Cognitive Domains in the National Social Life, Health, and Aging Project

Cognition consists of specific domains that are differentially linked to health outcomes. We provide guidance on how to derive cognitive domains in the National Social Life, Health, and Aging Project (NSHAP) study. We suggest the use of a bifactor analysis to derive cognitive domains. To support our view that a bifactor analysis is necessary, we created cognitive domains from a bifactor analysis and created cognitive domains through summing the cognitive items; we then regressed the cognitive domains created through both methods with functional health (activites of daily living [ADLs] and instrumental activities of daily living [IADLs]) and compared the regression results.

NSHAP is a nationally representative longitudinal study of U.S. older adults that started in 2005. Data from Round 3 (2015-2016) were used. The MoCA-SA contains 18 cognitive items to represent 6 cognitive domains: modified trail-making test-B, clock drawing test (3 items), rhinoceros naming, digit span (2 items), serial 7 subtractions, sentence repetition, phonemic fluency, abstraction, delayed recall (5 items), and orientation (2 items). We created cognitive domains through a bifactor analysis and through summing up the cognitive items. We used linear regression to examine how global cognition and each cognitive domain derived from both methods were associated with ADLs and IADLs. Analyses were restricted to respondents aged 50+ without dementia.

The global cognition score derived from both methods was statistically significantly associated with ADLs and with IADLs. All but the memory domain constructed from summing the cognitive items were associated with IADLs; only the visuospatial domain was associated with ADLs. None of the domains derived from a bifactor analysis were associated with ADLs or IADLs.

Researchers should derive cognitive domains using a bifactor analysis to reduce spurious associations.