Effect of earplugs on polysomnographic sleep and description of sleep-disturbing factors in critically ill subjects

Consequences of sleep deprivation in critically ill patients may include impaired lymphocyte and granulocyte function, reduction of natural killer cell and lymphokine-activated killer cell number and activity, and delayed weaning from mechanical ventilation. Studies of the impact of sleep-promoting interventions in mechanically ventilated subjects are not available. The primary objective of this research was to measure the impact of soft foam earplugs to reduce noise during the nighttime hours on the sleep of critically ill subjects. An increase in the sleep maintenance efficiency index (SMEI) and rapid eye movement (REM) sleep was expected when earplugs were worn. Other sleep disrupters and sleep during the day and afternoon hours were measured. Patients more than 18 years of age, who met the enrollment criteria, were studied using a randomized, unblinded, crossover design. Polysomnographic sleep was measured on 2 nights, with one "washout" night in between. Earplugs were randomly assigned to be worn on 1 night. A variety of environment, care content, and care process variables was also measured. Polysomnographic sleep was measured during the day and afternoon hours in 7 subjects. The sample consisted of 5 males and 8 females, with average ± SD for age = 56.9 ± 20 years, days in intensive care unit (ICU) at enrollment = 12.6 ± 8.3. Twelve of the subjects were mechanically ventilated. Significantly more REM sleep was obtained on the night earplugs were worn, and the power to detect a difference in SMEI was inadequate. All 13 subjects' sleep architecture was severely disturbed compared to normal sleep architecture, with or without earplugs. A clear pattern of increased opportunity for sleep on the night earplugs were worn was evident. Subjects slept during the day and afternoon hours; however, sleep architecture remained grossly fragmented, with markedly reduced REM and slow-wave sleep (SWS). The clinical significance of the increase in REM sleep seen in this study is open to debate. The study provides a reasonable basis to use earplugs to promote sleep because earplugs did provide more REM sleep. Over time, the effects might be of clinical importance. Before the efficacy of sleep in critically ill patients can be tested, an effective intervention must be developed.

THE EFFECT OF EARPLUGS ON POLYSOMNOGRAPHIC SLEEP AND DESCRIPTION OF SLEEP-DISTURBING FACTORS IN CRITICALLY ILL SUBJECTS by Carrie Jane Wallace A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy College of Nursing The University of Utah December 1998 Copyright to Carrie Jane Wallace 1998 All Rights Reserved THE UNIVERSITY OF UTAH GRADUATE SCHOOL SUPERVISORY COMMITTEE APPROVAL of a dissertation submitted by Carrie Jane Wallace This dissertation has been read by each member of the following supervisory committee and by majority vote has been found to be satisfactory. ,l o,/ ;q) /rY Kathleen M. Baldwin C1k~~<~ f~± f// Thomas D. East ]llmes C. Reading // James M. Walker \ \ THE UNIVERSITY OF UTAH GRADUATE SCHOOL FINAL READING APPROVAL To the Graduate Council of the University of Utah: I have read the dissertation of Carrie Jane Wallace in its final form and have found that (1) its format, citations, and bibliographic style are consistent and acceptable; (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the supervisory committee and is ready for submission to The Graduate School. Date Karin T. Kirchhoff Chair, Supervisory Committee Approved for the Major Department ~~U:&4 Linda K. Amos Chair/Dean Approved for the Graduate Council David S. Chapman Dean of The Graduate School ABSTRACT Consequences of sleep deprivation in critically ill patients may include impaired lymphocyte and granulocyte function, reduction of natural killer cell and lymphokine-activated killer cell number and activity, and delayed weaning from mechanical ventilation. Studies of the impact of sleep-promoting interventions in mechanically ventilated subjects are not available. The primary objective of this research was to measure the impact of soft foam earplugs to reduce noise during the nighttime hours on the sleep of critically ill subjects. An increase in the sleep maintenance efficiency index (SMEI) and rapid eye movement (REM) sleep was expected when earplugs were worn. Other sleep disrupters and sleep during the day and afternoon hours were measured. Patients more than 18 years of age, who met the enrollment criteria, were studied using a randomized, unblinded, crossover design. Polysomnographic sleep was measured on 2 nights, with one "washout" night in between. Earplugs were randomly assigned to be worn on 1 night. A variety of environment, care content, and care process variables was also measured. Polysomnographic sleep was measured during the day and afternoon hours in 7 subjects. The sample consisted of 5 males and 8 females, with average ± SD for age = 56.9 ± 20 years, days in intensive care unit (lCU) at enrollment = 12.6 ± 8.3. Twelve of the subjects were mechanically ventilated. Significantly more REM sleep was obtained on the night earplugs were worn, and the power to detect a difference in SMEI was inadequate. All 13 subjects' sleep architecture was severely disturbed compared to normal sleep architecture, with or without earplugs. A clear pattern of increased opportunity for sleep on the night earplugs were worn was evident. Subjects slept during the day and afternoon hours; however, sleep architecture remained grossly fragmented, with markedly reduced REM and slow-wave sleep (SWS). The clinical significance of the increase in REM sleep seen in this study is open to debate. The study provides a reasonable basis to use earplugs to promote sleep because earplugs did provide more REM sleep. Over time, the effects might be of clinical importance. Before the efficacy of sleep in critically ill patients can be tested, an effective intervention must be developed. v TABLE OF CONTENTS Page ABSTRACT ......................................... IV LIST OF TABLES ..................................... IX LIST OF FIGURES ..................................... xi ACKNOWLEDGMENTS ................................ xu Chapter 1. INTRODUCTION AND REVIEW OF THE LITERATURE ........ 1 Background and Significance . . . . . . . . . . . . . . . . . . . . . . . . . .. 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 1 Definitions of Sleep ................. . . . . . . . . . . . 2 Organizing Framework for Sleep Research in the Intensive Care Unit .................................. 5 The Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 20 Significance for Nursing . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 22 II. PRELIMINARY WORK. . . . . . . . . . . . . . . . . . . . . . . . . . . .. 23 Development and Testing the Earplug Intervention . . . . . . . . . . . .. 23 Clinician Ability to Predict Intensive Care Unit Stay . . . . . . . . . . .. 24 Reliability of Measurement Tools ....................... 25 Anxiety and Sensory Alteration Measurement Tools . . . . . . .. 25 Sleep Disturbance Measurement . . . . . . . . . . . . . . . . . . .. 28 Technical Adequacy of Sleep Monitoring. . . . . . . . . . . . . . . . . .. 29 Problems Encountered in Preliminary Work . . . . . . . . . . . . . . . .. 30 Use of Eye Covers . . . . . . . . . . . . . . . . . . . . . . . . . . .. 30 Use of Earplugs ............................. 31 Informed Consent ............................ 32 Chapter Page III. RESEARCH DESIGN AND METHODS . . . . . . . . . . . . . . . . . .. 34 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 34 Setting ....................................... 35 Sample Size and Patient Availability ..................... 35 Patient Selection and Enrollment . . . . . . . . . . . . . . . . . . . . . . .. 35 Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 38 Study Procedures ......... . . . . . . . . . . . . . . . . . . . . . . .. 38 Adequacy of Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 42 The Hearing Handicap Scale . . . . . . . . . . . . . . . . . . . . .. 42 The LDS Hospital Sleep Disorder Questionnaire . . . . . . . . .. 42 Illness Severity: APACHE II . . . . . . . . . . . . . . . . . . . .. 43 Noise and Lighting . . . . . . . . . . . . . . . . . . . . . . . . . . .. 43 Ambient Temperature . . . . . . . . . . . . . . . . . . . . . . . . .. 44 Intensity of Care . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 45 Medications and Surgical Procedures . . . . . . . . . . . . . . . .. 45 Pain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 46 Sleep Opportunity ............................ 47 Sleep Position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 47 Anxiety .................................. 48 Intermediate Health Outcomes ......................... 48 Sleep Deprivation ............................ 48 Sensory Alteration . . . . . . . . . . . . . . . . . . . . . . . . . . .. 49 Circadian Disruption . . . . . . . . . . . . . . . . . . . . . . . . . .. 50 Time on Mechanical Ventilation . . . . . . . . . . . . . . . . . . .. 50 Outcomes at Hospital Discharge: Survival and Length of Stay ..... 50 Data Analysis ................................... 51 IV. RESULTS ..................................... 52 Screening and Enrollment . . . . . . . . . . . . . . . . . . . . . . . . . . .. 52 Demographic Data ................................ 55 Integrity and Safety of the Earplug Intervention . . . . . . . . . . . . . .. 55 Sleep Instrumentation and Measurement ................... 59 Effect of Earplugs on Sleep Maintenance Efficiency Index and Rapid Eye Movement Sleep: Research Question 1 . . . . . . . . . . . .. 59 Other Potential Sleep Disrupters: Research Question 2 .......... 61 Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 61 Lighting ................................ .. 61 Ambient Temperature . . . . . . . . . . . . . . . . . . . . . . . . .. 64 Care Content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 65 Intensity of Care .. . . . . . . . . . . . . . . . . . . . . . . . . . .. 65 vii Chapter Page Medications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 67 Surgical Procedures ........................... 70 Care Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 70 Pain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 70 Sleep Opportunity ............................ 70 Sleep Position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 73 Anxiety .................................. 75 Descriptive Evaluation of Potentially Important Sleep Disrupters .... 77 Intermediate Health Outcomes ......................... 78 Sleep Deprivation ............................ 78 Sensory Alteration . . . . . . . . . . . . . . . . . . . . . . . . . . .. 81 Circadian Disruption . . . . . . . . . . . . . . . . . . . . . . . . . .. 81 Time on Mechanical Ventilation . . . . . . . . . . . . . . . . . . .. 82 Outcomes at Hospital Discharge . . . . . . . . . . . . . . . . . . .. 82 Day and Afternoon Sleep: Research Question 3 . . . . . . . . .. 85 V. DISCUSSION AND RECOMMENDATIONS ................ 87 Sample Characteristics and Outcomes . . . . . . . . . . . . . . . . . . . .. 87 Conclusions .................................... 88 Research Question 1 . . . . . . . . . . . . . . . . . . . . . . . . . .. 88 Research Question 2 . . . . . . . . . . . . . . . . . . . . . . . . . .. 89 Research Question 3 . . . . . . . . . . . . . . . . . . . . . . . . . .. 93 Study Design and Measurement Issues .................... 94 Study Design Issues .. . . . . . . . . . . . . . . . . . . . . . . . .. 94 Measurement Issues ................ . . . . . . . . . .. 95 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 97 Appendices A. SUMMARY OF RECENT SLEEP DEPRIVATION STUDIES ..... 99 B. PILOT STUDY MANUSCRIPT . . . . . . . . . . . . . . . . . . . . . .. 109 C. SLEEP-MONITORING EQUIPMENT CONFIGURATION ...... 126 D. DEFINITIONS OF STANDARD SLEEP MEASURES ......... 128 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 132 V111 LIST OF TABLES 1. Definitions and Approximate Percentage of Night Sleep Time Spent in Sleep Stages for Normal Adult Humans . . . . . . . . . . . . . . . 4 2. The Motor Activity Assessment Scale. . . . . . . . . . . . . . . . . . . .. 26 3. Disturbance Categories and Definitions . . . . . . . . . . . . . . . . . . .. 29 4. Earplug Feasibility Sleep Questionnaire Results . . . . . . . . . . . . . .. 32 5. Enrollment Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 36 6. Computation of the Sleep Deprivation Threshold .... . . . . . . . . .. 49 7. "Other" Reasons for Exclusion . . . . . . . . . . . . . . . . . . . . . . . .. 54 8. Demographics and Illness Severity. . . . . . . . . . . . . . . . . . . . . .. 56 9. The Effect of Earplugs on Sleep Maintenance Efficiency Index and Rapid Eye Movement Sleep (N = 13) . . . . . . . . . . . . . . . . .. 60 10. Preenrollment Sedative, Analgesic, and Paralytic Medications ...... 69 11. Benzodiazepine Use (Mg) . . . . . . . . . . . . . . . . . . . . . . . . . . .. 78 12. Subjective Evaluation of the Effect of Potentially Important Sleep Disrupters on Rapid Eye Movement and Sleep Maintenance Efficiency Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 79 13. Sleep Maintenance Efficiency Index and Percentage of Time Spent in Sleep Stages (N = 13) ............................ 80 14. Outcomes at Hospital and Rehabilitation Unit Discharge (N = 13) ... 84 15. Night, Day, Afternoon, and Twenty-Four-Hour Total Sleep Measures (N = 7) ...................................... 86 Table Page 16. Sleep Deprivation Literature Review Summary .............. 100 17. Intensive Care Unit Sound-Level Data ................... 116 18. Noise Reduction Devices Evaluated . . . . . . . . . . . . . . . . . . . .. 119 19. Sleep and Sound-Level Data Summary ................... 123 20. Significant Post Hoc Pairwise Comparisons ................ 125 x LIST OF FIGURES Figure Page 1. Organizing Framework for Intensive Care Unit Sleep Research ...... 6 2. Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 39 3. Reasons for Exclusion and Refusal. . . . . . . . . . . . . . . . . . . . . .. 53 4. Sound-Level Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . .. 62 5. Overhead Lighting ................................ 63 6. Lighting Comparisons .............................. 65 7. Ambient Temperature Comparisons ...................... 66 8. Intensity of Care Comparisons ......................... 67 9. Pain and Morphine Equivalent Comparisons . . . . . . . . . . . . . . . .. 71 10. Average Undisturbed Time and Disturbance Frequency .......... 72 11. Average Frequency, Length, and Time Between Disturbances ...... 74 12. Disturbance Frequency by Time . . . . . . . . . . . . . . . . . . . . . . .. 75 13. Sleep Position Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 76 14. Motor Activity Assessment Scale Summary ................. 77 15. Sensory Alteration Summary .......................... 82 16. Body Temperature Data ......... . . . . . . . . . . . . . . . . . . .. 83 17. Study Design (N = 6 Healthy Volunteers) . . . . . . . . . . . . . . . .. 121 18. Sleep Monitoring Equipment Block Diagram ............... 