Effects of ambulatory blood pressure monitoring on sleep in healthy, normotensive men and women

Abstract

Objectives:

To determine the effect of ambulatory blood pressure monitoring (ABPM) on sleep quality among healthy adults and to explore possible effect modification by demographic factors.

Methods:

We examined data from 192 relatively healthy young (median age:31, 33% men, 18% with clinic BP >130/80 mmHg) participants in an observational study of sleep and arterial stiffness. Demographic and health questionnaires were completed at baseline. A wrist-based accelerometer was used to assess sleep for 7 consecutive nights, from which sleep duration, wakefulness after sleep onset (WASO), sleep fragmentation (physical restlessness), sleep midpoint, and sleep efficiency were estimated. ABPM was conducted for one 36-hour period, including 1 actigraphy night.

Results:

Within-subject comparisons indicated WASO and fragmentation were higher, midpoint was later, and efficiency was lower on the ABPM night (Ps<0.001–0.038). Neither age nor sex moderated these associations. Among older adults, a later midpoint and greater fragmentation were observed with ABPM (Ps=0.002–0.010). There was also a main effect of sex: men demonstrated shorter sleep duration, greater WASO and fragmentation, and less efficiency than women (Ps=0.002–0.046). With ABPM, women had worse fragmentation and a later midpoint (Ps=0.002–0.049); for men, WASO and fragmentation were worse (Ps=0.003–0.023). Importantly, this study does not address whether the effect of wearing ABPM on sleep in turn affects BP during sleep.

Conclusions:

The present study demonstrates that ABPM modestly disturbs actigraphy-assessed sleep among healthy adults. Researchers and clinicians should consider the downstream effects of performing ABPM and whether these effects are more pronounced in those who typically experience sleep disturbance.

Introduction

Twenty-four-hour ambulatory blood pressure monitoring (ABPM) is a Grade A recommendation for guideline-directed clinical assessment of hypertension [1]. Relative to conventional sphygmomanometry or blood pressure (BP) measured at rest in a clinical setting, ABPM is preferable for stratifying patients by cardiovascular risk, as average ambulatory blood pressure (ABP) better correlates with target organ damage and related vascular markers [2]. As Medicare and Medicaid coverage of ABPM was recently expanded [3], this measurement technique will likely be implemented by increasing numbers of health care providers to evaluate and manage hypertension.

ABPM involves frequent, periodic cuff inflation and associated noise from the integral pump, which may, in turn, disturb sleep [4]. For example, in a study of relatively healthy individuals, 39% of participants reported that their sleep was “somewhat worse” or “much worse” than usual on a night with ABPM [5]. Given that disturbed sleep is associated with elevated BP [4,6], the potential impact of ABPM on sleep may affect the validity of ABPM measurements during sleep. While previous studies have reported associations between ABPM and sleep disturbance [7–17], each has significant methodological limitations that hinder interpretation of the findings. These include assessing sleep via self-report rather than objectively [15,16], small samples [8–12], examining men only [9,11], including patients with confounding conditions such as hypertension, sleep-disordered breathing, or depression [10,12–15,17], assessing sleep in a laboratory rather than in the person’s home [8–14], or only examining one night of sleep monitoring together with ABPM [7,13,14]. ABPM may also affect sleep more significantly among those who are vulnerable to sleep disturbance, such as individuals who are prone to insomnia or who identify as being light sleepers [18,19]. Thus, it remains to be demonstrated whether ABPM affects sleep among healthy men and women and if there are individual differences in these effects. Such information is central to widely implementing ABPM, and for refining clinical recommendations for the use of ABPM.

We analyzed baseline data from a sample of healthy men and women. Participants wore wrist-based accelerometers to assess sleep on 7 consecutive nights and ABPM was conducted during a single 36-hour period that included 1 of these nights. Our goals were to ascertain, 1) the effects of ABPM on actigraphy-assessed sleep, and 2) whether any such associations were moderated by age or sex. In exploratory analyses we also examined if potential ABPM-sleep associations differed among those with and without elevated clinic BP (≥120/80 vs. <120/80 mmHg). We hypothesized that multiple indices of actigraphy-assessed sleep would be worse on the night with vs. the nights without ABPM. We further hypothesized that these associations would be greater among older individuals and women, as these groups are more vulnerable to sleep disturbances [18,19]. Finally, it was predicted that ABPM-sleep associations would be greater for participants with vs. those without elevated clinic BP [20].

