Sleep Disordered Breathing, Daytime Symptoms, and Functional Performance in Stable Heart Failure

Abstract

Study Objectives:

To evaluate characteristics of sleep disordered breathing (SDB); clinical and demographic correlates of SDB; and the extent to which SDB explains functional performance and symptoms in stable heart failure patients receiving care in structured HF disease management programs.

Design:

Cross-sectional, observational study.

Setting:

Structured heart failure disease management programs.

Participants:

170 stable chronic heart failure patients (mean age = 60.3 ± 16.8 years; n = 60 [35%] female; n = 50 [29%] African American; left ventricular ejection fraction mean = 32 ± 14.6).

Interventions:

N/A

Measurements and Results:

Full polysomnography was obtained for one night on participants in their homes. Participants completed the 6-minute walk, 3 days of actigraphy, MOS-SF 36, Epworth Sleepiness Scale, Pittsburgh Sleep Quality Index, Multi-Dimensional Assessment of Fatigue Scale, and the Centers for the Epidemiological Studies of Depression Scale. Fifty-one percent had significant SDB; Sixteen (9%) of the total sample had central sleep apnea. Severe SDB was associated with a 4-fold increase in the likelihood of poor self-reported physical function (OR = 4.15, 95%CI = 1.19–14.57) and CSA was associated with low levels of daytime mobility (OR = 4.09, 95%CI = 1.23–13.62) after controlling for clinical and demographic variables. There were no statistically significant relationships between SDB and daytime symptoms or self-reported sleep, despite poorer objective sleep quality in patients with SDB.

Conclusions:

Severe SDB is associated with poor physical function in patients with stable HF but not with daytime symptoms or self-reported sleep, despite poorer objective sleep quality in patients with SDB.

Citation:

Redeker NS; Muench U; Zucker MJ; Walsleben J; Gilbert M; Freudenberger R; Chen M; Campbell D; Blank L; Berkowitz R; Adams L; Rapoport DM. Sleep disordered breathing, daytime symptoms, and functional performance in stable heart failure. SLEEP 2010;33(4):551-560.

SLEEP DISORDERED BREATHING (SDB), INCLUDING OBSTRUCTIVE AND CENTRAL SLEEP APNEA, IS COMMON IN PEOPLE WITH CHRONIC HEART FAILURE (HF) and appears to be associated with objective and self-report measures of functional performance,^1,2 excessive daytime sleepiness,^2,3 self-reported poor sleep,⁴ and depressive symptoms.¹ However, findings have been inconsistent,^5–7 and previous studies have not addressed the clinical or demographic factors that may contribute to both SDB and its daytime consequences in HF patients. Understanding the extent to which SDB may be associated with daytime symptoms and functional performance may help to identify patients who are at high risk for these problems and may benefit most from improvements in daytime function through treatment.

Depending on the population studied, obstructive apnea (OSA) and/or central sleep apnea (CSA) occur in 24% to 82% of HF patients.^5–14 The relative odds of HF in the Sleep Heart Health Study (SHHS), a study of sleep in cardiovascular cohorts, was 2.38 for people in the highest vs. lowest quartile of the respiratory disturbance index.¹⁵ As many as 50% of patients with either systolic HF^5,8,12,16 or HF with preserved systolic function¹⁷ have OSA. Between 15% and 62% of systolic HF patients^{5,8,9,12,18,19} and 20% of patients¹⁷ with preserved systolic function have CSA. Most previous studies have focused on patients with systolic dysfunction^19–22 and included only men or very small proportions of women^2,3,6,20,21; however, 40% to 71%^23,24 of patients with HF have preserved systolic function and women represent approximately 50% of patients with HF over the lifespan.

Although SDB was associated with lower actigraph-recorded daytime activity duration in male HF patients² and oxygen uptake,¹ but not the shuttle-walk test, in another study,¹ SDB has not been consistently related to self-reported physical function in HF patients.^15,6,25 A study of 700 HF patients¹² revealed that HF patients with CSA had lower 6-minute walk test (6m WT) distances than patients with OSA or patients with no SDB. However, the potentially confounding effects of age, gender, and clinical characteristics on SDB and functional performance were not evaluated.

SDB has been associated with objective,³² but not self-report measures of sleepiness^2,5–7,26 and depressive symptoms among HF patients.¹ There was a linear relationship between SDB and vitality in the SHHS²⁷ that included a small proportion of HF patients; but, to our knowledge, the extent to which SDB explains fatigue, a common and disabling symptom in HF patients, has not been examined.

The purposes of this study were to evaluate: (1) the characteristics of SDB in community-residing patients with stable HF; (2) the demographic and clinical correlates of SDB severity and predominant central vs. obstructive apnea; and (3) the extent to which SDB explained objective sleep characteristics, symptoms (fatigue, excessive daytime sleepiness, self-reported sleep quality, depression), and functional performance (self-report, 6m WT, daily mobility) in these patients.

METHODS

The study employed a cross-sectional design. Human subjects approval was obtained, and all participants provided informed consent.

Sample

The sample included patients with stable chronic HF recruited from 5 structured HF disease management programs in the Northeastern United States. Participants had stable HF (no hospital admissions within the previous month or titration of vasoactive medications within the past 2 weeks), were ≥ 18 years of age, and cognitively intact by clinical impression. Exclusion criteria included current pregnancy, unstable medical or psychiatric conditions, ongoing alcohol abuse, illicit drug use, history of Parkinson disease, obstructive valvular, hypertrophic, or surgically correctable valvular disease, renal failure, and previously identified sleep disorders. We also excluded participants who had hemiplegia affecting the non-dominant arm because of the potential for immobility to confound the wrist actigraph measurements.

