Effects of a Single-Evening Yoga Session on Sleep in US Adults With Insomnia Symptoms
Background:
Vinyasa yoga (VY), a form of yoga shown to meet guidelines for moderate physical activity, has been sparsely studied for its effects on sleep and autonomic function among adults with insomnia. In this study, we examine the effects of an acute evening VY session on sleep and nocturnal heart rate variability (HRV), a measure of cardiovascular autonomic function, in adults with insomnia symptoms.
Methods:
Thirty-three insufficiently active US adults (84.8% female; 78.8% White; age = 34.9 ± 10.6 years; body mass index = 28.9 kg/m2) with at least mild insomnia symptoms (Insomnia Severity Index ≥ 10) were randomized to either a single 60-minute session of VY (n = 17) or a quiet rest control condition (CON: n = 16). One night before and after the acute experimental session, participants wore a wrist accelerometer and chest heart rate monitor overnight and completed a sleep diary. Analyses compared presession with postsession changes in sleep and HRV between groups using linear mixed models and Cohen’s d effect sizes.
Results:
The change in actigraphy-assessed sleep efficiency (SE) and total sleep time (TST) did not differ from presession to postsession for VY (SE: 88.9 ± 5.7% to 88.9 ± 6.2%; TST: 434.08 ± 79.4 minutes to 428.30 ± 98.1 minutes) or CON (SE: 89.4 ± 5.7% to 88.7 ± 6.5%; TST: 425.44 ± 79.3 minutes versus 391.90 ± 101.5 minutes); group × time interactions were not statistically significant (each P > 0.48). No significant differences were found between groups in nocturnal root mean square of successive differences HRV (P = 0.94).
Conclusion:
An acute bout of VY performed in the early evening did not significantly change accelerometer- or diary-assessed sleep or impact cardiovascular autonomic function during sleep.
INTRODUCTION
Sleep is a biological process that is essential for health (1,2). However, approximately one-third of adults experience at least 1 insomnia symptom (e.g., difficulty initiating and/or maintaining sleep, early morning awakenings, nonrestorative sleep) (1–3). Physical activity (PA), performed either in acute bouts or on a regular basis, improves sleep in healthy adults and in those with sleep disturbances (e.g., insomnia, obstructive sleep apnea) (4). However, most researchers have focused on traditional modes of exercise such as aerobic (e.g., walking, cycling) or resistance training (4). Whether other modes of PA such as yoga have similar effects remains unclear.
Yoga is an increasingly popular form of PA (5,6), with approximately 16.9% of the US population reporting practicing yoga in 2022 (7). Numerous styles of yoga exist, but each shares a foundational principle that unifies the mind and body through physical postures, controlled deep breathing, relaxation, and meditation (8–10). Among these, vinyasa yoga (VY) is noteworthy for its emphasis on connecting breath with movement, creating a dynamic practice through poses and sequences, and potentially leading to a greater energy expenditure than other forms (5,6). Although the health benefits of habitual yoga practice are extensive, even a single yoga session has been shown to heighten mood (11), improve psychological and physiological markers of stress (12), enhance vascular function (13), increase energy expenditure (5,14), and decrease blood pressure (BP) (15). However, research examining the impact of a single session of yoga on sleep, particularly VY, is sparse.
To our knowledge, authors of only 1 study have examined whether an acute bout of yoga impacts sleep. Kudesia and Bianchi (16) conducted a pilot study examining the impact of Bikram yoga, a low-intensity style of yoga, on sleep parameters. They found that most sleep parameters did not differ between days with and without Bikram yoga over 14 days; however, the time to return to sleep after nocturnal awakening was significantly reduced on yoga days (16). It remains unknown whether an acute bout of VY, a style demanding greater energy expenditure (5,14), may improve sleep.
Research on how an acute session of yoga affects sleep is limited. However, various online non-peer-reviewed sources (e.g., blogs) suggest practicing vigorous yoga like VY in the morning and reserving restorative practices for later to prevent sleep disruption (17,18). However, to the best of our knowledge, these recommendations are largely speculative and lack substantial evidence. Authors of previous studies on evening yoga have primarily focused on restorative (or gentle) styles of yoga (19–22). Notably, authors of experimental studies involving traditional forms of PA have generally found that evening exercise does not impair and, in some studies, improves sleep (4). Thus, based on previous research on the impact of traditional PA modes and gentle yoga forms on sleep, VY may improve sleep.
