Osteosarcopenic obesity syndrome (OSO) is a disease that seriously endangers the health of older people. The rational classification of the disease can guide the clinical diagnosis and treatment. Therefore, classifying OSO based on inter-correlations of its diagnostic variables and exploring its structural properties may offer insights into clinical prevention and treatment of OSO.

To explore the structural properties of OSO, providing a theoretical basis for individualized diagnosis and treatment of the disease.

A cross-sectional study was conducted with a random sample of OSO patients (≥60 years old) who underwent physical examination in Physical Examination Center, the 2nd Affiliated Hospital of Harbin Medical University from January 2018 to December 2020. The data collected include 9 diagnostic variables for OSO〔skeletal muscle index, grip strength, body fat percentage, BMD of the lumbar spine (L1-L4), hip and femoral neck, BMI, waist circumference, walking pace〕, sociodemographic characteristics, lifestyle and prevalence of common chronic diseases. KMO test and Bartlett's test of sphericity were used to evaluate the suitability of diagnostic variables for factor analysis. The components with an eigenvalue equal to or greater than 1.000 were extracted by principal component analysis, and the varimax orthogonal rotation matrix was obtained by the varimax orthogonal rotation method. The common factors were named according to the orthogonal rotation matrix of factors. On the basis of factor analysis, thesum of squares and systematic cluster analysis were used to develop a dendrogram for classifying patients. The structural properties of OSO were analyzed by comparing the values of diagnostic variables and clinical features among patients of different categories.

A total of 107 cases were included. The KMO value (0.688) and the result of Bartlett's test of sphericity (χ²=492.374, P<0.001) indicated that the data of diagnostic variables were suitable for factor analysis. Three common factors (osteoporosis factor, muscle + body fat factor and obesity factor) with an eigenvalue greater than 1.000 were extracted, explaining 81.408% variance of the total. The load value of each diagnostic variable on its common factor ranged from 0.770 to 0.918. The patients were divided into 3 categoriesby cluster analysis using the common factors. The skeletal muscle index, grip strength, body fat percentage, BMD of L1-L4, hip and femoral neck, BMI and waist circumference varied significantly across patients of different categories (P<0.05). The values of BMD of L1-L4, hip and femoral neck of OSO patients in the first category were significantly lower than those of the other two categories (P<0.05). The BMI and waist circumference values of OSO patients in the second category were lower than those of the other two categories (P<0.05). OSO patients in the third category had higher values of skeletal muscle index, grip strength and BMD of L1-L4, hip and femoral neck, but lower body fat percentage than those of the other two categories (P<0.05). There were statistically significant differences in sex ratio, distribution of education level and total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), uric acidand creatinine in the serum among different categories of patients (P<0.05). OSO patients in the first category had higher prevalence of below the undergraduate education level than those in the third category (P<0.017). OSO patients in the second category had higher level of TC than those in the third category (P<0.05). In comparison with those in other two categories, OSO patients in the third category had higher personal monthly income equal to or greater than 5 000 yuan, and lower female ratio (P<0.017). Moreover, OSO patients in the third category also demonstrated higher levels of uric acid and creatinine in the serum (P<0.05) .

OSO diagnostic variables can be generalized and interpreted in terms of osteoporosis, muscle and body fat, and obesity. And OSO patients have different structural properties. The application of multivariate statistical methods to study the structural properties of OSO patients will contribute to the individualized management of such patients.

骨量肌量减少性肥胖综合征（OSO）是一种严重损害老年人健康的疾病，对疾病进行临床分型可为疾病的临床诊治提供指导。基于OSO诊断变量间的相关性对OSO进行分型，并探寻OSO的结构特征，可为OSO的防治提供新的思路。

探索OSO的结构特征，为实现OSO的个体化诊治提供理论依据。

本研究为横断面研究。于2018年1月至2020年10月，采用随机抽样法，选取在哈尔滨医科大学附属第二医院体检中心接受健康体检、年龄≥60岁的老年OSO患者作为研究对象，采集其OSO诊断变量〔四肢骨骼肌指数，握力，体脂百分比（BF%），腰椎1~4（L_1~4）、髋部、股骨颈骨密度（BMD），体质指数（BMI），腰围，步速〕、社会人口学特征、生活方式、常见慢性病患病情况等方面的资料。在利用因子分析法对OSO诊断变量数据进行分析前，采用KMO检验、Bartlett's球形检验评价OSO诊断变量数据是否适合进行因子分析。通过主成分分析法，提取特征值≥1.000的成分，并运用最大方差正交旋转法得出方差最大正交旋转矩阵。根据因子正交旋转矩阵，对公因子进行命名。基于公因子得分，利用离差平方和系统聚类法生成树状结构并对患者进行分类，通过比较不同类别患者间诊断变量水平和临床特征的差异，分析OSO的结构特征。

共纳入107例老年OSO患者。KMO值为0.688，Bartlett's球形检验χ²=492.374，P<0.001，表明OSO诊断变量数据适合进行因子分析；按特征根>1.000的标准可提取3个公因子（骨质疏松因子、肌肉+体脂因子、肥胖因子），3个公因子的累积方差贡献率为81.408%，各诊断变量在所属公因子上的载荷值为0.770~0.918。聚类分析结果显示，共将OSO患者分为3类。不同类别人群四肢骨骼肌指数、握力、BF%、BMD_L1~4、BMD_髋部、BMD_股骨颈、BMI、腰围比较，差异均有统计学意义（P<0.05）。其中第1类人群的BMD_L1~4、BMD_髋部、BMD_股骨颈均低于其他两类人群（P<0.05）；第2类人群的BMI和腰围均低于其他两类人群（P<0.05）；第3类人群的四肢骨骼肌指数、握力和BMD均高于其他两类人群，BF%低于其他两类人群（P<0.05）。不同类别人群性别、受教育程度、个人月收入分布、总胆固醇（TC）、高密度脂蛋白胆固醇（HDL-C）、尿酸（UA）、肌酐（Cr）水平比较，差异有统计学意义（P<0.05）。第1类人群中本科以下学历者占比高于第3类人群（P<0.017）；第2类人群TC水平高于第3类人群（P<0.05）；第3类人群个人月收入≥5 000元者占比，以及UA、Cr水平均高于其他两类人群，女性占比低于其他两类人群（P<0.05或P<0.017）。

可从骨质疏松、肌肉与体脂、肥胖3个方面对OSO诊断变量进行概括和解释；OSO患者具有不同的结构特征。应用多元统计学方法研究OSO患者的结构特征，有助于实现对不同类型OSO患者的个体化管理。