背景

随着人口老龄化问题日益显现，衰弱成为公共卫生的主要优先事项之一。衰弱在中年和老年人中均可识别到，在中老年人群中分析危险因素与衰弱的关联，并估计人群归因风险，以期为卫生专业人员制定衰弱的预防和干预措施提供依据。由于人口年龄结构的转变，衰弱与糖尿病和心血管疾病的关系也逐渐得到重视，特别是衰弱成为心血管疾病中一个高度优先的主题。然而，衰弱与糖尿病和心血管疾病发病风险关联的证据较为有限，尤其是缺少来自中国人群的证据。因此，本研究还将分析中老年人衰弱与糖尿病和心血管疾病发病风险的关联，以促进这些疾病中针对衰弱的预防和干预措施的制定。

方法

本研究基于中国健康与养老追踪调查（China Health and Retirement Longitudinal Study，CHARLS）2011-2018年的数据。衰弱指数（Frailty index）用于评估衰弱状态。

中老年人衰弱的危险因素研究：基于既往研究结果、CHARLS项目收集信息及公共卫生意义，选择12项与衰弱存在关联的常见的可改变危险因素，包括2项社会经济地位因素（教育程度和居住地区）、3项社会关系因素（婚姻状况、社会活动和生活满意度）、2项生活方式因素（吸烟和睡眠）和5项代谢危险因素（腹部肥胖、体质指数、高血压、糖尿病和血脂异常）。为每项危险因素分配0-1分，以获得社会经济地位、社会关系、生活方式和代谢风险因素的总分。研究结局是45岁及以上中老年人衰弱的发生，定义为衰弱指数≥0.25。基于CHARLS 2011-2018年的调查数据，在所有中老年人和三个年龄组人群（45-54岁、55-64岁和≥65岁），使用Cox比例风险模型分析单独的和综合的危险因素与衰弱风险的关联，并估计人群归因危险度百分比（Population attributable risk percentage，PAR%）。

中老年人衰弱与糖尿病或心血管疾病发病风险的关联性研究：基线（2011年）衰弱与糖尿病或心血管疾病发病风险的关联分析：基于CHARLS 2011-2015年（用于糖尿病）或2011-2018年（用于心血管疾病）的数据，将基线衰弱状态定义为3类：无衰弱（衰弱指数≤0.10）、衰弱前期（衰弱指数>0.10至<0.25）和衰弱（衰弱指数≥0.25）。Cox比例风险模型用于分析基线衰弱状态与糖尿病或心血管疾病发病风险的关联。限制性立方样条函数（Restricted cubic splines，RCS）用于分析基线衰弱指数作为连续变量与糖尿病或心血管疾病发病风险的非线性关联。衰弱发展轨迹与糖尿病或心血管疾病发病风险的关联分析：基于CHARLS 2011-2018年的数据，使用组基轨迹模型，以及2011、2013和2015年的调查数据确定衰弱发展轨迹。Cox比例风险模型用于分析衰弱发展轨迹与2018年糖尿病或心血管疾病发病风险的关联。

结果

1 中老年人衰弱的危险因素研究：排除基线存在衰弱，危险因素和衰弱发病信息缺失的参与者后，6852名参与者被纳入分析。在中位6.92（IQR 3.92-7.01）年随访期间内，共发生1133例衰弱事件。所有人群中，除吸烟和血脂异常外，其余所有的危险因素均与衰弱的风险显著增加有关，其中较低的教育程度与衰弱风险的关联程度最强（HR=1.61，95% CI=1.40-1.83）。综合的危险因素中，社会经济地位与衰弱风险的关联程度最强（HR=1.53，95% CI=1.41-1.67）。按年龄分组的结果显示，低教育程度与<65岁成年人中衰弱的关联性最强（45-54岁年龄组中HR=1.79；95% CI=1.37-2.34；55-64岁年龄组中HR=1.77；95% CI=1.44-2.19），而不健康睡眠与≥65岁成年人衰弱的关联性最强（HR=1.43；95% CI=1.15-1.77）。而综合的危险因素中，较低的社会经济地位与所有年龄组衰弱风险的关联性均为最强（45-54岁年龄组中HR为1.57 [95% CI=1.32-1.87]；55-64年龄组中HR为1.74 [95% CI=1.52-2.00]；≥65岁年龄组中HR为1.30 [95% CI=1.14-1.50]）。

