk of all-cause mortality in diabetes.
    衰弱与糖尿病全因死亡率风险增加 1.8 倍相关。
  • •
    Frailty was associated with a 2.0-fold higher risk of CVD mortality in diabetes.
    虚弱与糖尿病中心血管疾病死亡风险增加 2.0 倍相关。
  • •
    Of five domains, gait speed showed the strongest predictive value for mortality.
    在五个领域里，步行速度显示出对死亡率的最高预测价值。
  • •
    Integrating frailty assessment into routine diabetes care is recommended.
    建议将虚弱评估纳入常规糖尿病护理。

Abstract  摘要

Objective  目标

To investigate the association of frailty status with all-cause and cardiovascular disease (CVD) mortality in individuals with diabetes.
研究虚弱状态与糖尿病患者的全因死亡率和心血管疾病（CVD）死亡率之间的关联。

Methods  方法

Data was sourced from the Third National Health and Nutrition Examination Survey (NHANES III, 1988–1994), NHANES (1999–2006), and the UK Biobank. Frailty status was assessed using the Fried phenotype and classified as non-frailty, pre-frailty, and frailty. We further performed a meta-analysis involving 19 prospective cohort studies (753,480 patients) to summarize the existing evidence.
数据来源于第三次国家健康与营养调查（NHANES III，1988-1994 年）、NHANES（1999-2006 年）和英国生物样本库。使用 Fried 表型评估虚弱状态，并将其分为非虚弱、预虚弱和虚弱。我们进一步进行了荟萃分析，涉及 19 项前瞻性队列研究（753,480 名患者），以总结现有证据。

Results  结果

We included 31,225 diabetes patients from NHANES III (mean age 63.3 ± 0.8, 56.4 % female), NHANES 1999–2006 (mean age 61.6 ± 0.4, 49.7 % female), and the UK Biobank (mean age 59.6 ± 7.2, 39.5 % female). The prevalence of frailty was 9.9 %, 10.7 %, and 12.1 % across respective cohorts. During a follow-up period exceeding 13 years, we observed consistent results that frailty and pre-frailty were significantly associated with increased risks of all-cause and CVD mortality in diabetes. Notably, of the five domains used to assess frailty phenotypes, low gait speed showed the strongest association with all-cause and CVD mortality risks. Meta-analysis showed that, compared to non-frailty, frailty in patients with diabetes was associated with a 1.8-fold higher risk of all-cause mortality and a 2.0-fold higher risk of CVD mortality. Similarly, pre-frailty was associated with a 1.3-fold higher risk of all-cause mortality and a 1.4-fold higher risk of CVD mortality.
我们纳入了来自 NHANES III（平均年龄 63.3 ± 0.8 岁，女性占 56.4%）、NHANES 1999–2006（平均年龄 61.6 ± 0.4 岁，女性占 49.7%）和英国生物样本库（平均年龄 59.6 ± 7.2 岁，女性占 39.5%）的 31,225 名糖尿病 患者。在各个队列中，衰弱的患病率分别为 9.9%、10.7%和 12.1%。在超过 13 年的随访期间，我们观察到一致的结果：衰弱和衰弱前期与糖尿病中全因死亡率和心血管疾病（CVD）死亡率风险增加显著相关。值得注意的是，在用于评估衰弱表型的五个领域 中，步速缓慢与全因死亡率和心血管疾病死亡率风险的相关性最强。荟萃分析表明，与非衰弱患者相比，糖尿病患者的衰弱与全因死亡率风险增加 1.8 倍，CVD 死亡率风险增加 2.0 倍。类似地，衰弱前期与全因死亡率风险增加 1.3 倍，CVD 死亡率风险增加 1.4 倍。

Conclusions  结论

This study established a strong association between frailty, pre-frailty, and increased all-cause and CVD-related mortality in diabetes. Integrating frailty assessment into routine practice to identify frail and pre-frail status early on is recommended, followed by the implementation of targeted healthy lifestyle interventions to mitigate adverse outcomes.
本研究建立了衰弱、衰弱前期与糖尿病中全因死亡率和心血管疾病相关死亡率增加之间的强关联。建议将衰弱评估纳入常规实践，以早期识别衰弱和衰弱前期状态，随后实施有针对性的健康生活方式干预，以减轻不良后果。

Keywords  关键词

Frailty
Mortality
Cardiovascular mortality
Type 2 diabetes
Cohorts
Meta-analysis

衰弱死亡率心血管死亡率 2 型糖尿病队列荟萃分析

1. Introduction  1. 引言

Diabetes has become a pressing public health concern worldwide. According to data from the International Diabetes Federation, the projected number of adults afflicted with diabetes is set to surge to 783 million by 2045 (International Diabetes Federation, 2021). Patients with diabetes persist to face elevated risks of premature death, particularly due to cardiovascular causes (Wong and Sattar, 2023). Consequently, there is an imperative need to investigate the risk factors associated with the outcomes, aimed at crafting effective prevention strategies and extending the healthy life expectancy of individuals affected by diabetes.
糖尿病已成为全球紧迫的公共卫生问题。根据国际糖尿病联合会数据，预计到 2045 年，受糖尿病困扰的成年人数量将激增至 7.83 亿（国际糖尿病联合会，2021 年）。糖尿病患者持续面临过早死亡的风险，尤其是心血管原因（Wong 和 Sattar，2023 年）。因此，迫切需要研究与结果相关的风险因素，旨在制定有效的预防策略，延长受糖尿病影响人群的健康预期寿命。
Frailty represents a prevalent geriatric syndrome characterized by functional declines across multiple physiological systems and elevated vulnerability to stressors (Fried et al., 2001). Patients with diabetes are prone to developing frailty, with a prevalence of frailty as high as 20.1 % (Hanlon et al., 2020, Kong et al., 2021). Growing evidence shows that the coexistence of frailty and diabetes is related to an elevated risk of adverse outcomes, encompassing all-cause and cardiovascular disease (CVD) mortality (Castro-Rodríguez et al., 2016, Espeland et al., 2022, He et al., 2022, Kitamura et al., 2019, Weng et al., 2023). However, upon conducting a systematic review of the literature, we found that existing studies predominantly concentrate on the association between frailty and all-cause mortality in diabetes, yet there is a dearth of research pertaining to CVD mortality. Pre-frailty, signifying an intermediate and reversible transitional state between non-frailty and frailty, presents a potentially critical window for intervention to enhance the health outcomes (Mone et al., 2023, O’Caoimh et al., 2018, Rodríguez-Laso et al., 2022, Teh et al., 2021), while evidence regarding pre-frailty and mortality in diabetes is conflicting and insufficient. He et al. (2022) did not observe a significant association between pre-frailty and mortality in diabetes, while Chao et al. (2018) and O’Donovan et al. (2019) observed a significant positive association. Notably absent are high-quality meta-analyses concerning frailty and pre-frailty and mortality in diabetes, and only one pertinent meta-analysis exists, with a literature search conducted in 2018, and merely 4 included studies offering hazard ratios (Ida et al., 2019).
衰弱是一种常见的老年综合征，其特征是多个生理系统的功能下降以及对外界压力的易感性增加（Fried 等人，2001 年）。糖尿病患者易患衰弱，衰弱患病率高达 20.1%（Hanlon 等人，2020 年；Kong 等人，2021 年）。越来越多的证据表明，衰弱与糖尿病共存与不良结局风险升高相关，包括全因死亡率和心血管疾病（CVD）死亡率（Castro-Rodríguez 等人，2016 年；Espeland 等人，2022 年；He 等人，2022 年；Kitamura 等人，2019 年；Weng 等人，2023 年）。然而，在文献系统综述过程中，我们发现现有研究主要集中于衰弱与糖尿病全因死亡率之间的关联，而关于心血管疾病死亡率的研究却十分匮乏。 前期衰弱，表示非衰弱与衰弱之间的一个中间且可逆的过渡状态，为干预以改善健康结果提供了一个潜在的临界窗口（Mone 等，2023；O’Caoimh 等，2018；Rodríguez-Laso 等，2022；Teh 等，2021），然而关于糖尿病前期衰弱与死亡率的证据存在矛盾且不足。He 等（2022）未观察到糖尿病前期衰弱与死亡率之间存在显著关联，而 Chao 等（2018）和 O’Donovan 等（2019）观察到存在显著的正相关。值得注意的是，关于糖尿病衰弱与前期衰弱与死亡率的高质量荟萃分析目前缺失，仅存在一项相关荟萃分析，该文献检索于 2018 年，且仅纳入了 4 项研究提供风险比（Ida 等，2019）。
To fill these knowledge gaps, we sought to comprehensively investigate the associations of frailty and pre-frailty with all-cause and CVD mortality among middle-aged and older individuals with diabetes in the Third National Health and Nutrition Examination Survey (NHANES III, 1988–1996), NHANES (1999–2006), and the UK Biobank. We also performed a meta-analysis to systematically summarize and expound upon the existing evidence in this regard.
为填补这些知识空白，我们旨在全面调查在糖尿病的中老年人群中，衰弱和衰弱前期与全因死亡率和心血管疾病死亡率之间的关联，研究数据来源于第三次国家健康与营养调查（NHANES III，1988-1996）、NHANES（1999-2006）和英国生物样本库。我们还进行了荟萃分析，以系统性地总结和阐述现有相关证据。

