Obstructive sleep apnea risk and psychological health among non-Hispanic blacks in the Metabolic Syndrome Outcome (MetSO) cohort study

Abstract

Introduction This study assessed associations of depression and anxiety with risk of obstructive sleep apnea (OSA) among non-Hispanic blacks in the Metabolic Syndrome Outcome (MetSO) study.

Method A total of 1,035 patients participated. ARES^TM score ≥6 defined high OSA risk. Moderate depression was defined by a CES-D score ≥16. Moderate anxiety was measured by a BAI score ≥16.

Results The mean age was 62 ± 14 years; 70% were female. A total of 93% were diagnosed with hypertension; 61%, diabetes; and 72%, dyslipidemia; 90% were overweight/obese; 33% had a history of heart disease; and 10% had a stroke. Logistic regression analysis, adjusting for age and gender, showed that patients with depression had nearly two-fold increased odds of being at risk for OSA (OR 1.75, 95% CI 1.02–2.98, p < 0.05). Patients with anxiety had three-fold increased odds of being at risk for OSA (OR 3.30, 95% CI 2.11–5.15, p < 0.01). After adjusting for marital status and income, patients with anxiety had a 6% increase in OSA risk (OR 1.06, 95% CI 1.04–1.09, p < 0.05), but depression was no longer significant.

Conclusion Our results suggest that non-Hispanic blacks with metabolic syndrome who experience anxiety and/or depression should be screened for OSA.

Key messages

• This study assessed associations of moderate to severe depression and anxiety with risk of obstructive sleep apnea (OSA) among non-Hispanic blacks with metabolic syndrome.

• Patients with depression had nearly two-fold increased odds of being at risk for OSA.

• Patients with anxiety had three-fold increased odds of being at risk for OSA.

Keywords:

• Metabolic syndrome
• non-Hispanic blacks
• obstructive sleep apnea
• psychological health

Introduction

Obstructive sleep apnea (OSA) is a highly prevalent sleep-related breathing disorder caused by repeated episodes of airflow cessation (apneas) leading to arterial hypoxemia and sleep fragmentation. OSA is characterized by intermittent hypoxia, which can lead to oxidative stress (1), systemic inflammation (2), vascular endothelial dysfunction (3), and an increase in sympathetic nervous system (SNS) activity (4), thereby putting patients with OSA at risk for cardiovascular and cardiometabolic diseases (5). The physiological effects of OSA are severe, including hypoxia, hypercapnia, increased left ventricular afterload, and acute arterial hypertension (5). OSA is also associated with abnormalities in glucose metabolism as shown in one recent cross-sectional study of obese adults with short sleep duration. More severe OSA was associated with higher glucose concentration, higher fasting insulin levels, and higher plasma Adrenocorticotropic hormone (ACTH) levels (6). Additionally, several lines of evidence suggest that OSA is an independent risk factor for cardiovascular morbidity and mortality (5). The consequences of OSA include excessive daytime sleepiness (EDS) (7), impaired executive function (8), decreased vigilance (8), and impaired health-related quality of life (HRQoL). The long-term sequelae of OSA include: obesity, cardiovascular disease, diabetes, and stroke (5), which are among the leading causes of death in the United States (9).

In addition to the well-known relationship between neurocognitive deficits and OSA (10), recent literature suggests a relationship between OSA and psychological health. For instance, recent studies have shown an association between suspected OSA and psychological disorders (11), and between OSA and anxiety (12). A small study of adults with suspected OSA found an association between habitual short sleep duration and depression (13). Earlier studies point to the fact that non-Hispanic blacks have higher OSA rates than their white counterparts in community- and population-based studies; but it is unknown if untreated psychological symptoms are the mechanism of this disparity (14,15). In a hospital clinic-based study in Brooklyn, NY, USA, among non-Hispanic blacks, it was found that 25% of patients who underwent polysomnography reported comorbid depressive symptoms and 24% of patients reported stress-related symptoms (5). Several studies reveal that non-Hispanic blacks who report depressive and/or anxiety symptoms are more likely to experience negative cardiovascular consequences such as aortic calcification (16), hypertension (17), and cardiovascular disease (CVD)-related mortality (18); but the mechanism of these associations is unclear and could be mediated by OSA.

