Contribution of the Intensive Care Unit Environment to Sleep Disruption in Mechanically Ventilated Patients and Healthy Subjects

Transcription

1 Contribution of the Intensive Care Unit Environment to Sleep Disruption in Mechanically Ventilated Patients and Healthy Subjects Jonathan Y. Gabor, Andrew B. Cooper, Shelley A. Crombach, Bert Lee, Nisha Kadikar, Harald E. Bettger, and Patrick J. Hanly Department of Medicine, St. Michael s Hospital; Department of Critical Care Medicine, Sunnybrook and Women s College Health Sciences Centre, University of Toronto, Toronto, Ontario, Canada Recent studies have challenged the traditional hypothesis that ex- been only indirectly established through simulated laboratory cessive environmental noise is central to the etiology of sleep disrup- studies on healthy subjects (15 17) and correlations between tion in the intensive care unit (ICU). We characterized potentially noise levels and arousal frequencies in ICU patients (18). disruptive ICU noise stimuli and patient-care activities and deter- Studies on healthy subjects in sleep laboratories indicated mined their relative contributions to sleep disruption. Furthermore, that audiotape recordings of ICU noise disrupted sleep and we studied the effect of noise in isolation by placing healthy subjects that reducing noise, either through earplugs or on quiet in the ICU in both normal and noise-reduced locations. Seven mecontrol nights, improved sleep. However, a sleep laboratory chanically ventilated patients and six healthy subjects were studied by continuous 24-hour polysomnography with time-synchronized cannot simulate the full auditory and visual experience of environmental monitoring. Sound elevations occurred the ICU. In addition, the studies were only performed for times per hour of sleep and were responsible for % of an 8-hour nocturnal period rather than over 24 hours. Theretotal arousals and awakenings. Patient-care activities occurred fore, a more realistic evaluation of the effects of the ICU times per hour of sleep and were responsible for % environment would be obtained by studying patients and of total arousals and awakenings. Healthy subjects slept relatively healthy subjects in the ICU for at least a 24-hour period, with well in the typically loud ICU environment and experienced a quantipolysomnography (PSG) and time-synchronized recording of tative, but not qualitative, improvement in sleep in a noise-reduced, environmental variables to quantify their contribution to single-patient ICU room. Our data indicate that noise and patientsleep disruption. care activities account for less than 30% of arousals and awakenings and suggest that other elements of the critically ill patient s environated the contribution of noise to sleep disruption in ICU Two studies by Freedman and colleagues (8, 19) re-evalu- ment or treatment should be investigated in the pathogenesis of ICU sleep disruption. patients. A detailed questionnaire administered to patients Keywords: sleep disruption; intensive care unit noise; polysomnography after discharge from the ICU indicated that assessment of vital signs and phlebotomy were considered more disruptive Previous studies have determined that acutely ill patients in than noise (19). This study had a large sample size and was the intensive care unit (ICU) suffer unique sleep disturbances thorough in its design and analysis but was limited by poten- (1 8). Sleep is fragmented by frequent arousals and awakensleep tial recall bias and the lack of objective measurement of ings, resulting in decreased or absent slow-wave and REM quality. The subsequent study (8), using PSG and time- sleep. The circadian rhythm of sleep is distorted, with nearly synchronized recording of environmental noise, directly half of the total sleep time occurring during the daytime. linked noise to arousals from sleep and determined that noise Although sleep disruption in ICU patients has been unequiv- was responsible for only 15% of all arousals and awakenings. ocally confirmed, little is known about its etiology. However, Although it was the first study to demonstrate that common potential sleep-disrupting factors such as noise, frequent pa- noise elevations directly cause arousals in ICU patients, other tient-care activities, medication effects, acute and chronic environmental factors such as patient-care activities were not illness, and dyssynchrony between the patient and the ventilaassessed. Furthermore, the extent to which different noise tor are hypothesized as being the causes of this syndrome. sources, such as alarms or conversation, contributed to sleep Environmental stimuli, particularly noise, have been predisruption was not documented. sumed to be the most disruptive factors in the ICU. Although numerous studies have documented excessive noise levels in Therefore, the relative contribution of noise and other the ICU (9 14), the link to sleep disruption has, until recently, components of the ICU environment to the pathogenesis