OriginalInvestigation ... - obu.ch

Copyright 2015 American Medical Association. All rights reserved.

CPAP vs Mandibular Advancement Devices and Blood Pressurein Patients With Obstructive Sleep ApneaA Systematic Review and Meta-analysisDaniel J. Bratton, PhD; Thomas Gaisl, MD; Annette M. Wons, MD; Malcolm Kohler, MD

IMPORTANCE Obstructive sleep apnea is associated with higher levels of blood pressure (BP),which can lead to increased cardiovascular risk.

OBJECTIVE To compare the association of continuous positive airway pressure (CPAP),mandibular advancement devices (MADs), and inactive control groups (placebo or notreatment) with changes in systolic BP (SBP) and diastolic BP (DBP) in patients withobstructive sleep apnea.

DATA SOURCES The databases of MEDLINE, EMBASE, and the Cochrane Library weresearched up to the end of August 2015 and study bibliographies were reviewed.

STUDY SELECTION Randomized clinical trials comparing the effect of CPAP or MADs (vs eachother or an inactive control) on BP in patients with obstructive sleep apnea were selected byconsensus. Of 872 studies initially identified, 51 were selected for analysis.

DATA EXTRACTION AND SYNTHESIS Data were extracted by one reviewer and checked byanother reviewer. A network meta-analysis using multivariate random-effectsmeta-regression was used to estimate pooled differences between each intervention.Meta-regression was used to assess the association between trial characteristics and thereported effects of CPAP vs inactive control.

MAIN OUTCOMES AND MEASURES Absolute change in SBP and DBP from baseline to follow-up.

RESULTS Of the 51 studies included in the analysis (4888 patients), 44 compared CPAP with aninactive control, 3 compared MADs with an inactive control, 1 compared CPAP with an MAD,and 3 compared CPAP, MADs, and an inactive control. Compared with an inactive control, CPAPwas associated with a reduction in SBP of 2.5 mm Hg (95% CI, 1.5 to 3.5 mm Hg; P < .001) and inDBP of 2.0 mm Hg (95% CI, 1.3 to 2.7 mm Hg; P < .001). A 1-hour-per-night increase in meanCPAP use was associated with an additional reduction in SBP of 1.5 mm Hg (95% CI, 0.8 to2.3 mm Hg; P < .001) and an additional reduction in DBP of 0.9 mm Hg (95% CI, 0.3 to 1.4 mmHg; P = .001). Compared with an inactive control, MADs were associated with a reduction inSBP of 2.1 mm Hg (95% CI, 0.8 to 3.4 mm Hg; P = .002) and in DBP of 1.9 mm Hg (95% CI, 0.5 to3.2 mm Hg; P = .008). There was no significant difference between CPAP and MADs in theirassociation with change in SBP (−0.5 mm Hg [95% CI, −2.0 to 1.0 mm Hg]; P = .55) or in DBP(−0.2 mm Hg [95% CI, −1.6 to 1.3 mm Hg]; P = .82).

CONCLUSIONS AND RELEVANCE Among patients with obstructive sleep apnea, both CPAPand MADs were associated with reductions in BP. Network meta-analysis did not identifya statistically significant difference between the BP outcomes associated with these therapies.

JAMA. 2015;314(21):2280-2293. doi:10.1001/jama.2015.16303

Supplemental content atjama.com

Author Affiliations: Department ofPulmonology, University HospitalZurich, Zurich, Switzerland (Bratton,Gaisl, Wons, Kohler); Zurich Centerfor Integrative Human Physiology,University of Zurich, Zurich,Switzerland (Kohler); Center forInterdisciplinary Sleep Research,University of Zurich, Zurich,Switzerland (Kohler).

Corresponding Author: MalcolmKohler, MD, University HospitalZürich, Rämistrasse 100, 8091 Zürich,Switzerland ([email protected]).

Research

Original Investigation

2280 (Reprinted) jama.com


Downloaded From: http://jama.jamanetwork.com/ by a UZH Hauptbibliothek / Zentralbibliothek Zuerich User on 12/02/2015

http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2015.16303&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2015.16303


http://www.

mailto:[email protected]

http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2015.16303


O bstructive sleep apnea is characterized by recurringcessations or reductions in respiratory flow due to up-per airway collapse during sleep. The estimated preva-

lence of symptomatic obstructive sleep apnea in Western coun-tries is 2% to 4%; however, prevalence is increasing due to levelsof obesity in these populations.1 The condition is associatedwith oxygen desaturations and arousals from sleep, which canlead to increases in blood pressure (BP) and risk of cardiovas-cular disease.2

Continuous positive airway pressure (CPAP) has beenshown to be an effective treatment for improving symptomsof obstructive sleep apnea, such as daytime sleepiness,3 andmeta-analyses have shown it to be associated with a reduc-tion of about 2 mm Hg in BP.4-8 However, the estimated asso-ciation of trial-level characteristics (mean nightly CPAP use inparticular) with the effects of CPAP reported in individual ran-domized clinical trials (RCTs) has been less consistent.5,6

Alternative treatments often used by patients unable to tol-erate CPAP are mandibular advancement devices (MADs),which work by protruding the mandible and tongue to keepairways open during sleep.9 The association of use of MADswith reductions in BP is less clear. A recent meta-analysis thatincluded only 2 RCTs was inconclusive.10 To our knowledge,no meta-analysis has comprehensively compared CPAP vsMADs on change in BP, perhaps due to a lack of RCTs. The mostrecent meta-analysis5 briefly reviewed 2 trials,11,12 comparingthe effects of CPAP with MADs on change in BP, and found con-flicting results between the studies.

The primary aim of our study was to perform a networkmeta-analysis13 comparing the association of CPAP vs MADsand vs an inactive control (eg, placebo or no treatment) withchanges in systolic BP (SBP) and diastolic BP (DBP) in patientswith obstructive sleep apnea. A secondary aim was to ex-plore the association of trial-level characteristics, such as meannightly CPAP use, with the reported treatment effects of CPAPvs inactive control therapy on BP outcomes.

MethodsInclusion CriteriaThe studies must have randomized participants aged 18 yearsor older with a diagnosis of obstructive sleep apnea (definedby an apnea-hypopnea index of ≥5/h) to at least 2 of the fol-lowing treatments: (1) CPAP, (2) MADs, or (3) inactive control,such as sham CPAP, placebo MADs, or conservative treat-ment (no active therapy). Trials must also have measured andreported data on SBP or DBP at a follow-up visit and prefer-ably also at enrollment or randomization, or reported a treat-ment effect for either outcome. If 2 eligible studies containeda significant overlap in patients, the larger of the 2 studies wasused in the analysis. The protocol for this meta-analysis ap-pears in Supplement 1.

Identification of TrialsLiterature searches were performed independently by 2 ofthe authors (D.J.B. and A.M.W.) using the databases ofMEDLINE, EMBASE, and the Cochrane Library from incep-

tion to the end of August 2015. The RCTs were identifiedusing the Cochrane Collaboration highly sensitive searchstrategy (sensitivity-maximizing and precision-maximizingversion).14 The general electronic search strategy appears inSupplement 2. The bibliographies of all eligible trials andreview articles were also screened for relevant trials thatmight have been missed in the database search. Inclusionwas restricted to trials reported in English.

Selection of Studies and Data ExtractionTwo of the authors (D.J.B. and A.M.W.) assessed the eligibil-ity of studies found in the literature searches. One author(D.J.B.) extracted the relevant data from eligible studies, whichwas then independently checked by another author (T.G.). Trialcharacteristics, such as sample size, length of follow-up, typeof control group (eg, placebo or no treatment), and type of study(eg, crossover or parallel), were recorded. The main outcomeof interest was the change in SBP and DBP between baselineand follow-up in the 3 treatment groups (CPAP, MADs, inac-tive control). Treatment effects were extracted directly fromthe studies along with standard errors, 95% confidence inter-vals, or P values. If treatment effects were not reported, otherdata, such as mean (standard deviation) for SBP and DBP foreach treatment group at each visit or the change in SBP andDBP between visits in each group, were recorded and used toestimate the treatment effect of interest.

Measurements of BP during the daytime (while the pa-tient was ambulatory), during the morning, or during an of-fice visit were preferable (in that order). Otherwise, 24-hourambulatory BP measurements were used. Summary statis-tics for the following baseline data were also recorded: age,body mass index (calculated as weight in kilograms dividedby height in meters squared), apnea-hypopnea index, oxy-gen desaturation index, Epworth Sleepiness Scale score, andproportion of male participants. We also extracted the meannightly use of CPAP from trials comparing CPAP with an inac-tive control.

End PointsThe primary outcomes were the absolute changes in SBP andDBP from baseline to follow-up in each of the following treat-ment comparisons: (1) CPAP vs inactive control, (2) MADs vsinactive control, and (3) CPAP vs MADs. For the first compari-son (CPAP vs inactive control), we also investigated the asso-ciation of mean CPAP use with the treatment effects reportedin each trial for both outcomes of SBP and DBP. We also ex-plored the association between the reported treatment ef-fects and mean baseline apnea-hypopnea index, oxygen-desaturation index, baseline BP, length of follow-up, type ofcontrol group (sham CPAP, no treatment, or other placebo), andtype of BP measurement (daytime, morning, office, or 24 hour).Similar investigations for the second and third comparisonswere not made due to the insufficient number of studies di-rectly comparing these treatments.

Risk of Bias AssessmentTwo authors (D.J.B. and T.G.) evaluated the risk of bias ineach study included in the meta-analysis using the Cochrane

Treatment for Obstructive Sleep Apnea Original Investigation Research

jama.com (Reprinted) JAMA December 1, 2015 Volume 314, Number 21 2281







Collaboration tool for assessing risk of bias.15 This toolassesses studies on different sources of bias (selection, detec-tion, performance, attrition, or reporting bias) and catego-rizes studies by low, unclear, or high risk of bias in eachdomain. We then planned to compare the reported treatmenteffects on SBP and DBP in studies at low and high risk of biasin each domain using meta-regression.

