Association of Obstructive Sleep Apnea With Nighttime Blood Pressure in African Americans: The Jackson Heart Study

Abstract

BACKGROUND

Obstructive sleep apnea (OSA), nocturnal hypertension, and nondipping systolic blood pressure (BP) are each highly prevalent among African Americans. However, few data are available on the association between OSA and nighttime BP in this population.

METHODS

We examined the association of OSA with nighttime BP among African Americans who completed 24-hour ambulatory BP monitoring (ABPM) at Exam 1 (2000–2004) of the Jackson Heart Study (JHS) and subsequently participated in the JHS Sleep Study (2012–2016). Type 3 home sleep apnea testing was used to assess OSA measures, including respiratory event index (REI4%) and percent sleep time <90% oxygen saturation (nocturnal hypoxemia). Nocturnal hypertension was defined as mean asleep systolic BP (SBP) ≥120 mm Hg or diastolic BP (DBP) ≥70 mm Hg. Multivariable linear regression models were fit to estimate the association between each OSA measure and nighttime SBP and DBP.

RESULTS

Among 206 participants who completed ABPM and participated in the Jackson Heart Sleep Study, 50.5% had nocturnal hypertension and 26.2% had moderate to severe OSA (REI4% ≥15 events/hour). After multivariable adjustment, each SD (13.3 events/hour) increase in REI4% was associated with 1.75 mm Hg higher nighttime DBP (95% confidence interval (CI): 0.38, 3.11) and a prevalence ratio of 1.11 (95% CI: 1.00, 1.24) for nocturnal hypertension. Each SD (10.4%) increase in nocturnal hypoxemia was associated with a 1.91 mm Hg higher nighttime SBP (95% CI: 0.15, 3.66).

CONCLUSIONS

Severity of OSA and nocturnal hypoxemia were associated with high nighttime BP in African American participants in the JHS.

Obstructive sleep apnea (OSA) is associated with both prevalent and incident hypertension,^1,2 as well as treatment resistant hypertension.^3–5 A high proportion of African Americans have blood pressure (BP) phenotypes on ambulatory BP monitoring (ABPM), including nocturnal hypertension and nondipping systolic BP (SBP), that have a strong association with cardiovascular morbidity and mortality.^6–10 Mechanisms underlying these associations include repetitive hypoxemia and arousal-related sympathetic nervous system activation that result from OSA.^11–13

Two studies conducted among sleep clinic patients reported that a higher severity of OSA was associated with higher nighttime SBP and diastolic BP (DBP).^14,15 However, these studies included a relatively small sample of patients (n = 54 and n = 90). Another study conducted among 147 participants in the Wisconsin Sleep Cohort Study reported that mean awake and sleeping SBP and DBP were higher among White participants with, compared with those without, OSA.¹⁶ These studies did not include African American adults, a population with a high prevalence of nocturnal hypertension and nondipping SBP.^8,9 In a population-based study, White but not African American participants with a high likelihood of having OSA had a higher relative probability for nocturnal hypertension and nondipping SBP.¹⁷ In that study, African American participants with a low likelihood of having OSA had a high prevalence of nocturnal hypertension (48.8%) and nondipping SBP (38.4%), resulting in the lower probability of these outcomes. Identifying factors associated with higher nocturnal hypertension and nondipping SBP can be used to develop interventions to decrease their prevalence among African Americans. We conducted an analysis of the Jackson Heart Study (JHS), a community-based cohort study comprised exclusively of African American adults, to test the hypothesis that OSA is associated with higher nighttime BP, as well as a higher prevalence of nocturnal hypertension and nondipping SBP.

METHODS

Study population

The JHS is a community-based study designed to identify factors that explain the high rate of cardiovascular disease among African American adults and find approaches for reducing this risk. Details regarding the design of the JHS have been published elsewhere.^18,19 Briefly, between 2000 and 2004, the JHS enrolled 5,306 African Americans, aged 20–95 years, from urban and rural areas of the 3 counties (Hinds, Madison, and Rankin) that comprise the Jackson, Mississippi metropolitan area. Following an initial exam at baseline (Exam 1 in 2000–2004), Exam 2 was conducted between 2005 and 2008, and Exam 3 was conducted between 2009 and 2013. The current analysis included participants (N = 216) with a complete 24-hour ABPM recording (defined below) at Exam 1 who subsequently completed sleep apnea testing as part of the Jackson Heart Sleep Study (JHSS; 2012–2016).²⁰ Since OSA was only assessed once but changes in OSA severity track closely with body mass index (BMI),²¹ we excluded participants (N = 5) with a >30% change in BMI from Exam 1 to the JHSS to reduce the impact of large changes in BMI on OSA severity across examinations. We also excluded participants with missing data on BMI (N = 1), smoking status (N = 1), and statin or antihypertensive medication use (N = 3) at Exam 1. These exclusions resulted in a final analytic sample of 206 participants. The protocol for the JHS was approved by the institutional review boards at the participating institutions, including Jackson State University, Tougaloo College, and the University of Mississippi Medical Center and the JHSS was additionally approved by the Brigham and Women’s Hospital institutional review board. All participants provided written informed consent prior to participation.

