The Effect of Triazolam on the Arousal Response to Airway Occlusion during Sleep in Normal Subjects 1 , 2
RICHARD B. BERRY, CRAIG R. MCCASLAND, and RICHARD W. LIGHT
Introduction
T he termination of obstructive apnea in the obstructive sleep apnea syndrome is believed to depend largely on arousal (1-4). Evidence of arousal usually precedes or coincides with the preferential increase in upper airway tone that restores airway patency and terminates the obstructive apnea (2). The precise events leading to arousal and apnea termination are still not well understood. Central nervous system stimulation from a decrease in P02 , an increase in Pe0 2 , and sensations of inspiring against an occluded airway appear to interact to produce arousal and apnea termination (5-10). Recent evidence suggests that the amount of inspiratory effort may be the most important factor producing arousal from sleep (8-10). The amount of inspiratory effort may be gauged by measuring the deflections in esophageal pressure (8) or the airway pressure. In normal subjects, mask occlusion during nonrapid eye movement (NREM) sleep results in a progressive rise in the maximal airway suction pressure generated by each inspiratory effort until arousal occurs (11, 12). If one assumes that arousal will occur once a level of suction pressure is reached (arousal threshold), then the time to arousal also depends on the initial suction pressure and the rate of increase in maximal suction pressure with time. We will refer to these two factors as the respiratory response to airway occlusion. Central nervous system depressants would be expected to impair the arousal response to airway occlusion. In a previous study, we demonstrated that ethanol ingestion prolonged the time to arousal after airway occlusion in NREM sleep in normal subjects (12). The level of suction pressure associated with arousal was increased (increased arousal threshold), and the rate of increase in the maximal suction pressure developed during inspiratory efforts was decreased. Perhaps this is not surprising because ethanol is a central nervous system depressant that 1256
SUMMARY The purpose of this study was to assess the effect of triazolam (0.25 mg) on the arousal response to airway occlusion during non rapid eye movement sleep in normal subjects. Six male subjects (mean age ± SO, 28.1 ± 7.1 yr) had their arousal response tested by occluding a mask covering the nose with the mouth sealed. After an adaptation night, subjects were studied on two consecutive nights. They ingested trlazolam (0.25 mg) or placebo one-half hour before bedtime in a randomized double-blind crossover manner. Mask occlusion was performed 1 to 4 h after triazolam/placebo ingestion while the subjects breathed a mixture of air and oxygen adjusted to produce an arterial oxygen saturation of 98%. The maximal deflections in airway pressure were measured at a supraglottic location during airway occlusion to reflect the degree of Inspiratory effort. The time to arousal (mean ± SEM) was significantly longer on trlazolam nights (32.0 ± 5.2 versus 22.6 ± 3.2 s, p < 0.01).The maximal airway suction pressure preceding arousal was higher on trlazolam nights (26.5 ± 2.0 em H20 versus 20.0 ± 1.2 em H20 , p < 0.02). Conversely, the rate of Increase in inspiratory effort (maximal pressure) during occlusion was not decreased by triazolam. We conclude that trlazolam prolongs the time to arousal following airway occlusion by increasing the arousAM REV RESPIR DIS 1992; 146:1256-1260 al threshold.
also depresses the ventilatory response to hypoxia and hypercapnia (13). The effect of benzodiazepines on the arousal response to airway occlusion has not been studied. The benzodiazepine flurazepam has been demonstrated to increase the arousal threshold to auditory stimuli (14) and to hypercapnia (15) during sleep in normal subjects. We hypothesized that triazolam (a widely used benzodiazepine hypnotic) would increase the time to arousal after airway occlusion by increasing the arousal threshold (level of suction pressure preceding arousal). However, because triazolam has little or no effect on the ventilatory response to CO 2 (16, 17), we predicted that the rate of increase in the maximal airway suction pressure after airway occlusion would not be decreased by this hypnotic. To test these hypotheses, we studied the effect of triazolam on the arousal and respiratory response to mask occlusion in normal subjects during stage 3/4 NREM sleep, using a double-blind randomized crossover protocol. Methods Six normal male subjects werestudied on three consecutive nights. The first night was an adaptation to the sleep laboratory, and no airway occlusions were performed. On the second and third nights, subjects ingested in random order either placebo or 0.25 mg triazo-
lam 30 min before lights out. The project was approved by the Institutional Review Board of our hospital, and all subjects signed an informed consent agreement before participating in the study. The presence and stage of sleep were monitored using two pairs of electroencephalographic (EEG) leads (C4-Al, 02-Al), two pairs of electro-oculographic leads, and chin electromyographic (EMG) leads, using standard methods (18). An. electrocardiographic (ECG) lead was monitored, and arterial oxygen saturation was continuously measured using pulse oximetry (Ohmeda 3700; Ohmeda, Boulder, CO). Subjects wore a modified Adams continuous positive airway pressure (CPAP) circuit (Puritan Bennett, Los Angeles, CA) consisting of a nasal shell and soft pillows that were inserted into the nostrils. The nasal shell was held in place with head straps. The mask circuit was connected to C\ 6-ft length of lowcompliance respiratory tubing, which exited to the monitoring room. The length of tubing was connected to a pneumatic occlusion valve that could prevent inspiration. When (Received in original form February 28, 1992 and in revised form June 18, 1992) 1 From the Long Beach VeteransAdministration Medical Center, University of California, Irvine, California. 2 Correspondence and requests for reprints should be addressed to Richard B. Berry,M.D., Pulmonary Section HIP, Long Beach VAMedical Center, 5901 E. 7th Street, Long Beach, CA 90822.
