Distinguishing early cognitive impairment in older adults with CogniScan: a nine-item composite derived from the Mini-Mental State Examination and Clock Drawing Test.

AIMS: Identifying cognitive impairment in routine care is essential, yet many brief screening instruments exhibit ceiling or floor effects. To our knowledge, no widely used brief cognitive screening instrument has incorporated the qualitative Clock Drawing Test (CDT) command-copy discrepancy, in which the copied clock is better organized than the clock drawn to command, into a composite score. METHODS: In this cross-sectional study, patients with amnestic multidomain mild cognitive impairment (aMD-MCI) or early Alzheimer's disease (AD) were consecutively recruited from a tertiary memory clinic at Salamanca University Hospital, Spain, and compared with cognitively unimpaired community-dwelling controls. All participants underwent the Mini-Mental State Examination and the CDT. CogniScan was derived post hoc from nine elements retained through binary partial least squares regression: five temporal-orientation items, three delayed-recall words and one CDT command-copy discrepancy indicator. Factor maps from the fitted model were used to visualize group separation, and receiver operating characteristic analyses were used to estimate apparent discriminative performance and candidate cutoffs. RESULTS: The final sample included 183 older adults (median age, 75 years): 35 with aMD-MCI, 75 with early AD and 73 cognitively unimpaired controls. All nine CogniScan elements were strongly associated with cognitive status, with large effect sizes. In the primary binary analysis, discrimination was excellent (area under the curve [AUC] 0.99, 95% confidence interval 0.98-1.00). The Youden-optimized cutoff was ≥2 errors (sensitivity 96%, specificity 99%). In secondary exploratory analyses restricted to cognitively impaired participants, discrimination between aMD-MCI and early AD was lower (AUC 0.89, 95% confidence interval 0.82-0.97), consistent with partial overlap on the latent maps. In exploratory models, categorical educational level was not significantly associated with CogniScan score among controls and was not an independent predictor of cognitive status in the adjusted model. CONCLUSIONS: CogniScan showed excellent apparent discriminative performance, but external validation across broader cognitive phenotypes and wider age and educational ranges is needed. It appears best suited as a binary detector of impairment rather than as a definitive stage classifier.