研究背景：

慢性阻塞性肺疾病（简称“慢阻肺病”）是我国发病率高、疾病负担重的主要慢性病之一，常合并心血管疾病、高血压、糖尿病等多种共病，呈现出高度的多病共存特征。这些共病不仅加重疾病管理难度，也显著影响患者的生命质量和临床结局。作为典型的共病高度聚集性疾病，慢阻肺病兼具代表性、复杂性与干预潜力，是当前慢病管理从“单病视角”迈向“多病共管”范式转型的重要研究切入点。随着其被纳入国家基本公共卫生服务体系，社区逐渐成为慢阻肺病早筛、随访与长期健康管理的核心平台，为推进基层共病管理模式的实践探索提供了政策基础。

然而，当前我国在该领域仍存在诸多研究空白。已有研究多来源于国外数据，缺乏基于中国本土、全国性样本的共病流行特征与健康结局研究。面对慢阻肺病患者共病结构的高度异质性，传统分层方法难以支撑个性化管理，国内亦缺少基于共病特征的系统分型与分群方法。同时，社区在实际共病管理中面临多重挑战，如资源不足、信息割裂、协作机制缺失等，但相关基层视角下的定性研究较为稀缺。为此，本研究从慢阻病共病流行特征识别、聚类分型分析与社区管理优化三方面展开系统探讨，力图为我国多病共管体系建设提供数据支持与政策参考。

研究目的：

本研究以慢阻肺病共病为核心领域，构建“量化共病特征—识别患者聚类—优化共病管理”三位一体的研究路径，力求从流行特征、结构分型到服务实践三个层面，系统回应当前学术与实践中的不足。具体而言，第一，通过基于全国多中心数据的流行病学分析，识别慢阻肺病患者常见共病及其对健康结局的影响；第二，以共病特征为主进行无监督聚类，识别慢阻肺病患者群体的异质性分布与差异化特征；第三，从社区实践出发，采用定性方法识别基层共病管理中的关键阻碍与改进方向，最终为推进多病共管的本土化实施提供理论支持与策略建议。

研究方法：

本研究分别从慢阻肺病共病流行特征识别、患者聚类分析与社区疾病管理实践探索三个层面展开。前两部分为定量研究，基于“幸福呼吸”项目中2020至2023年全国范围内收集的多中心横断面调查数据。

第一部分量化共病，用描述性统计分析人口学、临床和生活方式变量的分布特征，组间差异通过Pearson卡方检验和Wilcoxon秩和检验评估。共病数量的影响因素使用泊松回归建模，进一步采用逻辑回归和泊松回归评估不同共病类型与慢阻肺急性加重发生风险和加重频率之间的关联，健康相关生命质量通过五水平五维健康量表测量（EuroQol 5-Dimension 5-Level version，EQ-5D-5L），并使用普通最小二乘回归评估其与共病的统计关联性。所有模型均调整混杂变量，并进行多重共线性诊断。

第二部分聚类研究，首先对纳入的27种共病和4项基本人口学特征（性别、年龄、BMI、吸烟频率）在内的31个变量应用多重对应分析进行变量降维，提取3个主成分，累计解释了数据的主要变异性。随后，采用K-means++与层次聚类两种无监督机器学习算法开展聚类分析，通过肘部法与轮廓系数法综合评估聚类效果，最终确定最优聚类数为4。K-means++聚类的平均轮廓系数为0.42，优于层次聚类（0.21），聚类结构明确。K-means++聚类结果经随机森林模型训练与100次重复验证，获得较高的一致性（平均调整兰德指数为 0.707， 95% CI: 0.478–0.936），验证了聚类的稳定性与可靠性。最后，构建多变量Logistic回归模型，评估不同聚类类型与EQ-5D五个维度生命质量结局之间的关联，调整相关社会人口学与临床混杂因素。

第三部分定性研究，以多重慢病战略框架为理论基础，在北京市城区社区卫生服务中心（站）采用目的性抽样策略，选取具有慢阻管理经验的医务人员和慢阻肺病共病患者进行半结构式一对一深度访谈，所有访谈经录音、转录后导入Nvivo 14.0软件，运用主题框架分析法进行系统性资料整理与编码，并结合理论模型逐步提炼社区慢阻肺病共病管理中的现实阻碍与优化建议等关键主题。