Statistical MethodsIf treatment effects or changes in SBP and DBP in each treat-ment group were not reported in studies, then the mean (stan-dard deviation) values for each outcome, in each group, andat each visit were extracted and used to estimate treatment ef-fects. To calculate standard error, an estimate of the correla-tion between SBP and DBP measurements at baseline andfollow-up was required. Previous meta-analyses6,7 have as-sumed this correlation to be 0.5, which might not be appro-priate. We instead estimated the correlation in all studies forwhich it was possible (ie, those reporting standard deviationsor standard errors of treatment effects or changes during fol-low-up and the standard deviations at each visit) and used themean correlation to impute the treatment effect standard er-ror in studies for which estimation was not possible. To as-sess the sensitivity of our results to this correlation, we re-peated the meta-analysis using the minimum and maximumcorrelations estimated from the studies. In crossover studiesnot reporting treatment effects from paired t tests (or tests ac-counting for the between-period correlation), the between-period correlation was assumed to be zero, slightly increas-

ing the conservatism of the analysis.16 In studies not reportingdata on baseline BP, conservative estimates of treatment ef-fects were obtained by estimating the differences betweentreatment groups in follow-up BP measurements.

Separate meta-analyses of direct evidence only (pairwisemeta-analyses) were conducted for each of the 3 treatmentcomparisons listed above using the metan command in Stataversion 14.0 (StataCorp).17 Heterogeneity was assessed usingthe estimated between-study variance (τ2), Cochran χ2 test,and the I2 statistic.18 Only random-effects models were usedto be consistent with the network meta-analysis. To assessthe association of trial characteristics and risk of bias with thereported effects of CPAP compared with an inactive control,we used random-effects meta-regression (metareg commandin Stata19), and only included studies directly comparingthese 2 interventions. To be consistent with the pairwiserandom-effects meta-analyses, we performed each meta-regression without the use of the adjustment method byKnapp and Hartung,20 but also applied the adjustment in asensitivity analysis.

Because there are relatively few trials directly comparingCPAP vs MADs on change in BP, a meta-analysis of only directevidence is likely to lack power. To strengthen this and all othertreatment comparisons, we performed a network meta-analysis. Unlike traditional meta-analyses, this method has theadvantage of allowing trials comparing CPAP or MADs withsome other common treatment (eg, placebo) to be incorpo-rated into the analysis, thus increasing power and enabling abetter comparison of CPAP and MADs to be made.13

We used multivariate, random-effects meta-regressionsto perform each analysis using the network family of com-mands in Stata.21 We first fitted a consistency model, whichassumes that treatment effects from direct and indirect com-parisons are in agreement. An unstructured between-studycovariance matrix was used to allow for the possibility ofunequal levels of heterogeneity in each comparison. We alsoperformed a sensitivity analysis for which heterogeneity wasassumed to be the same in each comparison. To test forinconsistency, design × treatment interactions were addedto the consistency model, in which design refers only to theset of treatments in a trial (4 sets in total). To further investi-gate the plausibility of the consistency assumption, we alsochecked whether potential effect modifiers were similaracross different designs.22 If inconsistency was not rejected,we estimated the probability of each treatment having thestrongest association with BP reduction by applying theparametric bootstrap procedure (with 5000 draws), whichwas described by White.23 Forest plots were used to summa-rize study level and pooled treatment comparisons andcomparison-adjusted funnel plots were used to assess publi-cation bias.24 All analyses were conducted at the 2-sided sig-nificance level of .05.

ResultsA total of 51 eligible studies (4888 patients) were identified andincluded in the network meta-analysis (Figure 1). Of these 51

Figure 1. Flowchart of Literature Search and Study Selection

357 Records excluded due to duplication

65 Articles selected for full-text review

51 Randomized trials included in meta-analysis

450 Records excluded

100 Review articles

97 Conference abstracts correspondingto other trial reports

96 Not randomized clinical trials inpatients with obstructive sleep apnea

72 Assessed other treatment comparisons

39 Did not assess blood pressure orreported insufficient data

23 Not reported in English

20 No data on adults

3 No data on humans

14 Excluded due to being substudies

872 Records identified

867 Through database searches

5 Through other sources

444 EMBASE

263 Cochrane Library

160 MEDLINE

515 Records screened

Research Original Investigation Treatment for Obstructive Sleep Apnea

2282 JAMA December 1, 2015 Volume 314, Number 21 (Reprinted) jama.com





Tabl

e1.

Base

line

Char

acte

ristic

sofI

nclu

ded

Tria

lsof

Cont

inuo

usPo

sitiv

eAi

rway

Pres

sure

(CPA

P)vs

Inac

tive

Cont

rol

Sour

ceM

ain

Incl

usio

nCr

iteria

Type

ofDe

sign

Type

ofCo

ntro

lTh

erap

ySa

mpl

eSi

zeLe

ngth

ofFo

llow

-up,

wk

Mea

nAg

e,y

Mal

eSe

x,%

Mea

nBM

IaM

ean

AHI,

/hES

SSc

oreb

Tim

eor

Loca

tion

ofBP

Mea

sure

men

t

Mea

nBa

selin

eBP

,mm

Hg

Syst

olic

Dias

tolic

Ari

aset

al,2

52

00

5A

HI

≥1

0/h

and

ES

Ssc

ore

≥1

0C

ross

ove

rS

ham

CP

AP

27

12

52

10

03

14

4N

AD

ayti

me

12

77

9

Bar

bé

etal

,26

20

01

AH

I≥

30

/han

dE

SS

sco

re≤

10

Par

alle

lS

ham

CP

AP

55

65

39

12

95

57

Day

tim

e1

29

81

Bar

bé

etal

,27

20

12

AH

I≥

20

/han

dE

SS

sco

re≤

10

Par

alle

lN

op

lace

bo

72

52

65

28

63

13

87

Off

ice

13

18

0

Bar

nes

etal

,28

20

02

AH

I5

-30

/hC

ross

ove

rO

ral

pla

ceb

o4

28

46

86

31

13

11

Day

tim

e1

32

84

Bec

ker

etal

,29

20

03

AH

I≥

5/h

and

ES

Ssc

ore

≥1

0P

aral

lel

Sh

amC

PA

P6

09

53

91

33

64

14

Day

tim

e1

41

86

Cam

po

s-R

od

rig

uez

etal

,30

20

06

Sys

tem

icar

teri

alh

yper

ten

sio

nan

dA

HI

≥1

0/h

Par

alle

lS

ham

CP

AP

72

45

76

03

55

91

42

4h

13

27

8

Co

mo

nd

ore

etal

,31

20

09

Min

imal

day

tim

esl

eep

ines

san

dA

HI

≥1

5/h

Cro

sso

ver

No

pla

ceb

o1

34

56

69

31

28

72

4h

13

88

4

Co

ug

hli

net

al,3

2

20

07

AH

I>

15

/han

dE

SS

sco

re≥

10

Cro

sso

ver

Sh

amC

PA

P3

56

49

10

03

64

0c

14

Mo

rnin

gN

AN

A

Cra

iget

al,3

32

01

2In

suff

icie

nt

day

tim

esy

mp

tom

sfo

rC

PA

Pan

dO

DI

>7

.5d

ips/

hP

aral

lel

No

pla

ceb

o3

91

26

58

78

32

13

d8

Day

tim

e1

30

81

Cro

sset

al,3

42

00

8A

HI

>1

5/h

and

ES

Ssc

ore

>1

0C

ross

ove

rS

ham

CP

AP

29

64

89

63

76

3N

AO

ffic

e1

43

80

de

Oli

veir

aet

al,3

5

20

14

Res

ista

nt

hyp

erte

nsi

on

and

AH

I>

15

/hP

aral

lel

Sh

amC

PA

P4

78

59

57

30

20

10

Day

tim

e1

50

90

Dra

ger

etal

,36

20

07

Mal

ese

xan

dA

HI

>3

0/h

Par

alle

lN

op

lace

bo

24

17

46

10

03

05

91

4D

ayti

me

12

37

9

Dra

ger

etal

,37

20

11

Pre

hyp

erte

nsi

on

or

mas

ked

hyp

erte

nsi

on

and

AH

I>

30

/hP

aral

lel

No

pla

ceb

o3

61

34

31

00

29

56

12

Day

tim

e1

27

83

Du

rán

-Can

toll

aet

al,3

82

01

0H

yper

ten

sio

nan

dA

HI

>1

5/h

Par

alle

lS

ham

CP

AP

34

01

25

28

13

24

41

0D

ayti

me

13

48

6

Eg

eaet

al,3

92

00

8C

hro

nic

hea

rtfa

ilu

re,

AH

I>

10

/h,

and

LVE

F≤

45

%P

aral

lel

Sh

amC

PA

P7

31

36

39

33

14

28

NA

12

47

6

En

gle

man

etal

,40

19

96

AH

I≥

5/h

and

≥2

OS

Asy

mp

tom

sC

ross

ove

rO

ral

pla

ceb

o1

33

51

85

36

49

NA

Day

tim

eN

AN

A

Facc

end

aet

al,4

1

20

01

Tw

om

ajo

rsy

mp

tom

so

fO

SA

and

AH

I≥

15

/hC

ross

ove

rO

ral

pla

ceb

o7

14

50

81

30

35

15

24

hN

AN

A

Go

ttli

ebet

al,4

2

20

14

Ber

lin

qu

esti

on

nai

resc

ore

of

2-3

,ch

ron

ich

eart

dis

ease

or

mu

ltip

leca

rdio

vasc

ula

rri

skfa

cto

rs,

and

AH

I≥

15

/h

Par

alle

lN

op

lace

bo

21

21

26

37

73

32

58

Day

tim

e1

28

73

Hal

let

al,4

32

01

4D

ysp

nea

,A

HI

>1

0/h

,an

dLV

EF

≤4

5%

Par

alle

lN

op

lace

bo

45

76

27

6N

A2

71

0O

ffic

e1

18

NA

Ho

yos

etal

,44

20

12

Mal

ese

x,

AH

I≥

20

/h,

and

OD

I>

15

dip

s/h

Par

alle

lS

ham

CP

AP

65

12

49

10

03

14

01

0O

ffic

eN

AN

A

Ho

yos

etal

,45

20

15

AH

I≥

25

/han

dO

DI

≥2

0d

ips/

hC

ross

ove

rS

ham

CP

AP

38

84

98

73

14

01

1M

orn

ing

12

67

7

Hu

ang

etal

,46

20

15

Hyp

erte

nsi

on

,ch

ron

ich

eart

dis

ease

,an

dA

HI

≥1

5/h

Par

alle

lN

op

lace

bo

83

15

76

28

22

82

99

Mo

rnin

g1

47

83

Hu

iet

al,4

72

00

6O

SA

-rel

ated

sym

pto

ms

and

AH

I≥

5/h

Par

alle

lS

ham

CP

AP

56

13

51

77

27

31

11

Day

tim

e1

28

84

Ipet

al,4

82

00

4M

ale

sex

and

AH

I≥

15

/hP

aral

lel

No

pla

ceb

o2

84

43

10

02

94

61

1M

orn

ing

12

37

6 (con

tinue

d)







Tabl

e1.