Sleep measures

Details regarding the sleep measures included in the JHSS have been published elsewhere.^20,22 Briefly, sleep apnea was assessed with a validated Type 3 home sleep apnea device (Embletta-Gold device; Embla, Broomfield, CO)^23,24 that recorded nasal pressure; thoracic and abdominal inductance plethysmography; finger pulse oximetry; body position; and electrocardiography. Sleep studies were scored according to published guidelines.^25,26 Obstructive apneas were determined to be present when the amplitude (peak to trough) of the nasal pressure signal was flat or nearly flat for >10 seconds and accompanied by respiratory effort on either the thoracic or abdominal inductance plethysmography bands.²⁶ Hypopneas were determined to be present if a ≥30% reduction of amplitude occurred in the nasal pressure signal or, if unclear, a reduction in the amplitude of respiratory effort in the respiratory inductance bands for ≥10 seconds, but did not meet criteria for apneas. Events were further classified based on the degree of associated desaturation (≥3% and ≥4%). The respiratory event index (REI) was calculated as the sum of all obstructive apneas plus hypopneas associated with 3% (REI3%) or 4% (REI4%) oxygen desaturation divided by the estimated sleep time.²⁵ Sleep apnea was categorized by the following standard REI categories: <5 (unaffected); ≥5 to <15 (mild), ≥15 to <30 (moderate), and ≥30 (severe). For the purpose of these analyses, we further categorized moderate or severe OSA based on the REI ≥15 using REI4%. Nocturnal hypoxemia was quantified as percent estimated sleep time with <90% oxyhemoglobin saturation during the sleep period.

Ambulatory blood pressure monitoring

ABPM was conducted following Exam 1 using the validated Spacelabs 90207 monitor (SpaceLabs Healthcare, Snoqualmie, WA) and an appropriately sized cuff.^18,27 BP was measured every 20 minutes over a 24-hour period. Consistent with the criteria used by the International Database of Ambulatory Blood Pressure in relation to Cardiovascular Outcome (IDACO), daytime was defined as 10:00 am to 8:00 pm and nighttime was 12:00 am to 6:00 am.²⁸ Participants with ≥10 valid daytime and ≥5 valid nighttime SBP and DBP measurements were considered to have a complete ABPM recording. We have previously shown that the prevalence of nocturnal hypertension and nondipping SBP was similar when a complete ABPM recording was defined using the IDACO criteria and the 2013 European Society of Hypertension criteria.^29,30 Mean daytime and nighttime SBP and DBP were each defined based on the mean of all available readings obtained during the respective daytime and nighttime periods. Nocturnal hypertension was defined as mean nighttime SBP ≥120 mm Hg or mean nighttime DBP ≥70 mm Hg.³¹ The percent declines in SBP and DBP from daytime to nighttime measurements were each calculated as one minus the ratio of the mean nighttime to mean daytime BP multiplied by 100.³¹ Nondipping SBP was defined as the percent decline from mean daytime to mean nighttime SBP <10% or an increase in SBP from daytime to nighttime.

