Recovery sleep following sleep restriction is insufficient to return elevated daytime heart rate and systolic blood pressure to baseline levels

Abstract

Objective:

Sleep restriction alters daytime cardiac activity, including elevating heart rate (HR) and blood pressure (BP). There is minimal research on the cumulative effects of sleep loss and the response after subsequent recovery sleep on HR and BP. This study examined patterns of HR and BP across baseline, sleep restriction, and recovery conditions using multiple daytime cardiac measurements.

Methods:

Participants (15 healthy men, M=22.3 years, SD=2.8) completed an 11-day inpatient protocol with three nights of 10 hours/night baseline sleep opportunity, five sleep restriction nights (5 hours/night sleep opportunity), and two recovery nights (10-hours/night sleep opportunity). Resting HR and BP were measured every two hours during wake. Multilevel models with random effects for individuals examined daytime HR and BP across study conditions and days into the study.

Results:

Mean daytime HR was 1.2±0.5 beats/minute lower during sleep restriction compared to baseline, p<0.001. During recovery, HR was 5.5±1.0 beats/minute higher, p<0.001, and SBP was 2.9±1.1 mmHg higher, p=0.009. When accounting for days into the study (irrespective of condition) and measurement timing across the day, HR increased 7.6 beats/minute and SBP increased 3.4 mmHg across the study period, p<0.001.

Conclusions:

Our findings suggest that daytime HR and SBP increase following successive nights of sleep restriction, even after accounting for measurement time of day. HR and SBP did not recover to baseline levels following two recovery nights of sleep, suggesting longer recovery sleep may be necessary to recover from multiple, consecutive nights of moderate sleep restriction.

INTRODUCTION

It is recommended that healthy adults regularly sleep at least seven hours per night¹, yet only 65% of adults in the United States sleep the recommended amount². Failure to regularly get adequate sleep is a modifiable health risk factor that may contribute to the development of cardiovascular disease³. Insufficient sleep is associated with a number of adverse cardiometabolic outcomes^4,5, including increased risk of obesity⁶, hypertension⁷, and insulin resistance and diabetes^8–10, all of which contribute to the development of cardiovascular disease. Insufficient sleep may increase cardiometabolic risk by elevating basal cardiac activity, such as heart rate (HR) and blood pressure (BP)^11,12.

Sleep insufficiency is associated with elevated daytime HR and BP^7,13–15. Studies of restricted sleep suggest that both degree and duration of sleep restriction contribute to elevation of HR and/or BP^15–20, with more nights of sleep restriction producing greater increases in daytime cardiac activity. Furthermore, elevated daytime cardiac activity due to sleep restriction may not return to baseline after full recovery sleep. HR, but not BP, failed to return to baseline levels after recovery sleep following sleep restricted to 4 hours/night¹⁵. However, an earlier analysis of the same sample only found increased HR during recovery, with HR marginally trending upwards during sleep restriction¹⁴. A similar study found that systolic BP (SBP), but not HR, remained elevated during recovery after more than a week of sleep restricted to 4 hours/night²⁰. Conversely, other studies have not found consistent changes in HR or BP after sleep restriction^15,16,21,22.

Findings from studies with five or fewer days of sleep restriction are mixed. Factors such as the study design and other procedures conducted during the study may have differentially affected whether and how cardiac activity changed during sleep restriction. Most prior studies of sleep restriction measured HR and BP once daily, typically in the morning, and additional procedures may have increased cardiac activity during sleep restriction^23,24. Despite potential confounds, there seems to be a dosage effect of accumulating sleep loss on daytime cardiac activity, possibly due to disruption of underlying autonomic nervous system activity⁷. Specifically, sleep-disrupted autonomic profiles may increase daytime HR and BP above basal levels with each successive night of sleep restriction. The current study adds to the existing literature by measuring HR and BP every two hours during each waking period throughout the duration of the study, including both sleep restriction and recovery, allowing for control of intraday changes in cardiac activity.

This study aimed to investigate whether successive nights of sleep restriction increase HR and BP when also accounting for the measurement timing of cardiac activity. Based on previous studies, we hypothesized that (1) daytime HR and SBP would be elevated during sleep restriction and recovery, compared to baseline levels; (2) daytime HR and SBP would increase with each successive night of additional sleep restriction; and (3) increases in daytime HR and SBP would remain after accounting for clocktime.