Base

line

Char

acte

ristic

sofI

nclu

ded

Tria

lsof

Cont

inuo

usPo

sitiv

eAi

rway

Pres

sure

(CPA

P)vs

Inac

tive

Cont

rol(

cont

inue

d)

Sour

ceM

ain

Incl

usio

nCr

iteria

Type

ofDe

sign

Type

ofCo

ntro

lTh

erap

ySa

mpl

eSi

zeLe

ngth

ofFo

llow

-up,

wk

Mea

nAg

e,y

Mal

eSe

x,%

Mea

nBM

IaM

ean

AHI,

/hES

SSc

oreb

Tim

eor

Loca

tion

ofBP

Mea

sure

men

t

Mea

nBa

selin

eBP

,mm

Hg

Syst

olic

Dias

tolic

Jon

eset

al,4

92

01

3A

HI

≥1

5/h

Cro

sso

ver

Sh

amC

PA

P5

31

24

65

83

03

11

3O

ffic

e1

28

76

Ko

hle

ret

al,5

02

01

1C

PA

Pu

se>

4h

/nig

ht

du

rin

gp

ast

year

and

OD

I>

10

dip

s/h

off

CP

AP

Par

alle

lS

ham

CP

AP

41

26

39

83

34

11

5M

orn

ing

13

18

2

Lam

etal

,51

20

10

Mal

ese

xan

dA

HI

≥1

5/h

Par

alle

lS

ham

CP

AP

61

14

61

00

28

32

11

Mo

rnin

g1

30

81

Lit

vin

etal

,52

20

13

Hyp

erte

nsi

on

and

AH

I>

30

/hC

ross

ove

rS

ham

CP

AP

44

35

67

73

86

3N

AO

ffic

e1

41

85

Lo

zan

oet

al,5

32

01

0R

esis

tan

th

yper

ten

sio

nan

dA

HI

≥1

5/h

Par

alle

lN

op

lace

bo

75

13

59

69

31

53

6D

ayti

me

13

37

8

Mar

tín

ez-G

arcí

aet

al,5

42

01

3R

esis

tan

th

yper

ten

sio

nan

dA

HI

≥1

5/h

Par

alle

lN

op

lace

bo

19

41

25

66

93

44

09

Day

tim

e1

46

85

McM

illa

net

al,5

5

20

14

Ag

e≥

65

y,E

SS

sco

re≥

9,

and

OD

I>

7.5

dip

s/h

Par

alle

lN

op

lace

bo

27

85

27

18

23

42

9d

12

Off

ice

14

07

8

Mo

nas

teri

oet

al,5

6

20

01

Ab

sen

ceo

fse

vere

day

tim

esl

eep

ines

san

dA

HI

10

-30

/hP

aral

lel

No

pla

ceb

o1

42

26

54

86

29

20

13

Off

ice

12

98

2

Mu

xfel

dt

etal

,57

20

15

Res

ista

nt

hyp

erte

nsi

on

and

AH

I≥

15

/hP

aral

lel

No

pla

ceb

o1

25

26

61

40

33

41

11

Day

tim

e1

31

77

Ng

uye

net

al,5

82

01

0E

SS

sco

re>

10

and

RD

I≥

15

Par

alle

lS

ham

CP

AP

20

13

53

90

30

35

NA

Off

ice

12

47

7

No

da

etal

,59

20

07

No

tta

kin

gan

tih

yper

ten

sive

med

icat

ion

and

AH

I≥

20

/hP

aral

lel

No

pla

ceb

o4

01

35

31

00

NA

47

NA

Off

ice

14

48

9

No

rman

etal

,60

20

06

AH

I>

15

/hP

aral

lel

Sh

amC

PA

P3

32

50

85

31

61

12

Day

tim

e1

29

78

Pam

idie

tal

,61

20

15

Pre

dia

bet

esan

dA

HI

≥5

/hP

aral

lel

Ora

lp

lace

bo

39

25

46

73

63

61

0D

ayti

me

14

08

3

Ped

rosa

etal

,62

20

13

Res

ista

nt

hyp

erte

nsi

on

and

AH

I≥

15

/hP

aral

lel

No

pla

ceb

o4

02

65

67

73

22

91

0D

ayti

me

14

78

7

Pep

per

ell

etal

,63

20

02

Mal

ese

x,

ES

Ssc

ore

>9

,an

dO

DI

>1

0d

ips/

hP

aral

lel

Sh

amC

PA

P1

18

45

11

00

35

37

d1

62

4h

13

48

5

Ro

bin

son

etal

,64

20

06

Hyp

erte

nsi

on

,E

SS

sco

re≤

10

,an

dO

DI

>1

0d

ips/

hC

ross

ove

rS

ham

CP

AP

35

45

48

93

32

8d

52

4h

14

28

6

Ro

ssie

tal

,65

20

13

CP

AP

use

>4

h/n

igh

td

uri

ng

pas

tye

aran

dO

DI

>1

0d

ips/

hP

aral

lel

Pro

ven

tp

lace

bo

45

26

27

83

43

91

4M

orn

ing

13

18

3

Ru

ttan

aum

paw

anet

al,6

62

00

8H

eart

fail

ure

,A

HI

≥2

0/h

,an

dLV

EF

≤4

5%

Par

alle

lN

op

lace

bo

33

46

09

13

14

3N

AM

orn

ing

12

66

6

Taka

esu

etal

,67

20

12

Pan

icd

iso

rder

and

AH

I>

20

/hC

ross

ove

rS

ham

CP

AP

12

44

19

22

64

1N

AO

ffic

e1

42

85

Wea

ver

etal

,68

20

12

AH

I5

-30

/han

dE

SS

sco

re>

10

Par

alle

lS

ham

CP

AP

28

18

51

59

34

13

15

Day

tim

e1

24

76

Abbr

evia

tions

:AH

I,ap

nea

hypo

pnea

inde

x;BP

,blo

odpr

essu

re;L

VEF,

left

vent

ricul

arej

ectio

nfr

actio

n;N

A,no

tav

aila

ble

orap

plic

able

;OD

I,ox

ygen

desa

tura

tion

inde

x;O

SA,o

bstr

uctiv

esle

epap

nea;

RDI,

resp

irato

rydi

stur

banc

ein

dex.

aBo

dym

assi

ndex

(BM

I)ca

lcul

ated

asw

eigh

tin

kilo

gram

sdiv

ided

byhe

ight

inm

eter

ssqu

ared

.b

The

Epw

orth

Slee

pine

ssSc

ale

(ESS

)has

asc

ore

rang

eof

0to

24(h

ighe

rsco

resi

ndic

ate

grea

terd

aytim

esle

eppr

open

sity)

.

cM

ean

RDIw

asus

edbe

caus

em

ean

AHIa

ndO

DIw

ere

unav

aila

ble.

dM

ean

OD

Iwas

used

beca

use

mea

nAH

Iwas

unav

aila

ble.







studies, 44 compared CPAP with an inactive control (4289patients),25-68 3 compared MADs with an inactive control(229 patients),69-71 1 compared CPAP with MADs (126patients),72 and 3 compared CPAP, MADs, and an inactivecontrol (244 patients)11,12,73 (eFigure 1 in Supplement 2). Allincluded studies assessed SBP as an outcome. One study43 didnot report DBP data. Summaries of the characteristics of theincluded trials appear in Table 1 and Table 2. There wassome evidence of a difference in the mean body mass indexbetween trials of different treatment comparisons (Table 3);however, the differences between the weighted means weresmall. In particular, mean baseline BP was similar across com-parisons. Thus, the consistency assumption in the networkmeta-analysis was not majorly violated in this regard.

Primary AnalysesThe mean between-visit correlation across the 16 trials forwhich it could be estimated was 0.69 (minimum, 0.47 andmaximum, 0.89) for SBP and 0.74 (minimum, 0.53 andmaximum, 0.85) for DBP with the mean values being usedto impute treatment effect standard errors in 25 studies. Noinconsistency was found in the network meta-analysis ofSBP (χ2

3 = 0.54, P = .91) or DBP (χ23 = 2.25, P = .52) and so the

main conclusions were drawn from the consistency model.Although these tests for inconsistency are likely to beunderpowered due to the small number of studies forsome treatment comparisons, the relatively small values ofthe χ2 statistics indicate that this is unlikely to be an issue inthis case.