Methods

Participants

Two hundred and forty-one adults were recruited from the community, using flyers posted on the campus of Yale University, and through emails to individuals who were registered as research volunteers with the Yale Center for Clinical Investigation. Participants were ≥21 years of age, fluent in English, and had a screening BP of <160/105 mmHg. Individuals were excluded from participation if they were taking medications known to affect BP (e.g., antihypertensive medications, steroids, sedative-hypnotics, and chronic anti-inflammatory), engaged in shift work in the evening or during the night, or if they self-reported clinical atherosclerotic disease (e.g., acute coronary syndrome, stroke), angina, impaired renal function, a diagnosis of sleep apnea, insomnia, bipolar disorder, chronic alcohol, or substance abuse, or any other major medical condition (e.g., cancer, chronic obstructive pulmonary disease, kidney disease, lung disease). Current smoking, hyperlipidemia, and diabetes were not exclusions. Individuals with recent (i.e., last 4 weeks) or planned trans-meridian travel completed study procedures during a time not affected by travel or residual desynchrony (i.e., ≥4 days after returning). The Yale University Human Investigation Committee approved this protocol and all participants provided written informed consent.

Study Design and Procedures

Eligible individuals were screened by phone and then invited to an in-person baseline visit to complete informed consent, determine final eligibility, and receive instructions for study procedures. During the baseline visit, BP, height, and weight were measured according to standard protocols and the latter two measures were used to calculate body mass index (BMI). Participants also received a wrist actigraph (Actiwatch-2, Philips Respironics), a device to assess sleep-disordered breathing (ApneaLink™ Plus, ResMed), and an ABP monitor (Oscar-2, SunTech Medical). Each device has been previously validated [21–23]. Participants were trained in and demonstrated the ability to perform self-instrumentation. Upon completion of their baseline visit, participants began a 7-day ambulatory monitoring period. Sleep was assessed each day and night via actigraphy and assessment of sleep-disordered breathing was completed the first night. On their choice of day (2–6), participants began 36-hours of ABPM. After the 7-day monitoring period, participants returned all devices and received compensation for their time and effort.

Measures

Sleep

At the baseline visit the Actiwatch-2 was placed on the participant’s non-dominant wrist. The device was programmed to record data in 1-minute epochs with a wake threshold of 40 activity counts. The number of epochs for sleep onset/offset was 10. Philips Actiware software v6.0 was used to derive the following parameters for each night: sleep duration (sleep onset to offset - periods of wakefulness), wakefulness after sleep onset (WASO; in minutes), fragmentation index (physical restlessness during sleep [percentage of mobile epochs + percentage of immobile epochs that are 1-minute in duration]), sleep midpoint (wake time-[0.5*sleep duration]), and sleep efficiency (100*[sleep duration/time in bed]). Pre-midnight sleep midpoints were coded as negative values (e.g., 11pm=−60 minutes). All participants completed a sleep diary to inform the determination of sleep onset and wake times assessed with actigraphy.

ABPM

At the baseline visit, participants provided their usual sleep and wake times. The Oscar-2 ABPM was programmed to measure BP every 30 minutes during awake hours and every 60 minutes during anticipated sleep hours, based on this self-reported sleep pattern. Participants were instructed to wear the Oscar-2 on their non-dominant arm. Systolic and diastolic BP readings were averaged separately for day and nighttime, according to actigraphy-assessed sleep times and sleep diaries. A single 24-hour BP variable was computed as a weighted average of the awake and sleep averages (weights=1-[time in bed/24 hours] and time in bed/24 hours, respectively). Average awake and sleep BP were computed for participants who had ≥10 valid awake BP readings and ≥5 valid sleep BP readings. BP dipping was characterized as the percentage difference between the awake and sleep periods in average systolic and diastolic BP ([awakeBP-sleepBP]/awakeBP).

