PATIENTS’ RECALL AND EVALUATION OF MECHANICAL VENTILATION: IMPACT OF SEDATION A DISSERTATION SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF THE UNIVERSITY OF MINNESOTA BY JILL LYNN GUTTORMSON IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY LINDA CHLAN, ADVISOR JUNE 2011 © Jill L. Guttormson (2011) i Acknowledgements I would like to thank Linda Chlan, my academic advisor, for her support, encouragement, and guidance throughout my doctoral studies. I would also like to thank the other members of my committee: Sue Henly for advice and patience over the last three months and Craig Weinert and Cynthia Gross for their time and encouragement. I would like to acknowledge my funding sources. My doctoral studies were funded in part by an NRSA fellowship from the National Institute of Nursing Research, National Institutes of Health. My dissertation research was funded in part by an American Association of Critical-Care Nurse mentorship grant. A special thank you to the wonderful nurses and staff at my research site who still greeted me with a smile even in the eighteenth month of recruitment. It was a pleasure to work with each of them. My deepest thank you to family and friends for their understanding and support. Thank you for the patience and good humor with which you gracefully accepted the unreturned phone calls and emails. To Karin, thank you for providing a laugh, a shoulder, or a diversion and for knowing when to offer each. To my parents, there is not enough space to describe the ways you have supported me over the last few years. I thank you for it all. To my daughters, Lea and Claire, who were perhaps even more excited than me when “that really big paper” was completed, thank you for letting me work but also for making sure that I played. To my husband, Eric, I can’t begin to describe what your constant support and encouragement has meant. Thank you for believing not only that I could but also being so firmly convinced that I should. ii Dedication This dissertation is dedicated to the patients that so generously shared their experiences and memories with me. iii Abstract Purpose: Mechanically ventilated patients routinely receive sedative medications to manage patient symptoms of anxiety and agitation. However, these medications are associated with multiple complications and the effectiveness of sedation for improving symptoms during mechanical ventilation has not been evaluated from the patient’s perspective. The purpose of this study is to examine the relationship between the pattern of sedation and mechanically ventilated patients’ evaluation of critical care and to identify non-pharmacologic interventions patients find effective during mechanical ventilation. Subjects: Sixty-nine mechanically ventilated patients were enrolled from a medical surgical intensive care unit in a Midwestern community hospital of whom 35 completed post ICU interviews. Subjects completing interviews had a mean age of 66 (SD 12.6) and were mechanically ventilated for an average of 4.5 days (SD 6.8). Methods: Level of arousal data utilizing the Motor Activity Assessment Scale and all sedative medication received were abstracted from the medical record. Subject interviews were conducted after ICU transfer and included the ICU Memory Tool and Intensive Care Experience Questionnaire. Latent Class Growth Analysis was used to classify subjects based on patterns of arousal and sedative exposure over the first 5 days of mechanical ventilation. Results: The most common memory of ICU was confusion. Hallucinations, difficult communication, and feelings of loss of control were the most disturbing memories for subjects. Subjects that were minimally arousable over the first five days of mechanical iv ventilation had more delusional memories of ICU. There was no difference in patient satisfaction, awareness of surroundings, frightening experiences, memories of negative feelings, or factual memories based on level of arousal. Sedative exposure did not impact patients’ memories of ICU. A lower level of arousal and higher sedative exposure were associated with increased time on the ventilator and in the ICU. Subjects reported family presence, receipt of information, and freedom of movement as helpful during mechanical ventilation. Conclusions: Deeply sedating mechanically ventilated patients may not be effective for improving the patient’s experience. Improving communication, frequent provision of information, providing choices to patients, and inclusion of patients’ families are basic interventions that may improve the patients’ experience of mechanical ventilation. v Table of Contents Acknowledgements ……………………………………………………………... i Dedication ………………………………………………………………………. ii Abstract …………………………………………………………………………. iii Table of Contents ………………………………………………………………... v List of Tables ……………………………………………………………………. viii List of Figures …………………………………………………………………… ix List of Abbreviations …… ……………………………………………………… x List of Symbols and Terms ……………………………………………….……. xi List of Equations ……………………………………………………………….. xii Chapter I: Introduction ………………………………………………………... 1 Chapter II: Literature Review …………………………………………………... 9 Patients’ evaluation of mechanical ventilation ……………………………. 9 Sedation of mechanically ventilated patients ……………………………... 15 Complications related to sedation …………………………………………. 20 Research of sedation methods …………………………………………........ 25 Sedation protocols or algorithms ………………………………………… 25 Daily interruption of sedation ………………………………………….... 31 Analgosedation ………………………………………………………….. 32 Sedation guidelines ……………………………………………………….... 33 Sedation Practice ……………………………………………………………. 35 Chapter III: Research Methods 39 Conceptual framework ……………………………………………………. 39 Specific aims …………………………………………………………….. 41 Setting and sample ………………………………………………………… 42 Data collection procedures, study variables, and measures ……………….. 43 Subject characteristics …………………………………………………. 43 Severity of Illness …………………………………………………….. 43 Pattern of sedation …………………………………………………….... 45 Sedative Exposure ……………………………………………………. 46 Level of Arousal ……………………………………………………… 47 Patients’ memories of ICU ……………………………………………… 49 Intensive Care Experience Questionnaire ……………………………. 49 Intensive Care Unit Memory Tool …………………………………… 49 vi Open-ended interview questions …………………………………….. 50 Non-pharmacologic interventions ……………………………………… 50 Human subjects protection and data management …………………………. 51 Data analysis …………………………………………………………………. 51 Sample size ……………………………………………………………….. 51 Analysis by specific aim ………………………………………………….. 52 Aim one: description of patient memories ...………………………….. 52 Aim two and three: pattern of sedation and relationship with memories 53 Building the LCGA Model ………………………………………… 53 Formal LCGA model specification ………………………………... 57 Aim four: interpersonal, physical, and environmental interventions … 59 Chapter IV: Results ….………………………………………………………….. 60 Subject Recruitment and Characteristics ……………………………………. 60 Description of Patients Memories of ICU …………………………………… 63 Memories and Patterns of Sedation: Level of Arousal ………………………. 74 Arousal trajectories: model selection .……………………………………. 74 Arousal latent trajectory class descriptions ………………………………. 78 Arousal trajectories, subject characteristics, and memories ……………… 79 Memories and Patterns of Sedation: Sedative Exposure .……………………. 83 Sedation intensity trajectories: model selection .…………………………. 83 Sedation intensity latent trajectory class description .…………………….. 86 Sedation intensity trajectories, subject characteristics, and memories …… 86 Relationship between arousal and sedation intensity trajectory classes …….. 86 Non-pharmacologic Interventions .…………………………………………... 91 Chapter V: Discussion ….………………………………………………………. 96 Patients Memories and Evaluations of ICU ………………………………… 96 Trajectories of Arousal and Sedative Exposure …………………………….. 100 Level of Arousal, Sedative Exposure, and Memories of ICU ………………. 101 Non-pharmacologic Interventions ………………………………………….. 104 Limitations …………………………………………………………………. 105 Implications for Research ………………………………………………….. 105 Implications for Nursing Practice ………………………………………….. 106 Conclusion …………………………………………………………………… 108 References ……………………………………………………………………… 109 vii Appendices ............................................................................................................ 132 Appendix A: Literature Review Search Strategies ………………………… 132 Appendix B: Studies Included in Literature Review ………………………. 135 Summary of Patient Experience Research Studies ……………………….. 135 Summary of Sedation Methods Research Studies ……………………... 142 Appendix C: Copies of Interview Instruments and Permissions for Use ….. 157 Intensive Care Unit Memory Tool ……………………………………... 157 Intensive Care Experience Questionnaire ……………………………….. 158 Permission for use of the Intensive Care Unit Memory Tool ……………. 161 Permission for use of the Intensive Care Experience Questionnaire ……. 162 Appendix D: Non-pharmacologic Instruments ………………………………. 