从人群水平来看，低教育程度和居住在农村地区对所有中老年人衰弱风险的贡献最大，PAR%分别为23.5%和22.2%。二者综合得出的社会经济地位也是对衰弱风险贡献最大的因素，PAR%为40.8%。按年龄分组的结果显示，低教育程度和居住在农村地区对衰弱的贡献随着年龄的增长而降低，二者是导致65岁以下成年人衰弱的主要危险因素（45-54岁年龄组中，低教育程度相关的PAR%为21.5%，居住在农村地区相关的PAR%为18.1%；55-64岁年龄组中，低教育程度相关的PAR%为30.3%，居住在农村地区相关的PAR%为31.6%）。然而，二者综合得出的社会经济地位仍是导致所有年龄组衰弱发生的主要危险因素（45-54、55-64和≥65岁年龄组中PAR%分别为38.2%、51.8%和28.0%）。社会关系对中老年人衰弱的影响主要来自于低社会活动，代谢危险因素对中老年人衰弱的影响主要来自腹部肥胖，可归因于社会关系和几项单独的代谢危险因素（血脂异常除外）的衰弱风险也随着年龄的增长而降低。生活方式对中老年人衰弱的影响主要来自不健康睡眠，与其他危险因素不同，不健康睡眠对衰弱的贡献随着年龄逐渐增加，并成为65岁及以上成年人衰弱发生的首要原因，PAR%为21.2%。

2 中老年人衰弱与糖尿病发病风险的关联性研究：基线（2011年）衰弱与糖尿病发病风险的关联：13689名基线无糖尿病的参与者被纳入分析，其中7468名（54.55%）归类为无衰弱，4954名（36.19%）归类为衰弱前期，1267名（9.26%）归类为衰弱。中位随访4.00年（IQR 3.60-4.00）期间，发生513例糖尿病事件。与基线无衰弱人群相比，基线为衰弱前期和衰弱人群的糖尿病风险升高，调整所有潜在的混杂因素后的HR（95% CI）分别为1.55（1.27-1.89）和2.09（1.56-2.79）。限制性立方样条分析表明衰弱指数≥0.09时，糖尿病的发病风险增加。衰弱发展轨迹与糖尿病发病风险的关联：9461名在2011-2015年期间无糖尿病的参与者被纳入分析。使用组基轨迹模型以及2011、2013和2015的调查数据确定3组不同的衰弱轨迹：低水平稳定组（n=6216，65.7%）、中等水平增加组（n=2673，28.3%）和高水平增加组（n=572，6.0%）。在中位3.00年（IQR 2.92-3.00）的随访期间，确定505例糖尿病事件。相比于低水平稳定组人群，中等水平增加组和高水平增加组人群糖尿病的发病风险增加，调整所有潜在的混杂因素后的HR（95% CI）分别为1.50（1.22-1.84）和1.70（1.21-2.38）。未观察到常规的糖尿病危险因素，包括吸烟、饮酒、睡眠、肥胖、高血压和血脂异常对基线衰弱或衰弱发展轨迹与糖尿病风险关联的影响。一系列的敏感性分析也未改变基线衰弱或衰弱发展轨迹与糖尿病风险的关联。

3 中老年人衰弱与心血管疾病发病风险的关联性研究：基线（2011年）衰弱与心血管疾病发病风险的关联：12575名基线时无心血管疾病的参与者被纳入分析，其中7255名（57.69%）归类为无衰弱，4360名（34.67%）归类为衰弱前期，960名（7.63%）归类为衰弱。中位6.92年（IQR 3.05-7.09）的随访期间，发生了2004例心血管疾病事件。与基线无衰弱人群相比，基线为衰弱前期和衰弱人群的心血管疾病风险增加，调整所有潜在的混杂因素后关联的HR（95% CI）分别为1.45（1.31-1.60）和2.05（1.76-2.38）。限制性立方样条分析表明衰弱指数≥0.09时，心血管疾病的发病风险增加。衰弱发展轨迹与心血管疾病发病风险的关联：8331名在2011-2015年期间无心血管疾病的参与者纳入分析。使用组基轨迹模型以及2011、2013和2015的调查数据确定3组不同的衰弱轨迹：低水平稳定组（n=5480，65.8%）、中等水平增加组（n=2393，28.7%）和高水平增加组（n=458，5.5%）。中位3.00年（IQR 2.92-3.00）的随访期间，确定了1008例心血管疾病事件。相比于低水平稳定组人群，中等水平增加组和高水平增加组人群心血管疾病的发病风险增加，调整所有潜在的混杂因素后的HR（95% CI）分别为1.85（1.61-2.13）和2.42（1.94-3.02）。未观察到常规的心血管疾病危险因素，包括吸烟、饮酒、睡眠、肥胖、高血压、糖尿病和血脂异常对基线衰弱或衰弱发展轨迹与心血管疾病风险关联的影响。一系列的敏感性分析也未改变基线衰弱或衰弱发展轨迹与心血管疾病风险的关联。