of ICU sleep disruption are unknown. Because previous studies have not used PSG for objective sleep quality assessment, or have used simulated ICU environments or have not evaluated factors other than noise, the impact of these potential sleep (Received in original form January 31, 2002; accepted in final form November 22, 2002) disruptors on sleep continuity has been incompletely as- Supported by St. Michael s Hospital Foundation; Canadian Intensive Care Foundasessed. The objectives of our study were to (1) determine tion; Department of Anesthesia, University of Toronto; and Department of Anaesthe prevalence of excessive noise and patient-care activities thesia, Sunnybrook and Women s College Health Sciences Centre. Correspondence and requests for reprints should be addressed to Patrick J. Hanly, over a 24-hour period, (2) determine the relative impact M.D., Room 6-049, St. Michael s Hospital, 30 Bond Street, Toronto, ON, M5B of these factors on sleep continuity, (3) monitor healthy, 1W8 Canada. hanlyp@smh.toronto.on.ca unattended individuals in the actual ICU environment to This article has an online supplement, which is accessible from this issue s table examine, in isolation, the effect of noise on sleep quality as of contents online at compared with that in critically ill patients, and (4) evaluate Am J Respir Crit Care Med Vol 167. pp , 2003 DOI: /rccm the effectiveness of a noise-reduction strategy by monitoring Internet address: these healthy subjects in a single room in the ICU.

2 Gabor, Cooper, Crombach, et al.: Sleep Disruption in the ICU 709 Figure 1. Polysomnographic example of noise-induced arousal. Arrow indicates abrupt increase in noise (68 db[a]) due to an alarm, followed by arousal from Stage 2 non-rem sleep. EOG, electrooculogram; EMG, electromyogram. C4 and O1 refer to the placement of electrodes over the central and occipital lobes, respectively. METHODS Subject Recruitment The study was approved by the research ethics boards of both participat- ing institutions. Patients or their substitute decision-makers provided written consent to participate in the protocol. Patients admitted to the Critical Care Unit of Sunnybrook and Women s College Health Sciences Centre were screened for eligibility according to previously published exclusion criteria for reliable PSG (7). Selection criteria were endotracheal intubation and anticipated mechanical ventilation for a further 24 hours. Healthy volunteers were recruited by advertising on the St. Michael s Hospital network and were excluded on the basis of previous medical history, sleep disorders, or exposure to the ICU environment. Study Protocol Mechanically ventilated patients were studied in the Critical Care Unit of Sunnybrook and Women s College Health Sciences Centre, an 18- bed ICU, with beds arranged in a semicircular open-plan, separated by curtains and a separate row of 2-patient and 4-patient rooms. Healthy subjects were studied in the medical/surgical ICU of St. Michael s Hos- pital, a 24-bed ICU, with 19 beds arranged in a semicircular open-plan, separated by curtains, and 5 beds in enclosed single-patient rooms. Healthy subjects were randomized to spend a 24-hour period in one of two ICU locations a single-patient, enclosed room or one of several beds in the open-plan ICU and were then crossed-over to the other location at least 1 week later to avoid acclimatization. After each study, healthy subjects completed a questionnaire to rate sleep quality and the main sources of sleep disruption. PSG All subjects were monitored with continuous and attended 24-hour PSG using Sandman DOS 2.4 software (NPB-Melville, Ottawa, ON, Canada). Polysomnographs and arousals from sleep were scored ac- cording to standard criteria (20, 21). Arousals or awakenings caused by noise were defined as those occurring within 3 seconds of the termination of the noise (Figure 1). Arousals or awakenings caused by patientcare activities were defined as those occurring within 3 seconds of the termination of the patient-care activity. Arousals and awakenings not caused by noise or patient-care activities were classified as unidentifi- able. Environmental Monitoring Each subject s immediate environment was continuously monitored using a sound meter and infrared camera synchronized to the PSG. In addition to background sound intensity, the magnitude, source, and disruptive effect of each abrupt sound elevation above 10 db(a) (A-weighted decibel scale) was recorded. Each interaction between the patient and a member of the critical care team, sleep technician, or visitor that occurred during sleep was classified by type, and its effect on sleep continuity was documented. Statistical Analysis All data are reported as mean standard deviation. Paired t tests were used to compare noise and sleep data between healthy subjects in the two ICU locations. Unpaired t tests were used to compare noise and sleep data between patients and healthy subjects. One-way analyses of variance with post hoc Bonferroni tests were used to compare arousal/ awakening thresholds among sleep stages and the contribution of different noise sources to sleep disruption. Chi-square tests were used to compare relative proportions of noise-source sleep disruption between ICU locations. Statistical calculations were performed using SPSS 10.1 (SPSS, Chicago, IL). RESULTS Prevalence of Eligible Patients In all, 3,443 patient-days (daily screening result for an individual patient) were screened, 1,048 (30.4%) of which were new,