163 viii List of Tables Table Page 1 Commonly Used Intravenous Sedative and Analgesic Medications 17 2 Comparison of Sedation Scales 27 3 Comparison of Sedation Guidelines 36 4 Reported Use of Sedation Scales, Daily Interruption Sedation, and Sedation Assessments 37 5 Study Variables, Measurement Instruments, and Measurement Frequency 44 6 Motor Activity Assessment Scale 48 7 Arousal LCGA Model Variables 56 8 Sedative Exposure LCGA Model Variables 57 9 Characteristics of Study Sample 62 10 Intravenous Sedative and Analgesic Medications Received 63 11 Participant Responses to the Intensive Care Unit Memory Tool 64 12 Participant Responses to the ICU Experience Questionnaire Subscales 65 13 Participant Responses to the ICU Experience Questionnaire Individual Items 66 14 Level of Arousal Growth Curve Models 75 15 Level of Arousal LCGA Unconditional Models 76 16 Level of Arousal LCGA Conditional Models 77 17 Level of Arousal Model Parameter Estimates 78 18 Level of Arousal Model Characteristics by Class 80 19 Sedation Intensity LCGA Unconditional Models 84 20 Sedation Intensity LCGA Conditional Models 85 21 Sedation Intensity Model Parameter Estimates 87 22 Sedation Intensity Model Characteristics by Class 88 23 Non-Pharmacologic Intervention Frequency: Patients’ Reports 93 24 Non-Pharmacologic Intervention Frequency: Patients’ Evaluations 94 25 Frequency of Non-Pharmacologic Interventions Reported by Nurses 95 B1 Summary of Patient Experience Studies Included in the Literature Review 135 B2 Summary of Sedation Methods Research Included in the Literature Review 142 ix List of Figures Figure Page 1 Conceptual Framework: Factors Influencing Patient Recall and Assessment of ICU 40 2 LCGA Framework for Level of Arousal Analysis 54 3 LCGA Framework for Sedation Intensity Analysis 55 4 Estimated means for the two classes of level of arousal 81 5 Adjusted estimated means and observed individual values for the easily arousable class 82 6 Adjusted estimated means and individual observed values for the minimally arousable class 82 7 Estimated means for the two classes of sedation intensity 89 8 Adjusted estimated means and individual observed values for the decreasing sedation intensity class 90 9 Adjusted estimated means and individual observed values for the increasing sedation intensity class. 90 x List of Abbreviations ARDS acute respiratory distress syndrome BPS behavior pain scale CNS central nervous system CVP central venous pressure DIS daily interruption sedation ETT endotracheal tube GABA gamma amino-butyric acid GCS Glasgow Coma Scale GCS-CP Glasgow Coma Scale-modified by Cook and Palma ICEQ Intensive Care Unit Experience Questionnaire ICU intensive care unit ICU-AW intensive care unit-acquired weakness ICUM Intensive Care Unit Memory Tool IV intravenous LOS length of stay MAAS Motor Activity Assessment Scale MODS Multiple Organ Dysfunction Score MV mechanical ventilation NRS numeric rating scale PRIS Propofol related infusion syndrome PTSD post-traumatic stress disorder RASS Richmond Agitation Sedation Scale RCT randomized controlled trial SAS Sedation-Agitation Scale SBT spontaneous breathing trial xi List of Symbols and Terms Observed and Latent Variables y Continuous outcome variables; arousal or sedation intensity x Covariates X Matrix of covariates in model (xage and xAPACHE) yik Vector containing observed variables for individual ith in the kth class q Number of covariates Subscripts and Counters i Subscript for individual subject p Number of measurement occasions (1,2,3, 4, or 5) m Number of growth parameters (intercept, slope, quadratic) c Latent class indicator k Number of latent classes Vectors and Matrices of Parameters νik Vector of y intercepts Λk Fixed design matrix (p by m) linking time to growth parameters ηik Vector of growth parameters for the ith individual in the kth class ε Matrix of residuals for y at each occasion; individual error αk Vector of m latent means for the kth class πi k dimensional vector of the probabilities of class membership given the covariates logit πi k - 1 dimensional vector of the log odds of class membership given the covariates αc k -1 vector of the intercepts in the model for the log Γc k -1 by q matrix containing the change in log odds of class membership per unit change in the covariates Y Vector of probability of class membership given X ik Vector of distal outcome means for individual i in the kth classes νc Vector of distal outcome means for trajectory classes xii List of Equations Equation Page Unconditional LCGA Model 1 Unconditional growth mixture ………………………………. 57 2 LCGA model restriction …………………………………….. 58 Conditional LCGA Model 3 Log odds of class membership ……………………………….. 58 4 Inverse logit transformation ………………………………….. 58 5 Probability of class membership……………………………… 58 6 Conditional LCGA Model Distal Outcome Means………………. 58 1 CHAPTER I Introduction “But what I see these days are paralyzed, sedated patients, lying without motion, appearing to be dead, except for the monitors that tell me otherwise. Why this syndrome of sedation and paralysis has emerged baffles me, because this was not the case in the past. ….The only drugs we used for sedation and pain were morphine and occasionally low doses of benzodiazepines" (Petty, 1998 p. 360). In the critical care setting, it is common practice to administer intravenous sedative medications often by continuous infusion to patients receiving mechanical ventilation, a supportive modality used to treat respiratory failure from a variety of causes. Patients are sedated to provide patient comfort, to promote patient safety, to induce patient amnesia, to decrease anxiety and agitation, and to prevent ventilator dysynchrony (Brush & Kress, 2009; Jacobi et al., 2002; Mehta et al., 2006; Rhoney & Murry, 2003; Weinert, Chlan, & Gross, 2001). The trend in sedation management described in the opening quotation, although written almost 14 years ago, appears to be continuing. The number of patients receiving continuous sedation during mechanical ventilation increased between 2001 and 2007 from 40 to 67% (Wunsch, Kahn, Kramer, & Rubenfeld, 2009). On day two of mechanical ventilation, 57% of patients were in a deep state of sedation (unresponsive or responding only to physical stimulation) (Payen et al., 2007). In a study where one third of nurse observations ranked patients as minimally arousable or unarousable, only 2.6% of nurse ratings indicated the patient 2 was oversedated (Weinert & Calvin, 2007). While sedative medications may be indicated at times, they contribute to physical and psychological complications both during and after intensive care.   Complications Related to Sedation Continuous intravenous sedation and greater sedative exposure is associated with a longer duration of mechanical ventilation, longer length of intensive care unit (ICU) stay, and more reintubations (Arroliga et al., 2005; Kollef et al., 1998; Kress, Pohlman, O'Connor, & Hall, 2000). Increased duration of mechanical ventilation increases the incidence of pneumonia, airway damage, decreased mobility, decreased self care ability, and increased health care costs (Douglas, Daly, Gordon, & Brennan, 2002; Kollef et al., 1998; Ostermann, Keenan, Seiferling, & Sibbald, 2000). The amount of sedation received while on the ventilator may contribute to adverse psychological outcomes. Increased duration of sedation is associated with increased symptoms of depression after critical illness (Nelson, Weinert, Bury, Marinelli, & Gross, 2000). As many as 69% of patients suffer from depression and continue to have depressive symptoms at two years post critical illness (Hopkins et al., 2005; Nelson et al., 2000). Sedation, particularly the use of lorazepam, increases the risk of developing delirium during critical illness which is associated with increased mortality and increased length of hospital stay (Ely et al., 2004; Pandharipande et al., 2006). Higher total doses of lorazepam (Girard et al., 2007) and prolonged duration of sedation (Jones et al., 2007) have been linked to the development of post-traumatic stress symptoms. 3 These physical and psychological complications of sedative therapy contribute to a prolonged recovery period and impact patients’ daily lives. Patients suffering from PTSD after ICU have lower health related quality of life compared to the general population particularly in the general health, social function, and mental health dimensions (Kapfhammer, Rothenhausler, Krauseneck, Stoll, & Schelling, 2004). Of patients mechanically ventilated for 48 hours or more, 57% needed caregiver assistance with activities of daily living and 16% remained unemployed due to disability at one year post ICU discharge (Chelluri et al., 2004). Five years after ICU discharge, acute respiratory distress syndrome (ARDS) survivors’ physical function remains below predicted normal values (Herridge, et al., 2011). Sedation Protocols and Guidelines Many factors such as age and severity of illness impact patient outcomes after ICU. Use of sedative medications, however, is a modifiable risk factor. The use of sedation protocols to guide sedation administration or daily sedative interruption has resulted in shorter duration of mechanical ventilation and ICU length of stay (Brook et al., 1999; De Jonghe et al., 2005; Kress et al., 2000; Robinson et al., 2008; Skrobik et al., 2010) . Patients randomized to a daily interruption of sedative medications had fewer complications during ICU and fewer symptoms of post-traumatic stress disorder after critical illness compared to a control group suggesting that decreased sedative exposure may improve patient outcomes (Kress et al., 2003; Schweickert, Gehlbach, Pohlman, Hall, & Kress, 2004). 