结论

1 社会经济地位是对中老年人衰弱风险贡献最大的因素，低教育程度和居住在农村地区是主要危险因素。然而，低教育程度和居住在农村地区对衰弱风险的影响随着年龄的增长而降低，二者是导致65岁以下人群衰弱的主要危险因素。相反，不健康睡眠对衰弱风险的影响随着年龄的增长而增加，并成为65岁及以上人群衰弱的主要危险因素。总之，危险因素对衰弱的影响存在年龄差异，优先考虑不同年龄阶段对衰弱影响较大的危险因素，将有助于最大化的降低衰弱的负担。

2 基线衰弱和随着时间增加的衰弱轨迹与糖尿病的发病风险呈正相关。这些结果独立于实际年龄和常见的糖尿病危险因素（包括吸烟、饮酒、睡眠、肥胖、高血压和血脂异常），一系列的亚组分析和敏感性分析也证明了研究结果的稳健性。在中老年人群中制定针对衰弱的筛查和干预措施，将有助于降低糖尿病的发病风险。

3 在中国中老年人群，基线衰弱和随时间增加的衰弱轨迹与心血管疾病的发病风险呈正相关。这些结果独立于个体的实际年龄和常见的心血管疾病的危险因素（包括吸烟、饮酒、睡眠、肥胖、高血压、糖尿病和血脂异常），一系列的亚组分析和敏感性分析也证明了研究结果的稳健性。在中老年人群中制定针对衰弱的筛查和干预措施，将有助于降低心血管疾病的负担。

Background

As the ageing of the population becomes more apparent, frailty is emerging as one of the most important priorities in public health. Frailty has been identified in both middle-aged and older adults; analysis of the association between risk factors and frailty and estimation of the population-attributable risk in middle-aged and older people will provide evidence for health professionals in frailty prevention and intervention. With the changing age structure of the population, the association of frailty with diabetes and cardiovascular disease has gained traction. In particular, frailty has become a high-priority issue in cardiovascular disease. However, evidence on the association of frailty with the risk of diabetes and cardiovascular disease is more limited, especially in the Chinese population. Thus, the association of frailty with the risk of diabetes and cardiovascular disease was analysed, with a view to the development of prevention and intervention strategies for frailty in these diseases.

Methods

This study was based on the data of the China Health and Retirement Longitudinal Study (CHARLS) from 2011 to 2018. Frailty index was used to assess frailty status.

Risk factors for frailty in middle-aged and older adults: Twelve modifiable risk factors associated with frailty were selected based on the results of previous studies, information collected in the CHARLS project, and public health implications, including two socioeconomic status (education level and residence), three social relationships (marital status, social activities, and life satisfaction), two lifestyle factors (smoking and sleep), and five metabolic risk factors (abdominal obesity, body mass index, hypertension, diabetes, and dyslipidemia). A score of 0-1 was assigned to each risk factor to give an overall score for each domain. The study outcome was the occurrence of frailty in middle-aged and older adults aged 45 years and older, defined as a frailty index≥0.25. Using the CHARLS data 2011 to 2018, Cox proportional hazards models were used to analyse the association of individual and combined risk factors with the risk of frailty among all middle-aged and older adults and in three age groups (45-54 years, 55-64 years, and ≥65 years), and to estimate the population attributable risk percentage (PAR%).