3 710 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL TABLE 1. PATIENT CHARACTERISTICS ON DAY OF Sleep Quality in the ICU POLYSOMNOGRAPHIC STUDY In comparison with healthy subjects in the open-plan ICU, there Age, yr was a tendency toward poorer sleep quality in the patient group M:F 7:0 (Table 3). Patients tended to have a higher awakening index Duration of stay, d and shorter sleep time compared with healthy subjects. The APACHE III score Mean arterial BP, mm Hg percentages of Stage 1 and Stage 2 non-rem and REM sleep Serum creatinine, mol/l were similar between patients and healthy subjects; however, BUN, mmol/l slow-wave sleep was significantly lower compared with healthy Positive blood cultures, : 5:2 subjects. Approximately half of patients sleep occurred during Inotropes, : 1:6 the daytime (6:00 a.m. to 10:00 p.m.), whereas the sleep of healthy Benzodiazepines, : 6:1 Opioids, : 4:3 subjects tended to be more nocturnal (Figures 2 and 3). Hypnotic agents, : 2:5 Healthy subjects in the single room experienced significant Antidepressants, : 1:6 improvements in total sleep time and night sleep compared with Butyrophenones/Phenothiazines, : 2:5 the open ICU (Table 3, Figure 4). However, sleep architecture and arousal and awakening indices were not significantly differ- Definition of abbreviations: APACHE III Acute Physiology and Chronic Health Evaluation III; BP blood pressure; BUN blood urea nitrogen. ent between the open ICU and the single room. No differences existed in any variable between those healthy subjects randomized to the single room first and those placed in the open ICU first. No healthy subject or patient demonstrated evidence of representing the first patient-day, and 2,395 (69.6%) of which sleep apnea syndrome. were repeats, representing subsequent patient-days. A total of Impact of Noise on Sleep Quality 1,296 (37.6%) patient-days passed screening criteria for reliable PSG. Of the excluded patient-days, 42.7% of exclusions were The majority of healthy subjects total arousals and awakenings due to central nervous system injury, 21.7% to Glasgow Coma in the open ICU were caused by sound peaks (Table 4), whereas Scale ranking below 10, and 8.8% to morphine equivalent sedasound. Of the sound peaks, abrupt elevations of over 10 db(a) a minority of patients arousals and awakenings were caused by tion above 10 g/kg/hour. Other factors, such as general anestheas defined in Methods, % resulted in an arousal or sia in the preceding 24 hours (4.2%), drug overdose (0.9%), and anticipated death in the following 24 hours (0.5%), were less awakening in the patients ICU. If peaks reaching 75 db or common reasons for exclusion. greater are considered separately, % resulted in an arousal or awakening in the patients open ICU. Peak decibel Subject Demographics thresholds that resulted in arousals or awakenings did not differ Seven male patients (24 to 82 years) and six healthy male subjects by sleep stage in any subject group. Although the magnitude of the sound level increase that resulted in an awakening was similar (23 to 65 years) were recruited. The number of days from ICU in all three locations, all other sound peak measures were signifiadmission to PSG ranged from 9 to 110. Five patients were admitcantly lower in the single room than in the other two locations. ted for respiratory insufficiency and two for multiple trauma. All In critically ill patients in the open ICU, no single noise patients were hemodynamically stable (systolic blood pressure source was predominantly responsible for sound-induced sleep 90 mm Hg). Mean total Acute Physiology and Chronic Health disruption. In the healthy subjects in the open ICU, alarm noises Evaluation III score (22) was (range 7 to 61). Four patients were less disruptive than were conversation or staff activities. were ventilated on pressure-support ventilation and three on The proportions of different noise types responsible for sleep assist-control ventilation (Table 1). disruption were not significantly different between patients and Sleep diaries of the healthy subjects indicated that they mainhealthy subjects in the open ICU. However, noise resulting from tained a regular nocturnal sleep schedule, with an estimated staff activities was the predominant source of sound-induced total sleep time of hours. Epworth Sleepiness Scale sleep disruption in the single room. The great majority of these scores were normal ( 6) for all subjects (mean ) (23). arousals and awakenings ( % and %) was All subjects were naive to PSG and the ICU environment and associated specifically with the opening of the main door of the had no history of sleep disorders. ICU, which was in close proximity to the single room. Without the main-door effect on the single room, the arousal and awaken- Sound Intensities in the ICU Locations ing indices would have been events/hour, which would Noise levels were similar in the open ICU for patients and healthy have been significantly different from the arousal and awakening subjects (Table 2), reflected by similar mean and mean maximum indices in the open ICU (p 0.05). noise levels, both during the day and night as well as during wakefulness and sleep. The percent of total sleep time in which Impact of Patient-Care Activities on Sleep Quality ambient sound levels exceeded 60 and 70 db, the maximum Patient-care interactions occurred times/hour of sleep decibel level attained by sound peaks, the magnitude of the (Table 5), mostly due to nursing activities such as dressing care, peaks increase, and the frequency of sound peaks greater than adjustment of intravenous drips, and administering medications. 75 db were also similar between the two locations. However, the Family visits and medical care activities that caused sleep disruptotal number of sound peaks per hour of sleep was significantly tion were of longer duration than those that did not (05:22 greater in the healthy subjects open ICU compared with the 05:23 minutes versus 01:00 01:24 minutes for family visits, p patients open ICU. In comparison with the open ICU, subjects 0.05; 00:58 01:05 seconds versus 00:35 00:58 seconds for in the single room experienced significantly reduced noise. Mean medical care activities, p 0.05). Of patient-care activities, and mean maximum decibel levels, as well as sound peak mea % resulted in sleep disruption and were responsible sures, were significantly lower in the single-room ICU versus the open ICU. for % of the total number of arousals and awakenings (Table 5). Despite our comprehensive monitoring, the etiology

4 Gabor, Cooper, Crombach, et al.: Sleep Disruption in the ICU 711 TABLE 2. SOUND INTENSITY IN THE THREE INTENSIVE CARE UNIT LOCATIONS Healthy Subjects (SMH) Patients (SWC) Open ICU Open ICU Single Room Mean (night), db Mean (day), db Mean maximum (night), db Mean maximum (day), db Mean (wakefulness), db Mean (sleep), db Mean maximum (wakefulness), db Mean maximum (sleep), db Wake 60 db, % Wake 70 db, % Sleep 60 db, % Sleep 70 db, % None Sound peaks/hr of sleep, No * Mean sound peak, db Mean increase of peak, db Peaks 75 db/hr wake, No Peaks 75 db/hr sleep, No Definition of abbreviations: ICU intensive care unit; SMH St. Michael s Hospital; SWC Sunnybrook and Women s College Health Sciences Centre. *p 0.05 between patients and healthy subjects in the open-icu. All differences between the single room and the open ICU are significant at the p 0.05 level except the peaks greater than 75 db per hour of wakefulness. healthy subjects and to estimate the effectiveness of a noise reduction strategy. Our findings demonstrate that although loud noise and frequent patient-care activities were prevalent in the ICU environment, they were responsible for only a small proportion of the observed sleep disruption. Healthy individuals slept relatively well in this potentially disruptive environment, and although noise accounted for a significant proportion of sleep disruption in this group, its extent was not pathologic. A quanti- tative improvement in sleep quality was observed as a result of noise reduction; however, there was no change in sleep architec- ture. Numerous studies have examined noise levels in the ICU, and all have concluded that they exceed Environmental Protection Agency recommendations for hospitals, which are less than 45 db(a) during the daytime and greater than 35 db(a) at night (24). Rather, mean sound levels are usually in the db(a) range, with sound peaks greater than 80 db(a) (8 14). In contrast to previous studies, we quantified all sound peaks that of the majority of arousals and awakenings ( %) could not be identified. Subjective Perception of Sleep-Disrupting Factors Sleep quality in the open ICU was rated as significantly worse than in the single room (p 0.05) and worse than at home (p 0.05, Table 6). In the open ICU, subjects identified alarms and staff conversation as the most disruptive environmental noises, a result identical to that observed by Freedman in a survey of ICU patients at discharge (19). Subjects rated noise from televisions and pagers as least disruptive, again similar to ICU patients (19). In the single room, noise and its component sources were not considered significantly disruptive to sleep. DISCUSSION This is the first study to systemically determine the contribution of the ICU environment to sleep disruption in both patients and TABLE 3. SLEEP ARCHITECTURE IN THE THREE INTENSIVE CARE UNIT LOCATIONS Healthy Subjects (SMH) Patients (SWC) Open ICU Open ICU Single Room Total sleep time, hr Night sleep, hr Day sleep, hr Day sleep, % Stage 1, %TST Stage 2, %TST SWS, %TST * REM, %TST Arousals per hr Awakenings per hr Arousals awakenings per hr Definition of abbreviations: ICU intensive care unit; SMH St. Michael s Hospital; SWC Sunnybrook and Women s College Health Sciences Centre; SWS slow-wave sleep; TST total sleep time. *p 0.05 versus healthy subjects in the open ICU. p 0.05 versus healthy subjects in the open ICU.