研究结果：

第一步部分量化共病研究共纳入11,145名慢阻肺病患者，以年龄≥70岁（占56.4%）、男性（76.9%）和农村居民（70.5%）为主，整体症状负担较重（CAT ≥10者占83.4%），平均共病数量为1.09种。59.4%的患者合并至少一种慢性疾病，泊松回归分析显示，共病数量随年龄增长、症状评分加重、BMI升高和环境暴露等因素显著增加。最常见的共病类型包括慢性支气管炎（32.1%）、高血压（17.8%）和肺气肿（17.8%）。女性患者在多种共病中的患病率显著高于男性，包括慢性支气管炎（女性34.9%，男性31.2%，P<0.001）、哮喘（女性8.7%，男性4.6%，P<0.001）、支气管扩张（女性5.2%，男性3.2%，P<0.001）、糖尿病（女性4.1%，男性3.1%，P=0.020）及缺血性心脏病（女性9.1%，男性7.5%，P=0.010）；而在肿瘤方面，男性患病率为1.53%，显著高于女性的0.58%（P<0.001）。在共病与健康结局的关联分析中发现，部分共病类型与慢阻肺病急性加重（Acute Exacerbation of Chronic Obstructive Pulmonary Disease，AECOPD）的发生风险和加重频率均呈显著正向统计关联，相关疾病主要分布于心血管系统（如高血压、缺血性心脏病、心力衰竭）、呼吸系统（如肺源性心脏病、慢性支气管炎、肺气肿、哮喘、支气管扩张、气胸、呼吸衰竭和肺炎）以及代谢与精神类疾病（如糖尿病、精神障碍）。逻辑回归结果显示，肺源性心脏病（OR=3.78，95%CI：2.73–5.24，P<0.001）、哮喘（OR=2.44，95%CI：2.03–2.94，P<0.001）和慢性支气管炎（OR=2.45，95%CI：2.23–2.68，P<0.001）等共病与AECOPD发生风险显著相关；而泊松回归分析进一步指出，这些共病亦与AECOPD年加重次数具有统计学上的正相关关系（如哮喘：RR=1.66，P<0.001）。此外，普通最小二乘回归模型结果表明，部分共病对EQ效用值具有统计学意义上的不良影响（p < 0.05），主要涉及心脑血管系统疾病（如缺血性心脏病、肺源性心脏病、中风、心力衰竭）、呼吸系统疾病（如肺气肿、哮喘、慢性支气管炎）及其他系统疾病（如前列腺增生、脊椎病）。哮喘（β =–0.03，P<0.001）、中风（β =–0.06，P<0.001）和肺源性心脏病（β =–0.02，P<0.001）与EQ-5D效用值显著负相关，提示其在统计上与患者生命质量下降相关。

第二部分聚类分析识别出四类具有代表性的慢阻肺患者亚群：年轻男性吸烟群体（N=5,453），男性占比98.3%，吸烟率高达87.6%，共病数量最少（0.38），CCI评分最低（3.47），EQ-5D效用值最高（0.74）；女性生物质燃料暴露群体（N=2,483），女性占比86.5%，生物质暴露率达50%，尽管共病数量中等（均值0.94），但EQ-5D效用值偏低（0.69）；呼吸系统共病群体（N=1,819），合并慢性支气管炎和肺气肿分别达83%和78%，共病数量最多（2.63），AECOPD发生率高（64.3%），EQ-5D效用值降至0.66；老年多重共病群体（N=1,390），70岁及以上占比最高（83.2%），合并高血压（68%）、缺血性心脏病（48%）等，CCI评分最高（4.97），EQ-5D效用值最低（0.65）。Logistic回归显示，相较于年轻男性吸烟群体，呼吸系统共病群体在EQ-5D五个维度中健康功能受损风险最高（如日常活动受限OR=1.97，95%CI：1.73–2.25），老年多共病群体在行动能力（OR=1.71）和疼痛不适（OR=1.68）方面亦显著不利。女性生物质暴露群体则在自理能力（OR=1.65）和行动能力（OR=1.62）方面存在较高风险。