METHODS

Participant recruitment and screening

Fifteen healthy men between the ages of 20 and 35 were recruited from the community within the central Pennsylvania, United States region, using several forms of advertisements (e.g., flyers, online advertisements). Participants were excluded if they had ongoing medical or psychiatric disorders or recent history of shift work, travel across more than two time zones within the previous three months, or drug or medication use. All participants provided informed written consent prior to undergoing screening/study procedures, and they were compensated for their participation in the study. During screening, participants underwent a physical examination and were excluded if they had risk factors associated with cardiovascular disease²⁵: a waist circumference > 102 cm²⁶, seated SBP ≥ 130 mmHg, and/or DBP ≥ 85 mmHg^27,28.

Participants were further screened for sleep disorders²⁹ and other medical conditions via questionnaires, and participant urine samples were screened for drug use and other substances. Habitual sleep was assessed via completion of a sleep/wake log, time-stamped call-ins at bedtime and wake time, and by actigraphy (Spectrum, Philips-Respironics, Murrysville, PA) for one week. Finally, a clinical psychologist evaluated each participant on their willingness to comply with study protocols and their ability to tolerate study conditions, resulting in 17 total participants, two of whom discontinued the study and were excluded from analyses. All procedures were conducted in accordance with the Declaration of Helsinki and were approved by the Institutional Review Board of the Pennsylvania State University.

Protocol

Pre-study sleep conditions were implemented to ensure participants maintained the sleep-wake behaviors of the inpatient 11-day protocol. Specifically, participants were instructed to keep the same 22:00 – 8:00 (±1 hour) 10-hours/night time-in-bed schedule at home for one week before beginning the study. Adherence to this schedule was verified by wrist actigraphy, a sleep/wake log, and time-stamped bedtime and wake time call-ins (see ³⁰). Data were collected between April 2016 and April 2018.

Participants stayed for 11 days (10 nights) at the Clinical Research Center at the Pennsylvania State University, University Park campus. Participants lived in sound-dampened, windowless private quarters that were furnished with a bed and desk, with attached bathroom. Study staff were responsible for putting participants to bed, connecting participants to polysomnography, turning lights off, and waking participants at the scheduled times for each condition. Study staff also monitored participant wakefulness throughout scheduled wake periods (excluding use of the bathroom and while showering). Participants were exposed to dim light conditions (<100 lux in the direction of gaze) during scheduled wake periods and darkness (0 lux) during scheduled sleep opportunities. Temperatures were maintained between 20°C and 22°C. During scheduled wake periods, participants were not permitted to sit or lay down on the bed (except during study-specific procedures) or to exercise except for light stretching. Participants were permitted to drink water ad libitum. The study dietician designed the controlled, in-laboratory diet using weighed foods with predetermined macronutrient and micronutrient content, including limited sodium intake (see ^31,32), thereby minimizing potential diet-related changes in cardiac activity. Meals were administered at 09:00, 12:00, and 18:00 each day (excepting metabolic procedures). Participants were given 30 minutes to complete their meals and no HR or BP measures were taken for at least 30 minutes after meal completion. Salivary cortisol was sampled from participants every fifteen minutes for the first hour after waking and then every hour during wakefulness on days when saliva sampling did not interfere with other procedures.

Participants underwent three sleep conditions, with sleep duration verified by polysomnography (see Figure 1 and Supplemental Digital Content, Table S1 for differences by condition). During the baseline condition, the first three nights of the protocol, participants had the opportunity to sleep between 22:00 and 08:00. On the fourth night, participants entered the sleep restriction condition, during which they had the opportunity to sleep between 00:30 and 05:30 each night for five consecutive nights. This was followed by the recovery condition for the final two nights, during which they again had the opportunity to sleep between 22:00 and 08:00. The sleep midpoint, an index of circadian timing of sleep, was maintained at 03:00 irrespective of condition.

Figure 1. Visualization of the 11-day Sleep Restriction Study protocol.