5 712 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL Figure 2. Twenty-four hour hypnograms of each patient in the open intensive care unit. Stages 1, 2, 3, and 4, of non-rem sleep. Many studies have demonstrated unequivocal sleep disruption in critically ill patients, characterized by extreme sleep frag- mentation, an overrepresentation of Stage 1 and Stage 2 non- REM sleep, reduced or absent slow-wave sleep and REM sleep, and circadian rhythm abnormalities (1 8). We observed similar findings in our patient cohort. Healthy subjects in the open ICU experienced a modest decrease in the proportion of slow-wave sleep and REM sleep and a concomitant increase in the propor- tion of Stage 1 and Stage 2 non-rem sleep, perhaps partly due increased by more than 10 db(a) (which represents a doubling of sound intensity) because the change in noise may be as important to the pathogenesis of sleep disruption as the actual decibel level achieved (8). We observed 37 and 72 sound peaks/hour of sleep in the open ICUs of patients and healthy subjects, respectively, which indicated frequent noise spiking. Consequently, we conclude that our open ICUs are sources of excessive noise that may cause sleep disruption, which is consistent with the previous literature. Figure 3. Twenty-four hour hypnograms of each healthy subject in the open intensive care unit. Stages 1, 2, 3, and 4 of non-rem sleep.

6 Gabor, Cooper, Crombach, et al.: Sleep Disruption in the ICU 713 Figure 4. Twenty-four hour hypnograms of each healthy subject in the single-room intensive care unit. Stages 1, 2, 3, and 4 of non-rem sleep. ties and mechanical ventilation did not exist. Previous work on arousal responses to acoustic stimulation in healthy individuals observed that 35% of sound spikes of 85 db intensity resulted in an arousal (26), which is similar to our results with a threshold of 75 db. These investigators observed sleep stage dependent changes in arousal frequency for a fixed sound intensity, whereas we observed no significant differences in the peak sound intensity that caused sleep disruption across sleep stages and between patients and healthy subjects. This is also the first study to directly quantify the effect of patient-care activities on sleep continuity. In the first study to characterize sleep disruption in the ICU by PSG, Hilton (1) monitored patient-care activities. However, she did not directly determine their contribution to sleep disruption with PSG syn- chronization. In our study, patient-care activities (which included to unrestricted daytime napping, as approximately 30% of sleep occurred during the day. The frequency of arousals and awakenings was within normal limits (25), supporting the more recent finding (8) that noise, which was the only known environmental factor intruding on the sleep of the healthy subjects, is not a significant source of sleep disruption. Only 20% of arousals and awakenings in our ICU patient cohort were identifiably due to noise peaks, demonstrating that, in contradiction of traditional hypotheses, environmental noise is not responsible for the majority of ICU sleep disruption instances. Although noise peaks occurred frequently, only 12% of these peaks, and 35% of peaks greater than 75 db(a), resulted in an arousal or awakening. Noise was responsible for the majority of sleep disruption in healthy subjects, likely because other potential sources of sleep disruption such as patient-care activi- TABLE 4. IMPACT OF NOISE ON SLEEP DISRUPTION Healthy Subjects (SMH) Patients (SWC) Open ICU Open ICU Single Room Arousals from sound, % * Awakenings from sound, % * Arousals and awakenings from peaks 75 db, % Sound-arousals from alarms, % Sound-arousals from talking, % Sound-arousals from ICU activities, % Sound-awakenings from alarms, % Sound-awakenings from talking, % Sound-awakenings from ICU activities, % Spike intensity causing waking, db * Spike intensity causing arousal, db Spike intensity with no effect, db Intensity increase causing waking, db Intensity increase causing arousal, db Intensity increase with no effect, db Definition of abbreviations: ICU intensive care unit; SMH St. Michael s Hospital; SWC Sunnybrook and Women s College Health Sciences Centre. *p 0.05 versus healthy subjects in the open ICU. All differences between healthy subjects in the open ICU and the single room are significant at the p 0.05 level except for the db increase that caused an awakening. The number of sound peaks 75 db during sleep was negligible in the single room.