Variables and Instruments

Functional Performance

Functional performance is the “day to day corporeal activities people do in the normal course of their lives to meet basic needs, fulfill usual roles, and maintain health and wellbeing.”²⁸ The 6m WT, daytime activity level (wrist actigraph; percent daily mobile time), and Medical Outcomes Study SF36 v2 (SF36) physical function (PF) component were used to evaluate objective and subjective attributes of functional performance, respectively.

The 6m WT²⁹ is an objective measure of the distance walked in 6 minutes under controlled conditions and is correlated with oxygen consumption during treadmill testing,³⁰ cycle ergometry, and self-reported functional status.²⁹ It was conducted using standard methods.³¹ The SF36v2 PF component^32,33 was used to elicit self-reported physical function. The SF36 has well-documented reliability and validity in healthy and chronically ill populations.^34–37 Construct, criterion-related, discriminant validity, and internal and re-test consistency have been supported in older adults.^37–39 A median internal consistency of 0.80 was reported over a large group of studies.³³ Internal consistency of the SF-36 sub-scales ranged from 0.81-0.92 in the current study. The PF component score was computed using published methods.⁴⁰

The Actiwatch-64 (Respironics Mini Mitter, Inc., Bend, OR) wrist actigraph was used as a measure of daytime mobility. Wrist actigraphs are reliably able to discriminate levels of activity associated with changes in speed and incline during treadmill testing and are able to distinguish across known physical activities of varying intensity.^41–43 Movement scores of HF patients were lower than those of age-matched controls and were correlated with self-reported activity (r = 0.72, P < 0.001). Movement scores and peak oxygen consumption during treadmill testing (r = 0.42, P < 0.01) were moderately correlated. Leg and wrist movement scores were highly correlated (r = 0.81, P < 0.01).⁴⁴

Daytime actigraph data were computed from the interval from self-reported morning lights on time to lights out time. For each of the 3 daily intervals, we computed the percent mobile time (percentage of daytime intervals during which there was one or more mobile counts/minute, with the Actiware Sleep v5 Program (Respironics Mini Mitter, Inc., Bend, OR).

Symptoms

We evaluated 4 common symptoms of HF and sleep disorders: excessive daytime sleepiness, fatigue, sleep disturbance, and depressive symptoms. The Epworth Sleepiness Scale (ESS), a self-report measure of propensity for sleepiness during activities occurring in every day life,^45–47 was used to measure excessive daytime sleepiness. The ESS has well-documented reliability in a variety of populations. Coefficient α was 0.77 on data obtained in this study. Consistent with the methods used in the Sleep Heart Health Study,⁴⁸ a score ≥ 11 was used to indicate excessive daytime sleepiness.

The Multi-Dimensional Assessment of Fatigue Scale (MAF)^49,50 was used to measure fatigue. It contains 16 items in a numeric rating scale format and measures 4 dimensions of fatigue: severity, distress, degree of interference in activities of daily living, and timing. The MAF was highly correlated with the fatigue subscale of the Profile of Mood States in HF patients (r = 0.81).⁵¹ Coefficient α was 0.93 in data obtained in the current study.

The Center for Epidemiological Studies Depression Scale (CESD)^52,53 was used to measure depressive symptoms. The CESD has high internal consistency (0.87), and adequate test-retest reliability⁵⁴ and sensitivity and specificity.⁵⁵ The CESD had an internal consistency of 0.84 in the current study. The total scale score and the dichotomized scale score (CESD ≥ 16), indicating likelihood of clinically relevant depression, were used in the analyses.

The Pittsburgh Sleep Quality Index (PSQI)⁵⁶ was used to obtain participants' perception of habitual sleep quality. Using a global PSQI score ≥ 5 as a measure of poor sleep, the instrument had a diagnostic sensitivity of 89.6% and specificity of 86.5%. Validity was also acceptable in comparison with polysomnography. Internal consistency was 0.79 in the current study.

Polysomnography

Unattended nocturnal polysomnography (PSG) was conducted for one night in participants' homes with the Safiro (Compumedics, Inc., Charlotte, NC), a battery-operated, miniaturized sleep recorder. We used 2 channels of electroencephalogram (C3/A2 and C4/A1), right and left electro-oculograms, and bipolar submental electromyograms. We measured respiratory effort, nasal flow with nasal cannula connected to a pressure transducer, and oxygen saturation; single bipolar electrocardiogram; heart rate; and body position. Bilateral piezo-electric sensors were used to evaluate leg movements. Studies were saved to a compact flash disk.

PSG studies were downloaded from the flashcard retrieved from the collection device and scored manually on a high-resolution monitor, using 30-sec epochs for sleep stages and 3-min epochs for respiratory and leg movement data. Sleep and respiratory/leg data were scored in separate passes through the data. Sleep stages were scored using Rechtschaffen and Kales criteria.⁵⁷ EEG arousals were defined according to standard criteria.^58,59 Respiratory events were scored from the nasal pressure signal, thermistor, and rib/abdomen channels. Respiratory events tabulated were apneas (flow < 10% of baseline for > 10 sec on both nasal and oral signals), hypopneas (flow < 70% of baseline on the nasal cannula signal for > 10 sec, associated with desaturation ≥ 4% within 30 sec),⁶⁰ and respiratory event related arousals (RERAs: flow < 70% of baseline for > 10 sec associated with an arousal but no desaturation ≥ 4%). Each apnea was characterized as either obstructive or central, based on the persistence of movement on the effort channels. In contrast, no attempt was made to characterize hypopneas as central or obstructive, as recommended by the recently published AASM scoring criteria.⁶⁰ Although it has been well established that during a hypopnea, the presence of a plateau on the inspiratory airflow waveform is virtually always a marker of obstruction, it is not clear whether the absence of this inspiratory flattening (“flow limitation”) indicates that an event is central or low resistance. In fact, preliminary work by the last author⁶¹ has shown that as often as 50% of the time, non–flow-limited breaths are associated with high esophageal pressures and have high resistance. For this reason, no attempt was made to classify the hypopneas in the present study as either obstructive or central. Hypopneas were used solely to calculate the apnea hypopnea index (AHI), which was calculated as the sum of all apneas and all hypopneas divided by total sleep time (TST). Similarly, RDI was calculated as the sum of apneas, hypopneas, and RERAs divided by TST.

