The Acute Effects of Continuous Positive Airway Pressure and Oxygen Administration on Blood Pressure during Obstructive Sleep Apnea

The Acute Effects of Continuous Positive Airway Pressure and Oxygen Administration on Blood Pressure during Obstructive Sleep Apnea

The Acute Effects of Continuous Positive Airway Pressure and Oxygen Administration on Blood Pressure during Obstructive Sleep Apnea* N. J Ali, B.M.; R...

1MB Sizes 0 Downloads 2 Views

The Acute Effects of Continuous Positive Airway Pressure and Oxygen Administration on Blood Pressure during Obstructive Sleep Apnea* N. J Ali, B.M.; R. J Q Davies, B.M.; J A. Fleetham, M.D.; and

J R. Stradling, M.D.

We have measured blood pressure continuously with a digital artery blood pressure monitor in eight patients with severe obstructive sleep apnea (OSA) during 30 min each of wakefulness, OSA, OSA with added oxygen to keep saturation above 96 percent at all times (OSA + OJ, and nasal continuous positive airway pressure (CPAP) therap~ Mean blood pressures were not different between wakefulness, OSA, OSA + 0., and CPAP, although the variability in blood pressure was signi6cantly greater during OSA and OSA + O. than during wakefulness and CPAP. The addition ofoxygen did not attenuate the variability in blood pressure. Using multiple linear regression modeling to further dissect out the priDcipai variables determining the postapneic blood pressure rise, we found that only apnea length ("'=0.28, p
sleep apnea is associated with O bstructive swings in arterial blood pressure, sometimes (OSA)

this was small by comparison with apnea length. We conclude that CPAP treatment of OSA does not lower mean blood pressure acutely, although it signi6cantly reduces the large oscillations in blood pressure seen in patients with untreated OSA. The rise in blood pressure following each apnea is not primarily due to arterial desaturation but is related to apnea length and may be caused by increased sympathetic activity secondary to arousal. (Chat 1992; 101:1526-32) CV=coefticient of variation; DBPmu=muimum diastolic pressure following apn~ termination calculated as for SBPmax; DBPten=mean diastolic blood pressure during the 6nt 10 s immediately after apnea termination, normalized as above; E O G = e l _ ; OSA+O.=obstrudive sleep a~ with added oxygen to keep saturation above 96_percent at all times; SBPmu = maximum systolic pressure following_ apnea termination, expressed as a percentage of the overall mean systolic blood pressure during the experiment, excluding wakefulness; SBPten = mean systolic pressure during the &nt 10 s immediately after the end of the apnea, also DOrmalized as above.

and examined the factors responsible for the postapneic blood pressure rises.

considerably in excess of 50 mm Hg. These swings are

of two types: (1) in time with the obstructed breaths due to the large oscillations in pleural pressure (pulsus paradoxus), and (2) rises that occur with the termination of each apnea. 1 Shepard2 suggested that these postapneic blood pressure rises were primarily dependent on the degree of arterial hypoxemia developing during the apnea and, to a lesser extent, on the actual arousal from sleep that accompanies the resumption of respiration. 2 The recent development ofa noninvasive continuous arterial blood pressure monitor has made intensive study of the mechanism of these apnea-related blood pressure swings possible without the. need for arterial cannulation. In this study, we investigated the acute effects of oxygen administration and continuous positive airway pressure (CPAP) on peripheral arterial blood pressure *From the Osler Chest Unit (Drs. Ali, Davies, and Stradling), Churchill Hospital, Oxford, United Kingdom, and Division of Respiratory Medicine (Dr. Fleetham), University Hospital, Vancouver, British Columbia, Canada. Supported by a British Lung Foundation Grant (Dr. Ali). Manuscript received November 19, 1990; revision accepted August 7.



Ibtients We studied eight patients (seven male) with moderate to severe OSA. Anthropometric and clinical data are summarized in Table 1. Six of the eight patients were studied on their first night of CPAP; two had been established on CPAP for two and four years, respectively, but bad discontinued treatment for three nights before the study. 7echniques

