Ability of the Fitbit Alta <scp>HR</scp> to quantify and classify sleep in patients with suspected central disorders of hypersomnolence: A comparison against polysomnography

Abstract

Measuring sleep duration and early onset rapid eye movement sleep (REMS) is critical in the assessment of suspected central disorders of hypersomnolence (CDH). Current multi-sensor activity trackers that integrate accelerometry and heart rate are purported to accurately quantify sleep time and REMS; however, their utility in suspected CDH has not been established. This investigation aimed to determine the ability of a current, multi-sensor tracker, Fitbit Alta HR (FBA-HR), to quantify and classify sleep in patients with suspected CDH relative to polysomnography (PSG). Forty-nine patients (46 female; mean age, 30.3 ± 9.84 years) underwent ad libitum PSG with concurrent use of the FBA-HR. FBA-HR sleep variable quantification was assessed using Bland-Altman analysis. FBA-HR all sleep (AS), light sleep (LS; PSG N1 + N2), deep sleep (DS; PSG N3) and REMS classification was evaluated using epoch-by-epoch comparisons. FBA-HR-detected sleep-onset rapid eye movement periods (SOREMPs) were compared against PSG SOMREMPs. FBA-HR displayed significant overestimation of total sleep time (11.6 min), sleep efficiency (1.98%) and duration of deep sleep (18.2 min). FBA-HR sensitivity and specificity were as follows: AS, 0.96, 0.58; LS, 0.73, 0.72;DS, 0.67, 0.92; REMS, 0.74, 0.93. The device failed to detect any nocturnal SOREMPs. Device performance did not differ appreciably among diagnostic subgroups. These results suggest FBA-HR cannot replace EEG-based measurements of sleep and wake in the diagnostic assessment of suspected CDH, and that improvements in device performance are required prior to adoption in clinical or research settings.