PROCEDURES

Participants were recruited during a routine visit to the HF program. A research assistant explained the study, obtained informed consent, reviewed medical records, and performed the 6m WT. Participants completed a packet of questionnaires including the MOS SF36 V2, ESS, CESD, PSQI, and MAF, and wore the wrist actigraph for 3 days. A sleep technician visited participants' homes in the early evening hours prior to their anticipated bed times, attached the electrodes and sensors, programmed and turned on the sleep recorder, and explained the procedure for wearing and removal of the device and sensors. Lights out time was recorded in a sleep diary, and the sleep study was recorded. Participants removed the sensors upon awakening in the morning. A member of the study team returned in the morning to retrieve the sleep recorder. Each participant received $50 upon completion of data collection.

Data Analysis

Data were double-entered into an SPSS data base, corrected for errors, and examined for the extent to which they met assumptions for parametric analysis. Severity of SDB was initially categorized with the apnea hypopnea index (AHI) as none (0 to < 5); mild (5 to < 15); moderate (15 to < 30); and severe (30+) for descriptive purposes. However, all bivariate and multivariate analyses were conducted using quartiles of AHI as indicators of severity. The extent to which patients were characterized as having predominantly central vs. obstructive apnea was calculated as follows: For each subject who had an apnea index (AI) ≥ 5, we calculated the percentage of apneas that were scored as central [central apneas/ (central + obstructive apneas)] × 100. Predominant central sleep apnea (CSA) was defined as ≥ 50% central apneas. Predominant obstructive sleep apnea (OSA) was defined as < 50% central apneas only in those individuals with AI ' 5. Individuals who had apnea indices < 5 and AHI ≥ 5 were classified as “indeterminant” because of the difficulties associated with determining whether hypopneas are obstructive or central.⁶¹ Descriptive statistics, cross-tabulations, analysis of variance with post hoc Bonferroni comparisons, and linear and logistic regression were performed to address the study aims. For the logistic analyses of the relationship between severity of SDB and functional performance, AHI quartile (quartiles 1-3 as referent) was the independent variable. For the analyses of the effects of predominant CSA vs. OSA, data from those with “indeterminant” SDB were excluded from the analyses. Physical function, 6m WT, and percent mobile time were dichotomized as: lowest quartiles vs. 2nd through 4th quartiles (referent). Daytime sleepiness (ESS ≥ 11), depression score (CESD ≥ 16), and sleep quality (PSQI ≥ 5) were dichotomized using standard methods. Multivariate analyses were statistically controlled for age, gender, comorbidity, and body mass index.

RESULTS

Sample

Potential study participants were referred through the cardiologists and nurse practitioners in the respective HF programs. Of the 324 patients who were approached to participate, 41 were ineligible for the study upon further screening, and 233 consented. Among these, a total of 170 patients provided usable PSG and other data. Reasons for incomplete data included death (n = 1), rehospitalization or deteriorating health (n = 3), intolerance of PSG monitoring equipment due to anxiety, dermatological problems or nosebleeds (n = 5), technical problems with equipment or sensors (n = 8), lost to follow-up (n = 11), and unwilling to continue in the study for unknown reasons (n = 35). The resulting sample consisted of 170 patients. Demographic and clinical characteristics are presented in Table 1. Approximately one-third were women. A total of 35.7% were Black (n = 50), Asian-Pacific Islander (n = 7), or reported more than one race (n = 3). Five percent reported Hispanic ethnicity. Ninety five (56%) had NY Heart Association Class (NYHA) II HF. The majority of participants were on prescribed diuretics, β-blockers, and ACE inhibitors or angiotensin receptor blockers (Table 2).

Table 1.

Comparison of AHI quartiles on clinical, demographic, sleep, symptom, and functional variables (N = 170)

