Characteristics of patients with positional OSA according to ethnicity and the identification of a novel phenotype—lateral positional patients: a Multi-Ethnic Study of Atherosclerosis (MESA) study

Abstract

Study Objectives:

We investigated the characteristics of obstructive sleep apnea (OSA) positional patients’ (PP) phenotypes among different ethnic groups in the Multi-Ethnic Study of Atherosclerosis (MESA) dataset. Moreover, we hypothesized the existence of a new OSA PP phenotype we coined “Lateral PP,” for whom the lateral apnea-hypopnea index is at least double the supine apnea-hypopnea index.

Methods:

From 2,273 adults with sleep information, we analyzed data of 1,323 participants who slept more than 4 hours and had at least 30 minutes of sleep in both the supine and the nonsupine positions. Demographics and clinical information were compared for the different PP and ethnic groups.

Results:

861 (65.1%) patients had OSA, and 35 (4.1%) were Lateral PP. Lateral PP patients were mainly females (62.9%), obese (median body mass index: 31.4 kg/m²), had mild–moderate OSA (94.3%), and mostly were non–Chinese American (97.1%). Among all patients with OSA, 550 (63.9%) were Supine PP and 17.7% were supine-isolated OSA. Supine PP and Lateral PP were present in 73.1% and 1.0% of Chinese Americans, 61.0% and 3.4% of Hispanics, 68.3% and 4.7% of White/Caucasian, and 56.2% and 5.2% of Black/African-American patients with OSA.

Conclusions:

Chinese Americans have the highest prevalence of Supine PP, whereas Black/African-American patients lean toward less Supine PP and higher Lateral PP. Lateral PP appears to be a novel OSA phenotype. However, Lateral PP was observed in a small group of patients with OSA and thus its existence should be further validated.

Citation:

Ben Sason Y, Oksenberg A, Sobel JA, Behar JA. Characteristics of patients with positional OSA according to ethnicity and the identification of a novel phenotype—lateral positional patients: a Multi-Ethnic Study of Atherosclerosis (MESA) study. J Clin Sleep Med. 2023;19(3):529–538.

BRIEF SUMMARY

Current Knowledge/Study Rationale: We studied the prevalence of obstructive sleep apnea (OSA) positional phenotypes among Black/African-American, White/Caucasian, Hispanic, and Chinese-American patients and we describe their demographic and polysomnographic characteristics.

Study Impact: In addition, we identify a novel OSA PP phenotype that we named “Lateral Positional Patients” (Lateral PP). These patients with OSA showed apnea and hypopnea events mainly in the lateral position. This was a small group of patients with OSA (4.1%) who were mainly obese, female, with mild–moderate OSA, and Black/African American. Despite similar levels of OSA severity in the 4 ethnic groups, Chinese Americans had a higher prevalence of a supine positional patients’ (PP) phenotype, whereas Black/African-American patients depicted a significantly lower prevalence of the supine PP phenotype compared with other ethnic groups.

INTRODUCTION

Obstructive sleep apnea (OSA) is characterized by periods of breathing cessation (apnea) and of reduced breathing effort (hypopnea) during sleep due to the complete or partial collapse of the upper airway. OSA is estimated to affect 9–38% of the adult population, and even more among elderly people.¹ Today, the medical definition of OSA is actively being challenged because of the different characteristics of individuals with OSA. A good example of different OSA phenotypes is related to the position of the patient—for instance, by comparing OSA positional patients’ (PP) characteristics vs non-PP (NPP) characteristics. PP have sleep-related breathing abnormalities, mainly when they sleep in the supine posture. These are patients with OSA who can benefit from positional therapy (PT)—that is, the avoidance of the supine posture during sleep. Contrarily, PT for NPP could only partially help this OSA group.² Thus, different phenotypes of OSA should be diagnosed and managed in a different manner. It is therefore critical to define these different phenotypes in order to refine the diagnosis and personalize the treatment of OSA. OSA is divided into 3 severity levels according to the apnea-hypopnea index (AHI), which corresponds to the average number of apnea or hypopnea events per hour: mild OSA with an AHI greater than 5 and lower than 15 events/h, moderate OSA with an AHI in the range of 15 to 30 events/h, and severe OSA with an AHI above 30 events/h.³

Many researchers have shown the presence of several OSA phenotypes in which symptom appearance and severity are highly related to sleep position.^4–9 The pioneer study of Cartwright⁹ in 1984 showed the worsening effect of the supine posture on the AHI in 30 patients with OSA. Oksenberg et al⁸ in 1997 described the demographic and polysomnographic characteristics of PP vs NPP and showed that PP are a dominant group of patients with OSA, representing 55.9% of all patients with OSA evaluated in a sleep unit. This study also showed that PP more frequently have mild–moderate OSA (ranging from 65% to 69%) and that PP are typically thinner and younger, have less severe breathing abnormalities, and experience less daytime sleepiness than NPP. Later, it was found that Asian patients with OSA have a much higher prevalence of a PP phenotype, reaching 75% of all patients with OSA and up to 87% in patients with mild OSA.¹⁰ Thus, PT could represent a simple, low-cost, and effective therapy to minimize the health burden of many patients with mild–moderate OSA.¹¹ It is of interest that PP variation across different ethnic groups has not been investigated. Therefore, we analyzed the prevalence and characteristics of PP according to ethnicity in a relatively large group of patients with OSA. In addition, we hypothesized the existence of a group of patients with OSA for whom the lateral AHI can be at least double the supine AHI and investigated the prevalence and characteristics of this novel PP phenotype that we coined “Lateral PP”.

