Actigraphy-Derived Sleep Efficiency Is Associated With Endothelial Function in Men and Women With Untreated Hypertension

Abstract

BACKGROUND

Poor sleep quality is increasingly recognized as an important and potentially modifiable risk factor for cardiovascular disease (CVD). Impaired endothelial function may be 1 mechanism underlying the association between poor sleep and CVD risk. The present study examined the relationship between objective measures of sleep quality and endothelial function in a sample of untreated hypertensive adults.

METHODS

Participants were 127 men (N = 74) and women (N = 53), including 55 African Americans and 72 White Americans, aged 40–60 years (mean age, 45.3 ± 8.5 years), with untreated hypertension (systolic blood pressure 130–159 mm Hg and/or diastolic blood pressure 85–99 mm Hg). Noninvasive brachial artery flow-mediated dilation (FMD) was assessed by ultrasound. Sleep parameters, including sleep efficiency (SE), total sleep time (TST), and subjective sleep quality, were assessed over 7 consecutive days by wrist actigraphy.

RESULTS

Participants averaged 7.76 ± 1 hours in bed, with an average SE of 78 ± 9%, and TST of 6 ± 1 hours. Brachial FMD averaged 3.5 ± 3.1%. In multivariate analyses controlling for sex, race, body mass index, clinic blood pressure, income, smoking, alcohol use, and baseline arterial diameter, SE was positively associated with FMD (β = 0.28, P = 0.012). Subjective sleep quality (β = −0.04, P = 0.63) and TST (β = −0.11, P = 0.25) were unrelated to FMD.

CONCLUSIONS

Poor sleep as indicated by low SE was associated with impaired FMD. These findings for SE are consistent with previous observations of other measures implicating poor sleep as a CVD risk factor. Interventions that improve sleep may also help lower CVD risk.

Poor sleep and insomnia have become recognized as important, and potentially modifiable, risk factors for development of cardiovascular disease (CVD) and its associated mortality.^1–3 Elevated blood pressure (BP) and changes in underlying vascular function may be 1 pathway through which poor sleep quality contributes to increased CVD risk.^4,5 Recent meta-analytic findings, spanning 22 studies and over 45,000 subjects, reported that poor subjective sleep quality was associated with greater average systolic and diastolic BP, as well as an increased likelihood of hypertension.⁶ Further, studies have shown that poor sleep quality, assessed both subjectively (i.e., by self-report) and objectively (i.e., via polysomnography and/or actigraphy), is associated with increased nighttime BP and BP nondipping.^7–12

In addition to its association with BP, growing evidence suggests that poor sleep quality is associated with impaired vascular endothelial function.¹³ For example, poorer subjective sleep quality has been associated with diminished forearm vasodilation capacity¹⁴ and elevated systemic vascular resistance during the nighttime sleep period.¹⁵ Impaired endothelium-dependent vasodilation has been characterized as a primary manifestation of endothelial dysfunction, an early indicator of atherosclerosis that is further associated with an increased risk of future CVD events, including myocardial ischemia and acute coronary syndromes.¹⁶ Studies in patients with obstructive sleep apnea, and other clinical populations have shown that poor sleep quality is associated with decreased flow-mediated dilation (FMD).¹⁷ Moreover, a recent systematic review reported that while poor sleep quality was generally associated with impaired endothelial function in relatively healthy cohorts, the assessment of sleep quality was highly heterogeneous, with most studies employing only subjective (i.e., self-report) measures.¹³

Given the limited nature of the existing literature, the present study aimed to investigate the association between objective measures of sleep using wrist actigraphy, and endothelial function assessed by brachial artery FMD in men and women with untreated hypertension. We hypothesized that poor sleep, as defined by lower sleep efficiency (SE) and shorter total sleep time (TST) would be associated with lower FMD.