1257
EFFECT OF TRIAZOLAM ON AROUSAL FROM SLEEP
60
t041 yr(mean ± SD,28.1 ± 7.1 yr). The body weight as a percentage of ideal 50 weight (20) ranged from 820/0 to 126% Fig. 1. The mean time to arousal dur(mean, 103 ± 15.6%). None of the sub< 40 ing mask occlusion for each subject on jects were noted to have apnea or snoring. placebo and triazolam nights. The solid 0" a:-g circles represent the group means. The < g 30 The time to arousal after triazolam in00> time to arousal was significantly longer I-~ gestion (mean ± SEM) was significant20 on the nights the subjects ingested tria:!: ly greater for the group of subjects (32.0 zolam (p < 0.01). i= ± 5.2 versus 22.6 ± 3.2 s, p < 0.01). The 10 means for the individual subjects are 0 - ' - - - - - -__, - - - - - - - - - - , - - - - - - shown in figure 1. The arterial oxygen PLACEBO TRIAZOLAM saturation (Sa02 ) preocclusion was 98 % on both placebo and triazolam nights by study design. The end-tidal Pe02 values the valve was open, a bias flow of room air Grass 78D polygraph (Grass, Quincy, MA) preceding mask occlusion in the triazowas provided by a continuous positive airway using a paper speed of 10 mm/s. A near in- lam and placebo conditions were 45.2 ± pressure device (BIPAP SD; Respironics, frared camera and video monitor system al- 0.40 and 44.8 ± 0.21 mm Hg, respectiveMurrysville, PA). The flow exited the mask lowed continuous visual inspection of the ly (p = NS). Arterial oxygen desaturacircuit at a one-way valve near the mask. A subjects. tion did not occur after any of the four-wayvalveat the outlet of the flow source occlusions. Airway Occlusion Procedure allowed adjustment of the amount of bias A typical polygraph tracing for an ocflow. The flow was adjusted for subject com- The inspiratory airway was closed during ex- clusion trial is shown in figure 2. There fort and to prevent rebreathing of CO2. The piration in stage 3/4 NREM sleep from six was a progressive rise in the maximal same amount of flow wasused on both study to 13 times between 1 and 4 h after placebo supraglottic pressure swings until arousnights. The mask pressureduring tidal breath- or triazolam ingestion. Before all occlusion ing was ± 1 em H 20. A flow of oxygen was trials, subjects were required to have been in al occurred. Arousal was manifested by also added to the circuit and adjusted to ob- stage 3/4 sleep without arousals for at least an increase in the EEG frequency and tain a baseline sleep arterial oxygen satura- 3 min, and the breathing pattern was required simultaneous increase in the EMG amtion of 98070. When the occlusion valve was to be regular and the end-tidal Pe0 2 stable. plitude. At arousal, there was also an activated, the subject could not inspire (low- The occlusion was quickly terminated when abrupt change in the supraglottic prescompliance tubing) but could continue to ex- there wasevidenceof arousal. An arousal was sure tracings. hale through the one-way expiratory valve defined as an abrupt increase in the EEG freThe mean maximal supraglottic presnear the mask. Outlets on and near the nasal quency and a simultaneous increase in the sure deflections on the initial occluded shell allowed measurement of end-tidal CO 2 EMG amplitude (4). breath (PmaxI) did not differ on placebo (capnograph, Model 223; Puritan Bennett) and triazolam nights (table 1). The maxData Analysis and mask pressure, respectively. A pneumotachygraph near the occlusion valve was The time to arousal was measured from the imal supraglottic pressure preceding used to qualitatively monitor inspiratory flow. start of the initial occludedinspiration to EEG arousal (PmaxF) on triazolam nights was This flow signal and the sum of ribcage and and EMG evidence of arousal. The maximal greater for the group of subjects and for abdominal band signals (Vsum)tracing from negativepressuredeflection(supraglotticpres- each individual subject (figure 3). Thus, respiratoryinductance plethysmography(Res- sure) of the initial occluded inspiratory ef- the suction pressure threshold for arouspigraph; NIMS, Miami, FL) wereused to veri- fort (PmaxI) and the final complete effort al was higher on triazolam nights. fy a stable pattern of respiration before preceding arousal (PmaxF) weremeasured. A In contrast, the rate of increase in occlusions. complete effort was one in which arousal did maximal suction pressure with time The airway suction pressure was measured not occur before a definite maximal pressure (~Pmax/T) did not differ between triaat a supraglottic location by a transducer- had been attained. Then a mean slope of max- zolam and placebo nights (table 1). There tipped catheter to reflect the level of inspira- imal pressure deflection (L\Pmax/T) was dewas some individual variability, some tory effort, using the same method as de- termined by dividing L\Pmax = PmaxF scribed in a previous study (12). Following Pmaxl by the elapsed time (T) between subjects having a lower and some a higher instillation of 1 to 2 ml of 2070 lidocaine in PmaxF and PmaxI. The L\Pmax/T was further ~Pmax/T on triazolam nights (figure 4). one nostril and placement of the catheter, the analyzed by factoring this variable into the Analysis of the two factors determining nasal pillows (mask) were inserted into the product of (L\Pmax/No.) and (No.lT), where ~Pmax/T (~Pmax/No. and No.lT) nostrils, and silasticpolymer (Factor II, Lake- No. is the number of occluded inspiratory ef- showed that neither factor differed for side, AZ) was applied until the entire nasal forts after the initial effort before arousal. the group as a whole on placebo and triaarea and mouth weresealed. Beforelightsout, The (L\Pmax/No.) is the mean increase in zolam nights. The No.lTwas essentially the mask circuit was occluded and the seal Pmax per breath and the (No.lT) is the rate the same on both nights. tested repeatedly to verify that no inspirato- of occluded efforts (number per second). The mean values of the time to arousal ry leaks werepresent. This was performed by instructing the subject to inspire forcefully (TTA), preocclusion end-tidal Pe02' PmaxI, Discussion with the inspiratory path occluded and listen- PmaxF, L\Pmax/T, L\Pmax/No., and No.lT for This study shows that triazolam increases ing carefully at the bedside for leaks. Similar each patient were compared between the leak tests werealso performed throughout the placebo and triazolam conditions with the the time to arousal after airway occlunight after every fourth occlusion. Lack of paired t test (19). A P < 0.05 was considered sion during slow-wave (NREM) sleep in normal subjects. This is associated with inspiratory flow during occlusions was also to show statistical significance. an increase in the maximal suction presdocumented by a flat Vsumsignal during airResults sure preceding arousal. If one assumes way occlusion. All variableswererecorded on a 12-channel The age of the subjects ranged from 21 that the amount of inspiratory effort is ...J
(f)
::J~
W
•
•
1258
BERRY, MCCASLAND, AND LIGHT
Fig. 2. Apolygraph tracing showing an occlusion trial. The maximal supraglottic pressure increased progressively with each inspiratory effort until arousal occurred. Arterial oxygen desaturation did not occur.
.-
>--------
TABLE 1 MAXIMUM PRESSURES AND RATES OF CHANGE DURING OCCLUSION* Placebo Pmaxl, cm H20 PmaxF, cm H20 L\PmaxlT, cm H20/s l\Pmax/No., cm H20/No. No.lT, No.ls
11.2 20.0 0.50 1.80 0.30
± 1.0 ± 1.2
± 0.07 ± 0.35 ± 0.02
Triazolam 11.4 26.5 0.60 2.2 0.29
± ± ± ± ±
1.0 2.0t 0.12 0.54 0.02
* Values are mean ± SEM. Pmaxl, PmaxF = maximum airway suction pressure on the initial and final occluded breaths. dPmaxlT = the rate of increase in Pmax with time. dPmax/No. = average increase in Pmax per occluded breath. No.1T = rate of occluded breaths. t Triazolam versus placebo. p < 0.02.
