Sighs During Sleep in Adult Humans

Transcription

1 Sleep. 6(3): Raven Press. New York Sighs During Sleep in Adult Humans Rogelio Perez-Padilla, Peter West, and Meir H. Kryger Department of Respiratory Medicine. St. Boniface General Hospital, Winnipeg, Manitoba, Canada Summary: We analyzed sighs (breaths with a tidal volume at least twice that of baseline breaths) during sleep in 12 normal adults. We found a total of 124 sighs in the group, with an average of 1.66 sighs/h of sleep, but with great intersubject variation (range: 1-25 sighs/night). There were sighs in all sleep stages, but there were more per hour in stage % of the sighs were associated with an increase in EMG activity or EEG frequency, starting either before or immediately after the sigh. The remainder of the sighs were not associated with any arousal or sleep stage changes. The normal variability of heart rate with breathing is exaggerated during sighs, probably because of the greater inflation and the associated arousal. Sighs have larger mean inspiratory flows (Vt/Ti), expiratory flows (Vt/Te), and a larger fraction of respiratory cycle spent in inspiration (Ti/Ttot) than the previous breaths, all evidence of a change in respiratory control. Sighs during sleep may occasionally be followed by central apneas, hypoventilation, or considerable slowing of respiratory rate. Although it has been shown that a sigh renders the respiratory centers refractory to another sigh, we found that sighs sometimes occur in pairs. Key Words: Sleep-Respiratory pattern-sighs-apnea-heart rate Arousal-Respiratory control. Normal respiration of most mammals includes breaths with a tidal volume (Vt) considerably larger than average (1,2). These large breaths, sighs, are considered an important component of normal breathing (3). It is well known that pulmonary compliance and functional residual capacity increase after a sigh (4-8), and this may be the result of the opening of collapsed alveoli (5,6). Atelectasis occurring when sighs are reduced may worsen gas exchange, producing VIO. inequality or increase in shunt (4,7). In animals, an inspiration-augmenting reflex, thought to be mediated through rapidly adapting vagal receptors, appears to be the main basis for the sighs (9). It is also known that the frequency of sighs is modified by hypoxemia, hypercapnia, several drugs, cutaneous stimuli, and, in humans, behavioral factors (3,10,11). All these factors are evidence of the complex regulation of sighs. This complexity may be greatest in humans because of the development of forebrain structures and control mechanisms. Sleep is associated with important changes in ventilatory control and with inhibition of many somatic and autonomic reflexes, factors that could affect the occurrence of sighs. Isolated sighs during sleep have been reported in rats 0), kittens (2), and Accepted for publication May Address correspondence and reprint requests to Meir H. Kryger, M.D., Department of Respiratory Medicine, 2C St. Boniface General Hospital, Tache Avenue, Winnipeg, Manitoba R2H 2A6, Canada. 234

2 SIGHS DURING SLEEP 235 humans (13). We wondered how often humans sigh during sleep, if sighing is associated with a particular sleep stage, and if the breathing pattern is altered after a sigh during sleep, as happens in anesthetized animals. METHODS The subjects were six men and six women, all healthy, age 25 to 34 (mean 28.5). All slept in the laboratory for two consecutive nights; the recording of data was done during the second night (Grass Instruments model #78D polygraph). For determination of sleep stage, the C3-A2 and C4-Al leads of the electroencephalogram (EEG), the electro-oculogram (EOG), and the mental electromyogram (EMG) were recorded. Other physiological variables recorded were arterial oxygen saturation (Hewlett-Packard oximeter model #47201A), electrocardiogram, beat-by-beat heart rate, relative thoracoabdominal cross-sectional area changes (Respiratory Inductance Plethysmograph, Respitrace, Ambulatory Monitoring Inc.), expired CO 2 as an indicator of air flow at the nose and