“My Fitbit tells me I don’t sleep”

Abstract

Background: There has been a rapid growth in wearables marketed to measure sleep. Fitbit trackers collect data through an internal accelerometer and use heart rate variability to estimate the sleep-wake state. There is a paucity of “real-world” data in patients who are being evaluated for sleep disorders.

Aim: To evaluate the agreement between Fitbit Charge3^TM & in-lab polysomnography (PSG) in patients who require a PSG for assessment.

Methods: A prospective study of patients attending a PSG through Epworth Camberwell Sleep Lab between 2020-2021 was conducted. Fitbit Charge3^TM was worn on the dominant wrist with concurrent PSG monitoring. Parameters measured included total sleep time; TST (min), Sleep onset latency; SOL (min), sleep efficiency; SE (%), wake after sleep onset; WASO (min) and time spent in N1, N2, N3 and REM sleep (min). 30 second epoch-by-epoch analyses were conducted.

Results: Seventy patients (male=52), median age of 55 years; IQR(45,67) completed the study. On average, the Fitbit significantly overestimated light sleep by 62.99 min, TST by 29.50 min, and SE by 3.33% and underestimated deep sleep by 40.55 min WASO by 28 min. Fitbit and PSG did not significantly differ on REM or SOL measurements.

Conclusion: Our findings may support the use of Fitbit Charge3^TM as an initial screening device to assess sleep duration and architecture in select patients attending sleep clinics

Table 1: Comparison of sleep variables between PSG and Fitbit Charge 3^TM