127 ACKNOWLEDGMENTS I am indebted to Dr. Karln Kirchhoff for chairing my committee and allowing me the latitude I needed to accomplish my goals. Special thanks to Dr. Tom East for invaluable assistance in helping me meet the grant submission deadline and to Dr. Jim Walker for equipment loans, laboratory space, and assistance with sleep monitoring. I gratefully acknowledge grant support from the LDS Hospital Deseret Foundation for the preliminary work and the Agency for Health Care Policy and Research (1 R03 HS 09335-01) for the patient study. I also thank the following companies: SensorMedics, Yorba Linda, CA, for sleep­monitoring equipment and software; Bose Corporation, Framingham, MA, for noise-cancellation devices; Larson-Davis, Provo, UT, for sound-level meters, statistical software, and technical advice; Standard Supply, Salt Lake City, UT, for the ohm meter; and Argus Corporation, Newark, DE, for super-soft-foam earplugs. Thanks are also extended to supportive colleagues at the LDS Hospital, particularly Vicki Spuhler and Terry Clemmer. Thanks to Dave Wallace for his support. To Carl and Lydia Wallace, my precious daughters, both born during the course of this work, thank you for bringing your light and with it a renewed sense of purpose into my life. CHAPTER I INTRODUCTION AND REVIEW OF THE LITERATURE Backaround and Si&nificance Introduction Existing evidence suggests that sleep deprivation is a well-known phenomenon in the intensive care unit (lCU) (Aurell & Elmqvist, 1985; Broughton & Baron, 1978; Dohno, Paskewitz, Lynch, Gimbel, & Thomas, 1979; Edwards & Schuring, 1993; Fontaine, 1989; Gottschlich et al., 1994; Helton, Gordon, & Nunnery, 1980; Hilton, 1976; Orr & Stahl, 1977; Richards & Baimsfather, 1988; Topf & Davis, 1993). Consequences of sleep deprivation may include impaired lymphocyte and granulocyte function (Palmblad, Petrini, Wasserman, & Akerstedt, 1979), reduction of natural killer cell and lymphokine-activated killer cell number and activity (Irwin et al" 1994; Irwin et al" 1996), disrupted thermoregulation (Landis, Savage, Lentz, & Brengelmann, 1997), reduced inspiratory muscle endurance (Chen & Tang, 1989), decreased ventilatory ability and loss of pulmonary reserve in patients with obstructive lung disease (Phillips, Cooper, & Burke, 1987), decreased ventilatory response to hypercapnia (Cooper & Phillips, 1982; Schiffman, Trontell, Mazar, & Edelman, 1983; White, Douglas, Pickett, Zwillich, & Weil, 1983), and sensory alteration (delirium, ICU psychosis) (Cooper & Phillips, 1982; Helton et al., 1980; Kloosterman, 1983; Schiffman et al" 1983; 2 White et al., 1983). Despite adequate documentation of the problem and urgent pleas for further research, reports of the impact of sleep-promoting interventions are scarce (Richards, 1995); in fact, none of the studies describes sleep in mechanically ventilated subjects. This research was designed to measure the impact of soft-foam earplugs to reduce noise during the nighttime hours on the quality and quantity of sleep in critically ill subjects. The research questions include the following: 1. Will the sleep maintenance efficiency index (SMEI) and rapid eye movement (REM) sleep increase with earplug use in critically ill subjects? 2. What environment, care content, and care process variables are potential sleep disrupters? 3. How much sleep do critically ill subjects obtain during the day and evening hours? The work was designed to lay the groundwork for future investigations of the impact of sleep-promoting interventions on sleep outcomes and other clinically important outcomes in critically ill patients. Definitions of Sleep The behavioral definition of human sleep is "a reversible behavioral state of perceptual disengagement from and unresponsiveness to the environment usually (but not necessarily) accompanied by postural recumbency, quiescence, closed eyes, and all the other indicators one commonly associates with sleeping" (Carskadon & Dement, 1994, p. 16). Behavioral indicators, including eyelid position (open or closed), body movements, respiratory pattern, and response to arousal stimuli, have been used in sleep research to assess the presence of human sleep (Carskadon & Dement, 1994; Fontaine, 1989). 3 The scientific definition of sleep in mammals and birds comes from polysomnograph data. The polysomnograph simultaneously measures and records electrical potentials from the brain (electroencephalogram), the outer canthus of the eye (electrooculogram), and the mental and submental muscles beneath the chin (electromyogram). The data are commonly divided into two parts: (a) non-REM (NREM or non-REM) and (b) REM. Detailed standards for the configuration and interpretation of polysomnograph sleep data are available (Rechtschaffen & Kales, 1977). Normal adult humans complete approximately five 90-minute (range 70 to 120 minutes) sleep cycles in an 8-hour night (Parker, 1995). Each cycle is divided into non-REM and REM sleep. Table 1 summarizes the definitions and approximate percentage of sleep time spent in each sleep stage over the course of a typical night's sleep period. Non-REM sleep begins with Stage 1, progresses through Stage 4, and then regresses to Stage 2. REM sleep follows non-REM sleep_ As sleep progresses from Stage 1 to Stage 4, the subject is increasingly difficult to arouse. (Stages 3 and 4 are termed slow wave sleep or SWS.) During REM sleep, the subject is more easily aroused. Early in the sleep period (first two to three sleep cycles), SWS periods predominate and REM sleep periods are short. Towards the end of the night SWS periods disappear and REM sleep periods 4 Table 1 Definitions and Approximate Percentage of Night Sleep Time Spent in Sleep Stages for Normal Adult Humansl Sleep stage Definition % of night sleep time Awake Irregular electrooculogram, rapid waves Less than 5 on the electromyogram, alpha rhythms (8 to 12 cycles per second) on the electroencephalogram 1 Slowing electrooculogram, relatively low 2 to 5 voltage, mixed frequency (4 to 6 cycles per second) electroencephalogram without rapid eye movements 2 12 to 14 cycles per second sleep spindles 45 to 55 and K -complexes on a background of relatively low voltage, mixed frequency electroencephalogram activity 3 Moderate amounts of high amplitude, 3 to 8 slow wave activity (1 to 2 cycles per second) 4 Large amounts of high amplitude, slow 10 to 15 wave activity REM Relatively low voltage, mixed frequency 20 to 25 electroencephalogram in conjunction with episodic rapid eye movements on the electroencephalogram, low amplitude electromyogram IA normal human adult is defined as one who "is living on a conventional sleep­wake schedule and who is without sleep complaints" (Carskadon & Dement, 1994, p.21). 5 become longer. During the last two sleep cycles of the night (approximately 05:00 to 07:00), the longest REM sleep periods occur (Carskadon & Dement, 1994; Evans & French, 1995). Organizing Framework for Sleep Research in the Intensive Care Unit An organizing framework was developed for the research after an extensive review of literature summarizing current sleep knowledge and sleep in the ICU (Figure 1). The variables within the framework were included because (a) they are known to or believed to influence sleep or circadian rhythms, or (b) they are outcomes known to or believed to be influenced by sleep quality or quantity. The variables within the framework are grouped as follows: (a) physiologic and psychosocial variables present on admission, (b) fixed and changeable variables within the care setting, (c) intermediate health outcomes during the hospital stay, and (d) health outcomes at hospital discharge. The following discussion presents the rationale and supporting literature for each variable in the organizing framework. Physiolo&ic Variables: A&e, Gender, Circadian Rhythms, Dia&nosis/lllness Severity, and Medications A&e. Age is one of the most important factors affecting sleep patterns. Newborn infants' and young children's sleep architecture is dramatically different from that of adults. During adolescence SWS decreases by approximately 40% (Carskadon & Dement, 1994). SWS continues to decline as adults age (Hayashi & ( PERSON )~ I------ PHYSIOLOGIC VARIABLES: PSYCHOSOCIAL VARIABLES: Age Gender Circadian Rhythms Diagnosis/Illness Severity Medications SETIING: Hospital Size Level of Services I I ICU Physical Characteristics: Size, Physcial Design " SETTING I I Prior Sleep History Psychiatric History ENVIRONMENT: Noise Lighting Ambient Tempemture CARE CONTENT: Intensity of Care Medications Surgical Procedures CARE PROCESS: Pain Sleep Opportunity Sleep Position Anxiety INlERMEDIA 1E HEALTH OUTCOMES: Sleep Deprivation Sensory Alteration Circadian Disruption LenJrth of Time on Ventilator HEALTH OUTCOMES AT HOSPITAL ., DISCHARGE: Survival Length of Stay Fi~ure 1. Organizing Framework for Intensive Care Unit Sleep Research. 6 7 Endo, 1982), although there is controversy over the age of onset and the magnitude of the decline (Bliwise, 1994; Webb, 1982). Age-related decline in the percentage of time spent in REM sleep is also a subject of controversy. Current data suggest that the magnitude of the decline in REM percentage is relatively unimportant (Bliwise, 1994). Zepelin, McDonald, and Zammit (1984) found that a significantly higher noise intensity threshold was necessary to awaken younger subjects compared with older subjects. Subjective sleep quality decreases with age; however, healthy elderly subjects have been shown to be "good" sleepers (Buysse et al., 1991). Healthy elderly subjects also tend to exhibit more frequent and more prolonged awakenings, shorter sleep stage periods, longer sleep latency, and more time in bed (Middlekoop, Smilde-van den Doel, Neven, Kamphuisen, & Springer, 1996; Webb, 1982). Subjects age 61 to 74 years nap less frequently, have shorter sleep latency, higher sleep efficiency, less wakefulness after sleep onset, and more SWS (Hoch et al., 1997). Gender. Gender bias in sleep research and sleep medicine has contributed to the paucity of information regarding differences between the sleep of men and women (Driver, Tobler, Prinz, & Moldofsky, 1998; Young, Hutton, Finn, Badr, & Palta, 1996). Wever (1984) reported a shorter mean polysomnograph sleep­wake cycle in women than men; women slept an average of 1 hour 21 minutes longer than men (N = 27 healthy volunteers studied in a laboratory environment). More recent studies suggest that elderly women may have fewer problems sleeping when compared to elderly men (Rediehs, Reis, & Creason, 1990) and that gender differences in sleep may appear between age 30 and 40 years (Ehlers & Kupfer, 1997). 8 Upper airway resistance during non-REM sleep has been reported to be significantly higher in young adult men (N = 14) compared to women (N = 14) in the same age group (18 to 25 years) (Trinder, Kay, Kleiman, & Dunai, 1997). Healthy, middle-aged subjects studied at home revealed greater frequency of daytime napping, more Stage 1 sleep, significantly less Stage 3 and 4 sleep, a lower sleep efficiency, less REM, and more awakenings from REM sleep in men than in women (Kobayashi et al., 1998). Elderly women have been shown to have higher sleep efficiency and more SWS during recovery from 36 hours of sleep deprivation (Reynolds et al., 1986). Circadian rhythms. Circadian rhythms are internal physiologic cycles occurring approximately every 24 hours (Moore-Ede, Sulzman, & Fuller, 1982). Circadian rhythms serve to organize a living system's physiologic function temporally and adaptation to fluctuating conditions in the environment and are, therefore, important to the health of an organism (Ellmore & Burr, 1993). Wakefulness, sensitivity of the respiratory center to carbon dioxide, temperature, urinary potassium excretion, blood cortisol concentration, and the timing of sleep onset and duration of sleep are just a few examples of the numerous biological processes that exhibit circadian rhythmicity (Felton, 1987; Moore-Ede, Czeisler, & Richardson, 1983; Sebilia, 1981). Electroencephalogram changes, during REM and non-REM sleep, have been closely associated with the melatonin and endogenous sleep consolidation circadian rhythms (Dijk, Shanahan, Duffy, Ronda, & Czeisler, 1997). Hospitalization disrupts circadian rhythmicity (Farr, Keene, Samson, & Michael, 1984), and usual sleep-wake schedules may change during hospitalization to reflect hospital routines (Floyd, 1984). Because the role of chronobiology (the study of rhythms and their temporal relationship in living systems) may be important in the etiology, diagnosis, and management of sleep­wake cycles in the critically ill, circadian rhythms are included in the framework. 9 Dia2nosis/illness severity. Aurell and Elmqvist (1985) suggested that sleep disruption in surgical patients may be more severe than in patients with myocardial infarction. It is likely that as illness severity increases, the frequency of sleep disruptions for medical and nursing care also increases. The influence of illness severity and diagnosis on mortality, particularly in the presence of chronic disease, is well-documented in the critical care literature (Knaus et al., 1991). Because diagnosis and illness severity may influence sleep outcomes, intermediate outcomes, and hospital discharge outcomes, they are included in the framework. Medications. A variety of commonly administered prescription and nonprescription medications affects sleep patterns. Benzodiazepines may suppress SWS, and their withdrawal is accompanied by a "rebound" effect in which SWS is increased (Herkes, Wszolek, Westmoreland, & Klass, 1992). Most antidepressant drug classes, particularly tricyclic antidepressants, suppress REM sleep (Nicholson, Bradley, & Pascoe, 1994). Monoamine oxidase inhibitors and alcohol also 10 suppress REM sleep; withdrawal is followed by increased REM sleep (Carskadon & Dement, 1994; Greenblatt, Harmatz, Zinny, & Shader, 1987). Neuroleptics with dopamine receptor blockade properties, including the phenothiazines, thioxanthenes, butyrophenones, and diobenzodiazepenes, may decrease wakefulness and increase SWS; their effect on REM sleep is variable (Nicholson et al., 1994). One of the primary side effects of stimulants, including xanthenes, caffeine, dopamimetic agents (amphetamines), and anorectics, is increased wakefulness, reduced SWS, and REM sleep disturbances (Nicholson et al., 1994). Sedative effects of anticonvulsants, analgesics, antihistamines (HI and H2 antagonists), and antiemetics have the ability to disturb the sleep-wake cycle (Nicholson et al., 1994). Beta adrenoreceptor antagonists such as propranolol, metoprolol, and pindolol tend to suppress REM sleep. Alpha adrenoreceptor agonists (clonidine and methyldopa) tend to produce drowsiness, prolong sleep times, and exhibit variable effects on REM sleep (Nicholson et al., 1994). The effects of ethyl alcohol on sleep vary with acute and chronic ingestion. In general, REM sleep is decreased, and more frequent awakenings and stage shifts are observed in patients with chronic alcohol ingestion (Nicholson et al., 1994). Psychosocial Variables: Prior Sleep History, Psychiatric History Prior sleep history. Persons deprived of sleep because of an irregular sleep schedule will experience a rebound increase in SWS and REM sleep during recovery from sleep loss. Chronic nocturnal sleep restriction or irregular sleep 11 schedules can disrupt sleep states such as REM episodes occurring at the onset rather than at the end of a sleep cycle (Carskadon & Dement, 1994). The Richards and Bairnsfather (1988) study of night sleep patterns in critically ill male subjects with cardiovascular disease described significantly greater Stage 1 sleep duration in those defined as night sleepers than in day sleepers. Other differences in sleep observed in their study, though not statistically significant, included less time spent awake and more time spent asleep in the night sleepers compared to the day sleepers. Patients with sleep disorders such as narcolepsy, sleep apnea, restless leg syndrome, and chronic medical disorders that result in frequent arousals from sleep will exhibit a variety of abnormalities in sleep measurements (Carskadon & Dement, 1994). Sleep or chronic medical problems may obscure the effects of sleep-promoting interventions. Psychiatric history. Panic disorders, generalized anxiety disorders, obsessive-compulsive disorders, and posttraumatic stress