40 W
a:~
:::>0
~:f' W E
30
a: u
0.-
z..J O~
~5
20
:::>a: CIl
a: ::::;;0 _LL XW
10
•
~
•
Fig. 3. The mean maximal suction pressure before arousal isshown for each subject on triazolam and placebo nights. The solid circles represent the mean values for the group. The mean PmaxF on triazolam nights was higher (p < 0.02).
:
.. ~ .. 0-
CIl
-0
0.2
0.0 PLACEBO
TRIAZOLAM
Fig. 4. The mean rate of increase of the maximal supraglottic pressure (slope of Pmax) during occluded inspiratory efforts (L\PmaxlT) is shown for each individual subject on placebo and trlazolam nights. The solid circles represent the mean values for the group. The group mean did not differ between placebo and triazolam nights. Some subjects had a higher L\PmaxlT and some alower value on triazolam nights.
the main stimulus for arousal (8), this suggests that triazolam increases the arousal threshold. However, even if the arousal threshold also depends on the arterial Pco, and Po 2, this conclusion can still be reached. First, arterial oxygen desaturation did not occur due to the use of supplemental oxygen. Second, the Pe02 at arousal on triazolam nights was probably slightly higher. This is likely, because the end-tidal Pco, preocclusion on placebo and triazolam nights was almost identical but the time to arousal was longer in the latter condition. If both the higher Pco, and the amount of inspiratory effort directly stimulated arousal, one would expect a lower amount of inspiratory effort (pressure) would be needed to induce arousal, assuming the arousal threshold was unchanged by triazolam. The fact that PmaxF was higher on triazolam nights shows that the arousal threshold was definitely increased by this drug. In any case, a difference of 10 s in occlusion times would be expected to produce, at most, a very small increase in the arterial Pco, at arousal on triazolam compared with placebo nights. Flurazepam, a long-acting benzodiazepine, has previously been shown to increase the arousal threshold to auditory stimuli (14) and to inhaled CO 2 during sleep (15). In a previous study, wedemonstrated that ethanol ingestion also increases the maximum suction pressure preceding arousal from sleep during mask occlusion in normal subjects (12). In both this study and the current study the subjects were breathing supplemental oxygen. In contrast to the effects of ethanol, we found that triazolam did not decrease the rate of increase in maximal suction pressure (~Pmax/T). Therefore, although the arousal response to airway occlusion was impaired by triazolam, the respiratory response to airway occlusion (PI, ~Pmax/T) was not changed. Triazolam, in the dose used in this study, has minimal if any depressive effects on the ventilatory response to CO 2 (16, 17) during wakefulness. The effect of triazolam on the ventilatory response to CO 2 during sleep has not been studied. Flurazepam did not decrease the response to hypoxia and hypercarbia during sleep in a study of normal subjects (15). Thus, it is not surprising that triazolam did not alter ~Pmax/T. In a previous study (12), ethanol almost doubled the time to arousal after airway occlusion in normal subjects. In
1259
EFFECT OF TRIAZOLAM ON AROUSAL FROM SLEEP
the present study, triazolam ingestion caused a more modest increase in the time to arousal, though it was almost 400/0 of the placebo value. The smaller increase with triazolam compared with ethanol may be due in part to the lack of decrease in ~Pmax/T with the former drug. Alternatively, this could simply be a doserelated phenomenon. However, triazolam at doses of 1.0 and 1.5 mg did not impair ventilation or the ventilatory response to hypercapnia in awake normal subjects (17). After oral ingestion of triazolam, peak levels are obtained in 1 to 2 h, and the half-life ranges from 2 to 6 h (21). This is the reason we chose to study occlusion 1 to 4 h after placebo/triazolam ingestion. The serum blood concentration of triazolam undoubtedly varied during the night. The effects demonstrated by this study, then, represent the average effect of a range of blood concentrations and could underestimate the effects of peak concentrations of the drug. The random order of placebo and drug nights minimized the chance that carryover effects of triazolam on a following placebo night would obscure differences between the drug and placebo conditions. Triazolam has no active metabolites, and its short half-life relative to the 24-h washout period should have made such effects unlikely. In any study of a small number of subjects, one must consider the possibility of type II error in evaluating nonsignificant comparisons. That is, perhaps our failure to find a significantly lower ~P max/T on triazolam nights was due to the small number of subjects used. We feel this is not likely inasmuch as the mean L\Pmax/T was actually slightly higher on the triazolam nights. This may have been due to a slightly higher Pco, in the latter part of the triazolam occlusions, which may have augmented the increase in ventilatory drive with time. Several studies have evaluated the effects of benzodiazepines on sleep-disordered breathing. Dolly and Block (22) found an increase in the number of apneas after normal subjects ingested 30 mg of flurazepam. Guilleminault and coworkers (23) found that 30mg of flurazepam increased the mean apnea duration in NREM sleep from 21.7 to 27.6 s in a population of elderly adults with minimal apnea on placebo nights. Another study found that triazolam had minimal adverse effects on respiration in a group of patients with chronic obstructive pul-
