Missing data and imputation in the international studies following the Harmonized Cognitive Assessment Protocol

Area of Science:

• Gerontology
• Epidemiology
• Cognitive Science

Background:

• Aging research faces significant challenges with missing data, particularly in large-scale, cross-national studies involving older adults with declining health.
• Cultural, linguistic, and educational variations across populations in international studies complicate data harmonization and analysis.
• Effective strategies for managing missing data in global aging and dementia research are not well-documented.

Purpose of the Study:

• To describe and document current practices for harmonizing and handling missing data in international Harmonized Cognitive Assessment Protocol (HCAP) studies.
• To identify challenges and mitigation strategies for missing data in cross-national aging research.
• To provide recommendations for best practices in managing missing data in dementia research.

Main Methods:

• Leveraged data from international Harmonized Cognitive Assessment Protocol (HCAP) studies across six diverse geographic contexts (US, England, Chile, Mexico, India, China).
• Documented methods used for harmonizing and handling missing data, including various missing codes and imputation techniques.
• Described imputation of item-level missingness, missing subinterviews, and dementia classification using chained equations.

Main Results:

• Missing data rates were generally low, ranging from approximately 7% in some countries to around 15% in studies from India and China.
• Item-level missingness and missing subinterviews (neurocognitive assessments and informant reports) were successfully imputed using chained equations.
• Dementia classification was also imputed for relevant individuals within the parent Health and Retirement Study-International Family Studies (HRS-IFS) substudies.

Conclusions:

• Current practices for handling missing data in Harmonized Cognitive Assessment Protocol (HCAP) studies proved effective.
• Developed concrete recommendations for study design, data collection, and post-hoc treatments to mitigate missing data challenges.
• Emphasized the importance of proper use and reporting of imputed data in cross-national aging and dementia research.