METHODS

Multi-Ethnic Study of Atherosclerosis

The Multi-Ethnic Study of Atherosclerosis (MESA) database is a polysomnography (PSG) database of 2,273 adults (age ≥ 54 years), collected between November 2010 and 2012 by the National Heart, Lung, and Blood Institute as a part of the MESA.¹² The data consisted of full PSG recordings collected at home (ie, type II study) and included a variety of ethnicities—namely, Black or African-American, White or Caucasian, Hispanic, and Chinese-American men and women. The MESA dataset was obtained from the National Sleep Research Resources.¹³ Figure 1 summarizes the patient inclusion and exclusion criteria. A total of 2,056 recordings had raw PSG data available. Among them, 1,323 recordings had at least 4 hours of total sleep time (TST) and at least 30 minutes spent in both the supine and nonsupine position. Among the 1,323 patients included in the analysis, a total of 861 patients had OSA. These 861 patients were included in the analysis. Permission to use retrospective medical databases was granted following internal Institutional Review Board (IRB #62-2019).

Figure 1. MESA patients according to condition and exclusion criteria.

Block diagram of MESA patients and exclusion criteria. AHI = apnea-hypopnea index, MESA = Multi-Ethnic Study of Atherosclerosis, NPP = non-positional patients, OSA = obstructive sleep apnea, PP = positional patients, siOSA = supine isolated obstructive sleep apnea, spOSA = supine predominant obstructive sleep apnea, TST = total sleep time.

Respiratory event definitions

Obstructive apnea and recommended hypopnea events were used to calculate the AHI. An obstructive apnea event was defined by a decrease of 90% or more in the amplitude of the airflow signal from a baseline period (ie, a regular breathing period with stable oxygen levels while breathing). This needed to be maintained for at least 90% of the event duration, and for a minimum event duration of 10 seconds. A recommended hypopnea event was defined as a reduction from baseline of more than 70% in breathing amplitude that lasted 10 seconds or more. Hypopnea events had to be associated with a 4% oxygen desaturation.

Definitions of PP and its derivatives

In relation to positionality, we divided the patients with OSA into the following 4 groups (Figure S2 ^{(1.6MB, pdf)} in the supplemental material):

Supine predominant OSA (spOSA)	$\frac{{\text{AHI}}_{\text{supine}}}{{\text{AHI}}_{\text{nonsupine}}}\ge 2$, and ${\text{AHI}}_{\text{nonsupine}}>5$.
Supine isolated OSA (siOSA)	$\frac{{\text{AHI}}_{\text{supine}}}{{\text{AHI}}_{\text{nonsupine}}}\ge 2$, and ${\text{AHI}}_{\text{nonsupine}}\le 5$.
Lateral predominant OSA (lpOSA)	$\frac{{\text{AHI}}_{\text{nonsupine}}}{{\text{AHI}}_{\text{supine}}}\ge 2$, and ${\text{AHI}}_{\text{supine}}>5$.
Lateral isolated OSA (liOSA)	$\frac{{\text{AHI}}_{\text{nonsupine}}}{{\text{AHI}}_{\text{supine}}}\ge 2$, and ${\text{AHI}}_{\text{supine}}\le 5$.

As shown in Figure S2 ^{(1.6MB, pdf)} , Lateral PP includes both lpOSA and liOSA, and Supine PP includes both spOSA and siOSA.

Statistical tests

We used the Kolmogorov-Smirnov test to compare the distribution of TST, TST in the supine position (${\text{TST}}_{\text{supine}}$), TST in nonsupine positions (TST_nonsupine), and the ratio of TST_supine by the TST $(\frac{{\text{TST}}_{\text{supine}}}{\text{TST}})$, between the Lateral PP groups and all the other patients. The Mann-Whitney nonparametric statistical test (U test) was used to assess the assumption that AHI_nonsupine of Lateral PP is greater than the AHI_nonsupine of the supine PP. For the multivariate test, to compare between the characteristics of the different OSA phenotypes, or different ethnic groups, we used the Kruskal-Wallis H-test for independent samples. We also used this statistical test to examine the prevalence of different OSA phenotypes among patients with OSA with different ethnicities. Statistical tests were made in Python 3 using the statistical functions (“stats”) of the SciPy 1.6.1 package¹⁴. A multivariate logistic model was performed using the generalized linear model framework in R (R Foundation for Statistical Computing, Vienna, Austria) to model PP vs NPP, and Lateral PP vs Supine PP, based on body mass index (BMI), sex, age, race, and comorbidities. Statistical significance was considered for 2-sided P values ≤ .05 and odd ratios were reported. Summary statistics are reported as medians and interquartile range (IQR).