Covariates

All covariates other than BMI and sleep measures were collected during an in-home interview and a clinic examination at Exam 1. BMI was measured at both Exam 1 and the JHSS. Interviewer-administered questionnaires were used to collect information on age, sex, highest level of education attained, current cigarette smoking, alcohol consumption, self-reported medication use, symptoms of depression, and history of cardiovascular disease. Antihypertensive medication use was determined by self-report, and statin use was defined based on a pill bottle review. During Exam 1 and the JHSS, trained staff measured height and weight. BMI was calculated as weight in kilograms divided by height in meters squared. Current cigarette smoking was defined by affirmative responses to the question “Do you now smoke cigarettes?” Alcohol consumption was determined by whether the participant reported consuming any alcoholic beverages within the previous 12 months. During Exam 1, clinic BP was measured by trained staff using a Hawksley random zero sphygmomanometer and Littmann stethoscope following a standardized protocol. The random zero BP measurements were calibrated to a semi-automated oscillometric device (Omron HEM-907XL, Omron Healthcare, Lake Forest, IL).³² Hypertension was defined as a clinic SBP ≥140 mm Hg, clinic DBP ≥90 mm Hg, or self-reported use of antihypertensive medication. Diabetes was defined as a fasting blood glucose ≥126 mg/dl, HbA1c ≥6.5% (48 mmol/mol), or use of insulin or other glucose lowering medications within 2 weeks prior to the examination. Estimated glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation.³³ History of cardiovascular disease was defined as a self-reported history of myocardial infarction or evidence of a myocardial infarction on the study electrocardiogram, self-reported history of any cardiac procedure, or self-reported history of stroke.

Statistical analyses

Participant characteristics were summarized overall and for individuals with and without nocturnal hypertension and with and without nondipping SBP, separately. The statistical significance of differences by nocturnal hypertension and nondipping SBP status was determined using t-tests and chi-square tests, as appropriate. Linear regression models were used to examine the associations of OSA measures (i.e., REI4% and nocturnal hypoxemia) with nighttime SBP and DBP in an unadjusted model (Model 1). Covariates were selected a priori and included age, sex, BMI, smoking status, and alcohol consumption (Model 2). In sensitivity analyses, we conducted regression models that included further adjustment for antihypertensive medication use, diabetes, and estimated glomerular filtration rate (Model 3), as well as these variables and clinic SBP when the outcome was nighttime SBP and clinic DBP when the outcome was nighttime DBP (Model 4). We also conducted tests of interactions between obesity status (i.e., BMI <30 kg/m² (nonobese) vs. BMI ≥30 kg/m² (obese)) and OSA measures on nighttime SBP and DBP. Given the small sample size and, therefore, the potential lack of power to detect significant interactions, we also stratified the analyses by obesity status. In secondary analyses, the association of OSA severity with nocturnal hypertension and nondipping SBP were estimated using Poisson regression models with robust variance estimators, deriving prevalence ratios (PRs) with 95% confidence intervals (CIs). Given the small sample size and number of outcomes, adjustment was restricted to the variables in Model 2 described above. The linear and Poisson regression model assumptions were assessed and were not violated. We examined interactions between obesity status and OSA severity on nocturnal hypertension and nondipping SBP and stratified analyses by obesity status, as described above. In sensitivity analyses, we repeated the above analyses using the 2017 American College of Cardiology/American Heart Association (ACC/AHA) BP guideline definition of nocturnal hypertension (mean nighttime SBP ≥110 mm Hg or mean nighttime DBP ≥65 mm Hg).³⁴ Analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC). Statistical significance was defined as a 2-tailed P value <0.05.

RESULTS

Of the 206 participants included in the current analysis, 104 (50.5%) had nocturnal hypertension and 135 (65.5%) had nondipping SBP (Table 1). Compared with participants without nocturnal hypertension, those with nocturnal hypertension were more likely to be male and have hypertension based on clinic BP measurements or antihypertensive medication use, and less likely to report alcohol consumption. Additionally, participants with nocturnal hypertension had higher clinic, daytime and nighttime SBP and DBP, a smaller decline in SBP from daytime to nighttime, and a greater severity of OSA (i.e., higher mean REI4%) compared with their counterparts without nocturnal hypertension. Participants with nondipping SBP were less likely to smoke cigarettes, had higher nighttime SBP and DBP, lower daytime DBP, and a smaller decline in SBP from daytime to nighttime compared with participants without nondipping SBP.

Table 1.

Characteristics of participants overall and with and without nocturnal hypertension or nondipping systolic blood pressure