disorders have all been linked to subjective sleep complaints, including insomnia, nightmares, and polysomnographic evidence of difficulty initiating and maintaining sleep (Uhde, 1994). Mood disorders, including bipolar disorders and depressive disorders (Benca, 1994), schizophrenia (Zarcone & Benson, 1994), eating disorders (Benca & Casper, 1994), and psychoactive substance abuse disorders (Gillin, 1994), are associated with a variety of subjective and objective sleep disturbances. A history of psychiatric illness, anxiety, depression, or ineffective coping skills has been 12 cited as a predisposing factor for development of the ICU syndrome (Crippen & Ermakov, 1992; Kloosterman, 1991). Descriptive studies of sleep in the ICU have excluded patients with a history of psychiatric disorders because of the potential for disrupted sleep (Aurell & Elmqvist, 1985; Fontaine, 1989; Helton et al., 1980). Settin2: Hospital Size, Location. Level or Services, leu Pbysical Desi2n The size, level of services provided (secondary and tertiary), location of the institution, and physical characteristics of the ICU, including the number of beds and the floor plan (open units versus private rooms), are fixed characteristics of the environment that may enhance or limit the opportunity for sleep and, therefore, affect sleep outcomes. Physical characteristics may be important contributors to the development of the "ICU syndrome" (Helton et al., 1980; Kloosterman, 1983). The physical setting may impact sleep opportunity, sleep outcomes, and other potentially important outcomes and, therefore, are included in the framework. Environment Noise and 1i2htin2. Shaver and Giblin (1989) cited continuous lighting and noise as the main environmental reasons for sleep disruption. Empirical evidence suggests that ICU noise levels are causally related to reduced sleep efficiency, longer sleep latency, more wake during sleep, suppressed REM sleep, and increased arousals from sleep (Aaron et al., 1996; Topf, 1992; Topf & Davis, 1993). Although the problem of excessive ICU noise (Bentley, Murphy, & Dudley, 1977; Hilton, 1976; Redding, Hargest, & Minsky, 1977) has been amply 13 documented, no studies of the effect of a noise reduction intervention on the sleep of critically ill subjects are currently available in the literature. A pilot study of the effect of a noise reduction intervention (earplugs) on sleep outcomes in 18 female patients was conducted on an acute surgical/gynecological ward using subjective measures of sleep outcomes (Haddock, 1994). Patients who used earplugs reported their sleep to be significantly improved compared with nonearplug users. The suprachiasmatic nuclei of the anterior hypothalamus regulates melatonin secretion by relaying light information to the pineal gland (Harrington, Rusak, & Mistlberger, 1994). White light continuously impinging on the retina inhibits melatonin synthesis by the pineal gland (Home, Donlon, & Arendt, 1991). The precise role of melatonin is unknown but has been linked to the regulation of circadian rhythms in mammals (Harrington et al., 1994) and has been shown to increase serotonin levels and to induce sleep (Wurtman, Axelrod, & Kelly, 1968). The timing of exposure to ordinary room light apparently influences the circadian pacemaker in humans (Czeisler et al., 1989). Preterm infants in an intensive care nursery, who were randomly assigned to a reduction in intensity of light and noise between the hours of 1900 and 0700, spent significantly more time sleeping and gained more weight even though significantly less time was spent feeding (Mann, Haddow, Stokes, Goodley, & Rutter, 1986). Ambient temperature. Mammals lose the ability to thermoregulate during REM sleep (Parmeggiani, 1980). Extreme temperatures in the environment tend to 14 disrupt sleep, especially during REM sleep. Sweating and shivering in response to temperature extremes during NREM sleep are absent in REM sleep (Carskadon & Dement, 1994). Sleep deprived women lost heat more rapidly when exposed to lowered ambient temperatures compared to women who were not sleep deprived (Landis et al., 1997). Stage 4 sleep reportedly has been reduced in sleep-deprived humans when ambient heat exposure was increased (Bach et al., 1994). Care Content Intensity of care. Intensity of care is related to illness severity and mortality outcome in critically ill patients (Cullen, Civetta, & Briggs, 1974; Keene & Cullen, 1983). Available data suggest that illness severity and sleep deprivation are also linked (Dohno et al., 1979); therefore, it stands to reason that more acutely ill patients experience more sleep disruption for medical and nursing care (Richards & Bairnsfather, 1988). Medications. The effects of a variety of prescription and nonprescription drugs on sleep and wakefulness has been discussed (see Physiologic Variables: Medications). Medication profiles in the ICU frequently include an array of medications that may affect sleep patterns, most notably sedatives and narcotics. Intravenous midazolam administered via bolus or continuous infusion has been shown to change the electroencephalogram in critically ill patients (e.g., induction of fast frequency electrical activity on the slower frequency patterns present in SWS) in ways that may affect the electroencephalogram interpretation (Herkes et al., 1992). 15 Sureical procedures. The hypothesis that disruption of the normal sleep-wake regulation is due to neurophysiologic compromise from anesthesia or surgery has been posited as a result of descriptive polysomnograph studies in postoperative patients. Measurement of sleep patterns in critically ill subjects after anesthesia and surgical procedures has shown severely reduced total sleep time (less than 2 hours per night on average) and almost complete suppression of SWS and REM sleep (Aurell & Elmqvist, 1985; Johns, Masterton, & Dudley, 1974; Orr & Stahl, 1977). Care Process: Pain. Sleep Opportunity, Sleep Position, and Anxiety Nurses have a profound influence on the care process because of their sustained patient contact. Webster and Thompson (1986) reviewed sleep literature in nursing practice and observed that "nurses are in a privileged position to ensure that each patient is in an optimum state to receive sufficient sleep" (p. 447). Providing uninterrupted rest periods, adequate pain treatment, being available to listen to the fears of an anxious/wakeful patient, and providing a comfortable sleep position have been proposed as potentially sleep-promoting interventions. Pain management has been specifically identified as a factor in obtaining adequate sleep and is strongly influenced by nurses in the critical care environment (Fontaine, 1989; Puntillo, 1990). Planning for and providing uninterrupted rest periods have been recurring recommendations strongly influenced by nurses (Barrie-Shevlin, 1987; Dracup, 1988; Fisher & Moxham, 1984; McGonigal, 16 1986). Placing patients in their usual sleep position may promote sleep, and lying in unfamiliar sleep positions or restricting movement by invasive monitoring equipment may contribute to discomfort (Webster & Thompson, 1986). Anxiety assessment and treatment are difficult management problems, in general, and in the leu, in particular (Bone et al., 1993). Anxiety is commonly treated with pharmacologic agents; however, nonpharmacologic measures can be added to or substituted for pharmacologic measures. Nonpharmacologic anxiety treatment may include massage, music therapy, relaxation therapy, and the institution of flexible visiting policies (Fontaine, 1993; Webster & Thompson, 1986), Intennediate Health Outcomes Intermediate health outcomes occur during the course of the hospital stay and have the potential to influence the care content and process (hence the feedback loop between intermediate health outcomes and environment, care content, and care process variables), as well as health outcomes at hospital discharge. Sleep deprivation. The unknown effect of sleep-promoting interventions upon sleep deprivation and other health outcomes makes sleep deprivation an important element in the organizing framework. The function of sleep has been examined primarily in mammals through studies of sleep behavior and sleep deprivation (Home, 1985; Rechtshaffen, Bergmann, Everson, Kushida, & Gilliland, 1989; Zepelin, 1989). Despite the volume of information available, the function of sleep is currently unknown. Rechtschaffen (1998), in his recent thoughtful review of current knowledge on the function of sleep, cogently argued that sleep is important enough that it "ultimately enhances survival. " 17 The restitution and energy conservation theories of sleep represent two main theoretical categories in the nursing literature. Restitution theorists have proposed that sleep is for body or brain tissue restoration (Adam & Oswald, 1984), with nurses tending to favor the body restitution hypothesis (Closs, 1988; Hemenway, 1980; Shaver & Giblin, 1989; Webster & Thompson, 1986). Energy conservation theorists have proposed that sleep is to offset the metabolic costs of endothermy (maintenance of constant body temperature) in mammals (Walker & Berger, 1980) or to force inactivity and thus "balance the energy budget" (Zepelin, 1989). Studies in rats have shown that prolonged total sleep deprivation leads to debilitated appearance, weight loss, increased energy expenditure, skin lesions, increased food intake, increased plasma norepinephrine, decreased plasma thyroxine, and (in the late stages) decreased body temperature and death (Rechtshaffen et al., 1989). Human total sleep deprivation studies have occurred for up to 11 days but most for 5 days or less, a time period too short to demonstrate severe physiological derangements (Rechtshaffen et al., 1989). There are no definitive data in humans to suggest that lack of sleep is harmful to any organ other than the brain (Home, 1985). Recent partial or total sleep deprivation and sleep fragmentation studies conducted on healthy humans have documented the effects on hormonal activity 18 (Baumgartner et al., 1993; La1 et al., 1997; Mullington, Hermann, Holsboer, & Pollmacher, 1996; Seifritz et al., 1997); performance (cognitive and motor); mood state (Aeschbach, Cajochen, Landolt, & Borbely, 1996; Hill, Welch, & Godfrey, 1995; Lorenzo, Ramos, Arce, Guevera, & Corsi-Cabrera, 1995; McCarthy & Waters, 1997; Pilcher & Huffcutt, 1996; Richardson et al., 1996); immune system activity (Irwin et al., 1994; Irwin et al., 1996; Uthgenannt, Schoolmann, Pietrowsky, Fehm, & Born, 1995); seizure activity (Fountain, Kim, & Lee, 1998); muscle contractile properties, exercise, and work capacity (Rodgers et al., 1995); body temperature (Landis et al., 1997); airway resistance and ventilation (Meurice, Marc, & Series, 1995; Series, Roy, & Marc, 1994); and electroencephalogram sleep activity (Daurat, Aguirre, Foret, & Benoit, 1997; Dijk, Hayes, & Czeisler, 1993; Roehrs, Merlotti, Petrucelli, Stepanski, & Roth, 1994). A summary listing the purpose, sample charactelistics, independent and dependent variables, and major findings can be found in Appendix A. None of the recent sleep deprivation studies in humans has shown definitive physiologic harm other than well-documented cognitive and motor performance deficits following partial and total sleep deprivation. No studies of the efficacy of sleep as a therapy in critically ill humans such as the effect on mortality, length of stay, and survival are currently available in the literature. Sensory alteration. Sleep deprivation is believed to play a major role in the genesis of "sensory alteration" in critically ill patients. Sleep-deprived well humans exhibit symptoms of anxiety, an inability to sleep, restlessness, irritability, 19 and confusion (Sassin, 1970). The symptoms often subside after one night of recovery sleep. Sensory alteration is a multifaceted problem, with no single cause. Predisposing factors include sleep deprivation, sensory deprivation, sensory overload, anxiety, operative procedures, drugs, immobilization, pain, loneliness and isolation from family meITlbers, a history of psychiatric illness or ineffective coping skills, physiologic derangements, and inadequate information concerning illness and treatment status (Kloosterman, 1983; MacKellaig, 1987; Moore, 1991; Wever, 1984). No single factor or cOITlbination of factors has been shown to produce sensory alteration in critically ill patients. The literature, particularly nurse-authored literature, implicates sleep deprivation, sensory deprivation, and sensory overload as major contributing factors (Blachly & Starr, 1965; Easton & MacKenzie, 1988; Kloosterman, 1983; Kornfeld, Zimberg, & Malm, 1965; Richards & Bairnsfather, 1988; Wilson, 1987). Circadian disruption. Consequences of circadian disruption are illustrated in well humans by the feelings of malaise and fatigue that accompany "jet lag" (Moore-Ede et al., 1982). Circadian disruption has been associated with insomnia, fatigue, poor sleep quality, increased incidence of gastritis and peptic ulcers, psychomotor disruption, anorexia, constipation, nervousness, and decreased mental alertness (Cole & Rogers, 1984). Studies describing or documenting the effects of circadian disruption in critically ill patients are scarce. Len21h of time on ventilator. The adverse effects of sleep deprivation on respiratory physiology have been studied in healthy and ill humans. Healthy male 20 and female subjects demonstrated significantly greater airway collapsibility following sleep fragmentation and total sleep deprivation (Series et aI., 1994). Healthy male subjects showed significant decreases in inspiratory muscle endurance and maximum voluntary ventilation after 30 hours of total sleep deprivation (Chen & Tang, 1989). Decreased ventilatory ability and loss of pulmonary reserve in patients with obstructive lung disease have been observed following total sleep deprivation (Phillips et al., 1987). Other investigations have revealed decreased ventilatory responsiveness to hypercapnia after sleep deprivation compared to baseline values (Cooper & Phillips, 1982; Schiffman et al., 1983; White et al., 1983). No studies have specifically examined the effect of sleep deprivation on length of mechanical ventilation time in critically ill subjects. However, such studies are warranted and could have interesting medical economic implications for the critically ill. Health Outcomes at Hospital Discharee: Survival and Leneth of Stay Survival and length of stay are among health outcomes of interest to intervention researchers. Large samples are typically required to demonstrate the effects of an intervention on survival and length of stay. The Problem Understanding the effects of sleep-promoting interventions upon outcomes in the critically ill presents a challenge because of the number and complexity of interacting variables that make isolating a sleep-promoting independent variable 21 difficult. The ICU environment provides a natural laboratory , albeit a "noisy" one with many uncontrolled variables, for studying sleep-promoting interventions in critically ill humans. Carefully designed intervention research is needed because the effect size of interventions is unknown and, thus, large scale clinical trials are difficult to plan. Measuring sleep presents another hurdle. Visual scoring of polysomnographic sleep tracings is the accepted scientific standard for measuring sleep. Extensive training is required to master the measurement and scoring techniques. Continuous monitoring of equipment by trained personnel during recording sessions is required in order to obtain tracings of suitable quality (Richards, 1987). Thousands of pages of recordings or large computer files are generated during polysomnographic sleep measurement, and the equipment is