RESULTS

OSA severity groups and PP prevalence

The 861 patients with OSA were relatively old, with a median age of 67 years, and with remarkably normal daytime sleepiness, with 84.7% of patients having a normal Epworth Sleepiness Scale (ESS) score (Table 1). Supine PP, following Cartwright’s definition,⁹ was found in 550 (63.9%) out of 861 patients with OSA. Supine PP was found in 306 of 443 patients (69.07%) with mild OSA, 158 of 254 patients (62.20%) with moderate OSA, and 86 of 164 patients (52.44%) with severe OSA. The total numbers of patients with siOSA were 232 with mild OSA (75.82% of the mild Supine PP and 52.37% of the total number of mild OSA), 48 with moderate OSA (30.38% of the moderate Supine PP and 18.90% of the total number of moderate OSA), and 13 with severe OSA (15.12% of the severe Supine PP and 7.93% of the total number of severe OSA). Figure S3 ^{(1.6MB, pdf)} shows the fraction of Supine PP, siOSA, spOSA, and NPP for each OSA severity group, and also the fraction of spOSA and siOSA in each severity’s Supine PP group.

Table 1.

Demographical, clinical, and polysomnographic data of 861 patients with OSA.

Variable	OSA, n (%)
N (%)	861 (100)
Sex, n (%)
Male	448 (52.0)
Race/ethnicity, n (%)
White, Caucasian	319 (37.0)
Chinese American	104 (12.1)
Black, African American	233 (27.1)
Hispanic	205 (23.8)
Cigarette smoking, n (%)
Never	400 (46.5)
Former	401 (46.6)
Current	57 (6.6)
None	3 (0.3)
Hypertension, n (%)
No	360 (41.8)
Yes	500 (58.1)
None	0 (0.1)
Asthma, n (%)
No	827 (96.1)
Yes	32 (3.7)
Don’t know	1 (0.1)
None	1 (0.1)
ESS, n (%)
Normal (0–10)	729 (84.7)
Mild (11–14)	91 (10.6)
Moderate (15–17)	24 (2.8)
Severe (18–24)	9 (1.0)
None	8 (0.9)
Emphysema or COPD,** n (%)
No	842 (97.8)
Yes	16 (1.9)
Don’t know	2 (0.2)
None	1 (0.1)
Atrial fibrillation, n (%)
No	847 (98.4)
Yes	8 (0.9)
None	6 (0.7)
AHI, median (IQR), events/h	14.5 (15.7)
Supine	23.6 (29.2)
Nonsupine	7.7 (12.6)
REM	4.6 (4.6)
NREM	9.3 (13.2)
Demographics, median (IQR)
Age, y	67 (14)
BMI, kg/m2	28.5 (6.9)
No. of events, median (IQR)
Apneas	15.0 (37.0)
Hypopneas	66.0 (71.0)
Longest event, median (IQR)
Apnea, s	37 (24.4)
Hypopnea, s	47.3 (21.5)
Sleep efficiency, median (IQR)
Average, %	91.6 (4.7)
Standard deviation, %	2.4 (1.7)
Body shape, median (IQR)
Waist, cm	100.0 (17.5)
Hip, cm	104.5 (14.8)
TST, median (IQR), h	6.3 (1.4)
Supine	2.6 (2.3)
Nonsupine	3.5 (2.3)
REM	1.1 (0.7)
NREM	5.1 (1.2)

Demographical, clinical, and polysomnographic data on 861 MESA patients with OSA. **Self-reported. AHI = apnea-hypopnea index, BMI = body mass index, COPD = chronic obstructive pulmonary disease, ESS = Epworth Sleepiness Scale, IQR = interquartile range, MESA = Multi-Ethnic Study of Atherosclerosis, NREM = non–rapid eye movement, OSA = obstructive sleep apnea, REM = rapid eye movement, TST = total sleep time.

Lateral PP analysis

Among the 861 participants with OSA, 35 patients (4.1%) were Lateral PP with 14 lpOSA and 21 liOSA (Table S1 ^{(1.6MB, pdf)} ). From these 35 patients, 12 patients slept less than 1 hour in either the supine or nonsupine position (10 in supine, 2 in nonsupine), as seen in Table S1 ^{(1.6MB, pdf)} . According to a 2-sample Kolmogorov-Smirnov statistical test, the distributions of the TST_supine, TST_nonsupine, and TST_supine/TST, according to Lateral and Supine PP groups, were significantly different (P values < .05). In a recent study¹⁵ conducted in 445 PSG patients and 416 home sleep apnea test patients, the proportion of sleep in the supine position was 44.1% and 44.6%, respectively. This study suggests no clear difference in the proportion of supine in PSG vs home sleep apnea test study, and it also suggests that patients spent more time in the nonsupine position, in accordance with our findings in Table 1. Table S1 ^{(1.6MB, pdf)} shows that most of the TST of the Lateral PP group was in the lateral position, with only 4 of 35 patients having TST_prone greater than 30 minutes and still with a meaningful amount of TST_lateral (between 1.48 and 2.85 hours). Figure S4 ^{(1.6MB, pdf)} and Figure S5 ^{(1.6MB, pdf)} show the distributions of the sleep times for different sleeping positions and for the Lateral PP group and the Supine PP group. The median [IQR] nonsupine AHI was significantly higher (P < .001) for Lateral PP (13.7 [13.3]) than for Supine PP (4.5 [7.6]).

