Pragmatic questionnaire-based evaluation of auditory function in individuals with major neurocognitive disorders and hearing loss in diverse contexts.

Background: Hearing impairment in older people is a significant risk factor for cognitive decline and dementia, while it is a source of bias in the diagnostic workup of cognitive complaints. Early detection and intervention are critical, yet audiometric equipment is often unavailable in primary healthcare- and/or community care-, as well as in low-resource settings across the globe. Objective: This study aims (i) to develop brief accurate instruments for capturing hearing loss severity based on items of the 25-item Hearing Handicap Inventory for the Elderly (HHIE) and its counterpart the Hearing Handicap Inventory for the communication partner (HHIE-SP) and (ii) to compare their usefulness as well as that of the 10-item screening version of HHIE (HHIE-S) in detecting hearing loss severity in people with dementia and hearing loss to HHIE and HHIE-SP. Methods: The study relies on screening- and baseline data of the Sense-Cog Trial, being a European, multi-center, observer-blind, 36-week long, randomized controlled trial (RCT) of people with dementia with sensory impairment and their companions. An exploratory data analysis was utilized to provide a comprehensive understanding of the data structure and the characteristics of the sample. Eight different proportional odds logistic regression models were computed to study the relationship between the pure-tone audiometry screen results and different versions of the HHIE, with or without consideration of demographic data of the person with dementia and his/her communication partner. Stratified repeated random subsampling was employed to create two new HHIE models. All models were assessed by calculating the Mean Squared Deviation (MSE) over 1,000 splits into 90% training and 10% test set. Results: Two separate HHIE-mini models were developed. HHIE-2 includes one item of the HHIE and one item of the HHIE-SP. HHIE-8 includes three items of the HHIE and five items of the HHIE-SP. The model including HHIE-S and demographic data demonstrated the highest performance (MSE = 6.818), followed by the model including HHIE-SP and demographic data (MSE = 7.065) and the HHIE-2 model which included age (MSE = 7.254) but not country of residence. The HHIE-8 model was less effective (MSE = 7.740), and the model including HHIE and no demographic data was the least reliable (MSE = 9.220). Conclusion: HHIE-S and HHIE-2 combined with demographic data are practical and more efficient tools for assessing hearing loss severity in people with dementia and hearing impairment compared to HHIE, HHIE-S and HHIE-SP in different European countries. They both address the specific challenges associated with dementia-related hearing assessments by limiting the cognitive load of the evaluation process. Particularly the ultra-brief HHIE-2 may be feasible for use in primary and community healthcare settings in different countries, since in a European cohort it is not affected by the country of residence of the individuals with dementia.