cumbersome. Medications such as atropine and hyoscyamine sulfate produce electroencephalogram charactelistics of sleep in alert patients (Johns et al., 1974). Benzodiazepines administered by continuous infusion or intermittent bolus have been shown to produce a number of electroencephalogram abnormalities that may affect its interpretation in critically ill ventilator-dependent patients (Herkes et al., 1992). Little is known about the problems of continuous long-term polysomnographic sleep recording in the ICU, but it is easy to conclude that sleep recording is difficult, time consuming, and expensive. The reliability and validity of observation or self-report methods are too questionable to be substituted for polysomnography (Fontaine, 1989; Johns et al., 1974; Richards, 1987). Even if observation techniques were reliable and valid, they still require the continuous presence of an observer, which is also expensive and time consuming. Perhaps the most important barrier to studying the effects of sleep­promoting interventions is the low priority placed on sleep promotion in the ICU environment. Care processes would require significant modification in order to provide sufficient sleep opportunity in the ICU. Si2nificance for Nursin2 22 Critical care nurses have identified sleep research as an important priority (Lewandowski & Kositsky, 1983), yet existing evidence indicates that sleep promotion is a low priority in critical care units (Dracup, 1987). Nursing care practices in the leU would be substantially different if providing ample sleep opportunity was a higher priority. Because nurses are the caregivers with the most sustained patient contact in the ICU, they are in a unique position to participate in sleep research and base their practice on the results of such research. CHAPTERll PRELIMINARY WORK Preliminary work conducted by the investigator in preparation for this study includes (a) development and testing of the earplug intervention; (b) a prospective study of ICU patients to determine clinician ability to predict transfer from the ICU or death within 3 days; (c) interrater reliability studies of anxiety, sensory alteration, and sleep disturbance measurement tools; (d) measurement of sleep in 4 critically ill patients to assure technical adequacy of recordings; and (e) a feasibility study of using earplugs in 5 critically ill patients. The LDS Hospital and University of Utah Institutional Review Board (IRB) approval and informed consent from the subjects were obtained for all preliminary work. Development and Testine the Earplue Intervention A pilot study was designed to develop and test the feasibility of using a noise reduction intervention to improve sleep in critically ill subjects. The study results suggest that sleep in healthy male volunteers is disrupted by exposure to simulated leu noise and that earplugs provide more REM sleep_ The study results provide a reasonable basis for examining the effect of earplugs to reduce noise on the sleep of critically ill patients. The manuscIipt, submitted to the American Journal of Critical Care, is included in Appendix B. Clinician Ability to Predict Intensive Care Unit Stay A convenience sample of patients in the four LDS Hospital leu s was screened by the investigator between 1 August 1997 and 7 August 1997 for the 24 purpose of determining how consistently leu clinicians could predict transfer from the leu or death within 3 days. The researcher was interested in the prediction because one of the exclusion criteria for the proposed research was that a patient is likely to transfer out of the leu or die within 3 days. A "yes," "no," or "not sure" answer to the question ("Is the patient likely to be in leu for 3 more days?") was recorded on a worksheet. Three days later, a "yes" or "no" answer to the question ("Was the patient in the leu for 3 more days?") was recorded on the worksheet. The leu day when the prediction was made, subject's age, predictor (registered nurse [RN], medical doctor [MD], or the investigator), and years of leu experience of the predictor were also recorded. Thirty-eight clinicians made 49/59 correct predictions in 59 patients (83 % correct). The 35 male and 24 female patients were in leu for an average of 7 ± 12 (SD) days (range from 0.04 to 47) when the predictions were made; their average age was 54 ± 20 (SD) years (range from 18 to 88). Ten predictions were made by the investigator, 48 by RN s, and 1 by an MD. The clinicians had an average of 9.5 ± 7.5 (SD), range from 0.5 to 23 years of experience. Two incorrect predictions were made, one by the researcher and one by an RN. Life support therapy was withdrawn in the case of the researcher's incorrect prediction that the patient would remain in leu for 3 more days. In the second 25 case, the RN predicted that the patient would be transferred within 3 days but because beds were not available elsewhere in the hospital the patient remained in ICU. Eight remaining responses, all from RNs, were uncertain. Four were uncertain about predicting death within 3 days (one patient died within 3 days), and four were uncertain that the patient would remain in ICU for 3 more days; one patient was transferred within 3 days. Clinicians were able to predict a 3-day stay reliably in the ICU but were less willing to predict death. Clinician judgment was adequate for the purpose of excluding patients from the present research. Reliability of Measurement Tools Anxiety and Sensory Alteration Measurement Tools Agitation defined as excessive motor activity associated with internal tension (Bone et al., 1993) may be a sign of anxiety amenable to pharmacologic treatment with benzodiazepines, barbiturates, and antipsychotic drugs, as well as nonpharmacologic treatment (Fontaine, 1993). The Motor Activity Assessment Scale (Clemmer, Wallace, & Bailey, 1995) (Table 2) consists of six items, ranging from 0 (unresponsive) to 6 (dangerously agitated. uncooperative). The scale was developed by clinicians (MDs and RNs) in the LDS Hospital Shock-Trauma ICU for the purpose of assessing anxiety and the need for sedation. lnterrater reliability (kappa = 0.83) and validity comparisons with a visual analog sedation scale (Spearman R = 0.9, n < .001) have been reported in mechanically ventilated, surgical ICU patients (Devlin et al., 1998). Table 2 The Motor Activity Assessment Scale Score o 1 2 3 4 5 6 Description Unresponsive Responsive to noxious stimulus Responsive to touch or name Calm and cooperative Restless and cooperative Agitated Dangerously agitated, uncooperative 26 Definition Does not move with noxious stimulus Opens eyes OR raises eyebrows OR turns head toward stimulus OR moves limbs with noxious stimulus (eyes closed) Opens eyes OR raises eyebrows OR turns head towards stimulus OR moves limbs when touched OR name is loudly spoken (eyes mostly closed) No external stimulus required to elicit movement AND patient adjusts sheets or clothes purposefully AND follows commands (opens eyes spontaneously) (eyes may be closed) No external stimulus required to elicit movements AND patient picks at sheets or tubes OR uncovers self AND follows commands (opens eyes spontaneously) (eyes mostly open) No external stimulus required to elicit movement AND attempts to sit up OR move limbs out of bed AND does not consistently follow commands (e.g., will lie down when asked but soon reverts back to attempts to sit up or move limbs out of bed) (wide eyed) No external stimulus required to elicit movement AND patient pulls at tubes or lines OR thrashes side to side OR strikes at staff OR tries to climb out of bed AND does not calm down when asked (wide eyed) 27 The Sensory Alteration Score was used for the purpose of assessing sensory alteration. The Sensory Alteration Score is a 5-point scale, with points given for each episode of the following occurrence during the recording intervals: (a) patient-caused dislodgment of invasive lines or tubes, (b) any restraint use during an 8-hour period, (c) one point for nonpurposeful motor activity reflecting agitation, and (d) two points for nonpurposeful motor activity reflecting dangerous agitation. The Sensory Alteration Score reflects recurring themes in the literature describing the nonpurposeful motor activity frequently observed in disoriented or agitated critically ill patients. Face validity of the Sensory Alteration Score was assessed by four critical care staff nurses and one critical care intensivist, and it was believed to be a reasonable reflection of sensory alteration in the critically ill. Interrater reliability for the Motor Activity Assessment Scale and Sensory Alteration Score for the present study was assessed by two sleep technicians and the investigator in 7 different subjects. A 1-hour videotape of each subject was made and viewed independently by the raters. Each rater scored the Motor Activity Assessment Scale and Sensory Alteration Score at 15-minute intervals, generating four observations per subject. The Motor Activity Assessment Scale data were divided into four sets of seven observations. Finn's r was computed as an index of reliability between all three pairs of raters for each set of observations, yielding 12 reliability coefficients ranging between 0.73 to 0.98. Finn's r is an alternative to intraclass correlation that can be interpreted "as the proportion of the correspondence of the observed ratings that is not due to chance" (Whitehurst, 28 1984, p. 25) and is preferable in cases in which distributions of ratings are skewed. Intraclass correlation coefficients were computed for the Sensory Alteration Score. The four reliability coefficients (one for each set of three raters) were 1.0, 0.76, 0.76, and 0.57. Adequate reliability for both tools was demonstrated. Sleep Disturbance Measurement Uninterrupted time for sleep has been commonly used in a variety of descriptive studies of leu patients' sleep (Helton et al., 1980; McFadden & Giblin, 1971; Richardson, 1986; Walker, 1972; Woods & Falk, 1974). The Sleep Disturbance Record was developed for the present study from Richardson's (1986) categorizations and Walker's (1972) tools. The Sleep Disturbance Record is a paper-and-pencil tool used to record four categories of disturbances external to the subject, originating in the subject's room, and related to subject care (Table 3). Timing of disturbances was measured using a digital stopwatch (Sportline, Campbell, CA) accurate to hundredths of seconds. The Sleep Disturbance Record was used to estimate the frequency and duration of sleep disturbances, time spent in the usual sleep position, and time room lights were on. Interrater reliability of the tool was evaluated by the investigator and two sleep technicians using simultaneous observation and independent scoring of 8 subjects for I-hour time periods. The sleep technicians observed 2 additional subjects for 3-hour time periods. Intraclass correlation coefficients for the total minutes of disturbances were computed for all three observers (N = 8 subjects, 29 Table 3 Disturbance Categories and Definitions Disturbance category Definition 1 (minimal) Staff moving int%ut of the room without touching the subject, equipment, or furniture in the room and without talking 2 (maximal) Staff providing direct, hands-on care to the subject OR talks to the subject OR talks to others in the room in normal or loud voice tones 3 (moderate) Staff moving int%ut of the room, touching equipment but not the subject OR talking quietly with others in the room 4 (other) Family interacting with the patient at the bedside. (If family is sitting quietly at the bedside, do not count.) r = 0.96) and for two observers (N = 10 subjects, r = 0.995). Intraclass correlation coefficients for the time room lights were on and time spent in the sleep position were 1.0 for all three observers (N = 8) and 1.0 for two observers (N = 10). The Sleep Disturbance Record demonstrated sufficient interrater reliability. Technical Adeguacy of Sleep Monitorinl Two technicians and the researcher began preparation for monitoring sleep in ICU subjects in November 1996. Each spent approximately 16 hours in the LDS Hospital Sleep Disorders Center observing and helping with the instrumentation of patients scheduled for polysomnograph and multiple sleep- 30 latency tests. Forty hours were spent practicing head measurement and polysomnograph instrumentation on a glass-head model and on each other. A head measurement procedure was developed according to the Ten-Twenty System of the International Federation (Carskadon & Dement, 1994; Cavallaro, 1992; Chusid, 1979; Jasper, 1958). Placement of head electrodes and preliminary recording of polysomnograph signals were checked by experienced EEG and sleep disorders technicians and were judged to be technically adequate for scoring. Polysomnograph recordings were completed on 4 critically ill subjects (2 male and 2 female, age 58 to 80) overnight for 6 nights. Signals were judged to be of acceptable quality by the director of the LDS Hospital Sleep Disorders Center (James M. Walker, PhD) in the first patient. Polysomnograph signals were completely unacceptable in the second patient because of a 6O-cycle electrical noise that could not be resolved, even with the help of experienced sleep disorders technicians. Sensor Medics (Yorba Linda, CA) was contacted and provided a new computer and data acquisition software with 60-cycle filtering capabilities. Subsequent recordings for 5 nights in 3 patients were technically acceptable. Appendix C illustrates the configuration of sleep-monitoring equipment. Problems Encountered in Preliminary Work Use of Eye Covers During preliminary polysomnograph recording, 2 subjects tried wearing eye covers (Sleep-Eze by Austin House, Oakville, Ontario, Canada) to eliminate room light. Both subjects removed the eye covers after 1 hour. The decision to dim the 31 room lights rather than use eye covers to eliminate light was made since dimming room lights at night constitutes usual nursing care. Use of Earplugs One of 2 subjects who participated in preliminary polysomnograph recordings and who were asked to wear earplugs all night requested that the earplugs be removed after 1 hour; the other patient wore the earplugs all night. Since earplugs was the planned intervention, a study to test the feasibility of wearing earplugs was conducted on 5 critically ill subjects. Six subjects or their next of kin were approached for consent, with 5 providing consent. One subject's next of kin declined to participate because of concern that the subject was already withdrawn and depressed and that the earplugs might eliminate an important source of meaningful sensory input. Earplugs were inserted by the investigator at 22:30 and removed between 06:00 and 08:00. An axillary temperature probe was placed as a substitute for tympanic temperature monitoring. An investigator-developed questionnaire was used to gather subjective data regarding earplug comfort and ease of use from the nurses and subjects. An otoscopic ear examination before insertion and after removal of the earplugs was also completed. A plan to use soft cloth ear covers to blind the caregivers to the presence of earplugs was abandoned during the preliminary study because nurses needed to know if communication with the subject was potentially compromised. When questioned the morning after earplug use, 4 subjects rated the earplugs comfortable, and 3 subjects liked the earplugs and wanted to keep them 32 for subsequent use. Subject 5 said the earplugs were comfortable but were not helpful for sleep and preferred not to wear them again. The nurses' comments were all favorable except for one who removed one earplug to compare axillary temperature and tympanic temperature. Table 4 summarizes the questionnaire results. The earplugs was a viable intervention and could be worn by critically ill subjects. Infonned Consent A final problem encountered in the preliminary work was a consistent pattern of subjects' next of kin declining to participate in poly somnograph y . Of 11 Table 4 Earplu~ Feasibi1it~ Sleen Questionnaire Resylts Date Subject Hours worn Reason not Comfort Ease of use Prior worn 8 hours (RN rating) (RN earplug use rating) by subject 1115197 1 8.00 1 Yes 1115197 2 4.75 Lab drawna NA 1 Yes 1115197 3 9.75 2 2 No 1116/97 4 10.00 1 1 No 1122/97 5 7.50 Temperature 2 1 No checkb Note. NA = not answered, 1 = very comfortable or very eas~ to use, and 2 = somewhat comfortable or somewhat eas~ to use. ARN removed the earplugs to talk to the subject; subject requested they be left out. bRN wanted to check tympanic temperature (see text). subjects approached, 5 subjects' next of kin declined to provide consent for sleep monitoring and continuous rectal temperature monitoring. The reason given for declining in all 5 cases was because they did not want anything else "hooked up" and were reluctant to allow insertion of a rectal temperature probe or changing a standard indwelling urinary catheter to a thermistor-tipped urinary catheter. 