Longitudinal Sleep Characteristics and Hypertension Status: Results from the Wisconsin Sleep Cohort Study

Abstract

Aims:

Sleep characteristics such as short sleep duration or sleep-disordered breathing are established predictors of hypertension. However, few studies have used in-lab polysomnography with a longitudinal design to measure how hypertension is associated with different sleep stages over time. The purpose of this study is to examine whether hypertension is associated with the longitudinal course of sleep quality over time.

Methods:

This study evaluated data from the Wisconsin Sleep Cohort Study, which consists of 1,525 adults in a community-based population of middle-aged to older adults followed for approximately 12–25 years. Sleep characteristics were objectively measured using polysomnography and subjectively assessed using a self-report questionnaire on insomnia complaints. We used linear mixed-effects regression models and cumulative logit models to assess whether the interaction of hypertension and time is associated with objective and subjective sleep.

Results:

We found people with hypertension exhibited a greater decline in total sleep time in REM sleep (%) over time than those without hypertension (p <0.05). Individuals with hypertension had less decline in % N3 sleep over time than those without hypertension (p <0.05). Among the subjective insomnia complaints, our findings indicate hypertensive individuals have a higher probability of having higher levels of “difficulties in falling asleep” compared to people without hypertension.

Conclusion:

These findings suggest that hypertension is associated with modified longitudinal changes of objective and subjective sleep characteristics.

Introduction

Approximately 50–70 million individuals suffer from insufficient sleep in the United States [1]. Sleep is composed of eye movement at four levels: N1, N2, N3, characterized as non-rapid eye movement (NREM), and rapid eye movement (REM) sleep. N1 typically accounts for 2–5% of sleep time, N2 accounts for 45–55%, N3 accounts for 13–23%, and REM accounts for 20–25% [2]. Numerous factors affect the quality and quantity of sleep including aging [3], sleep disorders [4], chronic conditions [5,6], and medications [7]. For instance, as individuals age, slow wave sleep and difficulties initiating and maintaining sleep increase (reflected as increases in N1 and N2 NREM sleep stages), and slow wave sleep (stage N3 sleep) and REM sleep decreases [3]. Sleep-disordered breathing, which is more prevalent in older aged individuals, also results in frequent arousals due to apneic episodes that interfere with sleep architecture [8]. Chronic conditions that are more likely in older individuals such as cardiovascular disease [9–11], pain [12], and psychiatric diseases [13] can also interfere with sleep. A combination of these factors may be associated with a decreased ability to sleep over time with advancing age.

Hypertension is a highly prevalent condition that affects one-third of the adult population in the U.S. and more than one-quarter of the adult population worldwide [14]. The consequences of hypertension include heart failure, coronary artery disease, stroke, renal disease, and peripheral arterial disease [14]. Prior longitudinal studies have indicated that specific sleep characteristics are predictive factors for developing hypertension, [15–19,19–26] including sleeping less than five hours, or more than nine hours per night, [20,23,25], longer sleep latency, and wake after sleep onset [22], shorter slow wave sleep [24], and a higher apnea hypopnea index score from sleep-discorded breathing [19,25–28]. For instance, prior clinical studies have demonstrated that fluid redistribution and increased sympathetic activity from sleep disordered breathing can increase blood pressure [27,28]. In addition, in a recent Wisconsin Sleep Cohort study, investigators reported that three conditions were associated with a relative risk of developing 24-hour ambulatory non-dipping blood pressure: (1) having 7–8 hours total sleep compared to 6–7 hours of sleep, (2) longer time awake after sleep onset, and (3) lower sleep efficiency [21]. While these studies provide insights on how sleep could contribute to hypertension, it is unclear how sleep changes longitudinally by hypertension.

Several findings from cross-sectional studies support that individuals with hypertension have poorer self-reported sleep quality [15]; higher sleep fragmentation, lower sleep efficiency, more insomnia, more frequent napping, higher rate of sleep disordered breathing [29]; and shorter sleep time, longer total time spent in N1 and N2, shorter total time spent in N3 and REM sleep compared to people without hypertension [30]. However, these studies did not comprehensively measure sleep characteristics using both subjective and polysomnography (PSG) and did not explore longitudinal changes in sleep characteristics by hypertension. Thus, the objective of this study was to investigate how hypertension is associated with PSG-assessed sleep characteristics and subjective insomnia complaints over time in a community-based population of middle-aged to older adults who were followed for up to 25 years. Our specific aims were 1) to examine how hypertension modifies longitudinal PSG-assessed sleep characteristics and 2) to examine whether hypertension is associated with changes in subjective insomnia complaints over time. We hypothesized that hypertension is associated with greater increases in sleep latency, time spent in N1, wake time after sleep onset (WASO), greater declines in total sleep time (TST), total time spent in N3, and REM sleep, and greater increases in subjective insomnia complaints over time compared to individuals without hypertension.