monary disease (24). To our knowledge, there have been no studies of triazolam in patients with significant obstructive sleep apnea. Bonnet and colleagues (25) did study the effect of triazolam on patients with predominant central sleep apnea. The central sleep apnea index (apneas/hour of sleep)decreased slightly but significantly. The number of obstructive apneas did not increase in this population. The investigators hypothesized that triazolam may have blunted the magnitude of postapneic ventilation associated with arousal. This may have prevented the Pco, from falling below the sleeping apneic threshold and subsequent central apnea when the patients returned to sleep after arousal. In some patients with obstructive sleep apnea, triazolam could conceivably increase the preapnea Pco, by decreasing the duration or magnitude of arousal postapnea and the associated increases in ventilation. This would tend to reduce apnea length. In the present study of normal subjects, the preocclusion Pco, was not different on placebo and triazolam nights. However, the ultimate effect of triazolam on apnea length in patients with obstructive sleep apnea may depend on the effect of this drug on the preapnea Pco, as well as changes in the arousal threshold. In this study, the end-tidal Pco, was measured at the nasal shell. The bias flow exited the circuit via a one-way expiratory valve several inches above the nasal shell. It is possible that this may have diluted the gas sampled for the CO 2 measurements; however, the same amount of bias flow was used on both study nights, and, therefore, this should not have caused a problem in comparing the preocclusion end-tidal Pco, values on triazolam and placebo nights. Upper airway sensation is important for maintaining airway patency and detecting airway occlusion (26, 27). However, we do not believe the use of lidocaine affected our conclusions regarding the effect of triazolam on arousal. We used a small amount of lidocaine, began the occlusion trials more than 1 h after lidocaine instillation (duration of action 1 to 2 h), and employed the same method of anesthesia on both placebo and triazolam nights. Thus, even if lidocaine increased the time to arousal in our study, there is no reason to believe it differentially affected the placebo and triazolam occlusions. We chose to perform occlusions in stage 3/4 rather than stage 2 sleep. It was
thought that awakening (one or more epochs of stage wake) would be less likely to occur from the deeper sleep stages. We found that if airway occlusion was quickly released at the first EEG and EMG sign of arousal, the subjects usually returned quickly to deep sleep. This allowed a large number of occlusion trials in a short period of time. In summary, our findings document that triazolam ingestion impairs the arousal response to airway occlusion during slow-wavesleep such that the time to arousal after airway occlusion is prolonged and the maximal suction pressure before arousal is increased. However, the respiratory response to airway occlusion was not impaired and thus the rate of increase in suction pressure with time was not decreased for the group of subjects studied. References 1. Guilleminault C, Rosekind M. The arousal threshold: sleep deprivation, sleep fragmentation, and obstructive sleep apnea syndrome. Bull Eur Physiopathol Respir 1981; 17:341-9. 2. Remmers JE, Degroot WJ, Sauerland EK, Anch AM. Pathogenesis of upper airway occlusion during sleep. J Appl Physiol 1978; 44:931-8. 3. Phillipson EA, Sullivan CEo Arousal: the forgotten response to respiratory stimuli. Am Rev Resp Dis 1978; 118:807-9. 4. West P, Kryger MH. Sleep and respiration: terminology and methodology. In: Kryger MH, ed. Clinics in chest medicine: sleep disorders. Philadelphia: W. B. Saunders, 1985; 691-2. 5. Phillipson EA, Sullivan CEo Ventilatory and arousal responses to hypoxia in sleeping humans. Am Rev Respir Dis 1982; 125:632-9. 6. Berthon-Jones M, Sullivan CEo Ventilation and arousal responses to hypercapnia in normal sleeping humans. J Appl Physiol 1984; 57:59-67. 7. Douglas NJ, White D, Weil JV, Pickett CK, Zwillich CWoHypercapnic ventilatory response in sleeping adults. Am Rev Respir Dis 1982; 126:758-62. 8. Gleeson K, Zwillich CW, White DP. Arousal from sleep in response to ventilatory stimuli occurs at a similar degree of ventilatory effort irrespective of the stimulus. Am Rev Respir Dis 1989; 142:295-300. 9. VinckenW, Guilleminault C, Silvestri L, Cosio M, Grassino A. Inspiratory muscle activity as a trigger causing the airways to open in obstructive sleep apnea. Am Rev Respir Dis 1987; 135:372-7. 10. Wilcox PG, Pare PD, Road JD, Fleetham JA. Respiratory muscle function during obstructive sleep apnea. Am Rev Respir Dis 1990; 142:533-9. 11. Issa FG, Sullivan CEo Arousal and breathing responses to airway occlusion in healthy sleeping adults. J Appl Physiol 1983; 55:1113-9. 12. Berry RB, Bonnet MH, Light RW. Effect of ethanol on the arousal response to airway occlusion in normal subjects. Am Rev Respir Dis 1992; 145:445-52. 13. Michiels TM, Light RW, Mahutte CK. Effect of ethanol and naloxone on control of ventilation and load perception. J Appl Physiol 1983; 55:929-34.
1260
14. Bonnet MH, Webb WB, Barnard G. Effect of flurazepam, pentobarbital, and caffeine on arousal threshold. Sleep 1979; 1:271-9. 15. Hedemark LL, Kronenberg RS. Flurazepam attenuates the arousal responsesto CO 2 during sleep in normal subjects. Am Rev Respir Dis 1983; 128:980-3. 16. Longbottom RT, Pleuvry BJ. Respiratory and sedative effects of triazolam in volunteers. Br J Anaesthesiol 1984; 56:179-85. 17. Skatrud JB, Begle RL, Busch MA. Ventilatory effects of single, high-dose triazolam in awake human subjects. Clin Pharmacol Ther 1988; 44:684-9. 18. Rechteschaffen A, Kales A. A manual of standardized terminology, techniques, and scoring system for sleep stages in human adults. Los Angeles: Brain Information Service/Brain Research Insti-
BERRY, MCCASLAND, AND LIGHT
tute, University of California, Los Angeles, 1968. 19. Ott L. An introduction to statistical methods and data analysis. Boston: Duxbury, 1984; 166-7. 20. Russell RM, McGandy RB, Jelliffe D. Reference weights. Am J Med 1984; 76:767-9. 21. McEvoy GK, ed. AHFS drug information. Bethesda, MD: American Societyof Hospital Pharmacists, 1992; 1330-1. 22. Dolly FR, Block AJ. Effect of flurazepam on sleep-disordered breathing and nocturnal oxygen desaturation in asymptomatic subjects. Am J Med 1982; 73:239-43. 23. Guilleminault C, Silvestri R, Mondidin S, Coburn S. Aging and sleep apnea: action of benzodiazepine, acetazolamide, alcohol, and sleepderivation in a healthy elderly group. J Gerontoll984; 39:655-61. 24. Timms RM, Dawson A, Hajukovic RM, Mi-
tler MM. Effect of triazolam on sleep and arterial oxygensaturation in patients with chronic obstructive pulmonary disease. Arch Intern Med 1988; 148:2159-63. 25. Bonnet MH, Dexter JR, Arand DL. The effect of triazolam on arousal and respiration in central sleep apnea patients. Sleep 1990; 13:31-41. 26. McNicholas WT, Coffey M, McDonnell T, O'Reagan R, Fitzgerald MX. Upper airwayobstruction during sleep in normal subjects after selective topical oropharyngeal anesthesia. Am Rev Respir Dis 1987; 135:1316-9. 27. Basner R, Ringler J, Garpestad E, Schwartzstein R, Weinberger S, Weiss JW. Upper airway anesthesia prolongs the time to arousal from airway occlusion induced during NREM sleep (abstract). Am Rev Respir Dis 1990; 141:AI27.
