RESUMEN
BACKGROUND: A large proportion of older adults assessed for cognitive impairment likely have hearing loss, potentially affecting accuracy of cognitive performance estimations. This study aimed to develop a hearing-impaired version of the Addenbrooke's Cognitive Examination-III (HI-ACE-III) and to assess whether the HI-ACE-III can accurately distinguish people with mild cognitive impairment (MCI) and dementia from cognitively intact controls. METHODS: The HI-ACE-III was developed by converting verbal instructions into a visual, timed PowerPoint presentation. Seventy-four participants over the age of 60 years were classified into three groups: 29 had MCI, 15 had mild to moderate dementia and 30 were cognitively intact controls. Receiver operating characteristic (ROC) curves were graphed to test screening accuracy. Concurrent validity was examined through correlations between HI-ACE-III domain scores and relevant, visually presented standardized neuropsychological measures. RESULTS: ROC analysis for dementia revealed an area under the curve (AUC) of 0.99, achieving excellent sensitivity (100%) and good specificity (93.3%) at an optimum cut-off of <87. The AUC for MCI was 0.86, achieving reasonable sensitivity (75.9%) and good specificity (86.7%) at an optimum cut-off of <92. HI-ACE-III subtests shared anticipated and statistically significant correlations with established measures of cognitive functioning. Internal consistency of the HI-ACE-III was excellent as verified with Cronbach's alpha (α = 0.904). CONCLUSIONS: Preliminarily, the HI-ACE-III showed good reliability, validity and screening utility for MCI and dementia in older adults in a hearing-impairment context. The adapted HI-ACE-III may offer accurate and reliable indication of cognitive performance, supporting timely diagnosis and research examining links between hearing loss and cognitive decline.