Annals of Medicine (Dec 2023)

Development and validation of prediction models for poor sleep quality among older adults in the post-COVID-19 pandemic era

  • Min Du,
  • Manchang Li,
  • Xuejun Yu,
  • Shiping Wang,
  • Yaping Wang,
  • Wenxin Yan,
  • Qiao Liu,
  • Min Liu,
  • Jue Liu

DOI
https://doi.org/10.1080/07853890.2023.2285910
Journal volume & issue
Vol. 55, no. 2

Abstract

Read online

AbstractBackground 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.Materials and Methods 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 (n = 7109, 70%) and an independent validation cohort (n = 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.Results 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.Conclusions 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.

Keywords