METHODS

Participants were 127 men (N = 74) and women (N = 53) between the ages of 40 and 60 years (M = 45.3 ± 8.5 years), including 55 African Americans and 72 White Americans. BP inclusion criteria were clinic systolic BP 130–159 mm Hg and/or diastolic BP 85–99 mm Hg, with all participants meeting the 2017 ACC/AHA criteria for hypertension.¹⁸ Exclusion criteria were body mass index >35 kg/m²; age <40 years or >60 years; current use of BP or CVD medications; diabetes mellitus; previously diagnosed obstructive sleep apnea; pacemaker; atrial fibrillation; myocardial infarction, percutaneous coronary intervention, or coronary artery bypass graft surgery within 6 months of enrollment; heart failure; severe uncorrected primary valvular disease; uncorrected thyroid heart disease; oral contraceptive use; pregnancy; hormone replacement therapy; alcohol or drug abuse within 12 months; renal or hepatic dysfunction; dementia; inability to comply with the assessment procedures; and inability to provide informed consent. In addition to these exclusion criteria, participants were further screened for sleep apnea, insomnia, hypersomnia, sleep–wake disorders, narcolepsy, and additional parasomnias using the Duke Structured Interview for Sleep Disorders.^10,19,20 Participants were recruited by advertisements in the Piedmont region of North Carolina. The study protocol was approved by the Institutional Review Board at Duke University Medical Center. All eligible individuals provided written informed consent prior to participation in the study.

Clinic BP screening

Clinic BP was determined on 3 separate visits, each approximately 1 week apart. After 5 minutes seated in a quiet, temperature-controlled room, 4 seated BP readings, each 2 minutes apart, were taken using a mercury sphygmomanometer and stethoscope. Systolic BP and diastolic BP for each visit were calculated as the means of the last 3 readings, and clinic BP eligibility was based upon whether the average of the 3 mean office BP readings met the study’s inclusion criteria.

Measures of sleep

Mini-Mitter Actiwatch wrist-watch style actigraphs (Mini-Mitter, Sunriver, OR) were used to derive objective estimates of sleep parameters. Actigraphy-derived sleep parameters have been shown to yield comparable estimates to measures obtained via polysomnography.²¹ The Actiwatch contains a calibrated accelerometer and a memory storage apparatus, housed in a casing that, in size and shape, resembles a wristwatch. The accelerometer samples movement/activity at a rate of 32 times per second. 24-Hour actigraphy data were recorded over 7 consecutive days. Total time in bed was measured using sleep logs and confirmed by actigraphy. Data from the actigraph were used to derive SE (the percent of time asleep during the sleep period participants allocated for themselves to be sleeping) and TST (the total time asleep during the sleep period). Subjective sleep quality also was assessed using the Pittsburg Sleep Quality Index (PSQI).²²

Assessment of vascular endothelial function

FMD of the brachial artery was assessed in the morning, following overnight fasting. Longitudinal B-mode images of the brachial artery, in the region 4–6 cm anterior to the antecubital fossa, were obtained using a 7–11 MHz linear-array transducer and Aspen ultrasound system. Images were captured by the same sonographer under the following conditions: (i) after 10 minutes of supine rest, (ii) during the first 120 seconds of reactive hyperemia, achieved by inflation of a pneumatic occlusion cuff, located around the forearm, to suprasystolic pressure (~200 mm Hg) for 5 minutes. End-diastolic images were stored and arterial diameters were measured as the distance between the proximal and distal arterial wall intima-media interfaces using PC-based software (Brachial Analyzer—Version 5.0, Medical Imaging Applications LLC, Iowa City, IA). Baseline arterial diameter was assessed from measurements obtained during the supine rest period. Peak reactive hyperemia response was assessed from 10 to 120 seconds postdeflation of the cuff, with peak arterial diameter quantified using polynomial curve fitting. FMD was defined as the maximum change in arterial diameter during reactive hyperemia relative to baseline diameter, expressed as a percentage.

Data reduction and statistical analysis

Sleep parameters were calculated for each of the 7 actigraphy monitoring days and averaged to yield a robust estimate of each sleep parameters for each participant. Means, SDs, and/or percentages were computed for demographic variables, clinic BP, baseline arterial diameter, FMD, and measures of sleep quality. Multiple linear regression analyses were used to examine the association of SE, TST, and subjective sleep quality with FMD in a model containing age, sex, race, body mass index, smoking, alcohol use in the past 30 days, clinic systolic BP, and baseline arterial diameter. All statistical analyses were conducted using the SAS system (SAS 9.4, SAS Institute, Cary, NC) with significance set at P = 0.05.

