Article Text

Download PDFPDF
P52 Calibrating cognitive tests across two Chinese ageing cohorts for dementia prevalence estimation: a confirmatory factor analysis
  1. Yuyang Liu1, 2,
  2. Yanjuan Wu1, 2,
  3. Jingheng Cai3,
  4. Yuntao Chen4,
  5. Tishya Venkatraman4,
  6. Sophia Lobanov-Rostovsky4,
  7. Yung-Jen Yang5,
  8. Yu-Tzu Wu6,
  9. Jing Liao1, 2,
  10. Eric Brunner4,
  11. Yun Huang7,
  12. Piotr Bandosz8,9,
  13. Yuantao Hao10
  1. 1Department of Medical Statistics, School of Public Health, Sun Yat-sen University, Guangzhou, China
  2. 2School of Public Health and Institute of State Governance, Sun Yat-sen University, Guangzhou, China
  3. 3Department of Statistics, Sun Yat-sen University, Guangzhou, China
  4. 4Department of Epidemiology and Public Health, University College London, London, UK
  5. 5Social Research Institute, Institute of Education,University College London, London, UK
  6. 6Population Health Sciences Institute, Newcastle University, Newcastle, UK
  7. 7Guangdong Provincial Center for Disease Control and Prevention, Guangzhou, China
  8. 8Department of Public Health and Policy, University of Liverpool, Liverpool, UK
  9. 9Department of Prevention and Medical Education, Medical University of Gdansk, Gdansk, Poland
  10. 10Peking University Center for Public Health and Epidemic Preparedness and Response, Peking University, Beijing, China


Background There is limited evidence on the nationwide estimates and existing epidemiological surveys generally focused on specific regions that might not be representative of the older population across the whole country. To estimate the nationwide prevalence of dementia using existing data, one of the key challenges is variation in the cognitive measures utilized across cohorts. This study aimed to calibrate cognitive tests into common latent cognitive construct for dementia prevalence estimation across these two population-representative Chinese ageing cohorts.

Methods Chinese Longitudinal Healthy Longevity Survey (CLHLS, N=12015) and China Health and Retirement Longitudinal Study (CHARLS, N=6623) are ongoing nationwide surveys on health and its risk factors of older adults, with different cognitive batteries. Based on participants aged 65–99 from the 2018 wave of CLHLS and CHARLS, confirmatory factor analysis was applied to identify latent cognitive construct and to estimate dementia prevalence. Dementia prevalence estimated by factor scores of the latent approach was compared to raw scores (i.e. the sum of original cognitive test scores), and Mini-Mental State Examination scores. Age-specific dementia prevalence estimated by the CFA models was compared with estimates reported by a nationwide survey, the COAST study and by the Global Burden of Disease (GBD) group.

Results Common three-factor cognitive constructs (orientation, memory, executive function and language) were verified in CLHLS (13 items) and CHARLS (9 items). Dementia prevalence estimated by factor scores was more consistent in CLHLS 5.9% (95%CI:5.7%-6.1%) and CHARLS 4.9% (95%CI: 4.7%-5.1%) than by raw scores and MMSE. Dementia estimates of CLHLS and CHARLS largely overlapped with each other and were in line with these of the cross-sectional COAST survey particularly in the age range between 65 to 80. For estimations above 80 years old, CLHLS and CHARLS estimates turned to be lower than the COAST study, and all lower than the GBD estimates.

Discussion Configural invariance established in this study facilitated the calibrating different cognitive measures across the two ageing cohorts in China and provide a potential approach to estimate dementia prevalence across surveys and over time at the national level. Our latent approach may serve further statistical harmonization of cognitive assessments across aging cohorts over time.

  • Cognitive impairment
  • Dementia
  • Confirmatory factor analysis

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.