Background. Obstructive sleep apnoea is commonly aggravated by the supine body position. e impact of body position on the severity
of mixed and central sleep apnoeas is understudied.
Objectives. To evaluate the impact of body position on obstructive, mixed and central apnoea indices in subjects presenting with this triform
of sleep apnoea during a single polysomnogram.
Methods. We retrospectively analysed 26 polysomnograms where obstructive, mixed and central apnoeas each occurred at a rate >5/hr.
Comparisons between lateral and supine body positions were made for obstructive apnoea index (OAI), mixed apnoea index (MAI), central
apnoea index (CAI), apnoea-hypopnoea index (AHI) and obstructive apnoea-hypopnoea index (OAHI).
Results. Mean (SD) apnoea indices were signicantly lower in lateral v. supine positions, respectively: MAI 15.06 (18.34) v. 32.09 (17.05);
p<0.001, CAI 11.82 (11.77) v. 23.82 (14.18); p<0.001, AHI 79.46 (31.17) v. 99.47 (26.33); p<0.001, OAHI 67.87 (28.25) v. 76.00 (23.21);
p=0.039. Unexpectedly, the converse was seen for OAI when comparing the lateral v. supine position: 53.10 (30.64) v. 43.58 (25.83); p=0.009,
Conclusion. It may be benecial for subjects with a combination of obstructive, mixed, and central apnoeas to avoid the supine body
position. In this triform phenotype, mixed apnoeas are neither purely obstructive nor purely centrally mediated. Furthermore, obstructive,
mixed, and central apnoeas may be dierent representations of a single respiratory abnormality.
Afr J Thoracic Crit Care Med 2019;25(4):141-144. https://doi.org/10.7196/AJTCCM.2019.v25i4.024
Sleep apnoea is a breathing disorder characterised by recurrent,
partial or complete pauses in respiration. Two major types of sleep
apnoea syndromes are recognised, each with its own underlying
pathophysiology. Obstructive sleep apnoea syndrome is largely the
result of upper airway instability and collapse, while central sleep
apnoea syndrome is more complex and is attributed to insucient
or absent central ventilatory drive, oen secondary to cardiac or
neurological disease. In overnight polysomnography, however,
regarded as the preferred laboratory test for the evaluation of sleep-
related breathing disorders, three major types of sleep apnoea
are measured: obstructive, mixed and central apnoeas. The
classification of mixed apnoeas remains unclear. The American
Academy of Sleep Medicine (AASM) suggests that mixed apnoeas
are part of obstructive apnoeas, despite displaying polysomnographic
(PSG) features of both central and obstructive apnoeas. Furthermore,
mixed apnoeas do not always respond well to continuous positive
airway pressure therapy, the preferred treatment method of
obstructive sleep apnoea syndrome. During the treatment of such
cases, the apnoea indices of patients with predominantly mixed or
obstructive apnoeas may either decrease in severity without achieving
normalisation, or an unfavourable central apnoea response may be
inadvertently induced. It is well established that body position during
sleep has an impact on the severity of obstructive sleep apnoea, where
the lateral position tends to be less severe than the supine position.
e occurrence of obstructive apnoeas can be twice as frequent in the
supine body position v. the lateral body position, an eect known as
positional dependent obstructive sleep apnoea. With the exception
of Cheyne-Stokes breathing, the impact of body position on other
forms of central sleep apnoea is an area that is understudied.
is study builds on previous sleep disorder studies by examining
the impact of body position on the severity of obstructive, mixed
and central sleep apnoeas. By not limiting our study participants to
only cardiovascular-compromised patients, we were able to evaluate
various other types of central sleep apnoea in addition to Cheyne-
Stokes breathing. In addition, this study investigated whether mixed
apnoeas, which display characteristics of both obstructive and central
apnoeas, were strictly of an obstructive origin.