33 The remaining 6 subjects were able to give their own consent and agreed to participate in sleep monitoring; they were not asked to participate in rectal temperature monitoring. One subject was not studied because of scheduled surgery. Subject 4 declined to continue after a first night of sleep monitoring, citing the equipment as cumbersome and "too stressful." Subject 5 asked, "How much longer will this take?," after 6 hours of monitoring on the first night but agreed to continue with a second night of monitoring. The other 3 subjects did not express discomfort with the sleep-monitoring procedure. Because of the difficulty obtaining consent and 2 subjects' indications that the equipment was bothersome, the design of the main study was changed from 72 consecutive hours of proposed sleep monitoring to 32 hours (8 hours the first night, no monitoring the second night, and 24 consecutive hours beginning the third night). Scalp electrodes were changed from nondisposable cup electrodes with lead wires permanently attached to disposable electrodes with detachable wires (Physiometrix, Billerica, MD). Axillary temperature monitoring was substituted for rectal temperature monitoring. The research design and methods are described in Chapter III. CHAPTERll RESEARCH DESIGN AND METHODS Purpose The purpose of this research was to measure the effect of soft-foam earplugs to reduce noise during the nighttime hours on the quality and quantity of sleep in critically ill subjects. The research questions include the following: 1. Will the sleep maintenance efficiency index (SMEI) and rapid eye movement (REM) sleep increase with earplug use in critically ill subjects? 2. What environment, care content, and care process variables are potential sleep disrupters? 3. How much sleep do critically ill subjects obtain during the day and evening hours? The organizing framework for the research was presented in Chapter 1. Each variable within the organizing framework was included in the study design. A minimum sample size of 6 subjects was selected because the research was a pilot study designed to provide effect size data necessary for a larger study of noise reduction in critically ill subjects. Setting The research was conducted in a 520-bed, tertiary care hospital providing Level I trauma services and comprehensive heart services (heart transplant and cardiac surgery) in the Western United States. The hospital is affiliated with the University of Utah teaching program for medicine, nursing, and medical informatics. Sixty leu beds in four leu s are all configured as large, private rooms arranged in a U shape around two central nursing stations. Sample Size and Patient Availability Results of a search of the hospital's computerized patient information 35 system between 1 August 1995 and 29 February 1996 yielded 3,099 patients admitted to the leUs. Five hundred eighty patients' length of stay was > 3 days, providing an adequate pool of subjects to carry out the research. Purposive sampling was used to select patients who were mechanically ventilated and likely to stay in the leu long enough to complete the study. Patient Selection and Enrollment All patients with a ~ 3 leu stay were screened each weekday morning to identify potential candidates. Enrollment criteria are listed in Table 5. If no evidence of hearing problems was documented in the medical record, the Hearing Handicap Scale (High, Fairbanks, & Glorig, 1964) was administered to the patient or a close relative. Subjects with a Hearing Handicap Scale score of ~ 25 % (more than a 25 % hearing loss) were excluded. Table 5 Enrollment Criteria Inclusion In ICU ~ 3 days, likely to remain in ICU for ~ 3 additional days Mechanically ventilated Age ~ 18 Exclusion History of current alcohol or drug abuse that carries the likelihood of acute withdrawal symptoms, sleep medication abuse, current psychiatric illness or anxiety disorder (panic attacks, phobias), and brain disorder Inability to pass hearing screening test or documented history of hearing problems History of sleep disorder or failed sleep disorders screening Other Rationale Subjects with ~ 3 day leu stays are more likely to need sleep-promoting intervention; study length is 3 days. 36 A paucity of information about sleep in mechanically ventilated patients is currently available. Sleep architecture is more likely to be stable and similar to adult sleep. Rationale Eliminate patients with sleep pattern variability due to medical, psychiatric, or anxiety disorder. ICU noise may have less effect on subjects with reduced auditory acuity. Eliminate patients with sleep disorders. Other reason not listed. Note. Detailed definitions were used and are available from the investigator. 37 After scoring the Hearing Handicap Scale, a standard questionnaire for the detection of sleep disorders from the LDS Hospital Sleep Disorders Center was administered in an interview format. Sleep disorders were considered to be present if the questionnaire results indicated that the subject's sleep habits prior to hospitalization could be due to an undiagnosed sleep disorder, including sleep apnea, narcolepsy, chronic insomnia, or restless leg syndrome. Specific responses that disqualified the subject were (a) night shift work and (b) a response of "frequently" or "almost always" to any of the following general questions pertaining to sleep habits: any problems with sleep, difficulty going to sleep, difficulty staying asleep, crawling or aching sensations in the legs or an inability to keep legs still when trying to fall asleep, loud snoring, pauses c:: 10 seconds in breathing pattern during sleep, awakening in the morning with a headache, and problems with daytime sleepiness. After sleep disorders screening was completed, technicians trained to monitor sleep and to use other measurement tools were scheduled to record sleep patterns and other study data. Available equipment and personnel precluded simultaneous enrollment of more than one subject. If more than one candidate was available, subjects were approached for consent in descending order of ICU length of stay. Subjects without exclusions and who provided informed consent were enrolled. A record of all subjects screened, all that met enrollment criteria, reasons for exclusions, subjects enrolled, and subjects who ultimately completed the study was maintained. 38 Experimental Desien A randomized, unblinded, within-subjects (crossover) study design (Figure 2) included all variables within the organizing framework (Figure 1). Physiolog~c variables (age, gender, circadian rhythms, diagnosis/illness severity, and medications) were controlled for by the within-subjects design. Patients with disturbed prior sleep patterns or evidence of a current psychiatric illness (psychosocial variables) were excluded. The study setting has been described and remained unchanged during the study. Environmental variables (noise, lighting, and temperature); care content variables (intensity of care, medications, and surgical procedures); care process variables (pain, opportunity for sleep, sleep position, and anxiety); and intermediate and hospital discharge outcome variables (sleep deprivation, sensory alteration, circadian disruption, length of time on ventilator, survival, and length of stay) were all measured and are described under Adequacy of Measures. Usual nursing care, defined as adequate pain treatment and dimming of room lights between the hours of 2230 and 0630, was measured. Standard sleep variables obtained from polysomnographic sleep recording are defined in Appendix D. Study Procedures After enrollment, the technicians collected demographic and illness severity data. The researcher placed scalp electrodes 6 to 10 hours prior to the study, according to the International 10-20 System for electrode placement, using 01 to 39 Screening and Enrollment ) + (n.=7) : Randomization (N=13) ~ (n.=6) ~, " Usual nursing care, Night 1: 2230 to 0630 Usual nursing care no earplugs plus earplugs " Night 2: "Washout" " Usual nursing care Night 3: 2230 to 0630 Usual nursing care, plus earplugs no earplugs .J Mean, Sll, effect size, power L.. -"'"I analysis for SMEI and REM I .... + Describe daytime (0630 to 1530) and evening (1530 to 2230) sleep patterns (n=7) Figure 2. Experimental Design. Note. Measure environment, care content, and care process variables on nights 1 and 3. 40 A2, 02 to AI, C3 to A2, and C4 to Al to record electroencephalogram (Rechtschaffen & Kales, 1977). A second experienced sleep technician checked scalp electrode placement. Electrode impedance was documented after initial scalp electrode placement and at the beginning of each study night. Initially, electrodes and lead wires were placed on the scalp and face at least 8 hours prior to the study in order to allow adaptation to the electrodes (Aaron et al., 1996). In 1 subject, problems with itching of the scalp and face 7 hours after electrode placement led to discontinuing the study. The practice of early-face electrode and lead-wire placement to allow for adaptation was subsequently abandoned. The sleep technicians placed the face electrodes and prepared all data collection instruments according to a detailed checklist 1 to 2 hours prior to the study. Sleep-monitoring equipment was located outside of the subject's room. All sleep recordings were made with SensorMedics Ampstar 4100 hardware and Somnostar Alpha data acquisition to score and report software (Appendix C). Signal calibration for all channels was 50#,v/cm. Time constants were 0.3, with a sensitivity at 5 for electrooculogram/electroencephalogram and .003 with a sensitivity at 2 for electromyogram. The low frequency filter was set at .5 Hz for electroencephalogram/electrooculogram and 10 Hz for electromyogram. The high frequency filter was set at 35 Hz for electroencephalogram, 15 Hz for electrooculogram, and 70 Hz for electromyogram. The technicians began collecting polysomnographic data at 2230 and stopped at 0630 hours for all subjects. The technicians started the daytime recordings between 0630 and 0700 hours, and the afternoon recordings between 1500 and 1530 hours. The study technicians used the (a) Sleep Disturbance Record to measure sleep opportunity, lighting, and sleep position; (b) Sensory Alteration Score and Motor Activity Assessment Scale to measure sensory alteration and anxiety; (c) Verbal Pain Assessment Scale to measure the subject's perception of pain before and after the nighttime recording periods; (d) Larson­Davis Model 720 sound-level meter to record sound levels; and (e) two Yellow Springs Instruments 700 Series thermistors, one for room and one for axillary temperature measurement. The bedside personnel could not be effectively blinded to the presence of the earplugs because of the need to communicate with the subject. However, it was more important that the person scoring the sleep data was unaware of the treatment assignment since the sleep data collection procedure is more objective than the scoring procedure. Analgesic medication was titrated by the subject's nurse. On the treatment night, the investigator completed an otoscopic ear examination, finding the earplugs in the subject's ears just before sleep recording began at 10:30 p.m. Following study completion (usually the next morning), a checklist tool was used to record the Therapeutic Intervention Severity Score (Keene & Cullen, 1983) as a measure of intensity of care. 41 Adeguacy of Measures The Hearing Handicap Scale 42 The Hearing Handicap Scale is a 40-item instrument with a 5-point ordinal scale to grade the response to each item and was developed for the purpose of quantifying hearing loss (High et al., 1964). The Hearing Handicap Scale is recommended as correlating (r = 0.7) well with measures of hearing sensitivity (Etienne, 1996). Internal consistency reliability estimated by the split-halves method in 50 cases was .96 (Pearson coefficient), and the total form internal consistency reliability was .98 (Spearman-Brown formula) (High et al., 1964). Each 20-item half is suitable for independent use because of the high internal consistency reliability. The Hearing Handicap Scale was selected for three reasons: (a) Ambient noise levels in the ICUs were too high to administer a simple hearing screening test using a 20 dB(A) tone at 500 to 4000 Hz to detect hearing loss (American Speech, Language, and Hearing Association, 1985); (b) the Hearing Handicap Scale can be administered in an interview format and answered by someone other than the subject if the subject is unable to provide the information; and (c) the results of the Hearing Handicap Scale are readily interpretable. The LDS Hospital Sleep Disorder Questionnaire The LDS Hospital Sleep Disorder Questionnaire was developed from existing tools and according to recommended standards for the detection of sleep disorders (Roffwarg & Erman, 1985). Reliability and validity data are not 43 available for the questionnaire; however it has been routinely used in a clinical sleep disorders center for more than 10 years and is recommended as an adequate screening tool (1. M. Walker, personal communication, March 2, 1996). Illness Severity: APACHE II The APACHE II scoring system (Knaus, Draper, Wagner, & Zimmerman, 1985) is an extensively validated point score (range 0 to 71) and was developed for the purpose of describing illness severity, predicting survival outcome, and prognostically stratifying critically ill patients to assist investigators in comparing the success of therapies, use of hospital resources, and efficacy of intensive care in different settings or over time. In this study, APACHE II was used as a descriptor of illness severity for each study night. Reliability of APACHE II has also been reported, with reliability coefficients and percentage agreement between raters consistently exceeding .90 and 86%, respectively (Knaus et al., 1991). Noise and Lighting Noise levels were continuously monitored with a Model 720 sound-level meter (Larson-Davis, Provo, UT) and associated statistical software. The Model 720 is a Type 2 precision integrating sound-level meter and statistical logger that exceeds all worldwide accuracy requirements for noise measurement (Larson­Davis, 1991). The sound-level meter measured (a) an integrated average sound pressure level in dB(A) for the nighttime periods; (b) instantaneous peak sound pressure level at I-minute intervals during the recording periods; and (c) sound- 44 level events defined as the number of events exceeding 80 dB(A) for more than 2 seconds. Sound-level descriptions were planned to include (a) comparing the integrated average sound pressure