ScholarWorks@성균관대학교

초록

Background: Subjective memory complaints are potential early indicators of Alzheimer's disease (AD), but their diagnostic utility is limited by high prevalence in cognitively normal older adults. The original Subjective Memory Complaints Questionnaire (SMCQ) shows suboptimal specificity and assigns equal weight to all items. Objective: To develop and validate a revised SMCQ (SMCQ-R) with improved diagnostic accuracy through systematic item selection and empirical weighting. Methods: Using a development dataset (252 AD patients, 252 age/sex-matched controls), we: 1) excluded items endorsed by >33% of controls, 2) applied Least Absolute Shrinkage and Selection Operator (LASSO) regression for item selection, and 3) derived integer weights from regression coefficients. Diagnostic performance was compared between SMCQ-R and original SMCQ using receiver operating characteristic (ROC) analysis. Hierarchical regression assessed incremental value with the Mini-Mental Status Examination (MMSE). Results: The 9-item SMCQ-R (score range: 0-30) showed higher AUC than the original SMCQ in the development dataset (0.868 versus 0.811, p < 0.001) and maintained comparable performance in an independent validation dataset. In the validation dataset, SMCQ-R showed modest but significant gains in specificity (0.850 versus 0.813, p = 0.031) and accuracy (0.816 versus 0.789, p = 0.061) with comparable area under the curve (AUC) and sensitivity. Combining SMCQ-R with MMSE improved classification accuracy to 89.5%, compared with 86.0% for MMSE alone. Conclusions: The SMCQ-R refines subjective memory assessment through evidence-based refinement and provides incremental diagnostic value for early symptomatic AD screening.

키워드