2. Research design and methods
2. 研究设计与方法

2.1. Study population  2.1. 研究人群

NHANES is a large-scale, stratified multistage, and nationally representative program conducted in the US. The UK Biobank is a large-scale cohort study of over 500,000 UK civilians. These projects both provide comprehensive demographic, socioeconomic, lifestyle, physical examination, and laboratory data and have obtained official approval (Centers for Disease Control and Prevention, 2024, Sudlow et al., 2015). NHANES was approved by the NCHS Ethics Review Board (Protocols 98–12, 2005–06, 2011–17, and 2018–01). The UK Biobank resource was approved by the North West Multi-centre Research Ethics Committee under reference number 21/NW/0157. All participants have signed informed consent forms.
NHANES 是一项在美国进行的全国性、分层多阶段代表性项目。英国生物样本库是一项包含超过 50 万名英国公民的大规模队列研究。这两个项目均提供全面的流行病学、社会经济、生活方式、体格检查和实验室数据，并已获得官方批准（美国疾病控制与预防中心，2024；Sudlow 等，2015）。NHANES 已获得 NCHS 伦理审查委员会批准（方案 98–12、2005–06、2011–17 和 2018–01）。英国生物样本库资源已获得西北多中心研究伦理委员会批准，编号为 21/NW/0157。所有参与者均签署了知情同意书。
In this study, we utilized the data from NHANES III (1988–1994), NHANES 1999–2006, and the UK Biobank, which provided information on frailty phenotype assessment. The participants with diabetes at baseline were defined as individuals who had been diagnosed with diabetes through self-reported doctor diagnosis, use of insulin or oral hypoglycemic medication, fasting blood glucose ≥ 7.0 mmol/L, postprandial plasma glucose ≥ 11.1 mmol/L, or glycated hemoglobin A1c (HbA1c) levels ≥ 6.5 %. The flow chart is presented in Supplementary Figure 1.
在本研究中，我们使用了来自 NHANES III（1988-1994 年）、NHANES 1999-2006 年和英国生物样本库的数据，这些数据提供了关于衰弱表型评估的信息。基线时患有糖尿病的参与者被定义为通过自我报告的医生诊断、使用胰岛素或口服降糖药物、空腹血糖≥7.0 mmol/L、餐后血浆葡萄糖≥11.1 mmol/L 或糖化血红蛋白 A1c（HbA1c）水平≥6.5%的个体。流程图展示在补充图 1 中。

2.2. Physical frailty phenotype assessment
2.2. 生理衰弱表型评估

We employed the five criteria proposed by Fried et al. (2001) to assess the frailty phenotype, namely, weight loss, exhaustion, weakness, slow gait speed, and low physical activity. These criteria were appropriately adapted based on previous studies utilizing NHANES (Crow et al., 2018, Wilhelm-Leen et al., 2009) and UK Biobank data (Hanlon et al., 2018, Petermann-Rocha et al., 2020). In NHANES, we included participants with four or more frailty domains, as previous research has demonstrated the applicability of this approach (Blodgett et al., 2015, Crow et al., 2018, Wilhelm-Leen et al., 2009). The participants were classified as non-frailty (0 scores), pre-frailty (1–2 scores), or frailty (≥3 scores) pursuant to the cutoffs described by Fried and colleagues (Fried et al., 2001). Detailed definitions are shown in Supplementary Tables 1–3.
我们采用了 Fried 等人（2001 年）提出的五个标准来评估衰弱表型，即体重减轻、疲劳、虚弱、步速缓慢和身体活动减少。这些标准根据先前使用 NHANES（Crow 等人，2018 年，Wilhelm-Leen 等人，2009 年）和英国生物样本库数据（Hanlon 等人，2018 年，Petermann-Rocha 等人，2020 年）的研究进行了适当调整。在 NHANES 中，我们纳入了衰弱领域达到四个或更多的参与者，因为先前研究表明这种方法具有适用性（Blodgett 等人，2015 年，Crow 等人，2018 年，Wilhelm-Leen 等人，2009 年）。根据 Fried 及其同事描述的截断值（Fried 等人，2001 年），参与者被分为非衰弱（0 分）、衰弱前期（1-2 分）或衰弱（≥3 分）。详细定义见补充表 1-3。

2.3. Ascertainment of mortality
2.3. 死亡率确定

In NHANES, mortality was ascertained by linking the data to National Death Index records until December 31, 2019. In the UK Biobank, mortality was obtained from the National Health Service (NHS) Information Centre in England and Wales and the NHS Central Register in Scotland until November 30, 2022. The International Classification of Diseases (ICD)-9 and ICD-10 were used to determine the causes of death and CVD mortality was defined as death due to heart diseases and cerebrovascular disease (Supplementary Methods). Follow-up duration for each participant was calculated as the difference between the baseline examination date and the time of death or censored date.
在 NHANES 中，通过将数据与国家死亡指数记录链接，直至 2019 年 12 月 31 日，确定死亡率。在英国生物样本库中，从英格兰和威尔士的国民医疗服务体系（NHS）信息中心和苏格兰的 NHS 中央登记处获取死亡率数据，直至 2022 年 11 月 30 日。使用国际疾病分类（ICD）-9 和 ICD-10 来确定死亡原因，心血管疾病（CVD）死亡率被定义为因心脏病和中风导致的死亡（补充方法）。每位参与者的随访持续时间计算为基线检查日期与死亡时间或删失日期之间的差值。