RICHARD B. BERRY, CRAIG R. MCCASLAND, and RICHARD W. LIGHT
Introduction
T he termination of obstructive apnea in the obstructive sleep apnea syndrome is believed to depend largely on arousal (1-4). Evidence of arousal usually precedes or coincides with the preferential increase in upper airway tone that restores airway patency and terminates the obstructive apnea (2). The precise events leading to arousal and apnea termination are still not well understood. Central nervous system stimulation from a decrease in P02 , an increase in Pe0 2 , and sensations of inspiring against an occluded airway appear to interact to produce arousal and apnea termination (5-10). Recent evidence suggests that the amount of inspiratory effort may be the most important factor producing arousal from sleep (8-10). The amount of inspiratory effort may be gauged by measuring the deflections in esophageal pressure (8) or the airway pressure. In normal subjects, mask occlusion during nonrapid eye movement (NREM) sleep results in a progressive rise in the maximal airway suction pressure generated by each inspiratory effort until arousal occurs (11, 12). If one assumes that arousal will occur once a level of suction pressure is reached (arousal threshold), then the time to arousal also depends on the initial suction pressure and the rate of increase in maximal suction pressure with time. We will refer to these two factors as the respiratory response to airway occlusion. Central nervous system depressants would be expected to impair the arousal response to airway occlusion. In a previous study, we demonstrated that ethanol ingestion prolonged the time to arousal after airway occlusion in NREM sleep in normal subjects (12). The level of suction pressure associated with arousal was increased (increased arousal threshold), and the rate of increase in the maximal suction pressure developed during inspiratory efforts was decreased. Perhaps this is not surprising because ethanol is a central nervous system depressant that 1256
SUMMARY The purpose of this study was to assess the effect of triazolam (0.25 mg) on the arousal response to airway occlusion during non rapid eye movement sleep in normal subjects. Six male subjects (mean age ± SO, 28.1 ± 7.1 yr) had their arousal response tested by occluding a mask covering the nose with the mouth sealed. After an adaptation night, subjects were studied on two consecutive nights. They ingested trlazolam (0.25 mg) or placebo one-half hour before bedtime in a randomized double-blind crossover manner. Mask occlusion was performed 1 to 4 h after triazolam/placebo ingestion while the subjects breathed a mixture of air and oxygen adjusted to produce an arterial oxygen saturation of 98%. The maximal deflections in airway pressure were measured at a supraglottic location during airway occlusion to reflect the degree of Inspiratory effort. The time to arousal (mean ± SEM) was significantly longer on trlazolam nights (32.0 ± 5.2 versus 22.6 ± 3.2 s, p < 0.01).The maximal airway suction pressure preceding arousal was higher on trlazolam nights (26.5 ± 2.0 em H20 versus 20.0 ± 1.2 em H20 , p < 0.02). Conversely, the rate of Increase in inspiratory effort (maximal pressure) during occlusion was not decreased by triazolam. We conclude that trlazolam prolongs the time to arousal following airway occlusion by increasing the arousAM REV RESPIR DIS 1992; 146:1256-1260 al threshold.
also depresses the ventilatory response to hypoxia and hypercapnia (13). The effect of benzodiazepines on the arousal response to airway occlusion has not been studied. The benzodiazepine flurazepam has been demonstrated to increase the arousal threshold to auditory stimuli (14) and to hypercapnia (15) during sleep in normal subjects. We hypothesized that triazolam (a widely used benzodiazepine hypnotic) would increase the time to arousal after airway occlusion by increasing the arousal threshold (level of suction pressure preceding arousal). However, because triazolam has little or no effect on the ventilatory response to CO 2 (16, 17), we predicted that the rate of increase in the maximal airway suction pressure after airway occlusion would not be decreased by this hypnotic. To test these hypotheses, we studied the effect of triazolam on the arousal and respiratory response to mask occlusion in normal subjects during stage 3/4 NREM sleep, using a double-blind randomized crossover protocol. Methods Six normal male subjects werestudied on three consecutive nights. The first night was an adaptation to the sleep laboratory, and no airway occlusions were performed. On the second and third nights, subjects ingested in random order either placebo or 0.25 mg triazo-
lam 30 min before lights out. The project was approved by the Institutional Review Board of our hospital, and all subjects signed an informed consent agreement before participating in the study. The presence and stage of sleep were monitored using two pairs of electroencephalographic (EEG) leads (C4-Al, 02-Al), two pairs of electro-oculographic leads, and chin electromyographic (EMG) leads, using standard methods (18). An. electrocardiographic (ECG) lead was monitored, and arterial oxygen saturation was continuously measured using pulse oximetry (Ohmeda 3700; Ohmeda, Boulder, CO). Subjects wore a modified Adams continuous positive airway pressure (CPAP) circuit (Puritan Bennett, Los Angeles, CA) consisting of a nasal shell and soft pillows that were inserted into the nostrils. The nasal shell was held in place with head straps. The mask circuit was connected to C\ 6-ft length of lowcompliance respiratory tubing, which exited to the monitoring room. The length of tubing was connected to a pneumatic occlusion valve that could prevent inspiration. When (Received in original form February 28, 1992 and in revised form June 18, 1992) 1 From the Long Beach VeteransAdministration Medical Center, University of California, Irvine, California. 2 Correspondence and requests for reprints should be addressed to Richard B. Berry,M.D., Pulmonary Section HIP, Long Beach VAMedical Center, 5901 E. 7th Street, Long Beach, CA 90822.
