Abstract

Background: Telephone and videoconference administration of cognitive tests introduce additional sources of variance compared to in-person testing. Reviews of test-retest reliability have included mixed neurocognitive and psychiatric populations with limited consideration of methodological and statistical contributions. Objective: We reviewed reliability estimates from comparison studies of older adults with and without dementia, considering test-retest analyses and study methods. Methods: Medline, Embase, PsycINFO, andWeb of Science were systematically searched from 1 January 2000 to 9 June 2020 for original articles comparing telephone or videoconference administered cognitive instruments to in-person administration in older adults with and without dementia or mild cognitive impairment. Results: Of 4,125 articles, 23 were included: 11 telephone (N = 2 dementia cohorts) and 12 videoconference (N = 4 dementia cohorts). Telephone administered subtest scores trended in the same direction as in-person with comparable means. Person-level data were scarce. Data on dementia was only available for MMSE, with resulting subtle modality bias. MMSE, SMMSE, Letter Fluency, and HVLT-R in healthy to mild-moderate Alzheimer's disease were particularly reliable for videoconference administration. Other tests show promise but require more observations and comprehensive analyses. Most studies used high-speed stable videoconferencing hardware resulting in a lack of ecological validity for home administration. Conclusion: Remote administration is often consistent with in-person administration but variable and limited at the person/test level. Improved statistical design and inclusion of dementia related cohorts in telephone studies is recommended. Reliability evidence is stronger for videoconferencing but with limited applicability to home administration and severe dementia. Improved reporting of administrative procedures is recommended.