第三部分定性研究共纳入33名受访者，包含19名具备慢病管理经验的社区医务人员（涵盖全科医生、护士和药师）和14名慢阻肺病共病患者，对其中慢阻肺病共病管理的实践障碍与应对策略进行了系统梳理。结果显示，当前社区共病管理面临多维度挑战。在卫生体系层面，存在政策支持不足、单病导向制度限制、多病协同机制缺失以及干预手段单一等结构性障碍；在患者层面，多重用药依从性差与健康素养不足共同限制了其自我管理能力；在工具与信息方面，信息系统割裂与专业培训匮乏导致管理效率与照护能力受限；在知识支撑方面，基层共病研究基础薄弱，缺乏标准化管理指南与流程。针对上述问题，研究归纳对应的优化路径建议，包括加强政策引导以提升基层慢阻肺病管理能力、推动全专协作机制建设、补足资源与能力配置、强化家庭医生团队与个体化健康教育、整合信息系统以支撑共病协同决策、以及制定基于证据的共病管理路径与干预策略体系。这些策略有望为基层医疗机构实现慢阻肺病共病的整合式照护提供实践依据与政策启发。

研究结论：

本研究基于全国多中心数据，系统识别了慢阻肺病患者的共病特征及其对急性加重和健康相关生命质量的影响，发现高血压、慢性支气管炎、肺气肿等为我国慢阻肺病人群中最常见的共病类型，且部分心血管、呼吸系统及代谢类共病对生命质量产生不良影响，急性加重发生频率显著相关。通过无监督聚类，划分出四类具有差异化共病特征和健康结局的慢阻肺病患者亚群，明确了呼吸系统共病群体与老年多重共病群体在生命质量受损和功能受限方面的高风险特征，提示人群分型有助于指导社区分层管理与个体化干预。定性访谈进一步揭示了当前社区共病管理中存在的制度、资源、协作及信息等多方面障碍，并提出针对性优化路径，包括全专协作机制建设、家庭医生团队完善、信息系统整合、个体化健康教育以及共病管理路径的制定等建议。综上，本研究从慢阻肺病共病特征、人群聚类到基层实践优化，提供了本土证据与可行性建议，为推动慢阻肺病共病的综合管理与我国多病共管体系建设提供了理论支持和实践参考。

Background：

Chronic obstructive pulmonary disease (COPD) is one of the major chronic diseases in China with high prevalence and heavy disease burden. It is frequently accompanied by multiple comorbidities such as cardiovascular diseases, hypertension, diabetes, presenting a high degree of multimorbidity. These comorbidities not only increase the difficulty of disease management but also significantly affect patients’ quality of life and clinical outcomes. As a typical disease with a high concentration of comorbidities, COPD is representative, complex, and has strong potential for intervention. It serves as a critical entry point for shifting chronic disease management from a single-disease perspective to a multimorbidity management paradigm. With its inclusion in the national basic public health service system, the community has gradually become the core platform for early screening, follow-up, and long-term health management of COPD. This provides a policy foundation for exploring comorbidity management models at the primary care level. 

However, there remain many research gaps in this field in China. Existing studies are mostly based on data from other countries and lack research on the epidemiological characteristics and health outcomes of comorbidities using domestic, national-level samples from China. Given the high heterogeneity of comorbidity structures among COPD patients, traditional stratification methods are insufficient to support personalized management. Moreover, there is a lack of systematic classification and clustering methods based on comorbidity characteristics in China. Meanwhile, community-level comorbidity management faces multiple challenges such as insufficient resources, fragmented information, and lack of collaborative mechanisms. Qualitative research from the primary care perspective remains scarce. Therefore, this study systematically explores three aspects: the identification of COPD comorbidity epidemiological patterns, clustering analysis, and optimization of community-based management, aiming to provide data support and policy references for multimorbidity management in China.

Objective: This study focuses on COPD comorbidities and establishes a integrated research pathway of “quantifying comorbidity characteristics—identifying patient clusters—optimizing comorbidity management.” It aims to systematically address current academic and practical gaps from three levels: epidemiological characteristics, structural classification, and service implementation. Specifically, first, it identifies common comorbidities among COPD patients and their impact on health outcomes through nationwide multi-center epidemiological analysis. Second, it uses unsupervised clustering based on comorbidity characteristics to reveal the heterogeneous distribution and differentiated features of COPD patient groups. Third, starting from community practices, it adopts qualitative methods to identify key barriers and improvement directions in primary care comorbidity management. Ultimately, the study aims to provide theoretical support and strategic recommendations for the localized implementation of multimorbidity management.

Methods: 

This study is conducted on three levels: identifying COPD comorbidity epidemiology, patient clustering analysis, and community chronic disease management exploration. The first two parts are quantitative, based on cross-sectional data collected nationwide from 2020 to 2023 under the “Happy Breathing” program.