4 Sedation guidelines recommend titration of medication to patient specific sedation levels with the recommended goal in both guidelines being a calm, easily aroused patient (Jacobi et al., 2002; Shapiro et al., 2007). Additionally the guidelines recommend regular assessment of sedation, pain, and delirium; the use of sedation algorithms or protocols to guide sedation administration; and intermittent dosing of medications with the use of continuous infusion only if there is inadequate symptom control (Jacobi et al., 2002; Shapiro et al., 2007). Sedation Practice Despite national guidelines and research evidence of the benefits of sedation protocols, changes to sedation practice in the ICU lag behind. Based on survey reports, only 23 to 71% of ICUs have sedation protocols in place (Egerod, Christensen, & Johansen, 2006; Martin, Franck, Sige, Weiss, & Spies, 2007; O'Connor, Bucknall, & Manias, 2010; Patel et al., 2009; Salluh et al., 2009; Tanios, de Wit, Epstein, & Devlin, 2009). These survey numbers reflect reports of protocols being in place, they do not reflect adherence or implementation. In one study, no sedation assessment was documented for 45% of patients receiving sedative medication and 41% of patients were in a deep state of sedation (unresponsive or moving only to physical stimulation) on day six of mechanical ventilation (Payen et al., 2007). Even when sedation goals are specified, the actual depth of sedation is greater than desired (Martin, Franck, Fischer, & Spies, 2006). Reasons for limited adoption or adherence to sedation protocols may include perceptions of health care workers. Reported barriers to use of sedation protocols 5 include lack of a physician order, lack of nursing acceptance, and a preference for more control over sedation than allowed within a protocol (Tanios et al., 2009). Attitudes of health care workers may also play a role. Nurses rated their own experience or intuition as more important than research based knowledge when making decisions about sedation (Randen & Bjork, 2010) and 81% of nurses agreed that sedation was necessary for patient comfort (Guttormson, Chlan, Weinert, & Savik, 2010). Patients’ Evaluation of Mechanical Ventilation Health care workers concern about mechanically ventilated patient comfort is not unwarranted. Patients report experiencing many physical and psychological stressors during mechanical ventilation. Physical discomfort. Many patients experience physical discomfort while on the ventilator with 58% of patients in one study reporting the level of general discomfort as distressful (Granja et al., 2005). Physical discomfort is most frequently related to the endotracheal tube and suctioning, but is also associated with other sources including procedures, immobility, thirst, and fatigue (Cochran & Ganong, 1989; Gries & Fernsler, 1988; Jablonski, 1994; Johnson & Sexton, 1990; Johnson, St John, & Moyle, 2006; Logan & Jenny, 1997; Rattray, Johnston, & Wildsmith, 2005; Rotondi et al., 2002; Wunderlich, Perry, Lavin, & Katz, 1999). Emotional and psychological distress. Patients attribute emotional and psychological distress to a lack of information and fear regarding the outcome of their illness ( Jablonski, 1994; Johnson & Sexton, 1990; Johnson et al., 2006; Wunderlich et al., 1999). Altered cognition, disorientation to person, place, and time, lack of recall of 6 ICU events, nightmares, and feeling dislocated from reality contribute to psychological distress (Bergbom-Engberg & Haljamae, 1993; Cochran & Ganong, 1989; Granja et al., 2005; Hafsteindottir, 1996; Hupcey & Zimmerman, 2000; Jablonski, 1994; Jenny & Logan, 1996; Johnson et al., 2006). Episodes of terror or panic are reported by 32% of patients in one study and are rated as the most bothersome of all ICU memories (Rotondi et al., 2002). Communication. Communication difficulties are reported as moderately to extremely bothersome by 57 to 77% of patients (Rotondi et al., 2002; Samuelson, Lundberg, & Fridlund, 2007). Patients are frustrated by their inability to communicate and a lack of received information (McKinley, Nagy, Stein-Parbury, Bramwell, & Hudson, 2002; Wunderlich et al., 1999). Patients’ inability to speak and clearly communicate needs contributes to negative emotions of anger and sadness (Bergbom- Engberg & Haljamae, 1993; Gries & Fernsler, 1988; Hafsteindottir, 1996; Jablonski, 1994; Johnson et al., 2006; Logan & Jenny, 1997). Non-pharmacologic interventions. Physical and interpersonal non- pharmacologic interventions have been identified by patients as helping to alleviate symptoms. Patients express a need for personalization of care and more information on their progress and plan of care (Hafsteindottir, 1996; Hupcey & Zimmerman, 2000; Jablonski, 1994; Logan & Jenny, 1997; McKinley et al., 2002; Wunderlich et al., 1999). More information helps to decrease their anxiety, stress, and feelings of vulnerability (Gries & Fernsler, 1988; Hafsteindottir, 1996; Hupcey & Zimmerman, 2000; Jablonski, 1994; Logan & Jenny, 1997; Wunderlich et al., 1999). Family presence and 7 involvement with care are frequently mentioned by patients as an effective intervention for improving their experience (Hafsteindottir, 1996; Jablonski, 1994; Logan & Jenny, 1997). Physical interventions identified by patients as reducing distressful symptoms are allowing patients to participate in care, limiting any activity restrictions, and providing frequent oral care (Gries & Fernsler, 1988; Jablonski, 1994; Logan & Jenny, 1997). Impact of Sedation and Level of Arousal on Memories Sedation and level of arousal over the course of mechanical ventilation impacts patients’ memories of ICU. Prolonged sedation (Jones et al., 2007) and higher sedative exposure averaged over the entire time of mechanical ventilation (Weinert & Sprenkle, 2008) are associated with increased delusional memories of ICU. Patients with greater proportions of time awake during mechanical ventilation have increased factual recall of ICU (Samuelson, Lundberg, & Fridlund, 2006; Weinert & Sprenkle, 2008). Patients with delusional memories and no factual memories of intensive care had a greater proportion of time at lower levels of arousal during mechanical ventilation (Samuelson et al., 2006). Patients experience multiple physical and psychological symptoms related to mechanical ventilation. Although sedation may be indicated at times, it is associated with multiple adverse outcomes. Despite increasing awareness of the adverse effects, changes in practice have lagged behind the clear evidence of the harm of excess sedation. Protocols shown to decrease complications from sedative therapy are not widely implemented. In practice, patients are often more deeply sedated than targeted 8 sedation levels increasing the risk of complications (Martin et al., 2006). Studies of the effect of sedation on patients’ memories have focused on the type of memory or count of memories (Jones et al., 2007; Samuelson et al., 2006; Weinert & Sprenkle, 2008). The effectiveness of sedation for management of patient symptoms during mechanical ventilation has not previously been evaluated with patient self-assessment. Therefore, the efficacy of sedation for relieving patient’s subjective symptoms during mechanical ventilation is primarily based on health care workers’ interpretations. Non- pharmacologic interventions that may reduce symptoms thereby alleviating or decreasing the need for sedation have not been systematically evaluated. The purpose of this study is to examine the impact of sedation on the patients’ evaluation of mechanical ventilation and to identify effective non-pharmacologic interventions for symptom management. The specific aims of this study are to: (1) describe mechanically ventilated patients’ memories of the intensive care unit; (2) describe the trajectory or pattern of sedation during mechanical ventilation; (3) determine the relationship between patients’ recall and evaluation of mechanical ventilation and sedation trajectories during mechanical ventilation; and (4) identify non- pharmacologic interventions that patients find effective for symptom management. This study contributes to the evidence base for the assessment, evaluation, and improvement of sedation administration practices and inclusion of non-pharmacologic interventions in symptom management for mechanically ventilated patients. 