Descriptive Characteristics and Covariates

A battery of self-report questionnaires was used to assess demographics and factors known to affect BP and sleep. Demographics included age, sex, race and ethnicity, partner or marital status, and smoking status. The Pittsburgh Sleep Quality Index [24] (PSQI) and Insomnia Severity Index [25] (ISI) were used to characterize self-reported sleep. The PSQI includes self-reported measures of sleep duration, sleep latency, and sleep efficiency, while the ISI is designed to assess the severity of daytime and nighttime symptoms of insomnia. Depressive symptoms were assessed with the Patient Health Questionnaire-8 (PHQ-8), which corresponds with the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V) diagnostic criteria for major depressive disorder [26].

Statistical Analysis

Analyses were performed using SAS software (V.9.4, SAS Institute, Cary, NC). All continuous variables are expressed as median (lower, upper quartiles; Q1,Q3) and categorical variables are expressed as frequencies (percentages). Forty-three participants (17.8%) were excluded from the analyses because of missing sleep, ABPM, or self-report data, and 6 participants (2.5%) were excluded for moderate sleep apnea as defined by an apnea-hypopnea index ≥15 events/hr, leaving a final sample of 192. Those excluded did not differ from those included in age, the distribution of sex, race, or ethnicity.

Univariate statistics were used to characterize the sample and to assess sleep variables for normality. The statistical significance of group differences was based on the Mann-Whitney test for continuous variables, Fisher’s Exact Test for binary variables, Pearson’s chi-square test for nominal variables with more than two categories and, the asymptotic test of the gamma statistic for ordinal variables. For each skewed sleep measure (WASO, fragmentation index, and sleep efficiency), a Box-Cox transformation was used to transform the variable into an approximately normal distribution [27].

The ApneaLink, worn on the first night, was expected to interfere with sleep. Thus, in initial tests we compared sleep on the first night to sleep on nights without ABPM. Our expectation was confirmed and, therefore, actigraphy data from the first night were excluded. SAS PROC MIXED software was used to estimate a linear repeated measures ANOVA model with a random person effect to compare the within-subject effect of the night with ABPM to the other five nights without ABPM. Covariates included age, sex, race (White vs. non-White), marital status (married/partnered vs. other), education (college graduate vs. other), smoking (current smoker vs. other), and BMI. An additional covariate was created based on whether ABPM occurred during a week or weekend night. We examined whether age (treated as a continuous variable) or sex moderates the effect of ABPM. In exploratory analyses, we examined sleep on the night with ABPM vs. nights without ABPM exclusively among younger and older age groups based on a median split, and separately for women and men. Although this was a relatively healthy population, the primary analyses were also repeated for participants with and without elevated clinic BP (<120/80 vs. ≥120/80 mmHg) per the 2017 AHA/ACC Guidelines [28]. Covariate-adjusted means of sleep variables for ABPM and non-ABPM nights are reported. Post hoc power estimates for the mixed ANOVA are also reported. Statistical significance required a P<0.05 level (two-tailed).

Results

The sample was 33% male, 65% White, and 13% Hispanic, with a median age of 31 years (Q1,Q3: 28,38); 34% were married or partnered, 76% were college graduates, and 75% were employed, with 30% reporting an income of <$30,000 (i.e., low income). Median BMI was 24 kg/m² (22,28; normal BMI: ≤25 kg/m²) and 9% of participants were smokers. Approximately 28% (n = 53) of participants exhibited a clinic BP in the elevated range (≥120/80 mmHg) and 18% (n = 34) of participants had a clinic BP indicative of Stage 1 hypertension (≥130/80 mmHg [28]). Median clinic systolic/diastolic BP (SBP/DBP) were 121/73 mmHg. Median awake and sleep SBP/DBP were 122/72 mmHg and 104/57 mmHg. Median sleep SBP/DBP dip were 13% and 14% (normal range:10–20% [29]). Four percent of participants (n = 7) reported symptoms meeting criteria for insomnia on the ISI (≥15) and 6% met the threshold for a depression diagnosis on the PHQ-8 (≥10; n = 12). Younger adults showed a significantly greater SBP dip, but older adults showed a greater diastolic dip (Ps = 0.026 and 0.047). Relative to women, men had significantly higher median clinic SBP/DBP (118/71 vs. 126/77, Ps < 0.001 and 0.017), awake SBP/DBP (118/71 vs. 128/76, Ps < 0.001), sleep SBP (103 vs. 111, P < 0.001), and greater sleep SBP/DBP dipping (12% and 12% vs. 16% and 18%, Ps < 0.002). Self-reported sleep variables assessed with the PSQI and the ISI did not differ by age or sex, and there were no statistically significant differences in depression on the PHQ-8 (Table 1).