level and defining a difference as > 3 dB(A) (L. Alvord, personal communication, April 18, 1996), (b) counting and comparing the number of instantaneous peak sound pressure levels ~ 80 dB(A), and (c) counting and comparing the number of sound-level events between the 2 nights. The 80 dB(A) threshold was selected for several reasons. Studies in healthy humans have shown that an 80 dB (A) (above the threshold of audibility) tone was required to produce a change in electroencephalogram sleep stage to a lighter sleep stage during Stage 4 sleep (Kryter, 1994). In the same study, a tone of 30 dB(A) was required to arouse a subject from Stage 2 sleep. In a separate study, peak sound levels > 80 dB(A) were found to be positively correlated and significantly related to sleep arousal (Aaron et al., 1996). An 80 dB(A), 2-second sound event threshold is commonly used in environmental noise-level measurement (G. Duersch, personal communication, February 20, 1996). The Sleep Disturbance Record included the total time (minutes) room lights were on. See Preliminary Work for reliability testing of the Sleep Disturbance Record. Ambient Temperature A Yellow Springs Instruments 700 Series thermistor was located at the head of the subject's bed and used to measure room temperature at hourly intervals during the recording periods in the first 11 subjects. In Subjects 12 and 13, the thermistor was centered inside a perforated, 11/2 in plastic ball and placed on the bed underneath the subject's gown at about waist level. The technique was recommended as a more valid ambient temperature measurement method because ambient room temperature does not reflect ambient temperature at the surface of covered skin (J. M. Walker, personal communication, November 12, 1997). Intensity of Care 45 The Therapeutic Intervention Severity Score (Cullen et al., 1974; Keene & Cullen, 1983) was selected to measure the intensity of care because it has been extensively used in critically ill patients. The tool is a checklist in which points for each therapeutic intervention are assigned and categories (Classes I, II, III, and IV) derived (Cullen et al., 1974). The Therapeutic Intervention Severity Score has been widely used to establish nurse-patient ratios in the ICU, assess hospital bed utilization, determine illness severity, and establish future needs for critical care beds (Keene & Cullen, 1983). Cullen and colleagues (1974) established validity of the tool. The researcher used the therapeutic intervention checklist tool to compute scores for a 24-hour period on the study day. Medications and Surgical Procedures ICU patients typically receive a large variety of medications. The influence of medications commonly administered in the ICU and their potential influence on measures of polysomnography are not well-described; in most cases, the effects are unknown. Methods for evaluating and comparing medication profiles in leu patients are not developed. All medications administered are available in the computerized hospital information system (Gardner & Shabot, 1990) and provide the medication data for each subject. The total dose per day (converted to dose equivalents in the case of narcotics) provided a simple method to describe and quantify the medication profile in ICU patients. Subjects were excluded if surgery was scheduled or anticipated during the study. The anesthesia and operative notes were used to describe surgical procedures. Pain The subjective nature of pain makes objective assessment very difficult. 46 The Verbal Pain Assessment Scale (pain scale) was used to measure the subject's perception of pain. The pain scale is recommended as an adequate assessment tool by the Agency for Health Care Policy and Research in a variety of adult pain assessment and management guidelines (Panel, 1994). Internal consistency reliability or interrater reliability cannot be assessed because the pain scale is a single-item scale used by the patient to rate a subjective phenomenon. Assessing the stability of the scale is not entirely appropriate because of the rapidly changing nature of pain. The study technicians used the pain scale to assess pain before and after each study night. Because most of the subjects were mechanically ventilated and could not speak, they were first asked if any pain was present. If not (indicated by shaking their head or mouthing the word "no"), a score of zero was recorded. 47 Subjects were asked to indicate the score by holding up their fingers; alternatively, the technicians verbally counted down, beginning with 10, until the subject nodded to indicate the score. The dose of analgesic medication administered (dose units per shift-narcotic analgesics converted to morphine equivalents) was used to measure pain management. Analgesic medication data were obtained from the hospital information system. Sleep Opportunity Uninterrupted time for sleep has been commonly used in a variety of descriptive studies of leu patients' sleep (Helton et al., 1980; McFadden & Giblin, 1971; Woods, 1972). The researcher developed the Sleep Disturbance Record from Richardson's (1986) categorizations and Walker's (1972) tool. The technicians timed the onset and duration of disturbances with a Sportline Model 220 stopwatch (laser tuned for l/l00-second accuracy) and then categorized and recorded all disturbances on the Sleep Disturbance Record. The opportunity for sleep was defined as the total undisturbed sleep time in minutes. Interrater reliability of the tool was described in Preliminary Work. Sleep Position The subject or a family member was asked to describe the subject's usual sleep position at the time of study enrollment. The sleep technicians recorded the amount of time during the nocturnal sleep periods that the subject spent in the supine, side-lying, and prone position using the Sleep Disturbance Record. 48 Interrater reliability for the tool was described in Preliminary Work. Anxiety Reliable and valid measurement tools to assess anxiety in critically ill patients have not been developed; in fact, development of these measures would be a prodigious undertaking (Rothenberg, 1993). Agitation, defined as excessive motor activity associated with internal tension (Bone et al., 1993), may be a sign of anxiety amenable to pharmacologic treatment with benzodiazepines, barbiturates, and antipsychotic drugs, as well as nonpharmacologic treatment (Fontaine, 1993). The Motor Activity Assessment Scale was used to assess anxiety. Interrater reliability for the Motor Activity Assessment Scale was described in Preliminary Work. The total dose per day of benzodiazepines, barbiturates, and antipsychotic drugs served as a descriptor of the pharmacologic treatment of anxiety. The data were obtained from the computerized hospital information system. Intermediate Health Outcomes Sleep Deprivation Polysomnographic data were used to assess the quantity and quality of sleep. The SMEI reflects the continuity or quality of sleep, that is, the proportion of time spent in bed sleeping after initial sleep onset (1. M. Walker, personal communication, March 2, 1996). The lower the SMEI, the poorer the quality of sleep. In the present study, sleep deprivation was defined as SMEI < 0.82. The SMEI threshold was computed as follows (Table 6): (a) Mean SMEI for Table 6 Computation of the Sleep Deprivation Threshold Age group Males Females Males X + SD. X + SD X - (3*SD) 30 to 39 0.98 ± 0.02 0.98 + 0.06 0.92 40 to 49 0.93 ± 0.06 0.98 ± 0.02 0.75 50 to 59 0.95 0.04 0.94 ± 0.07 0.83 Average 0.83 Females X - (3*SD) 0.80 0.92 0.73 0.82 N = 10 normal subjects in each of three age and gender groups was computed from the laboratory data of Williams, Karacan, and Hurst (1974) using the 49 standard formula for SMEI (total sleep time/time in bed minus time required to fall asleep); (b) three SDs were subtracted from the normal mean SMEI value for each group; (c) the average of the resulting three values was computed for each category; and (d) the lowest of the two values was selected. Other standard sleep outcome measures (Appendix D) were visually scored from the pol ysomnographic data by a single experienced rater. This rater was certified by the American Sleep Disorders Association and used standard scoring criteria (Force, 1992; Rechtschaffen & Kales, 1977). Sensory Alteration The Sensory Alteration Score was used to assess sensory alteration. One point was given for each episode of the following: (a) patient-caused dislodgment 50 of invasive lines or tubes, (b) any restraint use during an 8-hour period (1 point), (c) one point for nonpurposeful motor activity reflecting agitation, and (d) 2 points for nonpurposeful motor activity reflecting dangerous agitation. Reliability and validity of the Sensory Alteration Score was described in Preliminary Work. Circadian Disruption Measurement of circadian rhythms requires frequent data sampling over time. The most accessible measure of circadian rhythm available in this study was the circadian temperature cycle. A plot of digitally sampled axillary temperature measurements was used to measure circadian disruption. A square-shaped waveform varying between 36°C and 37.5°C with a rapid decrease around midnight and a rapid increase around 0600 is representative of a normal circadian temperature cycle (Moore-Ede et al., 1983). Time on Mechanical Ventilation Ventilator data were obtained from the LDS Hospital computerized health information system. The elapsed time between the first ventilator record and the last ventilator record was computed to obtain time on mechanical ventilation. Outcomes at Hospital Dischara:e: Survival and Len21h of Stay The survival outcome data were obtained at the time of hospital discharge. ICU length of stay was defined as the total number of days elapsed between ICU admission and discharge; transfers from one ICU to another were included in the 51 equation. Hospital length of stay was defined as the total number of days elapsed between hospital admission and discharge. Data Analysis All data were entered into Microsoft Excel, Version 5.0 (Microsoft Corporation, 1994) spreadsheets and checked for accuracy once by the technicians and a second time by the investigator. Data analyses were conducted using SPSS for Macintosh, Version 6.1.1 (SPSS, Inc., 1995) and nQuery Advisor (Elashoff, 1997). The data analysis is presented for each research question: 1. Will the SMEI and REM sleep increase with earplug use in critically ill subjects? The means and standard deviations were computed under each independent variable (usual nursing care and usual nursing care + earplugs) condition for SMEI and percentage of REM sleep; other standard sleep measures were also described. A one-tailed, paired 1 test with the associated power, effect size, and sample size (given a power of .80) were computed (Cohen, 1988; Elashoff, 1997). 2. What environment, care content, and care process variables are potential sleep disrupters? Multivariate data analysis was not possible because of the small sample size; therefore, a descriptive approach was taken. 3. How much sleep do critically ill subjects obtain during the day and evening hours? A table of means and standard deviations for the SMEI, percentage of REM sleep, and other standard sleep measures was constructed for the night, day, afternoon, and 24-hour time periods. CHAPTER IV RESULTS Screenin& and Enrollment During the enrollment period (6 February 1997 to 12 June 1997, 22 September 1997 to 12 January 1998), 573 subjects were screened and 540 were excluded. The first subject was ventilator dependent for 35 days prior to enrollment. No other subjects who had been in ICU for more than 21 days at the time of enrollment were studied in order to minimize variability in ICU length of stay at enrollment. One subject, who was in heart failure while awaiting cardiac transplantation but breathing spontaneously, was enrolled. Two subjects were mechanically ventilated on the first but not on the second study night. Thirty-three eligible subjects were approached for consent, 16 declined to participate, 17 were enrolled, and 13 completed the study. Reasons for exclusion and reasons for declining are shown in Figure 3. Table 7 lists "other" reasons for exclusion. Reasons for dropout after enrollment included (a) subject anxiety during scalp electrode placement, (b) scalp itching after scalp electrode placement, (c) subject removing face leads 1 hour after placement, and (d) severe shock with subsequent brain dysfunction after enrollment. Informants for prior drug use, psychiatric history, and sleep disorders screening were (a) the subject (n = 3); (b) the subject's spouse (n = 5); and Reasons for Exclusion (n=540) Other (n=56) 10% Hearing problem (n=18) 3% No ventilator 4~ (n=24) ~ Sleep 5% disorder (n=25) Brain disorder (!!=108) 1 =Do not want earplugs l=No trouble sleeping 1=End-of-life decision Reasons for Refusal (n=10) Transfer/die (n=309) Do not want sleep equipment (n=13) Figure 3. Reasons for Exclusion and Refusal. 53 Table 7 "Other" Reasons for Exclusion Reason Family unavailable for consent Subject in severe shock Other subject already enrolled Surgery scheduled or anticipated Psychiatric history Study personnel unavailable Unable to apply scalp electrodes due to injury Continuous intravenous drip sedation Drug overdose or alcohol withdrawal Severe otitis media precluding earplug use Reason not documented Patient transfer after long (~ 21 days) hospital stay Family unable to answer Hearing Handicap Scale Total 8 8 7 7 7 4 3 4 3 1 2 1 1 56 54 (c) the subject's mother, daughter, sister, son, and father (n = 1, respectively). All subjects passed sleep disorders screening and obtained at least 6.5 hours of sleep at night prior to hospital admission (X = 8.6, SD = 1.1, range = 6.5 to 10 hours). Three reported restless sleep on occasion due to chronic illness. All reported adherence to a regular bedtime schedule and no trouble or occasional trouble falling asleep or staying asleep. Seven subjects denied sleep medication use, 5 reported occasional sleep medication use, and 1 reported frequent sleep medication use but did not know the sleep medication used. None reported a psychiatric illness history and all passed the Hearing Handicap Scale (percentage hearing handicap, X = 2.6 ± 5.6, range = 0% to 14%). Demoeraphic Data 55 Eight females and 5 males participated in the study. Diagnoses included trauma (n = 2); repair of ruptured abdominal aortic aneurysm (n = 2); coronary artery bypass (n = 2); and pneumonia, bowel ischemia, cardiomyopathy, abdominal hernia repair, mitral valve replacement, coronary artery disease, and gastrointestinal bleeding (n = 1, respectively). Chronic illness and hospitalization within the past 6 months were documented in the medical history of 3 subjects (systemic lupus erythematosus, Crohn's disease, and idiopathic cardiomyopathy with congestive heart failure). Medications taken prior to admission were all discontinued for at least 3 days prior to enrollment. Six subjects were smokers, 6 reported alcohol intake once per month or less, and 1 reported alcohol intake once or twice per week, usually on weekends. One subject reported occasional marijuana use; all others denied any recreational drug use. Demographics are summarized in Table 8. Inteerity and Safety of the Earplue Intervention Sound levels on the night earplugs were worn were theoretically reduced by 25 dB(A), the noise reduction rating of the earplugs. It is well-known, however, that the noise reduction rating from a laboratory environment using standard 56 Table 8 Demographics and Illness Severity Days at enrollment Age (year) ICU Ventilator Surgery? APACHE II score Mean 56.9 12.7 11.9 Yes = 10 13.9 SD 20.0 8.2 8.9 No = 3 5.4 Minimum 18.0 3.0 3.0 5.0 Maximum 88.0 35.0 35.0 23.0 methods overestimates the actual noise