The following were recorded in all cases: (1) Electroencephalo-

gram (EEG), electro-oculogram (EOG), and submental electromy-

ogram (EMG) onto an eight-channel tape recorder (MPA I, Oxford Medical, Abingdon, United Kingdom). All-night sleep staging was according to standard criteria, initially by an automated method3 (Medilog9000, Oxford Medical, Abingdon, UK), but manual staging was used for the periods of sleep actually used in the subsequent analysis. 4 (2) Blood pressure wavefonn and pulse rate were measured (by the Obmeda 2300 Finapres, Englewood, Colo) with the cuff attached to the middle finger of the right hand. This device constantly clamps the arterial volume of the digital arteries at their "unloaded" point with a servo-controned pneumatic cuff, the internal pressure of which thus tracks the arterial blood pressure in the finger itseH: IS The patients remained in a supine position throughout the experiment. Hand position was kept constant across the abdomen with a sling to maintain the same height above the CPAP and O. during 0bstructNe Sleep Apnea (Ali et 81)

heart, and its position was checked during the study by video monitoring. The output from the Finapres was recorded onto a sixchannel chart recorder, and via the serial data port into a laptop computer (Zenith Data Systems Corporation, St Joseph, Mich). Blood pressure data were accurately synchronized with the polysomnogram for later analysis. (3) Oxygen saturation was measured by a pulse oximeter (Ohmeda Biox 37(0) and recorded onto the chart recorder. (4) For nasal mask pressure, a 8exible line was attached to one of the ports on the nasal mask (Respironics Inc, Monroeville Pa) and connected to a pressure transducer, the output of which went to the chart recorder.

Protocol Patients were acclimatized to the experimental conditions for 30 min before starting. They wore the nasal mask throughout the experiment with CPAP delivered from a Respironics Sleepeasy 11 and set initially at minimum pressure (1 to 2 em H.O). Recordings were made in the following order: (I) 30 min of maintained wakefulness in a well-lit room; (2) 30 min of established cyclical OSA; (3) 30 min of cyclical OSA with supplemental oxygen (OSA + OJ added via the second mask port to maintain oxygen saturation above 96 percent at all times; and 30 min CPAP at a pressure just sufficient to completely abolish all apneas, hypopneas, and snoring. The patient was then woken and blood pressure was recorded for a further 10 min. Blood pressure was also measured at the start and end of the experiment with a conventional mercury column sphygmomanometer. To avoid the confounding effects of different sleep stages on blood pressure, only periods of non-REM sleep were used in the subsequent analysis. Analym First the blood pressure values obtained by Finapres and cull' at the start and end of the experiment were compared by analysis of variance and the Duncan multiple range test, using the SAS suite of statistical programs.' The experimental data were examined in two ways. First, the mean ofthe beat-ta-beat blood pressure values, and their coefficients of variation (CV = lSD/mean) x 1(0), over the 3O-min experimental periods (wakefulness, OSA, OSA+O., and CPAP) were ca1cull!~ed and compared by analysis of variance with the Duncan multiple range test. Differences in sleep stage between OSA, OSA + 0., and CPAP and in the apnealhypopnea indices between OSA and OSA + O. were compared for statistically significant differences in the same way. The second approach involved selecting and analyzing individual apneas with their subsequent blood pressure rise. Twenty apneas of various lengths (ten from the period with OSA, matched as far as possible for length with ten from the period of OSA + OJ from each of the eight patients (total = 160) were examined in detail to determine the following: (1) the maximum systolic pressure follOwing apnea termination, which was normalized by expressing it as a percentage rise above the overall mean systolic blood pressure throughout the experimental period (excluding wakefulness), henceforth SBPmax; (2) the mean systolic blood pressure during the first 10 s immediately after the end of each apnea, also normalized as in step 1 above, henceforth SBPten; (3) the same procedure was followed for diastolic blood pressure to produce the two variables DBPmax and DBPten; and (4) apnea length, minimum oxygen saturation, actual percent fall in oxygen saturation from baseline, maximum pulse rate rise above the overall mean (of the whole experimental period excluding wakefulness). In addition, the maximum pulse rate in the postapneic period minus the minimum pulse rate during each apnea (le, the pulse rate rise with apnea termination, henceforth pulse change) and the length of ventilation from the resumption of respiration to the beginning of the next apnea were also calculated for each apnea, The interrelation among the variables was first examined by

FIGURE 1. Tracing from a patient with OSA, showing pulsus paradoxus and the postapneic rise in blood pressure. Pearson's single correlation coefficients and then by multiple linear regression (stepwise option), with SBPmax, SBPten, DBPmax, or DBPten as the dependent variable. The data for each individual were first examined alone in this way and then combined for the overall analysis. RESULTS

Figure 1 is a tracing from a patient with OSA showing the increasing pulsus paradoxus developing toward the end of each apnea with the progressively greater inspiratory efforts, and the postapneic rise in blood pressure. Finapres vs Cuff Blood ~ssure

The values obtained for arterial blood pressure, systolic and diastolic, were significantly lower when taken by Finapres than by cuff (p =0.032 and 0.001, respectively, see Table 1). There was no significant change in the measured arterial pressures from the start to the end ofthe experiment with either method.