		Nocturnal hypertension			Nondipping systolic blood pressure
Characteristic	Total (N = 206)	Yes (N = 104)	No (N = 102)	P value	Yes (N = 135)	No (N = 71)	P value
Demographics
Age, years	55.6 ± 9.9	56.3 ± 10.4	54.9 ± 9.4	0.30	56.0 ± 10.0	54.9 ± 9.8	0.46
Elapsed time from Exam 1 to sleep study, years	12.3 ± 1.5	12.5 ± 1.4	12.2 ± 1.6	0.17	12.3 ± 1.5	12.4 ± 1.5	0.76
Male	68 (33.0%)	42 (40.4%)	26 (25.5%)	0.02	42 (31.1%)	26 (36.6%)	0.42
Education at Exam 1				0.85			0.40
Less than high school	24 (11.6%)	13 (12.5%)	11 (10.8%)		15 (11.1%)	9 (12.7%)
High school or GED	33 (16.0%)	18 (17.3%)	15 (14.7%)		25 (18.5%)	8 (11.3%)
Some college/training	42 (20.4%)	19 (18.3%)	23 (22.5%)		24 (17.8%)	18 (25.3%)
College degree	107 (51.9%)	54 (51.9%)	53 (52.0%)		71 (52.6%)	36 (50.7%)
Anthropometrics
BMI at Exam 1, kg/m2	30.6 ± 5.3	31.2 ± 5.7	30.0 ± 4.7	0.12	31.1 ± 5.8	29.7 ± 4.0	0.04
BMI at sleep study, kg/m2	30.7 ± 6.0	31.3 ± 6.9	30.1 ± 4.9	0.15	31.2 ± 6.7	29.8 ± 4.4	0.06
Change in BMI from Exam 1 to sleep study, kg/m2	0.1 ± 3.1	0.1 ± 3.0	0.1 ± 3.2	0.89	0.1 ± 3.2	0.1 ± 2.9	0.99
Health behaviors and characteristics, Exam 1
Current cigarette smoking	15 (7.3%)	7 (6.7%)	8 (7.8%)	0.76	5 (3.7%)	10 (14.1%)	0.006
Alcohol consumption	92 (44.7%)	39 (37.5%)	53 (52.0%)	0.04	54 (40.0%)	38 (53.5%)	0.06
Hypertension	113 (54.8%)	66 (63.5%)	47 (46.1%)	0.01	74 (54.8%)	39 (54.9%)	0.99
Diabetes	36 (17.5%)	20 (19.2%)	16 (15.7%)	0.50	23 (17.0%)	13 (18.3%)	0.82
History of CVD	8 (3.9%)	4 (3.8%)	4 (3.9%)	0.99	7 (5.2%)	1 (1.4%)	0.27
Statin medication use	12 (5.8%)	4 (3.8%)	8 (7.8%)	0.22	6 (4.4%)	6 (8.4%)	0.35
Antihypertensive medication use	106 (51.5%)	56 (50.5%)	50 (49.0%)	0.49	68 (50.4%)	38 (53.5%)	0.67
eGFR, ml/min/1.73 m2	95.9 ± 17.7	95.0 ± 17.5	96.7 ± 18.0	0.50	94.4 ± 18.1	98.7 ± 16.8	0.10
Blood pressure measures, Exam 1
Clinic systolic blood pressure, mm Hg	124.5 ± 14.7	131.9 ± 12.9	117.0 ± 12.5	<0.001	125.8 ± 14.3	122.2 ± 15.3	0.10
Clinic diastolic blood pressure, mm Hg	78.7 ± 9.3	81.5 ± 9.2	75.9 ± 8.7	<0.001	78.9 ± 9.5	78.5 ± 9.1	0.77
Daytime systolic blood pressure, mm Hg	127.4 ± 12.4	133.6 ± 11.4	121.0 ± 9.9	<0.001	126.2 ± 12.1	129.7 ± 12.6	0.05
Daytime diastolic blood pressure, mm Hg	78.4 ± 8.8	82.1 ± 8.6	74.6 ± 7.3	<0.001	76.9 ± 9.2	81.2 ± 7.4	<0.001
Nighttime systolic blood pressure, mm Hg	118.2 ± 13.4	127.9 ± 11.0	108.3 ± 6.8	<0.001	122.3 ± 13.3	110.4 ± 9.6	<0.001
Nighttime diastolic blood pressure, mm Hg	68.5 ± 8.9	74.6 ± 7.7	62.2 ± 4.6	<0.001	70.6 ± 9.2	64.5 ± 6.7	<0.001
Nocturnal decline in SBP/DBP ratio from wake to sleep, %	7.1 ± 7.4	4.0 ± 7.1	10.2 ± 6.2	<0.001	3.0 ± 5.1	14.7 ± 4.2	<0.001
Sleep measures, sleep study
REI (4% desaturation), events/hour	7.6 ± 14.5	8.9 ± 20.1	5.7 ± 10.6	0.009	7.6 ± 15.0	7.5 ± 13.3	0.77
Categories of sleep apnea severity				0.06			0.69
None (0 ≤ REI4% < 5)	80 (38.8%)	32 (30.8%)	48 (47.1%)		53 (39.3%)	27 (38.0%)
Mild (5 ≤ REI4% < 15)	72 (34.9%)	38 (36.5%)	34 (33.3%)		46 (34.1%)	26 (36.6%)
Moderate (15 ≤ REI4% < 30)	30 (14.6%)	18 (17.3%)	12 (11.8%)		18 (13.3%)	12 (16.9%)
Severe (REI4% ≥ 30)	24 (11.6%)	16 (15.4%)	8 (7.8%)		18 (13.3%)	6 (8.4%)
Moderate or severe sleep apnea (REI4% ≥ 15)	54 (26.2%)	34 (32.7%)	20 (19.6%)	0.03	36 (26.7%)	18 (25.4%)	0.84
Nocturnal hypoxemia, %	0.5 ± 2.6	0.5 ± 3.3	0.5 ± 2.4	0.30	0.4 ± 2.5	0.7 ± 2.8	0.67