Association of frailty with risk of diabetes or cardiovascular disease in middle-aged and older adults: Association of baseline (2011) frailty with risk of diabetes or cardiovascular disease: Using CHARLS data from 2011 to 2015 (for diabetes) or 2011 to 2018 (for cardiovascular disease), baseline frailty status was defined in 3 categories: non-frail (frailty index ≤0.10), pre-frail (frailty index >0.10 to <0.25) and frail (frailty index ≥0.25). Cox proportional hazards models were used to analyse the association between baseline frailty and risk of diabetes or cardiovascular disease. Restricted cubic splines (RCS) were used for analysis of the non-linear association of baseline frailty index as a continuous variable with risk of diabetes or cardiovascular disease. Association of the developmental trajectories of frailty with the risk of diabetes or cardiovascular disease: Using CHARLS data from 2011 to 2018, the group-based trajectory modelling was combined with survey data from 2011, 2013 and 2015 to determine the developmental trajectories of frailty. Cox proportional hazards models were used to analyse the association between the developmental trajectories of frailty and risk of diabetes or cardiovascular disease in 2018.

Results

1 Risk factors for frailty in middle-aged and older adults: A total of 6852 participants were included in the analysis after exclusion of the presence of frailty at baseline, lack of information on risk factors and incident frailty. During a median follow-up period of 6.92 (IQR 3.92-7.01) years, 1133 frailty events occurred. In all populations, all risk factors were associated with a significantly increased risk of frailty, except for smoking and dyslipidemia. Among them, lower levels of education had the most significant association with the risk of frailty (HR=1.61, 95% CI=1.40-1.83). Among the combined risk factors, socioeconomic status had the strongest association with the risk of frailty (HR = 1.53, 95% CI = 1.41 - 1.67). Results by age group showed that low education was most strongly associated with frailty in adults aged <65 years (HR=1.79, 95% CI= 1.37-2.34 in the age group 45-54 years; HR=1.77, 95% CI=1.44-2.19 in the age group 55-64 years), while unhealthy sleep had the strongest association with frailty in adults aged ≥65 years (HR = 1.43; 95% CI = 1.15 to 1.77). Among the combined risk factors, lower socioeconomic status had the strongest association with frailty risk in all age groups (HR 1.57 [95% CI=1.32-1.87] in age group 45-54 years; HR in age group 55-64 years was 1.74 [95% CI=1.52-2.00]; HR in age group ≥65 years was 1.30 [95% CI=1.14-1.50]).

At the population level, low educational level and rural residence were the main contributors to frailty risk among all middle-aged and older adults, with PAR% of 23.5% and 22.2%, respectively. Combined socioeconomic status is also the factor that contributes most to frailty risk, with a PAR% of 40.8%. Results by age group showed that the contribution of low education and rural residence to frailty decreased with age, both being major risk factors for frailty in adults under 65 years old (the PAR% associated with low education was 21.5% and the PAR% associated with living in a rural area was 18.1% in the age group 45-54 years; the PAR% associated with low education was 30.3% and the PAR% associated with living in a rural area was 31.6% in the age group 55-64 years). However, combined socioeconomic status was still the most important risk factor for frailty in all age groups (PAR% was 38.2%, 51.8% and 28.0% in the 45-54, 55-64, and ≥65 age groups, respectively). The effect of social relationships on frailty in middle-aged and older adults was mainly due to low social activity, and the metabolic risk factors were mainly driven by abdominal obesity. There was also a decrease with age in the risk of frailty due to social relationships and several individual metabolic risk factors (except dyslipidemia). Unhealthy sleep was the main factors of lifestyle on frailty in middle-aged and older people. In contrast to other risk factors, the contribution of unhealthy sleep to frailty gradually increases with age. In adults aged 65 years and over, it was the main cause of frailty, with a PAR% of 21.2%.