2.4. Covariates  2.4. 协变量

Sociodemographic characteristics included age, sex, race/ethnicity, education, household income, and Townsend deprivation index. The Townsend deprivation index, a comprehensive and more representative indicator of socioeconomic status, equated high scores with greater socioeconomic deprivation. Behavioral factors and physical examination indicators encompassed smoking status, alcohol consumption, healthy diet score or healthy eating index (HEI), and body mass index (BMI), which refer to Supplementary Methods and Supplementary Table 4 for details. Previous studies have demonstrated associations of long-term morbidities with frailty and life span (Hanlon et al., 2018), and have indicated that patients with diabetes are prone to developing multiple comorbidities (Hanlon et al., 2020). The construction of long-term morbidities was initially developed for a large cross-sectional study in Scotland and subsequently adapted for use with the UK Biobank (Barnett et al., 2012, Hanlon et al., 2018), as we previously described (Chen et al., 2023). We also constructed the long-term morbidities score using available data in NHANES. Detailed definitions are provided in Supplementary Tables 5–7. We further considered diabetes-related factors, including diabetes duration, diabetes medication use, and HbA1c levels. Detailed variables included in each cohort are presented in Supplementary Table 8.
社会人口学特征包括年龄、性别、种族/民族、教育程度、家庭收入和 Townsend 剥夺指数。Townsend 剥夺指数是一个综合且更具代表性的社会经济地位指标，高分表示更严重的社会经济剥夺。行为因素和体格检查指标涵盖吸烟状况、饮酒量、健康饮食评分或健康饮食指数（HEI）以及体重指数（BMI），具体详情请参见补充方法及补充表 4。既往研究表明，长期疾病与衰弱和寿命存在关联（Hanlon 等，2018 年），并指出糖尿病患者易患多种合并症（Hanlon 等，2020 年）。长期疾病的构建最初是为苏格兰的一项大型横断面研究而开发的，随后被改编用于英国生物样本库（Barnett 等，2012 年；Hanlon 等，2018 年），如我们先前所述（Chen 等，2023 年）。我们还利用 NHANES 中的可用数据构建了长期疾病评分，详细定义见补充表 5-7。 我们进一步考虑了与糖尿病相关的因素，包括糖尿病病程、糖尿病药物使用和 HbA1c 水平。每个队列中包含的详细变量在补充表 8 中呈现。

2.5. Statistical analysis
2.5. 统计分析

Multivariable Cox proportional hazard regression models were used to evaluate hazard ratios (HRs) and 95 % confidence intervals (CIs). In NHANES, all statistical analyses accounted for sample weights, strata, and primary sampling units to reflect the nationally representative estimates. Three models were fitted according to the included covariates. Model 1 was adjusted for age (continuous), sex, race, and center (only in the UK Biobank). Model 2 was further adjusted for education level (only in NHANES), family income (only in NHANES), Townsend deprivation index (only in the UK Biobank), smoking status, alcohol consumption, healthy diet score (only in the UK Biobank) or HEI (only in NHANES), and BMI. Model 3 was further adjusted for diabetes-related factors and the number of long-term comorbidities. The missing values of the covariates were treated as dummy variables in the regression models. Stratified analyses were performed to explore differences between different subgroups of age (40–64.9 years or ≥65 years), sex, race (White or non-White), smoking status, alcohol consumption, BMI, diabetes duration (<5 years or ≥5 years), diabetes medication use, HbA1c (<7.0 % or ≥7.0 %), and the number of long-term comorbidities (0–1 or ≥2). In addition, several sensitivity analyses were conducted to examine the robustness of the main results by excluding patients who died within two years of follow-up, developed CVD at baseline, or with only four frailty domains. All analyses were performed using SAS version 9.4 (SAS Institute, USA). A two-sided P value < 0.05 was considered statistically significant.
采用多变量 Cox 比例风险回归模型评估风险比（HRs）和 95%置信区间（CIs）。在 NHANES 中，所有统计分析均考虑了样本权重、分层和初级抽样单位，以反映全国代表性估计。根据所包含的协变量拟合了三种模型。模型 1 调整了年龄（连续）、性别、种族和中心（仅在 UK Biobank 中）。模型 2 进一步调整了教育水平（仅在 NHANES 中）、家庭收入（仅在 NHANES 中）、Townsend 剥夺指数（仅在 UK Biobank 中）、吸烟状况、饮酒量、健康饮食评分（仅在 UK Biobank 中）或 HEI（仅在 NHANES 中），以及 BMI。模型 3 进一步调整了与糖尿病相关的因素和长期合并症的个数。协变量的缺失值在回归模型中作为虚拟变量处理。 进行了分层分析，以探讨不同年龄亚组（40-64.9 岁或≥65 岁）、性别、种族（白人或非白人）、吸烟状况、饮酒量、BMI、糖尿病病程（<5 年或≥5 年）、糖尿病药物使用、HbA1c（<7.0%或≥7.0%）以及长期合并症数量（0-1 或≥2）之间的差异。此外，还进行了多项敏感性分析，以通过排除随访两年内死亡的患者、基线时患有 CVD 或仅存在四个衰弱领域的患者，来检验主要结果的稳健性。所有分析均使用 SAS 版本 9.4（SAS Institute，美国）进行。双侧 P 值<0.05 被认为具有统计学意义。

2.6. Meta-analysis  2.6. 综合分析

We conducted a systematic literature search of PubMed, Web of Science, Embase, and Ovid MEDLINE® ALL, covering records published up to November 15, 2024. The search strategies were as follows: (frailty OR frailties OR frailness OR frailty syndrome OR debility OR debilities OR frail) AND (mortality OR death OR cardiovascular mortality OR adverse outcome) AND (diabetes OR diabetes mellitus OR type 2 diabetes). Details are provided in Supplementary Table 9. Literature screening and data extraction were performed by independent researchers (L.W. and L.C.). We included articles if they met the following criteria: 1) cohort design; 2) targeted patients with diabetes or a stratified analysis was performed among patients with diabetes; 3) baseline frailty status was measured; 4) results were all-cause mortality or CVD mortality; 5) HRs or ORs and 95 % CIs were provided; and 6) the confounders were adjusted. The quality of the identified eligible literature was evaluated using the New Castle-Ottawa Scale. Stata software (version 17.0) was used for the meta-analysis. Heterogeneity across studies was assessed using I2 statistics. A fixed-effect model was used to combine effect sizes when I2 < 50 %; otherwise, a random-effect model was adopted. Meta-regression was performed to determine the source of heterogeneity, and subgroup analysis were performed according to frailty measurements, age, and sex. Registration was completed before conducting the literature search (CRD42023417723).
我们对 PubMed、Web of Science、Embase 和 Ovid MEDLINE® ALL 数据库进行了系统文献检索，检索范围截至 2024 年 11 月 15 日。检索策略如下：(虚弱 OR 虚弱性 OR 虚弱状态 OR 虚弱综合征 OR 衰弱 OR 衰弱性 OR 虚弱) AND (死亡率 OR 死亡 OR 心血管死亡率 OR 不良结局) AND (糖尿病 OR 糖尿病或糖尿病性或 2 型糖尿病)。详细信息见补充表 9。文献筛选和数据提取由独立研究人员（L.W.和 L.C.）进行。如果文章符合以下标准，则被纳入研究：1）队列设计；2）针对糖尿病患者或对糖尿病患者进行分层分析；3）测量基线虚弱状态；4）结果为全因死亡率或 CVD 死亡率；5）提供 HRs 或 ORs 和 95% CI；6）调整混杂因素。使用纽卡斯尔-奥塔瓦量表评估所识别的符合条件的文献质量。使用 Stata 软件（版本 17.0）进行荟萃分析。使用 I 2 统计量评估研究间的异质性。 当 I² < 50% 时，采用固定效应模型合并效应量；否则，采用随机效应模型。通过 Meta 回归分析确定异质性的来源，并根据虚弱程度测量、年龄和性别进行亚组分析。在开展文献检索前已完成注册（CRD42023417723）。

3. Results  3. 结果

3.1. Baseline characteristics
3.1. 基线特征

A total of 31,225 eligible patients with diabetes were included at baseline (Table 1), comprising 2284 from NHANES III (mean age 63.3 ± 0.8 years, 56.4 % female), 2162 from NHANES 1999–2006 (mean age: 61.6 ± 0.4 years, 49.7 % female), and 26,779 from the UK Biobank (mean age 59.6 ± 7.2 years, 39.5 % female). Detailed baseline characteristics stratified by frailty phenotype are presented in Supplementary Tables 10 & 11. Across all three cohorts, the prevalence of frailty was 9.9 %, 10.7 %, and 12.1 % for NHANES III, NHANES 1999–2006, and the UK Biobank, while the prevalence of pre-frailty was notably high at 44.2 %, 38.0 %, 54.1 %, respectively. Patients with diabetes and frailty were characterized as being older, predominantly female, with lower family income and education levels, longer diabetes duration, and having more long-term morbidities compared to non-frailty counterparts.
基线时共纳入 31,225 名符合条件的糖尿病患者（表 1），包括来自 NHANES III 的 2284 名（平均年龄 63.3 ± 0.8 岁，女性占 56.4%）、来自 NHANES 1999–2006 的 2162 名（平均年龄 61.6 ± 0.4 岁，女性占 49.7%）以及来自英国生物样本库的 26,779 名（平均年龄 59.6 ± 7.2 岁，女性占 39.5%）。按衰弱表型分层的基本特征详细信息见补充表 10 & 11。在所有三个队列中，NHANES III、NHANES 1999–2006 和英国生物样本库的衰弱患病率分别为 9.9%、10.7%和 12.1%，而前衰弱患病率则显著较高，分别为 44.2%、38.0%和 54.1%。与衰弱非患者相比，糖尿病合并衰弱患者表现为年龄更大、女性居多、家庭收入和教育水平较低、糖尿病病程更长以及慢性病更多。