RESULTS

Study sample characteristics

The characteristics of the study sample are shown in Table 1. The 127 participants included 31 African American women, 24 African American men, 22 White American women, and 50 White American men, with a mean age of 45.3 ± 8.5 years, a mean clinic systolic BP of 140 ± 7, and clinic diastolic BP of 89 ± 5 mm Hg. The average baseline arterial diameter was 4.4 mm and the average FMD was 3.5%. Participants spent an average of 7.7 hours in bed, exhibited a mean SE of 78% and an average of 6.0 hours of TST.

Table 1.

Sample characteristics (mean ± SD, or %)

N	127	—
Age (years)	45.35	8.49
Sex (% male)	58%	—
Race (% African American)	43%	—
BMI (kg/m2)	28.14	3.69
Smoking (%)	11%	—
Alcohol use (%)	69%	—
Clinic SBP (mm Hg)	140.14	7.28
Clinic DBP (mm Hg)	89.66	5.22
% classified as “nondipper”	35%	—
Baseline arterial diameter (mm)	4.39	0.70
Flow-mediated dilation (mm)	3.51	3.11
Subjective sleep quality	5.8	3.61
Sleep efficiency (%)	78.39	9.14
Total time in bed (hours)	7.66	1.00
Total sleep time (hours)	6.01	1.04

Abbreviations: BMI, body mass index; DBP, diastolic blood pressure; SBP, systolic blood pressure.

Multivariate associations with FMD

To examine the association of sleep parameters (SE, TST, and PSQI) with FMD, multiple regression models, controlling for demographics, health behaviors, and brachial artery baseline diameter were conducted, examining each sleep parameter independently. As shown in Table 2, SE exhibited a significant association with FMD, with regression parameters indicating that higher SE was related to increased FMD (β = 28, P = 0.012). However, neither TST nor subjective sleep quality (PSQI) was related to FMD, as shown in Tables 3 and 4, respectively.

Table 2.

Regression model of sleep efficiency (SE) predicting FMD

	B	SE	P
Age (years)	−0.05	0.04	0.58
Sex (M/F = 0/1)	0.27	0.87	0.054
Race (B/W = 0/1)	0.12	0.68	0.25
BMI (kg/m2)	0.09	0.08	0.35
Smoking (N/Y = 0/1)	0.07	0.99	0.48
Alcohol use (N/Y = 0/1)	0.15	0.66	0.11
SBP (mm Hg)	−0.14	0.04	0.14
Arterial diameter (mm)	−0.32	0.60	0.02
Sleep efficiency (%)	0.28	0.04	0.012

Abbreviations: BMI, body mass index; FMD, flow-mediated dilation; SBP, systolic blood pressure.

Table 3.

Regression model of total sleep time (TST) predicting FMD

	B	SE	P
Age (years)	−0.02	0.04	0.85
Sex (M/F = 0/1)	0.22	0.89	0.12
Race (B/W = 0/1)	−0.03	0.66	0.76
BMI (kg/m2)	0.09	0.08	0.35
Smoking (N/Y = 0/1)	−0.02	0.94	0.82
Alcohol use (N/Y = 0/1)	0.13	0.68	0.15
SBP (mm Hg)	−0.13	0.04	0.18
BAD (mm)	−0.34	0.61	0.02
TST (hour)	−0.11	0.31	0.25

Abbreviations: BAD, baseline arterial diameter; BMI, body mass index; FMD, flow-mediated dilation; SBP, systolic blood pressure.

Table 4.

Regression model of subjective sleep quality (PSQI) predicting FMD

	B	SE	P
Age (years)	−0.01	0.04	0.92
Sex (M/F = 0/1)	0.25	0.89	0.08
Race (B/W = 0/1)	0.01	0.63	0.95
BMI (kg/m2)	0.10	0.08	0.32
Smoking (N/Y = 0/1)	−0.03	0.94	0.75
Alcohol use (N/Y = 0/1)	0.11	0.67	0.21
SBP (mm Hg)	−0.13	0.04	0.18
Arterial diameter (mm)	−0.34	0.61	0.02
Subjective sleep quality	−0.04	0.08	0.63

Abbreviations: BMI, body mass index; FMD, flow-mediated dilation; PSQI, Pittsburg Sleep Quality Index; SBP, systolic blood pressure.