2 Association between frailty and risk of diabetes in middle-aged and older adults: Association between frailty at baseline (2011) and risk of diabetes: Of the 13689 participants without diabetes at baseline included in the analysis, 7468 (54.55%) were classified as non-frail, 4954 (36.19%) as pre-frail, and 1267 (9.26%) as frail. During the median follow-up of 4.00 years (IQR 3.60-4.00), 513 diabetes events occurred. Compared with non-frail at baseline, pre-frail and frail at baseline had an increased risk of diabetes, with HRs (95% CI) adjusted for all potential confounders of 1.55 (1.27-1.89) and 2.09 (1.56-2.79), respectively. Restricted cubic spline analyses suggested that a frailty index ≥ 0.09 was associated with an increased risk of diabetes. Association between developmental trajectories of frailty and risk of diabetes: 9461 participants who were free of diabetes during 2011-2015 survey were included in this analysis. Using group-based trajectory modelling and survey data from 2011, 2013 and 2015, three distinct groups of frailty trajectories were identified: low-stable (n=6216, 65.7%), medium-increasing (n=2673, 28.3%), and high-increasing (n=572, 6.0%). During the median follow-up of 3.00 years (IQR 2.92-3.00), 505 diabetes events occurred. Compared with the low-stable group, both the medium-increasing and high-increasing groups had an increased risk of diabetes, the HRs (95% CI) after adjusting for all potential confounders were 1.50 (1.22-1.84) and 1.70 (1.21-2.38), respectively. No effect on the association of baseline or developmental trajectories of frailty with diabetes was observed for traditional diabetes risk factors, including smoking, alcohol consumption, sleep, obesity, hypertension, and dyslipidemia. The above results were robust through a series of sensitivity analyses.

3 Association between frailty and risk of cardiovascular disease in middle-aged and older adults: Association between frailty at baseline (2011) and risk of cardiovascular disease: Of the 12575 participants without cardiovascular disease at baseline included in the analysis, 7255 (57.69%) were classified as non-frail, 4360 (34.67%) as pre-frail, and 960 (7.63%) as frail. During a median follow-up period of 6.92 years (IQR 3.05-7.09), 2004 cardiovascular disease events occurred. Compared with the non-frail population at baseline, those who were pre-frail and frail at baseline had an increased risk of cardiovascular disease, the HRs (95% CI) after adjustment for all potential confounders were 1.45 (1.31-1.60) and 2.05 (1.76-2.38), respectively. Restricted cubic spline analyses suggested that a frailty index ≥ 0.09 was associated with an increased risk of cardiovascular disease. Association between developmental trajectories of frailty and risk of cardiovascular disease: 8331 participants who were free of cardiovascular disease during 2011-2015 were included in this analysis. Using group-based trajectory modelling and survey data from 2011, 2013, and 2015, three distinct groups of frailty trajectories were identified: low-stable (n=5480, 65.8%), medium-increasing (n=2393, 28.7%), and high-increasing (n=458, 5.5%). During a median follow-up of 3.00 years (IQR 2.92-3.00), 1008 cardiovascular disease events were identified. Compared with the low-stable group, the medium-increasing and high-increasing groups had an increased risk of cardiovascular disease, the HRs (95% CI) after adjusting for all potential confounders were 1.85 (1.61-2.13) and 2.42 (1.94-3.02), respectively. No effect on the association of baseline or developmental trajectories of frailty with cardiovascular disease was observed for traditional cardiovascular risk factors, including smoking, alcohol consumption, sleep, obesity, hypertension, diabetes, and dyslipidemia. The above results were robust through a series of sensitivity analyses.

Conclusion

1 Socioeconomic status was the factor which contributed most to the risk of frailty in middle-aged and older adults, with low education and rural residence were the main risk factors. However, the effect of low education and rural residence on the risk of frailty decreased with age, both being the main risk factors for frailty in adults under 65 years of age. In contrast, the effect of unhealthy sleep on the risk of frailty increased with age and was the main risk factor for frailty in people aged 65 years and over. In short, the effect of risk factors on frailty varied by age. Prioritising risk factors that have a greater impact on frailty at different ages will maximise the reduction of the burden of frailty.

2 Both frailty at baseline and the increasing trajectories of frailty over time were positively associated with the risk of diabetes. These results were independent of chronological age and common diabetes risk factors (including smoking, alcohol consumption, sleep, obesity, hypertension, and dyslipidemia), and a series of subgroup and sensitivity analyses also demonstrated the robustness of the findings. Developing frailty screening and interventions for middle-aged and older adults will help reduce the risk of diabetes.

3 Both frailty at baseline and the increasing trajectories of frailty over time were positively associated with the risk of cardiovascular disease in middle-aged and older Chinese populations. These results were independent of chronological age and common cardiovascular disease risk factors (including smoking, alcohol consumption, sleep, obesity, hypertension, diabetes, and dyslipidemia), and a series of subgroup and sensitivity analyses further demonstrated the robustness of the findings. Developing frailty screening and interventions in middle-aged and older adults will help reduce the burden of cardiovascular disease.