In the network meta-analysis and compared with aninactive control, CPAP was associated with a reduction inSBP of 2.5 mm Hg (95% CI, 1.5 to 3.5 mm Hg; P < .001) andMADs were associated with a reduction in SBP of 2.1 mm Hg(95% CI, 0.8 to 3.4 mm Hg; P = .002) (Table 4). In the net-work meta-analysis and compared with an inactive control,CPAP was associated with a reduction in DBP of 2.0 mm Hg(95% CI, 1.3 to 2.7 mm Hg; P < .001) and MADs were associ-ated with a reduction in DBP of 1.9 mm Hg (95% CI, 0.5 to3.2 mm Hg; P = .008). There was no significant differencebetween CPAP and MADs in their association with change inSBP (−0.5 mm Hg [95% CI, −2.0 to 1.0 mm Hg]; P = .55) or inDBP (−0.2 mm Hg [95% CI, −1.6 to 1.3 mm Hg]; P = .82).

The findings of the pairwise meta-analyses (Table 4) weresimilar to those of the network meta-analysis except for thecomparison of MADs with inactive controls for reductions inDBP for which a smaller difference of −1.1 mm Hg (95% CI, −2.4to 0.2 mm Hg; P = .11) was estimated. The results of each pair-wise and network meta-analysis for SBP appear in Figure 2 andFigure 3 and for DBP in Figure 4 and Figure 5.

A sensitivity analysis using the minimum and maxi-mum between-visit correlation estimates (instead of themean estimate) to impute missing treatment effect standarderrors did not greatly affect the results of the network meta-analysis (eTables 1 and 2 in Supplement 2). Another sensitiv-ity analysis in which the level of heterogeneity was assumedto be the same across comparisons only led to a slightlysmaller estimated difference between MADs and inactivecontrols (eTable 3 in Supplement 2). This was likely due to Ta

ble

2.Ba

selin

eCh

arac

teris

ticso

fInc

lude

dTr

ials

ofM

andi

bula

rAdv

ance

men

tDev

ices

(MAD

s)vs

Cont

inuo

usPo

sitiv

eAi

rway

Pres

sure

(CPA

P)vs

Inac

tive

Cont

rol

Sour

ceM

ain

Incl

usio

nCr

iteria

Type

ofDe

sign

Type

ofCo

ntro

lTh

erap

ySa

mpl

eSi

zeLe

ngth

ofFo

llow

-up,

wk

Mea

nAg

e,y

Mal

eSe

x,%

Mea

nBM

IaM

ean

AHI,

/hES

SSc

oreb

Tim

eor

Loca

tion

ofBP

Mea

sure

men

t

Mea

nBa

selin

eBP

,mm

Hg

Syst

olic

Dias

tolic

MAD

svsI

nact

ive

Cont

rol

An

dré

net

al,6

92

01

3H

yper

ten

sio

nan

dA

HI

≥1

0/h

Par

alle

lP

lace

bo

MA

D7

21

35

87

93

02

41

1D

ayti

me

14

58

9

Go

tso

po

ulo

set

al,7

0

20

04

Man

dib

ula

rp

rotr

usi

on

≥3

mm

,A

HI

≥1

0/h

,an

dO

SA

sym

pto

ms

Cro

sso

ver

Pla

ceb

oM

AD

67

44

87

92

92

7N

AD

ayti

me

13

28

1

Qu

inn

elle

tal

,71

20

14

AH

I5

-30

/han

dE

SS

sco

re≥

9C

ross

ove

rP

lace

bo

MA

D9

06

51

80

31

14

12

Off

ice

13

08

0

CPAP

vsM

ADs

Ph

illi

ps

etal

,72

20

13

AH

I>

10

/han

dO

SA

sym

pto

ms

Cro

sso

ver

NA

12

64

50

81

30

26

9M

orn

ing

12

98

2

CPAP

vsM

ADsv

sIna

ctiv

eCo

ntro

l

Bar

nes

etal

,11

20

04

AH

I5

-30

Cro

sso

ver

Ora

lpla

ceb

o1

04

13

46

79

31

22

11

24

h1

27

76

Dal

-Fab

bro

etal

,73

20

14

Man

dib

ula

rp

rotr

usi

on

≥7

mm

,A

HI

≥2

0/h

,an

dB

MI

≤3

5C

ross

ove

rP

lace

bo

MA

D3

94

47

83

28

42

NA

Day

tim

e1

33

83

Lam

etal

,12

20

07

AH

I5

-40

/han

dE

SS

sco

re>

9if

AH

I5

-20

/hP

aral

lel

No

pla

ceb

o1

01

10

46

78

27

21

12

Mo

rnin

g1

28

77

Abbr

evia

tions

:AH

I,ap

nea

hypo

pnea

inde

x;BP

,blo

odpr

essu

re;N

A,no

tava

ilabl

eor

appl

icab

le;O

SA,o

bstr

uctiv

esle

epap

nea.

aBo

dym

assi

ndex

(BM

I)ca

lcul

ated

asw

eigh

tin

kilo

gram

sdiv

ided

byhe

ight

inm

eter

ssqu

ared

.

bTh

eEp

wor

thSl

eepi

ness

Scal

e(E

SS)h

asa

scor

era

nge

of0

to24

(hig

hers

core

sind

icat

egr

eate

rday

time

sleep

prop

ensit

y).










the increased weight of the smaller studies directly compar-ing these interventions, which tended to show treatmenteffects closer to the null than larger studies.

By applying a bootstrap procedure with 5000 draws tothe main network analysis model, the probability of CPAPhaving the strongest association with SBP reduction wasestimated to be 72%, whereas it was 28% for MADs. Theprobability of CPAP having the strongest association withDBP reduction was 58%, whereas it was 42% for MADs.Comparison-adjusted funnel plots for the network meta-analysis appear in eFigure 2 in Supplement 2. There is asmall amount of asymmetry in the plot for DBP.

Meta-regression AnalysesMean CPAP use (hours/night) could be obtained from 44 ofthe 47 studies comparing CPAP with an inactive control. Theassociations between the mean CPAP use and the treatmenteffects on BP reported in these studies appear in Figure 6. A1-hour-per-night increase in mean CPAP use was associatedwith an additional reduction in SBP of 1.5 mm Hg (95% CI,0.8 to 2.3 mm Hg; P < .001) and an additional reduction inDBP of 0.9 mm Hg (95% CI, 0.3 to 1.4 mm Hg; P = .001).There was evidence of an association between length offollow-up and the reported effects of CPAP on SBP (slope,0.2 mm Hg per 1-week increase [95% CI, 0.1 to 0.3 mm Hg];

Table 3. Baseline Characteristics of Patients in Trials of Different Treatment Comparisons

Weighted Mean (SD)

P Value

CPAP vsInactive Control(n = 44 Trials)

MAD vsInactive Control(n = 3 Trials)

CPAP vs MAD(n = 1 Trial)a

CPAP vs MAD vsInactive Control(n = 3 Trials)

Length of follow-up, wk 15.1 (9.0)b 7.6 (4.6) 4.3 10.4 (3.7) .30

Age, y 55.3 (6.3) 52.3 (5.0) 49.5 46.2 (0.6) .14

Proportion of males 0.80 (0.13) 0.80 (0.01) 0.81 0.79 (0.02) >.99

Body mass indexc 32.1 (2.0) 29.8 (0.8) 29.5 29.1 (2.1) .06

Apnea-hypopnea index, /h 36.9 (12.7) 20.7 (7.0) 25.6 24.8 (9.4) .08

Oxygen desaturation index, dips/h 21.5 (9.9) 9.8a 20.8 12.8a .68

Epworth Sleepiness Scale scored 10.1 (3.0) 11.5 (0.6) 9.1 11.1 (1.0) .87

Blood pressure, mm Hg

Systolic 132.6 (6.6) 135.0 (7.9) 129.2 127.9 (2.5) .62

Diastolic 80.6 (4.1) 83.3 (4.8) 81.6 77.5 (2.9) .48

Abbreviations: CPAP, continuous positive airway pressure; MAD, mandibularadvancement device.a Standard deviation is not reported if data from only 1 study was available.b Two studies46,55 with length of follow-up of 1 year or longer were excluded

as outliers.

c Calculated as weight in kilograms divided by height in meters squared.d Score range of 0 to 24 (higher scores indicate greater daytime

sleep propensity).

Table 4. Pairwise and Network Meta-analyses Comparing Treatment Effects on Changes in Systolic and Diastolic Blood Pressure

No. ofTrials

SampleSize

Type ofMeta-analysis

Blood Pressure Difference(SE) [95% CI], mm Hg P Value

Cochran χ2

P Valuea I2, %aBetween-StudyVariance, τ2

Systolic Blood Pressure

CPAP vs inactive control 47 4533 Pairwise −2.6 (0.5) [−3.6 to −1.6] <.001 <.001 54 5.2

Network −2.5 (0.5) [−3.5 to −1.5] <.001 5.0

MAD vs inactive control 6 473 Pairwise −1.9 (0.7) [−3.2 to −0.6] .004 .57 0 0

Network −2.1 (0.7) [−3.4 to −0.8] .002 0.7

CPAP vs MAD 4 370 Pairwise 0.3 (0.6) [−1.0 to 1.5] .68 .37 5 0.1

Network −0.5 (0.8) [−2.0 to 1.0] .55 1.9

Diastolic Blood Pressure

CPAP vs inactive control 46 4488 Pairwise −2.1 (0.3) [−2.8 to −1.4] <.001 <.001 52 2.1

Network −2.0 (0.4) [−2.7 to −1.3] <.001 2.3

MAD vs inactive control 6 473 Pairwise −1.1 (0.7) [−2.4 to 0.2] .11 .11 45 1.2

Network −1.9 (0.7) [−3.2 to −0.5] .008 1.3

CPAP vs MAD 4 370 Pairwise 0.2 (0.4) [−0.6 to 0.9] .68 .46 0 0

Network −0.2 (0.7) [−1.6 to 1.3] .82 0.1

Abbreviations: CPAP, continuous positive airway pressure; MAD, mandibular advancement device.a Data correspond to pairwise meta-analysis only.








