Sleep deprivation is a common complication in the Intensive Care Unit (ICU), resulting from anxiety, fear, pain, medication use, and a disruptive environment (1). Clinical manifestations of sleep loss in the ICU include agitation, psychosis, and respiratory weakness as well as immune and endocrine dysfunction (2–5). Sleep deprivation may thus potentially alter outcomes in critically ill patients, delaying ICU discharge and increasing overall costs.
The rapid-acting sedative propofol has allowed physicians to continuously sedate patients for extended periods without significantly delaying emergence (6). Although this strategy has been advocated to promote sleep (7), data to support it are lacking. By altering the regulation of sleep and wakefulness, prolonged periods of sedation may interfere with the natural generation of sleep/wake cycles. As a result, continuous sedation may not promote, but instead adversely affect, the quality of concurrent sleep.
We hypothesized that continuous sedation prevents the normal restorative functions of physiologic sleep. A sufficiently long period of sedation would then result in a sleep-deprived state upon emergence. Such a state would be characterized by rebound increases in rapid eye movement (REM) and/or non-REM sleep (8). To test this hypothesis, we timed a 12-h period of continuous sedation in rats to overlap with their normal sleep period and subsequently monitored their behavior and electroencephalogram (EEG) for signs of sleep deprivation.
After approval from the animal care committee at our institution and in accordance with animal care guidelines, 12 male Sprague-Dawley rats weighing 275–300 g (Harlan Industries, Indianapolis IN) were anesthetized with intraperitoneal ketamine 80 mg/kg and xylazine 40 mg/kg. A venous catheter was placed in the jugular vein via a sterile cutdown and tunneled to exit through the scalp. A temperature sensor-transmitter (Minimitter Co., Sun River, OR) was implanted in the peritoneal cavity through a midline incision. Four stainless steel screws were then implanted in the skull as dural EEG electrodes and connected to an amphenol socket (Newark Electronics, Chicago, IL) by short lengths of Teflon-coated, stainless steel wire. Two other lengths of this wire were also implanted in the neck musculature as electromyographic (EMG) electrodes. The entire assembly of electrodes and amphenol socket was then cemented into place with dental acrylic. Wound edges were treated with a bacitracin, polymyxin, and neomycin ointment. Rats were allowed to recover for 1 wk in a light-controlled room with lights on between 8:00 am and 8:00 pm. Because rats are nocturnal animals, their normal sleep period in such an environment would be 8:00 am to 8:00 pm (9). Ad libitum access to food and water was provided.
After the 1-wk recovery period, the experimental rat was placed in the testing chamber 12 h before testing to acclimate to the environment. At the beginning of the wake period (8:00 pm) on the day of testing, the rat was monitored and the data were recorded. The EEG (bifrontal and fronto-occipital) data were monitored continuously for 8 h to obtain baseline characteristics (8:00 pm–4:00 am). A 12-in. cable (to allow the rat free movement) was used to attach the amphenol connector on the rat’s headset to the polygraph. Three traces representing bifrontal EEG, fronto-occipital EEG and EMG were recorded for each rat. The paper speed was 10 mm/s, and the pen was calibrated so that 1.0 cm of vertical deflection signified an electrical potential of 50 μV. The raw EEG was simultaneously digitized and fed to a Swan P66 computer running the Polyview data acquisition program (Astro-Med Graph, West Warwick, RI). Movement and temperature were also measured during this time using the Minimitter system (Minimitter Co. Sun River, OR). In this system, the directional FM transmitter implanted in the peritoneum emits a signal varying in frequency with changes in temperature. Fluctuations in the amplitude of the signal are read as movements. The signal is detected by using a receiver plate placed under the cage and decoded with the Minimitter computer algorithm.