Studies of the prevalence of depressive and anxiety symptoms among non-Hispanic blacks and non-Hispanic whites have reported mixed results. In 2007 Williams and colleagues found that 10.4% of African Americans, 12.9% of Caribbean blacks, and 17.9% of non-Hispanic whites had a major depressive disorder (MDD) diagnosis in their lifetime (19). Alternatively, based on the Behavioral Risk Factor Surveillance System (BRFSS) data from 2006 and 2008, the prevalence ‘any current depression’ including subclinical depression was contrastingly higher among non-Hispanic blacks (12.8%) than among their non-Hispanic white counterparts (7.9%) (20). The discrepancy in rates of depression may be due to the requirement for clinically significant impairment in the Diagnostic and Statistical Manual of Mental Disorders (DSM) IV-TR. For instance, in the 1999 National Health Interview Survey (NHIS), investigators found that although non-Hispanic blacks and non-Hispanic whites have a similar prevalence of depressive symptoms, non-Hispanic blacks are less likely to fulfill the criteria for a depressive disorder than their white counterparts (21). Non-Hispanic blacks may be less likely to seek clinical help or have less access to treatment. There is minimal information regarding the prevalence of anxiety symptoms in non-Hispanic blacks as well. Both non-Hispanic whites and non-Hispanic blacks do poorly at recognizing symptoms of anxiety (22), but again non-Hispanic blacks are less likely to seek help for anxiety symptoms as demonstrated by data from the National Ambulatory Medical Care Survey and the National Hospital Ambulatory Medical Care Survey (23). The inconsistencies in prevalence of mental health diagnoses may be largely due to help-seeking behavior and how that relates to the diagnostic criteria of depression and anxiety. Additionally this poses a challenge in assessing the actual prevalence of comorbid OSA and mental health, such as depressive and anxiety symptoms, among non-Hispanic blacks.

Numerous studies suggest an association between OSA and psychological symptoms (e.g. depression) (24–26). However, previous studies do not fully explain the overwhelming prevalence of OSA in psychiatric populations. While the prevalence of moderate to severe OSA (Apnea Hypopnea Index ≥15) in a community-based cohort is estimated to be 5% and 14% in women and men, respectively (26); previous studies have reported the prevalence of OSA in psychiatric patients to be anywhere from 10% to 59% (27–29). This variation in prevalence is likely due to the inclusion of specific psychiatric diagnoses, such as depression, as the prevalence appears to be highest in people with mood disorders (27,29).

One implication of this finding is that non-Hispanic blacks are at greater risk for OSA because of their comparatively higher prevalence of undiagnosed depressive symptoms, as suggested by Behavioral Risk Factor Surveillance System (BRFSS) and National Health Interview Survey (NHIS) data described above (20,21). The prevalence of moderate to severe OSA is 50%–60% in people with metabolic syndrome, which is diagnosed by at least three of the following: hypertension, diabetes, obesity, and dyslipidemia. Furthermore, recent literature suggests that OSA modulated cardiometabolic risk in people with obesity and metabolic syndrome (30). We hypothesize that the presence of depression and anxiety would have independent effects on the risk of OSA in a sample of non-Hispanic blacks with metabolic syndrome.

Methods

The present study utilized data from the Metabolic Syndrome Outcome (MetSO) study, a NIH-funded study of non-Hispanic blacks with metabolic syndrome (31,32). A total of 1035 patients provided data for the analysis. These included socio-demographic factors, health risks, and medical history. Physician-diagnosed conditions were obtained using an electronic medical record system (HealthBridge). Patients provided informed consent under the supervision of the IRB at SUNY Downstate Medical Center, the main study site, and at NYU Langone Medical Center.

Criteria for classifying metabolic syndrome

Patients were diagnosed with metabolic syndrome using the National Heart, Lung and Blood Institute and the American Heart Association guidelines (33). According to the joint interim statement, metabolic syndrome is diagnosed when a patient has at least three of the following conditions: hypertension, diabetes, obesity, and dyslipidemia (Table I).

Table I. Guidelines from the National Cholesterol Education Program (NCEP): NCEP ATP III definition of the metabolic syndrome.

Characteristic	NCEP ATP III
Hypertension	BP medication or BP >130/85 mm/Hg
Dyslipidemia	Plasma triglycerides >150 mg/dL or 1.7 mmol/L; HDL cholesterol <40 mg/dL or 1.03 mmol/L in men and <50 mg/dL or 1.29 mmol/L in women
Obesity	Waist circumference >40 inches or 94 cm in men and >35 inches or 80 cm in women
Diabetes	Fasting plasma glucose >110 mg/dL^a or 5.6 mmol/L
Requirements for diagnosis	Any three of the above disorders

a Fasting plasma glucose was recently updated to 100 mg/dL by the American Diabetes Association.

BP = blood pressure; HDL = high-density lipoprotein.

