Effects of a vibrating sleep mask on objectively measured sleep: a controlled, at-home study

Abstract

Purpose

This randomized, controlled, crossover, within-subjects study investigated effects of a gently vibrating sleep mask on objectively measured sleep outcomes.

Methods

Adults with mild sleep difficulties (n = 91, 69% women, mean age = 45) tracked sleep at home with a polysomnography-validated noncontact device for three 14-night periods (baseline, vibrating mask, control mask).

Results

When wearing the vibrating mask compared to the control mask, controlling for baseline, participants objectively spent more time in bed (+ 11.11 min), slept longer (+ 11.87 min), increased deep sleep (+ 3.09 min), and spent less of the night in light sleep (-1.8% relative difference).

Conclusions

A mask administering gentle vibration may be associated with improved sleep outcomes among adults with mild sleep difficulties.

Introduction

Previous research has shown that vibration can increase relaxation, induce drowsiness, and improve sleep among healthy individuals and those with insomnia [e.g., 1, 2]. Some of these studies have focused on whole-body vibration; however, vibration applied to the eye and temple area at bedtime may be particularly helpful for promoting relaxation and supporting sleep [2].

Additionally, wearing light-blocking eye masks during overnight sleep is a common nonpharmacological sleep intervention. Numerous studies have shown eye masks to be useful in hospital environments where artificial light disrupts sleep [3], and experimental research has shown that overnight eye mask use can improve daytime alertness [4].

When used together, therefore, vibration delivered during the initiation of sleep as part of a light-blocking mask could help people relax as they fall asleep and experience better sleep outcomes. Previous research on vibrating sleep masks has not been directly tied to sleep outcomes.

We hypothesized that vibration therapy delivered via a light-blocking mask would be associated with improved objectively measured sleep outcomes, compared to using a mask without vibration, among adults with mild sleep difficulties.

Methods

Participants

Selected users of SleepScore technology in the United States were invited by e-mail to complete an eligibility questionnaire if they were interested in a study testing two sleep masks. Participants were required to be between ages 18 and 75, with objective or self-reported indication of difficulty falling asleep and/or staying asleep. To recruit a sample of healthy participants, people reporting sleep disorders or other medical diagnoses affecting sleep were excluded, as were people taking medications or using substances that impact sleep. Additional exclusion criteria included shift work, having children under age one at home, and upcoming travel.

The 91 participants were 69% women; the most frequent racial group identifications were White (66%), Black or African American (11%), and Asian (10%); and the mean age was 45.

Design and procedures

Sterling Institutional Review Board (ID: 11496) approved the study; all participants provided informed consent. A randomized, controlled, crossover, within-subjects study design was implemented. To establish a baseline measure of sleep, participants tracked their sleep at home for 14 nights wearing no mask. The two intervention periods were 14 nights each. Participants were instructed to put on the assigned mask at bedtime and wear it throughout the night, while continuing to track their sleep. Then they crossed over to use the other mask for the next intervention period. They were not informed how the different masks might affect sleep differently. Participants were randomized as to which mask they received first. After completing the study, participants received a Theragun PRO^® as compensation.

The vibrating mask was the Therabody ^® SleepMask, which was designed to improve sleep via gentle vibrating massage to promote relaxation and by blocking out light. The eyepiece is a proprietary fabric knitted in a “butterfly net” pattern with 95% polyester and 5% spandex to maximize breathability when in contact with the skin. Participants were instructed to use their preferred settings from three vibration pattern options (i.e., pulse, wave, or constant) and two intensity options (i.e., low or high mode). The proprietary vibration patterns range include frequencies between 0 and 75 Hz. The pulse vibration treatment provides a tactile simulation of a heartbeat that is slowing in pulse rate and vibration intensity up to 50 Hz in low mode and up to 75 Hz in high mode. The wave vibration treatment provides a rhythmic breathwork pattern mimicking inhalation and exhalation over a 5 s period up to 50 Hz in low mode and up to 75 Hz in high mode. The constant vibration treatment provides 40 Hz in low mode and 75 Hz in high mode continuously. Although participants selected their preferred settings, we did not record these choices to avoid disrupting sleep onset or relying on potentially inaccurate morning recall. The control mask was constructed uniquely for the purpose of this study, to ensure that the masks would be functionally identical (e.g., same materials, shape, weight) except for the vibration. See Fig. 1 for an example of an individual wearing the vibrating mask.

Fig. 1.

Therabody^® SleepMask. Note: Example of Therabody^® SleepMask used by all participants. Image provided by Therabody, Inc. The control mask was identical except for the vibration feature.