For purposes of illustration, a follow-up analysis of covariance model was conducted to show differences in FMD according to different levels of SE, by dividing SE into tertiles yielding relative low (M = 67.37%, SD = 7.38%), moderate (M = 79.87%, SD = 1.93%), and high (M = 85.99%, SD = 2.55%) groups. As depicted in Figure 1, there was a dose response relationship between SE and FMD, F(2,124) = 4.41, P = 0.038. Of note, relative to participants with the lowest SE, participants with the highest SE exhibited significantly greater FMD (2.3% vs. 4.4%, P = 0.006).

Figure 1.

FMD (%) shown by tertiles of sleep efficiency in our untreated hypertensive sample (values presented as mean + standard error). FMD was significantly higher among individuals in the highest sleep efficiency tertile (M = 4.45, SE = 0.48) relative to those in the lowest tertile (M = 2.27, SE = 0.54), P = 0.006. FMD for the middle sleep efficiency tertile (M = 3.71, SE = 0.48) was marginally higher than the lowest tertile, P = 0.055. Abbreviation: FMD, flow-mediated dilation.

DISCUSSION

Poor sleep and insomnia are increasingly recognized as risk factors for the development of CVD morbidity and mortality.^23,24 Impaired vascular endothelial function has been suggested as a potential mechanism underlying this association.¹³ In the present study, low SE, an actigraphy-derived measure of poor sleep, was associated with impaired FMD, after controlling for age, sex, race, body mass index, clinic BP, and baseline arterial diameter. However, neither TST measured by actigraphy, nor subjective sleep quality measured by the PSQI, were associated with FMD.

Few studies have examined the relationship between objective measures of sleep and endothelial function in nonclinical populations. Notably, Cooper et al. found that a greater proportion of rapid eye movement sleep (assessed via polysomnography) was positively associated with FMD.²⁵ This finding is consistent with the notion that more consolidated sleep, as reflected in high SE, is associated with a greater FMD response. More recently, Hall et al.²⁶ reported that short sleep duration, assessed via polysomnography, prospectively predicted lower FMD over an average follow-up period of 19 years, in a sample of initially healthy adults; however, the effect for FMD was attenuated after adjusting for demographic, psychosocial, and CVD risk factors. Although we did not find sleep duration (TST) to be related to FMD, our results extend these previous findings by demonstrating that low SE is associated with impaired FMD (and conversely, high SE is associated with greater FMD), independently of demographic and clinical characteristics. Fifteen participants in our sample had insomnia disorder, which is one common source of low SE. When chronic and/or co-occurring with objectively short sleep duration, insomnia is associated with hypertension²⁷ and CVD risk.²⁸ This offers a modifiable target for increasing SE, and perhaps improving FMD, as insomnia is treatable.

Limitations of our study include its cross-sectional design, precluding any cause–effect inferences regarding the positive association of FMD with SE. Although obstructive sleep apnea was an exclusion criterion, in the absence of a polysomnographic sleep study, it is possible that some participants had undiagnosed sleep apnea. Although the somewhat restrictive inclusion/exclusion criteria helped eliminate potential confounders, they also should be recognized as limiting the generalizability of the study findings to a broader population of adults with various comorbidities. Strengths of the current study include the objective assessment of sleep quality using actigraphy over 7 consecutive days as a basis for determining characteristic sleep parameters for participants. Also, our entire sample was middle-aged, ethnically diverse, and hypertensive, suggesting that the positive association observed between SE and FMD was independent of age, race/ethnicity, and elevated clinic BP.

Endothelial dysfunction has been characterized as a summative measure encompassing the accumulative effects of traditional CVD risk factors (i.e., smoking, hypertension, and hyperlipidemia), as well as local hemodynamic forces (i.e., shear stress) and other potentially adverse influences.¹⁶ Previous research suggests that poor sleep quality may contribute to impaired endothelial function through several pathways including disruption of metabolic, immune, and autonomic regulatory processes.²⁵ There is some evidence that both metabolic (i.e., triglycerides, glucose, insulin, and hemoglobin A1c) and immune (i.e., C-Reactive Protein) functioning are improved following cognitive-behavioral treatment for insomnia.²⁹ Nonetheless, further research is needed to determine whether interventions focused on addressing insomnia and improving sleep also may result in improved endothelial function.

FUNDING

This work was supported by funding from the National Heart, Blood, and Lung Institute (HL49427, HL50774, and HL121708).

DISCLOSURE

The authors declared no conflict of interest.