Criteria for assessing OSA risk

Patients were assessed for OSA risk using the Apnea Risk Evaluation System (ARES^TM); individuals with an ARES^TM score ≥6 were considered at risk for OSA (34). The ARES questionnaire has a sensitivity of 0.94, specificity of 0.79 (based on a clinical cut-off of Apnea Hypopnea Index (AHI) >5), positive predictive value of 0.91, and negative predictive value of 0.86 (35). The questionnaire includes age, weight, height, neck circumference and reports of comorbid illness such as high blood pressure, heart disease, diabetes, stroke, depression, lung disease, insomnia, as well as sleep medication, pain medication, and the Epworth Sleepiness scale. The ARES^TM was used to identify OSA risk because of its accuracy in evaluating populations with a large pretest OSA probability (34).

Criteria for assessing psychological health

The presence of depressive and anxious symptoms was assessed to ascertain the patients’ psychological health. Depression was assessed using the Center for Epidemiological Studies–Depression (CES-D) scale, which is a 20-item questionnaire that asks individuals to rate how often over the past week they experienced symptoms associated with depression, such as restless sleep, poor appetite, and feeling lonely. Responses range from 0 to 3 for each item (0 = rarely or none of the time; 1 = some or little of the time; 2 = moderately or much of the time; 3 = most or almost all the time). Scores range from 0 to 60, with high scores indicating greater depressive symptoms. A CES-D score greater than or equal to 16 was used to identify patients with clinically meaningful depression. The CES-D-20 has excellent internal consistency (Cronbach’s α = 0.88–0.91), excellent test–retest reliability (ICC = 0.87) (36). It has demonstrated adequate validity in measuring mental health (Pearson’s r = 0.75) and good sensitivity of 80.0%. Anxiety was measured with the Beck Anxiety Inventory (BAI), which is a 21-item questionnaire assessing anxiety symptoms such as ‘wobbliness in legs’, ‘scared’, and ‘fear of losing control’ (37). Accordingly, respondents are asked to rate how much each of these symptoms bothered them in the past week, on a scale ranging from 0 (‘not at all’) to 3 (‘severely, I could barely stand it’). Scores range from 0 to 63, with a score of 16 or higher indicating moderate to severe anxiety. The scale was validated in a sample of 160 psychiatric outpatients with various anxiety and depressive disorders, diagnosed with the Structured Clinical Interview for DSM-III (38). The BAI has a high internal consistency (Cronbach’s α = 0.92) and a test–retest reliability over 1 week of 0.75 (37). The CES-D and BAI are both scales which identify depressive and anxiety symptoms, but they are not synonymous with a psychiatric diagnosis.

Statistical analysis

Frequency and measures of central tendency were used to describe the sample. In preliminary analyses, Pearson and Spearman correlations were used to explore relationships between variables of interest. To determine the associations between psychological health measures and OSA risk among non-Hispanic blacks with metabolic syndrome, we utilized multivariate-adjusted logistic regression modeling. Covariates entered in the model were age, sex, and income. Before constructing the model, correlational analyses were performed to assess associations between hypothesized predictors (i.e. depression and anxiety) and the dependent variable (i.e. OSA risk). Only factors showing significant correlations (p < 0.05) with the dependent measure were entered in the final regression model; this helped to reduce redundancy in the model. Effects of all factors entered in the model were simultaneously adjusted. Data were coded and analyzed using SPSS 19.0.

Results

A total of 1,035 patients with metabolic syndrome provided data for this study. The mean age of the sample was 62 ± 14 years (range 20–97); 70% were female, and 43% reported an annual income lower than $10,000. Of the sample, 93% were diagnosed with hypertension; 61%, diabetes; 72%, dyslipidemia; 90% were overweight/obese; 33% had a history of heart disease; and 10% had had a stroke. Descriptive characteristics of cardiometabolic parameters are presented in Table II.

Table II. Metabolic characteristics of the study participants.

Variable	Mean	SD
Systolic BP (mmHg)	134.98	16.39
Diastolic BP (mmHg)	75.77	10.55
LDL cholesterol (mg/dL)	105.6	36.88
HDL cholesterol (mg/dL)	48.03	16.49
Triglycerides (mg/dL)	134.98	73.24
Glucose (mg/dL)	138.38	68.27
HbA1c (mmol/L)	7.93	1.63
BMI (kg/m^2)	33.8	8.56

BP = blood pressure; BMI = body mass index in pounds; HbA1c = glycated hemoglobin; HDL = high-density lipoprotein; LDL = low-density lipoprotein.