Subjects were eligible for this study if they were referred to the
sleep laboratory (Pretoria, South Africa) as a standard referral
for the evaluation of possible sleep-disordered breathing between
2009 and 2016. Retrospective data were collected aer a search of
the sleep laboratory’s database for polysomnograms matching our
criteria. Verbal consent was obtained aer a structured telephonic
e eects of body position on the distribution of obstructive,
mixed and central sleep apnoea
G van der Col,1 M Tech ClinTech; P R Bartel,2 PhD; P Becker,3 PhD; L T Hazelhurst,1 MSc
1 Department of Biomedical Sciences, Faculty of Science, Tshwane University of Technology, Pretoria, South Africa
2 Steve Biko Academic Hospital, Pretoria, South Africa
3 Research Unit, Faculty of Health Sciences, University of Pretoria, South Africa
Corresponding author: G van der Col (email@example.com)
AJTCCM VOL. 25 NO. 4 2019 141
This open-access article is distributed under
Creative Commons licence CC-BY-NC 4.0.
142 AJTCCM VOL. 25 NO. 4 2019
interview of subjects. Formal ethical approval was obtained from the
Tshwane University of Technology Faculty of Science Committee
for Research Ethics (ref. no. FCRE 2016/04/003).e primary PSG
criteria for inclusion were an obstructive apnoea index (OAI), mixed
apnoea index (MAI) and central apnoea index (CAI) of >5 per hour
of sleep and occurring within a single overnight polysomnogram.
Subjects >18 years of age were required to spend a minimum of
30 minutes in (i) each of the lateral positions and (ii) the supine body
position. Subjects who were unable to change their body position at
will (for example, during post-operative posturing or in the case of
neuromuscular disorders) were excluded from the study.
Overnight polysomnography was performed either in a sleep
laboratory or in a hospital ward. No home-based sleep studies were
conducted. Subjects were not prompted to assume any specic body
position during the polysomnogram recordings. All polysomnograms
were acquired from dedicated PSG equipment (Alice PDx, Philips-
Respironics Inc., USA) and were analysed using the device’s soware
(Alice Sleepware, version 2.8.78, Philips-Respironics Inc., USA).
PSG measurements included electroencephalography (C3 and C4
placements), electrooculography, submandibular electromyography,
leg electromyography (anterior tibialis), electrocardiography, pulse
rate and oxygen saturation via pulse oximetry, nasal airflow and
snoring via nasal cannula, thoracic and abdominal respiratory eort
via inductance plethysmography belt sensors, and body position via
a built-in sensor within the recording device attached to the thorax
at mid-sternal level.
PSG data were standardised by manual reanalysis in accordance
with the criteria in the latest AASM scoring manual at the time
of the study (version 2.2). Rule 1A for scoring hypopnoeas was
used. No distinction was made between obstructive hypopnoeas
and central hypopnoeas as per AASM guidelines. For practical
reasons, oesophageal manometry was not performed during
polysomnography and all hypopnoeas were assumed to be
obstructive. Body positioning recording was derived from sensor
readings that dened supine, le and right body positions. e lateral
body position was recorded for le and/or right sensor readings. e
OAI was dened as the sum of obstructive apnoeas and hypopnoeas.
By this denition, the term obstructive apnoea when referring to
a specic type of sleep apnoea includes hypopnoeas. e CAI was
dened as the sum of central apnoeas and included Cheyne-Stokes
breathing. e MAI was dened as the sum of all respiratory events
that received a mixed apnoea scoring. e apnoea-hypopnoea index
(AHI) was dened as the sum of all types of sleep apnoeas and
detected in a polysomnogram.
e obstructive apnoea-hypopnoea index (OAHI) was dened
as the index derived from the cumulative values of all obstructive
apnoeas, hypopnoeas and mixed apnoeas as per AASM guidelines.
Measurements from each polysomnogram included the total sleep
time, and sleep time in le, right, lateral, and supine body positions.