7 714 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL TABLE 5. IMPACT OF PATIENT-CARE ACTIVITIES ON SLEEP DISRUPTION Event Type No. per hr of Sleep Percentage Causing Disruption Percentage of Total Disruption Sound Family visits Resp/Physio Suctioning RN visits Assess vitals Mx admin All medical care Apparatus/tech Unidentifiable Definition of abbreviations: Apparatus/tech. noises from the sleep/environmental monitoring equipment or actions by the attending research assistant; Mx admin. administering medication to a patient; Resp/physio any care provided by a respiratory therapist or physiotherapist; RN visits any care provided by a nurse; sound sound peaks. nursing visits, assessment of vital signs, and administering medi- Two previous studies have attempted to reduce noise in the cations) occurred approximately 8 times per hour of sleep, which ICU (27, 28). Walder and coworkers (27) were partially success- is in contrast with the occurrence of noise spikes 37 times per ful in reducing noise and light levels, and Kahn and colleagues hour of sleep. Approximately 20% of patient-care activities resulted (28), through a detailed and comprehensive behavior modifica- in an arousal or an awakening, which accounted for only tion program, significantly reduced mean peak noise levels and 7% of observed sleep disruption. Therefore, patient-care activities, the number of sound peaks less than 80 db(a). However, neither although frequent, were not a predominant source of sleep of these studies assessed the impact of these changes on sleep disruption in ICU patients. quality. Two overnight studies on healthy subjects by Topf and We used a questionnaire to subjectively assess the relative coworkers (15, 16) and one by Wallace and colleagues (17) contribution of different sleep-disrupting factors (19). In contrast simulated an ICU environment in the sleep laboratory by using to the findings of Freedman and coworkers, our healthy subjects audiotape-recorded ICU noise. Reduction of noise, either by perceived noise as highly disruptive to sleep in the open ICU, stopping the audiotape or the use of earplugs, was associated presumably because the more disruptive factors from their study with improved sleep quality. However, these studies were not (phlebotomy, assessment of vital signs) are not relevant to performed for 24 hours, and the sleep laboratory cannot simulate healthy subjects. As with Freedman and colleagues, however, the full auditory and visual experience of the ICU. In addition our healthy subjects perceived that sleep in the ICU was worse to being performed completely in the ICU, our study objectively than at home. We observed that conversation and alarms were assessed sleep quality in both the loud and noise-reduced environments. perceived to be the most disruptive noises, whereas noise from Although there was a quantitative improvement in televisions, telephones, and pagers were rated as the least disrup- sleep in the single room, sleep architecture was nearly identical tive. Subject perceptions did not match the objective PSG and arousal frequencies were normal in both locations. Mean environmental data. For example, alarms were rated second to and mean maximum sound intensities were significantly reduced conversation in terms of sleep disruption, yet they were responsi- in the single room, as were the number of extremely loud ( 75 ble for only 20 25% of arousals and awakenings. We suspect db[a]) sound peaks and the mean peak decibel level; however, that subjects may subconsciously bias their observations on the the frequency of sound spikes remained elevated, essentially the basis of the perceived degree of irritation from these noises while same as that experienced by our patient cohort. This may be one awake; clearly, the transient nature of arousals and awakenings explanation for a lack of improvement in sleep continuity despite makes it difficult for the subject to recall the event and its cause. a reduction in overall sound intensity and may suggest that the frequency and nature of sound peaks are important contributors to noise-induced sleep