According to ARES^TM data, 48% of the patients were at high risk for OSA. Of those who were at high risk for OSA, 27% were characterized by depressed moods, based on CES-D scores, and 41% showed clinically meaningful anxiety symptoms based on BAI scores. Results show that 49.8% of women with a waist circumference greater than 35 inches (89 centimeters), an indicator of visceral adiposity, were at risk for OSA and 50.2% of women with the same waist circumference were not at risk for OSA (Table III).

Table III. Cross-tab with MetS indicators, psychological factors, and OSA risk.

Variables	OSA risk	No OSA risk	Fisher exact significance (p value)
Insulin/glucose^a	48.6%	51.4%	n.s.
Dyslipidemia^b	48.6%	51.4%	n.s.
Elevated BP/hypertension (>130/85)^c	49.2%	50.8%	n.s.
Waist circumference (visceral adiposity) male^d	42.5%	57.5%	n.s.
Waist circumference (visceral adiposity) female^d	49.8%	50.2%	0.004
Depression (CES-D)^e	55.1%	44.9%	n.s.
Anxiety (moderate–severe levels on BAI)^f	61.9%	38.1%	n.s.

a Fasting plasma glucose >110 mg/dL or 5.6 mmol/L.

b Plasma triglycerides >150 mg/dL or 1.7 mmol/L; HDL cholesterol <40 mg/dL or 1.03 mmol/L in men and < 50 mg/dL or 1.29 mmol/L in women.

c BP medication or BP >130/85 mm/Hg.

d Waist circumference >40 inches or 94 cm in men and >35 inches or 80 cm in women.

e Depressive symptoms on CES-D.

f Anxiety symptoms on Beck Anxiety Inventory (BAI).

Logistic regression analysis (Table IV), adjusting for effects of age and gender, showed that patients with moderate to severe depression had a nearly two-fold increased odds of being at risk for OSA (OR 1.75, 95% CI 1.02–2.98, p < 0.05). Likewise, patients with moderate to severe anxiety had a three-fold increased odds of being at risk for OSA (OR 3.30, 95% CI 2.11–5.15, p < 0.01). Also, age was positively associated with OSA risk (OR 0.97, 95% CI 0.95–0.98, p < 0.01). After adjusting for marital status and income (Table V), analysis showed that patients with anxiety had a 6% increase in OSA risk (OR 1.06, 95% CI 1.04–1.09, p < 0.05), but depression was no longer significant.

Table IV. Multivariate logistic regression analysis indicating odds ratios (OR) for psychological health associated with OSA risk in the MetSo; n = 1,035.

Variables	OR	95% CI		p
Gender (M or F)	0.93	0.55	1.57	0.77
Age (years)^a	0.97	0.95	0.98	<0.01
Depression (CES-D)^a	1.75	1.02	2.98	<0.05
Anxiety (BAI)^a	3.30	2.11	5.15	<0.01

a Significant, p < 0.05.

Table V. Multivariate logistic regression analysis indicating odds ratios (ORs) for OSA risk in the MetSo; n = 1,035.

Variables	OR	95% CI		p
Gender (M or F)	0.88	0.64	1.22	0.452
Age (years)^a	0.97	0.96	0.98	<0.001
Marital status^b	1.17	0.85	1.61	0.340
Income^c	1.30	0.96	1.75	0.093
Depression^d	0.98	0.95	1.00	0.084
Anxiety^a^,^e	1.06	1.04	1.09	<0.001

a Significant, p < 0.05.

b Reference: married.

c Reference: >$10,000 per annum.

d CES-D ≥16.

e Beck Anxiety Inventory ≥16.

Discussion

Our results suggest that non-Hispanic blacks with metabolic syndrome who experience psychological distress, as evidenced by depression and anxiety, are at an increased risk for OSA. Even after adjusting for marital status and income, moderate anxiety was still significantly associated with increased risk of OSA. The depression–OSA relationship may be attributed to the shared risk factors of diabetes, hypertension, and cardiovascular disease (39). The neurobiological risk factor of decreased serotoninergic neurotransmission has also been implicated in the etiology of both depression and OSA (40). Additionally, OSA and depression share common symptoms such as sleep disturbance, irritability, cognitive impairment, and concentration difficulties (39). As a result, OSA may frequently be misdiagnosed as depression. In this study marital status and income confounded the association between depression and OSA risk. This was unexpected as one study did not show an effect of poverty level on OSA risk (41) and another large population-based study found an association between OSA and depression even after adjusting for poverty level (42). Alternatively, several studies have noted an association between marital status, especially being separated/divorced/widowed, and OSA risk (43–45). The relationship between OSA and anxiety is less apparent. Magnetic resonance T2 relaxometry identified more severe and additional brain deficits in patients exhibiting both anxiety and OSA than in patients with OSA alone (46). Nonetheless, anxiety is known to occur more frequently in OSA patients than within a general population (12). The most likely explanation for the association between OSA risk and depressive and anxiety symptoms is that these psychological symptoms are secondary to the physiological changes which occur in OSA. For instance, increased sympathetic tone would lead to anxiety symptoms. An alternative interpretation of this association is that moderate to severe depression and anxiety may cause OSA, perhaps via metabolic syndrome. However, utilizing a population of patients in which all have metabolic syndrome would mean that depression and anxiety induce OSA by another mechanism like serotonin neurotransmission.