Measurement

Objective sleep was measured with SleepScore Max, a noncontact monitoring device using respiratory signal and motion sensing to detect sleep. This bedside sensor uses ultra-low power radiofrequency waves to monitor body movement and respiration patterns. High resolution magnitude and duration data of gross movements, micro-movements, and full breathing cycles are captured and transformed into 30-second epoch sleep stage data (wake, light, deep, REM) using proprietary algorithms. Good agreement with gold-standard polysomnography (PSG) [e.g., 5], exceeding the agreement typically reported for actigraphy [6], has been documented. Like all consumer sleep tracking technologies, SleepScore Max has limitations. It is not an ambulatory measure of sleep-wake patterns. Nor does it classify sleep stages with the accuracy achievable through PSG.

Standard sleep metrics (e.g., sleep onset latency, total sleep time, wake after sleep onset) were calculated using the 30-second epoch data. Additionally, proprietary SleepScore technology measures reflecting sleep quality were calculated, ranging from 0 to 100: BodyScore represents a measure of deep sleep duration adjusted to reflect age-and-gender normalized values. Specifically, an individual’s measured deep sleep duration is normalized against established population norms, ensuring appropriate comparison across different age and gender categories. MindScore similarly quantifies the duration of REM sleep, also adjusted based on age-and-gender normalized values. Thus, MindScore provides a standardized measure of REM sleep duration relative to expected normative values. SleepScore is an aggregated “global” measure of objective sleep quality integrating multiple sleep parameters, each adjusted for age-and-gender normalized values. Validation of the sleep-wake and staged data underlying BodyScore, MindScore, and SleepScore have been assessed in both healthy populations and individuals with subclinical or clinically significant sleep-disordered breathing [e.g., 5–7].

Brief daily questionnaires were sent to track compliance and assess demographics and other self-report variables not included in this paper.

Statistical analysis

Analyses compared nights when participants used the vibrating mask to nights when they used the control mask, controlling for baseline sleep. Mixed-effects multilevel regression was used, taking into account the nested structure of the data, with random intercepts for participants. The order of mask assignment (vibrating first vs. control first) was not included as a fixed effect in the model, as the randomization process was expected to balance order effects across participants. The baseline period had 1061 nights of data included, the control mask period had 896 nights of data included, and the vibrating mask period had 894 nights of data included. Because the multilevel model averages across nights, participants who were missing more than 5 nights of any period of data collection were removed from analyses. Two-tailed significance testing was statistically powered to detect differences at p<.05.

Results

As shown in Table 1, when comparing the vibrating mask to the control mask, controlling for baseline sleep, use of the vibrating mask was associated with longer time in bed (2.7%1 difference, p=.024); longer sleep duration (2.9% difference, p=.032); more time spent in deep sleep (4.1% difference, p=.007); improved BodyScore (1.2% difference, p=.048); and a lower proportion of the night spent in light sleep (1.8% relative difference, p=.035).

Table 1.

Multilevel regression results (n = 2851 nights nested within 91 participants) comparing objectively measured sleep outcomes on nights using vibrating mask to nights using control mask, controlling for baseline sleep with no mask

	Observed mean (SD)			Estimated marginal means
	Vibrating mask	Control mask	Absolute difference	Intercept (SE)	beta	p-value
Total sleep time (minutes)	387.47 (55.26)	376.36 (57.76)	11.11	373.37 (3.21)	6.91	0.032
Time in bed (minutes)	453.06 (58.71)	441.19 (56.63)	11.87	437.20 (3.29)	7.41	0.024
Sleep onset latency (minutes)	19.76 (11.27)	20.25 (14.23)	0.49	19.40 (0.77)	0.19	0.802
Number of awakenings	4.58 (1.83)	4.49 (1.83)	0.09	4.45 (0.10)	0.03	0.732
Wake after sleep onset (minutes)	39.60 (24.82)	38.31 (22.70)	1.29	37.64 (1.25)	0.63	0.613
Sleep efficiency	0.86 (0.07)	0.85 (0.07)	0.01	0.86 (0.00)	0.00	0.771
Sleep maintenance	0.91 (0.06)	0.91 (0.05)	0.00	0.94 (0.00)	0.00	0.985
Light sleep (minutes)	236.49 (38.79)	231.40 (40.31)	5.09	232.88 (2.49)	1.11	0.655
Deep sleep (minutes)	76.28 (23.28)	73.19 (21.42)	3.09	70.31 (1.25)	3.39	0.007
REM sleep (minutes)	75.01 (17.90)	71.78 (15.45)	3.23	70.24 (1.31)	2.51	0.056
% Light sleep	55% (3.71)	56% (3.96)	1.00%	56.23 (0.34)	−0.71	0.035
% Deep sleep	18% (5.77)	18% (5.64)	0.00%	17.64 (0.30)	0.43	0.159
% REM sleep	18% (3.48)	17% (2.76)	1.00%	16.98 (0.27)	0.28	0.301
% Wake after sleep onset	9% (5.50)	9% (5.45)	0.00%	9.05 (0.27)	0.01	0.982
SleepScore (0–100)	80.37 (9.40)	79.11 (9.70)	1.26	78.83 (0.50)	0.78	0.117
BodyScore (0–100)	81.63 (9.11)	80.67 (9.12)	0.96	79.98 (0.48)	0.94	0.048
MindScore (0–100)	78.09 (8.13)	77.20 (7.93)	0.89	77.21 (0.62)	0.43	0.484

