Department of Epidemiology and Biostatistics, School of Public Health, Peking University, Beijing, China.
Anning First People's Hospital, Kunming University of Science and Technology, Yunan, China.
Ann Med. 2023;55(2):2285910. doi: 10.1080/07853890.2023.2285910. Epub 2023 Nov 27.
Corona Virus Disease 2019 (COVID-19) has a significant impact on sleep quality. However, the effects on sleep quality in the post-COVID-19 pandemic era remain unclear, and there is a lack of a screening tool for Chinese older adults. This study aimed to understand the prevalence of poor sleep quality and determine sensitive variables to develop an effective prediction model for screening sleep problems during infectious diseases outbreaks.
The Peking University Health Cohort included 10,156 participants enrolled from April to May 2023. The Pittsburgh Sleep Quality Index (PSQI) scale was used to assess sleep quality. The data were randomly divided into a training-testing cohort ( = 7109, 70%) and an independent validation cohort ( = 3027, 30%). Five prediction models with 10-fold cross validation including the Least Absolute Shrinkage and Selection Operator (LASSO), Stochastic Volatility Model (SVM), Random Forest (RF), Artificial Neural Network (ANN), and XGBoost model based on the area under curve (AUC) were used to develop and validate predictors.
The prevalence of poor sleep quality (PSQI >7) was 30.69% (3117/10,156). Among the generated models, the LASSO model outperformed SVM (AUC 0.579), RF (AUC 0.626), ANN (AUC 0.615) and XGBoost (AUC 0.606), with an AUC of 0.7. Finally, a total of 12 variables related to sleep quality were used as parameters in the prediction models. These variables included age, gender, ethnicity, educational level, residence, marital status, history of chronic diseases, SARS-CoV-2 infection, COVID-19 vaccination, social support, depressive symptoms, and cognitive impairment among older adults during the post-COVID-19 pandemic. The nomogram illustrated that depressive symptoms contributed the most to the prediction of poor sleep quality, followed by age and residence.
This nomogram, based on twelve-variable, could potentially serve as a practical and reliable tool for early identification of poor sleep quality among older adults during the post-pandemic period.
2019 年冠状病毒病（COVID-19）对睡眠质量有重大影响。然而，在后 COVID-19 大流行时代，睡眠质量的影响仍不清楚，且缺乏针对中国老年人的筛查工具。本研究旨在了解睡眠质量不佳的流行情况，并确定敏感变量，以开发有效的传染病暴发期间睡眠问题筛查预测模型。
北京大学健康队列纳入了 2023 年 4 月至 5 月期间招募的 10156 名参与者。采用匹兹堡睡眠质量指数（PSQI）量表评估睡眠质量。数据被随机分为训练-测试队列（n=7109，70%）和独立验证队列（n=3027，30%）。采用 10 倍交叉验证的 5 种预测模型（包括最小绝对值收缩和选择算子（LASSO）、随机波动模型（SVM）、随机森林（RF）、人工神经网络（ANN）和基于曲线下面积（AUC）的 XGBoost 模型）来开发和验证预测因子。
睡眠质量差（PSQI>7）的患病率为 30.69%（3117/10156）。在所生成的模型中，LASSO 模型优于 SVM（AUC 0.579）、RF（AUC 0.626）、ANN（AUC 0.615）和 XGBoost（AUC 0.606），AUC 为 0.7。最后，共有 12 个与睡眠质量相关的变量被用作预测模型的参数。这些变量包括老年人在 COVID-19 大流行后的年龄、性别、种族、教育水平、居住地、婚姻状况、慢性病史、SARS-CoV-2 感染、COVID-19 疫苗接种、社会支持、抑郁症状和认知障碍。列线图表明，抑郁症状对预测睡眠质量不佳的贡献最大，其次是年龄和居住地。
该基于 12 个变量的列线图可能成为识别后疫情时期老年人睡眠质量不佳的实用可靠工具。