Table 1. Baseline Characteristics of Participants with Diabetes in NHANES III, NHANES (1999–2006), and the UK Biobank.
表 1. NHANES III、NHANES（1999-2006）和英国生物样本库中糖尿病参与者的基线特征。

Characteristics  特征NHANES III (1988–1994)NHANES (1999–2006)UK Biobank (2006–2010)  英国生物样本库 (2006–2010)
No. of Participants  参与人数2284216226779
Age, years  年龄，岁63.3 (0.8)61.6 (0.4)59.6 (7.2)
≥ 65 years  ≥65 岁1318 (50.1)1123 (42.1)8051 (30.1)
Female  女性1249 (56.4)1042 (49.7)10569 (39.5)
White  白色940 (75.8)884 (66.5)23284 (87.0)
Low household income  低收入家庭777 (21.0)664 (22.3)-
Less than high school  未完成高中学业922 (22.8)981 (31.2)-
Townsend deprivation index
Townsend 剥夺指数		--−0.5 (3.4)
Never smoking  从不吸烟1053 (42.5)1002 (47.4)12079 (45.1)
Never drinking & special occasions only
从不饮酒，仅在特殊场合		1909 (81.1)1752 (80.0)9034 (33.7)
Healthy diet score or HEI
健康饮食评分或 HEI		66.9 (0.6)53.3 (0.3)3.3 (1.5)
Overweight & obesity  超重 & 肥胖1701 (72.3)1743 (82.5)23988 (89.6)
Diabetes duration ≥ 5 years
糖尿病病程 ≥ 5 年		650 (23.3)1053 (46.2)11300 (42.2)
Diabetes medication use  使用糖尿病药物758 (28.6)1378 (60.3)15351 (57.3)
HbA1c ≥ 7.0 %  HbA1c ≥ 7.0 %1422 (66.8)1098 (56.0)16076 (60.0)
The number of long-term morbidities ≥ 2
长期疾病数量 ≥ 2		1455 (65.7)2162 (74.1)13923 (52.0)
History of CVD  心血管疾病史563 (22.5)646 (28.1)5480 (20.5)
Pre-frailty  预衰弱1098 (44.2)850 (38.0)14482 (54.1)
Frailty  虚弱298 (9.9)261 (10.7)3252 (12.1)
Continuous variables (age, Townsend deprivation index, and healthy diet score or HEI) are presented as means (standard deviation). HEI, healthy eating index. The maximum values of the HEI are 100 in NHANES and the maximum values of the healthy diet score model are 10 in the UK Biobank. Categorical variables are presented as numbers (percentages). In NHANES, all estimates accounted for complex survey designs. HbA1c, glycated hemoglobin.
连续变量（年龄、Townsend 剥夺指数和健康饮食评分或 HEI）以均值（标准差）表示。HEI，健康饮食指数。NHANES 中 HEI 的最大值为 100，英国生物样本库中健康饮食评分模型的最大值为 10。分类变量以数量（百分比）表示。在 NHANES 中，所有估计值均考虑了复杂的调查设计。HbA1c，糖化血红蛋白。

3.2. Associations of frailty and pre-frailty with all-cause and CVD mortality in diabetes
3.2. 衰弱和前衰弱与糖尿病全因死亡率和心血管疾病死亡率的关系

The median follow-up durations were 13.9, 13.6, and 13.4 years for NHANES III, NHANES 1999–2006, and the UK Biobank, respectively. We documented 8276 deaths, containing 2626 CVD-related deaths. Diabetes patients with frailty exhibited the highest incidence rates of all-cause and CVD mortality compared to non-frail individuals. Following comprehensive adjustment for sociodemographic, behavioral, diabetes-related factors, and long-term morbidities, we observed consistent results that frailty was significantly associated with increased risks of all-cause mortality, with HRs of 1.72 (95 % CI: 1.28, 2.31), 2.18 (95 % CI: 1.63, 2.91), and 1.87 (95 % CI: 1.71, 2.04) across NHANES III, NHANES 1999–2006, and the UK Biobank, respectively. Similarly, frailty was significantly associated with increased risks of CVD mortality, with HRs of 1.80 (95 % CI: 1.24, 2.63), 2.32 (95 % CI: 1.38, 3.91), and 2.11 (95 % CI: 1.79, 2.50) for the respective cohorts (Table 2).
NHANES III、NHANES 1999–2006 和英国生物样本库的中位随访时间分别为 13.9、13.6 和 13.4 年。我们记录了 8276 例死亡事件，其中 2626 例与心血管疾病（CVD）相关。与非虚弱人群相比，患有糖尿病的虚弱患者在全因死亡率和心血管死亡率方面表现出最高的发病率。在全面调整了社会人口统计学、行为、糖尿病相关因素和长期疾病因素后，我们观察到一致的结果：虚弱与全因死亡率风险显著增加相关，NHANES III、NHANES 1999–2006 和英国生物样本库的 HR 值分别为 1.72（95% CI：1.28，2.31）、2.18（95% CI：1.63，2.91）和 1.87（95% CI：1.71，2.04）。同样，虚弱与心血管死亡率风险显著增加相关，相应队列的 HR 值分别为 1.80（95% CI：1.24，2.63）、2.32（95% CI：1.38，3.91）和 2.11（95% CI：1.79，2.50）（表 2）。

Table 2. HRs (95 % CIs) for All-cause and CVD Mortality according to Frailty Phenotype among Patients with Diabetes in NHANES III, NHANES (1999–2006) and the UK Biobank.
表 2. NHANES III、NHANES（1999-2006）和英国生物样本库中糖尿病患者的虚弱表型与全因和心血管疾病死亡率的风险比（95%置信区间）