Black bars represent the sleep opportunity interval in three different conditions: Ten hours (22:00-08:00) for three nights during the baseline condition (BL), followed by five hours (00:30-05:30) for five nights during the sleep restriction condition (SR), and ten hours (22:00-08:00) for two nights during the recovery condition (REC). Heart symbols denote approximately when heart rate and blood pressure were measured throughout the day. Solid black lines denote approximate overlap with blood collection and intravenous glucose tolerance test procedure.

Heart rate and blood pressure

HR and BP were used to measure daytime cardiac activity in participants. HR was measured in beats per minute (BPM) and SBP and DBP were measured in millimeters of mercury (mmHg). Measurements of the brachial artery in the upper arm were taken every two hours during wakefulness using a calibrated device (Dinamap Pro 100, GE Medical Systems Information Technologies, Inc., Milwaukee, Wisconsin). Participants were seated during each measurement to control for effects of posture on cardiac activity³³.

The means of each participant’s measurements across conditions and across days were calculated. Person-weighted means and standard errors of the mean (SE) were then calculated by averaging the mean of each participant’s measurements across conditions (Supplemental Digital Content, Table S2) or days (Figure 2A–C, Supplemental Digital Content, Table S3).

Figure 2. Heart rate (A), systolic blood pressure (B), and diastolic blood pressure (C) across the study.

Black lines show person-weighted mean cardiac activity on each day of the study (A-C) and predicted cardiac activity on each day of the study at 12:00 after accounting for wake-centered time and clocktime² (Model 5) across the study (D-E).

Dotted lines show standard error of the mean.

Note. SR = Sleep restriction, REC = Recovery

Time of measurements

HR and BP were measured approximately every two hours during wake each day³⁴. Due to the nature of the protocol, participants were awake longer during days with sleep restriction and therefore had more measurements throughout the day (up to 11 measurements). Staff documented the date and time at which HR and BP measurements were taken. Measurements from the first partial day of the study (i.e., when participants were admitted) were excluded to account for any changes in cardiac activity attributable to transitioning into the study (e.g., dietary change), but daytime measurements following the first night of inpatient sleep were included. All other baseline measurements were used (henceforth referred to as “baseline”) as a comparator for other days in the study (e.g., baseline [BL] vs. sleep restriction day 4 [SR4] or recovery day 1 [REC1]). Values measured between 00:00 and 00:45 were considered part of the prior day. A quadratic term for days (i.e., days*days = days²) was computed using baseline-centered days (see below) to account for a potential curvilinear relationship between days into the study and cardiac activity.

Recorded time of measurements was converted to decimal 24-hour clocktime for analysis. We then calculated wake-centered time by centering 24-hour clocktime around waketime during baseline (08:00), with values greater than 0 representing the number of hours after waketime during baseline. A quadratic term for time (i.e., clocktime*clocktime = clocktime²) was computed using wake-centered time to account for a curvilinear relationship between time of day and cardiac activity.

Statistical analysis

Analyses were conducted using PROC MIXED (SAS 9.4, Cary, NC) in a multilevel modeling framework, which has robust statistical power to handle randomly distributed missing data³⁵. Five mixed models were used to test whether there was a difference of HR, SBP, and DBP across time. Model 1 was an empty, intercept-only model to examine the initial fit of cardiac parameters within a multilevel model and was assessed using the intraclass correlation coefficient and reliability. Model 2 tested differences between conditions (i.e., baseline vs. sleep restriction vs. recovery), with p-values adjusted for multiple comparisons using the Tukey-Kramer post-hoc test. Model 3 tested the linear effect of days into the study. Model 3A tested the linear effect of days into the study and the quadratic effect of days², which was included in subsequent models if significantly associated with the outcome. Model 4 tested the linear effect of days into the study and the linear effect of wake-centered time during sleep restriction. Finally, Model 5 included days into the study, wake-centered time, and clocktime² to test whether there was a curvilinear relationship between time of day and cardiac activity. Condition and subject were defined as categorical variables. Days, days², wake-centered time, clocktime², HR, SBP, and DBP were all treated as continuous variables. Models 2-5 accounted for participants’ pre-study habitual sleep duration and BMI, both of which were centered around the sample mean.

