The Virtual Kitchen Challenge—Version 2: Validation of a Digital Assessment of Everyday Function in Older Adults
Background: Conventional methods of functional assessment include subjective self- or informant report, which may be biased by personal characteristics, cognitive abilities, and lack of standardization (eg, influence of idiosyncratic task demands). Traditional performance-based assessments offer some advantages over self- or informant reports but are time-consuming to administer and score. Objective: This study aims to evaluate the validity and reliability of the Virtual Kitchen Challenge—Version 2 (VKC-2), an objective, standardized, and highly efficient alternative to current functional assessments for older adults across the spectrum of cognitive aging, from preclinical to mild dementia. Methods: A total of 236 community-dwelling, diverse older adults completed a comprehensive neuropsychological evaluation to classify cognitive status as healthy, mild cognitive impairment, or mild dementia, after adjustment for demographic variables (age, education, sex, and estimated IQ). Participants completed 2 everyday tasks (breakfast and lunch) in a virtual kitchen (VKC-2) using a touchscreen interface to select objects and sequence steps. Automated scoring reflected completion time and performance efficiency (eg, number of screen interactions, percentage of time spent off-screen, interactions with distractor objects). Participants also completed the VKC-2 tasks using real objects (Real Kitchen). All participants and informants for 219 participants completed questionnaires regarding everyday function. A subsample of participants (n=143) performed the VKC-2 again in a second session, 4-6 weeks after the baseline, for retest analyses. Analyses evaluated construct and convergent validity, as well as retest and internal reliability, of VKC-2 automated scores. Results: A principal component analysis showed that the primary VKC-2 automated scores captured a single dimension and could be combined into a composite score reflecting task efficiency. Construct validity was supported by analyses of covariance results showing that participants with healthy cognition obtained significantly better VKC-2 scores than participants with cognitive impairment (all Ps