P = .003) and on DBP (slope, 0.1 mm Hg [95% CI, 0 to 0.2mm Hg]; P = .006) with the reported treatment effectstrending toward zero as length of follow-up increased (eFig-ure 3 in Supplement 2). Studies with smaller sample sizestended to report higher mean compliance (correlation withsample size, r = −0.30) and had shorter follow-up periods(r = 0.19). Thus, the reported treatment effect of CPAP waslikely to be larger in smaller studies, which may be the rea-son for the asymmetry in the funnel plots in eFigure 2rather than publication bias.

There were statistically significant associations betweenthe reported effect of CPAP on SBP and mean baseline SBP

(slope, −0.2 mm Hg [95% CI, −0.3 to 0 mm Hg]; P = .04) andbetween the reported treatment effect on DBP and meanbaseline DBP (slope, −0.2 mm Hg [95% CI, −0.4 to 0 mmHg]; P = .01) (eFigure 4 in Supplement 2). A total of 11 of the47 studies (23%) comparing CPAP with an inactive controlincluded only patients with some form of hypertension. In apost hoc analysis, we found no difference between thereported treatment effects in this subgroup of trials com-pared with those with no such inclusion requirement onSBP (−0.8 mm Hg [95% CI, −3.1 to 1.5 mm Hg]; P = .50) oron DBP (−0.6 mm Hg [95% CI, −2.2 to 1.0 mm Hg]; P = .47).There was no association of either the mean baseline

Figure 2. Treatment Effect for Change in Systolic Blood Pressure (SBP) in the Included Trials of ContinuousPositive Airway Pressure (CPAP) vs Inactive Control

–20 –5 10–10 5 150

Change in SBP Treatment Difference (95% CI), mm Hg

–15

Weight, %Favors

CPAPFavorsInactive ControlCPAP vs Inactive Control

2-Group trials

Change in SBP TreatmentDifference (95% CI), mm Hg

2.5Arias et al,25 2005 0 (–4.5 to 4.5)

2.0Barbé et al,26 2001 –2.0 (–7.5 to 3.5)

4.2Barbé et al,27 2012 –0.1 (–2.1 to 1.9)

0.8Barnes et al,28 2002 –2.9 (–13.1 to 7.3)

0.8Becker et al,29 2003 –10.3 (–20.6 to 0)

2.3Campos-Rodriguez et al,30 2006 –0.9 (–5.8 to 4.0)

0.4Comondore et al,31 2009 –3.6 (–18.4 to 11.3)

3.2Coughlin et al,32 2007 –6.7 (–10.1 to –3.3)

4.4Craig et al,33 2012 1.8 (0.1 to 3.5)

1.4Cross et al,34 2008 –3.8 (–11.1 to 3.5)

0.9de Oliveira et al,35 2014 –8.5 (–17.8 to 0.8)

1.8Drager et al,36 2007 –1.0 (–6.9 to 4.9)

2.4Drager et al,37 2011 –8.0 (–12.7 to –3.3)

4.4Durán-Cantolla et al,38 2010 –1.6 (–3.3 to 0.1)

1.4Egea et al,39 2008 3.7 (–3.6 to 11.0)

0.9Engleman et al,40 1996 –1.0 (–10.8 to 8.8)

4.2Faccenda et al,41 2001 –1.3 (–3.3 to 0.7)

3.2Gottlieb et al,42 2014 –1.7 (–5.1 to 1.7)

0.9Hall et al,43 2014 6.6 (–2.8 to 16.0)

2.1Hoyos et al,44 2012 –4.3 (–9.6 to 1.1)

2.1Hoyos et al,45 2015 –2.4 (–7.6 to 2.8)

2.9Huang et al,46 2015 –5.0 (–8.8 to –1.2)

1.9Hui et al,47 2006 –2.5 (–8.2 to 3.2)

1.4Ip et al,48 2004 0.3 (–7.0 to 7.6)

2.6Jones et al,49 2013 –3.0 (–7.3 to 1.3)

1.5Kohler et al,50 2011 –8.5 (–15.3 to –1.7)

2.6Lam et al,51 2010 –0.9 (–5.2 to 3.3)

3.1Litvin et al,52 2013 –5.0 (–8.5 to –1.5)

1.4Lozano et al,53 2010 –1.6 (–8.7 to 5.5)

2.8Martínez-García et al,54 2013 –1.1 (–5.2 to 3.0)

3.1McMillan et al,55 2014 3.7 (0.2 to 7.3)

2.2Monasterio et al,56 2001 –2.0 (–7.1 to 3.1)

2.6Muxfeldt et al,57 2015 –0.3 (–4.7 to 4.1)

0.9Nguyen et al,58 2010 1.6 (–7.9 to 11.1)

1.0Noda et al,59 2007 –11.1 (–20.0 to –2.2)

1.9Norman et al,60 2006 –6.0 (–11.6 to –0.4)

1.4Pamidi et al,61 2015 –9.7 (–16.9 to –2.5)

1.0Pedrosa et al,62 2013 –9.6 (–18.7 to –0.5)

3.6Pepperell et al,63 2002 –3.4 (–6.3 to –0.6)

2.1Robinson et al,64 2006 –0.1 (–5.3 to 5.1)

1.8Rossi et al,65 2013 –11.4 (–17.5 to –5.3)

0.8Ruttanaumpawan et al,66 2008 –14.0 (–24.7 to –3.3)

0.7Takaesu et al,67 2012 –8.1 (–18.9 to 2.7)

3.4Weaver et al,68 2012 –1.3 (–4.4 to 1.8)

3.9Barnes et al,11 2004 –0.9 (–3.3 to 1.5)

1.9Dal-Fabbro et al,73 2014 –1.4 (–7.2 to 4.4)

1.6Lam et al,12 2007 –6.1 (–12.8 to 0.6)

3-Group trials

Pairwise meta-analysis –2.6 (–3.6 to –1.6)

Network meta-analysis –2.5 (–3.5 to –1.5) The size of each data marker isproportional to the weight carried bythe corresponding study in therandom-effects pairwisemeta-analysis.









apnea-hypopnea index or the oxygen desaturation index ineach study with the reported treatment effects of CPAP onSBP and DBP (eFigures 5 and 6 in Supplement 2). Thus,despite there being some evidence of the mean apnea-hypopnea index differing between treatment comparisons(Table 3), this lack of association implies that the consis-tency assumption of the network meta-analysis was notlikely to be violated.

Of the 47 trials comparing CPAP with an inactive con-trol, 22 used sham CPAP as the comparator. The remaining25 studies used either no treatment (n = 18), an oral placebo(n = 5), a placebo oral appliance (n = 1), or an expiratorynasal resistance valve placebo (n = 1). There were no statisti-cally significant differences in the reported effects of CPAPon SBP or DBP between trials using sham CPAP, any othertype of placebo or no placebo as the comparator (eTable 4 inSupplement 2).

Data on daytime ambulatory BP measurements couldbe extracted from 20 of the 47 studies comparing CPAPwith an inactive control, and only 24-hour data could beobtained in 6 studies. A post hoc analysis showed that theassociation of CPAP with reduction in BP in these 6 studiesdid not differ compared with those in which daytime BPdata was obtained; however, there was some suggestionthat the effect of CPAP reported in the studies was larger inthose in which morning BP data was extracted (eTable 5 inSupplement 2). Applying the adjustment method by Knappand Hartung20 in a sensitivity analysis made little differencein the findings of the meta-regression analyses, increasing

the standard error of the meta-regression coefficients by nomore than 9% in the analyses for SBP and by no more than12% for DBP.

Risk of Bias AssessmentNo more than 5 of the included trials (<10%) were deemed tobe at high risk of bias in only 3 domains (allocation conceal-ment, incomplete outcome data, selective reporting) of theCochrane Collaboration risk of bias tool (eTable 6 and eFigure7 in Supplement 2). In most domains, the majority of trials wereat low risk, except for the allocation concealment category inwhich most trials were at an unclear risk due to inadequate re-porting of methods. Due to either the small number or the ab-sence of any high-risk studies in each domain, reported treat-ment effects were not compared between trials at high and lowrisk of bias.

DiscussionTo our knowledge, this is the first network meta-analysiscomparing CPAP, MADs, and inactive controls on BP inpatients with obstructive sleep apnea. We found that bothCPAP and MADs were associated with similar reductions inSBP and DBP compared with an inactive treatment. This ispartly in contrast to a previous meta-analysis,10 which didnot find a beneficial association with MADs, perhaps due toincluding only 2 RCTs and thus having inadequate power todetect a difference. Even though there was no statistically

Figure 3. Treatment Effect for Change in Systolic Blood Pressure (SBP) in the Included Trials of MandibularAdvancement Device (MAD) vs Continuous Positive Airway Pressure (CPAP) and vs Inactive Controls

–20 –5 10–10 5 150


–15

Weight, %Favors

MADFavorsInactive ControlMAD vs Inactive Control


10.8Andrén et al,69 2013 –1.8 (–5.8 to 2.2)

36.4Gotsopoulos et al,70 2004 –3.3 (–5.5 to –1.1)

13.1Quinnell et al,71 2014 –0.2 (–3.8 to 3.4)

2-Group trials

31.7Barnes et al,11 2004 –1.5 (–3.8 to 0.8)

4.8Dal-Fabbro et al,73 2014 1.5 (–4.4 to 7.4)

3.2Lam et al,12 2007 –2.4 (–9.7 to 4.9)

3-Group trials


Network meta-analysis –2.1 (–3.4 to –0.8)

–20 –5 10–10 5 150


–15

Weight, %Favors

CPAPFavorsMADCPAP vs MAD


64.7Phillips et al,72 2013 0.6 (–0.8 to 2.0)

2-Group trial

26.6Barnes et al,11 2004 0.6 (–1.7 to 2.9)


4.2Lam et al,12 2007 –3.7 (–9.7 to 2.3)

3-Group trials

Pairwise meta-analysis 0.3 (–1.0 to 1.5)

Network meta-analysis –0.5 (–2.0 to 1.0) The size of each data marker isproportional to the weight carried bythe corresponding study in therandom-effects pairwisemeta-analysis.











significant difference between the associations of CPAP andMADs with change in BP in our meta-analysis, CPAP had aconsiderably higher probability of having the strongestassociation with SBP reduction. The associations of bothCPAP and MADs with DBP reduction were more similar;however, the association of CPAP with reductions of bothSBP and DBP is likely to be greater in patients using CPAPfor longer periods at night or in those with higher baselineBP levels.