Table 1.

Characteristics of the overall sample, by a median split of age, and sex.

	Overall	Age			Sex
	(N = 192)	≤ 30 (n = 92)	> 30 (n = 100)	P	Women (n = 128)	Men (n = 64)	P
Sociodemographics
Age, y, median (Q1,Q3)	31 (28,38)	27 (25,29)	37 (33,49)	---	31 (28,39)	30 (27,37)	0.615
Sex, men, n (%)	64 (33.3)	33 (35.7)	31 (31.0)	0.651	0 (0)	64 (100)	---
Race, n (%)				0.744			0.951
White	124 (64.6)	58 (63.0)	66 (66.0)		82 (64.1)	42 (65.6)
Black	34 (17.7)	16 (17.4)	18 (18.0)		22 (17.2)	12 (18.8)
Other, Not reported	34 (17.7)	18 (19.6)	16 (16.0)		24 (18.8)	10 (15.6)
Hispanic, n (%)	24 (12.5)	12 (13.0)	12 (12.0)	0.999	18 (14.1)	6 (9.4)	0.486
Married/partnered, n (%)	66 (34.4)	32 (34.8)	34 (34.0)	0.999	46 (35.9)	20 (31.3)	0.523
College graduate, n (%)	146 (76.0)	66 (71.7)	80 (80.0)	0.174	100 (78.1)	46 (71.9)	0.273
Employed, n (%)	144 (75.0)	67 (72.8)	77 (77.0)	0.414	98 (76.6)	46 (71.9)	0.598
Income, n (%)				0.065			0.127
<$30,000	58 (30.2)	40 (43.5)	18 (18.0)		28 (21.9)	30 (46.9)
$30,000–$69,999	59 (30.7)	25 (27.2)	24 (24.0)		39 (30.5)	20 (31.3)
≥ $70,000	49 (25.5)	27 (29.3)	22 (22.0)		45 (35.2)	14 (21.9)
BMI (kg/m2)	24 (22,28)	25 (23,29)	24 (22,27)	0.070	24 (22,28)	24 (23,28)	0.465
Current smokers, n (%)	17 (8.9)	11 (12.0)	6 (6.0)	0.204	9 (7.0)	8 (12.5)	0.172
Blood Pressure
Clinic SBP, mmHg	121 (112,131)	123 (113,135)	120 (111,128)	0.168	118 (109,128)	126 (117,135)	<0.001
Clinic DBP, mmHg	73 (65,82)	74 (65,84)	72 (67,80)	0.934	71 (64,80)	77 (69,84)	0.017
ABPM SBP, Awake, mmHg	122 (113,128)	123 (115,132)	121 (111,126)	0.057	118 (111,125)	128 (122,139)	<0.001
ABPM DBP, Awake, mmHg	72 (67,78)	73 (68,79)	71 (66,76)	0.065	71 (66,76)	76 (69,81)	<0.001
ABPM SBP, Sleep, mmHg	104 (97,112)	105 (98,112)	104 (97,112)	0.872	103 (96,109)	111 (101,116)	<0.001
ABPM DBP, Sleep, mmHg	57 (53,63)	57 (52,63)	58 (53,63)	0.531	57 (53,63)	58 (53,63)	0.792
ABPM SBP, Dip, %	13 (7,17)	14 (11,18)	12 (8,16)	0.026	12 (9,16)	16 (12,20)	<0.001
ABPM DBP, Dip, %,	14 (9,18)	15 (11,19)	18 (11,22)	0.047	12 (8,16)	18 (11,22)	0.002
Self-reported Sleep (PSQI)
Sleep duration, min	480 (450,540)	480 (450,540)	480 (445,525)	0.341	480 (450,540)	480 (436,539)	0.234
Sleep latency, min	15 (10,30)	15 (10,30)	15 (10,30)	0.620	15 (10,30)	15 (10,30)	0.919
Sleep efficiency, %	90 (83,98)	92 (84,97)	90 (83,99)	0.863	89 (82,99)	92 (86,96)	0.588
Comorbidities
Insomnia, n (%)	7 (3.6)	2 (2.1)	5 (5.0)	0.447	6 (4.7)	1 (1.6)	0.427
Depression, n (%)	12 (6.3)	6 (6.5)	6 (6.0)	0.999	9 (7.0)	3 (4.7)	0.754

^*Data are presented as the median (Q1,Q3) or frequency (percentage).