Therefore, in this study, we aimed at examining the effects of an acute session of VY performed in the early evening on sleep and nocturnal heart rate variability (HRV), a measure of cardiovascular autonomic function, in adults with insomnia symptoms. Our primary hypothesis was that, compared with a nonactive control condition, a single session of VY would result in significantly greater actigraphy-assessed sleep efficiency (SE) and total sleep time (TST) that night. The secondary hypothesis was that, compared with a nonactive control condition, a single session of VY would result in significantly higher HRV values that night.
METHODS
The present study is a subcomponent of a randomized controlled trial in which the impact of 4 weeks of VY versus a nonactive control condition on sleep and cardiovascular health was examined. The results of the 4-week intervention will be published elsewhere. As a secondary aim of this trial, we report here our examination of the impact of an acute bout of VY on sleep and cardiovascular health. This study was registered as a clinical trial at clinicaltrials.gov (NCT05723211). The study was carried out between March and June of 2023.
Participants
Insufficiently active adults with self-reported insomnia symptoms were recruited for the study. Recruitment occurred in the Pittsburgh community via word of mouth, flyers, and advertising on the University of Pittsburgh Clinical and Translation Science Institute’s research registry platform. Initial eligibility was assessed using an online screener (Qualtrics, Provo, Utah). Participants were eligible if they were ages 18–55 years, reported past-month levels of PA below public health guidelines (i.e., the first 3 categories of the Leisure-Time Activity Categorical Item) (23), reported at least mild insomnia symptoms (i.e., Insomnia Severity Index ≥ 10) (24), and PA was not contraindicated (i.e., no affirmative responses to the Physical Activity Readiness Questionnaire for Everyone) (25). Exclusion criteria included regular VY and/or power yoga practice, diagnosed medical conditions requiring clearance, physical limitations or mobility restrictions, untreated major psychiatric disorders, elevated risk for sleep apnea (i.e., STOP-Bang score ≥ 5) (26), current or planned pregnancy within 3 months, current insomnia treatment, overnight shiftwork, or medication affecting heart rate (HR) during exercise.
Participants meeting initial eligibility criteria were contacted to arrange a visit to the research facility for further study explanation. Those willing to participate provided written informed consent. This study was approved by the institutional review board of the University of Pittsburgh (STUDY22110168).
Experimental Measures
Sleep Measurements
Before and after the acute experimental session, sleep was objectively assessed using an Actiwatch Spectrum Plus (Philips Respironics, Murrysville, Pennsylvania). The Actiwatch collected accelerometry, ambient light, and event marker data that were uploaded and edited in Actiware software (version 6.0.9). Data were collected in 30-second epochs (27). Rest intervals were manually edited following a standardized approach that used a hierarchical method with multiple inputs: event markers, ambient light, sleep diaries, and activity (28). Following rest interval editing, sleep/wake status for each epoch was determined with the Actiware algorithm using a wake threshold of 40 along with 5 minutes of sleep to establish sleep onset and 0 minutes to establish sleep offset (29). The primary sleep outcome was SE, defined as the percentage of time in bed that was spent asleep (27). The secondary outcome was TST.
Subjective sleep was assessed using the Expanded Consensus Sleep Diary for Evening (30). Participants reported daily naps, caffeine and/or alcohol intake, and medication usage each evening (30). Upon waking, participants recorded time in bed, time attempting to fall asleep, number and duration of awakenings, final awakening time, and TST. They also rated sleep quality (1 = very poor to 5 = very good) and feeling refreshed upon waking (1 = not at all rested to 5 = very well-rested) (30). The key diary-assessed secondary outcomes included SE, TST, and sleep quality.
Cardiovascular Measures
Immediately before and after the acute experimental session, daytime resting cardiovascular health was assessed via HR/HRV and BP by a trained research staff member. Nocturnal HRV was assessed before and after the acute experimental session during the sleep assessment.
Daytime and nocturnal HR/HRV were measured using a Polar H10 HR monitor (Polar Electro, Bethpage, New York). Participants were provided with a Bluetooth-enabled HR monitor strap for each assessment and instructed to wear the strap directly on the skin below the sternum at the fifth intercostal space.