In the current study, associations were more striking for anxiety symptoms than depressive symptoms, as the presence of anxiety increased an individual’s odds of OSA risk three-fold, and depression two-fold, before adjusting for marital status and income. While our findings relate to OSA risk, this is consistent with the literature indicating that individuals with severe mental illness (27,47) and/or psychological distress, such as depression or anxiety, have a higher prevalence of OSA. As our study shows an association between moderate depression and anxiety and risk of OSA, current literature supports a bidirectional relationship (8,24,25,28,29,47). Additionally, older age represented an independent predictor of OSA risk. Thus, older non-Hispanic black adults with metabolic syndrome who presented with anxiety or depression are identified as a high-risk group for OSA. Consequently anxiety and/or depression in this high-risk group should raise clinical suspicion for OSA and prompt screening for classic OSA symptoms like snoring, witnessed apnea, or daytime sleepiness (48).

The strengths of this study include the use of a well-characterized cohort of non-Hispanic blacks with metabolic syndrome, who has historically been reluctant to participate in clinical research studies for a variety of reasons (49). Additionally, this community-dwelling cohort was not taking medications that could exacerbate metabolic syndrome, such as antipsychotics, although information on antidepressant use was unavailable. The limitations of the study include the fact that this was a cross-sectional study. This was a clinical sample of people with metabolic syndrome and may not be generalizable to community-dwelling people without metabolic syndrome. Also, the sample is very specific, being composed of non-Hispanic blacks. Therefore the findings may not be generalizable to people of other ethnicities. Another important limitation relates to the fact that the majority of participants in the study were women, limiting our ability to assess gender-based differences, although prior studies indicate that women are more likely to report depressive symptoms than men (50).

Given our preliminary findings regarding OSA risk, we identify several areas for future research. First, future studies should investigate the magnitude of the relationship between OSA and psychological health among men, relative to women. Second, well-characterized longitudinal cohort studies should be utilized to evaluate the incidence of OSA in people with depressive or anxiety symptoms. Third, although evidence supports a bidirectional relationship between psychological symptoms and OSA, future studies should investigate the causal relationship between psychological factors and OSA in longitudinal study models (24). There have already been several longitudinal studies which address the relationship between mental health and OSA. One prospective cohort study followed participants up to 1 year and found that those with OSA have twice the risk of depression (51). Another smaller prospective study found that the persistence of anxiety was linearly associated with the Apnea Hypopnea Index (52). Lastly, there are no studies to evaluate the effect of treatment of depressive and anxiety symptoms in order to decrease OSA risk. In terms of evidence-based clinical practice, the ARES^TM questionnaire has been shown to be efficient and sensitive for screening OSA risk in a dental setting (53), and therefore future work should investigate how the ARES^TM can be incorporated for ambulatory settings when screening patients with metabolic syndrome. The STOP-BANG questionnaire is another brief screening tool, which is part questionnaire and part demographic and physical measures, with over 90% sensitivity for moderate to severe OSA (54). Also the availability of home-based portable channel monitoring in lieu of in-lab polysomnography has already been shown to be a practical and slightly less expensive modality to diagnosis of OSA (55).

Conclusion

The findings of the present study support previous studies suggesting that individuals with comorbid metabolic syndrome and moderate anxiety symptoms and possibly depressive symptoms should be screened for OSA. The link between psychological health and OSA may have serious medical implications and health risks, which include cardiac ischemia, cardiac arrhythmia, cerebrovascular disease, and stroke. This suggests that appropriate management of these medical factors might necessitate diagnosis and treatment of OSA, a common and manageable condition among non-Hispanic blacks.

Declaration of interest

The authors report no conflicts of interest.

This research was supported by funding from the National Institutes of Health: R25HL105444, R01HL095799, RO1MD004113, and U54NS081765. The funding source had no role in the design, conduct, or analysis of the study, or in the decision to submit the manuscript for publication.