	Overall	AHI QI (0–7.05)	AHI QII (7.06–15.70)	AHI QIII (15.71–31.28)	AHI QIV (31.29+)
Clinical & Demographic
Age***#^	60.3 (16.8)	54.2 (14.4)	62.8 (16.5)	57.4 (17.2)	67.1 (13.3)
Gender (female)*	60 (35.5)	22 (51.2)	15 (35.7)	16 (37.2)	7 (16.7)
Race (white)*	109 (64.1)	21 (50.0)	26 (59.5)	30 (69.8)	32 (78.0)
Body mass index*	30.7 (8.0)	29.1 (6.9)	29.0 (6.7)	33.0 (9.7)	31.9 (8.1)
Comorbidity*	2.5 (1.52)	1.9 (1.3)	2.7 (1.8)	2.3 (1.3)	2.9 (1.6)
Ischemic heart disease	104 (60.8)	23 (53.5)	25 (61.4)	24 (55.8)	32 (74.4)
Myocardial infarction	68 (40.2)	13 (30.2)	5 (36.6)	19 (44.2)	21 (50.0)
Hypertension*	100 (59.2)	22 (51.2)	20 (47.6)	27 (64.3)	31 (73.8)
Diabetes	49 (29.0)	8 (18.6)	11 (26.2)	12 (28.6)	18 (42.9)
LVEF < 45	128 (75.4)	35 (81.5)	28 (66.7)	34 (79.1)	31 (73.8)
NYHA Functional Class	2.5 (0.67)	2.3 (0.7)	2.6 (0.7)	2.3 (0.6)	2.5 (0.7)
Beta blockers**	101 (59.4)	33 (76.7)	20 (47.60	29 (67.4)	19 (45.2)
Diuretics	144 (85.2	33 (82.5)	41 (93.2)	38 (90.5)	32 (74.4)
ARBS	52 (31.0)	16 (39.0)	12 (29.3)	11(25.0)	13 (30.2)
ACE inhibitors	88 (52.1)	20 (48.8)	18 (43.9)	27 (61.4)	23 (53.5)
Sleep Variables
Time in bed (min)	422.6 (98.0)	436.0 (100.0)	413.7 (95.1)	415.9 (110.4)	424.8 (86.5)
Total sleep time (min)	322.8 (96.5)	352.5 (99.8)	304.4 (84.4)	326.6 (95.4)	306.8 (100.8)
Sleep latency (min)	30.6 (35.4)	30.7 (34.9)	30.6 (29.20	32.4 (37.0)	28.6 (40.8)
REM latency (min)	112.6 (80.5)	122.4 (83.6)	113.6 (80.2)	104.1 (88.4)	110.9 (70.2)
Sleep efficiency (%)*#	70.1 (16.4)	76.5 (13.8)	68.6 (15.6)	70.9 (16.4)	67.2 (18.4)
Wake after sleep onset %**#	23.8 (15.6)	18.8 (14.1)	26.5 (15.2)	21.1 (12.1)	28.9 (18.6)
Stage 1 %***#^@	20.2 (8.3)	15.2 (5.7)	18.5 (6.0)	19.8 (6.6)	27.1 (9.4)
Stage 2 % ***#$	39.5 (12.1)	45.2 (9.8)	40.3 (10.7)	39.5 (9.4)	32.9 (14.7)
Stage 3-4 %**#^	5.4 (6.2)	7.2 (6.7)	4.1 (5.8))	7.3 (6.6)	2.9 (4.2)
Stage REM%***#^	11.2 (6.1)	13.5 (6.9)	10.6 (5.4)	12.3 (5.6)	8.2 (4.8)
Arousal index***@#$	21.6 (11.1)	14.6 (4.9)	19.8 (8.6)	20.6 (7.8)	31.8 (13.7)
AHI***@#$^	21.8 (19.3)	4.0 (2.0)	11.1 (2.4)	21.4 (4.7)	51.1 (12.9)
Obstructive apnea index***#$^	5.3 (10.6)	0.2(0.2)	1.4 (1.5)	3.2 (3.4)	16.6 (16.4)
Central apnea index***#$^	3.7 (8.4)	0.3 (0,5)	0.6 (0.9)	1.4 (2.4 )	12.7 (15.6)
Hypopnea index***#$^@	12.7 (9.6)	3.6 (1.8)	9.1 (2.5)	16.5 (4.5)	21.9 (12.2)
% Time at O2 sat < 90%**#	11.8 (18.7)	4.9 (13.4)	9.5 (19.2)	13.3 (19.4)	19.5 (19.7)
Symptom Variables
PSQI Global Score	8.7 (4.2)	9.6 (4.6)	8.5 (4.1)	8.1 (4.0)	8.6 (3.9)
PSQI > 5	125 (73.1)	32 (78.0)	33 (75.0)	30 (69.8)	30 (69.8)
Depressive symptoms (CESD)	17.0 (11.0)	19.5 (11.5)	18.2 (11.2)	14.8 (10.0)	15.5 (11.4)
CESD > 16	79 (45.7)	22 (53.7)	24 (54.5)	15 (34.9)	17 (39.5)
Global Fatigue Index	29.8 (14.2)	30.2 (16.4)	30.4 (13.7)	29.4 (14.1)	28.8 (14.2)
Epworth Sleepiness Scale	8.3 (4.3)	7.9 (4.6)	7.2 (4.2)	9.3 (4.3)	8.8 (4.1)
ESS > 11	48 (28.1)	10 (24.4)	10 (22.7)	15 (34.9)	13 (30.2)
Functional Variables
6-Minute Walk Test (feet)	979.8 (436.7)	1028.5 (409.8)	913.8 (489.1)	1045.7 (453.3)	918.2 (384.2)
Physical Function (SF36)	26.4 (1.5)	26.7 (1.7)	26.4 (1.5)	26.5 (1.6)	26.1 (1.5)
% Mobile time (actigraph)*#	81.3 (12.3)	86.0 (9.0)	81.6 (12.2)	80.5 (11.6)	77.1 (14.8)

All values mean (SD)/ N (%). Overall tests:

^*P < 0.05;

^**P < 0.01;

^***P < 0.001.

Post hoc tests (Bonferroni) P < 0.05:

^@Q1-Q3;

^#Q1-Q4;

^$Q2-Q4;

^{^}Q3-Q4

Table 2.

Prescribed Medications (N = 170)

Medication	N (%)
ACE inhibitors	88 (52.1)
Angiotensin receptor blockers	52 (31.0)
Beta-blockers	101(59.4)
Calcium channel blockers	20 (11.8)
Nitrates	35 (20.6)
Insulin	18 (10.6)
Oral antidiabetics	24 (14.1)
Digoxin	81 (47.6)
Diuretics	144 (85.2)
Antidepressants	25 (14.7)
Anxiolytics	16 (9.4)
Hypnotics	10 (5.9)

Age was associated with comorbidity (r = 0.32, P < 0.01), BMI (r = −0.43, P < 0.001), LVEF (r = 0.31, P < 0.001), and NYHA (0.21, P < 0.05). Women had higher BMI than men, mean = 32.96 (9.50) vs. 29.57 (6.92), P = 0.009, and were less likely than men to have a history of myocardial infarction (n = 17/ 28.3% vs. n = 52/ 46.4%), P = 0.02, and ischemic heart disease (n = 30/ 50% vs. n = 75/ 66.4%), but there were no detectable gender differences in hypertension, comorbidity, LVEF, NYHA, or the proportion of patients with preserved systolic function (LVEF ≥ 45%). Thirty-one (52.5%) of the women were minority group members, compared to 30 (26.8%) of the men (P = 0.001). Although women were slightly younger than men, mean = 57.33 (16.25) vs. mean = 61.95 (15.81) years, the difference was not statistically significant. There was no association between BMI and systolic dysfunction and no association between LVEF and BMI in those with systolic dysfunction.