Empty CellNHANES III (1988–1994)NHANES (1999–2006)The UK Biobank
Characteristic  特征Non-frailty  非虚弱Pre-frailty  预衰弱Frailty  虚弱Non-frailty  非虚弱Pre-frailty  预衰弱Frailty  虚弱Non-frailtyPre-frailtyFrailty
All-cause mortality  全因死亡率
Incidence/person-years  发病率/人年643/15881940/15009280/2849480/14000506/9748206/23141360/1183642850/1847901011/39019
Model 1  模型 11 (ref.)  1（参考）1.51 (1.30, 1.76)2.61 (2.07, 3.29)1 (ref.)  1（参考）1.70 (1.47, 1.96)3.21 (2.50, 4.12)1 (ref.)1.48 (1.39, 1.58)2.79 (2.57, 3.03)
Model 2  模型 21 (ref.)  1 (参考文献)1.40 (1.21, 1.61)2.33 (1.82, 2.97)1 (ref.)  1 (参考文献)1.64 (1.41, 1.91)2.67 (2.03, 3.51)1 (ref.)1.36 (1.27, 1.45)2.22 (2.03, 2.42)
Model 3  模型 31 (ref.)  1 (参考文献)1.28 (1.13, 1.47)1.72 (1.28, 2.31)1 (ref.)  1 (参考文献)1.53 (1.29, 1.81)2.18 (1.63, 2.91)1 (ref.)1.29 (1.21, 1.38)1.87 (1.71, 2.04)
CVD mortality  心血管疾病死亡率
Incidence/person-years  发病率/人年224/15881388/15009127/2849165/14000170/974877/2314356/118364800/184790319/39019
Model 1  模型 11 (ref.)  1 (参考文献)1.72 (1.35, 2.18)3.09 (2.26, 4.24)1 (ref.)  1 (参考文献)1.53 (1.16, 2.02)3.75 (2.44, 5.78)1 (ref.)1.61 (1.42, 1.83)3.49 (2.99, 4.07)
Model 2  模型 21 (ref.)  1 (参考文献)1.60 (1.21, 2.13)2.70 (1.93, 3.77)1 (ref.)  1 (参考文献)1.43 (1.07, 1.93)3.03 (1.92, 4.79)1 (ref.)1.45 (1.27, 1.64)2.69 (2.29, 3.16)
Model 3  模型 31 (ref.)  1 (参考文献)1.43 (1.09, 1.89)1.80 (1.24, 2.63)1 (ref.)  1 (参考文献)1.32 (0.99, 1.76)2.32 (1.38, 3.91)1 (ref.)1.34 (1.18, 1.53)2.11 (1.79, 2.50)
Model 1 was adjusted for age (continuous), sex, and race across all three studies, and considered complex sampling designs across NHANES studies, and adjusted for center across the UK Biobank study.
模型 1 在所有三个研究中调整了年龄（连续变量）、性别和种族，并考虑了 NHANES 研究的复杂抽样设计，并在英国生物样本库研究中调整了中心。
Model 2 was further adjusted for education level (in NHANES), family income (in NHANES), Townsend deprivation index (in the UK Biobank), smoking status, alcohol consumption, healthy diet score or healthy eating index (in quartiles), and BMI (< 25.0, 25.0–29.9, ≥ 30.0 kg/m2).
模型 2 进一步调整了教育水平（在 NHANES 中）、家庭收入（在 NHANES 中）、Townsend 剥夺指数（在英国生物样本库中）、吸烟状况、饮酒量、健康饮食评分或健康饮食指数（按四分位数），以及 BMI（< 25.0，25.0–29.9，≥ 30.0 kg/m²）。
Model 3 was adjusted for model 2 plus diabetes duration, diabetes medication use, HbA1c (< 7 % or ≥7 %), and the number of long-term morbidities.
模型 3 在模型 2 的基础上进一步调整了糖尿病病程、糖尿病药物使用、HbA1c（< 7% 或 ≥ 7%）以及长期慢性病数量。
Figure Legends  图例
Notably, diabetes with pre-frailty also exhibited an elevated risk of all-cause mortality, with HRs of 1.28 (95 % CI: 1.13, 1.47), 1.53 (95 % CI: 1.29, 1.81), and 1.29 (95 % CI: 1.21, 1.38), and CVD mortality, with HRs of 1.43 (95 % CI: 1.09, 1.89), 1.32 (95 % CI: 0.99, 1.76), and 1.34 (95 % CI: 1.18, 1.53), across the respective cohorts (Table 2).
值得注意的是，患有前期衰弱的糖尿病患者的全因死亡率也显著升高，其风险比（HR）分别为 1.28（95%置信区间：1.13，1.47）、1.53（95%置信区间：1.29，1.81）和 1.29（95%置信区间：1.21，1.38），心血管疾病（CVD）死亡率的风险比分别为 1.43（95%置信区间：1.09，1.89）、1.32（95%置信区间：0.99，1.76）和 1.34（95%置信区间：1.18，1.53），这些数据来自不同的队列（表 2）。

3.3. Secondary analyses  3.3. 次要分析

In stratified analysis, diabetes patients with frailty and pre-frailty consistently demonstrated a significant association with increased risks of all-cause and CVD mortality across various subgroups (Supplementary Tables 12&13). Particularly noteworthy was the significant elevation in mortality risk among middle-aged patients (40–64.9 years) with frailty, with an HR of 2.88 (95 % CI: 1.82, 4.54), exceeding that observed in older patients (≥65 years) in NHANES 1999–2006 (P for interaction = 0.02). We observed a considerable proportion (NHANES III, 5.9 %, 41.0 %; NHANES 1999–2006, 9.1 %, 37.6 %; the UK Biobank, 12.4 %, 54.2 %) of patients with diabetes exhibited frailty or pre-frailty within the middle-aged group (Supplementary Table 14). Additionally, among patients with diabetes of White race, frailty was associated with an elevated risk of CVD mortality in NHANES III and the UK Biobank (P for interaction = 0.05 and 0.03, respectively). Sensitivity analyses, excluding patients who died within two years of follow-up or had CVD at baseline, yielded consistent results, except for the association between frailty and pre-frailty and CVD mortality in NHANES III, which became non-significant (P = 0.09 and 0.07, respectively, approaching 0.05). Furthermore, in NHANES, consistent results were observed after excluding patients with only four frailty domains (Supplementary Table 15).
在分层分析中，患有衰弱和前衰弱的糖尿病患者在各种亚组中始终显示出与全因死亡率和心血管疾病死亡率增加的显著关联（补充表 12&13）。特别值得注意的是，40-64.9 岁中年患者中衰弱的死亡率风险显著升高，风险比（HR）为 2.88（95%置信区间：1.82，4.54），超过了 1999-2006 年 NHANES 研究中≥65 岁老年患者的观察值（交互作用 P 值=0.02）。我们观察到相当一部分（NHANES III，5.9%，41.0%；1999-2006 年 NHANES，9.1%，37.6%；英国生物样本库，12.4%，54.2%）的糖尿病患者在中年组中表现出衰弱或前衰弱（补充表 14）。此外，在白人糖尿病患者中，衰弱与 NHANES III 和英国生物样本库中的心血管疾病死亡率升高相关（交互作用 P 值分别为 0.05 和 0.03）。 敏感性分析排除了随访两年内死亡或基线时患有心血管疾病的患者，结果一致，除了在 NHANES III 中，虚弱与预虚弱与心血管疾病死亡率之间的关联变得不显著（P 分别为 0.09 和 0.07，接近 0.05）。此外，在 NHANES 中，排除仅具有四个虚弱领域的患者后，观察到一致的结果（补充表 15）。

3.4. Associations of five frailty domains with all-cause and CVD mortality in diabetes
3.4. 五个衰弱维度与糖尿病全因和心血管疾病死亡率的关联

We conducted further analysis to examine the association of five frailty components with all-cause and CVD mortality (Supplementary Figure 3 & Fig. 1). After fully adjusting for sociodemographic, behavioral, diabetes-related factors, morbidities, and five frailty components (mutual adjustment), a robust association was observed between slow gait speed and all-cause mortality with HRs of 1.23 (95 % CI: 1.03, 1.46), 1.53 (95 % CI: 1.11, 2.10), and 1.53 (95 % CI: 1.44, 1.64) and CVD mortality, with HRs of 1.43 (95 % CI: 1.12, 1.84), 2.15 (95 % CI: 1.18, 3.92), and 1.80 (95 % CI: 1.59, 2.03) across NHANES III, NHANES 1999–2006, and the UK Biobank, respectively. In the UK Biobank, we also observed a significant association of low physical activity, low grip strength, and weight loss with mortality risk, whereas a null association was found for exhaustion(Fig. 1).
我们对五个衰弱成分与全因和心血管疾病死亡率之间的关联进行了进一步分析（补充图 3 和图 1）。在充分调整社会人口统计学、行为、糖尿病相关因素、疾病史和五个衰弱成分（相互调整）后，观察到慢速步行速度与全因死亡率之间存在显著关联，NHANES III、NHANES 1999–2006 和英国生物样本库的 HR 分别为 1.23（95% CI：1.03，1.46）、1.53（95% CI：1.11，2.10）和 1.53（95% CI：1.44，1.64）；与心血管疾病死亡率之间的关联 HR 分别为 1.43（95% CI：1.12，1.84）、2.15（95% CI：1.18，3.92）和 1.80（95% CI：1.59，2.03）。在英国生物样本库中，我们还观察到低身体活动、低握力和体重减轻与死亡率风险之间存在显著关联，而疲劳则未发现关联（图 1）。
Fig. 1
  1. Download: Download high-res image (377KB)
    下载：下载高分辨率图像（377KB）
  2. Download: Download full-size image
    下载：下载完整尺寸图像