Average Data Results for the overall mean blood pressures and pulse rates during the four study periods and their mean coefficients ofvariation (CVs) are shown in Table 2. Mean systolic and diastolic blood pressures were not significantly different during wakefulness, OSA, OSA + O2 , or CPAP. The CVs, however, were signifiCHEST I 101 16 I JUNE. 1992


Table 1-Clinical and Anthropomorphic Data ofSubjecta, Including Cuffand FifItJfJ'U Blood Preaure Meaaurementl* Subject! Age, yr

Weight, kg

BMI, kglm 2

cmH 20


>4% SaOl dipslb































7t155 8/60

107 87

48.2 31.9

7.5 8

50 52

Means, 52





Cuff BP S, E, S, E, S, E, S, E, S, E, S, E,

Finapres BP

85155 89155 125185 122188 120165 122/64

124/16 130182 122195 125192 129185 127/82 112'16 11M2 130190 125186 120158 11M2

105160 106159 117n8 119168 122/66 115152

.. .i

.. .i

S, 110/68 E, 112'68 S, 121.M8.2 E, 120.7n9.1

100/58 100154 110.6166.7 110.4/62.8

·S = start of study; E = end of study. tFemale. iData lost.

cantly higher during OSA and OSA + O 2 compared with wakefulness and CPAP (all p
Individual Data Figure 3 shows the mean systolic and diastolic blood pressures actually during the apnea; the SBPmax, SBPten, DBPmax, and DBPten following the apnea; and the mean apnea length of the 20 apneas selected for detailed analysis. The mean minimum saturation

and range for the ten apneas without added oxygen are also tabulated. The presence or absence of hypoxemia had no significant effect on the systolic and diastolic blood pressures actually during the apnea. Hypoxemia also had no significant effect on the higher postapneic blood pressures. Where blood pressures were higher in one or other condition (eg, subjects 2 and 5, Fig 3), this tended to mirror the differences in apnea length. Overall, there was no significant difference in the length of apneas selected for detailed multivariate analysis during the OSA period (mean length, 25.6 s; SD 15.3) and during the OSA + O 2 period (mean length, 27.4 s; SD 15.9), as would have been expected due to our attempt to match this variable in the two situations. Multiple Linear Regression

With SBPmax as the dependent variable, the individual correlations (r) and multiple linear regression modeling (r2) are shown in Table 3. Individually, apnea

Table 2-Mean Overall Blood Preuure, Pulse Bate, and Their Coefficienta of Variation during Wakefulnea, OSA, OSA + 0 1 , and CPAP* SBP

Wakefulness OSA OSA+02 CPAP


Pulse Rate










111.8 111.4 113.0 109.4

(16.1) (15.0) (17.5) (16.9)

5.5 10.7t 9.9t 5.6

62.5 59.9 61.3 62.3

(11.2) (11.9) (9.4) (10.1)

5.5 10.3t 9.6t 5.7

73.5 70.3 69.2 68.4*

(11. 7) (10.9) (11.2) (10.6)

4.3 5.4 5.2 2.81

·SBP = systolic blood pressure; OBP = diastolic blood pressure; CV = coefficient of variation; OSA = obstructive sleep apnea; OSA + O 2 = obstructive sleep apnea with added oxygen to abolish hypoxemia; and CPAP = continuous positive ailway pressure. tOSA and OSA + O 2 significantly higher than wakefulness or CPAP (p

CPAP and 0, during 0bstructNe Sleep Apnea (Ali et 81)

.changes explained a further 8 percent of the variance in SBPten, while desaturation explained only 5 percent of the variance in DBPmax.

70 taJ



ti Q.