Characteristics and Clinical and Demographic Correlates of SDB

Participants had none (0 to < 5) (n = 27/ 15.8%), mild (5 to < 15) (n = 57/ 33.3%), moderate (15 to < 30) (n = 41/ 24.5%), and severe (30+) (n = 46/ 26.3%) SDB, as indicated by the AHI. Descriptive statistics on the clinical, demographic, sleep, symptom, and functional performance variables for the overall sample and by AHI quartile are presented in Table 1. Severity of SDB, as indicated by quartiles, was associated with male gender, age, white race, comorbidity, body mass index, and history of hypertension. There was a non-significant trend (P = 0.09) for an association between diabetes and severity of SDB. Age, gender, and body mass index together explained 18.6% of the variance in the AHI in linear regression analysis (P < 0.001).

There was an association between severity of SDB and β-blocker use (P = 0.006), with the highest proportion in those in AHI quartile I and the lowest rate in quartile IV. However, there was not a consistent trend across levels of SDB. There was also a non-significant trend toward an association between severity of SDB and diuretic use (P = 0.062), but no associations between atrial fibrillation, use of angiotensin receptor blockers, ACE inhibitors, pacemakers, or implantable defibrillators and SDB severity. Comorbidity was not associated with AHI in the multivariate analysis. Left ventricular ejection fraction and NYHA were not associated with SDB severity in the overall sample or in the sub-group of participants with systolic dysfunction (LVEF < 45).

Descriptive statistics comparing the clinical, demographic, symptom, and functional characteristics of participants with no SDB, predominantly obstructive, predominantly central, and “indeterminate” apnea are in Table 3. Data from the “indeterminate” group were not included in the χ² or ANOVA procedures. Sixteen (9% of the total sample) had predominant CSA; 37 (21%) had predominant OSA; and 27 (16%) had no significant SDB (AI and AHI < 5). Those with CSA were, on average, 7 years older than those with OSA and 12 years older than those with no SDB. All but one of the patients with CSA were men. CSA was associated with lower BMI than OSA, and there was a non-significant trend toward an association between greater use of beta blockers by patients who had CSA (P = 0.056), but there were no associations between use of diuretics, ACE inhibitors, angiotensin receptor blockers, pacemakers, or implantable defibrillators and type of SDB.

Table 3.

Comparison of categories of SDB on clinical, demographic, sleep symptom, and functional performance variables (N = 170)

	No SDB	Obstructive Apnea	Central Apnea	Indeterminate SDB
	(AI < 5; AHI < 5)	(AI ≥ 5; < 50% central apneas)	(AI ≥ 5; ≥ 50% central apneas)	(AI < 5; AHI ≥ 5)
	N = 27	N = 37	N = 16	N = 90
Clinical & Demographic
Age**#	58.0 (16.1)	63.5 (14.7)	70.6 (12.0)	60.1 (16.8)
Gender (Female)**	13 (48.1)	9 (23.7)	1 (6.2)	37 (41.1)
Race (white)*	42 (53.8)	28 (77.8)	12 (75.0)	54 (60.7)
Body mass index*@#	29.1 (6.9)	33.0 (9.1)	27.7 (5.0)	30.8 (8.3)
Comorbidity	2.3 (1.6)	2.6 (1.6)	2.7 (1.0)	2.5 (1.5)
LVEF < 45	21 (77.8)	27 (71.1)	15 (93.8)	66 (73.7)
NYHA Functional Class	2.5 (0.7)	2.4 (0.6)	2.8 (0.8)	2.4 (0.6)
Ischemic heart disease	13 (48.1)	23 (62.2)	13 (81.2)	55 (61.1)
Myocardial infarction	6 (22.2)	16 (43.2)	9 (56.2)	38 (42.7)
Hypertension*	15 (55.6)	29 (78.4)	9 (56.2)	48 (53.3)
Diabetes*	6 (20.3)	14 (37.8)	5 (31.2)	24 (26.7)
Atrial fibrillation	1 (5.0)	4 (14.8)	1 (7.7)	3 (4.2)
Stroke/TIA	0	6 (15.8)	2 (12.5)	7 (7.8)
Implantable defibrillator	7 (25.9)	15 (40.5)	9 (56.2)	41 (46.1)
Pacemaker	9 (33.3)	8 (21.1)	8 (50.0)	39 (44.3)
Beta-blockers	7 (25.9)	14 (36.8)	10 (62.5)	23 (26.4)
Diuretics	21 (80.8)	30 (78.9)	12 (75.0)	80 (90.9)
ARBS	9 (33.3)	14 (36.8)	4 (25.0)	25 (28.7)
ACE-inhibitors	14 (51.9)	19 (50.0)	9 (56.2)	45 (51.7)
Sleep Variables
Time in bed (min)	435.2 (102.6)	432.6 (82.31)	437.1 (97.2)	411.9 (105.0)
Total sleep time (min)	349.5 (102.5)	333.5 (85.55)	306.8 (102.8)	314.5 (98.75)
Sleep latency	29.6 (30.0)	33.1 (85.9)	28.2 (30.4)	29.1 (33.1)
REM latency (min)	129.0 (101.4)	126.8 (87.9)	90.6 (51.4)	106.5 (74.1)
Sleep efficiency (%)	75.9 (15.2)	71.3 (13.7)	64.9 (14.0)	70.3 (17.8)
Wake after sleep onset%***	19.3 (15.5)	24.0 (14.0)	31.1 (13.8)	23.9 (16.2)
Stage 1%***@#$	13.6 (5.3)	22.3 (8.4)	28.4 (8.6)	18.4 (5.7)
Stage 2%***@#$	45.7 (10.3)	37.2 (12.9)	29.1 (11.0)	40.3 (11.1)
Stage 3-4 %*#	7.8 (7.4)	3.3 (4.1)	1.8 (2.0)	6.1 (6.6)
Stage REM %@#	13.5 (7.7)	9.4 (4.9)	9.5 (4.7)	11.2 (5.7)
Arousal index***	13.9 (4.4)	29.9 (12.2)	26.8 (12.9)	19.7 (9.2)
AHI***@$	2.8 (1.4)	41.6 (18.5)	44.5 (17.7)	15.5 (10.3)
Obstructive apnea index	0.1 (0.2)	17.5 (16.1)	6.3 (6.2)	1.7 (4.2)
Central apnea index	0.2 (0.3)	4.2 (6.4)	26.5 (15.7)	0.5 (0.7)
Hypopnea index	2.5 (1.3)	19.9 (10.7)	11.0 (15.7)	13.3 (8.4)
% Time at O2 sat ≤ 90%**@	7.2 (17.1)	16.8 (20.8)	13.2 (11.4)	12.7 (21.1)
Symptoms
PSQI Global Score	10.1 (5.1)	8.2 (3.3)	8.1 (3.5)	8.5 (4.3)
PSQI ≥ 5	60 (75.9)	55 (72.4)	10 (62.5)	65 (72.2)
Depressive symptoms*	19.6 (12.6)	14.9 (11.8)	12.6 (8.5)	17.9 (10.5)
CESD ≥ 16	44 (55.7)	28 (36.8)	6 (37.5)	47 (52.2)
Global Fatigue Index	30.7 (17.2)	29.1 (13.8)	25.0 (8.8)	30.3 (14.8)
Epworth Sleepiness Scale	8.4 (5.4)	7.9 (4.5)	9.8 (2.6)	8.2 (4.2)
ESS ≥ 11	19 (24.1)	24 (31.6)	5 (31.3)	25 (27.6)
Functional Performance
Six-Minute Walk Test (feet)	1083.3 (418.8)	968.3 (372.1)	1046.9 (483.5)	944.3 (458.6)
Physical Function	26.9 (1.6)	26.3 (1.6)	26.2 (1.4)	26.4 (1.6)
% mobile time*#	86.4 (9.3)	80.1 (13.4)	75.3 (13.0)	81.3 (12.0)