Numbers in the table are mean ± SD or number (percentage). Respiratory event index and nocturnal hypoxemia data were skewed and, therefore, reported as median ± interquartile range. Abbreviations: BMI, body mass index; CVD, cardiovascular disease; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; GED, general education diploma; REI, respiratory event index; SBP, systolic blood pressure. Nocturnal hypoxemia was calculated as the percent sleep time with less than 90% oxygen saturation. Nocturnal hypertension was defined as mean sleep systolic blood pressure ≥120 mm Hg or mean diastolic blood pressure ≥70 mm Hg. Nondipping blood pressure was defined as a decline in sleep systolic blood pressure relative to awake systolic blood pressure <10%.

P values <0.05 are bolded.

Nighttime SBP and DBP

A 1 SD (13.3 events/hour) higher REI4% was associated with a 2.64 mm Hg (95% CI: 0.84, 4.44) higher mean nighttime SBP level and a 1.98 mm Hg (95% CI: 0.79, 3.17) higher mean nighttime DBP level in an unadjusted model (Table 2). These estimates were attenuated after adjusting for age, sex, BMI, cigarette smoking, and alcohol consumption (1.71 mm Hg; 95% CI: −0.29, 3.71 for mean nighttime SBP and 1.76 mm Hg; 95% CI: 0.40, 3.12 for mean nighttime DBP). Each SD higher REI4% was associated with 1.69 mm Hg; 95% CI: −0.33, 3.71 and 1.75 mm Hg; 95% CI: 0.38, 3.11 higher mean nighttime SBP and DBP, respectively, after additional adjustment for antihypertensive medication use, diabetes, and estimated glomerular filtration rate.

Table 2.

Difference in nighttime systolic and diastolic blood pressure associated with obstructive sleep apnea severity and nocturnal hypoxemia

	Mean nighttime systolic blood pressure		Mean nighttime diastolic blood pressure
Sleep characteristic	Estimate (95 CI%) in mm Hg	P value	Estimate (95 CI%) in mm Hg	P value
REI (4%), per 13.3 events/hour
Model 1	2.64 (0.84, 4.44)	0.004	1.98 (0.79, 3.17)	0.001
Model 2	1.71 (−0.29, 3.71)	0.10	1.76 (0.40, 3.12)	0.01
Model 3	1.69 (−0.33, 3.71)	0.10	1.75 (0.38, 3.11)	0.01
Nocturnal hypoxemia, per 10.4%
Model 1	2.31 (0.51, 4.12)	0.01	1.31 (0.10, 2.51)	0.04
Model 2	1.93 (0.18, 3.67)	0.03	1.18 (−0.02, 2.38)	0.06
Model 3	1.91 (0.15, 3.66)	0.04	1.18 (−0.03, 2.38)	0.06

Abbreviations: CI, confidence interval; REI, respiratory event index. Nocturnal hypoemia was calculated as the percent sleep time with less than 90% oxygen saturation. Model 1 is unadjusted. Model 2 included adjustment for age, sex, body mass index, cigarette smoking, and alcohol consumption. Model 3 included adjustment for the Model 2 variables plus, antihypertensive medication use, diabetes, and estimated glomerular filtration rate.