9 CHAPTER II Literature Review The patient experience of mechanical ventilation includes many physical and psychological stressors. To manage symptoms during mechanical ventilation, the majority of patients routinely receive sedative medications. These medications are often associated with multiple complications. Methods of sedation administration that minimize complications have been investigated and sedation guidelines have been developed but changes in sedation practices have not been uniformly implemented outside of research studies. This review summarizes the current literature in two main areas: (a) patient’s evaluation of mechanical ventilation and (b) sedation of mechanically ventilated patients. The following topics related to sedation of mechanically ventilated patients are described: pharmacology of sedative medications, complications associated with sedative medications, impact of different sedation methods, sedation administration guidelines and recommendations, and the current state of sedation administration in critical care. Search strategies for each section of the literature review are described in Appendix A. Patients’ Evaluation of Mechanical Ventilation Across studies of the patient experience of mechanical ventilation, common themes emerged. Patients report physical, psychological, and emotional discomfort and frustration with communication. Details of individual studies included in the section are presented in Table B1 (Appendix B). 10 Physical discomfort. Physical discomfort was experienced by many patients while on the ventilator. In one study, 58% of patients reported the level of general discomfort during ICU as distressful (Granja et al., 2005). The number of patients that recalled experiencing pain varied across studies ranging from 12% to 64% of patients (Granja et al., 2005; Li & Puntillo, 2006; Magarey & McCutcheon, 2005; van de Leur et al., 2004). Discomfort was primarily associated with the endotracheal tube (ETT) (Cochran & Ganong, 1989; Gries & Fernsler, 1988; Jablonski, 1994; Johnson & Sexton, 1990; Johnson et al., 2006; Logan & Jenny, 1997; Wunderlich, Perry, Lavin, & Katz, 1999). However, patients reported pain or discomfort from multiple sources including needle sticks, urinary catheters, nasogastric tubes, thirst, immobility, fatigue, and procedures (Bergbom-Engberg & Haljamae, 1993; Cochran & Ganong, 1989; Granja et al., 2005; Gries & Fernsler, 1988; Hafsteindottir, 1996; Jablonski, 1994; Johnson & Sexton, 1990; Li & Puntillo, 2006; Logan & Jenny, 1997; Magarey & McCutcheon, 2005; Rotondi et al., 2002; van de Leur et al., 2004). Descriptions of irritation from the ETT varied from mild discomfort to pain with 58 to 79% of those who remembered the ETT reporting they were moderately or extremely bothered by it (Rotondi et al., 2002; Samuelson et al., 2007). Suctioning of the ETT was also described as a source of distress (Granja et al., 2005; Gries & Fernsler, 1988; Johnson & Sexton, 1990; Johnson et al., 2006; Logan & Jenny, 1997). Suctioning caused some patients to feel short of breath, gag, cough, or experience pain (Gries & Fernsler, 1988; Johnson & Sexton, 1990). Additionally, moderate dyspnea was reported by 40% of patients in one study (Li & Puntillo, 2006). 11 Immobility, body position, and physical restrictions were also a source of patient distress (Granja et al., 2005; Gries & Fernsler, 1988; Jablonski, 1994; Johnson & Sexton, 1990; Logan & Jenny, 1997). Although the use of restraints was an identified factor, even unrestrained patient experienced distress from immobility or activity restrictions (Gries & Fernsler, 1988). Profound fatigue was experienced by many patients with 68% to 80% of patients in two studies rating their fatigue from moderate to severe (Hafsteindottir, 1996; Higgins, 1998; Li & Puntillo, 2006). Sleep disturbances were commonly reported and were attributed to environmental noise, lights, activity on the unit, fear, and anxiety (Bergbom-Engberg & Haljamae, 1993; Cochran & Ganong, 1989; Hafsteindottir, 1996). Emotional and psychological distress. Patients experienced many emotional and psychological symptoms and stressors during ICU including hallucinations, altered cognition, nightmares, and paranoia (Hafsteindottir, 1996; Hupcey & Zimmerman, 2000; Jablonski, 1994; Jenny & Logan, 1996; Johnson & Sexton, 1990; Johnson et al., 2006; Li & Puntillo, 2006; Logan & Jenny, 1997; Wunderlich et al., 1999). Patients also recall experiencing fear and anxiety during mechanical ventilation with 53% of patients in one study reporting moderate to severe anxiety (Bergbom-Engberg & Haljamae, 1993; Hafsteindottir, 1996; Jablonski, 1994; Johnson & Sexton, 1990; Johnson et al., 2006; Li & Puntillo, 2006; Logan & Jenny, 1997; Wunderlich et al., 1999). Patients attribute their anxiety and fear to multiple sources including a lack of information, uncertainty about the outcome of their illness, fear of an incomplete recovery, and fear 12 of death (Granja et al., 2005; Jablonski, 1994; Johnson & Sexton, 1990; Johnson et al., 2006; Wunderlich et al., 1999). Additionally, for patients who recalled the ETT, 41 to 59% reported experiencing fear or anxiety related directly to the ETT or ventilator (Granja et al., 2005; Rotondi et al., 2002). Anxiety to the level of terror or panic was reported by 27 to 32% of patients with recall of ICU (Rotondi et al., 2002; Samuelson et al., 2007). This high degree of anxiety was the most bothersome of memories for patients in these two studies with 90 to 93% reporting they were bothered moderately to extremely by episodes of terror or panic (Rotondi et al., 2002; Samuelson et al., 2007). Patients also reported feelings of vulnerability related to difficulty communicating, lack of knowledge or understanding, use of restraints, and reliance on technology (Johnson et al., 2006; Rotondi et al., 2002; Wunderlich et al., 1999). This vulnerability may have contributed to patients’ reports in one study where feelings of not being in control were moderately to extremely stressful for 46% of those with memories of ICU (Rotondi et al., 2002). Altered cognition was commonly mentioned. Patients described hallucinations, nightmares, feelings of paranoia and persecution, and disorientation to person and place (Bergbom-Engberg & Haljamae, 1993; Hafsteindottir, 1996; Hupcey & Zimmerman, 2000; Jenny & Logan, 1996; Johnson et al., 2006; Rundshagen, Schnabel, Wegner, & Schulte am Esch, 2002). Indeed, of the 17 to 26% of patients in two studies recalling nightmares in ICU, 85 to 88% were bothered moderately to extremely by them (Rotondi et al., 2002; Samuelson et al., 2007). Hallucinations were reported as a source of discomfort by 32% of patients in a third study (van de Leur et al., 2004). Patients felt 13 dislocated from the real world with difficulty determining reality from unreality (Johnson & Sexton, 1990; Johnson et al., 2006). Additionally, patients experienced difficulty keeping track or marking the passage of time (Jablonski, 1994; Johnson & Sexton, 1990; Johnson et al., 2006; Logan & Jenny, 1997). This altered cognition may contribute to the inability to recall events of the intensive care unit which was distressful for patients (Cochran & Ganong, 1989; Hafsteindottir, 1996; Hupcey & Zimmerman, 2000; Jablonski, 1994). Patients expressed the need to fill in gaps in memory regarding their critical illness and actively worked to obtain information about their time in the ICU (Jablonski, 1994). Communication. Frustration with communication while on mechanical ventilation was twofold: inability to communicate and lack of information received (McKinley et al., 2002; Wunderlich et al., 1999). Patients were distressed by their inability to clearly communicate their needs due to the inability to speak (Bergbom- Engberg & Haljamae, 1993; Gries & Fernsler, 1988; Hafsteindottir, 1996; Jablonski, 1994; Johnson et al., 2006; Logan & Jenny, 1997). Communication difficulties were stressful and frustrating with 40 to 77% of patients reporting communication as moderately to extremely bothersome or difficult (Granja et al., 2005; Happ et al., 2011; Patak, Gawlinski, Fung, Doering, & Berg, 2004; Rotondi et al., 2002; Samuelson et al., 2007). Impaired communication led to negative emotions of anger and sadness (Hafsteindottir, 1996). Frustrated patients sometimes gave up trying to make their needs known or restricted communication to only essential information (Gries & Fernsler, 1988; Hafsteindottir, 1996; Patak et al., 2004). 14 Non-pharmacologicInterventions. These studies did not address patients’ evaluation of sedation to alleviate symptoms. However, physical and interpersonal non- pharmacologic interventions were identified. To decease anxiety and stress patients expressed a need for family presence; more information on their progress and the plan of care; and personalization of care (Hafsteindottir, 1996; Hupcey & Zimmerman, 2000; Jablonski, 1994; Logan & Jenny, 1997; McKinley et al., 2002; Wunderlich et al., 1999). Patients wanted to understand their illness and progress, know the names of health care personnel, receive reassurance, and be reoriented to their surroundings and situation (Hafsteindottir, 1996; Higgins, 1998; Hupcey & Zimmerman, 2000). Provision of information was seen as a way to decrease patients’ stress, anxiety, and feelings of vulnerability (Hafsteindottir, 1996; Wunderlich et al., 1999). Patience exhibited by the nurse and allowing time for the patient to communicate were additional actions seen as beneficial by patients (Hafsteindottir, 1996; Hupcey & Zimmerman, 2000; Patak et al., 2004; Wunderlich et al., 1999). The physical presence of their nurse provided patients with a feeling of safety (Karlsson & Forsberg, 2008). Communication by the nurse that included details either about the nurses’ own life or events happening outside of the hospital helped to link the patient to the outside world beyond ICU (Karlsson & Forsberg, 2008). Support of nurses was seen as the most important aspect of nursing care by patients in one study, actions seen as supportive by patients included relieving their fear, providing attention, providing comfort, and providing a sense of security (Hofhuis et al., 2008). 