Following a 5-minute seated rest period, a 5-minute artifact-free daytime HR/HRV was obtained for analysis. For nocturnal HR/HRV recordings, participants were instructed to wear the HR monitor throughout the night. The HR monitor was paired via Bluetooth with an ActiGraph GT9X (ActiGraph Corp, Pensacola, Florida) during initialization. HR was continuously measured and stored on the GT9X during the monitoring period at a beat-to-beat resolution of 1 millisecond and downloaded as interbeat R-R intervals. The nocturnal interval used for analysis was aligned with the actigraphy-derived rest interval. Python programming (version 3.11.1) was used to preprocess the nocturnal data, creating 5-minute segments throughout the night. Any segment with an HR of 0 b·min−1 (e.g., due to poor contact between the skin and electrodes of the HR monitor) was excluded from analysis. HRV parameters were analyzed during each 5-minute segment throughout the night, with the mean across all 5-minute segments retained for analysis. This approach reduces the potential influence of variations in nocturnal sleep duration (31).
HRV data from each 5-minute segment were imported as beat-to-beat RR intervals into Kubios HRV Premium software (version 3.5.0) for analysis (32). Erroneous RR interval values (e.g., ectopic beats) were detected and corrected using the automatic correction method within Kubios (33). Both time- and frequency-domain measures of HRV were captured. Time-domain measures quantify the time between successive beats, whereas frequency-domain measures (i.e., power spectral density) estimate the distribution of absolute power (quantified in ms2 units) or relative power (normalized units [n.u.]) into 4 frequency bands (i.e., ultralow frequency [ULF], very-low frequency [VLF], low frequency [LF], high frequency [HF]). The time-domain measure of root mean square of successive RR interval differences (RMSSD) was the primary HRV measure, as it is strongly linked to parasympathetic cardiac activation and HF HRV (34,35). Secondary measures for both daytime and nocturnal HRV measures included mean HR (b·min−1) and the frequency-domain measures of HF (ms2 [0.15–0.40 Hz]), LF (ms2 [0.04–0.15 Hz]), and LF/HF ratio.
Resting systolic and diastolic BP were measured using an automated BP system (Dinamap Carescape V100 Monitor; GE Healthcare, Chicago, Illinois) on the right arm in the seated position, with back supported and feet flat on the ground, after a ≥10-minute rest period. Two BP measurements were obtained with at least 1 minute between each measurement and averaged for analysis. If systolic and/or diastolic BP differed by ≥10 or ≥6 mmHg, respectively, a third measurement was taken and averaged with the previous 2 measurements (36).
Baseline Experimental Procedures
Baseline Sleep Assessments
Participants continuously wore the Actiwatch in conjunction to completing a sleep diary for 8 days. On 1 weeknight during this 8-day assessment period, HR was also continuously measured via telemetry. Data from the night when sleep and HR were assessed were considered the baseline comparisons for these analyses.
Baseline Daytime Cardiovascular Health Assessment
On the acute experimental study day, participants arrived at the research facility between 15:00 and 19:00. Participants were instructed to wear comfortable exercise clothing, refrain from moderate-to-vigorous PA and alcohol for ≥24 hours prior, and refrain from caffeine, nicotine, and eating (other than water) for ≥3 hours prior to the session; verbal compliance confirmation was obtained upon arrival. All procedures were performed in a controlled laboratory environment with room temperature maintained at 20.0–22.2°C.
Before the acute experimental condition, resting daytime cardiovascular function was assessed. Participants were fitted with a chest strap HR monitor and then instructed to sit quietly in an upright position with their feet uncrossed and flat on the ground with the back supported for a 5-minute rest period. A 5-minute artifact-free resting HRV measurement was collected, followed by an automated seated resting measurement of BP.
Experimental Conditions
After measuring resting daytime cardiovascular health, participants were randomly assigned to either VY or a nonactive control condition. Randomization sequences were stratified by sex (i.e., male, female) in a 1:1 ratio and blocks of 4 using SAS (v 9.4; Cary, North Carolina). A research staff member, unaffiliated with data collection, prepared randomization assignments in opaque envelopes.
VY Condition
Participants assigned to the VY condition were provided with a yoga mat and followed a prerecorded standardized yoga practice that included a person demonstrating each pose with verbal cues. The video was displayed on an 80-inch television in a quiet room. The VY sequence followed the Journey into Power sequence from Baptiste Power Yoga (37). Details of the VY protocol are provided in Supplemental Table 1. Briefly, the 60-minute VY protocol incorporated standing, seated, and supine postures and included the following order of sequences: integration, sun salutations, crescent lunge series, balancing, standing, back bending, and restorative series. Participants were instructed to follow the video cues and modify yoga poses for accessibility. A certified yoga instructor observed the participant’s fidelity to the practice and provided verbal assistance if needed.