With regard to Table 2, the Lateral PP patients were mainly obese (BMI: 31.4 [4.9] kg/m²), with a significantly higher (P < .005) BMI than patients with spOSA (29.0 [5.9] kg/m²) and patients with siOSA (26.9 [5.2] kg/m²). They had a significantly higher proportion of females than spOSA patients (62.9% vs 39.3%; P = .008). They also mostly had mild–moderate OSA (94.3%) and were mostly non–Chinese American (97.1%), as can be seen in Table 3. In addition, most of them were hypertensive (65.7%), former smokers (51.4%), and did not experience excessive daytime sleepiness according to the ESS (88.4%). However, for these 3 last parameters, the prevalence was not significantly different from that of the other OSA phenotypes. Twenty-one patients out of the 35 (60.0%) were liOSA, but 9 of them (42.9%) slept less than 1 hour in the supine position and the 15 remaining patients slept less than 2.05 hours in this position. Of the 35 Lateral PP, 10 slept less than 1 hour in the supine position and 2 slept less than 1 hour in the lateral position. Still, it is important to note that 25 out of the 35 Lateral PP slept more than 1 hour in the supine position. The multivariate logistic model between PP and NPP showed that age and BMI were significant predictors of NPP, with odds ratios of 1.02 (P = .004) and 1.1 (P = .017), respectively. This suggests that NPP patients are older and overweight. In addition, the multivariate logistic model between Supine PP and Lateral PP revealed that BMI was a significant predictor of Lateral PP, with an odds ratio of 1.25 (P = .019), suggesting that Lateral PP are more obese.

Table 2.

Demographics, clinical, and polysomnographic data per positional group.

Variable	spOSA	siOSA	Lateral PP	NPP
N (%)	257 (29.8)	293 (34.0)	35 (4.1)	276 (32.1)
Sex, n (%)
Male	156 (60.7)*	149 (50.9)	13 (37.1)	143 (52.9)
Race/ethnicity, n (%)
White, Caucasian	96 (37.4)	122 (41.6)	15 (42.9)	86 (31.2)
Chinese American	34 (13.2)	42 (14.3)	1 (2.9)	27 (9.8)
Black, African American	64 (24.9)	67 (22.9)	12 (34.3)	90 (32.6)
Hispanic	63 (24.5)	62 (21.1)	7 (20.0)	73 (27.4)
Cigarette smoking, n (%)
Never	124 (48.2)	173 (48.8)	13 (37.1)	120 (43.5)
Former	109 (42.4)	132 (45.1)	18 (51.4)	142 (51.4)
Current	23 (8.9)	17 (5.8)	3 (8.6)	14 (5.1)
None	0 (0)	1 (0.3)	1 (2.9)	0 (0.0)
Hypertension, n (%)
No	104 (40.5)	149 (50.9)*	11 (31.4)	96 (34.8)
Yes	153 (59.5)	144 (49.1)*	23 (65.7)	180 (65.2)
None	0 (0)	0 (0)	1 (2.9)	1 (0.3)
Asthma, n (%)
No	250 (97.3)	283 (96.6)	32 (91.4)	262 (94.9)
Yes	6 (2.3)	9 (3.1)	2 (8.6)	14 (5.1)
Don’t know	0 (0.0)	0 (0)	0 (0)	0 (0)
None	1 (0.4)	1 (0.3)	0 (0)	0 (0)
ESS, n (%)
Normal (0–10)	209 (81.3)	263 (89.8)	29 (88.4)	228 (82.6)
Mild (11–14)	29 (11.3)	23 (7.8)	3 (8.6)	36 (13.0)
Moderate (15–17)	11 (4.3)	3 (1.0)	2 (5.7)	8 (2.9)
Severe (18–24)	4 (1.6)	2 (07.)	1 (2.9)	2 (0.7)
None	4 (1.6)	2 (0.7)	0 (0)	2 (0.7)
Emphysema or COPD,** n (%)
No	250 (97.3)	288 (98.3)	34 (97.1)	270 (97.8)
Yes	6 (2.3)	4 (1.4)	1 (2.9)	5 (1.8)
Don’t know	0 (0)	1 (0.3)	0 (0)	1 (0.4)
None	1 (0.4)	0 (0)	0 (0)	0 (0.0)
Atrial fibrillation, n (%)
No	253 (98.4)	289 (98.7)	34 (97.1)	271 (98.2)
Yes	3 (1.2)	1 (0.3)	0 (0)	4 (1.4)
None	1 (0.4)	3 (1.0)	1 (2.9)	1 (0.4)
AHI, median (IQR), events/h	19.9 (18.2)*	9.4 (6.4)*	11.2 (8.6)*	17.2 (22.6)*
Supine	39.9 (32.5)*	18.6 (16.8)*	3.8 (4.6)*	19.9 (27.0)*
Nonsupine	9.8 (9.0)*	1.8 (2.7)*	13.7 (13.3)*	16.0 (20.2)*
REM	5.1 (4.6)*	3.2 (3.8)*	6.2 (4.6)	5.9 (5.4)*
NREM	14.8 (15.8)*	5.8 (6.4)*	4.0 (4.7)*	10.6 (21.4)*
Demographics, median (IQR)
Age, y	67 (14)	67 (15)	66 (11.5)	68 (14)
BMI, kg/m2	29.0 (5.9)*	26.9 (5.2)*	31.4 (4.9)*	30.5 (8.0)*
No. of events, median (IQR)
Apneas	24 (44.0)	12.0 (27.0)	10.0 (22.5)	15.5 (44)
Hypopneas	92.0 (77.0)	44.0 (33.0)	51.0 (57.0)	80.0 (89.5)
Longest event, median (IQR)
Apnea, s	41.4 (25.5)*	34.3 (20.9)*	30.7 (18.9)	37.7 (25.5)
Hypopnea, s	49.6 (19.2)	46.9 (25.2)	44.6 (30.3)	46.6 (18.7)
Sleep efficiency, median (IQR)
Average, min	91.0 (5.0)	92.1 (4.6)	91.4 (3.5)	91.5 (4.5)
Standard deviation, min	2.5 (1.9)	2.3 (1.7)	2.4 (1.5)	2.4 (1.6)
Body shape, median (IQR)
Waist, cm	100 (16.0)*	95.9 (14.5)*	103.3 (18.6)	104 (18.2)*
Hip, cm	104.2 (13.0)*	101.8 (11.0)*	109.5 (10.3)*	108.5 (18.5)*
TST, median (IQR), h	6.3 (1.7)	6.3 (1.2)	6.4 (1.4)	6.3 (1.3)
Supine	2.2 (2.2)*	3.1 (2.4)*	1.5 (1.6)*	2.3 (2.5)*
Nonsupine	3.9 (2.2)*	3.0 (2.3)*	4.7 (2.3)*	3.6 (2.2)*
REM	1.1 (0.7)	1.2 (0.6)*	1.2 (0.7)	1.1 (0.6)
NREM	5.1 (1.4)	5.0 (1.1)	5.2 (1.2)	5.3 (1.2)