The total number of apnoeas, apnoeas for each apnoea type and
apnoeas in each body position was determined separately. erefore,
it was possible to determine the apnoea index of the le, right, lateral
and supine body positions, for each of the OAI, CAI, MAI, OAHI,and
AHI. ese within-subject measurements were allocated to strata
according to the apnoea index of each apnoea type respectively,
and additionally, according to the apnoea index of the respective
body positions. Statistical analysis was performed on Stata soware
(version 14, Stata Corp LLC,USA). In line with previous ndings, [9,10]
no signicant dierences in the apnoea indices between le and right
body positions were found. erefore, the le and right body position
indices were combined, and dened the lateral body position index,
which was henceforth used for comparing the severity of sleep apnoea,
in terms of the apnoea index, of any given apnoea type, between the
lateral and supine body position.
Inter-body position comparisons (lateral v. supine), of obstructive,
central and mixed apnoeas, by means of the observation vector (OAI,
CAI, MAI) was assessed using Hotelling’s paired T2-test at a 0.05 level
of signicance, while specic dierences between body positions were
assessed using Student’s paired t-tests at a 0.0167 Bonferroni adjusted
level of signicance. Furthermore, inter-body position comparisons
of the AHI and OAHI (which are linear functions of the OAI, CAI,
and MAI), were respectively assessed using Student’s paired t-tests at a
0.05 level of signicance. To compare obstructive, central, and mixed
apnoeas by body position (intra-body position comparisons), random-
eects Generalised Least Squares (GLS) regression was performed at a
0.05 level of signicance, with the index type (OAI, CAI, and MAI) as
xed eect, and participant specied as the random component with an
intercept. Linear predicted means, together with their 95% condence
intervals (CI) were reported by index over body position.
A total of 31 potential candidates (1.1%) were identified from a
database of 2 802 patients, of whom 5 could not be reached for consent.
Included in the study were 26 subjects between the ages of 28 and 85
years of whom 24 were male. e demographic data of these subjects
are summarised in Table 1. Subjects were generally obese, with the
body mass index (BMI) exceeding 30 kg/m2 in 21 (81%) subjects.
Cheyne-Stokes breathing was present in 2/26 subjects.
PSG analysis determined a mean (SD) total sleep time of 473.06
(50.97) minutes. Subjects had longer sleep times in the le v. right
body position: mean (SD) 168.94 (125.21) v. 110.42 (86.59) minutes
respectively and in the lateral v. supine body position: mean (SD)
279.36 (117.13) v. 178.00 (109.63) minutes, respectively. Four subjects
completely avoided sleeping in the left position, while another
4 subjects avoided the right; however, they still met the inclusion
Table 1. Demographic data (N=26)
Variable Mean (SD) 95% CI Range
Age, years 50.22 (14.15) 44.50 - 55.94 28.90 - 85.20
BMI, kg/m234.41 (5.59) 32.15 - 36.67 26.70 - 49.40
SD = standard deviation; CI = condence interval; BMI = body mass index.
AJTCCM VOL. 25 NO. 4 2019 143
criterion of >30 minutes of sleep in either one of the lateral positions.
e AHI was severely abnormal (>30/hour) in all 26 subjects: mean
(SD) index 88.16/hour (26.89). Mean (SD) indices for each body
position for AHI were as follows: le 79.19/hour (29.77), right 78.27/
hour (32.60), lateral 79.46/hour (31.17) and supine 99.47/hour (26.33).
Inter-positional apnoea index comparisons showed that the CAI, MAI,
AHI and OAHI were signicantly lower in the lateral than the supine
body position (Table 2). Unexpectedly, the OAI was signicantly higher
in the lateral compared to the supine body position. A Hotelling’s paired
-test with OAI + CAI + MAI as a single vector showed signicant
dierences between lateral and supine body positions (p<0.001).