Values are mean (SD)/N (%). Overall tests:

^*P < 0.05;

^**P < 0.01;

^***P < 0.001.

Post hoc tests (Bonferroni), P < 0.05:

^@No: OSA;

^#No: CSA;

^$CSA: OSA.

“indeterminant” group not included in Χ² or ANOVA tests

Participants with predominant OSA were more likely than those with CSA or without SDB to have hypertension and diabetes, but no more likely to have a history of stroke or transient ischemic attack. There was a non-statistically significant trend (P = 0.06) for group-related differences in history of myocardial infarction, with the highest rate in those with CSA.

Table 4 presents the distribution of the categories of SDB, demographic characteristics, and BMI by LVEF. There was no statistically significant association between groups categorized by LVEF on type of SDB. However, the largest difference between groups was in CSA, with a greater proportion in the group with systolic dysfunction having CSA (11.71%), compared to 2.38% of those with preserved systolic function (LVEF ≥ 45). Among men with systolic dysfunction (n = 83), 14 (16.7%) had CSA.

Table 4.

Distribution of categories of SDB based on left ventricular ejection fraction^*

	No SDB	Obstructive Apnea	Central Apnea	Indeterminate SDB
	(AI < 5; AHI < 5)	(AI ≥ 5; < 50% central apneas)	(AI ≥ 5; ≥ 50% central apneas)	(AI < 5: AHI ≥ 5
	N = 27	N = 37	N = 16	N = 90
LVEF < 45 (N = 128)	21 (16.40)	26 (20.30)	15 (11.71)	66 (51.56)
Female 45 (35.16)	11 (8.65)	8 (6.25)	1 (0.05)	25 (19.53)
Age 57.55 (15.51)	50.00 (12.71)	60.07 (14.85)	69.53 (11.61)	56.14 (15.99)
BMI 31.25 (8.28)	30.34 (6.78)	33.62 (8.78)	28.04 (4.98)	32.37 (8.95)
EF 26.23 (8.37)	28.19 (9.72)	29.52 (7.83)	26.00 (8.16)	24.58 (7.83)
LVEF ≥ 45 (N = 42)	6 (14.29)	11 (26.19)	1 (2.38)	24 (57.14)
Female 15 (35.7)	2 (4.76)	1 (2.38)	0	12 (28.57)
Age 69.07 (14.75)	56.50 (16.13)	70.27 (12.40)	87.00+	60.08 (16.81)
BMI 29.07 (7.02)	24.42 (4.51)	31.64 (9.35)	27.76+	29.32 (5.86)
EF 56.57 (10.28)	58.67 (2.31)	53.00 (7.45)	50.00+	58.05 (11.79)

All values mean (SD)/N (%).

^*Percentages were calculated using the sample size for each LVEF sub-group as the denominator.

⁺Standard deviation could not be calculated because N = 1 in this group.

