Cross-national neuropsychological research is needed to understand the social, economic, and cultural factors associated with cognitive risk and resilience across global aging populations. Memory and language have been shown to be sensitive to age-related cognitive decline and pathological cognitive aging processes and may be more sensitive to subtle cognitive decline than measures of global cognitive function. Thus, we aimed to derive and validate harmonized cognitive domain scores for memory and language across population-based studies in the US and Mexico.

Data came from the Health and Retirement Study (HRS) Harmonized Cognitive Assessment Protocol (HCAP) and the Mexican Health and Aging Study (MHAS) Ancillary Study on Cognitive Aging (Mex-Cog). We used confirmatory factor analysis methodology to create statistically co-calibrated cognitive domains of memory and language. We performed differential item functioning (DIF) analysis to evaluate measurement differences across studies, using a cultural neuropsychological approach to identify comparable items across studies (i.e., cross-study anchors). We evaluated harmonized scores by examining their relationship to age and education in each study.

We included 3347 participants from the HRS-HCAP study [Mage=76.6(7.5), 60% female] and 2042 participants from the Mex-Cog study [Mage=68.1(9.0), 59% female]. Education was classified according to the International Standard Classification of Education in the following categories (HRS-HCAP and Mex-Cog, respectively): none or early childhood education: (0.7%; 50.5%), primary education (4.1%; 22.3%), lower secondary education (7.1%; 15.7%), upper secondary education (41.1%; 3.0%), and any college (47.1%; 8.5%). DIF analyses revealed that 5 out of the 7 memory items and 1 out of the 12 language items demonstrated statistical evidence of measurement differences across studies, meaning that these items measured each underlying cognitive construct differently across studies. After adjusting for DIF by not allowing the items with DIF to be cross-study anchors, harmonized memory and language scores showed generally the expected associations with age and education in each study. Increasing age was associated with lower memory (r=-0.40 in HRS-HCAP; r=-0.44 in Mex-Cog) and language (r=-0.31 in HRS-HCAP and r=-0.67 in Mex-Cog) scores. Increasing years of education was associated with better memory and language scores, with mean scores ranging from z=-0.86 and z=-0.29 among those with a primary education or lower to z=0.33 and z=0.90 among those with any college, for HRS-HCAP and Mex-Cog, respectively.

A cultural neuropsychology approach to statistical harmonization facilitates the generation of harmonized measures of cognitive functioning in cross-national studies. Future work can utilize these harmonized cognitive scores to investigate determinants of late-life cognitive decline and dementia in the US and Mexico.