15 An interesting dichotomy was found in one study between the patients’ report of their overall satisfaction and their report of specific experiences. Ninety-four percent of the sample described the overall ICU environment as calm and friendly (Granja et al., 2005). However a significant number of patients also recalled experiences as stressful such as: pain (64%), suctioning of the ETT (81%), the nasogastric tube (75%), fear of dying and uncertainty about the future (64%), and general discomfort (58%) (Granja et al., 2005). Even though research methods in the reviewed literature have changed over time to include more quantitative measures such as questionnaires, the patient’s reports of their experience remains markedly similar for physical, psychological, and communication difficulties. This stability of patient’s reported experiences over time suggests that we have yet to identify or utilize successful interventions and that our current use of medications such as sedative, narcotics, and anxiolytics may not be improving the patients’ experiences of mechanical ventilation as much as ICU practitioners believe. None of these studies evaluated the patient experience in relation to sedative medications received during mechanical ventilation therefore the efficacy of sedation for alleviating patients’ subjective symptoms is currently based on health care workers interpretations and assumptions. Sedation of Mechanically Ventilated Patients It is common practice to administer intravenous sedative medications to mechanically ventilated patients to enhance patient comfort, to ensure patient safety, to induce patient amnesia, to decrease anxiety and agitation, to decrease oxygen 16 consumption, and to prevent ventilator dysynchrony (Brush & Kress, 2009; Gehlbach & Kress, 2002; Jacobi et al., 2002; Mehta et al., 2006; Rhoney & Murry, 2003; Weinert et al., 2001). Although these medications may be effective in achieving these goals, they can contribute to complications both during and after ICU. To minimize complications different methods of sedation administration including sedation protocols or algorithms, daily interruption of sedation (DIS), and analgosedation have been investigated and guidelines developed. However adoption of guidelines and changes in practice are limited outside of research studies. This section of the literature review presents an overview of commonly utilized intravenous (IV) medications and reviews the research regarding complications related to sedation. Current research, recommendations, and guidelines for sedation administration are described. Finally, sedation practice recommendations and guidelines are contrasted with the current state of ICU practice. Medications. The most common sedative and pain medications administered to mechanically ventilated patients are opioids, benzodiazapines, and propofol (Arroliga et al., 2005; O'Connor et al., 2010; Patel et al., 2009; Payen et al., 2007; Randen & Bjork, 2010). Individual drug information regarding onset, duration, and specific side effects is presented in Table 1. It is important to note that the table provides average onset and duration, however multiple factors may impact critically ill patients responses to medications including: age, altered liver and renal function, effects of underlying disease, polypharmacy, slowed metabolism, and increased volume of distribution (Siegel, 2003). 17 Table 1 Commonly Used Intravenous Sedatives and Analgesic Medications Drug Name Drug Class Onset * Duration*a Fentanyl opioid analgesic 1-2 min. 10-120 min. Remifentanil opioid analgesic 1-3 min. 10-20 min. Morphine opioid analgesic 3-10 min. 1-4 hrs. Hydromorphone opioid analgesic 3-10 min. 1-4 hrs. Midazolam benzodiazepine 1-10 min. 1-4 hrs. Lorazepam benzodiazepine 5-10 min. 6-8 hrs. Diazepam benzodiazepine 1-3 min. 30-60 min. Propofol hypnotic/anesthetic .5-5 min. 2-10 min. Dexmedetomidine alpha 2 adrenergic agonist/sedative immediate approx. 6 min. Haloperidol butyrophenone 2-5 min. 120 min. Note. * ranges are a compilation from the following sources: (Gehlbach & Kress, 2002; Sessler & Varney, 2008; Siegel, 2003);a with repeated doses, continuous infusion, or extended administration duration may increase (Arnold, Hollands, Skrupky, & Mice, 2010; Gehlbach & Kress, 2002; Siegel, 2003) Opioids. Opioids bind to receptors in the central nervous system (CNS) causing analgesia (Gehlbach & Kress, 2002; Gommers & Bakker, 2008; Siegel, 2003). As doses of opioids increase, sedation occurs (Gommers & Bakker, 2008). Due to the action in the CNS, opioids cause dose dependent respiratory depression (Gehlbach & Kress, 2002; Gommers & Bakker, 2008) which may be increased when opioids are given in conjunction with benzodiazepines (Gommers & Bakker, 2008). Other common side effects include constipation due to decreased GI motility, pruritis, and nausea and vomiting (Gehlbach & Kress, 2002; Siegel, 2003). Although hypotension with opioid administration may occur this is more common in patients who are hypovolemic 18 (Gehlbach & Kress, 2002; Siegel, 2003). Dependence and withdrawal symptoms may occur for patients receiving opioid medications for a prolonged period of time (Gommers & Bakker, 2008; Siegel, 2003). Morphine, hydromorphone, and fentanyl are the most frequent IV opioids utilized in the ICU although remifentanil is becoming more common. Benzodiazepines. Benzodiazepines interact with the gamma amino-butyric acid (GABA) receptors in the central nervous system (CNS) to produce anxiolysis, sedation, and amnesia (Gehlbach & Kress, 2002; Gommers & Bakker, 2008). Side effects include dose dependent respiratory depression and hypotension in hypovolemic patients (Gehlbach & Kress, 2002). Benzodiazepines and opioids in combination may have a synergistic effect that allows decreased doses of both medications (Hogarth & Hall, 2004; Siegel, 2003) but may also increase respiratory depression (Gehlbach & Kress, 2002). Parodoxical agitation may occur particularly in the elderly (Gehlbach & Kress, 2002; Siegel, 2003). Specific to high doses of lorazepam, polypropylene glycol (a carrier in IV solution) toxicity, characterized by acute tubular necrosis and metabolic acidosis, may develop (Gehlbach & Kress, 2002; Gommers & Bakker, 2008; Siegel, 2003). Propofol. Propofol is a lipid soluble IV anesthetic sedative hypnotic agent (Gehlbach & Kress, 2002) which produces mild anxiolysis and amnesia but no analgesia (Gehlbach & Kress, 2002; Siegel, 2003). It’s mechanism of action is unclear however it is believed to affect the GABA receptors in the CNS (Gehlbach & Kress, 2002). Effect of the drug is dose dependent—the sedative effect on patients ranges 19 from a mild decrease in responsiveness to non-responsiveness (Gehlbach & Kress, 2002). Hypotension is the most common adverse effect occurring particularly in hypovolemic or hemodynamically unstable patients (Gehlbach & Kress, 2002; Gommers & Bakker, 2008; Siegel, 2003). Hypertriglyceridemia can occur especially in patients receiving propofol for extended periods (Gehlbach & Kress, 2002; Siegel, 2003). Infection related to propofol infusion is possible but can be avoided with careful management of IV lines (Gehlbach & Kress, 2002; Siegel, 2003). Propofol related infusion syndrome (PRIS) is a rare but serious complication that is typically seen in patients receiving high doses of propofol (> 80 mcg/kg/min) for greater than 48 hours (Gehlbach & Kress, 2002; Gommers & Bakker, 2008). PRIS has a high mortality rate and is characterized by dysrhythmia, heart failure, metabolic acidosis, hyperkalemia, and rhabdomyolysis (Gehlbach & Kress, 2002; Gommers & Bakker, 2008). Dexmedetomidine. A newer drug which is increasingly seen in the ICU is dexmedetomidine, a centrally acting selective alpha-2 agonist that has anxiolytic and mild analgesic effects (Gehlbach & Kress, 2002; Gommers & Bakker, 2008; Hogarth & Hall, 2004). Dexmedetomidine allows patients to be sedated when undisturbed yet still easily awakened when aroused (Gehlbach & Kress, 2002; Hogarth & Hall, 2004). Side effects include bradycardia and hypotension due to the sympatholitic effects of the drug (Gehlbach & Kress, 2002; Siegel, 2003). Haloperidol. Haloperidol is used for sedation particularly in patients where delirium is suspected or confirmed or those who have paradoxical reactions to benzodiazepines (Siegel, 2003). Haloperidol is believed to act as a dopamine antagonist 20 primarily in the basal ganglia (Brush & Kress, 2009; Gehlbach & Kress, 2002). Side effects can include mild hypotension and extrapyramidal effects (Gehlbach & Kress, 2002). Prolonged QT intervals are a rare but serious complication that may result in torsade de pointes (Gehlbach & Kress, 2002; Siegel, 2003). Haloperidol can lower the seizure threshold and should be used with caution in patients at risk for seizures (Siegel, 2003). Complications Related to Sedative Administration Although these medications have some positive effects for patients including decreasing anxiety they are also associated with physical and psychological complications. Amount or duration of sedation is associated with a longer duration of mechanical ventilation, longer ICU stays, and more reintubations (Arroliga et al., 2005; Kollef et al., 1998; Kress et al., 2000). Increased duration of mechanical ventilation increases the incidence of pneumonia and airway damage; increases health care costs; and negatively impacts mobility and self-care ability (Douglas et al., 2002; Kollef et al., 1998; Ostermann et al., 2000). ICU acquired weakness (ICU-AW), a broad term that encompasses weakness developing during ICU caused by myopathies, neuropathies, or a combination of muscle and nerve involvement, occurs in 25 to 60% of ICU patients (Bercker et al., 2005; De Jonghe et al., 2002; Garnacho-Montero, Amaya-Villar, Garcia-Garmendia, Madrazo-Osuna, & Ortiz-Leyba, 2005) and contributes to prolonged rehabilitation and problems related to physical function at one year post ICU discharge (Leijten, Harinck- de Weerd, Poortvliet, & de Weerd, 1995; van der Schaaf, Beelen, & de Vos, 2004). 