Even though the results of the pairwise and networkmeta-analyses were mostly similar, the biggest differencewas seen in the comparison of MADs with inactive controlson DBP with the network model estimating a larger associa-

tion than the pairwise meta-analysis. This was most likelybecause the data from the direct comparisons of CPAP andMADs tended to favor MADs, and so incorporating these datain the network meta-analysis increased the differencebetween MADs and inactive controls on change in DBP. Inaddition, because the difference between CPAP and inactivecontrols in the pairwise analyses was substantially largerthan that for MADs, the comparison between CPAP andMADs changed to favor CPAP in the network meta-analysis(albeit not to a statistically significant extent). The precisionof the comparison between CPAP and MAD was slightly lowerin the network meta-analyses than when considering directevidence alone, which may be due to the large amount of

Figure 4. Treatment Effect for Change in Diastolic Blood Pressure (DBP) in the Included Trials of ContinuousPositive Airway Pressure (CPAP) vs Inactive Control

–20 –5 10–10 5 150

Change in DBP Treatment Difference (95% CI), mm Hg

–15

Weight, %Favors

CPAPFavorsInactive ControlCPAP vs Inactive Control

2-Group trials

Change in DBP TreatmentDifference (95% CI), mm Hg

3.5Arias et al,25 2005 0 (–2.1 to 2.1)

2.8Barbé et al,26 2001 –1.0 (–3.8 to 1.8)

4.3Barbé et al,27 2012 –1.2 (–2.7 to 0.2)

0.4Barnes et al,28 2002 –2.6 (–12.8 to 7.6)

0.6Becker et al,29 2003 –11.2 (–19.5 to –2.8)

2.7Campos-Rodriguez et al,30 2006 –0.7 (–3.5 to 2.2)

0.6Comondore et al,31 2009 –0.7 (–8.5 to 7.1)

2.5Coughlin et al,32 2007 –4.9 (–8.0 to –1.8)

4.7Craig et al,33 2012 –0.4 (–1.5 to 0.7)

1.9Cross et al,34 2008 0 (–3.9 to 3.9)

1.3de Oliveira et al,35 2014 –4.5 (–9.6 to 0.6)

1.6Drager et al,36 2007 –1.0 (–5.4 to 3.4)

2.3Drager et al,37 2011 –6.0 (–9.4 to –2.6)

4.6Durán-Cantolla et al,38 2010 –1.1 (–2.3 to 0.1)

1.2Egea et al,39 2008 –1.4 (–6.7 to 3.9)

0.6Engleman et al,40 1996 –2.0 (–10.3 to 6.3)

4.3Faccenda et al,41 2001 –1.5 (–3.0 to –0.1)

3.5Gottlieb et al,42 2014 –2.5 (–4.6 to –0.4)

1.9Hoyos et al,44 2012 0.4 (–3.4 to 4.3)

2.7Hoyos et al,45 2015 –3.0 (–5.8 to –0.2)

1.0Huang et al,46 2015 –1.0 (–7.2 to 5.2)

2.1Hui et al,47 2006 –1.8 (–5.3 to 1.8)

0.7Ip et al,48 2004 –8.9 (–16.4 to –1.4)

2.9Jones et al,49 2013 0 (–2.6 to 2.6)

1.3Kohler et al,50 2011 –6.9 (–11.9 to –1.9)

2.2Lam et al,51 2010 –0.6 (–4.0 to 2.8)

3.0Litvin et al,52 2013 –6.0 (–8.6 to –3.4)

1.5Lozano et al,53 2010 –0.8 (–5.3 to 3.7)

3.1Martínez-García et al,54 2013 –2.3 (–4.8 to 0.2)

3.3McMillan et al,55 2014 0.2 (–2.1 to 2.5)

2.7Monasterio et al,56 2001 –1.0 (–3.8 to 1.8)

2.5Muxfeldt et al,57 2015 0.1 (–3.0 to 3.2)

0.8Nguyen et al,58 2010 8.9 (1.9 to 15.9)

1.1Noda et al,59 2007 –7.0 (–12.6 to –1.4)

1.7Norman et al,60 2006 –5.3 (–9.5 to –1.1)

1.7Pamidi et al,61 2015 –8.2 (–12.5 to –3.9)

1.1Pedrosa et al,62 2013 –6.6 (–12.4 to –0.8)

3.6Pepperell et al,63 2002 –3.3 (–5.3 to –1.3)

1.2Robinson et al,64 2006 –1.5 (–6.7 to 3.8)

1.6Rossi et al,65 2013 –7.3 (–11.7 to –2.9)

0.6Ruttanaumpawan et al,66 2008 –5.0 (–13.2 to 3.2)

0.7Takaesu et al,67 2012 –5.9 (–13.2 to 1.4)

3.8Weaver et al,68 2012 –1.9 (–3.8 to 0)

4.2Barnes et al,11 2004 –0.6 (–2.2 to 1.0)


1.6Lam et al,12 2007 –2.0 (–6.3 to 2.3)

3-Group trials


Network meta-analysis –2.0 (–2.7 to –1.3) The size of each data marker isproportional to the weight carried bythe corresponding study in therandom-effects pairwisemeta-analysis.







between-study variation observed in the indirect compari-sons being incorporated into the analysis.

Our meta-regression analyses showed that studies inpatients with higher mean use of CPAP or higher baseline BPlevel tended to report more beneficial treatment effects ofCPAP on SBP and DBP. However, it should be noted thatmeta-regression analyses of mean patient characteristics donot necessarily demonstrate the dose-response relationship

at a patient level and so it is possible that the same associa-tions found herein would not have been observed in manyof the individual trials.74 For instance, in contrast to ourfindings, a previous meta-analysis75 investigating the effectof CPAP in asymptomatic patients using individual patientdata did not detect a difference between outcomes inpatients using CPAP more or less than 4 hours/night com-pared with controls. Repeating our network meta-analysis

Figure 6. Association Between Use of Continuous Positive Airway Pressure (CPAP) on Change in Systolic Blood Pressure and Diastolic Blood Pressure

10

5

0

–5

–10

32–15

654 7

Rep

ort

ed C

han

ge

in S

BP

Tre

atm

ent

Dif

fere

nce

, m

m H

g

Systolic blood pressure(SBP; n = 44 studies)

Mean CPAP Use, h/Night

SBP slope = –1.5 mm Hg (95% CI –2.3 to –0.8), P <.001

10

5

0

–5

–10

32–15

654 7

Rep

ort

ed C

han

ge

in D

BP

Tre

atm

ent

Dif

fere

nce

, m

m H

g

Diastolic blood pressure(DBP; n = 43 studies)

Mean CPAP Use, h/Night

DBP slope = –0.9 mm Hg (95% CI –1.4 to –0.3), P = .001

Circles represent individual results for each trial with the size of the circle beingproportional to its weight in the random-effects meta-analysis. Eachmeta-regression line (solid lines) and its 95% confidence interval (shaded areas)

were estimated using a random-effects linear meta-regression model withmean use of CPAP as the covariate. The dashed lines represent the overalldifference. Treatment difference data are CPAP minus inactive control.

Figure 5. Treatment Effect for Change in Diastolic Blood Pressure (DBP) in the Included Trials of MandibularAdvancement Device (MAD) vs Continuous Positive Airway Pressure (CPAP) and vs Inactive Controls

–20 –5 10–10 5 150


–15

Weight, %Favors

MADFavorsInactive ControlMAD vs Inactive Control


15.5Andrén et al,69 2013 –1.2 (–3.9 to 1.5)

23.4Gotsopoulos et al,70 2004 –3.4 (–5.2 to –1.6)

18.5Quinnell et al,71 2014 0.3 (–2.0 to 2.6)

2-Group trials

26.1Barnes et al,11 2004 –1.0 (–2.5 to 0.5)

8.8Dal-Fabbro et al,73 2014 0.7 (–3.3 to 4.7)

7.7Lam et al,12 2007 0.4 (–3.9 to 4.7)

3-Group trials

Pairwise meta-analysis –1.1 (–2.4 to 0.2)

Network meta-analysis –1.9 (–3.2 to –0.5)

–20 –5 10–10 5 150


–15

Weight, %Favors

CPAPFavorsMADCPAP vs MAD


70.0Phillips et al,72 2013 0.3 (–0.6 to 1.2)

2-Group trial

23.1Barnes et al,11 2004 0.4 (–1.2 to 2.0)


3.3Lam et al,12 2007 –2.4 (–6.5 to 1.7)

3-Group trials

Pairwise meta-analysis 0.2 (–0.6 to 0.9)

Network meta-analysis –0.2 (–1.6 to 1.3) The size of each data marker isproportional to the weight carried bythe corresponding study in therandom-effects pairwisemeta-analysis.







using individual patient data rather than aggregate datawould improve the assessment of the association of CPAPuse and other patient characteristics with each treatmentcomparison but would be challenging to conduct due to thelarge number of RCTs from which to acquire data. Alterna-tively, conducting an RCT in which patients are given vari-ous durations of CPAP therapy each night will provide anunbiased assessment of whether there is a dose-responserelationship with BP but also might prove challengingto conduct.