Nonactive Control Condition
Participants were instructed to sit quietly for 60 minutes and watch a nature documentary (Growing Up Wild, Disneynature, 2016), which was shown on an 80-inch television in a quiet room. This control condition was chosen to account for the potential impact of the experimental protocol on study outcomes. A research staff member ensured participants stayed seated and refrained from using electronic devices.
Postacute Experimental Procedures
Immediately after the experimental session, the same HRV and BP procedures were repeated, starting with a 5-minute seated rest. Before leaving, participants were instructed to wear the actigraphy watch and HR monitor overnight and complete a sleep diary.
Analytical Approach
All analyses were performed using IBM SPSS Statistics (v 29.01.0; IBM, Armonk, New York). Statistical significance was set at P < 0.05. Data are presented as means ± standard deviations.
Descriptive characteristics were compared between yoga and control conditions. Independent t-tests and χ2 tests were used for continuous and categorical variables, respectively. The Shapiro-Wilk test assessed normality; if violated, the Wilcoxon rank-sum test was applied.
Linear mixed models with unstructured covariance were employed for all analyses. Fixed effects included condition (yoga, control), time point (baseline, postexperimental session), and the condition × time interaction, with participants as random effects. Given the preliminary nature of the study, P values were not adjusted for multiple outcomes.
Cohen’s d effect sizes were used to calculate the magnitude of difference for between- and within-groups comparisons. Effect sizes were interpreted as small (d = 0.20), medium (d = 0.50), and large (d ≥ 0.80) (38).
RESULTS
A summary of participant flow is provided in Figure 1. Characteristics of the 33 participants are summarized in Table 1. The sample had a mean Insomnia Severity Index score of 15.88 ± 3.93, indicating moderate severity insomnia symptoms. No significant differences were found in participant characteristics between the 2 experimental conditions (each P ≥ 0.144). All 33 participants completed all the experimental procedures.
Citation: Journal of Clinical Exercise Physiology 14, 3; 10.31189/CEPH-25-00005
Changes in sleep from baseline to postacute experimental session are summarized separately for actigraphy-assessed sleep (Table 2) and diary-assessed sleep (Table 3). Neither SE nor TST nor additional actigraphy-assessed variables differed over time between conditions (SE: F[1,30.56] = 0.90, P = 0.351; TST: F[1,31.28] = 0.51, P = 0.481), and the effect size changes for each of the actigraphy-assessed sleep variables were deemed negligible to small in magnitude for within- and between-groups comparisons. Similar findings were observed between the 2 conditions from before to after the acute experimental session for each diary-assessed sleep variable (each P ≥ 0.115; Table 3).
The changes in daytime and nocturnal cardiovascular outcomes are presented in Table 4. Significant condition × time interactions were observed for daytime cardiovascular outcomes of diastolic BP (F[1,30.57] = 8.81, P = 0.006), HR (F[1,25.64] = 7.91, P = 0.009), LF HRV (F[1,25.31] = 7.16, P = 0.013), and HF HRV (F[1,24.32] = 7.88, P = 0.010). While HR increased following yoga relative to quiet rest control, diastolic BP, LF, and HF HRV decreased to a greater extent following yoga than quiet rest control. The change in nocturnal RMSSD from baseline to the night after the acute experimental session was similar between the 2 conditions (F[1,57.78] = 0.06, P = 0.939). Within- and between-groups effect sizes for nocturnal RMSSD were negligible in magnitude.
DISCUSSION
In the present study, we examine the effects of a single VY session performed in the early evening compared with a quiet rest condition on sleep-related and cardiovascular outcomes. We hypothesized that, compared with a nonactive control condition, a single session of VY would result in significantly greater actigraphy-assessed SE that night. Secondarily, we hypothesized that VY would result in significantly higher nocturnal HRV that night. Overall, an acute bout of VY performed in the early evening did not significantly change accelerometer- or diary-assessed sleep or impact cardiovascular autonomic function during sleep.
Contrary to our hypothesis, changes in actigraphy-assessed SE and TST did not significantly differ between groups from before to after the acute experimental session. Moreover, no statistically significant changes were observed for any of the additional actigraphy- or diary-assessed sleep variables. However, medium-sized between-groups effects were observed for diary-assessed sleep, presenting mixed findings. Although sleep quality improved to a greater extent in the yoga group, yoga also led to a greater increase in sleep onset latency and delay in bedtime relative to the quiet rest condition.