Study condition with compound symmetric covariance structure of subjects was treated as a repeated effect in Model 2. Random intercepts were included for subjects across all models. We used an alpha level of p<0.05 (two-tailed) to determine whether each model statistically significantly fit the data, and Tukey-Kramer adjusted p-values were evaluated for Model 2, due to multiple pairwise comparisons of condition. We also used the ratio of difference in covariance between models (i.e., pseudo-R²) to calculate the amount of variability accounted for by primary predictors³⁵. Residuals from every model were normally distributed (skewness < |3|; kurtosis < |10|). Data and analysis script will be made available upon request.

In addition to the above analyses, we conducted sensitivity analyses to assess the effect of blood draws on HR and SBP (see Figure 1). We used bivariate correlations to assess the associations of percentage change in hematocrit and hemoglobin levels with change in cardiac activity across conditions. There were no significant correlations, as shown in Supplemental Digital Content, Table S4. We also assessed whether salivary cortisol (μg/dL) was associated with cardiac activity by including cortisol as a covariate in Models 3-5. Cortisol levels were not associated with HR or SBP in these models, as shown in Supplemental Digital Content, Tables S5 and S6, but were associated with increased DBP when accounting for clocktime, as shown in Supplemental Digital Content, Table S7. Finally, shown in Supplemental Digital Content, Table S8, we tested Model 2 using only measurements that were common across all conditions (i.e., between 08:00 and 22:00), with p-values adjusted for multiple comparisons using the Tukey-Kramer post-hoc test.

RESULTS

Descriptive statistics

Fifteen healthy men completed the study (mean±SD age: 22.3±2.8 years; BMI 24.7±3.0 kg/m²). Two individuals discontinued participation during the study and were excluded from all analyses. Descriptive statistics of participants’ HR and BP are shown by condition in Table S2 and by days into the study in Figure 2A–C and Table S3; one participant was missing one measurement of HR on the second sleep restriction day (SR2). Overall, the person-weighted mean±SE of HR was 69.2±1.8 BPM; SBP was 117.5±2.3 mmHg; and DBP was 64.9±1.7 mmHg. All measurements of cardiac activity were normally distributed.

Heart Rate

Between-person differences accounted for approximately 40.8% of the variability in HR with a reliability of 0.98 over an average of 77 observations per person (Model 1).

Condition.

There was an effect of sleep condition on HR (Model 2), as shown in Table S2. HR was 1.2±0.5 BPM lower during sleep restriction than during baseline, p=0.014, whereas HR during recovery was 5.5±1.0 BPM higher than baseline, p<0.001. HR was 6.7±1.0 BPM higher during recovery than during sleep restriction, adjusted p<0.001. The effect of conditions accounted for 7.4% of the within-person variability in HR.

Days into study.

Results from the association between days into the study and HR are shown in Table 1. HR increased nearly 1 BPM with each successive day into the study relative to baseline (Model 3), p<0.001. When accounting for both days into the study and days² (Model 3A), HR increased overall with each successive day into the study, p<0.001. Predicted HR was 69.0±1.7 BPM during baseline but was 77.7±2.4 BPM by the second day of recovery, REC2, suggesting that HR had not returned to baseline values after two nights of recovery sleep. The linear and quadratic effects of days into the study accounted for 9.8% of the within-person variability in HR.

Table 1.

Associations of time on heart rate and systolic blood pressure

	Heart rate (BPM)				Systolic blood pressure (mmHg)
	Model 3a	Model 3Ab	Model 4c	Model 5d	Model 3a	Model 4e	Model 5f
	b±SE	b±SE	b±SE	b±SE	b±SE	b±SE	b±SE
Intercept	66.7±1.6***	69.0±1.7***	70.3±1.6***	69.5±1.6***	116.2±1.8***	115.1±1.6***	114.9±1.6***
Days into study	0.9±0.2***	−1.9±0.4***	−2.1±0.4***	−1.6±0.4***	0.5±0.1**	0.5±0.1***	0.5±0.1***
Days2	-	0.5±0.1***	0.5±0.1***	0.4±0.1***	-	-	-
Wake-centered time	-	-	−0.2±0.1*	0.6±0.2**	-	0.2±0.1*	0.6±0.3*
Clocktime 2	-	-	-	−0.1±0.0***	-	-	−0.03±0.0†
Pseudo-R 2	0.05	0.10	0.11	0.16	0.01	0.02	0.03
Obs	1158				1159
N	15				15
Obs = total observations BPM = beats per minute mmHg = millimeters of mercury	a. Days into study b. Days into study, days2 c. Days into study, days2, wake-centered time d. Days into study, days2, wake-centered time, clocktime2				a. Days into study e. Days into study, wake-centered time f. Days into study, wake-centered time, clocktime2

Note.