Fig. 1. Individual components of frailty and their association with all-cause and CVD mortality. Associations of frailty domains and (A) all-cause and (B) CVD mortality. Multivariable adjusted for age, sex, race, center (in the UK Biobank), education (in NHANES), family income (in NHANES), Townsend deprivation index (in the UK Biobank), smoking, alcohol consumption, healthy diet score (in the UK Biobank) or healthy eating index (in NHANES), BMI, diabetes duration, diabetes medication use, HbA1c, morbidities, and five frailty components (mutual adjustment). We also considered complex sampling designs in NHANES studies.
图 1. 虚弱个体成分及其与全因和心血管疾病死亡率的关联。虚弱领域与（A）全因和（B）心血管疾病死亡率的关联。多变量调整了年龄、性别、种族、中心（在英国生物样本库中）、教育（在 NHANES 中）、家庭收入（在 NHANES 中）、Townsend 剥夺指数（在英国生物样本库中）、吸烟、饮酒、健康饮食评分（在英国生物样本库中）或健康饮食指数（在 NHANES 中）、BMI、糖尿病病程、糖尿病药物使用、HbA1c、疾病和五个虚弱成分（相互调整）。我们还考虑了 NHANES 研究中的复杂抽样设计。

3.5. Meta-analysis  3.5.荟萃分析

Following database searching, 7294 records were identified, and after removing duplicates, and screening titles, abstracts and full-text articles, 17 articles (Cacciatore et al., 2013, Castro-Rodríguez et al., 2016, Chao et al., 2018, Chode et al., 2016, Ferri-Guerra et al., 2020, He et al., 2022, Huang et al., 2023, Kitamura et al., 2019, Leung et al., 2021, Mickute et al., 2023, Nguyen et al., 2021, O’Donovan et al., 2019, Qin and Zheng, 2023, Thein et al., 2018, Wang et al., 2014, Weng et al., 2023, Xiong et al., 2024) met the inclusion criteria (Supplementary Figure 4). Among these, He’s article (He et al., 2022) comprised two cohort studies from different countries, and Mickute’s study (Mickute et al., 2023) mirrored ours by utilizing data from the UK Biobank. Xiong’s article (Xiong et al., 2024), which used the same population and frailty assessments as Qin’s study (Qin and Zheng, 2023), was included in the main meta-analysis (adjusting for more confounding factors and providing up-to-date evidence), while Qin’s study was only considered in subgroup analysis. The main meta-analysis encompassed 19 cohort studies (a total of 753,480 patients), including NHANES III, NHANES 1999–2006, and the UK Biobank studies featured in our present analysis. Supplementary Table 16 presents the characteristics of the included studies, which scored 6–9 points based on the Newcastle-Ottawa quality assessment scale.
数据库检索后，共识别出 7294 条记录，去除重复记录后，通过筛选标题、摘要和全文文章，最终有 17 篇文章（Cacciatore 等，2013；Castro-Rodríguez 等，2016；Chao 等，2018；Chode 等，2016；Ferri-Guerra 等，2020；He 等，2022；Huang 等，2023；Kitamura 等，2019；Leung 等，2021；Mickute 等，2023；Nguyen 等，2021；O’Donovan 等，2019；Qin 和 Zheng，2023；Thein 等，2018；Wang 等，2014；Weng 等，2023；Xiong 等，2024）符合纳入标准（补充图 4）。其中，He 的文章（He 等，2022）包含来自不同国家的两项队列研究，Mickute 的研究（Mickute 等，2023）通过使用英国生物样本库数据与我们的研究相似。Xiong 的文章（Xiong 等，2024），使用了与 Qin 的研究（Qin 和 Zheng，2023）相同的队列和衰弱评估方法，被纳入主要荟萃分析（调整了更多混杂因素并提供最新证据），而 Qin 的研究仅在亚组分析中被考虑。 主要荟萃分析涵盖了 19 项队列研究（总计 753,480 名患者），包括 NHANES III、NHANES 1999–2006 以及我们本次分析中包含的英国生物样本库研究。补充表 16 列出了纳入研究的特征，这些研究根据纽卡斯尔-渥太华质量评估量表评分为 6–9 分。
Forest plots depicted that among patients with diabetes, the pooled HRs (95 % CI) for frailty and pre-frailty and all-cause mortality were 1.83 (1.55, 2.16), and 1.29 (1.17, 1.42), and for CVD mortality were 1.97 (1.51, 2.57), and 1.35 (1.21, 1.51), respectively (Fig. 2). Notably, significant heterogeneity was observed across 19 included studies related to frailty and pre-frailty and all-cause mortality (I2 were 91.4 % and 67.8 %), with meta-regression analysis attributing 12.2 % and 91.4 % of the heterogeneity to frailty measurements. Subgroup analysis showed that the pooled HRs (95 % CI) for frailty and pre-frailty (defined based on frailty phenotype) and all-cause mortality were 2.15 (1.81, 2.55), and 1.32 (1.25, 1.39) respectively (Supplementary Figure 5). Summary results for measurements, age and sex subgroups on pooled HRs are shown in Supplementary Figures 5–8.
森林图显示，在糖尿病患者中，虚弱和预虚弱与全因死亡率和心血管死亡率合并的 HRs（95% CI）分别为 1.83（1.55，2.16）和 1.29（1.17，1.42），以及 1.97（1.51，2.57）和 1.35（1.21，1.51）（图 2）。值得注意的是，在 19 项与虚弱和预虚弱及全因死亡率相关的研究中观察到显著的异质性（I²分别为 91.4%和 67.8%），荟萃回归分析将异质性的 12.2%和 91.4%归因于虚弱测量。亚组分析显示，基于虚弱表型的虚弱和预虚弱与全因死亡率合并的 HRs（95% CI）分别为 2.15（1.81，2.55）和 1.32（1.25，1.39）（补充图 5）。关于测量、年龄和性别亚组的合并 HRs 的汇总结果分别显示在补充图 5-8 中。
Fig. 2
  1. Download: Download high-res image (393KB)
    下载：下载高分辨率图片（393KB）
  2. Download: Download full-size image
    下载：下载完整尺寸图片

Fig. 2. Meta-analysis of the association between frailty status and all-cause and CVD mortality. Forest plots of the associations between (A) frailty and all-cause mortality, (B) pre-frailty and all-cause mortality, (C) frailty and CVD mortality, (D) pre-frailty and CVD mortality.
图 2. 虚弱状态与全因死亡率和心血管疾病死亡率之间关联的荟萃分析。虚弱与全因死亡率（A）、前期虚弱与全因死亡率（B）、虚弱与心血管疾病死亡率（C）、前期虚弱与心血管疾病死亡率（D）之间关联的森林图。

