Cross-national harmonization of cognitive measures across HRS HCAP (USA) and LASI-DAD (India)

Abstract

Background: As global populations age, cross-national comparisons of cognitive health and dementia risk are increasingly valuable. It remains unclear, however, whether country-level differences in cognitive function are attributable to population differences or bias due to incommensurate measurement. To demonstrate an effective method for cross-national comparison studies, we aimed to statistically harmonize measures of episodic memory and language function across two population-based cohorts of older adults in the United States (HRS HCAP) and India (LASI-DAD).

Methods: Data for 3,496 HRS HCAP (≥65 years) and 3,152 LASI-DAD (≥60 years) participants were statistically harmonized for episodic memory and language performance using confirmatory factor analysis (CFA) methods. Episodic memory and language factor variables were investigated for differential item functioning (DIF) and precision.

Results: CFA models estimating episodic memory and language domains based on a priori adjudication of comparable items fit the data well. DIF analyses revealed that four out of ten episodic memory items and five out of twelve language items measured the underlying construct comparably across samples. DIF-modified episodic memory and language factor scores showed comparable patterns of precision across the range of the latent trait for each sample.

Conclusions: Harmonization of cognitive measures will facilitate future investigation of cross-national differences in cognitive performance and differential effects of risk factors, policies, and treatments, reducing study-level measurement and administrative influences. As international aging studies become more widely available, advanced statistical methods such as those described in this study will become increasingly central to making universal generalizations and drawing valid conclusions about cognitive aging of the global population.

Fig 1. Differential Item Functioning (DIF) impact; the boxplots represent the difference scores of initial and DIF-modified scores per domain and the vertical lines represent 1x the standard error of measurement of the sample, i.e., the threshold for salient DIF.

Fig 2. Information curves for the Differential Item Functioning (DIF)-modified episodic memory and language factors (reliability = 1–1/information) (upper panel).

The histograms are the population distribution on the latent trait (lower panel). With mostly continuous factor indicators for the episodic memory latent trait, reliability is constant over the range of theta (ability). With mostly categorical indicators for the language latent trait, reliability varies over the range of theta, as shown by a peak where most of the item difficulty parameters are.