attenuation in the workplace by 30% to 60% (Carter & Upfold, 1993). Improper fit or placement of the earplugs is a reason commonly cited for the discrepancy between the noise reduction rating and actual noise attenuation in the field (Carter & Upfold, 1993). The researcher placed the earplugs in every subject on the night earplugs were worn in order to assure consistent and proper placement within the ear canal and to maximize noise attenuation. The critical aspect of the placement procedure was to roll the earplug between the thumb and index fingers and to position the tightly compressed earplug parallel to the ear canal, allowing little resistance to insertion. The earplug was then held in place for approximately 15 seconds while it expanded and filled the ear canal. Subjects who were able responded to questioning (verbally or by nodding their head) about the adequacy of the seal; all affirmed that the earplugs felt sealed and reduced the sound levels. Subject 9 removed the earplugs 3 hours after the study began. The study technician said that the subject complained that 57 the earplugs were uncomfortable. The subject replaced the earplugs after approximately 1 hour, leaving the integrity of the earplug intervention questionable for Subject 9. The next morning, the subject denied discomfort from the earplugs and did not remember removing or replacing them during the night. The investigator completed otoscopic ear examinations prior to earplug placement and after earplug removal in all subjects. Tympanic membranes were intact with a normal appearance and were unchanged before and after earplug placement in 10 subjects. In Subject 1, the postearplug otoscopic examination on the final study night was forgotten until the subject's mother found an earplug in the left ear 4 days later. Otoscopic examination of both ears was normal after removal of the earplug, and no infectious or other complications of the ears were noted during the remainder of the hospital course. The study technicians had been instructed to remove the earplugs at 06:30 hours. When questioned, the technician reported being unable to see the left earplug. Nurses were apparently taking left tympanic temperature readings without noticing the earplug for 4 days. Tympanic temperature measurements were not significantly different (analysis of variance [ANOVA], n = .46) before (N = 45, X = 38.5, SD = 0.6, range = 37.2 to 39.4) and after earplug removal (N = 42, X = 38.8, SD = 0.5, range = 37.6 to 39.8). Otoscopic ear examinations were conducted the morning after earplug use in all other subjects. In Subject 5, tympanic membrane visualization was prevented by an earwax plug in the left ear, and the right tympanic membrane was intact but dull prior to 58 earplug insertion. Postearplug examination of Subject 5 revealed green drainage from the right ear, with a ruptured right tympanic membrane and an intact left tympanic membrane. The attending physician was contacted and confirmed the examination, believing that the otitis media was most likely due to a sinusitis from indwelling nasal tubes and that the tympanic membrane rupture was not caused by earplug use. The ear drainage was cultured and appropriate antibiotic therapy was in place. No further complications of otitis media were noted upon hospital discharge. Tympanic membrane visualization was precluded in Subject 10 by earwax obstructing the ear canal (before and after earplug use). Because of a dim otoscope light and an inability to obtain an otoscope with a brighter light, the tympanic membranes were not well-visualized in Subject 11 during the preearplug otoscopic examination. Postearplug examination revealed blood behind the left tympanic membrane, an intact and normal right tympanic membrane, and a small cut in the skin of the proximal right ear canal. Consultation with a nurse practitioner and critical care intensivist1 confirmed the findings; both clinicians believed that the findings were unrelated to earplug use. Subject 6 stated that discomfort from the earplugs prevented him from turning to his usual right side­lying position. IThe private attending physician was not available. 59 Sleep Instrumentation and Measurement Subject 12, a young trauma victim with unstable cervical spine fractures, was lying on a Roto Resr' bed; head movement was strictly prohibited. The head measurement procedure was modified with the help of experienced electroencephalogram technicians; occipital electrodes were placed symmetrically and slightly lateral to the usual 01 and 02 positions. Electrode impedances were within acceptable limits « 10,000 ohms) in the majority of cases. High measurements « 20,000 ohms) were noted in a few cases. All signals were judged to be technically acceptable for scoring by a single, experienced, board-certified sleep technician. Scalp electrode application required approximatel y 1 hour. The collodion glue (cellulose nitrate 5 %, ethanol 25 %, and ether 70 %) used to affix the scalp electrodes was noxious to one of the staff members. Consequently, a nurse with chronic airways obstruction had to leave the room during scalp electrode gluing because of watering eyes and chest "tightness." One subject was bothered by the smell of collodion remover (dipropylene glycol methyl). Effect of Earplues on Sleep Maintenance Efficiency Index and Rapid Eye Movement Sleep; Research Question 1 Table 9 summarizes the X ± SD, one-tailed, paired! tests with associated effect-size data and power calculations (Elashoff, 1997) for the SMEI and REM sleep. Other polysomnographic sleep measures (Appendix D) are shown for Table 9 The Effect of Earplugs on Sleep Maintenance Efficiency Index and Rapid Eye Movement Sleep (N = 13) No earplugs Earplugs Difference Paired t power Effect size Variable Units t! if 1-B=0.8 X ± SO X ± SD X ± SD I! (1 - B) SMEI Ratio 0.54 ± 0.25 0.59 ± 0.23 - 0.05 ± 0.22 0.44 0.18 0.22 127 REM % SPT 2.40 ± 5.60 5.60 ± 8.00 - 3.20 ± 5.70 0.04 0.60 0.56 22 Other standard sleep measures (for descriptive purposes only) SPT Minimum 432.40 ± 55.00 426.50 ± 105.40 5.90 ± 133.30 0.88 0.06 0.04 3157 SEI Ratio 0.51 ± 0.25 0.55 ± 0.24 - 0.04 ± 0.26 0.61 0.12 0.15 292 SAl Ratio 15.80 ± 8.60 12.80 ± 9.10 3.00 ± 13.10 0.43 0.19 0.23 120 Awakenings t! 47.50 ± 22.60 39.00 ± 21. 60 8.50 ± 35.70 0.41 0.20 0.24 112 Sleep latency Minimum 22.40 ± 28.30 34.40 ± 79.70 -12.00 ± 91. 50 0.65 0.11 0.13 363 REM latency Minimum 137.70 ± 106.30 123.00 ± 28.50 14.70 ± 85.00 0.79 0.14 0.17 210 Time awake Minimum 162.40 ± 104.70 185.50 ± 91. 20 -23.10 ± 90.00 0.37 0.21 0.26 96 NREM sleep Minimum 239.40 ± 113.40 246.20 ± 107.00 - 6.80 ± 114.40 0.83 0.07 0.06 1752 Stage 1 (%) % SPT 47.50 ± 33.90 47.50 ± 38.20 0.10 ± 42.80 1.00 '* * * Stage 2 (%) % SPT 49.50 ± 30.90 46.70 ± 32.30 2.80 ± 38.30 0.79 0.08 0.07 1120 Stage 3 (%) % SPT 0.60 ± 1.20 0.30 ± 1.20 0.20 ± 0.80 0.28 0.27 0.31 66 Stage 4 (%) % SPT 0 0 0 NA NA NA NA Note. SD = standard deviation, SMEI = sleep maintenance efficiency index, REM = rapid eye movement, SPT = sleep period time, SEI = sleep efficiency index, SAl = sleep arousal index, NREM = non rapid eye movement, 6 = probability of a Type II error. *The power analysis program would not calculate effect size because of the small mean difference. g 61 descriptive purposes only. Other Potential Sleep Disrupters: Research Question 2 Noise Ambient sound pressure levels for both nights exceeded 35 to 40 dB(A) (Figure 4), levels recommended for adequate sleep (U.S. Environmental Protection Agency, 1974; Walker, 1978), and were not statistically significantly different on either night (average sound pressure level, 12 = .244; number of events, 12 = .57; and number of sound peaks, 12 = .648). Only 12 subjects were included in the sound data analysis because the sound level meter battery was depleted after 2 hours of recording on the final night in Subject 13. Attenuated sound levels were estimated by subtracting 80% of the 25 dB(A) noise reduction rating [20 dB(A)] from ambient sound pressure level on the night earplugs were worn. Paired 1 tests were significantly different or trending toward significance when attenuated sound levels on the earplug night were compared with ambient sound levels on the nonearplug night (average sound level, 12 < .001; number of events, 12 = .07; and number of sound peaks, 12 = .004). Figure 4 summarizes the sound-level data. Lighting Overhead lights (Figure 5) were Octron® 800 4100K white fluorescent tubes (4100 Spectrum, Color Rendering Index = 82) aligned in double or single columns and recessed into white receptacles covered with an acrylic lens (light diffuser). Double overhead lights were wired separately from the single overhead light. A -. Go) t;:! (.) Cr.l -< -cA' ...8... ~ B ~ Cr.l -. Go) t;:! (.) Cr.l -< -cA' ...8... (.) Go) B ~ ~ Cr.l Actual Sound Pressure Levels (SPL) 8-hour Nocturnal Sleep Periods 250 0 SPL (mean) ~ Events (mean n.) • Peaks >79 (mean n.) 200 *(12 > 0.2, t 12 > 0.5, +12 > 0.6) + 150 -+ "M"" N - - 100 \*0 0*0 If) If) 50 -t t ~ 0 - 0 Earplugs No Earplugs (N=12) (N=12) Estimated Sound Pressure Level (SPL) Attenuation with Earplugs (Earplug Night J, 20 Decibels) 250 200 150 100 50 0 8-hour Nocturnal Sleep Periods o SPL(mean) rn Events (mean n) • Peaks >79 (mean n) *( 12 < .001, t 12 <. 07, +12 <. 004) t o Earplugs (N=12) 0*0 If) t \0 o No Earplugs (N=12) Figure 4. Sound-Level Comparisons. 62 Note. Events = number of times sound pressure level ~ 80 dB(A) for > 2 seconds and peaks = number of peaks sound pressure level (measured at I-minute intervals) ~ 80 dB(A). White fluorescent --Ji;[=========Jl Single overhead tube .... -10 inch X 45 inch ~ .. ... I..~ 1..1.. f Double overhead 21.5 inch X 45 inch 30 foot candles Head of Bed o Light Meter <10 foot candles when used as a night light Foot of Bed 32 foot candles ...... ~ - I.. .. f Double overhead 21.5 inch X 45 inch 30 foot candles Figure 5. Overhead Lighting. 63 64 halogen quartz examination light with a jointed, adjustable arm and light dome was affixed over the bed (Figure 5). The examination lights were wired to a dimmer switch located on the dome. Examination lights, when used during the study nights, were used as a night light and were positioned away from the patient's bed and pointed toward the wall or ceiling. A General Electric BIAXC> electronic bulb (15W) night light was also recessed into the wall approximately 2 feet inside of the room entrance, 14 inches from the floor and covered with a flat 1.25-inch thick 7.5-inch by 3-inch translucent glass lens. Illumination in foot candles was measured on two separate occasions using a General Electric Type 214 light meter positioned at the head of the bed near the patient's eyes in unobstructed room light during 2 of the study nights. Double overhead lights emitted 25- to 30-foot candles, the single overhead light emitted 30- to 32-foot candles, and the examination light (used as a night light) was < 10-foot candles. The night light emitted very low light levels; the light meter dial was very difficult to read when only the night light was on. The night light was not considered to be a sleep­disturbing factor and was included in the "lights-out" time. Paired 1 tests for all lighting data were not significantly different (12 > 0.1). Figure 6 summarizes the lighting data. Ambient Temperature Room temperature was measured hourly at the head of the bed in the first 11 patients. Temperature at the skin surface, underneath the gown or covers, was measured in Subjects 12 and 13. Paired 1 tests for the room temperature data were 500 400 v.I a 300 .S ::a 200 100 o -t--'--- Lighting (alln > 0.1) 8-hour Nocturnal Sleep Periods D Lights out • 1 overhead ~ 20verhead t2] Exam (dim) Earplugs <N=13) No Earplugs <N=13) Figure 6. Lighting Comparisons. 65 not statistically significant (12 = .14, N = 88) and for the skin temperature data were significantly different W = .003, N = 16). Figure 7 summarizes the temperature data. Most of the subjects were covered with a hospital gown only; however, detailed data about covers were not collected. Care Content Intensity of Care Intensity of care, as reflected by the average Therapeutic Intervention Severity Score, was not significantly different between the 2 nights. A change in the intensity of care was defined a posteriori as a change in total points ~ 4 and a 66 Ambient Temperature in Room 40 - (n = 38 , S u b~' ects 1 to 11) - (n=88) (n=88) 20 22.9 23.1 ± ± 1.5 1.0 o I Earplugs No Earplugs Ambient Temperature at Skin Swface 40 . (12 < 01 , Subjects 12 and 13) (n=16) (n=16) 31.2 32.6 ± ± 1.0 1.3 o Earplugs No Earplugs Figure 7. Ambient Temperature Comparisons. 67 change in classification category. On the night earplugs were worn, intensity of care was lower in Subject 3 and higher in Subject 5. Figure 8 summarizes the Therapeutic Intervention Severity Score. Medications Five subjects (1, 2, 5, 8, and 12) had seven continuous intravenous sedative infusions prior to enrollment. Subject 1 had fentanyl and diazepam infusions that were discontinued 8 and 16 days prior to enrollment, respectively. Subject 12 had simultaneous fentanyl and lorazepam infusions that were both discontinued 2.3 days prior to enrollment. All patients were mentally responsive at the time of enrollment. 40 - 20 - - - o Intensity of Care T 19.2 ± 5.5 Earplugs (N=13) U1. -.5 9) T 19.8 ± 5.3 No Earplugs (N=13) Figure 8. Intensity of Care Comparisons. 68 Table 10 summarizes the preenrollment intravenous sedative, analgesic, and paralytic drug data. The total dose per day for all medications administered between 0600 the morning before until 0600 the next morning was recorded. Seventy-six different medications were administered. After eliminating medications with exactly equivalent doses from both study days, 48 different medications remained. Each of the 48 medications was classified using the American Hospital Formulary Service's Pharmacologic-Therapeutic ClassificationCl (1996, 1998). The uses, cautions, and adverse effects were reviewed for sleep­related central nervous system effects, including sleeplessness, restlessness, insomnia, drowsiness, anxiety, and nervousness. After eliminating medications without sleep-related central nervous system effects, 28 remained. Twenty of the 28 remaining medications were eliminated because the incidence of sleep-related central nervous system effects was s 1 %; therefore, it was unlikely that the subjects experienced any sleep-related central nervous system effects from the medications. The remaining 8 medications included sedatives, analgesics, or hypnotic medications. Medication doses given on the study nights between the hours of 1800 to 0600 were reviewed. Fentanyl was converted to morphine equivalents (fentanyl 700 mcg = .7 mg X 10 = morphine 7 mg). Five subjects' (1, 2, 3, 5, and 6) medication doses were equivalent on both study nights. Four subjects (7, 9, and 10) received more sedatives and analgesics on the nonearplug night. Three subjects (4, 12, and 13) received more of some but less of other sedatives and analgesic medications on both study nights. Two subjects (8 and 11) Table 10 Preenrollment Sedativell Analgesic, and Paralvtic Medications Bolus IV doses Continuous IV infusions Days from last t! doses Days from Infusion dose to study infusion dIe duration Drug name t! subjects entry Minimum t! subjects to study Minimum (days) X ± SO X ± SO X ± SD X ± SD Diazepam 5 8.5 ± 5.1 2.6 15.8 ± 14 15.5 ± 15.5 6.5 ± Lorazepam 7 5.8 ± 7.8 0.4 6.0 ± 5 2 6.0 ± 5.3 2.3 6.6 ± 0.2 Midazolam 10 7.3 ± 8.4 0.2 17.2 ± 16 2 4.9 ± 0.9 4.2 4.1 ± 3.9 Morphine 12 4.8 ± 7.2 0.3 35.0 ± 18 0 ± ± Fentanyl 3 7.7 ± 7.6 0.3 12.6 ± 18 2 5.0 ± 3.8 2.3 10.8 ± 5.1 Meperidine 2 8.9 ± 3.5 6.5 22.5 ± 21 0 ± ± Paneuronium 7.3 ± 7.3 27.0 ± 0 ± ± Veeuronium 4 14.2 ± 3.6 12.3 5.0 ± 4.1 7.2 ± 7.2 5.1 ± $ received more sedatives and analgesics on the earplug night; however, the doses and differences between the 2 nights were small. Surgical Procedures 70 Eleven subjects had 15 surgical procedures. All procedures were conducted under general anesthesia (X + SD duration = 3.3 + 1.9 hours and X ± SD time from completion to enrollment = 13.3 ± 9.1 days). The minimum time from surgery to enrollment was 2 days. Care Process Pain No significant differences in the Verbal Pain Assessment Scale or morphine equivalents between the study nights were noted (Figure 9). Equivalent doses of tramadol were given to Subject 12 on both nights; otherwise no other nonnarcotic analgesics were given. Three subjects (3, 5, and 11) seemed to have difficulty understanding the pain-scoring process, presumably due to confusion or fatigue. Subject 3 reported a pain score of 8, even though her behavior was outwardly calm and relaxed, and she fell asleep almost immediately after recording began. Sleep Opportunity The average time spent undisturbed was 17 minutes higher <n < .05), and disturbances were less frequent on the night earplugs were worn (Figure 10). Disturbances were defined as external to the subject; related to subject care; and categorized as minimum, moderate, maximum, and other (Table 3). Minimum 0 -a u CI) .... r::: ~ c:n ~ c:n -< .