Propofol was diluted to 5 mg/mL with 0.9 normal saline and administered IV via a Baxter AS40 infusion pump (Baxter, Deerfield, IL) through PE-20 polyethylene tubing connected to the tunneled venous catheter. The initial infusion rate for propofol was 25 mg · kg−1 · h−1 (416 μg · kg−1 · min−1). Once the clinical endpoints of immobility, loss of righting reflex, and willingness to accept the clip-style pulse oximeter probe were reached, the infusion rate was titrated downward by 5% every 15 min until the rat began to move spontaneously. At that point, the infusion rate was then increased until clinical endpoints were reestablished. This process continued throughout the 11.5-h recording period. EEG, temperature, and pulse oximetry were continually monitored and recorded every 15 min during the sedation period. A 25-W heating lamp was used intermittently to maintain temperature >36°C. Rats were sedated continuously for 11.5 h, until 7:30 pm (Fig. 1). The sedative infusion was then discontinued and the rat was allowed to recover for 30 min. To determine the effects of the intralipid vehicle alone, another 12 rats underwent the identical protocol, except that intralipid (5%) was infused instead of propofol at a rate equal to the average infusion rate in the Propofol group. EEG and movement data were collected as in the Propofol group.
After the 30-min recovery period (7:30 pm–8:00 pm) EEG, EMG, and movement data were again recorded for 8 h (8:00 pm–4:00 am) as described previously.
EEG recordings were hand-scored in 30-s epochs and collected into 4-h segments to quantify the percentage of time spent in REM sleep, non-REM sleep, and wakefulness. The average duration of non-REM sleep bouts (number of consecutive epochs of non-REM sleep) and the number of transitions between states were also computed. Scoring was performed according to criteria established previously (8) by two trained scorers blinded to the origin of the recording. To obtain δ activity during non-REM sleep, portions scored as non-REM sleep were visually identified on a 100-sample/s digitized copy of the EEG by correlation with the paper record. The Polyview data analysis program (Astro-Med Graph) was then used to obtain both total power and δ (0.5–4 Hz) power for sequential 10-s segments by Fourier analysis of the raw waveform. Delta activity for each epoch was expressed as percent of total power, and averaged over four 2-h segments. Movements were recorded as movements/min and averaged over 1-h intervals to obtain an average movement value for that time period. Both movement and δ power data were compared with baseline levels to determine the change in activity resulting from drug administration.
The time spent in non-REM or REM sleep after propofol or intralipid administration was expressed as a percentage of the total monitoring time and averaged among all rats receiving the same treatment. For each sleep stage, group means after drug infusion were compared with baseline values by using a paired, two-tailed t-test with a hypothesized difference of zero. Movement and δ power data for rats receiving propofol and intralipid were compared with baseline values by using the Wilcoxon’s ranked sum test.
Propofol infusion rates ranged between 228 and 561 μg · kg−1 · min−1 in all rats. The average infusion rate was 368 ± 22 μg · kg−1 · min−1, corresponding to approximately 1.3–1.8 mL/h. All rats recovered spontaneous motion and righting reflex within 30 min of discontinuation of the sedative infusion. The mean ages of the rats were comparable between the Propofol (82.2 ± 2.4 days) and Intralipid (84.1 ± 2.3 days) groups (P = 0.84).
In the Propofol group, EEG data were obtained on all 12 rats, and movement and Fourier analysis data were obtained on 11 of the 12 rats. Incomplete data collection prevented the recording of movement and Fourier analysis data on one rat. Complete data were obtained for all 12 rats in the control (Intralipid) group.
When compared with baseline, a statistically significant decrease in both non-REM and REM sleep was noted during the first 4 h after sedation (Table 1). This difference was not significant during the subsequent 4 h. No increase in any stage of sleep was noted in either segment. Control rats receiving intralipid showed no statistically significant change from baseline in either non-REM or REM sleep (Table 1).
When compared with baseline, the average duration of non-REM sleep bouts after sedation did not differ (P = 0.89 for propofol). Fewer transitions between states (non-REM, REM, and wake) were observed in the first 4 h after sedation only (P = 0.001), but not in the last 4 h (P = 0.28). This finding likely resulted from reduced total non-REM sleep. No differences from baseline in either sleep-bout duration (P = 0.49) or number of transitions (P = 0.83) were noted in rats receiving intralipid.