21 Although sedative medications have not been directly linked to the development of ICU-AW, one study identified duration of mechanical ventilation as a predictor of ICU- AW development (De Jonghe et al., 2002). However, other studies conversely conclude that ICU-AW weakness contributes to prolonged mechanical ventilation (Bercker et al., 2005; De Jonghe S., Sharshar, Outin, & Brochard, 2004; Garnacho-Montero et al., 2005). With the complexity of critically ill patients causality between duration of mechanical ventilation, sedation administration, and development of ICU-AW may be difficult to establish. Sedation of mechanically ventilated patients will however impact the implementation of rehabilitation early in the ICU course. Early mobilization has been proposed as a means to decrease severity or incidence of ICU-AW (De Jonghe, Lacherade, Sharshar, & Outin, 2009; Needham, 2008). Outcomes from early mobilization with mechanically ventilated patients have been promising, with patients in the early mobilization group having shorter hospital and ICU-LOS (Needham & Korupolu, 2010) and a greater likelihood of functional independence at hospital discharge (Schweickert et al., 2009). Early mobility strategies however cannot be successfully implemented unless a patient is alert and cooperative. Sedation may also impact neurocognitive and psychological outcomes during and after ICU. Delirium—defined as an acute change in mental status, cognition, and attention—occurs frequently, developing in 32 to 82% of patients in the ICU (Ely et al., 2004; Micek, Anand, Laible, Shannon, & Kollef, 2005; Ouimet, Kavanagh, Gottfried, & Skrobik, 2007). Development of delirium during ICU is associated with higher doses or continuous infusions of midazolam and fentanyl (Granberg Axèll, Malmros, 22 Bergbom, & Lundberg, 2002; Micek et al., 2005). Midazolam and lorazepam have been shown to be independent risk factors for the development of delirium (Pandharipande et al., 2006; Pandharipande et al., 2008). Delirious patients have longer durations of mechanical ventilation (Micek et al., 2005), longer ICU and hospital lengths of stay (Milbrandt et al., 2004; Ouimet et al., 2007), greater cognitive impairment at time of hospital discharge (Ely et al., 2004), and higher six month and one year mortality (Ely et al., 2004; Pisani et al., 2009). Up to 12 to 64% of patients experience depression after critical illness (Dowdy et al., 2009; Jackson et al., 2010; Myhren, Ekeberg, Toien, Karlsson, & Stokland, 2010). Studies with acute lung injury patients have found the incidence of depression to be associated with more days of sedation (Nelson et al., 2000) and a high average daily dose of midazolam (Dowdy et al., 2009). Post-traumatic stress disorder (PTSD) occurs in 5 to 63% of patients after intensive care (Jackson et al., 2007). Sedation, type or content of memories, and level of arousal during mechanical ventilation have all been linked to PTSD post ICU. Prolonged duration of sedation (Jones et al., 2007) and a higher total dose of lorazepam (Girard et al., 2007) have been associated with the development of PTSD symptoms post ICU. Additionally, sedation administration may indirectly impact the development of PTSD symptoms through the influence of sedation on patients’ memories of ICU. Delusional memories seem to develop more commonly when patients receive prolonged sedation (Jones et al., 2007) or higher sedative exposure (Weinert & Sprenkle, 2008). Development of delusional memories may also be impacted by the proportion of time at 23 different levels of arousal during mechanical ventilation. Patients with paranoid delusional memories of ICU had higher proportions of agitation and restlessness while on the ventilator while patients with delusional memories and no factual memories had a higher proportion of decreased arousal (unresponsive to arouse to touch) (Samuelson et al., 2006). The link between patients’ memories of ICU and development of PTSD symptoms is not yet clearly understood. Delusional memories of ICU are associated with an increased incidence of PTSD symptoms (Granja et al., 2008; Jones, Griffiths, Humphris, & Skirrow, 2001; Jones et al., 2007). It has been proposed that the presence of factual memories may have some protective effect against the development of PTSD (Jones, Griffiths, Humphris, & Skirrow, 2001), however this has not been consistent across other studies (Granja et al., 2008). Myhren et al. (2010) reported memories of pain and factual recall as independent predictors of PTSD symptoms. Additionally, Schelling et al. (1998) found an increased number of remembered traumatic events (respiratory distress, anxiety/panic, pain, nightmares) to be associated with increased PTSD symptoms. Although the presence of delusional memories was not directly measured in this study, nightmares were included and may have some crossover with delusional memories. The quality of memories may also play a role in development of PTSD symptoms. Blurred memories, frightening memories, and a low awareness of surroundings during ICU are associated with higher scores on both the Impact of Events Scale and Hospital Anxiety and Depression Scale at time of hospital discharge (Rattray, Crocker, Jones, & Connaghan, 2010). 24 One study has directly evaluated the association between level of arousal and development of PTSD symptoms post ICU. Weinert and Sprenkle (2008) found no relationship between sedative exposure and PTSD symptoms. However, there was a nonlinear relationship between PTSD and wakefulness with those at the highest and lowest levels of wakefulness having the lowest prevalence of PTSD symptoms. These physical and psychological complications of sedative therapy contribute to a prolonged recovery period. As many as 69% of patients suffer from depression and continue to have depressive symptoms at two years post critical illness (Hopkins et al., 2005; Nelson et al., 2000). Patients with PTSD symptoms after ICU have a lower health related quality of life in general health, social function, and mental health dimensions when compared to the general population (Kapfhammer et al., 2004). At five years after ICU discharge, patients have still not returned to normal levels of physical function (Herridge et al., 2011). The percentage of patients requiring discharge to a nursing home after ICU is increasing (Carson, Cox, Holmes, Howard, & Carey, 2006). Of patients mechanically ventilated for 48 hours or more, 57% needed caregiver assistance with activities of daily living and 16% remained unemployed due to disability at one year post ICU discharge (Chelluri et al., 2004). Patients randomized to a daily interruption of sedative medications had fewer complications during ICU and fewer post-traumatic stress symptoms after critical illness compared to a control group suggesting that decreased sedative exposure may improve patient outcomes (Kress et al., 2003; Schweickert et al., 2004). Multiple factors such as severity of illness, age, and the disease process contribute to 25 complications experienced by mechanically ventilated patients during and after ICU. However, sedation administration is a factor that can be modified to improve patient outcomes. Research of Sedation Methods. This section summarizes the findings from numerous studies that compare different methods of sedation and analgesia administration to mechanically ventilated patients including daily interruption of sedation (DIS), initiation of sedation protocols or algorithms, and analgosedation or analgesia based sedation. Twenty-nine studies were identified that met inclusion criteria (Appendix A). Details of each study are presented in Table B2 (Appendix B). Sedation Protocols or Algorithms. Twelve studies compared usual sedation administration to mechanically ventilated patients (physician directed or unrestrictive sedative regimens) to protocol directed sedation