mouth (Hewlett-Packard Capnometer model #472 loa), and breath sounds (Elema-Schonander surface microphone, model #EMT25B). The staging was done according to conventional criteria, except that we used a 15-s moving window rather than a page-dependent epoch (14). For this study we monitored the thoracoabdominal changes in volume with a respiratory inductance plethysmograph (RIP). To confirm the detection of large breaths with a RIP, we examined the relationship of Vt measured by pneumotachygraph and by single belt, 15-cm wide, placed on the upper abdomen and covering the lower ribs. A total of 182 breaths with Vt ranging from 0.2 to 3 L were examined, in three positions (supine, prone, lateral). In a given position, the relationship was linear (mean r = 0.95 ± O.OI-SEM) and the scatter small (SEM/mean volume = ± 0.007). Change in position resulted in a new linear relationship. Thus a single upper abdominal RIP belt accurately detects large breaths in a given posture. The inspiratory time (Ti), expiratory time (Te), and cycle duration (Ttot) measured with the RIP were equal to those measured by the pneumotachygraph in the three positions and in all ranges of tidal volume. We examined the 12 breaths before and after each sigh, comparing the relative volume changes. This type of comparison is valid if the body and belt positions remain fixed during the breaths studied. In the absence of body movements changes in the RIPvolume relationship within the 25 breaths are unlikely. Because all the sighs reported were associated with air flow at nose or mouth, we discarded the possibility of registering, as true breaths, abdominal volume changes with respiratory efforts during upper airway obstruction. The RIP record contained segments during which breathing was regular (SD of Vt < 10% of mean value), as well as periods during which tidal volume was irregular (SD of Vt > 10% mean) or displayed cyclic increases and decreases of regular periodicity (periodic breathing). In addition, there were apneic periods during which the RIP record was stable and there were no fluctuations in CO 2 concentration at the nose and mouth. Movement artefacts were evident as large swings in the RIP record associated with high frequency and amplitude in EEG, EMG, and EOG activity. For the sake of defining a sigh, we eliminated periods of movement artefact and big breaths following an apneic period. For the remaining segments a sigh was defined as a tidal volume at least twice as large as the mean Vt during regular breathing, or twice as large as the largest breath in the preceding 30 s in the case of irregular or periodic breathing. Two SI!'cp. Vol. 6.. Vo. 3.!91!3

3 236 R. PEREZ-PADILLA ET AL. typical examples are shown in Figs. 1 and 2. We measured cycle duration (Ttot), inspiratory time (Ti), expiratory time (Te), and RIP amplitude (as an indicator of Vt) of the sigh, the three breaths preceding the sigh (pre 1-3), and the three following the sigh (post 1-3). We used the mean Vt, Ti, and Te of the pre 1-3 as our baseline measurements. For our calculations we normalized the volume and timing measurements of the sigh and post-sigh breaths by expressing them as a ratio to the baseline values. By this procedure all our baseline measurements equal 1. We classified sighs as single or double. Within the single sighs we identified those with an augmented pattern (11). In these latter big breaths, sigh onset occurs during the inspiration of an apparently normal breath. This type of sigh is detected by an abrupt increase in flow during inspiration. We examined the changes in EEG, EMG, EOG, heart rate, and oxygen saturation of hemoglobin associated with these sighs. We recorded the mean heart rate (HR) during the 12 breaths before and after the sigh, the variation in HR associated with respiration before the sigh, the HR at the start and at the end of the sigh, the peak HR during the sigh, and the minimum HR associated with the sigh (which occurred either during the sigh or within 5 s after the sigh). Arousal was defined as a transient period of increased EMG