4. Discussion  4. 讨论

Our analysis across three prospective cohorts consistently revealed a significant association of frailty and pre-frailty with increased risks of all-cause and CVD mortality in patients with diabetes. Through a comprehensive meta-analysis encompassing a sizable cohort of 753,480 patients, we substantiated that frailty was associated with a 1.8-fold risk of all-cause mortality and a 2.0-fold risk of CVD mortality. Similarly, pre-frail individuals exhibited a 1.3-fold risk of all-cause mortality and a 1.4-fold risk of CVD mortality compared to their non-frailty counterparts.
我们的分析跨越三个前瞻性队列，一致显示虚弱和前虚弱与糖尿病患者的全因死亡率和心血管疾病死亡率风险增加存在显著关联。通过一项涵盖 753,480 名患者的大型荟萃分析，我们证实虚弱与全因死亡率风险增加 1.8 倍、心血管疾病死亡率风险增加 2.0 倍相关。同样，与非虚弱人群相比，前虚弱个体在全因死亡率风险上增加 1.3 倍，在心血管疾病死亡率风险上增加 1.4 倍。
Diabetes patients persistently confront elevated hazards of premature death (Wong and Sattar, 2023). Therefore, identifying mortality-associated risk factors and predictors is imperative to guide the management of diabetes. Growing evidence underscores that frailty is prevalent among diabetes and associated with adverse outcomes (Castro-Rodríguez et al., 2016, Espeland et al., 2022, Hanlon et al., 2020, He et al., 2022, Kitamura et al., 2019, Kong et al., 2021, Weng et al., 2023). Due to convenience and ease of implementation, frailty phenotype stands as the most utilized metric for assessing frailty (Fried et al., 2001, Hanlon et al., 2020). In this study, we evaluated baseline frailty phenotype among patients with diabetes in three prospective cohorts and observed significant associations between frailty and increased risks of mortality, consistent with previous studies (Leung et al., 2021, Thein et al., 2018). Diverging from previous approaches that solely classified individuals as frail or non-frail (Ferri-Guerra et al., 2020, Huang et al., 2023, Leung et al., 2021, Nguyen et al., 2021, Qin and Zheng, 2023, Thein et al., 2018, Weng et al., 2023), we incorporated pre-frailty into our analysis and consistently observed a 1.3-fold to 1.5-fold augmented risk of mortality across three cohorts. Notably, while a quarter of pre-frail cases can be reversed, the reversal rate among frail individuals is merely 3 % (Kojima et al., 2019). This underscores the imperative of early screening for pre-frailty, particularly as an increasing number of studies have been interested in its association with mortality. A recent investigation with 2510 diabetes spanning two prospective cohorts from China and the UK found no significant link between baseline pre-frailty and mortality over a follow-up period exceeding six years (He et al., 2022). Conversely, another study involving 17,252 patients recently observed a significant association between pre-frailty and increased mortality risk (Mickute et al., 2023). These inconsistent findings prompt us to identify whether pre-frailty can serve as a risk factor or predictor of premature death in patients with diabetes. Therefore, we systematically summarized available evidence alongside the three cohorts’ findings, affirming that pre-frailty among patients with diabetes is associated with a 1.3-fold increased risk of all-cause mortality.
糖尿病患者持续面临过早死亡的风险升高（Wong and Sattar, 2023）。因此，识别与死亡率相关的风险因素和预测因素对于指导糖尿病管理至关重要。越来越多的证据表明，虚弱在糖尿病患者中普遍存在，并与不良结局相关（Castro-Rodríguez et al., 2016, Espeland et al., 2022, Hanlon et al., 2020, He et al., 2022, Kitamura et al., 2019, Kong et al., 2021, Weng et al., 2023）。由于方便且易于实施，虚弱表型成为评估虚弱的常用指标（Fried et al., 2001, Hanlon et al., 2020）。在本研究中，我们评估了三个前瞻性队列中糖尿病患者的基线虚弱表型，观察到虚弱与死亡风险增加之间存在显著关联，这与先前的研究结果一致（Leung et al., 2021, Thein et al., 2018）。 与以往仅将个体分为虚弱或非虚弱（Ferri-Guerra 等，2020；Huang 等，2023；Leung 等，2021；Nguyen 等，2021；Qin 和 Zheng，2023；Thein 等，2018；Weng 等，2023）的方法不同，我们将预虚弱纳入分析，并在三个队列中持续观察到死亡率增加了 1.3 至 1.5 倍。值得注意的是，虽然四分之一的预虚弱病例可以逆转，但虚弱个体的逆转率仅为 3%（Kojima 等，2019）。这突出了早期筛查预虚弱的重要性，特别是随着越来越多的研究对其与死亡率的相关性感兴趣。一项涉及 2510 名糖尿病患者的最新研究，跨越中国和英国的两个前瞻性队列，发现基线预虚弱与超过六年的随访期间死亡率之间没有显著关联（He 等，2022）。相反，另一项涉及 17252 名患者的研究最近观察到预虚弱与死亡率风险增加之间存在显著关联（Mickute 等，2023）。 这些不一致的发现促使我们确定前衰弱是否可以作为糖尿病患者的死亡风险因素或预测因素。因此，我们系统地总结了现有证据以及三个队列的研究结果，证实糖尿病患者的前衰弱与全因死亡率风险增加 1.3 倍相关。
CVD remains the leading cause of death for diabetes (Wong and Sattar, 2023). However, evidence supporting the association between frailty and CVD mortality in diabetes remains insufficient. A secondary analysis of a randomized controlled trial is the first to investigate such association (Nguyen et al., 2021). Nevertheless, the generalizability of this evidence was limited due to the study design, e.g., strict exclusion criteria and intervening factors. In contrast, our study provided robust and representative evidence based on large, nationally representative cohorts with long follow-up durations. Our findings demonstrated that the association between frailty and CVD mortality exhibited some attenuation post-adjustment for diabetes-related parameters and long-term morbidities but remained substantial. Additionally, our meta-analysis pooled results from three cohorts and represents the first systematic evidence in this field, elucidating the association between frailty and CVD mortality in diabetes.
心血管疾病仍然是糖尿病的主要死亡原因（Wong 和 Sattar，2023 年）。然而，关于虚弱与糖尿病心血管疾病死亡率之间关联的证据仍然不足。一项随机对照试验的二次分析首次研究了这种关联（Nguyen 等人，2021 年）。然而，由于研究设计（例如严格的排除标准和干预因素），该证据的普适性有限。相比之下，我们的研究基于大型、具有全国代表性的队列和较长的随访时间，提供了可靠且有代表性的证据。我们的研究结果表明，在调整了糖尿病相关参数和长期疾病后，虚弱与心血管疾病死亡率之间的关联有所减弱，但仍保持显著。此外，我们的荟萃分析整合了三个队列的结果，代表了该领域的首次系统证据，阐明了糖尿病中虚弱与心血管疾病死亡率之间的关联。
In stratified analyses, we observed that middle-aged patients in frailty displayed a significantly elevated risk of mortality, approaching or even exceeding that observed in older patients. In contrast to the age-related decline in physical function prevalent in the elderly, the manifestation of frailty in middle-aged patients may stem from significant underlying illnesses or disabilities, profoundly curtailing life expectancy (Brummel et al., 2017). Additionally, we observed that Caucasian patients in frailty exhibited a heightened vulnerability to CVD mortality compared to their counterparts of other racial/ethnic backgrounds. While prior studies have indicated a higher incidence of CVD among non-Hispanic White diabetes patients (An et al., 2021), the interplay among race, frailty, and CVD mortality remains largely unknown. We found a lower prevalence of frailty among White patients relative to non-White counterparts, aligning with preceding evidence (Nik-Ahd et al., 2022). However, Whites with diabetes confronted heightened susceptibility to CVD mortality upon frailty onset compared to non-Whites. Unraveling the underlying mechanisms driving this disparity warrants exploration in future investigations.
在分层分析中，我们发现虚弱的中年患者表现出显著升高的死亡率，接近甚至超过老年患者的观察结果。与老年人普遍存在的身体功能随年龄下降的现象相反，中年患者虚弱的表现可能源于严重的潜在疾病或残疾，极大地缩短了预期寿命（Brummel 等人，2017）。此外，我们发现白人虚弱患者对心血管疾病（CVD）死亡率的易感性显著高于其他种族/民族背景的患者。虽然先前研究表明非西班牙裔白人糖尿病患者的心血管疾病发病率较高（An 等人，2021），但种族、虚弱和心血管疾病死亡率之间的相互作用仍 largely 未知。我们发现白人患者的虚弱患病率低于非白人患者，这与先前证据一致（Nik-Ahd 等人，2022）。然而，与非白人相比，糖尿病患者一旦出现虚弱，对心血管疾病死亡率的易感性显著升高。 揭示导致这种差异的潜在机制需要在未来的研究中进行探索。
In meta-analysis, we synthesized data from diverse populations, thereby providing broader generalizability and greater robustness to the findings. While our three cohort studies consistently identified strong associations between frailty and mortality, the magnitude of increased risk varied across these studies. The meta-analysis resolved this variability and provided precise estimates of the increased risk of all-cause and CVD mortality associated with frailty in individuals with diabetes. This clarification could offer valuable insights into the potential role of frailty assessments in diabetes management. At the same time, the meta-analysis also summarized the association between other frailty assessment methods and mortality risk in diabetes, which increased the clinical application value of this study. Due to the high prevalence of frailty in diabetes, clinicians managing diabetes will encounter frailty regardless of clinical setting. Although the consensus on diagnosis and clinical management of frailty individuals in diabetes is lacking at present, the health benefits of early frailty assessment and interventions are visible. A recent study published in European Heart Journal found that frail participants who recovered to pre-frail or non-frail status presented decreased risks of incident CVD (He et al., 2024). Among patients with diabetes, an increase in the frailty index is associated with an elevated risk of premature death (Espeland et al., 2022). Frailty assessment of diabetes management would be an important step in understanding if and how interventions might mitigate frailty and increase healthy longevity, and individualized interventions are recommended for patients with frailty and diabetes (Hanlon et al., 2020).
在荟萃分析中，我们从不同人群综合了数据，从而为研究结果提供了更广泛的普适性和更强的稳健性。尽管我们的三项队列研究一致地识别了虚弱与死亡率之间的强关联，但这些研究之间风险增加的幅度存在差异。荟萃分析解决了这种变异性，并提供了与糖尿病患者的全因死亡率和心血管疾病死亡率相关的虚弱风险增加的精确估计。这种澄清可以为虚弱评估在糖尿病管理中的潜在作用提供有价值的见解。同时，荟萃分析还总结了其他虚弱评估方法与糖尿病患者死亡率风险之间的关联，这增加了本研究的临床应用价值。由于虚弱在糖尿病中的高患病率，糖尿病临床管理者无论在何种临床环境下都会遇到虚弱问题。尽管目前缺乏对糖尿病患者虚弱个体诊断和临床管理的共识，但早期虚弱评估和干预的健康益处是显而易见的。 最近发表在《欧洲心脏杂志》上的一项研究发现，恢复到前衰弱或非衰弱状态的衰弱参与者其心血管疾病（CVD）发病风险有所降低（He 等人，2024）。在糖尿病患者中，衰弱指数的增加与过早死亡风险升高相关（Espeland 等人，2022）。对糖尿病管理的衰弱评估是理解干预措施是否以及如何减轻衰弱并增加健康长寿的重要步骤，建议对衰弱和糖尿病的患者进行个体化干预（Hanlon 等人，2020）。
Our study possesses two noteworthy strengths. First, we leveraged data from representative populations in the US, a large-scale cohort from the UK, and a meta-analysis encompassing all available studies, thereby furnishing the most comprehensive evidence in this regard. Second, utilizing these high-quality data, we were able to consider numerous confounders, particularly comorbidities, often neglected in previous studies. However, several limitations should be considered. First, assessment of the frailty phenotype was limited to baseline measurements, precluding the association between changes in frailty status and mortality. Second, although previous studies have predominantly focused on frailty among the older population, it is intriguing to observe a considerable prevalence of frailty among middle-aged patients with diabetes in all three cohorts. This prevalence closely resembles that observed in older counterparts, similarly increasing the risks associated with frailty-related all-cause and CVD mortality. Third, due to the observational nature of the three studies utilized, direct causality inference should be cautious.
我们的研究具有两个显著的优点。首先，我们利用了美国代表性人群、英国大规模队列以及包含所有可用研究的荟萃分析数据，从而提供了最全面的证据。其次，利用这些高质量数据，我们能够考虑许多混杂因素，特别是既往研究中常被忽视的合并症。然而，仍需考虑一些局限性。首先，虚弱表型的评估仅限于基线测量，无法探讨虚弱状态变化与死亡率之间的关联。其次，尽管既往研究主要关注老年人群的虚弱，但观察发现所有三个队列中糖尿病中年患者中虚弱现象相当普遍。这种患病率与老年人相似，同样增加了与虚弱相关的全因和心血管疾病死亡率风险。第三，由于所使用的三个研究均为观察性研究，直接因果关系推断应谨慎。