Random-effects GLS regression was used to compare the intra-
positional distribution of obstructive, central and mixed apnoeas; thus,
the OAI, CAI, and MAI were compared within the lateral and supine
body positions, respectively (Table 3). While the CAI and MAI were
comparable, both were signicantly lower than the OAI irrespective of
body position. Of additional interest was to investigate whether body
position aected the severity of obstructive, central and mixed sleep
apnoea during rapid eye movement (REM) sleep; however, only one
subject met all the inclusion criteria during REM sleep and therefore
this analysis was not performed.
The sample group in this study was representative of a severe
obstructive sleep apnoea syndrome(mean AHI 88.16/hour),
presenting mostly with obstructive apnoeas but also with a signicant
proportion of mixed and central apnoeas. is triform phenotype of
sleep apnoea may be considered a rare phenotype, with a prevalence
of less than 5%. e most important nding in this study was that
while all apnoea indices were above the normal threshold in both the
supine and lateral body positions, 4/5 apnoea indices (CAI, MAI,
OAHI and AHI) were signicantly lower in the lateral body position.
e exception was the OAI, where the severity of obstructive apnoeas
was signicantly higher in the lateral compared to the supine body
position. While it is well established that obstructive sleep apnoea is
more severe in the supine position, this was evident in less than
a third (27%) of the subjects in this study. For the OAHI, however,
which represented all obstructive-mediated apnoeas collectively,
sleep severity relative to body position was consistent with ndings
from other studies. Considering that the OAHI, MAI and OAI are
calculated from obstructive-mediated events, it is unclear why the
OAI deviates from the MAI and the OAHI ndings with respect to
severity in supine v. lateral body positions. ree possible explanations
are proposed: (i) in subjects who present with a combination
of obstructive, mixed and central apnoeas, reflex inhibition of
respiratory eort, mediated via receptors in the mucosa of the upper
airway, may be triggered upon airway collapse associated with
obstructive apnoeas, specically while in the supine position. is
may result in an increase in the occurrence of central and mixed
apnoeas during supine sleep, as upper airway collapse is known
Table 2. Comparison of apnoea indices between lateral and supine body position (N=26)
Index Type Body Position Mean (SD)* 95% CI p-value†
OAI Lateral 53.10 (30.64) 40.72 - 65.47 0.009
Supine 43.58 (25.83) 33.14 - 54.01
CAI Lateral 11.82 (11.77) 7.07 - 16.58 <0.001
Supine 23.82 (14.18) 18.09 - 29.54
MAI Lateral 15.06 (18.34) 7.65 - 22.47 <0.001
Supine 32.09 (17.05) 25.20 - 38.98
AHI Lateral 79.46 (31.17) 66.87 - 92.05 <0.001
Supine 99.47 (26.33) 88.84 - 110.11
OAHI Lateral 67.87 (28.25) 56.46 - 79.28 0.039
Supine 76.00 (23.21) 66.62 - 85.38
SD = standard deviation; CI = condence interval; OAI = obstructive apnoea index; CAI = central apnoea index; MAI = mixed apnoea index; AHI = apnoea-hypopnoea index, OAHI = obstructive apnoea-
*Apnoea index units in events/hour.
†Comparisons were made between body positions using a Student’s paired t-test with: (i) p<0.0167 considered signicantaer Bonferroni correction for OAI, CAI and MAI; (ii) p<0.05 considered signicant
for overall apnoea index (AHI or OAHI).
Table 3. Intra-position comparison of the apnoea indices (events/hour) between apnoea types (N=26)
Body Position Index Type Mean (95% CI)*
v. OAI v. CAI
OAI 53.10 (44.75 - 61.44) - -
CAI 11.82 (3.48 - 20.17) <0.001 -
MAI 15.06 (6.71 - 23.40) <0.001 0.591
OAI 43.58 (36.02 - 51.13) - -
CAI 23.82 (16.26 - 31.37) <0.001 -
MAI 32.09 (24.53 - 39.64) 0.035 0.129
CI = condence interval; OAI = obstructive apnoea index; CAI = central apnoea index; MAI = mixed apnoea index.