activity associated with a change in activity in any of the EEG channels (a decrease in amplitude, increase in frequency, or appearance of alpha rhythm, but without artefacts due to body movements). We do not have sufficient information about awake supine breathing, owing to the short sleep latency in most of our subjects. Statistical analysis of the data included two-way analysis of variance and Duncan's multiple range test. RESULTS Characteristics of sleep Sleep duration was within normal limits, with a mean sleep time of 381 ± 56.4 (SD) min and a normal distribution of sleep stages (Table 1). Incidence of sighs There were 124 sighs during sleep in the whole group, with great intersubject variability (1-25 per night). The mean number of sighs per subject was ± 7.85 (SD) per night, and 1.66 ± 1.36 per hour of sleep. One hundred fourteen sighs (92%) were single and 10 (8%) double (Figs. 1 and 2). Only 5 sighs (4%) had the typical augmented pattern. Distribution of sighs in sleep stages (Table 1) We found sighs in all sleep stages. We calculated the "expected number" of sighs in each sleep stage according to the distribution of sleep stages during the night. We TABLE 1. Distribution of sighs during sleep Stage 1 Stage 2 Stage 3-4 Stage REM % Sleep time % Sigh SO Sighslscoreable hour in each sleep stagel person Distribution differs from the sleep stage distribution by X 2 test (p < 0.01) Sleep, Vol. 6, No.3, 1983

4 SIGHS DURING SLEEP 237 EMG EEG (C3-A2) ",i,,'''hni\vi<i'''''i<''''rl'r(\'''"'''' '':wa; jilv"vy'i': II'I"':V'",",, """".'..,I :,...,:,:,,/,,: EEG (C4-+ 1) I II.,Vf'vMJr..."".,,\\I/""I/\V:\III".,,\\..'l.r:\.,''vv-\'''''',')'' '''-'''-i'- ''''''..JI.J-Iw,,:.'?; "': EOG, i. """"'I YA."'-./I,,'.I""',NV,'\.\.''''''''''AI... J'NW,J.V'V.,''''''...,. ;...,.-...,...'...I.,,;,',';../ WJlliU.lli.JJ.JJJJJltlJ,j,!J;j... l..!.j.j.!. EXPIRED l/vv-- RESPITRACE 2 sec FIG. 1. Single sigh in stage 2 sleep, followed by a central apnea. compared the expected number with the observed number of sighs in each sleep stage by the X 2 test. The expected and observed distributions are significantly different (p <.01). This is mainly a result of an increased frequency of sighs in stage 1. Associated changes in EEG or EMG A transient increase in EEG frequency was observed to precede sighs in 33.1 % of the cases and to follow sighs in 31.4% of the cases. These transient changes occurred within 5 s of the sighs. In the remaining sighs (35.5%), changes in EEG or EMG were not detected. During sleep there was an average of 47.2 arousals/subject; of these, 6.7 arousals/subject were associated with a sigh. Since the occurrences of sighs (about to/night) and arousals (47/night) were both rare during the night, if the relationship between them was random one would expect the coincidence of sighs and arousals to be a very rare event. We estimated that the random coincident occurrence of sighs and arousals would be only 0.09 time per night. In fact, 6.7 arousals per night were associated with sighs. Thus the simultaneous occurrence of arousal and sigh was 74 times more frequent than one would expect on the basis of chance alone (p <.01, X2 test). Comparison between single sighs and the breaths around them (Table 2) By definition, sighs have significantly higher relative volume when compared with all other breaths. The mean Ti of the sighs was longer than Ti of baseline (pre 1-3) and post 1-3 breaths. Sigh Te was longer than Te of baseline breaths, but not longer than the post 1-3 breaths. The mean inspiratory flow (VtlTi) has been used as a measure of inspiratory drive (15). We calculated the relative VtlTi of the sighs by using the normalized volume and Ti as described above. The relative Vt/Ti of the sighs is significantly higher than the Vt/Ti of the baseline (Fig. 3). VtlTi is also higher in sighs than in the post 1-3. The relative mean expiratory flow (VtlTe) of the sighs (calculated Sleep. Vol. 6, No.3, 1983

