基于体成分和体适能指标的动脉粥样硬化性心血管疾病风险评估模型构建和验证

Construction and validation of a risk assessment model for atherosclerotic cardiovascular disease based on body composition and physical fitness indicators

摘要:

目的 基于体成分和体适能指标，研究体检人群动脉粥样硬化性心血管疾病（atherosclerotic cardiovascular disease，ASCVD）发生风险呈中高危的状况，研发ASCVD发生风险的评估工具，为ASCVD的预防和干预提供合理的建议。方法 抽取2020年7月至2022年1月重庆医科大学第二附属医院的1 620名年龄为35~59岁体检人群的体检资料，其中建模集1 200例，验证集420例。根据ASCVD 10年发病危险评估流程彩图计算调查对象的ASCVD发生风险，基于体成分和体适能指标构建ASCVD风险评估模型，继而对模型进行外部验证，并对模型进行区分度、一致性以及临床实用性进行评价。结果 本研究所建立ASCVD发生风险评估模型为Logit（P）=-0.298+2.151×性别+0.046×肝脏脂肪均值-0.016×骨密度均值+0.005×腹部肌肉面积+0.330×腹部脂肪比-0.027×握力体质指数-0.025×平衡力。模型特异度74.9%，灵敏度为81.2%，曲线下面积为0.845（95%CI=0.820~0.865）。结论 本研究所建立的ASCVD发生风险评估模型虽具有较好的区分度和临床实用性，但在不同的数据集中校准度并不是很好，后续有必要扩大样本量进一步对模型进行改善。

Abstract:

Objective To investigate the moderate or high risk of atherosclerotic cardiovascular disease（ASCVD） in the health check-up population based on body composition and physical fitness indicators，to develop a risk assessment tool for ASCVD，and to provide reasonable suggestions for the prevention and intervention of ASCVD.Methods Physical examination data were collected from 1 620 individuals，aged 35-59 years，who underwent physical examination in The Second Affiliated Hospital of Chongqing Medical University from July 2020 to January 2022，among whom 1200 were included in the modeling set and 420 were included in the validation set. The risk of ASCVD in the respondents was calculated according to the color diagram of the 10-year risk assessment process of ASCVD. A risk assessment model for ASCVD was constructed based on body composition and physical fitness indicators，and then internal validation was performed for the model. Finally the model was assessed in terms of discriminatory ability，consistency，and clinical practicability.Results In this study，the risk assessment model for ASCVD was established as Logit（P）=-0.298+2.151×sex+0.046×mean liver fat - 0.016×mean bone mineral density + 0.005×abdominal muscle area +0.330×visceral adipose tissue/subcutaneous adipose tissue ratio -0.027×grip strength values of body mass index -0.025×balance force. This model had a specificity of 74.9%，a sensitivity of 81.2%，and an area under the ROC curve of 0.845（95%CI：0.820-0.865）.Conclusion Although the risk assessment model for ASCVD established in this study has good discriminatory ability and clinical practicability，it lacks a good calibration degree in different datasets，and therefore，it is necessary to expand the sample size in the future to further improve the model.