1257
EFFECT OF TRIAZOLAM ON AROUSAL FROM SLEEP
60
t041 yr(mean ± SD,28.1 ± 7.1 yr). The body weight as a percentage of ideal 50 weight (20) ranged from 820/0 to 126% Fig. 1. The mean time to arousal dur(mean, 103 ± 15.6%). None of the sub< 40 ing mask occlusion for each subject on jects were noted to have apnea or snoring. placebo and triazolam nights. The solid 0" a:-g circles represent the group means. The < g 30 The time to arousal after triazolam in00> time to arousal was significantly longer I-~ gestion (mean ± SEM) was significant20 on the nights the subjects ingested tria:!: ly greater for the group of subjects (32.0 zolam (p < 0.01). i= ± 5.2 versus 22.6 ± 3.2 s, p < 0.01). The 10 means for the individual subjects are 0 - ' - - - - - -__, - - - - - - - - - - , - - - - - - shown in figure 1. The arterial oxygen PLACEBO TRIAZOLAM saturation (Sa02 ) preocclusion was 98 % on both placebo and triazolam nights by study design. The end-tidal Pe02 values the valve was open, a bias flow of room air Grass 78D polygraph (Grass, Quincy, MA) preceding mask occlusion in the triazowas provided by a continuous positive airway using a paper speed of 10 mm/s. A near in- lam and placebo conditions were 45.2 ± pressure device (BIPAP SD; Respironics, frared camera and video monitor system al- 0.40 and 44.8 ± 0.21 mm Hg, respectiveMurrysville, PA). The flow exited the mask lowed continuous visual inspection of the ly (p = NS). Arterial oxygen desaturacircuit at a one-way valve near the mask. A subjects. tion did not occur after any of the four-wayvalveat the outlet of the flow source occlusions. Airway Occlusion Procedure allowed adjustment of the amount of bias A typical polygraph tracing for an ocflow. The flow was adjusted for subject com- The inspiratory airway was closed during ex- clusion trial is shown in figure 2. There fort and to prevent rebreathing of CO2. The piration in stage 3/4 NREM sleep from six was a progressive rise in the maximal same amount of flow wasused on both study to 13 times between 1 and 4 h after placebo supraglottic pressure swings until arousnights. The mask pressureduring tidal breath- or triazolam ingestion. Before all occlusion ing was ± 1 em H 20. A flow of oxygen was trials, subjects were required to have been in al occurred. Arousal was manifested by also added to the circuit and adjusted to ob- stage 3/4 sleep without arousals for at least an increase in the EEG frequency and tain a baseline sleep arterial oxygen satura- 3 min, and the breathing pattern was required simultaneous increase in the EMG amtion of 98070. When the occlusion valve was to be regular and the end-tidal Pe0 2 stable. plitude. At arousal, there was also an activated, the subject could not inspire (low- The occlusion was quickly terminated when abrupt change in the supraglottic prescompliance tubing) but could continue to ex- there wasevidenceof arousal. An arousal was sure tracings. hale through the one-way expiratory valve defined as an abrupt increase in the EEG freThe mean maximal supraglottic presnear the mask. Outlets on and near the nasal quency and a simultaneous increase in the sure deflections on the initial occluded shell allowed measurement of end-tidal CO 2 EMG amplitude (4). breath (PmaxI) did not differ on placebo (capnograph, Model 223; Puritan Bennett) and triazolam nights (table 1). The maxData Analysis and mask pressure, respectively. A pneumotachygraph near the occlusion valve was The time to arousal was measured from the imal supraglottic pressure preceding used to qualitatively monitor inspiratory flow. start of the initial occludedinspiration to EEG arousal (PmaxF) on triazolam nights was This flow signal and the sum of ribcage and and EMG evidence of arousal. The maximal greater for the group of subjects and for abdominal band signals (Vsum)tracing from negativepressuredeflection(supraglotticpres- each individual subject (figure 3). Thus, respiratoryinductance plethysmography(Res- sure) of the initial occluded inspiratory ef- the suction pressure threshold for arouspigraph; NIMS, Miami, FL) wereused to veri- fort (PmaxI) and the final complete effort al was higher on triazolam nights. fy a stable pattern of respiration before preceding arousal (PmaxF) weremeasured. A In contrast, the rate of increase in occlusions. complete effort was one in which arousal did maximal suction pressure with time The airway suction pressure was measured not occur before a definite maximal pressure (~Pmax/T) did not differ between triaat a supraglottic location by a transducer- had been attained. Then a mean slope of max- zolam and placebo nights (table 1). There tipped catheter to reflect the level of inspira- imal pressure deflection (L\Pmax/T) was dewas some individual variability, some tory effort, using the same method as de- termined by dividing L\Pmax = PmaxF scribed in a previous study (12). Following Pmaxl by the elapsed time (T) between subjects having a lower and some a higher instillation of 1 to 2 ml of 2070 lidocaine in PmaxF and PmaxI. The L\Pmax/T was further ~Pmax/T on triazolam nights (figure 4). one nostril and placement of the catheter, the analyzed by factoring this variable into the Analysis of the two factors determining nasal pillows (mask) were inserted into the product of (L\Pmax/No.) and (No.lT), where ~Pmax/T (~Pmax/No. and No.lT) nostrils, and silasticpolymer (Factor II, Lake- No. is the number of occluded inspiratory ef- showed that neither factor differed for side, AZ) was applied until the entire nasal forts after the initial effort before arousal. the group as a whole on placebo and triaarea and mouth weresealed. Beforelightsout, The (L\Pmax/No.) is the mean increase in zolam nights. The No.lTwas essentially the mask circuit was occluded and the seal Pmax per breath and the (No.lT) is the rate the same on both nights. tested repeatedly to verify that no inspirato- of occluded efforts (number per second). The mean values of the time to arousal ry leaks werepresent. This was performed by instructing the subject to inspire forcefully (TTA), preocclusion end-tidal Pe02' PmaxI, Discussion with the inspiratory path occluded and listen- PmaxF, L\Pmax/T, L\Pmax/No., and No.lT for This study shows that triazolam increases ing carefully at the bedside for leaks. Similar each patient were compared between the leak tests werealso performed throughout the placebo and triazolam conditions with the the time to arousal after airway occlunight after every fourth occlusion. Lack of paired t test (19). A P < 0.05 was considered sion during slow-wave (NREM) sleep in normal subjects. This is associated with inspiratory flow during occlusions was also to show statistical significance. an increase in the maximal suction presdocumented by a flat Vsumsignal during airResults sure preceding arousal. If one assumes way occlusion. All variableswererecorded on a 12-channel The age of the subjects ranged from 21 that the amount of inspiratory effort is ...J
(f)
::J~
W
•
•
1258
BERRY, MCCASLAND, AND LIGHT
Fig. 2. Apolygraph tracing showing an occlusion trial. The maximal supraglottic pressure increased progressively with each inspiratory effort until arousal occurred. Arterial oxygen desaturation did not occur.
.-
>--------
TABLE 1 MAXIMUM PRESSURES AND RATES OF CHANGE DURING OCCLUSION* Placebo Pmaxl, cm H20 PmaxF, cm H20 L\PmaxlT, cm H20/s l\Pmax/No., cm H20/No. No.lT, No.ls
11.2 20.0 0.50 1.80 0.30
± 1.0 ± 1.2
± 0.07 ± 0.35 ± 0.02
Triazolam 11.4 26.5 0.60 2.2 0.29
± ± ± ± ±
1.0 2.0t 0.12 0.54 0.02
* Values are mean ± SEM. Pmaxl, PmaxF = maximum airway suction pressure on the initial and final occluded breaths. dPmaxlT = the rate of increase in Pmax with time. dPmax/No. = average increase in Pmax per occluded breath. No.1T = rate of occluded breaths. t Triazolam versus placebo. p < 0.02.