Acronyms and abbreviations: ABPM, Ambulatory blood pressure monitor; BMI, Body Mass Index; DBP, Diastolic blood pressure; ISI, Insomnia Severity Index; PHQ-8, Patient Health Questionnaire-8; PSQI, Pittsburgh Sleep Quality Index; SBP, Systolic blood pressure

On the night with vs. the nights without ABPM, participants exhibited on average 4.2 minutes greater WASO (48.4 vs. 44.2, 95% CI: 0.22, 8.15, P = 0.038), 4.3% greater fragmentation (33.2% vs. 28.9%, 95% CI: 2.36, 6.31, P < 0.001), a 15-minute later sleep midpoint (261.0 vs. 245.6 minutes past midnight, 95% CI: 3.26, 27.5,P = 0.013), and poorer sleep efficiency (86.6% vs. 87.8%, 95% CI: −2.28, −0.23, P = 0.016), all covariate-adjusted. The difference in sleep duration, 2.4 minutes, was not statistically significant (ABPM night: 437.8 vs. non-ABPM nights: 435.4, 95% CI: −13.01, 17.80, P = 0.761; see Table 2).

Table 2.

Actigraphy-assessed sleep for ABPM night and non-ABPM nights. Presented as covariate-adjusted means, standard errors, differences of least squares means (beta), and 95% confidence intervals for the total sample and stratified by age and sex.

	ABPM M (SE)	Non-ABPM M (SE)	β (95% C.I.)	P
Overall*
Duration, min	437.80 (8.16)	435.41 (11.00)	2.39 (−13.01, 17.80)	0.761
WASO, min	48.38 (2.48)	44.19 (3.12)	4.19 (0.22, 8.15)	0.038
Sleep Fragmentation, %	33.19 (1.33)	28.86 (1.64)	4.33 (2.36, 6.31)	<0.001
Midpoint, min†	261.00 (10.04)	245.63 (11.54)	15.36 (3.26, 27.46)	0.013
Sleep Efficiency, %	86.56 (0.70)	87.82 (0.86)	−1.26 (−0.23, −2.28)	0.016
Age
≤ 30 years (n = 92)
Duration, min	440.39 (8.82)	440.07 (13.24)	0.32 (−21.17, 21.81)	0.977
WASO, min	48.05 (2.68)	44.70 (3.70)	3.35 (−2.17, 8.88)	0.234
Sleep Fragmentation, %	32.53 (1.44)	28.37 (1.92)	4.16 (1.41, 6.91)	0.003
Midpoint, min	258.85 (10.89)	250.31 (13.31)	8.54 (−8.34, 31.29)	0.321
Sleep Efficiency, %	86.71 (0.75)	87.94 (1.00)	−1.22 (−2.65, 0.21)	0.094
> 30 years (n = 100)
Duration, min	434.38 (9.31)	429.88 (8.82)	4.50 (−17.50, 26.50)	0.688
WASO, min	48.90 (2.85)	43.93 (3.92)	4.96 (−0.70, 10.63)	0.086
Sleep Fragmentation, %	34.11 (1.44)	29.62 (2.04)	4.49 (1.67, 7.31)	0.002
Midpoint, min	262.96 (11.62)	240.33 (14.18)	22.63 (5.34, 39.92)	0.010
Sleep Efficiency, %	86.34 (0.80)	87.62 (1.07)	−1.28 (−2.75, 0.19)	0.087
Sex
Women (n = 128)
Duration, min	443.84 (8.73)	449.20 (12.58)	−5.36 (−24.60, 13.87)	0.584
WASO, min	44.11 (2.66)	41.89 (3.53)	2.22 (−2.73, 7.17)	0.379
Sleep Fragmentation, %	30.54 (1.43)	26.57 (1.85)	3.97 (1.50, 6.43)	0.002
Midpoint, min	253.85 (10.82)	238.74 (12.91)	15.11 (0.01, 30.23)	0.049
Sleep Efficiency, %	87.91 (0.75)	89.07 (0.97)	−1.16 (−2.45, 0.12)	0.075
Men (n = 64)
Duration, min	432.35 (9.49)	416.23 (15.13)	16.12 (−9.44, 41.68)	0.216
WASO, min	52.79 (2.89)	45.14 (4.19)	7.65 (1.08, 14.23)	0.023
Sleep Fragmentation, %	35.87 (1.56)	30.88 (2.18)	4.99 (1.71, 8.26)	0.003
Midpoint, min	268.17 (11.78)	252.35 (14.92)	15.81 (−4.30, 35.93)	0.123
Sleep Efficiency, %	85.21 (0.81)	86.63 (1.14)	−1.42 (−3.12, 0.28)	0.102

