Predictive validity and diagnostic stability of mild cognitive impairment subtypes

Abstract Background Mild cognitive impairment (MCI) is subclassified into four subtypes by the presence of impairment in the memory domain (amnestic vs nonamnestic) and the number of impaired cognitive domains (single vs multiple). However, predictive validity for outcomes of these criteria and the diagnostic stability of the subtypes are questionable. Methods We investigated the outcomes of 140 patients with MCI who participated in the baseline study of the Korean Longitudinal Study on Health and Aging and completed the 18-month follow-up evaluation (mean duration of follow-up = 1.57 ± 0.24 years). We evaluated the predictive validity of the criteria using multinomial logistic regression analyses, and the diagnostic stability of MCI subtypes using annual conversion rates between subtypes. Results Compared with the single-domain type (MCIs), the multiple-domain type (MCIm) had a lower chance of reversion to normal cognition (MCIm = 10.94%, MCIs = 43.42%) and higher risk of conversion to dementia (MCIm = 23.44%, MCIs = 5.26%). The difference in the reversion rate between the multiple- and single-domain type was statistically significant (odds ratio = 0.233, 95% confidence interval = 0.070–0.771, P = .017). However, neither the chance of reversion nor the risk of conversion was different between amnestic and nonamnestic subtypes. Among the 81 participants who neither converted to dementia nor reverted to normal cognition, 39 converted to different subtype (annual conversion rate = 17.74%). Conclusions The number of impaired cognitive domains, but not the presence of memory impairment, predicted poor outcomes in people with MCI. However, MCI subtype was diagnostically unstable.