with sedation algorithms to guide medication administration and a targeted level of sedation using a sedation scale. Sedation scales used in these studies included: Ramsay, Richmond Agitation Sedation Scale (RASS), Sedation Agitation Scale (SAS), Cambridge, Adaption to Intensive Care Environment (ATICE), and Glasgow Coma (GCS-CP) (Table 2). Five studies explicitly reported incorporating pain scales into their protocol and two studies explicitly reported including delirium scales. Only two studies evaluating patient outcomes with sedation protocols were randomized controlled trials (RCTs). All other studies were a pre-post design. Nine of the studies found improved patient outcomes with the use of sedation protocols including: 26  more ventilator free days (Robinson et al., 2008)  shorter duration of mechanical ventilation (MV) (Brattebø et al., 2004; Brook et al., 1999; De Jonghe et al., 2005; Jakob et al., 2007; Mascia, Koch, & Medicis, 2000; Robinson et al., 2008; Skrobik et al., 2010)  shorter hospital length of stay (LOS) (Brook et al., 1999; De Jonghe et al., 2005; Quenot et al., 2007; Robinson et al., 2008; Skrobik et al., 2010)  shorter ICU-LOS (Brook et al., 1999; Mascia et al., 2000; Quenot et al., 2007; Skrobik et al., 2010). Additional benefits of protocol use included a lower tracheotomy rate (Brook et al., 1999), lower cumulative and average doses of sedatives (De Jonghe et al., 2005), increased successful extubations (Arias-Rivera et al., 2008), fewer pressure sores (De Jonghe et al., 2005), decreased resource utilization (Jakob et al., 2007), decreased direct drug costs (Mascia et al., 2000), and decreased subsyndromal delirium defined as patients that have positive features of delirium but score less than the threshold for a clinical diagnosis (Skrobik et al., 2010). Not all studies of sedation protocols have reported positive results. An RCT of a goal directed sedation protocol using SAS sedation scale versus usual care found no difference between groups for MV duration, ICU-LOS, hospital LOS, or tracheotomy rate (Bucknall, Manias, & Presneill, 2008). Another pre-post study reported an Table 2 Comparison of Sedation Scales Scale Scoring Total Score Range 1 unarousable 4 calm and cooperative/awakens easily 1 to 7 2 arouse to physical stimulation 5 agitated 3 arouse to verbal stimuli 6 very agitated 7 dangerously agitated Sedation-Agitation Scale (SAS) 1 awake/anxious/restless 4 asleep/brisk response to glabellar tap 2 awake/cooperative/tranquil 5 asleep/sluggish response to glabellar tap Ramsay Scale 3 responds only to commands 6 asleep/no response to glabellar tap 1 to 6 +4 combative -1 drowsy +3 very agitated -2 brief awakening +2 agitated -3 movement to voice +1 restless -4 movement to physical stimulation Richmond Agitation Sedation Scale (RASS) 0 alert and calm -5 unarousable -5 to +4 1 agitated 4 arouse to suctioning 2 awake 5 unarousable Cambridge Scale 3 arouse to voice 6 paralyzed 1 to 6 27 Table2 (continued) Scale Scoring Total Score Range Conciousness Domain (Awakeness and Comprehension) Awakeness > 0 eyes closed; no response Comprehension (score one point for each command followed) 1 eyes closed; response to pain open/close your eyes (1) 2 eyes open to strong pain open your mouth (1) 3 eyes open to light pain look at me (1) 4 eyes open to voice nod yes with head (1) 5 eyes open spontaneously close eyes and open mouth (1) Tolerance Domain (Calmness, Ventilator Synchrony, and Face Relaxation) Calmness …0 dangerously agitated Ventilator Synchrony (score one point for each) 1 agitated/no response commands no dysychrony inspiratory phase; no cough 2 agitated/follows commands respiratory rate < 30; no accessory muscles 3 calm Face Relaxation 0 permanent grimacing 1 severe provoked grimacing 2 moderate provoked grimacing Adaptation to the Intensive Care Environment (ATICE) 3 relaxed face 0 to 20 (higher scores indicate better adaptation to ICU environment) 28 29 Table 2 (continued) Scale Scoring Total Score Range Eye Opening Cough 1 none 1 none 2 to pain 2 to suction only 3 to speech 3 spontaneous weak 4 spontaneous 4 spontaneous strong Response to Nursing Procedures Respiration 1 none 1 no respiratory efforts 2 non-purposeful extension 2 respiration against ventilator 3 non-purposeful flexion 3 intermittent triggering ventilator 4 purposeful 4 spontaneous respiration Glasgow Coma Scale modified by Cook and Palma (GCS-CP) 5 obeys comands 5 to command 4 to 18 0 unresponsive 4 restless and cooperative 1 responds to pain 5 agitated 2 responds to voice/touch 6 dangerously agitated Motor Activity Assessment Scale (MAAS) 3 calm and cooperative 1 to 6 Note.Source Arias-Rivera et al. (2008); Quenot et al. (2007); Stawicki (2007); SIMV: spontaneous intermittent mandatory ventilation 30 increased duration of MV and ICU-LOS and a trend toward more tracheotomies post initiation of a protocol targeting a Ramsay score of 3 (patient responsive to commands) (Elliott, McKinley, Aitken, & Hendrikz, 2006). MacLaren et al. (2000) reported a longer duration of sedation and a trend toward a longer duration of MV post protocol initiation however patients on protocol were also more often at a Ramsay Score of 4 (asleep, arouse easily) and exhibited less agitation, pain, restlessness, and anxiety. Of the three studies reporting no improvement or poorer outcomes from protocol initiation, two (Bucknall et al., 2008; Elliott et al., 2006) were conducted in Australia suggesting that cultural variation in intensive care units may be a factor to consider when assessing and implementing a sedation protocol. It is also worth noting that in the study by Elliot et al. (2006) in only 18% of audits did the sedation scale documentation and medication administration comply with guidelines. Adherence is an important aspect. In a unit with a sedation protocol in place, the addition of daily pharmacist rounds to evaluate MV patients receiving continuous sedation and provide recommendations to increase protocol adherence resulted in a decreased MV duration, ICU-LOS, and hospital LOS (Marshall, Finn, & Theodore, 2008). Similar results were found with a nursing education intervention involving daily rounds on mechanically ventilated and sedated patients (Southworth et al., 2008). None of the studies reviewed reported differences in adverse events with protocol versus usual care. Potential adverse events measured included self extubation, mortality, reintubation, and acquired organ dysfunction (Brattebø et al., 2004; Brook et al., 1999; Bucknall et al., 2008; De Jonghe et al., 2005; Quenot et al., 2007). 31 The studies above all incorporated sedation scales within sedation algorithms or protocols to guide medication administration. Other investigators have evaluated the implementation of sedation or pain scales without concurrent initiation of a medication protocol with some positive results. The introduction of the RASS sedation scale and two pain scales (numeric rating scale (NRS) and behavioral pain scale (BPS)) in a French medical ICU resulted in a shorter duration of MV post implementation (Chanques et al., 2006). Similarly, Payen et al. (2009) reported an independent association of pain assessment with a shorter duration of mechanical ventilation in a post hoc analysis of a large multi-site study conducted in France and Luxembourg. In contrast the institution of RASS and BPS scales in an Australian general ICU resulted in no change to MV duration (Williams et al., 2008). Daily Interruption of Sedation (DIS). Seven studies investigated the impact of daily interruption of sedation in isolation or in combination with sedation protocols. Daily interruption of sedatives (DIS) when compared to usual sedation as prescribed by the ICU team was shown to reduce duration of MV and ICU-LOS in one study (Kress et al., 2000) while another study found no significant difference (Anifantaki et al., 2009). Additionally, the DIS group in the study by Kress et al. had significantly fewer diagnostic tests to assess mental status and more patients were discharged to home (59% vs 40%) although the latter difference was not statistically significant. Results from the addition of DIS to sedation protocols or targeted sedation have also been mixed. One study reported trends toward shorter duration MV and ICU-LOS when DIS and a protocol was used in combination (Stewart, Castillo, Laine, Herlihy, & 32 Warren, 2006) while another small (n = 65) feasibility study reported no difference between groups (Mehta et al., 2008). In a comparison of protocol with a spontaneous breathing trial (SBT) and a protocol with SBT in combination with DIS, the DIS group had more ventilator free days and shorter ICU and hospital LOS (Girard et al., 2008). Only one study favored the protocol without DIS group with the protocol only group having a shorter duration MV and ICU and hospital LOS (de Wit, Gennings, Jenvey, & Epstein, 2008). Additionally, more agitation and higher mortality were reported for the protocol plus DIS group (de Wit et al., 2008). The Data Safety Monitoring Board could not determine a common etiology for mortality in the DIS group and the DIS group’s Sequential Organ Failure Assessment scores, a measure of severity of illness, did