Differential Item Functioning and Clinical Utility of the Subjective Memory Complaints Questionnaire in a Multi-Ethnic Cohort.

Abstract

Introduction: This study evaluated psychometric properties of the Subjective Memory Complaints Questionnaire (SMCQ) in a non-Hispanic White (NHW) and Mexican American (MA) sample from Texas in the United States.

Methods: Data were obtained from the Health and Aging Brain Study - Health Disparities (HABS-HD; N=1691, age=66.5±8.7, education=12.4±4.8, 60.6% female, 33.2% MA Spanish-speaking). Unidimensionality of the SMCQ was evaluated with confirmatory factor analysis. Differential item functioning (DIF) of the SMCQ was assessed across age, sex, education, and ethnicity/language using item response theory (IRT) / logistic ordinal regression. Associations of the SMCQ in relation to cognitive status, Alzheimer's disease (AD) blood-based biomarkers, and psychological distress were examined.

Results: The SMCQ showed excellent fit in a single-factor model (CFI=0.97, TLI=0.97, RMSEA [95% CI] = 0.05 [0.04, 0.05], SRMR=0.07). Significant item-level DIF was detected by education level and ethnicity/language, but not by age or sex; when detected, DIF was not salient (i.e., adverse). The SMCQ was associated with greater psychological distress, worse Clinical Dementia Rating scores, and greater disease burden as measured by total tau and neurofilament light.

Conclusions: Practically negligible item-level bias was identified across education and ethnicity/language. Detected DIF can be described as benign, indicating that some items manifested differently between groups but had minimal impact on measurement properties. These results demonstrate that the SMCQ performs appropriately across demographic variables. Our findings also provide support for the associations of SMCQ scores with self-reported mood, cognitive status, and AD blood-based biomarkers.