A 1 SD higher nocturnal hypoxemia (10.4%) was associated with a 2.31 mm Hg (95% CI: 0.51, 4.12) higher mean nighttime SBP and a 1.31 mm Hg (95% CI: 0.10, 2.51) higher mean nighttime DBP in an unadjusted model. These estimates were attenuated after adjustment for age, sex, BMI, cigarette smoking, and alcohol consumption (1.93 mm Hg; CI: 0.18, 3.67 for mean nighttime SBP and 1.18 mm Hg; CI: −0.02, 2.38 for mean nighttime DBP). After additional adjustment for antihypertensive medication use, diabetes, and estimated glomerular filtration rate, each SD higher nocturnal hypoxemia was associated with 1.91 mm Hg; 95% CI: 0.15, 3.66 higher mean nighttime SBP and 1.18 mm Hg; 95% CI: −0.03, 2.38 higher mean nighttime DBP.

In a sensitivity analysis with additional adjustment for clinic BP (Model 4), a 1 SD (13.3 events/hour) higher REI4% was associated with a 1.24 mm Hg (95% CI: −0.53, 3.01) higher mean nighttime SBP level and a 1.45 mm Hg (95% CI: 0.16, 2.74) higher mean nighttime DBP level (Supplementary Table S1 online). A 1 SD higher nocturnal hypoxemia (10.4%) was associated with a 1.66 mm Hg (95% CI: 0.12, 3.20) higher mean nighttime SBP and a 1.17 mm Hg (95% CI: 0.04, 2.29) higher mean nighttime DBP. We also examined potential effect modification by obesity status. Statistical tests of interaction were not significant (all interaction P values >0.05). However, analyses stratified by obesity status suggested stronger associations for obese vs. nonobese participants (Supplementary Table S2 online).

Nocturnal hypertension and nondipping SBP

Each SD higher REI4% (SD: 13.3 events/hour) was associated with a higher probability for nocturnal hypertension in an unadjusted model (PR = 1.18; 95% CI: 1.08, 1.29; Table 3). This association was attenuated after adjustment for age, sex, BMI, cigarette smoking, and alcohol consumption (PR = 1.11; 95% CI: 1.00, 1.24). There was no evidence of an association between nocturnal hypoxemia and nocturnal hypertension in an unadjusted model (PR = 1.03; 95% CI: 0.92, 1.15) or after adjustment for age, sex, BMI, cigarette smoking, and alcohol consumption (PR = 1.01; 95% CI: 0.90, 1.13). There was no evidence of an association between either REI4% or nocturnal hypoxemia and nondipping SBP. In secondary analyses, there was an interaction by obesity status for the association between REI4% and nondipping SBP (Supplementary Table S3 online). In obesity-stratified adjusted analyses, associations tended to be stronger among the obese group. However, within-stratum associations were not significant. All other statistical tests of interaction were not significant (interaction P values >0.05).

Table 3.

Prevalence ratios for nocturnal hypertension and nondipping systolic blood pressure associated with obstructive sleep apnea severity and nocturnal hypoxemia

	Nocturnal hypertension		Nondipping systolic blood pressure
Sleep characteristic	PR (95% CI)	P value	PR (95% CI)	P value
REI (4%), per 13.3 events/hour
Model 1	1.18 (1.08, 1.29)	<0.001	1.02 (0.93, 1.12)	0.68
Model 2	1.11 (1.00, 1.24)	0.06	1.02 (0.93, 1.12)	0.69
Nocturnal hypoxemia, per 10.4%
Model 1	1.03 (0.92, 1.15)	0.63	1.02 (0.95, 1.10)	0.54
Model 2	1.01 (0.90, 1.13)	0.89	1.02 (0.97, 1.08)	0.38

Abbreviations: CI, confidence interval; PR, prevalence ratio; REI, respiratory event index. Nocturnal hypoxemia was calculated as the percent sleep time with less than 90% oxygen saturation. Nocturnal hypertension is defined as mean sleep systolic blood pressure ≥120 mm Hg or mean diastolic blood pressure ≥70 mm Hg. Nondipping systolic blood pressure was defined as a decline in sleep systolic blood pressure relative to awake systolic blood pressure <10%. Model 1 is unadjusted. Model 2 included adjustment for age, sex, body mass index, cigarette smoking, and alcohol consumption.