SDB and Sleep Characteristics

Severity of SDB was positively associated with percentages of wake after sleep onset, and stage 1 sleep, arousal index, time at oxygen saturation less than 90%, and inversely related to sleep efficiency and percentages of stages 2, 3-4, and REM sleep (Table 1). Participants in AHI quartile IV had 46 minutes less total sleep time than those in AHI quartile I, but there was no statistically significant difference overall. Post hoc tests revealed that there few differences in sleep continuity and architecture between quartiles II and III. Figure 1 depicts the number of minutes of time in bed, sleep time, and sleep stages across quartiles of SDB.

Figure 1.

Comparison of quartiles of AHI on sleep stages in minutes

Participants with CSA had higher percentages of wake after sleep onset, stage 1 sleep, and less stage 2 and 3-4 sleep than those with no apnea or predominant OSA (Table 3). Both OSA and CSA were associated with EEG arousals and desaturation (oxygen saturation ≤ 90%; Table 2). Figure 2 depicts the number of minutes of time in bed, sleep time, and sleep stages for each group.

Figure 2.

Comparison of OSA, CSA, and no SDB on sleep stages in minutes

SDB, Functional Performance, and Symptoms

There was an inverse relationship between severity of SDB and daily mobility level (Table 1). However, the group-related differences in mobility were not statistically significant when age, gender, comorbidity, body mass index, and β-blocker and diuretic drugs were statistically controlled in the regression analyses. There was no linear relationship between severity of SDB and self-reported PF. However, logistic regression analysis revealed that severe AHI was associated with a 4-fold likelihood of poor PF when age, gender, BMI, β-blocker use, diuretic use, and comorbidity were statistically controlled (Table 5). The odds ratios for AHI in quartiles II or III were not statistically significant, and there was no relationship between severity of SDB and 6m WT.

Table 5.

Ratio of quartiles of AHI to SF-36 Physical Function Component, odds ratios with 95% confidence intervals (CI)

Quartiles of AHI
AHI < 7.05 referent	OR	CI
AHI Quartile II (7.06-15.70)	1.89	0.58–6.22
AHI Quartile III (15.71–31.28)	1.23	0.59–6.21
AHI Quartile IV (31.29+)	4.15	1.19–14.57

There was a statistically significant overall difference among the participants with central, obstructive, and no apnea in percent of daily mobile time, with the lowest level in participants with CSA (Table 3). CSA was associated with > 4-fold likelihood of having low mobility (1st quartile) (OR = 4.09, 95%CI = 1.23–13.62), compared with no SDB or OSA, when age, gender, comorbidity, use of β-blocker drugs, and BMI were statistically controlled in the analyses.

There were no statistically significant relationships between severity of SDB and symptoms, including self-reported sleep quality, fatigue, excessive daytime sleepiness, or depression in the linear or logistic analysis. Although there was an overall statistically significant difference in depressive symptoms between types of SDB, with participants with predominant CSA having less depression, this difference was not statistically significant when controlling for age and gender. Type of SDB was not associated with fatigue, sleep quality, or excessive daytime sleepiness in linear or logistic analyses.

DISCUSSION

The high rate of moderate to severe SDB in this study of a clinically and demographically diverse group of stable HF patients is similar to past reports.^1,5,6,8–13 Our study extended this line of research by including a larger proportion of women, minority group members, and patients with preserved systolic function than recent studies.^5,12 Therefore, our sample may be more representative of these groups. We also incorporated relevant clinical and demographic characteristics and cardiovascular medications in the multivariate analyses and controlled for these potential confounding influences on the relationships between SDB, symptoms, and functional performance. Similar to population-based studies, clinical and demographic correlates of severity of SDB were gender, age, body mass index, comorbidity, and hypertension. These findings, as well as the high levels of obesity in this sample and the absence of associations between severity of SDB (quartiles) and LVEF and New York Heart classification, suggest that the correlates and risk factors for obstructive sleep apnea are similar in HF patients to those in the general population.

Based on our categorization of central vs. obstructive sleep apnea, only 9% of the overall sample and 11.71% of those with systolic dysfunction had significant CSA. The 16.7% rate of CSA in male HF patients with systolic HF in our study is comparable to the 15% overall rate reported in a recent study⁵ of stable community residing HF patients with LVEF < 45%, of whom 77% were male and the mean age was similar to the age of the participants in our study.

The reasons for the low rates of CSA in our study are not known, but these findings may be explained by the strict criteria we used to define central vs. obstructive SDB, the clinically stable nature of the patient population recruited from structured HF disease-management programs, or the diversity of our sample on age, gender, race/ethnicity, and left ventricular ejection fraction. The high proportion of men and high rates of ischemic heart disease, myocardial infarction, systolic dysfunction, and lower rates of hypertension in the group with CSA suggest that CSA may be associated with a primarily ischemic vs. hypertensive etiology of HF. Moreover, the gender difference may be explained by the higher rate of history of myocardial infarction (46.4% vs. 28.3%) and ischemic heart disease (66.4% vs. 50%) in men than women.

The 4-fold increased likelihood of having poor physical function among those in the highest quartile of SDB is higher than the odds ratio obtained in the Sleep Heart Health Study²⁷ of over 6000 individuals recruited from cardiovascular cohorts and may suggest that HF patients are more vulnerable to the negative consequences of SDB. Like the SHHS investigators, we found little increased risk for poor physical function for those with mild or moderate SDB. Given the absence of a relationship between severity of SDB and 6-minute walk test performance, this finding suggests that the primary impact of SDB is on physical and role-related activities occurring during everyday life, such as those measured by the SF-36 physical function component, rather than on performance in the laboratory/ clinic setting, as measured by 6m WT, or on daily mobility levels.

The association of CSA with daily mobility, when adjusted for the potentially confounding effects of age and gender, extends previous work that did not control for these covariates,² but the absence of an association between CSA and 6m WT contrasts with the findings of another recent study.¹² Due to the fact that the actigraph is also sensitive to sleep, the lower levels of daytime activity may reflect sleepiness or napping, but detailed information on the timing of daytime naps was unavailable. This interpretation is consistent with past reports that CSA is associated with objectively recorded sleepiness^2,3 and may also reflect the increased wake time and poorer sleep architecture in these patients.