40 W
a:~
:::>0
~:f' W E
30
a: u
0.-
z..J O~
~5
20
:::>a: CIl
a: ::::;;0 _LL XW
10
•
~
•
Fig. 3. The mean maximal suction pressure before arousal isshown for each subject on triazolam and placebo nights. The solid circles represent the mean values for the group. The mean PmaxF on triazolam nights was higher (p < 0.02).
:
.. ~ .. 0-
CIl
-0
0.2
0.0 PLACEBO
TRIAZOLAM
Fig. 4. The mean rate of increase of the maximal supraglottic pressure (slope of Pmax) during occluded inspiratory efforts (L\PmaxlT) is shown for each individual subject on placebo and trlazolam nights. The solid circles represent the mean values for the group. The group mean did not differ between placebo and triazolam nights. Some subjects had a higher L\PmaxlT and some alower value on triazolam nights.
the main stimulus for arousal (8), this suggests that triazolam increases the arousal threshold. However, even if the arousal threshold also depends on the arterial Pco, and Po 2, this conclusion can still be reached. First, arterial oxygen desaturation did not occur due to the use of supplemental oxygen. Second, the Pe02 at arousal on triazolam nights was probably slightly higher. This is likely, because the end-tidal Pco, preocclusion on placebo and triazolam nights was almost identical but the time to arousal was longer in the latter condition. If both the higher Pco, and the amount of inspiratory effort directly stimulated arousal, one would expect a lower amount of inspiratory effort (pressure) would be needed to induce arousal, assuming the arousal threshold was unchanged by triazolam. The fact that PmaxF was higher on triazolam nights shows that the arousal threshold was definitely increased by this drug. In any case, a difference of 10 s in occlusion times would be expected to produce, at most, a very small increase in the arterial Pco, at arousal on triazolam compared with placebo nights. Flurazepam, a long-acting benzodiazepine, has previously been shown to increase the arousal threshold to auditory stimuli (14) and to inhaled CO 2 during sleep (15). In a previous study, wedemonstrated that ethanol ingestion also increases the maximum suction pressure preceding arousal from sleep during mask occlusion in normal subjects (12). In both this study and the current study the subjects were breathing supplemental oxygen. In contrast to the effects of ethanol, we found that triazolam did not decrease the rate of increase in maximal suction pressure (~Pmax/T). Therefore, although the arousal response to airway occlusion was impaired by triazolam, the respiratory response to airway occlusion (PI, ~Pmax/T) was not changed. Triazolam, in the dose used in this study, has minimal if any depressive effects on the ventilatory response to CO 2 (16, 17) during wakefulness. The effect of triazolam on the ventilatory response to CO 2 during sleep has not been studied. Flurazepam did not decrease the response to hypoxia and hypercarbia during sleep in a study of normal subjects (15). Thus, it is not surprising that triazolam did not alter ~Pmax/T. In a previous study (12), ethanol almost doubled the time to arousal after airway occlusion in normal subjects. In
1259
EFFECT OF TRIAZOLAM ON AROUSAL FROM SLEEP
the present study, triazolam ingestion caused a more modest increase in the time to arousal, though it was almost 400/0 of the placebo value. The smaller increase with triazolam compared with ethanol may be due in part to the lack of decrease in ~Pmax/T with the former drug. Alternatively, this could simply be a doserelated phenomenon. However, triazolam at doses of 1.0 and 1.5 mg did not impair ventilation or the ventilatory response to hypercapnia in awake normal subjects (17). After oral ingestion of triazolam, peak levels are obtained in 1 to 2 h, and the half-life ranges from 2 to 6 h (21). This is the reason we chose to study occlusion 1 to 4 h after placebo/triazolam ingestion. The serum blood concentration of triazolam undoubtedly varied during the night. The effects demonstrated by this study, then, represent the average effect of a range of blood concentrations and could underestimate the effects of peak concentrations of the drug. The random order of placebo and drug nights minimized the chance that carryover effects of triazolam on a following placebo night would obscure differences between the drug and placebo conditions. Triazolam has no active metabolites, and its short half-life relative to the 24-h washout period should have made such effects unlikely. In any study of a small number of subjects, one must consider the possibility of type II error in evaluating nonsignificant comparisons. That is, perhaps our failure to find a significantly lower ~P max/T on triazolam nights was due to the small number of subjects used. We feel this is not likely inasmuch as the mean L\Pmax/T was actually slightly higher on the triazolam nights. This may have been due to a slightly higher Pco, in the latter part of the triazolam occlusions, which may have augmented the increase in ventilatory drive with time. Several studies have evaluated the effects of benzodiazepines on sleep-disordered breathing. Dolly and Block (22) found an increase in the number of apneas after normal subjects ingested 30 mg of flurazepam. Guilleminault and coworkers (23) found that 30mg of flurazepam increased the mean apnea duration in NREM sleep from 21.7 to 27.6 s in a population of elderly adults with minimal apnea on placebo nights. Another study found that triazolam had minimal adverse effects on respiration in a group of patients with chronic obstructive pul-