Methods

Design, sample, and procedure

The University of Wisconsin Health Sciences Institutional Review Board approved the Wisconsin Sleep Cohort Study, and all participants provided informed consent. Data from 1,557 individuals were used for this study. The Wisconsin Sleep Cohort Study is a longitudinal, community-based study of the causes, consequences, and natural history of sleep behaviors and common sleep disorders in adults [19]. The investigators mailed a survey regarding sleep habits, health, and demographics to 6,947 state employees in Wisconsin between 30 and 60 years old in the initial study in 1988, with an initial response rate of 72%. From the stratified sample of the initial survey respondents (n = 2,884), respondents were invited to an overnight PSG study, and 1,557 (53%) participants successfully completed the baseline sleep study. Overnight protocols, which included PSG, height, weight, blood pressure, body habitus measurements, and questionnaires on lifestyle, health, and medications were conducted in the Clinical Research Unit at the University of Wisconsin Hospitals and Clinics.

Visits

All participants had a baseline study between 1988 and 2000 and were subsequently invited at approximately 4-year intervals for follow-up visits with overnight sleep studies over a ~15–25 year period (up to eight visits). This study evaluated data from 1,557 adults from the Wisconsin Sleep Cohort Study.

Measurements

All participants underwent an overnight, in-laboratory, 18-channel PSG (Grass model 78; Quincy, MA). During the PSG, subjects’ sleep state was recorded using (1) electroencephalography, electrooculography, and electromyography; (2) breathing, using respiratory inductance plethysmography (Respitrace; Ambulatory Monitoring, Ardsley, NY), nasal and oral airflow (ProTec thermoocouples; Hendersonville, TN), and nasal pressure transducers; and (3) oxyhemoglobin saturation, using pulse oximetry (Ohmeda Biox 3740; Englewood, CO). The sleep stage, apnea hypopnea index (AHI), and hypopnea events were scored by trained technicians. Apnea was defined as the cessation of nasal and oral airflow for 10 seconds or more, and hypopnea was characterized as a discernable reduction in breathing (sum of chest and abdominal excursions) with a reduction in oxyhemoglobin saturation of 4% or greater. Key sleep variables were examined: hours of total sleep time (TST) in minutes, percentage of each stage of sleep (i.e., N1, N2, N3, and rapid eye movement sleep (REM)), sleep efficiency (SE), and wake after sleep onset (WASO).

Subjective insomnia complaints

Participants completed questionnaires to measure four insomnia symptoms with questions related to usual (1) difficulty getting to sleep, (2) waking up during the night and having a hard time getting back to sleep, (3) waking up repeatedly at night, and (4) waking up too early in the morning and not being able to get back to sleep. Response categories for each symptom were as follows: “0 = never,” “1 = rarely” (once/month), “2 = sometimes” (2–4 times/month), “3 = often” (5–15 times/month), and “4 = almost always” (16–30 times/month).

Self-report sleep time and naps

Participants completed a questionnaire on their total sleep time (hours per day), naps (hours of sleep per week from naps), and total sleep time including naps (hours per day).

Hypertension status

Participants were considered to have hypertension if either of the following conditions were met: (1) auscultatory-assessed blood pressure was equal to or above 140/90 mmHg, measured by research staff at the PSG study visit; or (2) then-current use of antihypertensive medications. Their blood pressure was taken after 5 minutes of sitting, and repeated 2–3 times at the beginning of an in-person visit (night PSG, overnight, in-home). Further details on blood pressure measurement is included in the Supplementary digital contents.

Demographic and clinical information

Data on medical history, hypertension medication use, age, sex, continuous positive airway pressure (CPAP) use, current smoker (yes/ no), ever having been told of the following diagnoses (e.g., heart failure, diabetes, atherosclerosis, and myocardial infarction), number of caffeinated beverages per day, and the number of alcoholic drinks per week were obtained through interviews and questionnaires during the overnight PSG laboratory study visit.