For rats receiving propofol, EEG power in the δ (0.5–4 Hz) band was significantly reduced from baseline only in the first 2 h after sedation. In control rats, δ power after sedation did not differ from baseline. A repeated measures analysis of variance did not find statistical significance between Propofol and Intralipid groups in the change in δ activity (P = 0.25) or movement frequency (P = 0.85).
When compared with baseline, movement frequency was reduced in the first hour after propofol sedation, but returned to baseline values for each subsequent hour (P = 0.018, Fig. 2). Control rats demonstrated no difference in movement frequency before and after intralipid infusion.
Although difficult to obtain in the ICU (1), adequate sleep is necessary for physiologic and behavioral homeostasis in humans (8). The behavioral effects of sleep deprivation include irritability, disorientation, agitation, and psychosis (2). Sleep deprivation also alters respiratory muscle strength (5), CO2 and O2 homeostasis (10), and immune system function (4). These perturbations are readily reversible with a period of physiologic sleep (8).
In addition to being sleep deprived, patients in the ICU are often sedated continuously for days or weeks to facilitate care. By allaying anxiety and pain, such therapy has been advocated to facilitate physiologic sleep and thus reverse the detrimental consequences of sleep loss (7). Sedative-induced unresponsiveness, however, differs both behaviorally and physiologically from naturally occurring sleep. As a result, the physiologic state produced by continuous sedation may not effectively reverse the consequences of sleep deprivation, but instead produce a chronically sleep-deprived state.
The relationship between physiologic characteristics of sleep and the ability of sleep to reverse effects of sleep deprivation is incompletely understood. There is no clear correlation between sleep stage and reversal of sleep deprivation. In rats, selective deprivation of either REM or non-REM sleep is ultimately lethal (11,12). In humans, however, tricyclic antidepressants often improve daytime sleepiness in depressed patients despite pharmacologically suppressing REM sleep (13). Two aspects of the sleep state do affect the ability of sleep to reverse effects of sleep deprivation. Studies in volunteers and patients with sleep apnea indicate that limiting the duration of sleep (14), and repeatedly interrupting that sleep (15) both worsen its restorative effect. By raising the threshold for arousing stimuli, continuous sedation may thus potentially improve the restorative effects of sleep.
Conversely, sedation may disrupt the normal neurophysiology of the sleep state. Sleep in animals and humans is characterized by regular, synchronized electrical oscillatory behavior in thalamic neurons. Sedative-type drugs might disrupt this behavior (16), possibly reducing the ability of sleep to discharge accumulated sleep need.
We used a well defined rat model of sleep to differentiate between these possibilities. Rats respond to short periods of deprivation (less than four days) with rebound increases primarily in non-REM sleep and δ activity (8). Longer periods of deprivation, however, result primarily in REM rebound with little or no increase in non-REM sleep or δ activity (8). To maximize the potential for sedative-induced disruption of sleep in our study, we timed the sedation period to overlap completely with the rat’s normal sleep phase. Complete disruption of naturally occurring sleep by concurrent sedation would thus result in rebound increases in non-REM sleep and δ activity without inducing the tolerance to sedation and disuse neuropathy characteristic of longer periods of sedation.
We noted a statistically significant decrease in non-REM and REM sleep during the first four hours after the sedation period. Two possibilities explain this surprising result. First, lingering effects of propofol may perturb sleep during recovery, preventing the rat from expressing an increased need for sleep. We believed that this explanation was unlikely. Although we noted decreased movement in the first hour after emergence, the rapid offset of propofol would render any prolonged sedative effect unusual. Also, because the normal response to sleep deprivation is increased drowsiness, any postulated after-effect of propofol would have to enhance wakefulness to explain the reduction in sleep we observed. In addition, we noted no difference from baseline in the duration of non-REM sleep bouts during recovery, further arguing against a drug-induced perturbation in sleep regulation. Although it was possible that sedation delayed the onset of recovery sleep, no increase in sleep was observed in the eight hours after sedative discontinuation.