We found no differences in self-reported sleep quality, depressive symptoms, fatigue, self-reported excessive daytime sleepiness, or 6m WT distance between groups characterized by severity of SDB or categorized as CSA, OSA, or no-SDB, when adjusted for age and gender. These findings may reflect the non-specificity of these symptoms to SDB and their multifactorial etiology in HF patients. The lack of an association between SDB severity and self-reported sleep quality is particularly surprising, given the strong associations between SDB and sleep architecture. However, these results are consistent with past studies of more racially and ethnically homogeneous groups of ambulatory HF patients.^5,6,7

Our findings suggest that daytime symptoms or self-reported sleep disturbance may not be good indicators of the presence of SDB and should not be used to determine the need for referral for SDB screening in HF patients. Clinical and demographic characteristics, including more advanced age, male gender, comorbidity, and body mass index may be more predictive of the presence of SDB in HF patients.

A strength of this study was inclusion of a large and diverse group of ambulatory HF patients recruited from structured HF disease management programs. The use of multidimensional measures of sleep and functional performance enabled us to evaluate behavioral, perceptual, and physiological characteristics of sleep. This study extended past work by addressing the role of clinical and demographic covariates of sleep, symptoms, and functional performance.

The use of full PSG studies in the home environment provided ecological validity, as PSG is more likely to reflect sleep in normal environments rather than the laboratory setting. We obtained only one night of PSG on each participant because one night has been shown to be adequate to screen for SDB in stable HF patients⁶² and reduces subject burden and costs associated with measurement. However, the use of PSG and associated discomfort and lack of familiarity with the equipment may have had a negative impact on sleep architecture.

Our method of categorizing CSA vs. OSA was designed to clearly delineate participants with these conditions. We acknowledge that this method was somewhat arbitrary and may have underestimated the prevalence of CSA that might be manifested in central hypopneas, but presently there is little agreement or published validation for the separation of hypopneas into central and obstructive sub-types without the use of invasive technology.⁶⁰ HF patients manifest both central and obstructive apneas, and seldom do OSA and CSA occur in isolation. Although our overall sample was large, categorization of OSA, CSA, and no-SDB resulted in small groups of OSA, CSA and no-SDB that limited statistical power to determine potential group-related differences in demographic, clinical, symptom, or functional performance variables.

The focus of this study was on patients who had clinical heart failure, regardless of the extent of systolic or diastolic dysfunction and the extent to which SDB contributed to important daytime symptoms and function. We do not expect our findings to contribute to understanding of the pathophysiology of SDB and HF. Rather, we used LVEF levels obtained within the past six months for descriptive purposes. Given the dynamic nature of HF pathophysiology, it is possible that the actual LVEF at the time of the sleep study varied from the obtained measures. Reported LVEF levels may reflect underlying preserved systolic function or improvements in systolic function that may have occurred with HF treatment in this group of patients who were selected due to their “stable” status. Therefore caution must be observed in interpreting the nature of systolic function in this sample.

The cross-sectional nature of the study precludes inferences about causality. However, the temporal relationships between CSA and OSA and the development of HF and its symptom and functional consequences likely differ. Through its contributions to hypertension, OSA is a pathway to the development of HF and its negative functional consequences. Our findings suggest that only at severe levels does SDB have an impact on physical function. In contrast, CSA is a consequence of the pathophysiology of HF and reflects exacerbation of this condition. In the latter case, functional impairments may be comorbid with and not a consequence of CSA.

Patients with HF take many prescribed medications that cross the blood-brain barrier and are likely to have an impact on sleep and SDB, as well as symptoms, functional performance, and cardiovascular function. There was more use of β-blocker drugs in participants in the lowest quartile than those in the highest quartile, but no consistent patterns across quartiles II and III. There was also an apparent, albeit not statistically significant, trend toward greater use of β-blockers in those with CSA. These findings contrast with those in a recent study that suggested that there were no differences in rates of SDB or CSA with greater use of β-blockers.¹⁴ The extent to which associations found in our study are consequences of unmeasured differences in cardiovascular pathophysiology between the SDB groups or whether β-blocker drugs have effects on SDB are not known.

It is possible that improvements in cardiac function with positive airway pressure (PAP) may improve CSA, symptoms, and functional performance. PAP may also directly improve symptoms and functional performance in CSA and OSA directly through improvements in sleep. However, studies of the effects of PAP on SDB have had inconsistent results. For example, one group found improvements in 6-minute walk,²¹ but the impact on quality of life measures has been inconsistent.^63,64

CONCLUSIONS

SDB is common in a diverse sample of stable community-residing HF patients and associated with clinical and demographic characteristics similar to those in the general adult population. Clinical and demographic characteristics, such as male gender, aging, obesity, and hypertension, rather than self-reported symptoms of sleepiness, poor sleep, fatigue, or depression, may be indicators of the need for evaluation and treatment of SDB. Severe SDB is associated with poor self-reported, but not objective functional performance, despite poorer objective sleep quality in patients with SDB.

DISCLOSURE STATEMENT

This was not an industry supported study. Dr. Redeker is Associate Editor for the journal Heart & Lung. Dr. Rapoport has received research support from Fisher Paykel Healthcare, Meditronics, Inc., Ventis Medical, Advanced Brain Monitoring, and SleepEx. Dr. Rapoport holds multiple US and foreign patents covering techniques and analysis algorithms for the diagnosis of OSAHS and techniques for administering CPAP. Several of these have been licensed to Biologics, Fisher Paykel Healthcare, Advanced Brain Monitoring, and Tyco. The other authors have indicated no financial conflicts of interest.