^***p<.001

^**p<.01

^*p<.05

^†p<.10

all linear mixed models accounted for mean-centered habitual sleep duration and BMI

Days into study and clocktime.

Examining the effects of days into the study, days², and wake-centered time on HR (Model 4), shown in Table 1, accounted for 11.2% of the within-person variability in HR and showed main effects of days, days², and wake-centered time. HR increased overall with each successive day into the study, p<0.001. Finally, as shown in Figure 2D, we examined the effects of days into the study, days², wake-centered time, and clocktime² on HR (Model 5), which accounted for an overall 15.9% of the within-person variability in HR. HR increased overall with each successive day into the study, p<0.001, and decreased overall across the day, p≤0.005. For example, predicted HR was 68.3±1.5 BPM at 08:00, 69.8±1.6 BPM at 12:00, and 66.0±1.8 BPM at 21:00 on SR1, whereas HR was 71.1±1.7 BPM at 08:00, 72.6±1.8 BPM at 12:00, and 68.8±2.0 BPM at 21:00 on SR5.

Systolic Blood Pressure

Between-person differences accounted for approximately 46.1% of the variability in SBP with a reliability of 0.99 over an average of 77 observations per person (Model 1).

Condition.

We tested whether daytime SBP would be elevated during sleep restriction and recovery, compared to baseline levels (Model 2). SBP was 2.9±1.1 mmHg higher during recovery, p=0.009, than during baseline, as shown in Table S2, but was only marginally higher during sleep restriction (1.4±0.8 mmHg), p=0.072. Tukey-Kramer post-hoc comparisons revealed no differences in SBP between sleep restriction and recovery.

Days into study.

Results from the association between days into the study and SBP are shown in Table 1. SBP increased approximately 0.5 mmHg with each successive day into the study (Model 3), p=0.002. Predicted SBP was 116.2±1.8 mmHg during baseline and was 119.5±1.9 mmHg by REC2, suggesting a modest increase in SBP across the study. There was no effect of days² (Model 3A) so days² was not included in subsequent models.

Days into study and clocktime.

We then examined the effects of days and wake-centered time on SBP (Model 4), shown in Table 1. SBP increased with each successive day into the study, p<0.001, and increased overall across the day for each hour past waketime, p=0.025. Finally, we included clocktime² in addition to days and wake-centered time (Model 5), as shown in Figure 2E. There continued to be main effects of days, p<0.001, and wake-centered time, p=0.027, but clocktime² was only marginally associated with SBP, p=0.076. While accounting for both wake-centered time and clocktime², SBP increased approximately 0.5 mmHg with each successive day. For example, predicted SBP at 12:00 was 116.6±1.7 mmHg during baseline but increased to 117.1±1.7 mmHg by SR1, 119.0±1.9 mmHg by SR5, and 120.0±2.0 mmHg by REC2. Including the effects of days, wake-centered time, and clocktime² accounted for 2.8% of the within-person variability in SBP.

Diastolic Blood Pressure

Between-person differences accounted for approximately 53.9% of the variability in DBP with a reliability of 0.99 over an average of 77 observations per person (Model 1). There was no effect of condition on DBP (Model 2). DBP did not change across days into the study (Model 3). There was no effect of days² (Model 3A) and days² was not included in subsequent models. When accounting for wake-centered time (Model 4), there were main effects of wake-centered time, p=0.011, but not days, p=0.143. DBP increased 0.1±0.0 mmHg for each hour past waketime. Finally, we included clocktime² (Model 5), which was not associated with DBP, p=0.091. There were no main effects of wake-centered time, p=0.377, or days, p=0.090, on DBP after accounting for clocktime². Including days of sleep restriction, wake-centered time, and clocktime² accounted for 2.1% of the within-person variability in DBP.