disruption. The single room was chosen as a realistic location for noise reduction because it is physically TABLE 6. SUBJECTIVE PERCEPTION OF SLEEP QUALITY BY isolated from the main open ICU. Any ICU noise registered in HEALTHY SUBJECTS IN THE OPEN INTENSIVE CARE UNIT the single room originated from leakage from the main ICU. AND SINGLE ROOM In our case, this was particularly evident with respect to the main Open ICU Single Room door of the ICU, which was responsible for a disproportionate amount of sleep disruption. However, when sleep disruption Sleep quality ICU * Sleep quality home caused by the main door was excluded, the arousal and awaken- Noise (all sources) * ing indices of subjects in the single room were significantly lower Staff conversation * than the arousal and awakening indices of subjects in the open Alarms * ICU. Despite the effect of the main door, total sleep time and Suctioning noises * nocturnal sleep time improved significantly in the single room, Telephone * Light perhaps because the noise-reduced location is more conducive Pagers to a return to sleep after an arousal or awakening. Alternatively, Television the comparative lack of visual and other distractions in the single room may have also played a role. Reductions in the frequency Definition of abbreviation: ICU intensive care unit. of irritating noises, noise intensity, and visual distractions, in Values are on a1to10scale. For ratings of sleep quality, 10 is best possible sleep, 1 is worst possible sleep. For ratings of sleep disruption, 10 is most addition to a longer total sleep time, may have contributed to disruptive, 1 is least disruptive. the subjective rating of improved sleep quality in the single room *p 0.05 between the open ICU and single room. despite no improvement in arousal and awakening indices.

8 Gabor, Cooper, Crombach, et al.: Sleep Disruption in the ICU 715 Our study has a number of limitations, which should be noted. PJ. Sleep in critically ill patients requiring mechanical ventilation. First, our sample sizes were small, which limited the power of Chest 2000;117: Freedman NS, Gazendam J, Levan L, Pack AI, Schwab RJ. Abnormal our statistical analyses. Nevertheless, our adherence to strict sleep/wake cycles and the effect of environmental noise on sleep disinclusion and exclusion criteria enabled us to avoid a wide heter- ruption in the intensive care unit. Am J Respir Crit Care Med 2001;163: ogeneity in our data, which resulted in relatively consistent PSG findings despite the fact that we recruited patients and healthy 9. Falk SA, Woods NF. Hospital noise levels and potential health hazards. NEnglJMed1973;289: subjects over a wide age range. Second, the design of the study 10. Bentley S, Murphy F, Dudley H. Perceived noise in surgical wards and did not allow healthy subjects to adapt to the ICU environment an intensive care area: an objective analysis. BMJ 1977;2: and habituate to noise intensity, which may have occurred in our 11. Redding JS, Hargest TS, Minsky SH. How noisy is intensive care? Crit patients admitted to the ICU for prolonged periods. However, Care Med 1977;5: habituation to noise among our healthy individuals would be 12. Soutar RL, Wilson JA. Does hospital noise disturb patients? BMJ 1986; 292:305. expected to reduce the prevalence of noise-related sleep disrup- 13. Balogh D, Kittinger E, Benzer A, Hackl JM. Noise in the ICU. Intensive tion even further. Therefore, our finding that excessive noise Care Med 1993;19: did not pathologically disrupt the sleep of healthy subjects, with- 14. Meyer TJ, Eveloff SE, Bauer MS, Schwartz WA, Hill NS, Millman RP. out the opportunity to habituate to noise, coupled with our Adverse environmental conditions in the respiratory and medical ICU observations in ICU patients, strengthens our suggestion that settings. Chest 1994;105: Topf M. Effects of personal control over hospital noise on sleep. Res noise is not a major sleep disrupter in the ICU. Nurs Health 1992;15: Given that the elements of the ICU environment assessed in 16. Topf M, Davis JE. Critical care unit noise and rapid-eye-movement this study, namely sound