Compared with the most recent meta-analysis5 on thistopic with similar inclusion criteria, our study includes18 more RCTs of CPAP and includes at least an additional2700 patients. Therefore, we had considerably more powerto assess the association of trial-level characteristics, suchas mean CPAP use, with the reported treatment effects onSBP and DBP. We have also used data from 6 trials compar-ing MADs with an inactive control, which is considerablymore than the 2 trials used in a previous meta-analysis.10

Although only 4 RCTs directly compared CPAP with MADs,we have attempted to strengthen this and all other compari-sons by incorporating indirect evidence using a networkmeta-analysis. Furthermore, in contrast to separate pair-wise analyses, we have been able to rank each treatmentbased on the strength of its association with reductions inSBP and DBP.

A limitation of our meta-analysis is that we only investi-gated 2 active treatments (ie, CPAP and MADs) and excludedother treatments, such as weight loss interventions,76 whichare likely to have beneficial effects on BP because they havebeen shown to have favorable effects on decreasing theseverity of obstructive sleep apnea. However, few trials ofother interventions exist and so including them would haveincreased the sparseness of the network meta-analysis,which can lead to modeling problems, particularly withregard to estimating the between-trial variance of treat-

ments that were not directly compared. Another limitationwas that we were unable to extract daytime ambulatory BPdata from all studies and thus had to use the availablemorning, office, or 24-hour measurements. Although thismay have increased heterogeneity, it allowed all of the rela-tively few studies investigating MADs to be incorporatedinto the analyses. A subgroup analysis showed some evi-dence that studies from which we extracted morning BPreported slightly larger treatment effects than in other stud-ies. However, there was no difference with studies in whichwe extracted 24-hour ambulatory measurements and so theeffect of any nighttime BP variability was negligible. Futuremeta-analyses could analyze each BP measurement sepa-rately to better understand whether each treatment is asso-ciated with greater reductions in BP during the daytimeor nighttime.

Our results were robust to the assumed between-visitcorrelation, which was estimated from parallel trials. How-ever, for simplicity, we did not estimate the between-periodcorrelation from crossover trials (reporting treatment effectsfrom paired t tests) and then use that estimate when calcu-lating treatment effects in other crossover studies. Althoughthis could be deemed a limitation of our analyses, assuminga between-period correlation of zero is arguably reasonablewhen considering changes from baseline.77 In addition,because only a small proportion of crossover studies weretreated this way, we do not believe that our results are sen-sitive to this assumption.

ConclusionsAmong patients with obstructive sleep apnea, both CPAP andMADs were associated with reductions in BP. Network meta-analysis did not identify a statistically significant differencebetween the BP outcomes associated with these therapies.

ARTICLE INFORMATION

Author Contributions: Dr Bratton had full access toall of the data in the study and takes responsibilityfor the integrity of the data and the accuracy of thedata analysis.Study concept and design: Bratton, Kohler.Acquisition, analysis, or interpretation of data: Allauthors.Drafting of the manuscript: Bratton.Critical revision of the manuscript for importantintellectual content: All authors.Statistical analysis: Bratton.Obtained funding: Kohler.Administrative, technical, or material support: Gaisl,Kohler.Study supervision: Kohler.

Conflict of Interest Disclosures: The authors havecompleted and submitted the ICMJE Form forDisclosure of Potential Conflicts of Interest andnone were reported.

Funding/Support: This research was supported bygrant 32003B_143365/1 from the Swiss NationalScience Foundation and by funding from theUniversity of Zurich Clinical Research PriorityProgram Sleep and Health.

Role of Funder/Sponsor: The sponsors had no rolein the design and conduct of the study; collection,management, analysis, and interpretation of thedata; preparation, review, or approval of themanuscript; and decision to submit the manuscriptfor publication.

REFERENCES

1. Young T, Peppard PE, Gottlieb DJ. Epidemiologyof obstructive sleep apnea. Am J Respir Crit Care Med.2002;165(9):1217-1239.

2. Kohler M, Stradling JR. Mechanisms of vasculardamage in obstructive sleep apnea. Nat Rev Cardiol.2010;7(12):677-685.

3. Giles TL, Lasserson TJ, Smith B, et al. Continuouspositive airways pressure for obstructive sleepapnoea in adults. Cochrane Database Syst Rev.doi:10.1002/14651858.CD001106.pub3.

4. Bazzano LA, Khan Z, Reynolds K, He J. Effect ofnocturnal nasal continuous positive airway pressureon blood pressure in obstructive sleep apnea.Hypertension. 2007;50(2):417-423.

5. Fava C, Dorigoni S, Dalle Vedove F, et al. Effect ofCPAP on blood pressure in patients withOSA/hypopnea. Chest. 2014;145(4):762-771.

6. Haentjens P, Van Meerhaeghe A, Moscariello A,et al. The impact of continuous positive airwaypressure on blood pressure in patients withobstructive sleep apnea syndrome. Arch Intern Med.2007;167(8):757-764.

7. Montesi SB, Edwards BA, Malhotra A, Bakker JP.The effect of continuous positive airway pressuretreatment on blood pressure. J Clin Sleep Med.2012;8(5):587-596.

8. Schein AS, Kerkhoff AC, Coronel CC, et al.Continuous positive airway pressure reduces bloodpressure in patients with obstructive sleep apnea.J Hypertens. 2014;32(9):1762-1773.

9. Bennett LS, Davies RJ, Stradling JR.Oral appliances for the management of snoring andobstructive sleep apnoea. Thorax. 1998;53(suppl2):S58-S64.

10. Iftikhar IH, Hays ER, Iverson MA, et al. Effect oforal appliances on blood pressure in obstructivesleep apnea. J Clin Sleep Med. 2013;9(2):165-174.





http://www.ncbi.nlm.nih.gov/pubmed/11991871




http://dx.doi.org/10.1002/14651858.CD001106.pub3













11. Barnes M, McEvoy RD, Banks S, et al. Efficacy ofpositive airway pressure and oral appliance in mildto moderate obstructive sleep apnea. Am J RespirCrit Care Med. 2004;170(6):656-664.

12. Lam B, Sam K, Mok WY, et al. Randomised studyof three non-surgical treatments in mild tomoderate obstructive sleep apnoea. Thorax. 2007;62(4):354-359.

13. Caldwell DM, Ades AE, Higgins JP. Simultaneouscomparison of multiple treatments. BMJ. 2005;331(7521):897-900.

14. Lefebvre C, Manheimer E, Glanville J.Chapter 6: Searching for studies. In: Higgins JPT,Green S, eds. Cochrane Handbook for SystematicReviews of Intervensions. London, England: CochraneCollaboration; 2011.

15. Higgins JPT, Altman DG, Sterne JAC.Chapter 8: Assessing risk of bias in included studies.In: Higgins JPT, Green S, eds. Cochrane Handbookfor Systematic Reviews of Interventions. London,England: Cochrane Collaboration; 2011.

16. Higgins JPT, Deeks JJ, Altman DG.Chapter 16: Special topics in statistics. In: HigginsJPT, Green S, eds. Cochrane Handbook forSystematic Reviews of Interventions. London,England: Cochrane Collaboration; 2011.

17. Harris R, Bradburn M, Deeks J, et al.Metan: fixed- and random-effects meta-analysis.Stata J. 2008;8(1):3-28.

18. Higgins JP, Thompson SG. Quantifyingheterogeneity in a meta-analysis. Stat Med. 2002;21(11):1539-1558.

19. Harbord RM, Higgins JPT. Meta-regression inStata. Stata J. 2008;8(4):493-519.

20. Knapp G, Hartung J. Improved tests for arandom effects meta-regression with a singlecovariate. Stat Med. 2003;22(17):2693-2710.

21. White IR, Barrett JK, Jackson D, Higgins JPT.Consistency and inconsistency in networkmeta-analysis. Res Synth Methods. 2012;3(2):111-125.

22. Cipriani A, Higgins JP, Geddes JR, Salanti G.Conceptual and technical challenges in networkmeta-analysis. Ann Intern Med. 2013;159(2):130-137.

23. White IR. Multivariate random-effectsmeta-regression. Stata J. 2011;11(2):255-270.

24. Chaimani A, Higgins JPT, Mavridis D,Spyridonos P, Salanti G. Graphical tools for networkmeta-analysis in STATA. PLoS One. 2013;8(10):e76654.

25. Arias MA, García-Río F, Alonso-Fernández A,et al. Obstructive sleep apnea syndrome affects leftventricular diastolic function. Circulation. 2005;112(3):375-383.

26. Barbé F, Mayoralas LR, Duran J, et al.Treatment with continuous positive airwaypressure is not effective in patients with sleepapnea but no daytime sleepiness. Ann Intern Med.2001;134(11):1015-1023.

27. Barbé F, Durán-Cantolla J, Sánchez-de-la-TorreM, et al. Effect of continuous positive airwaypressure on the incidence of hypertension andcardiovascular events in nonsleepy patients withobstructive sleep apnea. JAMA. 2012;307(20):2161-2168.

28. Barnes M, Houston D, Worsnop CJ, et al.A randomized controlled trial of continuous positiveairway pressure in mild obstructive sleep apnea. AmJ Respir Crit Care Med. 2002;165(6):773-780.

29. Becker HF, Jerrentrup A, Ploch T, et al. Effect ofnasal continuous positive airway pressuretreatment on blood pressure in patients withobstructive sleep apnea. Circulation. 2003;107(1):68-73.

30. Campos-Rodriguez F, Grilo-Reina A,Perez-Ronchel J, et al. Effect of continuous positiveairway pressure on ambulatory BP in patients withsleep apnea and hypertension. Chest. 2006;129(6):1459-1467.

31. Comondore VR, Cheema R, Fox J, et al. Theimpact of CPAP on cardiovascular biomarkers inminimally symptomatic patients with obstructivesleep apnea. Lung. 2009;187(1):17-22.

32. Coughlin SR, Mawdsley L, Mugarza JA, et al.Cardiovascular and metabolic effects of CPAP inobese males with OSA. Eur Respir J. 2007;29(4):720-727.

33. Craig SE, Kohler M, Nicoll D, et al. Continuouspositive airway pressure improves sleepiness butnot calculated vascular risk in patients withminimally symptomatic obstructive sleep apnoea.Thorax. 2012;67(12):1090-1096.