Metrics are reported in either minutes (absolute values) or percentages (relative to sleep period time), as specified in the first column. For the vibrating mask and control mask periods, each average was calculated by averaging nights across participants, and then averaging those participants’ averages to a single simple average, listed under Observed in the table. Listed under Estimated are the outcomes of the multilevel regression analyses. Intercept represents the estimated value of the sleep metric during the control mask period. The beta coefficient can be interpreted as the increase or decrease in sleep outcome in the metric of that outcome (e.g., minutes) when comparing the vibrating mask period to the control mask period. P-value represents statistical significance. Bold indicates significant at p<.05. A mixed-effects model with random intercept was used. Regression model was as follows: SleepMeasure_ij = β₀ + β₁*BaselineSleepMeasure_ij + β₂*TestPeriod_ij + u_0j + e_ij. TestPeriod was coded as 0 for nights during the baseline period when participants did not use a mask, 1 for nights when participants used the control mask, and 2 for nights when participants used the vibrating mask

Discussion

This study supports previous research suggesting that vibration therapy can help sleep among individuals reporting difficulty falling asleep or staying asleep in the absence of clinical sleep disorders [e.g., 2]. Statistically significant improvements were seen in objectively measured sleep when wearing a light-blocking mask that gently vibrated with gradually decreasing intensity and was worn overnight, compared to when sleeping overnight wearing an identical mask that did not vibrate.

Interestingly, participants spent more time in bed (11 min per night on average) and slept longer (12 min per night on average) during the period when the vibrating mask was used. This suggests that longer time in bed may have driven the observed effects; however, other differences suggest subtle changes in sleep architecture: a slight but significant reduction in the proportion of the night spent in light sleep and increases in two deep sleep metrics (i.e., deep sleep measured in minutes and BodyScore, an age- and gender-adjusted measure of deep sleep). The latter results are consistent with previous research linking changes in deep sleep both to benefits of vibration [8] and to benefits of sleeping in an eye mask [4]. To what extent these effects may be synergistic would be an interesting topic for future research. Similarly, future research should explore underlying mechanisms affecting sleep architecture; for example, is relaxation directly increasing time spent in deep sleep.

Longitudinal objective data were collected in-home, carrying the advantage of more ecologically valid results than laboratory research. However, there may have been some bias in that participants were recruited from a database of individuals interested in this type of research; this population may be especially comfortable using technology, which could limit generalizability of the findings. Participants were aware that the masks had the potential to affect their sleep, which could have led to expectancy effects. Future studies could reduce expectancy effects by including a sham condition that mimics the intervention without delivering therapeutic vibration. We did not ask participants which vibrational settings they used, so as not to disrupt sleep or obtain unreliable information the next morning. We acknowledge this is a limitation, since different settings could have affected relaxation and sleep differently, both physiologically and psychologically. Also, although we instructed participants to track their sleep in their usual environment with their usual routine, there could have been behavioral changes that influenced sleep outcomes. A final study limitation is that participants resided only in the United States.

Although future investigation is needed to identify underlying mechanisms of the observed effects, the current findings have applied value, highlighting potential for a mask administering gentle vibration to be associated with improved sleep outcomes among adults with mild sleep difficulties. While the mean observed changes in the current study were modest, falling within the range of normal night-to-night variability, even small gains in sleep duration and deep sleep for some individuals may translate into meaningful benefits for next-day functioning, quality of life, and positive health outcomes [9, 10].

Funding

Therabody, Inc provided products and funding support to SleepScore Labs to design, conduct, and report the study independently. Therabody, Inc was not involved in any aspect of data collection, analysis, or interpretation. All data were analyzed by SleepScore Labs researchers without input or oversight from Therabody, and all interpretations and conclusions are solely those of the authors.

Declarations

Conflict of interest

The authors are employed by SleepScore Labs.

Ethical approval/Human participants

The study was approved by Sterling Institutional Review Board (ID: 11496). All procedures were in accordance with the ethical standards of the institutional review board and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

All research participants provided informed consent.