Statistical analysis

Descriptive statistics including means, standard deviations, and frequencies were calculated. We examined distributions of continuous variables for normality and homogeneity of variance. We used t-tests and chi-square tests to compare the demographic and clinical variables. Among study participants there was minimal data missingness so we did not impute missing data in the model. To examine whether participants’ hypertension modifies longitudinal, PSG-assessed sleep characteristics or self-report sleep time (i.e., total sleep time, naps, total sleep time including naps), we entered the interaction term (hypertension × time) in linear mixed-effects regression models (SAS PROC MIXED). For the continuous variables, we estimated the weighted averages of the cross-sectional and within-subject associations of hypertension and PSG sleep data. This method allowed us to maximize the study power and account for intra-subject correlation from the multiple assessments of the majority of the participants. To examine whether participants’ hypertension modifies longitudinally assessed subjective sleep insomnia complaints, we again entered the interaction term (hypertension × time) in cumulative logit models (SAS PROC GENMOD). We estimated the logit of being in a higher subjective insomnia complaints category relative to a lower subjective insomnia complaint category by having hypertension. PROC GENMOD allowed us to analyze non-normally distributed data structures and allowed for specification of both time-varying and individual difference variables in the longitudinal data analysis [31–33].

We controlled for potential confounding variables selected from previous literature including age, sex, body mass index (BMI), apnea-hypopnea index (AHI), CPAP use, diabetes, atherosclerosis, heart failure, and current smoker, as these factors have been associated with both sleep characteristics and hypertension in previous studies [5,15,19,34,35]. We also evaluated several other covariates that could impact both hypertension and sleep variables (e.g., use of hypertensive medications [e.g., beta-blockers, angiotensin converting enzyme blockers, angiotensin receptor blockers, diuretics], number of caffeinated beverages per day, and the number of alcoholic drinks per week). However, including these factors in the model did not substantially change the associations of interest, and these covariates were not significant. All the analyses were conducted using SAS 9.4 software (SAS Institute Inc., Cary, NC).

Results

Participant characteristics

Demographic and clinical characteristics are summarized in Table 1. A total of 479 individuals were identified as hypertensive, and 1,068 individuals were identified as non-hypertensive at baseline. There were 1,478 participants were white, 30 were Black or African American, 1 Native Hawaiian, 14 Asian, and 19 American Indian/ Alaska Native, and 22 were more than one race. On average, individuals were followed for 12 (SD= 8.1) years. Individuals who were identified as hypertensive were older, had higher BMIs, higher AHIs, and higher prevalence of diabetes, atherosclerosis, and heart failure at baseline. Table 2 and Supplementary digital content (Table S1) summarize the mean sleep measure values in the first four visits. The incidents of hypertension increased over time. As in Table 2, only 31% of the sample at visit 1 had hypertension, but the incidents increased to 35% at visit 2, 45% at visit 3, and 51% at visit 4. Hypertensive individuals had shorter sleep times, greater WASO, worse sleep efficiency, and longer sleep latency (p<0.05) than non-hypertensive participants. Hypertensive individuals had longer % time spent in N1 and shorter % time spent in N3 and REM sleep compared to non-hypertensive individuals at each visit (p<0.05) (Table 2). However, we did not observe any meaningful differences in mean subjective sleep complaints between the two groups at each visit (Table A).

Table 1.

Demographic and clinical information at the baseline visit ( n = 1,557)

	Hypertension ( n= 479)	Non-Hypertension (n=1068)
	Mean ± SD or % (n)
Age (years)	50.0 ± 8.0	46.4 ± 7.9
Body mass index, kg/m2	32.5 ± 6.8	28.7 ± 6.3
Sex (% Female)	36.4 (168)	48.4 (512)
CPAP use (yes)	1.6 (7)	1.1 (11)
Diabetes mellitus (yes)	7.2 (33)	1.5 (16)
Atherosclerosis (yes)	2.0 (9)	0.3 (3)
Heart failure (yes)	0.9 (4)	0.1(1)
Chronic Kidney Disease (No CKD eGFR >= 60, Stage 3 CKD (eGFR <60)	1 (5)	0.7 (7)
Current smoker (yes)	14.3 (66)	19.7 (208)
Systolic blood pressure (mmHg)	137.5 ± 14.5	119.4 ± 7.6
Diastolic blood pressure (mmHg)	89.0 ±10.1	77.9 ± 7.6
Use of antihypertensives (yes)	51.4 (237)	0 (0)
Beta blockers	18 (82)	0 (0)
Angiotensin Converting Enzyme Inhibitors	15( 71)	0 (0)
Angiotensin Receptor Blockers	1(0.2)	0 (0)
Diuretics	19(86)	0 (0)
Alpha blockers	4(17)	0 (0)
Number of caffeinated beverages per day	3.5 ± 3.1	3.4 ± 3.0
Number of alcoholic drinks per week	4.5 ± 5.8	3.7 ± 6.2