and patient-care activities, together (REM) sleep. Heart Lung 1993;22: accounted for less than 30% of observed sleep disruption, a 17. Wallace CJ, Robins J, Alvord LS, Walker JM. The effect of earplugs on sleep measures during exposure to simulated intensive care unit noise. significant proportion of sleep disruption in mechanically venti- Am J Crit Care 1999;8: lated ICU patients remains unexplained. One possibility is dys- 18. Aaron JN, Carlisle CC, Carskadon MA, Meyer TJ, Hill NS, Millman RP. synchrony between the patient and the ventilator. Preliminary Environmental noise as a cause of sleep disruption in an intermediate data (29) suggest that this may be dependent on the mode of respiratory care unit. Sleep 1996;19: ventilation, but this requires further study. If patient ventilator 19. Freedman NS, Kotzer N, Schwab RJ. Patient perception of sleep quality and etiology of sleep disruption in the intensive care unit. Am J Respir dyssynchrony is determined to be a major source of sleep disrup- Crit Care Med 1999;159: tion, this may be corrected by altering the mode of mechanical 20. Rechtschaffen A, Kales A. A manual of standardized terminology, techniques ventilation with the hope and expectation that improvement and scoring system for sleep stages of human subjects. Los of sleep quality will ultimately benefit the clinical outcome of Angeles: Brain Information Service/Brain Research Institute, UCLA; NIH Publication No critically ill patients. 21. American Sleep Disorders Assocation. EEG arousals: scoring rules and Acknowledgment : The authors are indebted to Dr. Hans Kunov and Dr. Taha examples. Sleep 1992;15: Jaffer at the Institute for Biomaterial and Biomedical Engineering and the Edwards 22. Knaus W, Wagner D, Draper E, Zimmerman J, Bergner M, Bastos P, S. Rogers Department of Electrical and Computer Engineering, University of Toprognostic system: risk prediction of hospital mortality for critically Sirio C, Murphy D, Lotring T, Damiano A, et al. The APACHE III ronto, who designed and built the apparatus for time-synchronization of PSG and environmental variables, and the Canadian Intensive Care Foundation and the ill hospitalized adults. Chest 1991;100: St. Michael s Hospital Foundation for financial support of this work. 23. Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep 1991;14: References 24. Environmental Protection Agency. Information on levels of environmental noise requisite to protect public health and welfare with an adequate 1. Hilton BA. Quantity and quality of patients sleep and sleep-disturbing margin of safety. Washington, DC: U.S. Government Printing Office; factors in a respiratory intensive care unit. J Adv Nurs 1976;1: No. 550/ Orr WC, Stahl ML. Sleep disturbances after open-heart surgery. Am J 25. Mathur R, Douglas NJ. Frequency of EEG arousals from nocturnal sleep Cardiol 1977;39: in normal subjects. Sleep 1995;18: Broughton R, Baron R. Sleep patterns in the intensive care unit and 26. Carley DW, Applebaum R, Basner RC, Onal E, Lopata M. Respiratory on the ward after acute myocardial infarction. Electroenceph Clin and arousal responses to acoustic stimulation. Chest 1997;112: Neurophysiol 1978;45: Walder B, Francioli D, Meyer JJ, Lancon M, Romand JA. Effects of 4. Kavey NB, Altshuler KZ. Sleep in herniorrhaphy patients. Am J Surg guidelines implementation in a surgical intensive care unit to control 1979;138: nighttime light and noise levels. Crit Care Med 2000;28: Aurell J, Elmqvist D. Sleep in the surgical intensive care unit: continuous 28. Kahn DM, Cook TE, Carlisle DO, Nelson DL, Kramer NR, Millman polygraphic recording of sleep in nine patients receiving postoperative RP. Identification and modification of environmental noise in an ICU care. BMJ 1985;290: setting. Chest 1998;114: Richards KC, Bairnsfather L. A description of night sleep patterns in 29. Parthasarathy S, Malik I, Jubran A, Tobin MJ. Does ventilator mode the critical care unit. Heart Lung 1988;17: influence sleep quality in critically ill patients? Am J Respir Crit Care 7. Cooper AB, Thornley KS, Young GB, Slutsky AS, Stewart TE, Hanly Med 2001;163:A483.