A second explanation for the reduction in sleep after sedation is that the lack of rebound sleep accurately reflects a reduced need to sleep. Our results might then be interpreted to mean that sedation enhanced the normal restorative effect of the sleep cycle. This possibility is plausible because rats normally sleep approximately 80% of the time during their normal sleep phase (9). By increasing that fraction with the use of sedation, sleep duration (and therefore its restorative effect) may be improved.
Alternatively, sedation may itself reverse the effects of sleep deprivation. Locating the sedation period during the waking phase rather than the sleep phase may differentiate between these possibilities. If sedation were itself restorative, for example, administration during the waking phase might then reduce the need to sleep during the subsequent sleep phase. Evidence that hypothalamic administration of propofol induces sleep in a manner similar to benzodiazepines (17) implies that propofol may induce sleep regardless of the natural propensity to sleep.
Our finding of reduced sleep after sedation differs slightly from previous studies of the effects of general anesthesia on subsequent sleep in humans (18–20). In those studies, recovery sleep after anesthesia resulted in lighter stages of sleep and a delayed REM rebound, but no change in total sleep times. Several factors may have accounted for the differences in our study. Patients were anesthetized only for a limited time, and during their natural waking period. Most underwent surgery as well as anesthesia. In contrast, we sedated rats during their natural sleep period, and did not apply a surgical stimulus. Previous studies also used inhaled anesthetics, which may have had different effects on restorative sleep than the use of propofol in our rats. Isoflurane, for example, prevents the electrical oscillatory behavior in thalamic neurons characteristic of sleep (21).
We did not measure blood levels of propofol in our rats, but titrated the infusion to behavioral endpoints. The average infusion rate in all rats was similar when adjusted for weight, and time to recovery of righting reflex was <30 minutes in all cases. We also monitored the EEG for only eight hours after the sedation period. Although EEG evidence of the rebound response to sleep deprivation usually occurs within that time period (12), a rebound effect occurring after the eight-hour measurement period may have been missed. This possibility would be unlikely, however, given the rapid clearance of propofol.
In the operating room, where anesthetics rarely exceed 6–8 hours, the interaction between sleep and anesthesia may not affect outcomes. In the ICU, where patients may be anesthetized for days to weeks, the possible distorting effects of sedation on sleep may play a larger role. Minimum alveolar anesthetic concentration requirements in rats are 10%–15% less during the circadian night than in the daytime (22), suggesting an interaction between circadian rhythms and the effects of inhaled anesthetics. Patterns of cortisol (23) and melatonin (24) secretion are nonphysiologic during prolonged sedation, also demonstrating an effect of sedation on normal circadian behavior. Although the long-term effects of such perturbations are unknown, one recent study found that allowing continuously sedated patients to emerge from sedation daily resulted in a reduced duration of mechanical ventilation and ICU length of stay (25). Although these data suggest that continuous sedation should not be administered for extended periods, they do not contraindicate the use of shorter periods of sedation, as modeled in our study. Instead, such data imply that a series of shorter periods of sedation shorten recovery when compared with longer periods.
We conclude that a 12-hour period of continuous sedation with propofol in the rat is compatible with a restorative process similar to that of naturally occurring sleep. We cannot distinguish whether sedation itself reduces the degree of sleep deprivation, or whether sedation facilitates the simultaneous generation of restorative sleep. Nevertheless, these data suggest that short periods of continuous sedation in the ICU do not prevent, and may enhance, the recovery from sleep deprivation.
The authors thank Dr. Jonathan Moss for review of the manuscript, and acknowledge the excellent technical assistance of Ms. Jaimee Huth.
1. Krachman SL, D’Alonzo GE, Criner GJ. Sleep in the intensive care unit. Chest 1995; 107: 1713–20.
2. Kornfeld DS. Psychiatric view of the intensive care unit. BMJ 1969; 1: 108–10.
3. Spiegel K, Leproult R, Van Cauter E. Impact of sleep debt on metabolic and endocrine function. Lancet 1999; 354: 1435–9.