*Apnoea index units in events/hour.
†Random-eects Generalised Least Squares regression with p<0.05 considered signicant.
144 AJTCCM VOL. 25 NO. 4 2019
to be aggravated by this body position; (ii) subjects presenting
with this triform of sleep apnoea evaluated in the current study,
have been associated with altered, or increased respiratory control
instability, that may have contributed to the atypical distribution of
obstructive apnoeas;[4,14] (iii) obstructive apnoea indices vary in the
lateral position whereas are more stable in the supine position.
e OAHI, which represents all obstructive-mediated apnoeas 
was signicantly lower in the lateral compared to the supine body
position as expected. Therefore, it could be argued that mixed
apnoeas contributed to the obstructive-mediated apnoea index
severity in relation to body position. Furthermore, mixed apnoeas
displayed characteristics of central apnoeas with respect to severity
and body position. erefore, based on patterns in severity between
supine and lateral sleeping positions, the MAI displayed both
obstructive and centrally mediated sleep apnoea characteristics.
Given the similarities between the OAHI, CAI and MAI with
respect to severity and body position, it is feasible that obstructive,
mixed and central apnoeas are of common origin, manifesting
as this triform phenotype of sleep apnoea. is hypothesis is in
keeping with that of Issa and Sullivan, who suggested that in a
specic subgroup of sleep apnoea, where subjects presented with
a combination of obstructive, mixed and central apnoeas, the
occurrence of dierent types of sleep apnoea (that appear to be
body position dependent) are probably dierent representations of
a single respiratory abnormality. While this study was not restricted
to heart-failure patients and those with Cheyne-Stokes respiration
occurred only in a minority of subjects, both the MAI and CAI were
signicantly lower in the lateral v. supine body position, respectively,
which supports previous ndings.[9,16] Szollosi et al. also showed that
in heart-failure patients, the CAI (central apnoeas associated with
Cheyne-Stokes breathing) was more than 50% less severe in the lateral
compared to the supine body position. We noted that in participants
presenting with the triform phenotype of sleep apnoea, a change in
body position (from supine to lateral) resulted in a reduction of ~50%
for the CAI and ~47% for the MAI.
Other studies have noted that those suering from Cheyne-Stokes
breathing and congestive heart failure prefer sleeping in the right body
position, [9,10,16] however only two subjects in this study presented with
Cheyne-Stokes breathing, therefore the nding that most subjects
preferred the le sleeping position was interesting. Subjects spent
more sleep time in the lateral body position as well, supported by
reports by Szollosi et al. In contrast, Eiseman et al. found that
more than half of the sleep time recorded within a sleep laboratory
environment is expected to be in the supine body position. is may
be a result of major dierences in subject selection between studies.
Both Szollosi et al. and the present study included subjects where
central apnoeas were prominent PSG features (median 124 events/
hr), whereas Eiseman et al. specically studied obstructive apnoeas
and central apnoeas were negligible in their study (median 2 events/
hr). is raises the question of whether subjects with PSG features
that include a signicant number of central apnoeas, compensate for
their disturbed breathing by favouring a lateral body position. Further
studies in this regard are warranted.
e lateral sleeping position is associated with a decrease in the severity
of obstructive-mediated (OAHI) and centrally mediated (CAI)
apnoeas in subjects who present with a combination of obstructive,
mixed and central sleep apnoeas. Furthermore, we propose that in this
specic sleep apnoea phenotype, mixed apnoeas are neither purely
obstructive nor purely central-mediated in origin. Obstructive, mixed
and central sleep apnoea may be dierent representations of a single
respiratory abnormality in patients presenting with this triform of
sleep apnoea. Further studies on the eectiveness of positional therapy
as a potential treatment option in this phenotype may be of value.