5 238 R. PEREZ-PADILLA ET AL. EMG FIG. 2. Double sigh in stage 2 sleep, with an associated arousal (increase in EMG activity and change in the frequency and amplitude of EEG). Note that the changes in heart rate precede the sigh and are more closely related to the onset of the arousal. EXPIRED C02 I\(\/\/' V\(\//\V/' as above for Vt/Ti) is also greater than the baseline and the post 1-3 Vt/Te. Ti/Ttot of the sighs was also significantly greater than Ttot of the baseline and post 1-3 breaths (Fig. 4). Comparison between breaths before and after the sigh There were no significant differences between the Vt of the breaths immediately preceding and following the sigh. Ttot preceding the sigh was significantly shorter than Ttot of the post-sigh breaths. Baseline Ti was longer, while Te was shorter than post 1-3. Vt/Ti in the breath immediately after the sigh was significantly higher than baseline and the other post-sigh breaths. Furthermore, the Vt/Te of the first post-sigh breath was greater than those of the second and third, but not different from baseline. Ti/Ttot was greater in the breaths before the sigh than in subsequent breaths. Effect of sleep stage, sex, and intersubject variation in sigh characteristics There was intersubject variation in Ti of the sigh, but not in Te, Ttot, Vt/Ti, Vt/Te, or Ti/Ttot (all expressed as a ratio to the baseline value). For the same variables, the only sex difference was in the relative VtlTi of the sighs, which was larger in males. No sleep stage-related differences were found in the characteristics of the sighs. Sao2 and heart rate changes associated with single sighs (Table 3) There was never a change in Sao2 of more than 2% before or after the sighs. There were significant increases and decreases in HR compared with the mean HR before the sigh. There were greater increases in HR after a sigh during sleep stages 1, 4, and REM than in 2 and 3. In general, onset of the increase in HR was closely related to the onset of the sigh (68.5%). In the remaining cases, the increase in HR preceded the sigh, and was usually associated with an arousal. On the other hand, the minimum HR occurred after the Sleep, Vol. 6, No.3, 1983

6 SIGHS DURING SLEEP en 1.0 z ; w >..J 0.5 w a: SIGH BREATHS "--. INSPIRATION. "-.. BASELINE --. EXPIRATION o FIG. 3. Relative flow rates in sighs and after sighs. Relative flow is relative volume divided by time in seconds. Time was not normalized (with base line = 1) in this figure, as it was in Table 2. This was done to show differences in inspiratory and expiratory flow rates. Expiratory time exceeded inspiratory time, hence the lower expiratory flow rates. There is a marked increase in flow rates in both inspiration and expiration of the sigh. (Asterisk indicates significant difference from baseline by two-way ANOYA, p < 0.05; n = 114 single sighs). sigh had finished in 68.5% of cases, most frequently during the expiration of the first post-sigh breath. In 24% of the sighs, the minimum HR was associated with an EEG or EMG return to a sleep pattern after an arousal. The variation in heart rate associated with normal breathing before a sigh (a HR) was 6.2 ± 0.28 beats/min, whereas the variation in HR associated with a sigh was significantly higher (21 ± 0.42 beats/min). Respiratory arrhythmias associated with sighs Fifteen of the sighs (12%) were immediately followed by an important alteration in respiratory rhythm (Fig. 1): 6 (4.8%) by central apneas longer than 5 s (mean duration 12 ± 2.4 s-sem), 4 (3.2%) by a reduction in Vt to less than 25% of the base line (mean duration 13.7 ± 2.5 s-sem), and 5 (4.0%) by a decrease in respiratory rate (RR) greater than 25% (mean duration 19 s, mean percentage decrease in RR 37%). Central apneas were closely related to sighs. If the incidence of central apneas after normal breaths was the same as after sighs (4.8%), we would expect about 240 central apneas/subject per night, whereas the number of central apneas/subject was only 4. I (p <.01 by X 2 test). The respiratory arrhythmias were not associated with significant decreases in Sao2; the greatest drop was 1%, after the longest apnea of 20.5 s. FIG. 4. Respiratory timing before and after sighs. Sighs are associated with a significant increase in Ti/Ttot, whereas the breaths after a sigh have a reduced Tiff tot. (Asterisk indicates significant difference from baseline by two-way ANOYA, p <.05; n = 114 single sighs).... fe... i= SIGH BREATHS AFTER SIGH 1st 2nd 3rd ' Sleep. Vol. 6. No