4. Brown R, Pang G, Husband AJ, King MG. Suppression of immunity to influenza virus infection in the respiratory tract following sleep disturbance. Reg Immunol 1989; 2: 321–5.
5. Chen HI, Tang YR. Sleep loss impairs inspiratory muscle endurance. Am Rev Respir Dis 1989; 140: 907–9.
6. Kress JP, O’Connor MF, Pohlman AS, et al. Sedation of critically ill patients during mechanical ventilation: a comparison of propofol and midazolam. Am J Respir Crit Care Med 1996; 153: 1012–8.
7. Mendel PR, White PF. Sedation of the critically ill patient. Int Anesthesiol Clin 1993; 31: 185–200.
8. Rechtschaffen A, Bergmann BM, Gilliland MA, Bauer K. Effects of method, duration, and sleep stage on rebounds from sleep deprivation in the rat. Sleep 1999; 22: 11–31.
9. Bergmann BM, Mistlberger RE, Rechtschaffen A. Period-amplitude analysis of rat electroencephalogram: stage and diurnal variations and effects of suprachiasmatic nuclei lesions. Sleep 1987; 10: 523–36.
10. White DP, Douglas NJ, Pickett CK, et al. Sleep deprivation and the control of ventilation. Am Rev Respir Dis 1983; 128: 984–6.
11. Kushida CA, Bergmann BM, Rechtschaffen A. Sleep deprivation in the rat. IV. Paradoxical sleep deprivation. Sleep 1989; 12: 22–30.
12. Gilliland MA, Bergmann BM, Rechtschaffen A. Sleep deprivation in the rat. VIII. High EEG amplitude sleep deprivation. Sleep 1989; 12: 53–9.
13. Dunleavy DL, Brezinova V, Oswald I, et al. Changes during weeks in effects of tricyclic drugs on the human sleeping brain. Br J Psychiatry 1972; 120: 663–72.
14. Rosenthal L, Roehrs TA, Rosen A, Roth T. Level of sleepiness and total sleep time following various time in bed conditions. Sleep 1993; 16: 226–32.
15. Roehrs T, Zorick F, Wittig R, et al. Predictors of objective level of daytime sleepiness in patients with sleep-related breathing disorders. Chest 1989; 95: 1202–6.
16. Destexhe A, Contreras D, Steriade M. Cortically induced coherence of a thalamic-generated oscillation. Neuroscience 1999; 92: 427–43.
17. Tung A, Mendelson W. Are anesthetics sleeping pills? Sleep induction by injection of propofol into the medial preoptic area of the rat Anesth Analg 2000; 90: S474.
18. Moote CA, Knill RL. Isoflurane anesthesia causes a transient alteration in nocturnal sleep. Anesthesiology 1988; 69: 327–31.
19. Rosenberg-Adamsen S, Skarbye M, Wildschiodtz G, et al. Sleep after laparoscopic cholecystectomy. Br J Anaesth 1996; 77: 572–5.
20. Knill RL, Moote CA, Skinner MI, Rose EA. Anesthesia with abdominal surgery leads to intense REM sleep during the first postoperative week. Anesthesiology 1990; 73: 52–61.
21. Ries CR, Puil E. Isoflurane prevents transitions to tonic and burst firing modes in thalamic neurons. Neurosci Lett 1993; 159: 91–4.
22. Munson ES, Martucci RW, Smith RE. Circadian variations in anesthetic requirement and toxicity in rats. Anesthesiology 1970; 32: 507–14.
23. Von Werder K, Stevens WC, Cromwell TH, et al. Adrenal function during long-term anesthesia in man. Proc Soc Exp Biol Med 1970; 135: 854–8.
24. Reber A, Huber PR, Ummenhofer W, et al. General anaesthesia for surgery can influence circulating melatonin during daylight hours. Acta Anaesthesiol Scand 1998; 42: 1050–6.
25. Kress JP, Pohlman AS, O’Connor MF, et al. Daily interruption of sedative infusions in critically ill patients undergoing mechanical ventilation. N Engl J Med 2000; 342: 1471–7.