~ en x. ~


The patients we studied are typical of those with severe OSA; they were predominantly obese, middleaged, and male. They were taking no medications. The Finapres device, an infrared volume clamp photoplethysmograph mounted in a finger cuff allows continuous beat-to-beat monitoring of arterial blood pressure and pulse rate. The device has been studied extensively and although absolute blood pressure has been found to be lower than that obtained by radial intra-arterial measurements, it accurately tracks changes in blood pressure. 5 •7 Although the Finapres has not been compared directly with invasive measurement ofarterial blood pressure in OSA, it has been studied in conditions where there are rapid changes in blood pressure and peripheral vascular tone and shown to be reliable8 and we are thus confident that our measurements are valid. The Finapres measures digital artery pressure and therefore any changes in hand position in relation to the heart will give rise to hydrostatic pressure differences that will be reflected in the measured digital arterial pressures. 5 In this stud~ patients remained in a supine position throughout the experiment and bandages were used to secure the hand in a fixed position relative to the heart and avoid this potential error. The hand was fixed across the abdomen at the level of the sternal angle, which in obese subjects is 10 to 15 cm above the brachial artery, where the blood pressure was measured with a sphygmomanometer. This and the tendency of the Finapres to underestimate central arterial blood pressures are the likely explanations for the lower blood pressures we observed with the Finapres compared with the cuff measurements at the start and end of the experiment. REM sleep is associated with variability in systemic blood pressure and we therefore excluded these periods . from the analysis. 9


~ Q





Q. t&.

0 &-e

20 10



o_ _





•• significantly less wakefulness in CPAP than OSA (p
length, pulse rate rise above mean, pulse change, minimum saturation, and fall in saturation are all significantly correlated with the rise in SBPmax. Following multiple linear regression, however, only apnea length and pulse rate rise above mean remain highly significantly correlated with the SBPmax. Forty-three percent of the variance in SPBmax is thus explicable by apnea length and pulse rate changes. Analysis of subjects' results individually revealed that apnea length completely displaced hypoxemia as the chief determinant of SBPmax in five of eight patients. Desaturation was an independent determinant in only two of eight patients. Table 4 summarizes the multiple linear regression analysis with SBPten, DBPmax, and DBPten as the dependent variables. Apnea length remained the most important explanatory variable for all three. Pulse rate

Table 3-Percent Change in Poatapneic SyBtolic Blood PraauR (MtJDmum): lndioidual Correlationa Grad Multiple lMaeGr Begreaion Motklirag* Multiple Linear Regression

Individual Correlations

Apnea length Pulse rate rise above mean Pulse change Minimum 0 1 saturation Desaturation Postapnea ventilation time



0.53 0.38 0.16 -0.23 0.20 0.15

<0.0001 <0.0001 0.033 0.008 0.022 O.09NS


0.28 0.15


<0.0001 <0.0001 >0.5 NS >0.5 NS >0.5NS >0.5 NS

*r =individual correlation; and r& =multiple linear regression modeling. CHEST I 101 I 6 I JUNE, 1992


_ 160


Ninsal (mean)=85~ Range = 80 - 90';


~ 140



E 120

120 -



100 80 taJ g: 60 § 40 ~ 20 ::> ~

80 60 40


o ~"~~DIP~-::....... ~~~~---L.-..-L...J..---.J_..L..18P\~ ~:h DUrwtC APNEA

160 140 -

3 10~
















Mansal (mean)=86'; Range =84-88~



+----...__ SlIP










D...... POST - APNEA




11 Api••

Mmsal (mean)=81 '7. Range =72-90'7. ,--







Il 64


40 20







_ _L....L.._

"Insel (meen).85~ Ranle =79-92'7.

__L..-Io-_ _-""""_ _I.-...L..---'.....L-_ _. . L . . - _ - - L -



7 "ansal (mean)=89'7. Range =86-93~

S8PInu SIPte. DePm.. o.te. POST - APNEA







Mlnsal (mean)=92~ Rente =89-94~


120 100


60 40 20





160 140


N ansa l ( mean ) - 81~ Ranee = 72 - 90'7.






100 80


160 140 120


160 140 120 -

100 80 60





Wlnsal (mean): 87~ Range =84- 90~


120 100 80 40



60 40 20










SlP\en DIP..... DBPtea POST - APNEA



FIGURE 3. Individual blood pressure during and after apnea termination, apnea length, and O. desaturation data used in multiple linear regression calculations. Aplen = mean apnea length in seconds.