CPAP: continuous positive airway pressure

Table 2.

Objective sleep characteristics by hypertension status (n= 1,557)

	Visit 1 (Baseline)		Visit 2 (4 Year follow Up)		Visit 3 (8 Year follow Up)		Visit 4 (12 Year follow Up)
	Hypertensive group	Non-Hypertensive group	Hypertensive group	Non-Hypertensive group	Hypertensive group	Non-Hypertensive group	Hypertensive group	Non-Hypertensive group
n (%)	479 (31%)	1068 (69%)	394(35%)	731(65%)	432 (45%)	520 (55%)	303 (51%)	288(49%)
	Mean		Mean		Mean		Mean
	Mean difference ( 95% CL)		Mean difference ( 95% CL)		Mean difference ( 95% CL)		Mean difference ( 95% CL)
Total sleep time (hour)	5.9	6.1	6.2	6.4	7.2	7.2	6.0	6.3
	−0.2* (−0.3, −0.1)		−0.2* (−0.3, −0.1)		−0.04 (−0.2, 0.07)		−0.3 * (−0.5, −0.2)
Sleep efficiency (%)	81.9	84.3	82.0	85.1	79.0	82.6	78.7	81.9
	−2.4* (−3.5, −1.3)		−3.1* (−4.3, −1.9)		−3.7* (−5.0, −2.3)		−3.2* (−4.9, −1.6)
Wake time after sleep onset (min)	63.3	53.8	65.6	55.4	77.5	63.6	76.8	65.6
	9.5* (5.3, 13.7)		10.2* (5.5, 14.8)		13.9* (8.3, 19.4)		11.1* (4.4, 17.9)
% total sleep time in stage N1	11.0	9.1	10.6	9.3	11.1	9.2	11.2	10.0
	1.9 * (1.2, 2.6)		1.3* (0.6, 2.0)		1.9* (1.1, 2.7)		1.2* (0.2. 2.21)
% total sleep time in stage N2	61.8	61.2	62.3	60.3	65.1	64.4	67.9	66.6
	0.6 (−0.4, 1.7)		1.9* (0.7, 3.2)		0.6 (−0.6. 1.8)		1.3 (−0.1, 1.3)
% total sleep time in stage N3	10.1	11.4	10.1	12.2	7.7	8.6	5.9	6.4
	−1.2* (−2.2, −0.2)		−2.2* (−3.3, −1.0)		−0.9 (−2.0, 0.1)		−0.4 (−1.6, 0.7)
% total sleep time in REM	16.8	18.2	16.9	18.0	16.1	17.7	15.0	17.1
	−1.4 * (−2.4, −0.7)		−1.1* (−1.9, −0.4)		−1.6* (−2.3, −0.8)		−2.1* (−3.1, −1.1)
Apnea hypopnea index	8.5	3.6	8.8	4.1	8.9	5.2	10.2	5.9
	4.8 * (3.7, 6.0)		4.6* (3.3, 5.9)		3.7* (2.4, 5.1)		4.3* (2.6, 6.1)

^*indicates p < 0.05 after comparing the hypertension and non-hypertension groups using t-test; REM = rapid eye movement; CL= confidence limits