monary disease (24). To our knowledge, there have been no studies of triazolam in patients with significant obstructive sleep apnea. Bonnet and colleagues (25) did study the effect of triazolam on patients with predominant central sleep apnea. The central sleep apnea index (apneas/hour of sleep)decreased slightly but significantly. The number of obstructive apneas did not increase in this population. The investigators hypothesized that triazolam may have blunted the magnitude of postapneic ventilation associated with arousal. This may have prevented the Pco, from falling below the sleeping apneic threshold and subsequent central apnea when the patients returned to sleep after arousal. In some patients with obstructive sleep apnea, triazolam could conceivably increase the preapnea Pco, by decreasing the duration or magnitude of arousal postapnea and the associated increases in ventilation. This would tend to reduce apnea length. In the present study of normal subjects, the preocclusion Pco, was not different on placebo and triazolam nights. However, the ultimate effect of triazolam on apnea length in patients with obstructive sleep apnea may depend on the effect of this drug on the preapnea Pco, as well as changes in the arousal threshold. In this study, the end-tidal Pco, was measured at the nasal shell. The bias flow exited the circuit via a one-way expiratory valve several inches above the nasal shell. It is possible that this may have diluted the gas sampled for the CO 2 measurements; however, the same amount of bias flow was used on both study nights, and, therefore, this should not have caused a problem in comparing the preocclusion end-tidal Pco, values on triazolam and placebo nights. Upper airway sensation is important for maintaining airway patency and detecting airway occlusion (26, 27). However, we do not believe the use of lidocaine affected our conclusions regarding the effect of triazolam on arousal. We used a small amount of lidocaine, began the occlusion trials more than 1 h after lidocaine instillation (duration of action 1 to 2 h), and employed the same method of anesthesia on both placebo and triazolam nights. Thus, even if lidocaine increased the time to arousal in our study, there is no reason to believe it differentially affected the placebo and triazolam occlusions. We chose to perform occlusions in stage 3/4 rather than stage 2 sleep. It was
thought that awakening (one or more epochs of stage wake) would be less likely to occur from the deeper sleep stages. We found that if airway occlusion was quickly released at the first EEG and EMG sign of arousal, the subjects usually returned quickly to deep sleep. This allowed a large number of occlusion trials in a short period of time. In summary, our findings document that triazolam ingestion impairs the arousal response to airway occlusion during slow-wavesleep such that the time to arousal after airway occlusion is prolonged and the maximal suction pressure before arousal is increased. However, the respiratory response to airway occlusion was not impaired and thus the rate of increase in suction pressure with time was not decreased for the group of subjects studied. References 1. Guilleminault C, Rosekind M. The arousal threshold: sleep deprivation, sleep fragmentation, and obstructive sleep apnea syndrome. Bull Eur Physiopathol Respir 1981; 17:341-9. 2. Remmers JE, Degroot WJ, Sauerland EK, Anch AM. Pathogenesis of upper airway occlusion during sleep. J Appl Physiol 1978; 44:931-8. 3. Phillipson EA, Sullivan CEo Arousal: the forgotten response to respiratory stimuli. Am Rev Resp Dis 1978; 118:807-9. 4. West P, Kryger MH. Sleep and respiration: terminology and methodology. In: Kryger MH, ed. Clinics in chest medicine: sleep disorders. Philadelphia: W. B. Saunders, 1985; 691-2. 5. Phillipson EA, Sullivan CEo Ventilatory and arousal responses to hypoxia in sleeping humans. Am Rev Respir Dis 1982; 125:632-9. 6. Berthon-Jones M, Sullivan CEo Ventilation and arousal responses to hypercapnia in normal sleeping humans. J Appl Physiol 1984; 57:59-67. 7. Douglas NJ, White D, Weil JV, Pickett CK, Zwillich CWoHypercapnic ventilatory response in sleeping adults. Am Rev Respir Dis 1982; 126:758-62. 8. Gleeson K, Zwillich CW, White DP. Arousal from sleep in response to ventilatory stimuli occurs at a similar degree of ventilatory effort irrespective of the stimulus. Am Rev Respir Dis 1989; 142:295-300. 9. VinckenW, Guilleminault C, Silvestri L, Cosio M, Grassino A. Inspiratory muscle activity as a trigger causing the airways to open in obstructive sleep apnea. Am Rev Respir Dis 1987; 135:372-7. 10. Wilcox PG, Pare PD, Road JD, Fleetham JA. Respiratory muscle function during obstructive sleep apnea. Am Rev Respir Dis 1990; 142:533-9. 11. Issa FG, Sullivan CEo Arousal and breathing responses to airway occlusion in healthy sleeping adults. J Appl Physiol 1983; 55:1113-9. 12. Berry RB, Bonnet MH, Light RW. Effect of ethanol on the arousal response to airway occlusion in normal subjects. Am Rev Respir Dis 1992; 145:445-52. 13. Michiels TM, Light RW, Mahutte CK. Effect of ethanol and naloxone on control of ventilation and load perception. J Appl Physiol 1983; 55:929-34.
1260
14. Bonnet MH, Webb WB, Barnard G. Effect of flurazepam, pentobarbital, and caffeine on arousal threshold. Sleep 1979; 1:271-9. 15. Hedemark LL, Kronenberg RS. Flurazepam attenuates the arousal responsesto CO 2 during sleep in normal subjects. Am Rev Respir Dis 1983; 128:980-3. 16. Longbottom RT, Pleuvry BJ. Respiratory and sedative effects of triazolam in volunteers. Br J Anaesthesiol 1984; 56:179-85. 17. Skatrud JB, Begle RL, Busch MA. Ventilatory effects of single, high-dose triazolam in awake human subjects. Clin Pharmacol Ther 1988; 44:684-9. 18. Rechteschaffen A, Kales A. A manual of standardized terminology, techniques, and scoring system for sleep stages in human adults. Los Angeles: Brain Information Service/Brain Research Insti-
BERRY, MCCASLAND, AND LIGHT
tute, University of California, Los Angeles, 1968. 19. Ott L. An introduction to statistical methods and data analysis. Boston: Duxbury, 1984; 166-7. 20. Russell RM, McGandy RB, Jelliffe D. Reference weights. Am J Med 1984; 76:767-9. 21. McEvoy GK, ed. AHFS drug information. Bethesda, MD: American Societyof Hospital Pharmacists, 1992; 1330-1. 22. Dolly FR, Block AJ. Effect of flurazepam on sleep-disordered breathing and nocturnal oxygen desaturation in asymptomatic subjects. Am J Med 1982; 73:239-43. 23. Guilleminault C, Silvestri R, Mondidin S, Coburn S. Aging and sleep apnea: action of benzodiazepine, acetazolamide, alcohol, and sleepderivation in a healthy elderly group. J Gerontoll984; 39:655-61. 24. Timms RM, Dawson A, Hajukovic RM, Mi-
tler MM. Effect of triazolam on sleep and arterial oxygensaturation in patients with chronic obstructive pulmonary disease. Arch Intern Med 1988; 148:2159-63. 25. Bonnet MH, Dexter JR, Arand DL. The effect of triazolam on arousal and respiration in central sleep apnea patients. Sleep 1990; 13:31-41. 26. McNicholas WT, Coffey M, McDonnell T, O'Reagan R, Fitzgerald MX. Upper airwayobstruction during sleep in normal subjects after selective topical oropharyngeal anesthesia. Am Rev Respir Dis 1987; 135:1316-9. 27. Basner R, Ringler J, Garpestad E, Schwartzstein R, Weinberger S, Weiss JW. Upper airway anesthesia prolongs the time to arousal from airway occlusion induced during NREM sleep (abstract). Am Rev Respir Dis 1990; 141:AI27.