Average Data

The first analysis, looking at the mean values (systolic and diastolic), showed no differences between wakefulness, OSA, OSA + O 2 , and CPAE However, the oscillations in blood pressure (represented by the CV) were clearly higher in the two OSA periods. Although the apnealhypopnea index was higher during OSA than OSA + O 2 , the CVs of blood pressure were not significantly different, suggesting that hypoxemia does not play a significant part in the generation of the blood pressure swings. This finding, apparently at variance with the published literature, is discussed later with the more detailed results from multiple linear regression analysis. 1530

The finding of no fall in mean systolic and diastolic blood pressure following the institution of CPAI: despite the abolition of the swings in blood pressure and deeper sleep, is surprising. Normal sleep is associated with a fall in blood pressure developing over the first 1.5 to 2.5 h. 9 Since we examined only the first 30 min of CPAP therapy, we may have missed a later fall in blood pressure. The raised catecholamine levels found in patients with OSAIO may take a while to return to normal, although the time course of this fall is not known. An alternative explanation is that the large negative pleural pressures that develop during the obstructed inspiratory efforts reduce the mean arterial pressure during OSA. Thus, mean CPAP and Or dwing 0b8tructNe Sleep Apnea (Ali et 8/)

Table 4-Summary of Multiple Linear Regreuion AnalyBiBfor Dependent Variables SBPten, DBPmax, and DBPten*

Apnea length Pulse rate rise above mean Pulse change Minimum O 2 saturation Desaturation Postapnea ventilation time Total model r2 =



SBPten r2






0.16 0.06 0.02

<0.0001 0.0005 0.033










*rI = multiple linear regression modeling.

peripheral blood pressure during OSA remains the same, as it contains values both higher and lower than during CPAP therap~ Multiple Unear Regression As in the above analysis, blood pressures during the 160 apneas selected for detailed analysis were not significantly different regardless of whether hypoxemic or not.Thus, hypoxemia does not play an important role in the generation of the blood pressure swings seen in OSA. Although single regressions of the SBPmax vs apnea length, pulse rate changes, and measures ofhypoxemia were all significant, only apnea length and the pulse rate rise above mean proved to be independent predictors after multiple linear regression. Forty-three percent of the variance in postapneic systolic blood pressure rise was explicable on the basis of these two variables alone. Thus, like Shepard,2 we found that blood pressure rise was apparently correlated with the degree of hypoxemia developing during an apnea, but by using multiple linear regression to dissect out the principal variables, we found that this was probably due to its own dependence on apnea length. The persistence of simila~sized blood pressure swings despite correction of hypoxemia found in the first analysis confirms this. In an early study, Schroeder et all reported that oxygen administration blunted the blood pressure swings seen during OSA. The disparity between this finding and our study may be explained by that article's small study group of only four patients (one of whom had central sleep apnea). The pattern offactors predicting SBPten was similar to those factors for SBPmax, although less of the variance is explicable by the two variables, apnea length and pulse rate rise above mean, and a small contribution is made by pulse change (difference between minimum pulse rate during the apnea and maximum pulse rate after apnea termination). Hypoxemia did appear to independently predict the size of DBPmax, although it contributed considerably less than did apnea length. For DBPten, the results were similar to those for systolic blood pressure, with apnea length and pulse rate changes being

the only independent predictors. It is not clear what the mechanisms relating apnea length to the subsequent rise in blood pressure are, but it is likely to be due in large part to arousal from sleep. Waking normal subjects from sleep increases blood pressure transiently. II We have reported the case of a patient with recurrent arousals from sleep caused by periodic movement of the legs during sleep that was associated with rises in blood pressure, similar to those observed in patients with OSA.12 Perhaps the longer the apnea the more established and deeper is sleep and consequently the "greater" the arousal and the larger the blood pressure rise. Alternatively, a long apnea may mean that greater arousal forces exist at the point of breaking (eg, ventilatory drive) and these may influence the size of the postapnea blood pressure rise. Such a hypothesis is difficult to test at present as the criteria that would allow quantification of the magnitude of an arousal are lacking. As we did not measure the rise in arterial PaC02 during the apneas, we cannot be sure that this is not an important determinant of the postapneic blood pressure rises. The fact that PaC02 reaches a virtual plateau at the mixed venous level within 30 s ofbreathholding would mean that the pressor response to rising CO2 would not have been expected to continue to produce larger blood pressure rises for apneas of 90 s than for apneas of 30 s. The cardiovascular effects of hypercapnia have not been studied in sleep, although in awake subjects, blood pressure changes very little during the first minute of breath-holding. 13 Thus, it seems unlikely that hypercapnia is an important determinant of the postapneic blood pressure rise. It may be argued that the inspiratory efforts associated with obstructive apneas or with the resumption of ventilation I lead to aspiration of venous blood into the heart (by the so-called respiratory pump), which would increase cardiac output and therefore blood pressure by the Frank-Starling mechanism. Accumulating evidence, however, suggests that in the supine posture, and in the absence of a raised central venous pressure, the respiratory pump may not operate (the opposite may indeed be true), negative intrathoracic pressure leads to virtual occlusion of the inferior vena CHEST I 101 I 6 I JUNE, 1992