In the first part, comorbidities are quantified through descriptive statistics of demographic, clinical, and lifestyle variables. Group differences are evaluated using Pearson's chi-square tests and Wilcoxon rank-sum tests. Poisson regression models assess influencing factors of comorbidity count. Logistic and Poisson regression analyses were further applied to assess associations between comorbidity types and the risk and frequency of acute exacerbations of COPD (AECOPD). Health-related quality of life (HRQoL) is measured by EQ-5D-5L utility values, and associations with comorbidities are assessed using ordinary least squares (OLS) regression. All models adjust for confounders and conduct multicollinearity diagnostics (VIF < 10).

In the second part, multiple correspondence analysis was applied to reduce the dimensionality of 31 variables, including 27 comorbidities and 4 basic demographic features (sex, age, BMI, smoking frequency). Three principal components were extracted, cumulatively explaining most of the data variance. K-means++ and hierarchical clustering algorithms were then applied for unsupervised clustering. The elbow method and silhouette coefficient were used to determine the optimal number of clusters, which was finally set at four. The mean silhouette coefficient of K-means++ clustering was 0.42, which was better than hierarchical clustering (0.21), indicating a clearer structure. The K-means clustering results were trained using a random forest model with 100 repetitions, achieving high consistency (mean adjusted Rand index ARI = 0.707, 95% CI: 0.478–0.936). This validated the stability and reliability of the clustering. Finally, multivariable logistic regression models were constructed to assess the associations between different cluster types and EQ-5D five-dimension health outcomes, adjusting for relevant sociodemographic and clinical confounders.

In the third part, guided by a multiple chronic conditions strategic framework, purposive sampling was conducted at community health service centers (stations) in urban Beijing. Medical staff experienced in chronic disease management and COPD patients with comorbidities were selected for semi-structured in-depth one-on-one interviews. All interviews were audio-recorded, transcribed, and imported into Nvivo 14.0 for systematic data organization and coding using thematic framework analysis. Key themes related to practical barriers and optimization suggestions in community COPD comorbidity management were progressively refined in combination with theoretical models.

Results: 

In the first part, a total of 11,145 COPD patients were included. The population was predominantly aged ≥70 years (56.4%), male (76.9%), and rural residents (70.5%). The overall symptom burden was high, with 83.4% of patients having a CAT score ≥10.

The mean number of comorbidities was 1.09. A total of 59.4% of patients had at least one chronic disease. Poisson regression analysis showed that comorbidity counts significantly increased with age, higher symptom scores, elevated BMI, and environmental exposures. The most common comorbidities were chronic bronchitis (32.1%), hypertension (17.8%), and emphysema (17.8%). The prevalence of several comorbidities was significantly higher in female patients than in male patients, including chronic bronchitis (34.9% in females vs. 31.2% in males, P<0.001), asthma (8.7% in females, 4.6% in males, P<0.001), bronchiectasis (5.2% in females, 3.2% in males, P<0.001), diabetes (4.1% in females, 3.1% in males, P=0.020), and ischemic heart disease (9.1% in females, 7.5% in males, P=0.010). For tumors, the prevalence in males (1.53%) was significantly higher than in females (0.58%) (P<0.001). The analysis of comorbidities and health outcomes showed that certain comorbidities were significantly positively associated with the risk and frequency of AECOPD. The associated diseases were mainly from the cardiovascular system (e.g., hypertension, ischemic heart disease, heart failure), respiratory system (e.g., cor pulmonale, chronic bronchitis, emphysema, asthma, bronchiectasis, pneumothorax, respiratory failure, and pneumonia), as well as metabolic and mental disorders (e.g., diabetes, mental disorders). Logistic regression results indicated that comorbidities such as cor pulmonale (OR=3.78, 95%CI: 2.73–5.24, P<0.001), asthma (OR=2.44, 95%CI: 2.03–2.94, P<0.001), and chronic bronchitis (OR=2.45, 95%CI: 2.23–2.68, P<0.001) were significantly associated with the risk of AECOPD. Poisson regression further showed that these comorbidities were also positively associated with the annual number of AECOPD events (e.g., asthma: RR=1.66, P<0.001). In addition, ordinary least squares regression indicated that some comorbidities had significant adverse effects on EQ-5D utility scores (p<0.05). The main comorbidities involved cardiovascular diseases (e.g., ischemic heart disease, cor pulmonale, stroke, heart failure), respiratory diseases (e.g., emphysema, asthma, chronic bronchitis), and other system diseases (e.g., prostatic hyperplasia, spondylosis). For example, asthma (β =–0.03, P<0.001), stroke (β =–0.06, P<0.001), and cor pulmonale (β =–0.02, P<0.001) were significantly negatively associated with EQ-5D utility scores, indicating their statistical correlation with reduced quality of life.

