Premium
Development of a short‐form version of the Reading the Mind in the Eyes Test for assessing theory of mind in older adults
Author(s) -
Chander Russell J.,
Grainger Sarah A.,
Crawford John D.,
Mather Karen A.,
Numbers Katya,
Cleary Rhiagh,
Kochan Nicole A.,
Brodaty Henry,
Henry Julie D.,
Sachdev Perminder S.
Publication year - 2020
Publication title -
international journal of geriatric psychiatry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.28
H-Index - 129
eISSN - 1099-1166
pISSN - 0885-6230
DOI - 10.1002/gps.5369
Subject(s) - psychology , test (biology) , receiver operating characteristic , cohort , theory of mind , reading (process) , reliability (semiconductor) , cognition , medicine , power (physics) , psychiatry , machine learning , computer science , paleontology , political science , law , biology , physics , quantum mechanics
Background The Reading the Mind in the Eyes test (RMET) is a 36‐item assessment for theory of mind (ToM) performance. While this measure has been shown to be sensitive to age‐related ToM difficulties, there are no established cutoffs or guidelines currently available that are specific to older adults. This article seeks to validate a short‐form version of the RMET appropriate for use in such populations. Methods Cross‐sectional data from 295 participants (mean age 86 years) from the Sydney Memory and Ageing Study, a longitudinal community observational cohort. Participants underwent an assessment battery that included the RMET. Individuals who scored >1SD below the RMET scores of cognitively normal participants were deemed to have below average RMET scores. Various model‐building methods were used to generate short‐form solutions of the RMET, which were compared with previously validated versions in their predictive power for below average full RMET performance. Results Individuals with below average RMET performance tended to be older and have poorer global cognition. Of the eight short‐form solutions, the 21‐item version generated using genetic algorithm exhibited the best classification performance with an area under the receiver operating curve (AUROC) of 0.98 and had 93.2% accuracy in classifying individuals with below average ToM. A shorter 10‐item solution derived by ant colony optimization also had acceptable performance. Conclusion We recommend the 21‐item version of the RMET for use in older adult populations for identifying individuals with impaired ToM. Where an even shorter version is needed with a trade‐off of slightly reduced performance, the 10‐item version is acceptable.