An approach to understand osteoarthritis, comorbidity and excess mortality in the population.
An approach to understand osteoarthritis, comorbidity and excess mortality in the population.
复制标题
一种了解人群中骨关节炎、合并症和超额死亡率的方法。
DOI：
10.1093/rheumatology/kead172
复制
发表时间：
2023
期刊：
影响因子：
--
通讯作者：
Golightly,YvonneM
中科院分区：
文献类型：
--
作者：
Nelson,AmandaE;Golightly,YvonneM
正如2016年国际骨关节炎研究协会（OARSI）白色白皮书所指出的，骨关节炎是一种严重疾病[1]。这部分是由于残疾和疼痛的严重负担，沿着这种缺乏有效治疗的疾病的进行性。OA的严重性在很大程度上也是由于伴随OA的多种慢性疾病的高频率，以及管理这些多种伴随问题所带来的挑战。这种慢性疾病的合并发生可能是由于疾病的高患病率（例如OA和糖尿病）、共同的风险因素（例如肥胖和年龄）和/或直接的因果关系（例如慢性疼痛和抑郁）[2]。最重要的是，OA会通过对疼痛、功能、生活质量和治疗决策的影响，加重个体的健康负担[2]。在这种复杂的背景下，确定这些慢性疾病的分组以及这些疾病的各种组合如何不同地影响重要结果（如死亡率）可能具有重要意义。为此，Pineda-Moncusi等人在新诊断为OA的个体中确定了合并症的集群以及这些集群与死亡率的关联[3]。作者在西班牙使用了一个大型基于人群的初级保健数据库，其中包括研究期间（2006- 2020年）首次诊断为OA的> 60万人。他们利用聚类技术来提供对合并症分组的理解，并检查这些分组与外部变量（聚类算法中未包括的关键特征）和死亡率之间的关系。重要的是，他们使用两种不同的聚类方法（K均值和潜在类分析）提供完整的结果，沿着替代解决方案（聚类数），提高了方法和结果的透明度。K-means是一种数据驱动的方法，旨在通过最小化从每个观察到指定聚类的质心的距离来识别聚类。相比之下，潜在类别分析（LCA）是一种有限混合模型，它为每个观察提供属于一个或多个聚类的概率。生命周期评价方法更灵活，可以更好地反映医学数据的异质性和复杂性。根据具体情况，这些方法可能会产生相似或截然不同的结果，
As noted in the OA Research Society International (OARSI) White Paper from 2016, OA is a serious disease [1]. This is in part due to the significant burden of disability and pain, along with the progressive nature of this condition which lacks effective therapies. The serious nature of OA is also largely due to the high frequency of multiple chronic conditions accompanying OA, and the challenges presented by managing these multiple concomitant issues. Such co-occurrence of chronic conditions can be due to the high prevalence of the conditions (eg OA and diabetes), shared risk factors (eg obesity and age) and/or direct causal relationships (eg chronic pain and depression)[2]. Most importantly, OA can exacerbate the health burden on an individual through its impact on pain, function, quality of life and therapeutic decisions [2]. Against this complex backdrop, it may be of great relevance to identify groupings of such chronic conditions and how various combinations of these may differentially impact important outcomes, such as mortality.To this end, Pineda-Moncusi et al. identified clusters of comorbid conditions among individuals newly diagnosed with OA and the associations of these clusters with mortality [3]. The authors employed a large population-based primary care database in Spain which included> 600 000 individuals with a first diagnosis of OA in the study period (2006–20). They utilized clustering techniques to provide an understanding of groupings of comorbidities and to examine relationships between these grouping and external variables (key characteristics not included in the clustering algorithm) and mortality. Importantly, they present full results using two different approaches to clustering (K-means and latent class analysis) along with alternate solutions (cluster number) improving the transparency of their approach and results. K-means is a data-driven method that aims to identify clusters by minimizing the distance from each observation to the centroid of the assigned cluster. In contrast, latent class analysis (LCA) is a finite mixture model that provides a probability of belonging to one or more clusters for each observation. LCA is more flexible and may better represent the heterogeneity and complexity of medical data. Depending on the situation, these methods can produce similar or vastly different results, and there is a