Abstract Introduction We assessed the performance and usability of an electroencephalography headband (HB) compared with polysomnography (PSG) in persons with cognitive impairment and sleep disturbances. Methods Twenty-five older adults with mild cognitive impairment or dementia and symptoms of insomnia and/or daytime sleepiness, supported by 22 caregivers, wore HBs (Beacon Biosignals) up to 7 nights and completed home-based PSGs (NOX Medical). We compared total sleep time (TST), wake after sleep onset (WASO), sleep latency (SOL), sleep efficiency (SE), and sleep stages from nights with simultaneous HB and PSG recordings. We calculated intraclass correlation coefficients (ICCs) and examined feasibility of HB use via usability scales and open-ended responses. Results Average age was 75.3SD=5.4 years, 10 (40%) had Alzheimer’s disease, 7 (28%) were female, and 24 (96%) were non-Hispanic White. MoCA scores averaged 21.7 (range=6–29). On PSG, averages for TST, WASO, and SE were 427 minutes, 89 minutes, and 79%. Six participants were excluded from analyses due to poor signal quality (i.e., EEG channel quality below 50% across all channels or off-head time 10% of recording). Excluded persons were not different with respect to cognitive scores and all had caregiver support. Among 19 participants, TST and WASO had good agreement with PSG (ICCs ranging 0.61-0.77), while agreement for SOL and SE was lower (ICCs 0.39-0.55). ICCs for N3 and REM sleep were 0.52 and 0.61, respectively, while light (N1 and N2) sleep showed poor agreement (ICC=-0.28). On average, participants wore the HB for 61.4 nights, with 36.4% completing ≥ 7 nights. Most participants found the HB easy to use: 22 reported it was not cumbersome, and 21 said it was not overly technical. The lowest-scoring items were related to comfort and sleekness of the HB. Open-ended responses highlighted issues with fit (n=9), manual start of recordings (n=6), and using the HB with continuous positive airway pressure therapy (n=2). Conclusion The HB demonstrated good agreement with PSG for TST and WASO and lower agreement for sleep stages. Most participants used the HB for multiple nights. Improvements in feasibility and algorithm performance of the HB are needed to improve sleep assessment in cognitively impaired older adults. Support (if any)
Chen et al. (Fri,) studied this question.