^*P-values are based on covariate-adjusted linear mixed models of actigraphy-assessed sleep measures. Severely skewed measures were transformed to a normal distribution. Covariates included age, sex, race, marital status, education, smoking status, body mass index, and a variable representing if ABPM occurred during a week or weekend night.

^†Minutes past midnight

Acronyms and abbreviations: ABPM, ambulatory blood pressure monitoring; WASO, wake after sleep onset

In covariate-adjusted models, age was not associated with any of the actigraphy-assessed sleep measures (Ps = 0.32–0.76) and there were no significant Age*ABPM interactions (Ps = 0.08–0.85). Exploratory analyses based on a median split of age (≤30 years and >30 years) indicated that individuals in the older group had significantly greater sleep fragmentation (34.1% vs. 29.6%, β = −4.49%, 95% CI: 1.67, 7.31; P = 0.002) and a later sleep midpoint (263.0 vs. 240.3 min., β = 22.70, 95% CI: 5.34, 39.92; P = 0.010) on the ABPM night compared to their nights without ABPM. The younger group also showed significantly greater sleep fragmentation on the ABPM night (32.5% vs. 28.4%; β = 4.16%, 95% CI: 1.41, 6.91; P = 0.003). Figure 1 displays covariate-adjusted mean differences between actigraphy-assessed sleep on ABPM and non-ABPM nights for each age group and overall.

Figure 1.

Covariate-adjusted mean differences and standard errors for actigraphy-assessed sleep on a night with ABPM vs. non-ABPM nights. Sleep variables are displayed for age groups based on a median split (≤ 30 years and > 30 years) and overall. Data presented for sleep duration and sleep midpoint can be multiplied by five to obtain actual values.

There were significant main effects of sex. Compared to women, men had significantly shorter sleep duration (446.5 vs. 424.3; β = −22.23 min., 95% CI: −42.33, −2.13; P = 0.030), greater WASO (43.0 vs. 49.0 min.; β = 5.96, 95% CI: 0.12, 11.81; P = 0.046) and fragmentation (28.7% vs. 33.4%; β = 4.82%, 95% CI: 1.71, 7.93; P = 0.002), and worse sleep efficiency (88.49% vs. 85.92%; β = −4.20%, 95% CI: −2.58, −0.95; P = 0.002). However, there were no statistically significant Sex*ABPM interactions predicting the sleep variables (Ps = 0.19–0.96). When comparing the ABPM night vs. the non-ABPM nights, women showed greater fragmentation (30.5% vs. 26.6%; β = 3.975, 95% CI: 1.50, 6.43; P = 0.002) and a later sleep midpoint (253.9 vs. 238.7; β = 15.11, 95% CI: 0.01, 30.23; P = 0.049). Men exhibited greater WASO (52.8 vs. 45.1 min.; β = 7.65 min., 95% CI: 1.08, 14.23; P = 0.023) and fragmentation (35.9% vs. 30.9% min.; β = 4.99%, 95% CI: 1.71, 8.26; P = 0.003) on the night with ABPM relative to nights without (see Figure 2).

Figure 2.