7 ;;- '".0-.'" TABLE 2. Characteristics of single sighs a Ti Te Ttot Volume Vt/Ti Vt/Te TiiTtot b Sigh 1.46 ± 0.04 e 1.12 ± 0.04 e 1.32 ± 0.03" 2.62 ± O.lO e 1.84 ± 0.08 e 2.54 ± 0.15 e 0.44 ± 0.01" Post I 0.92 ± 0.02 e 1.10 ± 0.04 e 1.01 ± ± ± 0.04 e 1.10 ± ± e Post ± O.Ole 1.15 ± 0.03 e 1.06 ± ± ± ± ± e Post ± 0.02; 1.16 ± O.OY 1.07 ± ± ± ± ± e a Means ± SEM from 114 sighs. Except for Ti/Ttot, the values in the table are ratios with the base line (mean of the 3 breaths before the sigh). The baseline values equal I. b Ti/Tot of the base line, 0.38 ± e Significantly different from the base line. TABLE 4. Double sighs a Tid Ted Ttot Volume VtlTi d Vt/Te d TiiTtot b d 1st sigh 1.57 ± 0.14e 0.82 ± ± (UO 3.19 ± 0.65 e 2.06 ± 0.32e 4.05 ± 0.65 e 0.53 ± 0.02 e 2nd sigh 1.34 ± O.13 e 1.22 ± ± ± 0.44 e 2.56 ± 0.33 e 3.05 ± 0.4g e 0.40 ± 0.02 a Means ± SEM from 10 double sighs. Except for TiiTtot, the values are expressed as ratios with the base line (mean of the 3 breaths previous to the sigh). b Ti/Ttot of the base line was e Significant difference from the base line. d Significant difference between the two sighs. N -I::.. C ::0 "'t:l Rl o t;: t"rl..., ;:t.. t"-<

8 SIGHS DURING SLEEP 241 TABLE 3. Heart rate changes associated with single sighs during sleep Heart rate Mean ± SEM Before sigh At start of sigh Peak during sigh End of sigh Minimum After sigh a p < (0.75) 64.5 (1.00) 76.0 (0.98)a 64.2 (1.28) 55.0 (0.73)a 62.0 (0.76) Double sighs (Fig. 2, Table 4) Both breaths in double sighs had higher Vt, Vt/Ti, Vt/Te, and shorter Ttot than single sighs. TiiTtot of the first double sigh was greater and from the second double sigh was smaller than Ti/Ttot of single sighs. Changes in EEG, EMG, and HR are also associated with double sighs. One of the double sighs was followed by a central apnea. On the average, the first sigh had the same Vt and Ttot as the second, but with longer Ti and shorter Te. Accordingly, the first sigh had a higher Vt/Te and Ti/Ttot and lower Vt/Ti than the second one. DISCUSSION Sighs have been described in most adult mammals (1-3). Ontogenically, sighing efforts appear almost coincidentally with the first respiratory movements in fetal lambs (16,17). The adaptive value of big breaths seems to be in preventing atelectasis (3,6). Gas exchange in anesthetized humans has been found to worsen after a period of time without hyperinflation; therefore, that normal breathing during sleep includes sighs is not surprising (4). Although all our subjects sighed during sleep, there was great intersubject variation in sigh frequency, as Bendixen et al. reported in awake humans (3). Sighs are present in all sleep stages, including REM, which is associated with loss of muscle tone and inhibition of several spinal and brain stem reflexes (18). The inspiration augmenting reflex, essential in the regulation of sighs in animals and probably in newborn humans, may be less important in adult humans, with the development of more sophisticated respiratory control mechanisms (7,8,17). Sighs due to the inspiratory augmenting reflex resemble control breaths in their early stages, and relatively few of the sighs we observed had this characteristic feature. The sigh frequency we found during sleep (1.66/h) is lower than that reported during wakefulness (lo/h). This may be partly due to the fact that we did not analyze time periods coincident with body movements, which may have an associated increased incidence of sighs. The frequency of sighing during wakefulness may also be influenced by behavioral factors, such as boredom (3). However, our findings of more sighs during light sleep and their strong association with arousals, gives some evidence that sighing is facilitated in higher arousal states. On the other hand, central nervous system state and afferent stimuli associated with sighs might in turn cause an arousal, a point of view supported by the appearance of arousals, changes of EEG frequency, or increases in muscle tone after some of the sighs. In dogs, passive inflation of the 'lungs does not cause arousals, but during a spon- Sleep. Vol. 6. No.3, /983