34. Cross MD, Mills NL, Al-Abri M, et al. Continuouspositive airway pressure improves vascular functionin obstructive sleep apnoea/hypopnoea syndrome.Thorax. 2008;63(7):578-583.

35. de Oliveira AC, Martinez D, Massierer D, et al.The antihypertensive effect of positive airwaypressure on resistant hypertension of patients withobstructive sleep apnea. Am J Respir Crit Care Med.2014;190(3):345-347.

36. Drager LF, Bortolotto LA, Figueiredo AC, et al.Effects of continuous positive airway pressure onearly signs of atherosclerosis in obstructive sleepapnea. Am J Respir Crit Care Med. 2007;176(7):706-712.

37. Drager LF, Pedrosa RP, Diniz PM, et al. Theeffects of continuous positive airway pressure onprehypertension and masked hypertension in menwith severe obstructive sleep apnea. Hypertension.2011;57(3):549-555.

38. Durán-Cantolla J, Aizpuru F, Montserrat JM,et al. Continuous positive airway pressure astreatment for systemic hypertension in people withobstructive sleep apnoea. BMJ. 2010;341:c5991.

39. Egea CJ, Aizpuru F, Pinto JA, et al. Cardiacfunction after CPAP therapy in patients withchronic heart failure and sleep apnea. Sleep Med.2008;9(6):660-666.

40. Engleman HM, Gough K, Martin SE, et al.Ambulatory blood pressure on and off continuouspositive airway pressure therapy for the sleepapnea/hypopnea syndrome. Sleep. 1996;19(5):378-381.

41. Faccenda JF, Mackay TW, Boon NA, Douglas NJ.Randomized placebo-controlled trial of continuouspositive airway pressure on blood pressure in thesleep apnea-hypopnea syndrome. Am J Respir CritCare Med. 2001;163(2):344-348.

42. Gottlieb DJ, Punjabi NM, Mehra R, et al. CPAPversus oxygen in obstructive sleep apnea. N Engl JMed. 2014;370(24):2276-2285.

43. Hall AB, Ziadi MC, Leech JA, et al. Effects ofshort-term continuous positive airway pressure onmyocardial sympathetic nerve function andenergetics in patients with heart failure andobstructive sleep apnea. Circulation. 2014;130(11):892-901.

44. Hoyos CM, Killick R, Yee BJ, et al.Cardiometabolic changes after continuous positiveairway pressure for obstructive sleep apnoea. Thorax.2012;67(12):1081-1089.

45. Hoyos CM, Yee BJ, Wong KK, et al. Treatmentof sleep apnea with CPAP lowers central andperipheral blood pressure independent of thetime-of-day. Am J Hypertens. 2015;28(10):1222-1228.

46. Huang Z, Liu Z, Luo Q, et al. Long-term effectsof continuous positive airway pressure on bloodpressure and prognosis in hypertensive patientswith coronary heart disease and obstructive sleepapnea. Am J Hypertens. 2015;28(3):300-306.

47. Hui DS, To KW, Ko FW, et al. Nasal CPAPreduces systemic blood pressure in patients withobstructive sleep apnoea and mild sleepiness. Thorax.2006;61(12):1083-1090.

48. Ip MS, Tse HF, Lam B, et al. Endothelialfunction in obstructive sleep apnea and response totreatment. Am J Respir Crit Care Med. 2004;169(3):348-353.

49. Jones A, Vennelle M, Connell M, et al. Theeffect of continuous positive airway pressuretherapy on arterial stiffness and endothelialfunction in obstructive sleep apnea. Sleep Med.2013;14(12):1260-1265.

50. Kohler M, Stoewhas AC, Ayers L, et al. Effectsof continuous positive airway pressure therapywithdrawal in patients with obstructive sleepapnea. Am J Respir Crit Care Med. 2011;184(10):1192-1199.

51. Lam JC, Lam B, Yao TJ, et al. A randomisedcontrolled trial of nasal continuous positive airwaypressure on insulin sensitivity in obstructive sleepapnoea. Eur Respir J. 2010;35(1):138-145.

52. Litvin AY, Sukmarova ZN, Elfimova EM, et al.Effects of CPAP on “vascular” risk factors in patientswith obstructive sleep apnea and arterialhypertension. Vasc Health Risk Manag. 2013;9:229-235.

53. Lozano L, Tovar JL, Sampol G, et al. Continuouspositive airway pressure treatment in sleep apneapatients with resistant hypertension. J Hypertens.2010;28(10):2161-2168.

54. Martínez-García MA, Capote F,Campos-Rodríguez F, et al. Effect of CPAP on bloodpressure in patients with obstructive sleep apneaand resistant hypertension. JAMA. 2013;310(22):2407-2415.

55. McMillan A, Bratton DJ, Faria R, et al.Continuous positive airway pressure in older peoplewith obstructive sleep apnoea syndrome(PREDICT). Lancet Respir Med. 2014;2(10):804-812.

56. Monasterio C, Vidal S, Duran J, et al.Effectiveness of continuous positive airwaypressure in mild sleep apnea-hypopnea syndrome.Am J Respir Crit Care Med. 2001;164(6):939-943.

57. Muxfeldt ES, Margallo V, Costa LM, et al. Effectsof continuous positive airway pressure treatmenton clinic and ambulatory blood pressures inpatients with obstructive sleep apnea and resistanthypertension. Hypertension. 2015;65(4):736-742.

58. Nguyen PK, Katikireddy CK, McConnell MV,et al. Nasal continuous positive airway pressureimproves myocardial perfusion reserve andendothelial-dependent vasodilation in patients withobstructive sleep apnea. J Cardiovasc Magn Reson.2010;12:50.














































































59. Noda A, Nakata S, Koike Y, et al. Continuouspositive airway pressure improves daytimebaroreflex sensitivity and nitric oxide production inpatients with moderate to severe obstructive sleepapnea syndrome. Hypertens Res. 2007;30(8):669-676.

60. Norman D, Loredo JS, Nelesen RA, et al.Effects of continuous positive airway pressureversus supplemental oxygen on 24-hourambulatory blood pressure. Hypertension. 2006;47(5):840-845.

61. Pamidi S, Wroblewski K, Stepien M, et al.Eight hours of nightly continuous positive airwaypressure treatment of obstructive sleep apneaimproves glucose metabolism in patients withprediabetes. Am J Respir Crit Care Med. 2015;192(1):96-105.

62. Pedrosa RP, Drager LF, de Paula LK, et al.Effects of OSA treatment on BP in patients withresistant hypertension. Chest. 2013;144(5):1487-1494.

63. Pepperell JC, Ramdassingh-Dow S,Crosthwaite N, et al. Ambulatory blood pressureafter therapeutic and subtherapeutic nasalcontinuous positive airway pressure for obstructivesleep apnoea. Lancet. 2002;359(9302):204-210.

64. Robinson GV, Smith DM, Langford BA, et al.Continuous positive airway pressure does notreduce blood pressure in nonsleepy hypertensiveOSA patients. Eur Respir J. 2006;27(6):1229-1235.

65. Rossi VA, Winter B, Rahman NM, et al.The effects of Provent on moderate to severeobstructive sleep apnoea during continuouspositive airway pressure therapy withdrawal. Thorax.2013;68(9):854-859.

66. Ruttanaumpawan P, Gilman MP, Usui K, et al.Sustained effect of continuous positive airwaypressure on baroreflex sensitivity in congestiveheart failure patients with obstructive sleep apnea.J Hypertens. 2008;26(6):1163-1168.

67. Takaesu Y, Inoue Y, Komada Y, et al. Effects ofnasal continuous positive airway pressure on panicdisorder comorbid with obstructive sleep apneasyndrome. Sleep Med. 2012;13(2):156-160.

68. Weaver TE, Mancini C, Maislin G, et al.Continuous positive airway pressure treatment ofsleepy patients with milder obstructive sleepapnea. Am J Respir Crit Care Med. 2012;186(7):677-683.

69. Andrén A, Hedberg P, Walker-Engström ML,et al. Effects of treatment with oral appliance on24-h blood pressure in patients with obstructivesleep apnea and hypertension. Sleep Breath. 2013;17(2):705-712.

70. Gotsopoulos H, Kelly JJ, Cistulli PA. Oralappliance therapy reduces blood pressure inobstructive sleep apnea. Sleep. 2004;27(5):934-941.

71. Quinnell TG, Bennett M, Jordan J, et al.A crossover randomised controlled trial of oral

mandibular advancement devices for obstructivesleep apnoea-hypopnoea (TOMADO). Thorax.2014;69(10):938-945.

72. Phillips CL, Grunstein RR, Darendeliler MA,et al. Health outcomes of continuous positiveairway pressure versus oral appliance treatment forobstructive sleep apnea. Am J Respir Crit Care Med.2013;187(8):879-887.

73. Dal-Fabbro C, Garbuio S, D’Almeida V, et al.Mandibular advancement device and CPAP uponcardiovascular parameters in OSA. Sleep Breath.2014;18(4):749-759.

74. Thompson SG, Higgins JP. How shouldmeta-regression analyses be undertaken andinterpreted? Stat Med. 2002;21(11):1559-1573.

75. Bratton DJ, Stradling JR, Barbé F, Kohler M.Effect of CPAP on blood pressure in patients withminimally symptomatic obstructive sleep apnoea.Thorax. 2014;69(12):1128-1135.

76. Chirinos JA, Gurubhagavatula I, Teff K, et al.CPAP, weight loss, or both for obstructive sleepapnea. N Engl J Med. 2014;370(24):2265-2275.

77. Elbourne DR, Altman DG, Higgins JP, et al.Meta-analyses involving cross-over trials. Int JEpidemiol. 2002;31(1):140-149.



































Date post:	15-Mar-2022
Category:	Documents
Upload:	others
View:	5 times
Download:	0 times

OriginalInvestigation ... - obu.ch

Documents