Hypertension and sleep stages

Table 3 summarizes the parameters resulting from linear mixed-effects regression models; Figure 1 shows the linear sleep characteristics over time and hypertension status. We examined the relationship between hypertension, time and sleep characteristics at each visit, and the interaction between hypertension and time. A significant interaction between time and hypertension (time × hypertension) indicated that having hypertension was associated with a greater rate of decline in total time spent in REM over time: β = − 0.3, 95% confidence limits (CL) (0.1, 06), and total time spent in N3, β = 0.3 95% CL (0.1, 0.6). As shown in Figure 1, the estimated slope of % total sleep time in N3 was larger in the non-hypertensive group than those with hypertension, although the intercept was higher in the hypertensive group. Although the interaction between time and hypertension for total sleep time was not statistically significant, there was a significant main effect of time indicating that total sleep time increased as individuals progressed through the study: β = 3.6, 95% CL (2.1, 5.1). Although the interaction between time and hypertension did not reach statistical significance for sleep efficiency, a significant main effect of hypertension indicated that individuals with hypertension had lower sleep efficiency: β = −1.1, 95% CL (−2.1, −0.1).

Table 3.

Unstandardized parameters resulting from linear mixed effects regression models including time and interaction (n=1,557)

Variables	Total sleep time (min)	Sleep efficiency	Wake after sleep onset	%Total sleep time in stage N1	%Total sleep time in stage N2	%Total sleep time in stage N3	%Total sleep time in stage REM	Sleep Latency
	β (95% CL)
Intercepts	420.1*(401.7, 438.4)	99.4* (96.7, 102.1)	−12.2*(−21.0, −1.4)	5.4* (3.7, 7.0)	54.7* (52.1, 57.3)	17.9* (15.5, 20.2)	23.0* (21.3, 24.6)	3.6 (−3.1, 10.3)
Hypertension (ref = no)	−4 .9 (−11.1, 1.3)	−1.1* (−2.1, −0.1)	3.1 (−1.0, 7.1)	0.6 (−0.02,1.2)	0.2 (−0.8, 1.3)	−0.9 (−1.8,0.007)	0.1 (−0.6,0.7)	−0.02 (−3.4, 3.3)
Time	3.6* (2.1, 5.1)	0.1 (−0.1, 0.3)	−0.2 (−1.1, 1.2)	0.008 (−0.1, 0.2)	0.2 (−0.1, 0.4)	−0.1 (−0.3, 0.2)	−0.04( −0.2, 0.1)	−0.2 (−0.8, 0.4)
Time × Hypertension	−1.3 (−3.0, 0.4)	−0.1 (−0.4, 0.1)	0.1 (−1.0, 1.2)	−0.03 ( −0.2, 0.1)	−0.03 ( −0.3, 0.3)	0.3* (0.1, 0.6)	−0.3*(−0.4, −0.1)	0.3 (−0.4, 1.1)

The models are adjusted for age, sex, body mass index, apnea hypopnea index, continuous positive airway pressure use, diabetes, atherosclerosis, heart failure, and current smoker Note:

^*indicates p values < 0.05; REM = rapid eye movement; CL = confidence limits

Figure 1. Linear association between time and sleep characteristics by hypertension over time.

Note: Models were adjusted for age, sex, body mass index, apnea hypopnea index, continuous positive airway pressure use, diabetes, heart failure, and current smoker. Meaningful differences in total sleep time in rapid eye movement (REM) sleep (%) and N3 by change in time and hypertension status were observed. People with hypertension showed a greater reduction in total sleep time in REM sleep (%) than those without hypertension (p <0.05). People with hypertension had an increase in total time in N3 by time, whereas people without hypertension had a decrease in N3 (p <0.05). Total sleep time (min) increased over time but did not differ by hypertension status. The rate of change in sleep efficiency differed by hypertension but not over time. The rate of changes in total sleep time in N1 and N2 and wake after sleep onset (min) did not differ over time nor by hypertension status.

Hypertension and Subjective Insomnia Complaints

Table 4 shows the logit estimates for the different levels of subjective insomnia complaints (“rarely,” “sometimes,” “often,” or “almost always”) relative to the reference category (“never”). We found a significant interaction effect between time and hypertension for “difficulty falling asleep,” indicating that for those with hypertension, progression of time was associated with a lower likelihood (β = −0.1, 95% confidence interval (CI) (−0.003, −2.0); odds ratio (OR) = 0.9 95% CI (0.8, 1.0)) of not having subjective complaints (“never”). Time was a statistically significant factor for “wake up too early in the morning and can’t fall back asleep” (β = −0.1, 95% CI (−0.2, −0.1); OR = 0.9 95% CI (0.8, 0.9)), and “wake up repeatedly at night” (β = −0.1, 95% CI (−0.003, −2.1); OR = 0.9, 95% CI (0.8, 0.9)).” As follow-up time progressed, individuals were more likely to report higher levels for each of these symptoms. We also calculated the probability of reporting insomnia symptoms over time by hypertension status (Figure 2). In the hypertension group, the probability of answering “never” or “rarely” increased over time, whereas the probability of answering “sometimes,” “often,” or “almost always” decreased over time. In contrast, the estimated probability of remaining in the same “difficulty falling asleep” category in individuals without hypertension was stable over time.

