Representability in longitudinal aging studies is biased by cognitive functioning
Longitudinal epidemiological studies on aging track the same participants over time to assess various topics, including dementia risk factors and consequences of dementia. A key concern is the generalisability of results, which can be biased by study attrition – when study participants drop out of the study. Previous research has shown that participants who drop out tend to have lower cognitive functioning compared to those who remain. However, the results are inconsistent regarding which specific areas of cognitive ability differ between these groups.
This paper examined the differences in performance across different domains of cognitive functioning between participants who dropped out of studies and those who remained. Data from 10 epidemiological longitudinal studies on aging, which included 209,518 persons from several Western countries, were used. These included two NEAR-studies: Swedish Adoption Twin Study of Aging (SATSA) and the Swedish section of the Survey of Health, Ageing and Retirement in Europe (SHARE-Sweden).
Photo: Marjon Besteman from Pixabay
The analyses were carried out by comparing, for each pair of consecutive study waves, the difference in cognitive scores between participants who dropped out at the next study wave and those who remained.
Results showed that compared with those who remained in the survey, people who dropped out had significantly lower levels of cognitive functioning in all the analysed domains of cognitive functioning. Consequently, differences in cognitive abilities might cause selection bias in long-term aging studies, making it hard to draw accurate conclusions. More research is needed to understand how cognitive impairment and related functional limitations lead to participants dropping out of these studies.
Publication
Hernandez, R., Jin, H., Lee, P. J., Schneider, S., Junghaenel, D. U., Stone, A. A., … & Zelinski, E. M. (2024). Attrition from longitudinal ageing studies and performance across domains of cognitive functioning: an individual participant data meta-analysis. BMJ open, 14(3), e079241. https://doi.org/10.1136/bmjopen-2023-079241