Demographics, clinical and polysomnographic data of 861 MESA patients with OSA, divided into 4 groups: spOSA, siOSA, Lateral PP, and NPP. *Statistically significantly different between the other 2 or 3 OSA phenotype groups. Statistical difference was computed using Kruskal-Wallis H-test for independent samples, and significance was considered for P values < .05. **Self-reported. AHI = apnea-hypopnea index, BMI = body mass index, COPD = chronic obstructive pulmonary disease, ESS = Epworth Sleepiness Scale, IQR = interquartile range, MESA = Multi-Ethnic Study of Atherosclerosis, NPP = non-positional patients, NREM = non–rapid eye movement, OSA = obstructive sleep apnea, PP = positional patients, REM = rapid eye movement, siOSA = supine isolated OSA, spOSA = supine predominant OSA, TST = total sleep time.

Table 3.

Positional OSA data according to ethnic group.

Ethnicity	n	OSA, n (%)	Supine PP, n (%)	spOSA, n (%)	siOSA, n (%)	Lateral PP, n (%)	NPP OSA, n (%)
White/Caucasian	493	319 (64.7)	218 (68.3)	96 (30.1)	122 (38.2)	15 (4.7)	86 (27.0)*
Chinese American	167	104 (62.3)	76 (73.1)*	34 (32.7)	42 (40.4)	1 (1.0)	27 (26.0)
Black/African American	367	233 (63.5)	131 (56.2)*	64 (27.5)	67 (28.8)*	12 (5.2)	90 (38.6)*
Hispanic	296	205 (69.3)	125 (61.0)	63 (30.7)	62 (30.2)	7 (3.4)	73 (35.6)

spOSA, siOSA, Lateral PP, and NPP OSA prevalence of each MESA participant’s ethnicity. The percentages of the first 3 are calculated by conditioned n (eg, spOSA n) out of the total number of patients with OSA of the relevant ethnicity. *Statistically significantly different in OSA phenotype incidence between the other 2 or 3 examined ethnic groups. Statistical difference was computed using Kruskal-Wallis H-test for independent samples, and significance was considered for P values < .05. MESA – Multi-Ethnic Study of Atherosclerosis, OSA = obstructive sleep apnea, NPP = non-positional patients, PP = positional patients, siOSA = supine isolated OSA, spOSA = supine predominant OSA.

For Lateral PP, the trend of positional AHI in rapid eye movement (REM) sleep and the trend of positional AHI in non-REM (NREM) sleep were opposite to all the other groups, and it reflected the same effect of position on positional AHI in general—that is, supine AHI was greater than nonsupine AHI for Supine PP (eg, for spOSA: 39.9 [32.5] > 9.8 [9.0] events/h), and for Lateral PP, nonsupine AHI was higher (13.7 [13.3] > 3.8 [4.6] events/h). That is, for Lateral PP, AHI REM in the nonsupine position was higher than AHI REM in the supine position, and AHI NREM in the nonsupine position was higher than AHI REM in the supine position (21.4 [29.5] > 0 [8.6] and 17.0 [15.3] > 14.0 [17.8] events/h, respectively). Moreover, almost no apnea or hypopnea events occurred in the supine posture during REM sleep for most of the Lateral PP (0 [8.6]). We did not find a statistical difference in the prevalence of hypertension between Lateral PP compared with the other groups. However, hypertension prevalence among siOSA was significantly lower than in all other groups (49.1% vs 59.5–65.7%; P < .05).