To our knowledge, with this study, we are the first to investigate the effects of a single bout of yoga on sleep-related outcomes on the same night. The most appropriate comparison is with studies in which authors have examined the effect of acute bouts of traditional exercise (i.e., aerobic or resistance activity) on sleep. In a meta-analysis, Kredlow and colleagues (39) found that acute exercise resulted in negligible to small improvements in sleep across a wide variety of variables (e.g., d = 0.22 for TST, d = 0.25 for SE); however, in the acute studies included in the meta-analysis, polysomnography was used to assess sleep.
Our results are clinically meaningful, as neither actigraphy-assessed nor self-reported sleep variables were adversely affected by acute VY. Traditionally, VY is recommended for the morning, while calming and restorative practices are advised later in the day to avoid sleep disruption (17,18). Authors of only a limited number of training studies have designated participants to perform yoga at specific times of the day, and those have primarily focused on gentle forms of yoga, leaving the recommendations largely speculative (19–22). Moreover, authors of previous studies have documented that an acute bout of moderate-intensity exercise in the evening does not negatively impact sleep in individuals with insomnia disorder (40–42). Our study is novel, as we are the first to investigate the effects of VY practiced in the early evening on sleep the same night in those with mild to moderate insomnia symptoms.
In our study, we assessed cardiovascular measures immediately before and after the acute experimental session. We found a significant increase in HR and decrease in diastolic BP, LF HRV, and HF HRV in the yoga group compared with the control condition. Our findings are consistent with those of Brinsley and colleagues (11) and Thrower and colleagues (15), who examined the acute effects of VY on cardiovascular outcomes. Authors of both studies reported an increase in HR at postsession assessment, with Brinsley and colleagues (11) demonstrating that the cardiovascular stimulus of an acute VY session was similar to cycling exercise. Moreover, Thrower and colleagues (15) found reductions in RMSSD and HF HRV immediately following an acute VY session. In our study, alongside previous research, we demonstrated that acute sessions of moderate-intensity yoga elicit a cardiovascular response like traditional exercise. Over time, consistent engagement may contribute to improvements in cardiovascular health (43,44).
Additionally, we assessed nocturnal HRV before and after the acute experimental session. We hypothesized that the VY group would show a greater increase in nocturnal RMSSD HRV than the control group. However, the results did not support our hypothesis; nocturnal RMSSD HRV did not change from baseline to after the experimental session between conditions. Moreover, none of the other nocturnal HRV parameters that were assessed were significantly changed following a single bout of VY. While our hypothesis was not supported, it is important to note that we did not observe any impairments in nocturnal HRV, which would have indicated altered autonomic function (particularly decreased parasympathetic nervous system activity) overnight. In healthy individuals, nonrapid eye movement sleep is a time of parasympathetic dominance, resulting in a decrease in HR and an increase in HF and RMSSD HRV as one progresses from lighter to deeper nonrapid eye movement sleep stages (45,46). Conversely, individuals with insomnia have been found to exhibit lower nocturnal HF HRV and higher LF HRV values across all stages of sleep (45). The present findings align with those focused on traditional aerobic exercise (i.e., cycling, running), in which acute bouts have a minimal impact on nocturnal HRV across various intensities (i.e., light, moderate, vigorous) (47,48). Our findings suggest an acute bout of VY in the early evening likely does not impact parasympathetic function during sleep.
The strengths of our study include using a standardized video with demonstration and verbal cues for yoga sessions in a controlled environment. We recruited participants with insomnia symptoms and assessed sleep through self-reports and device-based measures. The randomized controlled trial design enabled examination of between-groups and within-groups changes under acute experimental conditions.
Despite these strengths, the study had limitations, including a small sample size insufficient for testing multiple outcomes and a predominance of White, female participants, limiting generalizability. Participants recruited for insomnia symptoms were not formally diagnosed with insomnia disorder, potentially explaining minimal baseline sleep impairment and limited improvement postyoga. Participants were generally healthy with normal HR and BP values, restricting significant cardiovascular health improvements after a single yoga session.
CONCLUSIONS
The acute practice of VY in the early evening did not affect sleep or nocturnal autonomic function. Although our hypothesis of improved sleep was unsupported, this indicates that VY can be safely practiced in the evening without affecting sleep or nocturnal cardiovascular function. Authors of future studies should compare VY with gentle yoga to assess the impact of different styles on sleep and cardiovascular health and include longer interventions to help prescribe yoga as complementary therapy.
Summary of participant flow through the study. ISI = Insomnia Severity Index.
Contributor Notes
Conflicts of Interest and Source of Funding: Sally A. Sherman is an ambassador for lululemon.