9 242 R. PEREZ-PADILLA ET AL. taneous sigh, the activity of the brain stem respiratory neurons or inspiratory muscle contraction may be involved in a neuroiogicai arousai (i9). During CO2 rebreathing a hyperbolic relationship between tidal volume and Ti and Te has been described in humans (20). As the Paco2 increases, Vt, Vt/Ti, and Vt/Te increase, while Ti and Te remain constant or are shortened (20,21). A similar relationship occurs when ventilation is stimulated by hypoxia (15). In cats, sighs have a larger Vt and Vt/Ti and longer Ti than the previous breaths, therefore lying outside the Vt/Ti plot during CO2 rebreathing (11). We found the same for sigh Vt/Ti and Vt/Te in humans during sleep. In awake humans, it has been reported that there is a slight decrease of breathing frequency after a sigh, which is attributed to an increase in lung compliance (3). In cats, there is an increase in frequency after the sigh (8). We found examples of both increases and decreases of RR after sighs, so that on the average there was no significant change (Table 2). Sighs may be associated with an increased risk of disordered respiratory rhythm. We found some episodes of apnea, hypoventilation, or slowing of frequency after a sigh. Apneas after sighs have been reported in dogs, in kittens, and in humans (12,13,22,23). In dogs, the apnea may be related to a decrease in Paco2 after the sigh, because it is abolished after CO2 rebreathing (23). However, reflex effects of the sigh might also be involved. The Ti/Ttot of the three breaths following the sigh is less than in the breaths before the sigh. This is further evidence that the pattern of breathing changes after a sigh. The change probably is more important in the first breath after sighs, which has greater Vt/Ti and Vt/Te than the second and third breaths. We do not have data about how long the changed breathing pattern remains after the third post-sigh breath. However, the changes in RR after a sigh reported in cats and in awake humans last for less than 10 breaths, which may be true also for the variables we measured (3,8). Also of interest is the shorter Ttot of the breath immediately preceding the sigh. Bendixen et al. found that the breath before a sigh may vary in amplitude or timing (3). Some of the breathing pattern changes occurring before and after the sighs may be due to changes in lung compliance; changes in neural activity in the lower brain stem may also occur. Eldridge and Gill-Kumar have demonstrated an afterdischarge of respiratory neurons following a transient increase in respiratory drive by chemoreceptor or cutaneous stimulation (24). This respiratory afterdischarge may occur even after very transient stimuli (4-7 s). A similar reverberating neural network stimulated during sighs might be a causal factor in the increased flows observed in the following breath, and also modify the breathing pattern. Heart rate oscillations with respiration are well known. This is mainly due to a vagal reflex thought to be mediated by lung receptors (25). HR changes more during a sigh than during a normal breath, because of the greater associated lung inflation. Arousals during sleep are associated with increases in HR, mainly through vagal inhibition (26). The changes in HR during a sigh are probably the result of an interaction between the lung inflation and the arousal with which it is frequently associated. In some cases, the minimum HR associated with the sigh frequently occurred during the expiratory phase of the sigh, but in the majority of cases it occurred during expiration of the next breath. This may be due to increased vagal tone because of expiration and deepening sleep. In general, double sighs were associated with the changes described for single sighs. Sleep, Vol. 6, No.3, 1983