Ethnicity analysis

Table 3 shows that Black/African-American patients with OSA had a prevalence of 28.8% siOSA. This was significantly lower than Chinese Americans and White/Caucasians, who had a prevalence of 40.4% and 38.2% (P ≤ .036), respectively. The incidence of spOSA among Black/African Americans was the least frequent (27.5%); however, this did not show statistical significance. Overall, Chinese Americans tended to be more Supine PP (73.1%) and less Lateral PP (1.0%) compared with other ethnic groups. The Black/African-American patients had a lower proportion of Supine PP (56.2%) and a higher proportion of Lateral PP (5.2%) compared with other ethnic groups. These observations were statistically significant (P = .003) for Supine PP but not for Lateral PP. We also found that Black/African-American patients had a significantly higher prevalence of NPP than Chinese Americans and Whites/Caucasians (38.6% vs 26.0% and 27.0%, respectively; P ≤ .004) and that Whites/Caucasians were also statistically significantly less likely to be NPP than Hispanics (27.0% vs 35.6%; P = .036).

DISCUSSION

The results of this study highlight 2 important novel findings: we present for the first time, in a relatively large group of patients with OSA, the PP characteristics for multiple ethnic groups. In addition, we report the existence of a new phenotype of OSA PP—namely, Lateral PP—and provide demographic and PSG characteristics for this group. The prevalences of Supine PP among OSA severity groups were as follows: 69.1% for mild, 62.2% for moderate, and 52.4% for severe OSA. These results are in line with the literature showing that Supine PP are mainly dominant in the less severe groups of patients with OSA.^2–11,16 Moreover, our findings on Supine PP and siOSA prevalence among OSA in the MESA were also aligned with previous reports.^6,16 Overall, up to 34.0% of all individuals with OSA were siOSA, and by using PT efficiently, they could resolve their OSA condition. In addition, another 29.8% of all patients with OSA were spOSA, and therefore could have their OSA condition remarkably improved by effectively using this behavioral therapy.

Characteristics of OSA PP according to ethnicity

We found that Chinese Americans had a higher prevalence of Supine PP, whereas Black/African-American patients had significantly less Supine PP compared with other ethnic groups. In addition, we also found that Black/African-American patients had a significantly higher prevalence of NPP than Chinese Americans and Whites/Caucasians. Also, compared with Hispanics, Whites/Caucasians were significantly less NPP. The relatively significantly low prevalence of Supine PP and high prevalence of NPP among Blacks/African Americans might be partially explained by some clinical, demographic, and PSG differences with respect to the other ethnic groups (Table 4). Blacks/African Americans in MESA had a significantly higher BMI and waist circumference than did Whites/Caucasians and Chinese Americans. The Black/African-American group experienced significantly more hypertension and had a more severe daytime sleepiness level (ESS) compared with all other ethnic groups. They had a lower supine AHI compared with Chinese Americans and Hispanics. On the contrary, the AHI in REM sleep for Blacks/African Americans was significantly higher than for Whites/Caucasians and Chinese Americans.

Table 4.

Demographics, clinical, and polysomnographic information of patients with OSA according to ethnic group.