5. Conclusions  5. 结论

In summary, this study identified an association between frailty and pre-frailty and increased all-cause and CVD-related mortality in middle-aged and older individuals with diabetes. Integrating frailty assessment into routine practice to identify frail and pre-frail status is recommended for diabetes managment, followed by the implementation of targeted healthy lifestyle interventions to mitigate adverse outcomes.
总之，本研究确定了虚弱和预虚弱与糖尿病中年和老年个体全因死亡率和心血管疾病相关死亡率增加之间的关联。建议将虚弱评估纳入常规实践，以识别虚弱和预虚弱状态，为糖尿病管理，随后实施有针对性的健康生活方式干预，以减轻不良后果。

Funding  资助

This research was funded by the Key Special Projects of the Ministry of Science and Technology (grant numbers: 2022YFA1103600), the National Natural Science Foundation of China (grant numbers: 82473618 & 92357305), and the China Association for Science and Technology (grant numbers: 2021QNRC001 & 2020QNRC001).
本研究由科学技术部重点专项（资助编号：2022YFA1103600）、国家自然科学基金（资助编号：82473618 & 92357305）和中国科学技术协会（资助编号：2021QNRC001 & 2020QNRC001）资助。

Declaration of Competing Interest
利益冲突声明

No potential conflicts of interest relevant to this work were reported.
未报告与本研究相关的潜在利益冲突。

Acknowledgments  致谢

The authors thank the participants and staff of the NHANES and the UK Biobank for their valuable contributions. Data from the UK Biobank is available to all researchers upon submission of application. This research has been conducted using the UK Biobank Resource under Application 63454.
作者感谢 NHANES 和英国生物样本库的参与者和工作人员的贡献。英国生物样本库的数据在提交申请后对所有研究人员开放。本研究是在英国生物样本库资源（申请号 63454）的支持下进行的。

Author contributions  作者贡献

L.C. and Y.L. proposed and designed the study. L.W. and X.T. conducted the statistical analysis. L.W., L.C., X.L. and Y.A. completed the literature search and data extraction. L.W. and L.C. drafted the first version of the manuscript. Y.L., X.T., X.L. and Y.A. participated in the critical revision of the manuscript. L.C. is the guarantor of this work and, as such, has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors reviewed and approved the final version of the manuscript.
L.C.和 Y.L.提出了研究方案并设计了研究。L.W.和 X.T.进行了统计分析。L.W.、L.C.、X.L.和 Y.A.完成了文献检索和数据提取。L.W.和 L.C.起草了文稿的初稿。Y.L.、X.T.、X.L.和 Y.A.参与了文稿的审阅修订。L.C.是本研究的保证人，因此可以完全访问研究中的所有数据，并对数据的完整性和数据分析的准确性负责。所有作者均审阅并批准了文稿的最终版本。

Appendix A. Supplementary material
附录 A.补充材料

Download: Download Word document (3MB)

Supplementary Figure 1. Supplementary material

References

Cited by (1)

  • Associations between thyroid function, thyroid hormone sensitivity indices, and frailty: Insights from a large cross-sectional study
    甲状腺功能、甲状腺激素敏感性指标与衰弱的关系：一项大型横断面研究的启示

    2025, Experimental Gerontology
    2025 年，《实验老年学》
View Abstract  查看摘要
© 2025 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
© 2025 Elsevier B.V. 保留所有权利，包括文本和数据挖掘、人工智能训练以及类似技术的权利。