Mikula-Noble Nataysia, Cormie Vicki, McCowan Rebecca Eilidh, McCowan Colin
Division of Population and Behavioural Sciences, School of Medicine, University of St Andrews, St Andrews, United Kingdom.
School of Medicine, The Chancellor's Building, University of Edinburgh, Edinburgh, United Kingdom.
PLoS One. 2025 Aug 22;20(8):e0329794. doi: 10.1371/journal.pone.0329794. eCollection 2025.
The prevalence of multimorbidity has been growing due to the ageing population and increasingly unhealthy lifestyles. There is interest in identifying clusters of disease and how they are influenced.
This systematic review aims to (i) investigate the most common clusters in the adult population with multimorbidity (ii) identify methods used to define clusters (iii) examine if clusters differ based on age, sex and socioeconomic status.
We searched Medline, Embase, SCOPUS, Web of Science Core Collection, and CINAHL using concepts of multimorbidity and clustering techniques to identify relevant papers. Secondary data, including commonly reported clustering techniques, identified clusters, and other characteristics were extracted. All studies were quality assessed using the Newcastle-Ottawa Bias scale.
From a total of 24,231 papers, 125 were included in the review. There was a total of 918 different clusters identified, which were categorized into 59 broad groups. A cardiometabolic cluster appeared most frequently within the identified studies and across age strata. The most common clustering technique was Latent Class Analysis (n = 51). Disease cluster prevalence appeared to differ based on age, whereas no differences could be identified by sex.
Across the 125 papers identified, irrespective of clustering method, a relatively common set of clusters of disease were found. The Cardiometabolic cluster was the most frequently identified cluster across all age groups. Studies that stratified participants by age or sex identified distinct clusters within each subgroup, which differed from those observed in clusters formed from the general adult population (18+).Latent class analysis was the most common clustering technique within this review, but it was not explored if different clustering methods led to different clusters. Further work is needed to distinguish the most prevalent clusters within specific stratified cohorts of different ages, sex, and socioeconomic status; nonetheless, data strongly suggests that there are different clusters that arise dependent on stratifications. With the expected increasing burden of multimorbidity, healthcare services may need to think about the most prevalent disease combinations within certain strata and how joint-specialist services can be tailored to treat those common conditions.
由于人口老龄化和日益不健康的生活方式，多种疾病并存的患病率一直在上升。人们对识别疾病集群及其影响因素很感兴趣。
本系统评价旨在（i）调查患有多种疾病的成年人群中最常见的集群；（ii）确定用于定义集群的方法；（iii）检查集群是否因年龄、性别和社会经济地位而异。
我们使用多种疾病并存和聚类技术的概念在Medline、Embase、SCOPUS、Web of Science核心合集和CINAHL中进行检索，以识别相关论文。提取了包括常用聚类技术、已识别集群和其他特征在内的二次数据。所有研究均使用纽卡斯尔-渥太华偏倚量表进行质量评估。
在总共24231篇论文中，有125篇被纳入本评价。总共识别出918个不同的集群，这些集群被分为59个宽泛的组。在已识别的研究和不同年龄层中，心脏代谢集群出现得最为频繁。最常用的聚类技术是潜在类别分析（n = 51）。疾病集群患病率似乎因年龄而异，而按性别未发现差异。
在已识别的125篇论文中，无论聚类方法如何，都发现了一组相对常见的疾病集群。心脏代谢集群是所有年龄组中最常被识别出的集群。按年龄或性别对参与者进行分层的研究在每个亚组中识别出了不同的集群，这些集群与在一般成年人群（18岁及以上）中形成的集群不同。潜在类别分析是本评价中最常用的聚类技术，但未探讨不同的聚类方法是否会导致不同的集群。需要进一步开展工作，以区分不同年龄、性别和社会经济地位的特定分层队列中最普遍的集群；尽管如此，数据强烈表明，根据分层会出现不同的集群。随着多种疾病并存负担预计会增加，医疗服务可能需要考虑某些阶层中最普遍的疾病组合，以及如何定制联合专科服务来治疗这些常见病症。