~ ~ -a .0 ~ > c:n ~ ::= ~ 3 2.5 2 1.5 1 0.5 0 3 - 2.5 - 2 - 1.5 - 1 - 0.5 - 0 Pain Scores Prestudy and Poststudy Night (aliI! > 0.1) ~• Post -.:::t ....-1 -.:::t ....-1 ~ ....-1 l': 0 Earplugs No Earplugs (N=13) (N=13) Morphine Equivalents Administered During Study (n = .77) V'1 ....-1 'I'" Earplugs <N=13) I 00 ....-1 No Earplugs <N=13) Figure 9. Pain and Morphine Equivalent Comparisons. 71 I 400 300 r.n B::s .5 200 ~ 100 o 160 140 120 cu~ 100 g 80 1 60 40 20 0 - - - - 00 ~ Average Undisturbed Time 8-hour Nocturnal Sleep Periods <n < .05) - 408 Earplugs (N=13) I .. 391 No Earplugs (N=13) Frequency of Disturbances by Category 8-hour Nocturnal Sleep Periods ('f") Earplugs (N =13) \0 ~ No Earplugs (N=13) I o Minimum ~ Maximum • Moderate ~ Other Fifjure 10. Average Undisturbed Time and Disturbance Frequency. 72 73 disturbances were eliminated from further comparisons because they were equivalent on both nights and were unlikely to have contributed to sleep disruption. Further analysis revealed fewer disturbances, shorter length of disturbances, and longer periods of time between disturbances on the night earplugs were worn. On average, there were significantly fewer disturbances, and the average disturbance length was significantly lower on the night earplugs were worn (Figure 11) W < .05). The average time between disturbances was higher (X ± SEM = 44.5 ± 15.6 minutes versus 20.6 ± 1.9 minutes) but not statistically significant (Figure 11). Seven subjects were given nine opportunities to complete one 90- minute sleep cycle once per night (N = 4), twice per night (N = 2), and four times per night (N = 1) when earplugs were worn. One subject was given one opportunity and another subject was given three 90-minute sleep cycle opportunities when earplugs were not worn. The frequency of disturbances displayed by time between disturbances revealed some interesting patterns (Figure 12). On the night earplugs were worn, there were fewer disturbances separated by 1 to 60 minutes, an equal number 61 to 90 minutes apart, and a greater number of disturbances separated by more than 90 minutes. Sleep Position No significant differences in the time spent in the usual sleep position or any other sleep position were present; most of the sleep time was spent in the supine position. Figure 13 summarizes the sleep-position data. >. u c: 0 ::s 0" J: 0 e !;! ..( Con £ ::s .5 ~ 40 - 35 - 30 - 25 - 20 - 15 - 10 - 5 - 0 Average Disturbance Frequency 8-hour Nocturnal Sleep Periods U2 < .05) T 1 - X=18 Range=2 to 42 Earplugs (N=13) I T 1 - X=24 Range= 13 to 40 No Earplugs (N=13) I Average Disturbance Length and Time Between Disturbances 100 80 * 60 40 20 0 (*11 < .05, til = .12) 0 Length t Earplugs (N=13) * • Time between t No Earplugs (N=13) Figure 11. Average Frequency, Length, and Time Between Disturbances. 74 75 Frequency by Time Between Disturbances 00 Minutes .0... -1 Between 100 D 1-5 ~ 6-10 • 11-30 80 f2J 31-60 0 61-90 >. 60 • >90 uc: 0 ::s 1 40 20 0 Earplugs (N=13) No Earplugs (N=13) Fi~ure 12. Disturbance Frequency by Time. Anxiety Sixteen Motor Activity Assessment Scale ratings were recorded at 30- minute intervals from 2300 to 0630 each study night. Eighty-eight percent (368/416) of Motor Activity Assessment Scale ratings were between two and three, indicating quiescence or a calm state. (See Table 6 for detailed definitions.) Motor Activity Assessment Scale ratings indicating agitation were higher in 16 cases, lower in 6 cases, and equivalent in 2 cases on the night earplugs were worn compared to the nonearplug night. Three subjects (4, 7, and 11) all accounted for the Motor Activity Assessment Scale ratings, indicating restlessness and agitation. 500 400 tI:l £ 300 ::s .S ::E 200 100 Sleep Position 8-hour Nocturnal Sleep Periods (all12 > .13) Earplugs (N=13) No Earplugs (N=13) Figure 13. Sleep Position Summary. D Usual ~ Supine II Right ~ Left § Other 76 Motor Activity Assessment Scale ratings were consistently higher in Subjects 4 and 7 and consistently lower in Subject 11 on the night earplugs were worn. Figure 14 illustrates more frequent restlessness and agitation ratings on the night earplugs were worn. Pharmacologic anxiety management (benzodiazepines) was not used on either night in 8 subjects. Benzodiazepine doses were lower in 2 subjects (7 and 10), equivalent in 2 subjects (2 and 13), and higher in 1 subject (9) on the night earplugs were worn, although the midazolam dose difference in Subject 9 was not impressive (fable 11). >. u t:: Q) ::s ~ ~ ~ 120 100 80 60 40 20 0 Motor Activity Assessment Scale (MAAS) Frequency N M ~ Earplugs (N=13) MAASRating o Quiescent g Calm • Restless Ea Agitated ~ Combative No Earplugs (N=13) Figure 14. Motor Activity Assessment Scale Summary. Descriptive Evaluation of Potentially Important Sleep Disrupters A multivariate analysis was precluded by the small sample size in this study; therefore, a descriptive approach to the data analysis was completed. For 77 individual subjects, the researcher subjectively compared each of the environment, care content, and care process variables on both study nights. The intent was to gain some understanding of whether or not the effect of earplugs on REM and SMEI sleep outcomes may have been obscured by other variables. Table 12 presents a summary of the analysis, illustrating a clear pattern of increased sleep opportunity and reduced noise on the night earplugs were worn. The table also 78 Table 11 Benzodiazepine Use (M~) Midazolam Other Subject Earplugs No earplugs Earplugs No earplugs 2 7 9 10 13 aLorazepam. bDiazepam. 2a Ob 3.0 0 4.5 10 3.5 3 illustrates the complexity of understanding the contribution of other sleep­disrupting factors given the small sample. Intennediate Health Outcomes Sleep Deprivation 2- 20b All data were scored by a single, experienced rater who was unaware of subject assignment and was board certified for sleep measurement and scoring by the American Sleep Disorders Association. Sleep architecture was strikingly disturbed in all subjects on both nights (Table 13). Eleven subjects met the criterion for sleep deprivation (SMEI < 0.82) on both study nights. Subject 9 met the criterion on the nonearplug night (SMEI = 0.27) but not on the night earplugs Table 12 Subjective Evaluation of the Effect of Potentially Important Sleep Disrupters on Rapid Eye Movement and Sleep Maintenance Efficiency Index Subject II Noise Light Amb terne Int of care Meds OR Pain SleeE 0Ee SleeE EOS Anx REM SMEI t ? t J. 2 t ? J. t J. 3 t ? t 0 t 4 t ? t J. 0 t 5 t ? J. J. 0 J. 6 t ? t J. t t 7 t ? J. t J. 0 t 8 t ? t t t t 9 ? ? 0 J. 10 t t ? t 0 J. 11 t ? t 0 J. 12 t t ? t t t 13 t ? t 0 t All t t t Note. Amb temp = ambient temperature (C), int = intensity, meds = sedative/analgesics, OR = surgical procedures, opp = opportunity, pos = position, anx = anxiety, REM = rapid eye movement, SMEI = sleep maintenance efficiency index, ? = unknown, t = variable changed to favor sleep on the night earplugs were worn, and , = variable changed to oppose sleep on the night earplugs were worn. -....l \0 Table 13 Sleep Maintenance Efficiency Index and Percentage of Time Spent in Sleep Stages (N = 13) % sleep SMEI REM Stage 1 Stage 2 Stage 3 Stage 4 P NP P NP P NP P NP P NP P Mean 0.56 0.48 47.5 47.5 46.7 49.5 0.3 0.6 0 0 5.6 SD 0.23 0.26 38.2 33.9 32.3 30.9 1.1 1.1 0 0 8.0 Minimum 0.16 0.01 05.2 07.8 00.0 00.0 0.0 0.0 0 0 0.0 Maximum 0.94 0.86 100.0 100.0 78.3 85.6 4.2 3.6 0 0 21.2 Note. SMEI = sleep maintenance efficiency index, P = earplugs, NP = no earplugs, and REM = rapid eye movement. NP 2.4 5.6 0.0 18.6 00 o 81 were worn (SMEI = 0.82). The SMEI in Subject 13 exceeded 0.82 on both nights; however, all of the sleep time was spent in Stage 1 or Stage 2. No Stage 4 sleep was evident in any subject. Stage 3 sleep was evident in 3 subjects (1, 5, and 6). Sensory Alteration Sixteen Sensory Alteration Scores were recorded at 30-minute intervals from 2300 to 0630 each study night, yielding 16 per night (N = 416, Sensory Alteration Scores). Ninety-nine percent (410/416) of Motor Activity Assessment Scale scores were :$; 1, indicating the presence or absence of restraints. Figure 15 illustrates the similarity in Sensory Alteration Scores on both study nights. Restraint use did not appear to be uniformly related to agitated behavior. Subjects 4 and 11 were given a Sensory Alteration Score of 2 because of dangerous agitation. Restraints were not used in Subject 4. Circadian Disruption Body temperature was measured with a tympanic thermometer and an axillary thermistor on the nonearplug night. The tympanic thermometer was not used on the night earplugs were worn. Figure 16 illustrates the temperature curves from both nights in the first subject. Displacement of the axillary probe was not uncommon and even when intact the axillary temperature tended to underestimate tympanic temperature and rendered the data unusable for documentation of the temperature curve. Sensory Alteration Score (SAS) 1 1 20 O-t------- Earplugs No Earplugs Figure 15. Sensory Alteration Summary. Time on Mechanical Ventilation SAS [J]0 II 1 ~ 2 82 The average length of mechanical ventilation in ICU was 24.3 ± 19.6 (SD) days. Three subjects were still ventilator dependent at ICU and hospital discharge; they were transferred to a rehabilitation unit on the ventilator. Average length of mechanical ventilation on the rehabilitation unit was 50.7 ± 23.5 days, and average length of all mechanical ventilation was 47.3 ± 28.5 days (Table 14). Outcomes at Hospital Discharge Survival to hospital discharge was 69% (9/13). Four subjects were transferred to a rehabilitation unit and 2 died on the rehabilitation unit or after 38 37.5 37 0 1 36.5 '::1 c 0 36 u til 0 35.5 ~ 8 35 34.5 34 40 39.5 i 39 '::1 c 0 U 38.5 til ~ 8 38 37.5 37 Body Temperature Subject 1 (No Earplugs) ~ I \ \ I \ \ I I- a·- • A~Pxirlloarby e I 08 12 16 20 4 February 1997 Body Temperature Subject 1 (Earplugs) \ I \ I M ()() \ I!J 04 I · Em: Probe I - B- - Axillary C 08 12 16 20 ()() 06 6 February 1997 Figure 16. Body Temperature Data. 83 Table 14 Outcomes at Hospital and Rehabilitation Unit Discharge (N = 13) ICU LOS Hosp LOS RU LOS ICU vent days RU vent days Total vent days Hosp SD RU SD Discharge disposition Mean SD Minimum Maximum 30.0 17.3 16.0 79.0 40.5 17.2 18.0 80.0 51.1 26.1 13.0 71.0 24.3 19.6 1.0 79.0 50.7 23.5 25.0 56.0 45.2 30.9 1.0 105.0 S-9 D4 S-2 D-2 3 = home 4 = died on vent 2 = vent withdrawn Note. leu = intensive care unit, LOS = length of stay, vent = mechanical ventilation, S = survived, and D = died. 00 ~ readmission to the hospital. Survival to rehabilitation unit discharge was 54 % (7/13). leu, hospital, and rehabilitation unit length of stay are summarized in Table 14. Day and Afternoon Sleep: Research Question 3 85 After the second study night (2230 to 0630), sleep in 7 subjects was monitored during the day (0630 to 1430) and afternoon (1500 to 2230) hours. The recording was interrupted for approximately 1 hour to provide the technicians a break and archive the sleep data from the computer. During the interruption, all patients were awake (sitting in a chair or having procedures); therefore, the sleep data probably represent total sleep during a 24-hour period, even though the total recording time was 22.6 ± 0.4 (X ± SD) hour. Total sleep time during the 24- hour period was 8.1 ± 1.8 hour (X ± SD); most of the time was spent in non­REM sleep (7.7 ± 1.5 hour). Sleep architecture was markedly disturbed; that is, the sleep obtained was not normal, consolidated sleep. The majority of time in non-REM sleep was spent in Stage 1 (45.1 % ± 28.6%) and Stage 2 (51.7% ± 26.1 %) sleep. Table 15 summarizes the day, night, afternoon, and 24-hour sleep measures. Table 15 Ni~ht. Day. Afternoon. and Twenty-Four-Hour Total Sleep Measures (N = 7) Time period time TRT (hour) TST (hour) Non-REM (hour) REM (hour) Stage (% TST) SEI SMEI SAl Hours awake r! awakenings Sleep latency 2 3 4 REM Night ~230 to 0630 X ± SO = 8.00 ± 0.00 3.40 ± 0.60 3.30 ± 0.60 .13 ± 0.10 45.30 ± 12.00 52.10 ± 11.00 0.20 ± 0.20 0.00 ± 0.00 2.40 ± 2.00 .43 ± .08 .44 ± .08 7.05 ± 4.08 3.90 ± .47 47.60 ± 11.50 111.70 ± 39.40 Day 0630 to 1430 X ± SO 7.50 ± 0.50 1.80 ± 1.10 1.60 ± 0.80 .10 ± .19 45.90 ± 30.10 50.30 ± 27.90 1.40 ± 3.60 0.00 ± 0.00 3.00 ± 5.50 .25 ± .14 .27 ± .13 4.30 ± 2.90 4.43 ± 1.02 30.30 ± 15.00 Afternoon 1500 to 2230 24-hour total 2230 to 2230 X ± SO X ± SO == 7.30 ± 0.40 22.60 ± 0.40 2.90 ± 2.00 8.10 ± 1.80 2.86 ± 1.86 7.70 ± 1.50 0.17 ± 0.24 0.40 ± 0.50 44.00 ± 28.40 45.10 ± 28.60 52.60 ± 25.30 51. 70 ± 26.10 0.10 ± 0.20 0.50 ± 2.10 0.10 ± 0.30 0.00 ± 0.20 3.40 ± 4.80 2.90 ± 4.90 .44 ± .27 .39 ± .06 .44 ± .30 ± 6.10 ± 4.60 6.10 ± 4.60 3.00 ± 2.00 11.40 ± 2.00 29.00 ± 22.40 106.80 ± 50.00 Note. TRT = total recording time, TST = total sleep time, REM = rapid eye movement, SEI = sleep efficiency index, SMEI = sleep maintenance efficiency index, SAl = sleep arousal index, and N = number. SMEI could not be computed for 24 hours. Sleep latency was not a meaningful daytime sleep measure in this study. 00 0\ CHAPTER V DISCUSSION AND RECOMMENDATIONS The major conclusions of this study are the following: (a) Five of 13 critically ill subjects had more REM sleep on the night earplugs were worn; and (b) all 13 subjects had severely disturbed, fragmented sleep architecture compared to normal sleep architecture-with or without earplugs. Very little REM sleep was obtained, and SWS sleep (Stages 3 and 4) was virtually absent. One could argue that the lack of SWS was expected in older subjects; that is, SWS, as conventionally scored, diminishes with age. However, SWS does not completely disappear in older individuals; 31 % (4/13) of the patients in this study were <45 years old, and the decline of age-related SWS is controversial (Bliwise, 1994). The following discussion presents the sample characteristics and outcomes, conclusions to the research questions, study design and measurement issues, and future research. Sample Characteristics and Outcomes The sample was a heterogeneous group of critically ill subjects who were, with one exception, treated with long-term mechanical ventilation. The average APACHE II score at the time of enrollment (X, SD = 13.9 + 5.4 days) is typical of patients who are medically stable but in the lCU because of ventilator 88 dependence or high nursing care needs (Ward, 1995). Patients who require long­term mechanical ventilation (defined as more than 3 days) represent 3 % to 5 % of critical care admissions and consume 30% to 50% of critical care resources (Cohen & Chalfin, 1