In the second part, clustering analysis identified four representative COPD clusters.

The first cluster was young male smokers (N=5,453), with 98.3% males and a smoking rate of 87.6%. They had the fewest comorbidities (0.38), the lowest CCI score (3.47), and the highest EQ-5D utility score (0.74). The second cluster was females exposed to biomass fuel (N=2,483), with 86.5% females and a biomass exposure rate of 50%. Although they had a moderate number of comorbidities (mean 0.94), their EQ-5D utility scores were relatively low (0.69). The third cluster was the respiratory comorbidity cluster (N=1,819), with high rates of chronic bronchitis (83%) and emphysema (78%). They had the most comorbidities (2.63), high AECOPD rates (64.3%), and low EQ-5D utility scores (0.66). The fourth cluster was the elderly with multiple comorbidities (N=1,390), with 83.2% aged 70 or older. They had a high prevalence of hypertension (68%) and ischemic heart disease (48%), the highest CCI score (4.97), and the lowest EQ-5D utility score (0.65). Logistic regression showed that compared with young male smokers, the respiratory comorbidity cluster had the highest risk of functional impairment across EQ-5D dimensions (e.g., limitation in usual activities OR=1.97, 95%CI: 1.73–2.25). The elderly multimorbidity cluster also showed significantly higher risks in mobility (OR=1.71) and pain/discomfort (OR=1.68). The female biomass exposure cluster had higher risks in self-care (OR=1.65) and mobility (OR=1.62).

In the third part, the qualitative study included 33 respondents, comprising 19 community health professionals with chronic disease management experience (including general practitioners, nurses, and pharmacists) and 14 COPD patients with comorbidities. The study systematically summarized the practical barriers and coping strategies in COPD comorbidity management. The results showed that community comorbidity management currently faces multi-dimensional challenges. At the health system level, structural barriers included insufficient policy support, single-disease-oriented management, lack of multimorbidity coordination mechanisms, and limited intervention tools. At the patient level, poor medication adherence and low health literacy jointly limited self-management capacity. Regarding tools and information, fragmented information systems and insufficient professional training restricted management efficiency and care capacity. In terms of knowledge support, the foundation for primary care comorbidity research was weak, and standardized management guidelines and protocols were lacking. In response to these issues, the study proposed corresponding optimization pathways, including strengthening policy guidance to enhance COPD management capacity at the primary level, building collaborative mechanisms between general practitioners and specialists, improving resource and capacity allocation, enhancing family doctor teams and personalized health education, integrating information systems to support coordinated comorbidity decision-making, and developing evidence-based comorbidity management pathways and intervention strategies. These strategies are expected to provide practical and policy insights for implementing integrated COPD comorbidity care in primary care settings.

Conclusions: 

This study, based on national multi-center data, systematically identified the comorbidity characteristics of COPD patients and their associations with AECOPD and HRQoL. Hypertension, chronic bronchitis, and emphysema were found to be the most common comorbidities among COPD patients in China. Some cardiovascular, respiratory, and metabolic comorbidities were shown to have adverse effects on quality of life and were significantly associated with the frequency of acute exacerbations. Unsupervised clustering identified four distinct COPD subgroups with differentiated comorbidity profiles and health outcomes. The respiratory comorbidity group and the elderly multimorbidity group were found to have higher risks of impaired quality of life and functional limitations. This suggests that population stratification may help guide community-based hierarchical management and individualized interventions. Qualitative interviews further revealed that current community comorbidity management faces barriers at the system, resource, collaboration, and information levels.
Targeted optimization strategies were proposed, including the development of collaborative mechanisms between generalists and specialists, strengthening family doctor teams, integrating information systems, promoting personalized health education, and formulating comorbidity management pathways. In conclusion, this study provides local evidence and feasible recommendations from comorbidity characteristics, population clustering, to community practice optimization. It offers theoretical support and practical references for advancing integrated COPD comorbidity management and building a multimorbidity management system in China.