Table 4.

Unstandardized parameters resulting from multi-level logit models for subjective sleep characteristics (n=1,539)

Variables	Difficulty Falling Asleep	Waking up at night with difficulty falling back to sleep	Feelings of excessive daytime sleepiness	Wake up too early in the morning and cant’ fall back asleep	Wake up repeatedly at night
	β (95% CL)
Odds ratios (95% CI)
Intercept 4 “Rarely”	−3.7* (−4.5, −2.8)	−4.4*(−5.4, −3.6)	2.6* (−4.7, 3.0)	−3.9* (−4.7, −3.0)	−2.1* (−2.8, −1.3)
Intercept 3 “Sometimes”	−2.1* (−2.9, −1.3)	−2.6*(−1.8, −0.2)	0.8*(−3.1, −1.5)	−2.3* (−3.1, −1.5)	−1.1* (−1.9, −0.3)
Intercept 2 “Often”	−0.3(−1.1, 0.5)	−1.0*(−1.8, −0.2)	−0.7*(−1.7, −0.1)	−0.9* (−1.7, −0.1)	−0.1 (−0.9, 0.7)
Intercept 1 “Almost Always”	1.7* (0.9, 2.5)	1.0*(0.2, 1.8)	−2.3 (−0.02, 1.6)	0.8 (−0.02, 1.6)	1.4* (0.6, 2.1)
Ref = “Never”	-	-	-	-	-
Hypertension	0.4* (0.1,0.8)	0.3* (−0.07, −0.1)	0.09 (−0.2, 0.4)	−0.05(−0.4, 0.3)	0.1(−0.2, 0.4)
(ref =no)	1.5 (1.1, 2.1)	1.3 (1.0, 1.8)	1.1 (0.8,1.4)	0.9 (0.7, 1.3)	1.0(0.8, 1.5)
Time	−0.007 (0.06, −0.2)	−0.06 (−0.1, 0.002)	−0.02 (−0.008,0.04)	−0.1*(−0.2, −0.1)	−0.1*(−0.003, −2.1)
	1.0 (0.9, 1.1)	0.9 (0.9, 1.0)	1.0 (0.9, 1.0)	0.9 (0.8, 0.9)	0.9 ( 0.9, 1.0)
Time X hypertension	−0.1*(−2.0, −0.003)	−0.05 (−0.1, 0.03)	−0.02(−0.1, 0.05)	0.04(−0.03, 0.11)	0.02 (−0.04, 0.1)
	0.9 (0.8, 1.0)	0.9* (0.9, 1.0)	1.0 (0.9,1.0)	1.0 (0.8, 0.9)	1.0 (1.0, 1.1)

The models are adjusted for age, sex, body mass index, apnea hypopnea index, continuous positive airway pressure use, diabetes, heart failure, and current smoker. Note:

^*indicates p values < 0.05; CL= confidence limits

Figure 2. Probabilities of subjective insomnia complaints categories by time and hypertension.

Note: Y axis indicates the probability of being in one or more subjective insomnia complaint category. The X axis indicates time. The model was adjusted for age, sex, body mass index, apnea hypopnea index, continuous positive airway pressure use, diabetes, heart failure, and current smoker. In the hypertension group, the probabilities of indicating “never” or “rarely” increased over time, whereas the probabilities of being in “sometimes,” “often,” “almost always” decreased over time. However, the probabilities of being in one category did not change over time in individuals without hypertension.

Hypertension and sleep time

Supplementary digital content Table S2 shows the parameters resulting from linear mixed-effects regression models on self-report sleep time, including total nighttime sleep, naps, and total sleep time including naps. However, the interaction between time and hypertension and the main effects of time and hypertension did not reach statistical significance on total sleep time, naps, or subjective total sleep time.