10 SIGHS DURING SLEEP 243 The appearance of double sighs in our subjects is quite curious. It is well established that the occurrence of a sigh renders the respiratory centers refractory for development of another sigh (7,27). To our knowledge this is the first report of two consecutive sighs. In double sighs the TilTtot, Vt/Ti, and Vt/Te are different in the first and second sighs. The change in pattern could be due in some cases to the associated arousal response, but reflex effects of the first sigh or a change in lung compliance may also be important. Acknowledgment: This work was supported by the Medical Research Council, operating grant no. MA REFERENCES I. McCutcheon FH. The mammalian.breathing mechanism. J Cell Comp PhysioI1951;37: McCutcheon FH. Atmospheric respiration and the complex cycles in mammalian breathing patterns. J Cell Comp Physiol 1953;41: Bendixen HH, Smith GM, Mead l. Pattern of ventilation in young adults. J Appi Physioi 1964;19: Bendixen HH, Hedley Whyte J, Laver MD. Impaired oxygenation in surgical patients during general anesthesia with controlled ventilation. N Engl J Med 1963;269: Ferris Be, Pollard DS. Effect of deep and quiet breathing on pulmonary compliance in man. J Clin Invest 1960;39: Mead J, Collier C. Relation of volume history of lungs to respiratory mechanics in anesthetized dogs. J Appl PhysioI1959;14: Reynolds LB. Characteristics of an inspiration augmenting reflex in anesthetized cats. J Appl Physiol 1962;17: Reynolds LB, Hilgeson MD. Increase in breathing frequency following the reflex deep breath in anesthetized cats. J Appl Physiol 1965;20(3): Larrabee MG, Knowlton GC. Excitation and inhibition of phrenic motoneurones by inflation of the lungs. J Physiol (Lond) 1946;147: Bartlett D. Origin and regulation of spontaneous deep breaths. Respir PhysioI1971;12: Glogowska M, Richardson PS, Widdicombe JC, Winning AJ. The role of the vagus nerves, peripheral chemoreceptors and other afferent pathways in the genesis of augmented breaths in cats and rabbits. Respir PhysioI1972;16: McGinty DJ, London MS, Baker TL, et a!. Sleep apnea in normal kittens. Sleep 1979;1: Tabachnik E, Miller NL, Bryan CH, Levison H. Changes in ventilation and chest wall mechanics during sleep in normal adolescents. J Appl PhysioI1981;51(3): Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and scoring for sleep stages of human subjects. Washington, D.C.: Department of Health, Education and Welfare, 1968 (NIH Pub!. 204). 15. Remmers JF. Analysis of ventilatory response. Chest 1976;70(Suppl): Dawes GC, Fox HE, Leduc BM, Liggins GC, Richards RT. Respiratory movements and REM sleep in foetal Iamb. J Physiol (Lond) 1972;220: Thach BT, Taeusch HW. Sighing in newborn human infants: role of inflation augmenting reflex. J Appl Physiol 1976;41 : Phillipson EA. Control of breathing during sleep. Am Rev Respir Dis 1978;118: Sullivan C, Kozar L, Murphy E, Phillipson EA. Arousal, ventilatory and airways responses to bronchopulmonary stimulation in sleeping dogs. J Appl PhysioI1979;41(1): Gardner WN. The relation between tidal volume and inspiratory and expiratory times during steadystate carbon dioxide inhalation in man. J Physiol (Lond) 1977;272: Clark Fl, von Euler C. On regulation of depth and rate of breathing. J Physiol (Lond) 1972;222: Hoff HF, Breckenridge CG. Levels of integration of respiratory patterns. J Neurophysiol 1952;15: Reininger EJ, Segall P. Effect of inhalation of O 2 and CO 2 gas mixtures on spontaneous gasps and apnea in unanesthetized dog. Physiologist 1970(a);13: Eldridge FL, Gill-Kumar P. Central neural respiratory drive and afterdischarge. Respir Physiol 1980;40: Anrep GV, Pascual W, Rossier R. Respiratory variations in the heart rate. I. The reflex mechanisms of the respiratory arrhythmia. Proc R Soc Lond [BioI) 1936;119: Baust W, Bohnert B. The regulation of heart rate during sleep.. Exp Brain Res 1969;7: Cherniack NS, von Euler C, Glogowska M, Homma I. Characteristics and rate of occurrence of spontaneous and provoked augmented breaths. Acta Physioi Scand 1981;111: Sleep. Vo/. 6. No