DISCUSSION

Our study examined the effects of 5 hours/night sleep restriction for five nights on three measures of cardiac activity (HR, SBP, and DBP) using better temporal resolution of measurements compared to previous research. By measuring cardiac activity every two hours during every waking period throughout the duration of the study, we were able to account for measurement timing and reveal the effects of sleep restriction and subsequent recovery on cardiac activity compared to baseline. We found that HR decreased slightly during sleep restriction but was elevated in the recovery condition, and that HR increased overall across the duration of the study. SBP increased with each successive day into the study. Both HR and SBP patterns remained significant after accounting for the curvilinear intraday effect of measurement timing. DBP did not change across conditions or with each successive day into the study. Most importantly, neither HR nor SBP recovered to baseline levels following the two nights of recovery sleep. Recovery sleep of longer duration and/or more nights may be necessary to recover from multiple, consecutive nights of moderate sleep restriction.

HR increased across the duration of the study, primarily due to elevations during recovery from sleep restriction. This finding aligns with prior studies that found increased HR after five or more nights of sleep restricted to fewer than five hours/night^14–17, notably studies that showed that HR was elevated during recovery compared to baseline^14,15. Based on these studies, we hypothesized that HR would increase during the sleep restriction condition, but we instead found that HR slightly decreased. Other studies that examined five or fewer nights of sleep restriction found either no increases^22,36 or marginal decreases in HR³⁷. Although we do not yet know the precise mechanism underlying this change, we speculate that decreases in HR during sleep restriction may serve as a compensatory mechanism that is eventually overwhelmed by increased sleep drive. Alternatively, lower HR during sleep restriction may be a lagged indicator of participants being sleep replete following the baseline condition. This phenomenon should be investigated in future pathophysiological studies examining HR during sleep restriction.

We found that SBP increased across the study in contrast to prior studies that generally found no increases in SBP during sleep restriction, most of which measured SBP one time in the morning^{14,15,17,36,37}. One prior study also found increases in SBP but only after seven nights of sleep opportunity restricted to fewer than 4.2 hours/night¹⁶. Another study examined four blocks of sleep restricted to 4 hours/night for three nights, followed by 8 hours of recovery sleep for one night, and they found increased SBP during the first block of sleep restriction but not during subsequent blocks²². More recent studies found that SBP increased after a week or more of mild sleep restriction (1.5 fewer hours/night)^18,19 and severe sleep restriction (4 hours/night)²⁰.

DBP marginally increased across the study after accounting for measurement timing. Only one other study found increased daytime DBP during sleep restriction²². The findings of that prior study may be due to measuring BP using continuous beat-to-beat monitoring of arteries in the fingers, which differs from BP measured using the brachial artery of the upper arm, as in the current study³⁸. Overall, our study corroborates the findings of other studies that found no change in DBP during sleep restriction or recovery from sleep restriction^14–17.

Contribution

This study substantially contributes to the literature because it examined whether HR and SBP increase with successive nights of sleep restriction using measurements that assess bihourly changes in daytime HR and BP. Many prior studies measured cardiac activity only once per day, which captures day-to-day variation but fails to account for the timing of measurements across the day. By comparison, we measured cardiac activity during wakefulness every two hours throughout this study, which exceeds current semi-automated BP monitoring minimum recommendations³⁴. We therefore were able to account for measurement timing and diurnal effects that may otherwise obfuscate acute changes in cardiac activity in healthy adults.

It is also paramount to use protocols that reflect the reality of the national sleep crisis to understand potential antecedents of cardiovascular disease^4,5. Indeed, our protocols tested the effect of five nights of sleep restriction to five hours/night, a duration that may be representative of what people experience during the work week in everyday life compared to prior studies with shorter nighttime sleep opportunity. Specifically, adults in the United States sleep on average 6.8 hours per night on weekdays, with approximately 12% sleeping fewer than five hours^2,39. Studies that employ sleep restriction protocols of only four hours of sleep per night capture the experience of only a small proportion of adults². Five hours/night sleep restriction protocols, therefore, may strike a balance of capturing changes in cardiac activity due to sleep loss while better representing typical self-restricted sleep duration.