Variable	White/Caucasian	Chinese American	Black/African American	Hispanic
n	319	104	233	205
Sex, n (%)
Male	170 (53.3)	65 (62.5)	112 (48.1)	101 (49.3)
Cigarette smoking, n (%)
Never	133 (41.7)	74 (71.1)*	97 (41.6)	96 (46.8)
Former	170 (53.3)	24 (23.1)*	113 (48.5)	94 (45.8)
Current	15 (4.7)	6 (5.8)	22 (9.4)	14 (6.8)
None	0 (0.3)	0 (0)	1 (0.4)	1 (0.3)
Hypertension, n (%)
No	152 (47.6)	58 (55.8)*	61 (26.2)*	89 (43.4)*
Yes	166 (52.0)	46 (44.2)*	172 (73.8)*	116 (56.6)*
None	1 (0.3)	0 (0)	0 (0)	0 (0)
Asthma, n (%)
No	312 (97.8)	101 (97.1)	220 (94.4)	280 (94.6)
Yes	6 (1.9)	3 (2.9)	13 (5.6)	15 (5.1)
Don’t know	0 (0)	0 (0)	0 (0)	1 (0.3)
None	1 (0.3)	0 (0)	0 (0)	0 (0)
ESS, n (%)
Normal (0–10)	280 (87.8)*	90 (86.5)*	182 (78.1)*	260 (87.8)
Mild (11–14)	25 (7.8)*	11 (10.6)*	35 (15.0)*	25 (8.5)
Moderate (15–17)	10 (3.1)*	0 (0)*	10 (4.3)*	6 (2)
Severe (18–24)	2 (0.6)*	1 (1.0)*	4 (1.7)*	4 (1.4)
None	2 (0.6)	2 (1.9)	3 (1.3)	1 (0.3)
Emphysema or COPD,** n (%)
No	483 (98)	103 (99.0)	227 (97.4)	202 (98.5)
Yes	9 (1.8)	1 (1.0)	5 (2.1)	2 (1.0)
Don’t know	0 (0)	0 (0)	1 (0.4)	1 (0.5)
None	1 (0.2)	0 (0)	0 (0)	0 (0)
Atrial fibrillation, n (%)
No	312 (97.8)	166 (99.4)	103 (99.0)	203 (99.0)
Yes	5 (1.6)	1 (0.6)	1 (0.4)	1 (0.5)
None	3 (0.6)	0 (0)	0 (0)	1 (0.5)
AHI, median (IQR), events/h	13.0 (14.6)*	15.8 (15.9)	14.5 (14.5)	15.8 (18.7)
Supine	23.1 (29.0)	28.0 (31.5)	20.0 (26.7)*	25.0 (28.7)
Nonsupine	6.6 (12.1)*	7.1 (12.3)	8.9 (13.4)	8.7 (13.1)
REM	3.9 (4.5)*	3.8 (3.6)*	5.1 (4.9)*	5.7 (5.5)*
NREM	9.0 (12.0)	12.0 (16.9)	8.8 (13.4)	9.3 (14.8)
Demographics, median (IQR)
Age, y	68 (15.0)	64.5 (14.0)	68.0 (14.0)	67 (15.0)
BMI, kg/m2	27.7 (6.3)*	24.4 (3.8)*	30.4 (7.1)*	30.5 (6.7)*
No. of events, median (IQR)
Apneas	12.0 (31.0)	21.0 (49.3)	16.0 (36.0)	18.0 (44.0)
Hypopneas	62.0 (75.2)	59.0 (51.5)	65.0 (70.0)	75.0 (78.0)
Longest event, median (IQR)
Apnea, s	35.2 (23.9)*	43.4 (58.8)*	34.4 (20.0)*	41.8 (28.4)*
Hypopnea, s	47.3 (24.0)	53.1 (25.0)*	44.8 (18.4)*	47.5 (18.8)*
Sleep efficiency, median (IQR)
Average, min	92.4 (4.3)*	91.5 (4.4)	90.6 (4.8)	91.4 (4.8)
Standard deviation, min	2.3 (1.5)*	2.6 (1.6)	2.8 (1.9)	2.4 (1.7)*
Body shape, median (IQR)
Waist, cm	99.5 (17.4)*	89.2 (9.3)*	102.8 (16.7)*	103.0 (14.0)*
Hip, cm	104.6 (12.8)*	95.7 (7.2)*	109.0 (16.8)*	105.6 (13.8)*
TST, median (IQR), h	6.5 (1.5)*	6.2 (1.4)	6.1 (1.5)	6.3 (1.3)
Supine	2.4 (2.4)	2.7 (2.0)	2.7 (2.4)	2.7 (2.4)
Nonsupine	3.8 (2.4)*	3.3 (1.8)	3.2 (2.4)	3.6 (2.4)
REM	1.2 (0.6)	1.0 (0.6)*	1.1 (0.7)	1.2 (0.6)
NREM	5.2 (1.3)	5.0 (1.2)	5.3 (1.2)	5.1 (1.2)

Demographics, clinical, and polysomnographic information of 861 MESA patients with OSA, divided into 4 ethnic groups: White/Caucasian, Chinese American, Black/African American, and Hispanic. *Statistically significantly different between the other 2 or 3 examined ethnic groups. Statistical difference was computed using Kruskal-Wallis H-test for independent samples, and significance was considered for P values < .05. **Self-reported. AHI = apnea-hypopnea index, BMI = body mass index, COPD = chronic obstructive pulmonary disease, ESS = Epworth Sleepiness Scale, IQR = interquartile range, MESA = Multi-Ethnic Study of Atherosclerosis, NREM = non–rapid eye movement, OSA = obstructive sleep apnea, REM = rapid eye movement, TST = total sleep time.

It has already been reported many times in the literature that Supine PP are patients with OSA with less severe disease compared with NPP.^8,17,18 In our data, Chinese Americans had a higher prevalence of PP, whereas Blacks/African Americans had a high prevalence of NPP. However, these positional prevalence differences were not reflected by the AHI across ethnic groups (Table 4). These results suggest that the differences in positional prevalence found between ethnic groups in this study, although statistically significant, are not sufficient to translate into establishing AHI severity differences.

The Black/African-American group had a significantly higher BMI compared with Whites/Caucasians and Chinese Americans. They also had a high prevalence of hypertension and higher self-reported sleepiness during the daytime when compared with all 3 other ethnic groups. In 1995, Ancoli-Israel et al¹⁹ found that more Blacks/African Americans had severe OSA, with a relative risk of 2.13 compared with Whites/Caucasians. Overall, this supports the notion that Blacks/African Americans would be an OSA group with a supposedly more severe form of the disease. Nevertheless, in our dataset, Black/African-American patients did not have more severe disease (expressed by the AHI) than the other ethnic groups. However, it was noticed that they did have the highest REM AHI in the supine position.

Asians patients with OSA had a high Supine PP prevalence (∼75%). The high prevalence in this group was previously reported by Mo et al.¹⁰ A possible explanation is the short cranial base and retrognathia that are more common among Asians and may lead to an airway structure that tends to collapse in the supine position.²⁰ Whites’/Caucasians’ Supine PP prevalence in the literature has been reported to be in the range of 26.7–55.9%.²⁰ There is a lack of research regarding other ethnic groups in the context of Supine PP. According to our findings, the Chinese-American ethnic group was strongly affected by sleeping position, with 74.1% of all patients having either Supine or Lateral PP, as compared with 61.4–73% for the other ethnic groups.