Discussion

The objective of this study was to investigate how hypertension associates longitudinally PSG-assessed sleep characteristics and subjective insomnia complaints in a community-based population of middle-aged to older adults who were followed up to 25 years. This study is one of the few studies to examine longitudinal sleep characteristics by hypertension status in a population-based sample using in-laboratory PSG. The findings revealed that some objective and subjective sleep characteristics are associated with hypertension longitudinally after adjusting for covariates, including AHI and CPAP use. The results indicated that hypertension is significantly associated with longitudinal changes in the percentage of sleep time spent in N3 and REM sleep. Individuals with hypertension had greater rates of decline in % time spent in REM sleep, but had lower rates of decline in % time spent in N3 sleep compared to those without hypertension. For subjective insomnia complaints, compared to individuals without hypertension, individuals with hypertension had a higher likelihood of having at least some level of difficulty falling asleep.

Our findings also indicated that individuals with hypertension may have greater rates of decline in % time spent in REM sleep over time. However, sleep time (i.e., PSG and self-report) and naps did not reveal a statistically significant longitudinal change over time. A prior cross-sectional study found that people with hypertension had shorter proportions of REM sleep as well as longer sleep latency, shorter sleep duration, and lower sleep efficiency in a sample of 106 individuals aged 45–65 years [30]. However, Javaheri et al. (2018) did not find significant baseline mean differences in proportions of REM sleep between individuals who developed hypertension and those who did not. In general, adults spend 20 to 25% of their night time sleep cycle in REM, with cortical arousal, dreaming, and loss of muscle tone as well as elevations in blood pressure, heart rate, cardiac output, and peripheral vascular resistance [2]. Prior research has also suggested that REM sleep changes with age; in particular, REM sleep remains stable from the end of adolescence until age 60 and starts to decrease thereafter [3]. Further, the linear slope of % time spent in REM sleep in total sleep time decreases by 0.6% per decade over the adult life span [36]. Future investigation is needed to identify how age-related hypertension affects different sleep stages and characteristics.

Interestingly, this study revealed that individuals with hypertension had lower rates of declines in time spent in N3 sleep over time than those without hypertension. This finding is different from our hypothesis that hypertensive individuals would have greater declines in slow wave sleep than those without hypertension. Previous studies comparing baseline sleep characteristics by hypertension have demonstrated shorter N3 sleep in individuals with hypertension compared to those without hypertension. Liao et al. (2016) found that people with hypertension had longer N3 sleep than those without the condition [30]. A recent study from the Sleep Heart Health Study cohort also revealed that there were significant baseline differences in the % time spent in slow-wave sleep (hypertensive group: 16.9 ± 11.6 vs. the non-hypertensive group: 18.1 ±10.9, p = 0.0007) and N2 (hypertensive group = 57.1 ± 11.4 vs. the non-hypertensive group = 55.4 ± 10.9, p = 0.002) when compared by hypertension status involving 550 individuals with incident hypertension and 1,334 individuals who did not develop hypertension [24]. These studies characterized the reduction of N3 sleep as a risk factor for developing hypertension, but found baseline differences by later life hypertension, although they did not follow sleep patterns for a longer period of time. Our t-test results showed significant mean differences in N3 in baseline and visit 2, but the mean differences were not significant in visits 3 and 4.

We found that individuals with hypertension have a higher likelihood of greater difficulty falling asleep over time compared to individuals without hypertension. We also found that people without hypertension had persistent sleep complaints over 12 years. In contrast, individuals with hypertension had a higher probability of answering “never” or “rarely” over time, whereas the probability of answering “sometimes,” “often,” and “almost always” decreased over time. Prior studies examining hypertension status and sleep quality have indicated that hypertensive individuals have subjectively poorer sleep quality than control groups [15,37]. Bruno et al. (2013) reported that people with treatment-resistant hypertension had higher rates of sleep latency longer than 30 minutes and daytime dysfunction measured by the Pittsburgh Sleep Quality index (PSQI)[37]. However, our PSG-assessed sleep latency was not associated with an interaction between time and hypertension. Our findings indicated that some subjective insomnia characteristics were associated with progression of time but were not associated with hypertensive status. As time progressed in this study, individuals had a higher likelihood of at least some level of subjective insomnia complaints on questions regarding “feeling of excessive daytime sleepiness,” “wake up too early in the morning and can’t fall back asleep,” and “wake up repeatedly at night.” Our findings are consistent with previous research on age-related sleep changes. Older age has been associated with sleep-related changes, including declines in the ability to fall asleep, increased awakening in night time sleep, waking up too early in the morning, and increased daytime sleepiness, but the magnitude of these changes are not typically large and healthy older adults often do not report sleep problems [38].