The characteristics of participants with and without nocturnal hypertension using the 2017 ACC/AHA BP guideline definition of nocturnal hypertension (i.e., mean nighttime SBP ≥110 mm Hg or mean nighttime DBP ≥65 mm Hg) are presented in Supplementary Table S4 online. Each SD higher REI4% (SD: 13.3 events/hour) was associated with a higher probability for nocturnal hypertension in an unadjusted model (PR = 1.08; 95% CI: 1.04, 1.13; Supplementary Table S5 online). After adjustment for age, sex, BMI, cigarette smoking, and alcohol consumption, the PR was 1.04; 95% CI: 0.99, 1.09. There was no evidence of an association between nocturnal hypoxemia and nocturnal hypertension in an unadjusted model (PR = 1.00; 95% CI: 0.93, 1.08) or after adjustment for age, sex, BMI, cigarette smoking, and alcohol consumption (PR = 0.99; 95% CI: 0.92, 1.06). Statistical tests of interaction were not significant (all interaction P values >0.05; data not shown).

DISCUSSION

The prevalence of nocturnal hypertension, nondipping SBP, and moderate or severe OSA was high in this sample of African Americans enrolled in the JHS. Moreover, these data suggest that individuals with more severe OSA or more nocturnal hypoxemia are also more likely to have higher nocturnal BP. After multivariable adjustment, more frequent respiratory disturbances (REI4%) were associated with higher nighttime DBP and more severe hypoxemia was associated with higher nighttime SBP. Our findings are also consistent with possibly stronger associations among obese vs. nonobese individuals.

Studies have consistently demonstrated an association between OSA and higher clinic BP.^2,35,36 However, few data are available on the association between OSA and nighttime BP, with a particular paucity of research in African Americans. Moderate to severe OSA was associated with higher nighttime SBP, higher daytime and nighttime DBP, and lower percent dipping in a case–control study of 45 patients recruited from a sleep clinic.¹⁵ In a sample of 54 patients from a different sleep clinic, higher severity of OSA was associated with higher mean 24-hour and nocturnal BP, and a nondipping BP profile.¹⁴ In addition to small sample sizes, patients in these studies were enrolled from a sleep clinic, which may limit the generalizability of these results. Furthermore, both studies were conducted outside of the United States and, therefore, did not include African Americans. The current findings extend the results from prior studies to a community-based sample of African Americans who completed in-home sleep testing and ABPM.

Acutely, nocturnal BP is likely particularly reflective of responses to OSA-related stressors.^11,12,37 Studies have demonstrated that BP surges transiently, as high as SBP/DBP of 240/130 mm Hg,³⁷ with each respiratory disturbance, changes that are attributable to both hypoxemia and arousal-related sympathetic nervous system activation.^11,12 Furthermore, persistently high daytime and/or nighttime BP may result from recurrent oxidative stress, vascular remodeling, and tonic chemoreflex activation, all of which are known sequelae of untreated OSA. Therefore, measurement of 24-hour BP profiles may provide evidence for early changes in vascular health, as well as be more sensitive to OSA-related stressors. The findings from the present study provide further support to recommendations for the use of ABPM, particularly among African Americans, a population with a high prevalence of OSA and hypertension.^6–9,20 In secondary analyses, associations between OSA severity and nighttime SBP and DBP were somewhat stronger among obese compared with nonobese individuals. While these data need to be cautiously interpreted because there was not a statistically significant interaction, they suggest that obese individuals with OSA may have higher nighttime SBP and DBP than nonobese individuals with OSA. The stronger association among obese individuals may be due to differences in background levels of sympathetic activation, endothelial dysfunction, and insulin resistant, or possibly due to differences in the underlying OSA phenotype itself.

The observed differential association between REI4% and nighttime DBP vs. nocturnal hypoxemia and nighttime SBP are of potential interest. Although it is important not to over-interpret findings from the current study, prior research suggests that the presence of OSA may result in elevations in DBP that precede elevations in SBP.³⁸ Additional research is needed to better understand the specific impact of OSA-related stressors on vascular function, such as those related to breathing disturbances (which cause both intermittent hypoxemia and arousal) compared with those associated primarily with hypoxemia.

We also examined nocturnal hypertension and nondipping SBP in secondary analyses, as these are clinically meaningful phenotypes.³⁹ Higher REI4% was associated with a higher probability for nocturnal hypertension in an unadjusted model, as well as after adjustment for age, sex, BMI, cigarette smoking, and alcohol consumption. The lack of associations with nondipping SBP may reflect several factors, including less consistent patterns of daytime and clinic SBP compared with nocturnal BP levels, temporal factors between ABPM and sleep assessments, the influence of age-related changes in SBP that may mask effects of sleep apnea,⁴⁰ and the relatively modest sample size.