Author contributions. GC co-wrote the manuscript and collected the
data. PRB co-wrote the manuscript. PRB and LH assisted with the design
and planning of the study. PB performed the statistical analysis and
also assisted with the planning of the study. PRB was the project head-
supervisor and LH co-supervised the study.
Funding. First author, GC, funded the study.
Conicts of interest. None.
1. Spriggs, WH. Essentials of polysomnography: A training guide and reference for sleep
technicians. 2nd ed. Burlington: Jones & Bartlett Learning, 2015(7-24):199-213.
2. White LH, Bradley TD. Role of nocturnal rostral uid shi in the pathogenesis of
obstructive and central sleep apnoea. J Physiol 2013; 591(5):1179-1193. https://doi.
3. Berry RB, Wagner MH. Sleep Medicine Pearls. 3rd ed. Philadelphia: Elsevier/
4. Yamauchi M, Tamaki S, Yoshikawa M, et al. Dierences in breathing patterning during
wakefulness in patients with mixed apnea-dominant vs obstructive-dominant sleep
apnea. Chest 2011;140(1):54-61. https://doi.org/10.1378/chest.10-1082
5. Morgenthaler T, Gay PC, Gordon N, et al. Adaptive servoventilation versus
noninvasive positive pressure ventilation for central, mixed, and complex sleep apnea
syndromes. Sleep 2007;30(4):468-475. https://doi.org/10.1093/sleep/30.4.468
6. Eijsvogel MM, Ubbink R, Dekker J, et al. Sleep position trainer versus tennis ball
technique in positional obstructive sleep apnea syndrome. J Clin Sleep Med 2015;
7. Zaharna M, Rama A, Chan R, et al. A case of positional central sleep apnea. J Clin
Sleep Med 2013;9(3):265-268. https://doi.org/10.5664/jcsm.2496
8. Berry RB, Brooks R, Gamal do CE, et al. e AASM Manual for the Scoring of Sleep
and Associated Events: Rules, Terminology and Technical Specications, Version 2.2.
Illinois: American Academy of Sleep Medicine, 2015.
9. Szollosi I, Roebuck T, ompson B, et al. Lateral sleeping position reduces severity of
central sleep apnea/Cheyne-Stokes respiration. Sleep 2006;29(8):1045-1051. https://
10. Leung RST, Bowman ME, Parker JD, et al. Avoidance of the le lateral decubitus
position during sleep in patients with heart failure: Relationship to cardiac size and
function. JACC 2003;41(2):227-230. https://doi.org/10.1016/S0735-1097(02)02717-1
11. Oksenberg A, Khamaysi I, Silverberg DS, et al. Association of body position with
severity of apneic events in patients with severe nonpositional obstructive sleep apnea.
Chest 2000;118(4):1018-1024. https://doi.org/10.1378/chest.118.4.1018
12. De Vries GE, Hoekema A, Do MHJ, et al. Usage of positional therapy in adults with
obstructive sleep apnea. J Clin Sleep Med 2015;11(2):131-137. https://doi.org/10.5664/
13. Issa FG, Sullivan CE. Reversal of central sleep apnea using nasal CPAP. Chest
14. Xie A, Bedekar A, Skatrud JB, et al. e heterogeneity of obstructive sleep apnea
(predominant obstructive vs pure obstructive apnea). Sleep 2011;34(6):745-750.
15. Chou YT, Yang TM, Lin CK, et al. Pay attention to treating a subgroup of positional
obstructive sleep apnea patients. J Formos Med Assoc 2017;116(5):359- 365. https://
16. Sahlin C, Svanborg E, Stenlund H, et al. Cheyne-Stokes respiration and supine
dependency. Eur Respir J 2005;25(5):829-833. https://doi.org/10.1183/09031936.05
17. Eiseman NA, Westover MB, Ellenbogen JM, et al. e impact of body posture and
sleep stages on sleep apnea severity in adults. J Clin Sleep Med 2012; 8(6):655-666.
Accepted 15 October 2019.