One hypothesis to explain differences in sleep characteristics between people with and without hypertension is that the negative feedback of autonomic regulation may alter the sleep-wake cycle, especially REM sleep [39]. In hypertensive states, increased sympathetic nerve traffic may stimulate the cortical area to be more active, causing people to wake up or have more disrupted sleep [40]. Sympathetic hyperactivity is more prominent in nocturnal blood pressure non-dippers, and it may be associated with a higher risk of future cardiovascular events [41]. Although blood pressure increases with REM, baroreflex sensitivity is more effective in buffering increased blood pressure and sympathetic activation at the end of the sleep period, especially during REM sleep before morning awakening [42]. Hypertensive individuals may have decreased baroreflex sensitivity because of shorter REM sleep and disrupted homeostasis of autonomic modulation [27]. A second possible reason for different sleep characteristics by hypertension status could be that hypertension may affect neuronal networks in the brain that may alter the circadian rhythm and sleep by structural brain alterations beyond aging. High blood pressure activates central inflammation signaling (e.g., increased NOD-like receptor family pyrin domain containing 3 (NLRP3), caspase-1 and interleukin-1 beta levels in the paraventricular nucleus of the hypothalamus) [43]. In addition to the inflammation process, endothelial dysfunction from prolonged high blood pressure could increase white matter hyperintensities [44]. People with hypertension have presented alterations in the prefrontal cortex, hippocampus, brainstem, thalamus, inferior temporal cortex, and inferior parietal lobule. These structural neural alterations could lead to changes that are potentially linked to sleep quality [45–47]. A third potential explanation for the differences in sleep characteristics by hypertensive status could be that people with hypertension have a higher prevalence of sleep-disordered breathing. In this study, we controlled for the AHI and CPAP use. The sleep stages may be fragmented by frequent arousals with an increase in stage N1 sleep and a decrease in stage N3 and REM sleep [48].

The strengths of this study include a large, 12–25 year longitudinal, population-based community sample and the use of in-laboratory PSG. We conducted both cross-sectional and longitudinal data analyses to better understand how hypertension is associated with longitudinal sleep characteristics. However, there are some important caveats. First, our definition of hypertension was based on sitting blood pressure without an ambulatory blood pressure measurement. However, our ad hoc analysis (Supplementary digital content Tables S3 and S4) demonstrates moderate correlations (r = 0.57–0.74) between sitting blood pressure and ambulatory blood pressure in the subset of the sample. Future studies incorporating ambulatory blood pressure would provide more comprehensive 24-hour information on blood pressure. Second, the use of in-lab PSG does not capture night-to-night variability of objective sleep characteristics. Although we were able to acquire accurate information regarding sleep characteristics on a given night, this may not reflect individuals’ average sleep data in a natural setting. Additional objective measures, such as actigraphy or in-home PSG, could help determine individuals’ sleep characteristics outside of a clinical setting. Third, the current study analysis models are associative and cannot determine causality. We cannot determine whether hypertension precedes alterations in sleep characteristics or the other way around. However, our approach provides some insights on how hypertension predicts sleep characteristics by assessing the data on sleep characteristics and hypertension over 12–25 years. More longitudinal studies are needed to understand the nature of sleep characteristics in hypertensive people. Fourth, we used a definition of hypertension as blood pressure equal to or above 140/90 mmHg. Although 140/90mmHg may reflect strict hypertension, future studies using different cut off with broader definition of hypertension are needed[49]. Finally, while we were able to account for many sleep-related and other covariates in our models, including sleep apnea and CPAP use, additional unmeasured factors may affect the measured associations.

The results from this study suggest that hypertension may modify changes over time in objective and subjective sleep characteristics in community dwelling, middle-aged to older adults. Additional research is needed to investigate whether these differences in sleep characteristics by hypertension are part of the pathway to future cardiovascular disease events. Larger studies with racial/ethnic and geographically diverse populations may guide the development of targeted interventions to improve sleep characteristics and lower cardiovascular disease risk, among individuals with hypertension.