Covariate-adjusted mean differences and standard errors for actigraphy-assessed sleep on a night with ABPM vs. non-ABPM nights, displayed by sex and overall. Data presented for sleep duration and sleep midpoint can be multiplied by five to obtain actual values.

Finally, actigraphy-assessed sleep was compared on ABPM and non-ABPM nights among participants with elevated clinic BP (≥120/80 mmHg; n = 53) and those with normotensive clinic BP (<120/80 mmHg). Overall, there was no effect modification by clinic BP category (Ps = 0.30–0.53). On the ABPM night compared with non-ABPM nights, those with elevated clinic BP showed significantly greater sleep fragmentation (36.3% vs. 31.3%, β = 4.35%, 95% CI: 1.89, 6.80, P < 0.001), and a marginally significant later midpoint (257.4 vs. 231.4 min., β = 26.0 min., 95% CI: −1.89, 53.81, P = 0.067). Among participants with a normal clinic BP, sleep fragmentation was significantly greater on the ABPM night vs. non-ABPM nights (31.9% vs. 28.6%, β = 3.33%, 95% CI: 0.81, 5.85, P = 0.010).

In post-hoc power analyses, a sample size of 192 provided >90% power to detect differences between the ABPM night and non-ABPM nights consisting of 26 minutes in sleep duration, 6 minutes in WASO, 4% in fragmentation, 20 minutes in sleep midpoint, and 2% in sleep efficiency.

Discussion

We compared actigraphy-assessed sleep quality on 1 night with ABPM to 5 nights without ABPM in 192 healthy, young and middle-aged adults. Study participants demonstrated small but statistically significant decrements in sleep on the ABPM night, including increased WASO and sleep fragmentation, a later sleep midpoint, and poorer sleep efficiency, supporting the study hypothesis. Sleep duration did not significantly differ with ABPM. Overall, associations between ABPM and actigraphy-assessed indices of sleep may be subtle. Yet, our findings are comparable in magnitude to the effects observed in previous intra-individual investigations of behavioral and environmental influences (e.g., alcohol, caffeine) on actigraphy-assessed sleep [30–32]. Although there were no significant moderating effects of age, some of the normative changes in sleep that occur with age (i.e., earlier sleep midpoint) do not occur on ABPM nights, when older participants had about the same sleep midpoint as younger participants [18]. Finally, sex and elevated clinic BP did not modify ABPM-sleep associations.

Although women often show greater sleep disturbances than men [19], our hypothesis that women would experience more profound effects of ABPM on actigraphy-assessed sleep was not supported. Even after controlling for differences in BMI, we observed greater ABPM-related sleep disturbance among men, including increased WASO and fragmentation. Women also demonstrated increased fragmentation and a later sleep midpoint with ABPM, indicating that each group’s sleep was negatively affected by ABPM, albeit in slightly different ways. Our study did not utilize the gold standard for measuring sleep, polysomnography, and there may be simple effects of ABPM on sleep architecture for both sexes - e.g., as measured by electroencephalogram (EEG) spectral analysis [33]. Based on these results, we would expect to see main effects of ABPM using other sleep measures as well; while we cannot rule out the possibility that the effects of performing ABPM on sleep architecture and other sleep parameters differs by sex, the present study suggests that these differences may be small or non-existent.

When comparing ABPM-sleep associations among participants with elevated vs. normal clinic BP, although there was a slight tendency for those with elevated BP to exhibit a greater adverse effect of wearing the ABPM on sleep quality, the difference between groups did not approach statistical significance. Within each group, only sleep fragmentation was significantly worse on the ABPM night than on non-ABPM nights. Only two known studies have examined the sleep-related effects of performing ABPM among young and middle-aged adults; both investigated patients with hypertension, but had smaller sample sizes and low statistical power to detect such effects (i.e., N=70[34] and 121[35]), and neither found associations with sleep. ABPM effects on sleep may be greater among individuals with hypertension, potentially due to more intrusive cuff inflation or to the effects of medication use (e.g., antihypertensives) [20]. Thoroughly exploring those associations will require a much larger sample size and a higher proportion of individuals with Stage 1 and 2 hypertension than examined previously.

