JMIR Publications

ABSTRACT

Background:

Multi-sensor fitness trackers offer the perspective to longitudinally estimate sleep quality in a home environment that can outperform traditional actigraphy. To benefit from these new tools to objectively assess sleep for clinical and research purposes, multi-sensor wearable devices require careful validation against the gold standard of sleep polysomnography (PSG). Naturalistic studies in samples drawn from clinical populations favor validation.

Objective:

With this purpose in mind, we conducted a validation study of Fitbit Charge 2TM against portable home PSG in a shift-work population composed of 59 first-responder police officers and paramedics.

Methods:

Reliable comparison between the two measurements was ensured through data-driven alignment of the time series recorded in each night. Epoch-by-epoch analyses (EBE), together with Bland-Altman plots were used to assess sensitivity, specificity, accuracy, matthews correlation coefficient, bias and limits of agreement.

Results:

Sleep onset and offset, total sleep time, and the durations of rapid-eye-movement (REM) and non-rapid-eye-movement (NREM; N1 + N2 and N3) sleep stages displayed unbiased estimates, yet with non-negligible limits of agreement. By contrast, the proprietary Fitbit algorithm overestimated REM sleep latency by 29.4 min and wake time after sleep onset (WASO) by 37.1 min. EBE analyses indicated better specificity than sensitivity, with a higher accuracy for WASO (0.82) and REM sleep (0.86) than for N1 + N2 (0.55) and N3 (0.78) sleep. Fitbit heart rate (HR) displayed a small underestimation of 0.9 beats-per-minute (bpm) with a limited capability to capture sudden HR changes because of the reduced time resolution when compared to PSG. The underestimation was smaller in N2, N3, and REM sleep stages (0.6-0.7 bpm) compared to N1 sleep (1.2 bpm) and wake (1.9 bpm), indicating a state-specific bias. Finally, Fitbit suggested a distribution of all sleep episode durations that was different to that derived from PSG and showed non-biological discontinuities, indicating potential limitations of the staging algorithm.

Conclusions:

We conclude that following careful data processing, Fitbit Charge 2TM can provide reasonably accurate mean values of sleep and HR estimates in shift-workers under naturalistic conditions. The value of this consumer-grade multi-sensor wearable to tackle clinical and research questions could be enhanced with open-source algorithms, raw data access and the ability to blind participants from their own sleep data.