cava just below the level of the diaphragm and a reduction in venous return. 14 This suggested mechanism is therefore unlikely to be important, especially as similar blood pressure rises are seen following central apneas l and periodic leg movements,12 where there are no large pleural pressure swings. In man the fall in cardiac output during the diving reflex is mediated by a fall in pulse rate while stroke volume is unchanged}S In this stud~ pulse change (which takes into account vagally mediated cardiac slowing as well as the postapneic cardioacceleration) was not an important determinant of SBPmax, DBPmax, or DBPten and made only a small contribution to SBPten. It is unlikely therefore that release of the diving reflex associated with apnea termination plays a significant part in generating the blood pressure swings. However, the rise in heart rate above the overall mean did seem to contribute to the postapneic rise in systolic blood pressure. This postapneic cardioacceleration may simply he an indirect marker of the degree of arousal. In summary, we have shown that mean systemic blood pressure is unchanged between wakefulness, OSA with and without added oxygen, and short periods of CPAE and we have discussed some of the possible reasons for this. We have also demonstrated that the postapneic rises in blood pressure (except for DBPmax) are not dependent on the development of hypoxemia, but relate to apnea length and pulse rate changes and have suggested that these variables may reflect the magnitude of the arousal from sleep.

2 3


5 6 7 8

9 10 11 12 13 14

ACKNOWLEDGMENT: Dr. M. Goldman kindly provided us with the initial software used in the analysis of the Finapres arterial recordings.



in sleep apnea. In: Guilleminault C, Dement WC, eds. Sleep apnea syndromes. New York: Alan R Liss Inc, 1978; 177-96 Shepard JW Gas exchange and hemodynamics during sleep. Med Clin North Am 1985; 69:1243-64 Huelscher TJ, Vaughan McCall ww, Owell J, Marsh GR, Erwin CW: Two methods of scoring sleep with the Oxford Medilog 9000: comparison with conventional paper scoring. Sleep 1989; 12:133-39 Rechtschaffen A, Kales A, eds. A manual of standardized terminology, technique and scoring system for sleep stages in human sleep. Los Angeles: UCLA Brain Information Service/ Brain Research Institute, 1968 Van Egmond J, Hasenbos M, Crul JF. Invasive v. non invasive measurement of arterial pressure. Br J Anaesth 1985; 57:434-44 Cody ~ Smith JK. Applied statistics and the SAS programming language, 2nd ed. New York: Elsevier, 1987 Smith NT, Wesseling KH, de Wit B. Evaluation oftwo prototype devices producing non invasive, pulsatile, calibrated blood pressure from a finger. J Clin Monit 1985; 1:17-29 Parati G, Casadei R, Groppelli A, Di Rienzo M, Mancia G. Comparison of finger and intra-arterial blood pressure monitoring at rest and during laboratory testing. Hypertension 1989; 13:647-55 Snyder F, Hobson jA, Morrison DF, Goldfrank F. Changes in respiration, heart rate and systemic blood pressure in human sleep. J Appl Physiol 1964; 19:417-22 Fletcher EC, Miller J, Schaaf J~ Fletcher JG. Urinary catecholamines before and after tracheostomy in patients with obstructive sleep apnea and hypertension. Sleep 1987; 10:35-44 Katri M, Freis ED. Hemodynamic changes during sleep. J Appl Physioll967; 22:867-73 Ali NJ, Davies RJO, Fleetham J, Stradling JR. Periodic movements of the legs during sleep associated with rises in systemic blood pressure. Sleep 1991; 14:163-65 Hong SK, Lin YC, Lally B, et ale Alveolar gas exchanges and cardiovascular functions during breath holding with air. J Appl Physiol1971; 30:540-47 Griffiths DJ. Principles of 80w through collapsible tubes: venous haemodynamics. In: Gardner AMN, Fox RH, eds. The return of blood to the heart: venous pumps in health and disease. London: John Libbey and Company Ltd, 1989; 115-26 Widdowson AN, Chapman BJ. Changes in cardiac output during the diving re8ex in man. Clin Sci 1990; 79(suppl23):17p

1 Schroeder JS, MottaJ, Guilleminault C. Haemodynamic studies


CPAP and O. during 0b8tructNe Sleep Apnea (All et aI)