ABPM-sleep associations are of interest for both methodological and clinical considerations. Recurrent arousals from sleep due to ABPM may iatrogenically affect cardiovascular biomarkers or behavioral factors assessed in concert with ABP. For example, observational epidemiological studies and those in which sleep fragmentation was experimentally manipulated demonstrate that one or two nights of amplified sleep fragmentation is associated with a greater heart rate, increased endocrine markers including cortisol, cholesterol, and catecholamines, decreased glucose metabolism and O₂/CO₂ metabolism, and elevated cellular and genomic markers of inflammation among healthy adults [36−39]. Many of those patterns resemble the physiological signatures of sleep-related breathing disorders such as sleep apnea [40]. Behaviorally, fragmentation reduces the restorative value of sleep and is directly associated with daytime sleepiness [41]. Thus, interpreting ABPM data requires a consideration of potential effects of the device on both sleep and related measures that may be influenced by sleep deprivation.

Individuals with disturbed sleep (sleep duration, continuity, and/or architecture) may have less BP dipping (i.e., increased sleep vs. awake BP) and a greater likelihood of hypertension [4]. A recent review suggests that ABP cuff inflation may induce a reactive increase in BP for some adults [4], an association that may be modulated by sleep disturbance [16] or by sleep stage [42]. If the conduct of ABPM disrupts sleep sufficiently to alter BP during sleep, it would be difficult to evaluate the effects of sleep on sleep BP unless using brachial artery occlusion or plethysmography. Although more recent BP monitors and other emerging tools do not rely on cuff inflation, likely a key factor in the observed sleep disturbances [4], those technologies are often expensive, too sensitive to an individual’s movements, or require further validation before they are adopted [6]. Continuing to evaluate the dynamic effects of ABPM on objective sleep is necessary to understand how measuring sleep BP with ABPM may affect the estimated BP one is trying to measure. Alternately, if ABPM cuff inflation has relatively little effect on sleep, then effects on BP level during sleep are probably small to negligible.

There are limitations to these findings. For one, the sample was relatively healthy, limiting generalizability. We did not assess momentary effects of ABPM-associated sleep disturbance on BP level and ABPM-sleep associations may be underestimated due to methodological or individual differences, such as the young average age and/or relatively low average BP level of this group. In another example, older ABPM inflation and cuff designs that detect oscillations during deflation, including the Oscar 2 used in this study, may affect sleep more than monitors that assess BP during inflation. Similarly, individuals who have a larger arm circumference, and hence require a larger BP cuff size, may experience longer cuff inflation times, creating a wider window for ABPM to affect sleep. Sleep was assessed using actigraphy rather than by polysomnography, and questions concerning sleep architecture and quantitative EEG could not be pursued. Polysomnography was used in investigations with older ABP technologies [9–11,13], yet we do not know how today’s ABPM technology influences physiological indices of sleep, and individual differences were not previously examined. Our sleep diary did not include questions about satisfaction with sleep or daytime functioning, so it is unclear if ABPM affected self-reported sleep or fatigue the next day [34]. We also did not collect information about alcohol intake or diabetes and could not control for those factors. Thus, the observed ABPM-sleep associations warrant replication in the broader population. Finally, there may be subgroups for whom sleep is more affected by ABPM (e.g., minorities or those who already suffer from sleep disturbance or posttraumatic stress disorder [35,43]).

Conclusions

ABPM is a valuable tool for cardiovascular research and to assess for hypertension. While many patients and research participants self-report that ABPM negatively affects their sleep, there is scant data comparing actigraphy-assessed sleep quality from nights with vs. without ABPM. In the present study of healthy adults, ABPM was associated with modest within-subject nightly decrements in several indices of actigraphy-assessed sleep, including increased arousals and alterations in sleep timing. The sleep-related consequences of ABPM are subtle but relevant to both clinical and research stakeholders. Although performing ABPM influences sleep, it likely has only modest, transient, downstream effects for most individuals, and may or may not affect the level of BP measurements. Those who are interested in sleep BP or in the simultaneous assessment of sleep should evaluate such downstream effects in healthy individuals and determine whether ABPM-sleep associations are more robust in individuals with disturbed sleep, with the hope of better interpreting ABPM data, understanding BP pathophysiology, and advancing practices for cardiovascular prevention.