Limitations

There are several major limitations. This study was limited to young, healthy adult men, making it comparable to most sleep restriction studies examining cardiac activity but not as generalizable as studies that include both men and women^20,22,37. We speculate that populations with persistently elevated cardiac activity may experience comparatively greater increases in HR and BP. Furthermore, the absence of a control group challenges our ability to disentangle the true effects of sleep restriction from effects due to experimental procedures unrelated to cardiac activity (see Figure 1). We attempted to account for these factors by including clocktime variables in analyses and found that both HR and SBP continue to increase during the sleep restriction condition. The inclusion of clocktime variables also serves to account for changes in HR or BP due to circadian timing^40,41. Future studies should attempt to replicate these findings across both men and women and with comparison to an independent control group.

This study is also limited by other factors, most of which are consistent with prior studies^14–17. Systematic behavioral (e.g., walking or posture before measurements) or environmental factors (e.g., recent meals, experimental procedures) may have confounded the changes seen in cardiac activity throughout the study. The data examined here are part of a larger study on the effects of sleep restriction on metabolism, some of the results of which have already been published (see^31,32). Repeated frequent blood draws were taken via intravenous catheters during baseline and on SR4, SR5, REC1 and REC2, which correspond with increased HR, as seen in Figure 2A. Specifically, intravenous catheters were used during a standardized meal test on the evenings of BL3, SR4, and REC1 and again during intravenous glucose tolerance tests (IVGTTs) on the mornings of BL4, SR5 and REC2. No HR or BP measurements were taken during the IVGTTs. Blood draws during most of the standardized meal test were spaced at 30-minute intervals; any BP and HR measurements during the procedure were recorded at least 15 minutes after the most recent blood draw. Decreased circulating levels of red blood cells due to blood draws may have driven the changes seen in HR^23,24, although the total amount of blood drawn across the study was less than a pint. Hematocrit and hemoglobin levels were monitored throughout the study, but changes in levels were not associated with cardiac activity. Salivary cortisol levels were also monitored but were not associated with the changes seen in cardiac activity. However, other related factors, such as discomfort during blood draws, may have contributed to elevated cardiac activity. Results of the sensitivity analyses are shown in Tables S3–S6.

Future studies may expand upon this study by measuring HR and BP throughout the entire day and night during sleep restriction. Doing so may allow for the identification of shifts in circadian timing or attenuation of rhythms (e.g., blunted nocturnal dipping of BP) due to accumulated sleep debt^22,42. Future studies should also include additional cardiac parameters, such as pre-ejection period and high-frequency heart rate variability, which are orthogonally mediated by the sympathetic and parasympathetic branches of the autonomic nervous system, respectively. Autonomic function is associated, at least in part, with HR and BP⁴³, and may explain the association between insufficient sleep and cardiac activity. Finally, future studies should be conducted with comparison groups and fewer study procedures unrelated to cardiac activity measurement (e.g., blood draws) to better ascertain the effects of sleep restriction on HR and BP²¹.

Conclusion

Our study demonstrates that sleep restriction is a challenge to the cardiovascular system in the absence of other perturbations like exercise, cardiometabolic dysfunction, or interpersonal stress. Indeed, we found changes in cardiac activity after accumulating sleep loss via a sleep restriction protocol, the consequence of which over time may precipitate cardiovascular disease. Furthermore, this study reproduces findings that HR is elevated during recovery but not during sleep restriction^14–16, while providing novel evidence that HR increases with successive nights of sleep restriction when accounting for time of measurement. Most interestingly, both HR and SBP failed to recover to baseline levels after two nights of recovery sleep following sleep restriction, suggesting that a weekend of catch-up sleep after a work week of sleep restriction may not be sufficient recovery for this population (i.e., young, healthy adult men).

Abbreviations:

HR

heart rate

BP

blood pressure

SBP

systolic blood pressure

DBP

diastolic blood pressure

BPM

beats per minute

mmHg

millimeters of mercury

BL

baseline

SR

sleep restriction

REC

recovery

IVGTT

intravenous glucose tolerance test