Demographic and PSG characteristics of Lateral PP

Our analysis highlights a novel phenotype of PP that we coined “Lateral PP,” which included patients with OSA having a nonsupine AHI more than double the supine AHI. This was a small group of patients (4.1%) among the 861 patients with OSA. Lateral PP were mainly obese, female, and with mild–moderate OSA. Chinese Americans comprised the group of patients with OSA with the lowest prevalence of Lateral PP compared with the other ethnicity groups. Approximately one-third of Lateral PP slept less than 1 hour in the supine position and therefore it is possible that the reason they did not show apneas/hypopneas in the supine position is the limited time slept in this position. This critical issue could be solved by having these patients sleep several nights in the laboratory or their home in order to investigate whether the results remain consistent. One possibility is that these are Supine PP with OSA who consciously avoid the supine posture since they know, based on their experience, that this sleeping position worsens their OSA condition. Nevertheless, another possibility is that this could also be a specific group of patients with OSA for whom the condition is truly worse in the nonsupine position compared with the supine position. Yet, it is important to note that the majority of the Lateral PP (71.4%) slept more than 1 hour in the supine position, and, for them, it is less likely that the reason they did not show apneas/hypopneas in the supine position is due to their sleep time in the supine position. We do not know if these patients had some unique cephalometric characteristics, and this is certainly also a topic for future research. It is important to note that the nonsupine AHI of Lateral PP was significantly higher than the nonsupine AHI of Supine PP, and most of the breathing abnormalities occurred during REM sleep, a known characteristic of many women with OSA.²¹ All of these observations require confirmation from the study of other datasets of patients with OSA and, ideally, the study of those Lateral PP over several nights to investigate the night-to-night prevalence variability of this novel positional phenotype.

If this new PP phenotype, named “Lateral PP,” is confirmed, it may have clinical consequences for the treatment recommendations for this group of patients. Lateral PP would be advised to adopt the supine posture and avoid the lateral position (ie, the opposite recommendation given to Supine PP). Indeed, PT could be an appropriate therapy for Lateral PP. Nevertheless, how effective PT would be in Lateral PP remains to be studied. We know that compliance with PT is a challenge. During many years the Tennis Ball Technique was a popular PT for avoiding the supine posture during sleep, but this therapy was uncomfortable and compliance was low.²² Nevertheless, during recent years, a new generation of devices for PT that provide a subtle vibrating stimulus that prevents patients from adopting the supine position has been used successfully by many Supine PP with OSA, and the results are good for short-term use but still limited for long-term use.^23,24

Limitations

The study considered only 1-night PSG examination, and it is known that consecutive-night PSG study can change the prevalence of PP and NPP.²⁵ Moreover, another aspect that may influence the results on the prevalence of OSA and its phenotypes (eg, Supine PP, Lateral PP) is the definition used for apnea and hypopneas. The definitions used in this research were those from the MESA and similar to previous publications^12,26,27 reporting on the MESA database. However, there exist some differences in the definition of hypopnea with the latest American Academy of Sleep Medicine 2012 recommendations and thus it will be valuable in future work to validate our hypothesis in a multiethnic dataset annotated with the latest American Academy of Sleep Medicine guidelines. We chose to use the most common definition for PP (Cartwright’s definition), but using other definitions for OSA PP could provide different results. It should be mentioned that we cannot completely refute that the Lateral PP phenotype is not a consequence of different positional sleep time, as we rejected Kolmogorov-Smirnov’s null hypothesis for supine sleep time, for lateral sleep time, and for TST_supine/TST, meaning that the distribution of those statistics for Lateral PP and Supine PP were different. This requires further investigation. Finally, the Lateral PP phenotype was observed in a small group of patients with OSA and thus its existence should be further validated.

CONCLUSIONS

In this study we describe for the first time demographic and polysomnographic characteristics of positional patients, over different ethnic groups in a relatively large population of patients with OSA. In addition, we report a novel OSA PP phenotype, namely Lateral PP, and characterize those patients demographically and using PSG. Future research will need to confirm the results of this preliminary report.

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. Work for this study was performed at the Technion Institute of Technology, Haifa, Israel. Y.B.S. and J.A.B. acknowledge the financial support of the Technion Machine Learning and Intelligent Systems center (MLIS). J.A.S. was partially supported by The Milner Foundation, founded by Yuri Milner and his wife Julia. J.A.S. acknowledges the Placide Nicod Foundation for their financial support. The authors report no conflicts of interest.

ABBREVIATIONS

AHI,

apnea-hypopnea index

BMI,

body mass index

ESS,

Epworth Sleepiness Scale

IQR,

interquartile range

liOSA,

lateral-isolated obstructive sleep apnea

lpOSA,

lateral-predominant obstructive sleep apnea

MESA,

Multi-Ethnic Study of Atherosclerosis

NPP,

non-positional patients

NREM,

non–rapid eye movement

OSA,

obstructive sleep apnea

PP,

positional patients

PSG,

polysomnography

PT,

positional therapy

REM,

rapid eye movement

siOSA,

supine-isolated obstructive sleep apnea

spOSA,

supine-predominant obstructive sleep apnea

TST,

total sleep time