Cognitive impairment is common in patients with type 2 diabetes mellitus (T2DM) and negatively impacts effective diabetes self-care. However, it is unclear whether self-care level can predict cognition status in T2DM patients. We aimed to identify self-care scores that would provide information on the patients’ cognition status.
Methods: We collected diabetes self-care and cognition data from 50 T2DM patients (age 55.9±7.6 years, diabetes duration 10.1±8.2 years, A1C 7.1±1.4%, 54% female) using the Scale for Summary of Diabetes Self-Care Activities Assessment (SDSCA) and the Montreal Cognitive Assessment (MoCA) . A mixed effects logistic regression model for repeated measures used both baseline and 6 months follow-up observations with adjustment for within-subject correlation (SAS GLMMIX) . First, from nine potential covariates, an iterative backward elimination process selected a parsimonious subset of covariates for predicting MoCA level (normal vs. impairment) , optimizing area under the curve (AUC) . Four covariates were identified (education, income, ethnicity, sleep quality; AUC=.75) . Then an iterative process calculated AUC for MoCA level predicted from self-care level, using each possible self-care score cut point from .to 6.95 in .increments, controlling for the four identified covariates. An optimal cut point was selected that produced the highest AUC value.
Results: The optimal self-care cut point was 2.95 (AUC=.76) , producing good discrimination of MoCA level. As a sensitivity analysis, the second best cut point of 3.95 produced AUC=.75.
Conclusion: Self-care cut points, accounting for education, income, ethnicity, and sleep quality, could be used to gain information on T2DM patients’ cognition status, which may prompt clinicians and researchers to further evaluation with formal cognitive tests.
S.E.Choi: None. M.Freeby: Research Support; Abbott Diabetes, Novo Nordisk. B.Roy: None. M.A.Woo: None. R.Kumar: None. M.Brecht: None.
National Institutes of Health (RNR017190)