Limited understanding of the association between sleep apnea and nocturnal BP is a substantial gap in scientific knowledge given the high prevalence of both conditions among African Americans, as well as their association with cardiovascular morbidity and mortality.^6–9,20 A high prevalence of nocturnal hypertension among African Americans in the current study is consistent with results for African Americans in the Coronary Artery Risk Development in Young Adults (CARDIA) study (50.3%), which was almost double that of White participants in the CARDIA study.⁹ While the selection and geographical location of JHS and CARDIA are different, the high prevalence of nocturnal hypertension among African Americans in both studies suggests that the current findings are not specific only to the JHS. A separate study among CARDIA participants reported that African American participants with a low likelihood of having OSA had a high prevalence of nocturnal hypertension (48.8%) and nondipping SBP (38.4%).¹⁷ The high prevalence of nocturnal hypertension and nondipping SBP among African American participants with a low likelihood of OSA suggests that other factors, including salt sensitivity,⁴¹ may be a strong contributor to high asleep BP in this population and may explain the modest association between OSA/hypoxemia and BP (<2 mm Hg) observed in this study.

The current study has several important strengths. Data on ambulatory BP and sleep were collected in a large community-based sample of African Americans following a standardized protocol. Additionally, use of home sleep testing allowed for objective assessments of OSA severity and nocturnal hypoxemia. Extensive data collection permitted adjustment for multiple confounders. Despite these strengths, the results should be interpreted in the context of potential limitations. Although this study included a large community-based sample of African Americans, the sample may not be representative of individuals of lower socioeconomic status (51.9% of this sample had a college degree), African Americans in other communities or other racial/ethnic groups. Further, this sample was comprised of more women than men (67% vs. 33%, respectively). Only 206 of 5,306 individuals (3.9%) completed both ABPM and the JHSS, which reduced the statistical power to detect small to modest associations. Multiple correlated outcomes and exposures were tested, which may have resulted in Type I error. Only a single 24-hour ABPM and a single sleep study were performed, which may have resulted in misclassification of participants’ nocturnal hypertension and nondipping BP status as well as OSA status. In addition, OSA was assessed approximately 12 years after ABPM measurements. We assumed that individuals with OSA at the sleep examination would likely also have had OSA when studied at the time of ABPM. In a prior study of adults, median 5-year change in REI was estimated to be less than 1 event/hour,²¹ although REI may change to a greater degree with large changes in BMI. To minimize misclassification in the current study, we excluded participants whose BMI changed by >30% between the conduct of ABPM and assessment of OSA (n = 5). Nonetheless, some degree of misclassification is likely, which would have resulted in a more conservative, underestimate of true associations. Despite these limitations, this study provides new data that highlight an association of OSA severity and nocturnal hypoxemia with nocturnal BP among African Americans, a population with a high prevalence of hypertension-associated morbidities.

In conclusion, the prevalence of nocturnal hypertension, nondipping SBP, and OSA was each high in this community-based sample of African Americans. Additionally, more severe OSA, as measured by a higher REI4%, was associated with higher nighttime DBP and more severe hypoxemia was associated with higher nighttime SBP. Previous studies have reported a high prevalence of OSA and high nighttime BP in the JHS.^8,20 The current study extends these previous findings to suggest that OSA severity and nocturnal hypoxemia is associated with high nighttime BP in African Americans, a population in whom there is under-recognition and under-treatment of OSA. These findings support the need for developing efficient strategies for screening populations at risk for both hypertension-associated morbidities and OSA, and testing the role of OSA treatment in high-risk groups on reducing health disparities.

FUNDING

The Jackson Heart Study (JHS) is supported and conducted in collaboration with Jackson State University (HHSN268201800013I), Tougaloo College (HHSN268201800014I), the Mississippi State Department of Health (HHSN268201800015I), and the University of Mississippi Medical Center (HHSN268201800010I, HHSN268201800011I, and HHSN268201800012I) contracts from the National Heart, Lung, and Blood Institute (NHLBI) and the National Institute for Minority Health and Health Disparities (NIMHD). This work was also supported by the following grants: R01 HL110068 (SR), 5R35 HL135818 (SR), R01 HL117323 (PM), K01 HL138211 (DJ), and K23 HL141682-01A1 (MA) from NHLBI and 18AMFDP34380732 (MA), 15SFRN2390002 (SJT, JB, DC, and PM), and 19CDA34660139 (SJT) from the American Heart Association. This work was funded, in part, by the Intramural Program at the National Institutes of Health (NIH), National Institute of Environmental Health Sciences Z1A ES103325-01 (CLJ).

DISCLOSURE

The authors declared no conflict of interest.