Background: The cognitive-functional composite (CFC) was designed to improve the measurement of clinically relevant changes in predementia and early dementia stages. We have previously demonstrated its good test-retest reliability and feasibility of use. The current study aimed to evaluate several quality aspects of the CFC, including construct validity, clinical relevance, and suitability for the target population. Methods: Baseline data of the Capturing Changes in Cognition study was used: an international, prospective cohort study including participants with subjective cognitive decline (SCD), mild cognitive impairment (MCI), Alzheimer's disease (AD) dementia, and dementia with Lewy bodies (DLB). The CFC comprises seven existing cognitive tests focusing on memory and executive functions (EF) and the informant-based Amsterdam Instrumental Activities of Daily Living Questionnaire (A-IADL-Q). Construct validity and clinical relevance were assessed by (1) confirmatory factor analyses (CFA) using all CFC subtests and (2) linear regression analyses relating the CFC score (independent) to reference measures of disease severity (dependent), correcting for age, sex, and education. To assess the suitability for the target population, we compared score distributions of the CFC to those of traditional tests (Alzheimer's Disease Assessment Scale-Cognitive subscale, Alzheimer's Disease Cooperative Study-Activities of Daily Living scale, and Clinical Dementia Rating scale). Results: A total of 184 participants were included (age 71.8 ± 8.4; 42% female; n = 14 SCD, n = 80 MCI, n = 78 AD, and n = 12 DLB). CFA showed that the hypothesized three-factor model (memory, EF, and IADL) had adequate fit (CFI = .931, RMSEA = .091, SRMR = .06). Moreover, worse CFC performance was associated with more cognitive decline as reported by the informant (β = .61, p < .001), poorer quality of life (β = .51, p < .001), higher caregiver burden (β = − .51, p < .001), more apathy (β = − .36, p < .001), and less cortical volume (β = .34, p = .02). Whilst correlations between the CFC and traditional measures were moderate to strong (ranging from − .65 to .83, all p < .001), histograms showed floor and ceiling effects for the traditional tests as compared to the CFC.

Introduction To improve the detection of changes in Alzheimer's disease (AD), we designed the cognitive-functional composite (CFC). As a first validation step, we investigated its test–retest reliability and feasibility of use. Methods We performed a test–retest study with 2–3 weeks between assessments, including patients with mild cognitive impairment (MCI) or mild AD dementia and cognitively healthy participants. We calculated intraclass correlation coefficients (ICCs) type absolute agreement for all CFC measures and compared baseline and retest scores using paired-samples t -tests. We evaluated feasibility by interviewing participants. Results Forty-three patients (40% female, mean age = 69.9) and 30 controls (50% female, mean age = 65) were included. Subtest intraclass correlation coefficients ranged from .70 to .96. We found negligible improvements after retesting on only two subtests. Overall, patients perceived the administration of the CFC as feasible. Discussion The CFC is a stable and feasible measure in MCI and mild AD dementia, and thereby meets important quality metrics for clinically meaningful outcome measures.

Objective: Since computerized cognitive test performance may be influenced by computer experience, correction for this measure might be needed. This study examined how to correct for computer experience by examining its influence on online and traditional tests. Method: 248 healthy adults completed an online neuropsychological test battery and 70 adults completed traditional equivalents of the tests. Computer experience was assessed by a performance-based and a self-report measure. Regression analyses were applied to examine their influence on the online and traditional tests. Results: After correction for demographics, the performance-based measure was associated with online and traditional, predominantly speed-based, tests. The self-report measure was also associated with speed-based online tests but not with most traditional tests. Conclusions: Correcting computerized neuropsychological tests using a performance-based measure of computer experience would be unwise, because this measure also seems to tap into cognitive functions. A correction using a self-report measure might be better and is appropriate.