not improve as quickly as the control group’s despite being equal at baseline (de Wit et al., 2008). This study was terminated early due to the significantly decreased duration of MV in the control group not due to mortality. With the exception of the study by De Wit et al. (2008), there were no differences in adverse events between groups in the reviewed studies. AnalgoSedation. Five studies reviewed investigated analgosedation or analgesia based sedation versus conventional combination sedation and analgesia. Analgesia based intervention protocols were all similar with the upward titration of analgesia first and sedatives added only as necessary for continued agitation. Three studies used remifentanil as the only analgesia in the intervention group (Breen et al., 2005; Park, Lane, Rogers, & Bassett, 2007; Rozendaal et al., 2009) while other studies used either morphine alone (Strøm, Martinussen, & Toft, 2010) or a combination of fentanyl, morphine, or 33 remifentanil (Egerod, Jensen, Herling, & Welling, 2010). Two studies using remifentanil showed a decrease in duration of MV (non-significant in one study) and a trend toward shorter ICU-LOS (Breen et al., 2005; Rozendaal et al., 2009). The morphine based analgosedation protocol had patients with more ventilator free days, shorter ICU and hospital LOS, and lower delirium rates when compared to targeted sedation with propofol and DIS in combination (Strøm et al., 2010). These differences occurred despite 18% of the intervention group receiving Propofol at some point during MV (Strom, 2010). Although an analgesia first protocol introduced in a neurological ICU did not impact duration of MV or duration of sedation, the proportion of Ramsay scores at level 2-3 (oriented, cooperative) were higher and the proportion of Ramsay scores of 1 (agitated) were lower in analgosedation group. Pain intensity scores were also decreased (Egerod et al., 2010). Similarly, in the study by Park et al. (2007) patients in the analgesia based group spent a greater percentage of time at an awake or easily arousable level of sedation. The majority of studies found improved patient outcomes with sedation practice changes regardless of the method of change: DIS or sedation protocols. Results from analgosedation and sedation protocols with DIS in combination did not consistently show improvement in patient outcomes but warrant further investigation. Sedation Guidelines. In response to the increasing awareness of the potential complications related to sedation and results of research on sedation methods, numerous articles have been written with recommendation for the sedation of mechanically ventilated patients. This section reviews commonalities of these sedation 34 recommendations and guidelines. Although specifics may vary, some general areas of consensus are apparent:  Medications should be titrated to a specific targeted sedation level or goal (Bateman & Grap, 2003; Gehlbach & Kress, 2002; Kress, Pohlman, & Hall, 2002; Sessler & Varney, 2008; Siegel, 2003)  The ideal sedation goal or target is a patient whom is calm, cooperative, and easily arousable or alert (Hogarth & Hall, 2004; Siegel, 2003; Yagan, White, & Staab, 2000)  Recommend use of sedation protocols or algorithms and DIS either in combination or separately (Bateman & Grap, 2003; Dotson, 2010; Fuchs & Von Rueden, 2008; Hogarth & Hall, 2004; Kress et al., 2002; Sessler & Varney, 2008; Siegel, 2003)  Pain should be treated first before starting sedative medications (Hogarth & Hall, 2004; Sessler & Varney, 2008). Two sedation guidelines have been published. One by the Society of Critical Care Medicine (SCCM) in 2002 and the other by the Inflammation and Host Response to Injury Large Scale Collaborative Research Project in 2007 (Jacobi et al., 2002; Shapiro et al., 2007; Shapiro et al., 2007). Although exact dosing and medication recommendations vary, these two guidelines share many features. Both recommend titrating medications to a patient specific sedation level. The recommended goal in both guidelines is a calm patient whom is alert or easily aroused although the goal should be individualized to each patient as needed. Additionally, both protocols recommend reassessment of the sedation 35 target daily (Jacobi et al., 2002; Shapiro et al., 2007). Regular assessment and documentation of sedation, pain, and delirium and the use of sedation algorithms or protocols to guide medication administration is another shared recommendation. Both guidelines recommend intermittent dosing of medications initially with the use of continuous infusion only if there is inadequate symptom control. One difference is in the emphasis that the SCCM guideline places on pain control, advocating for analgesia and pain control first before initiating any sedative medication (Jacobi et al., 2002). The guideline by Shapiro et al. (2007) does not make this distinction but appears to advocate for the correct identification of pain, anxiety, or delirium and the correct treatment based on this assessment. Table 3 presents a side by side comparison of the two protocols. Sedation Practice. Despite clear evidence in the research literature and consensus in guidelines, changes in sedation practice in the ICU lag behind recommendations. Following is a review of the current state of sedation administration practice in the ICU. The number of patients receiving continuous sedation during mechanical ventilation is increasing from 39.7% in 2001 to 66.7% in 2007 (Wunsch et al., 2009). Although use of sedation protocol appears to be increasing (Martin et al., 2007), based on reports of surveys only 23 to 71% of ICUs have a sedation protocol in place and 31 to 76% have a protocol of DIS with the highest percentages for both reported in North America (Table 4) (Egerod et al., 2006; Martin et al., 2007; O'Connor et al., 2010; Salluh et al., 2009; Tanios et al., 2009). Reported use of sedation assessment scales is somewhat higher ranging from 44 to 88% (Egerod et al., 2006; Martin et al., 2007; O'Connor et al., 2010; Salluh et al., 2009; Tanios et al., 2009). It is important to Table 3. Comparison of Sedation Guidelines Society of Critical Care Medicine Jacobi et al. (2002) Inflammation and Host Response to Injury Large Scale Collaborative Research Project (Shapiro et al. 2007) assessment sedation target established for each patient; reassess goal/target daily; look for correctable causes of agitation/ discomfort; regular assessment of pain and sedation level sedation target established for each patient; reassess goal daily; pain and sedation assessment Q15 minutes; until desired level reached then at least Q4h pain scale Pain Scale: NRS or Behavioral pain scale Pain Scale: not specified sedation scale Three mentioned: SAS, MAAS, VICS RASS: Q15 minutes until stable than at least Q4h optimal level of sedation defined for each patient but a "common level is a calm patient that can be easily aroused." (p.124) RASS 0 to -2: maintains eye contact for ten seconds/follows simple commands or is alert and calm Sedation management Sedation only after adequate analgesia and treatment of reversible causes of agitation; titrate to established target/sedation goal; consider DIS or systematic tapering of sedation; use protocol to guide sedation For patients at a RASS of -3 (unresponsive or no eye contact) then DIS until patient able to follow simple commands/make eye contact. Stop DIS if patient agitated/uncomfortable; use protocol to guide sedation Pain Medications fentanyl, hydromorphone, or morphine/ scheduled or continuous infusion preferred over PRN Fentanyl initially as bolus; if pain control not achieved after three hours start continuous infusion Sedative Medications acute agitation: midazolam; ongoing sedation: lorazepam or propofol; intermittent dosing initially for lorazepam; if doses > Q2h consider infusion Propofol for MV < 48h or frequent neurological assessment; Intermittent Lorazepam for MV > 48h; start infusion if sedation goal not met in 24 hours Management of Delirium Assessment: CAM-ICU; Medication: Haloperidol IVP PRN then scheduled Assessment not specified; Medication: haloperidol PRN for 6 hours if goal unmet start haldol Q6h Notes.RASS: Richmond Agitation Sedation Scale; NRS: numeric rating scale; DIS: daily sedative interruption; SAS: Sedation Agitation Scale; MAAS: Motor Activity Assessment Scale; VICS: Vancouver Interaction and Calmness Scale; MV: mechanical ventilation 36 37 note that these survey numbers reflect reports of protocol, assessment scales, and DIS being in place, they do not reflect adherence or implementation. A few authors have tried to capture actual use of protocols, assessments, and DIS. When DIS is assessed based on utilization in 75% or more of patients, percentages decreased to 23% (O'Connor et al., 2010; Tanios et al., 2009). In one study, for 45% of patients receiving sedative medications, no sedation assessment was documented (Payen et al., 2007). Furthermore, 57% of patients were rated as being in a deep state of sedation (unresponsive or moving only in response to physical stimulation) on day two of mechanical ventilation. This number decreased over time but was still at 41% on day six (Payen et al., 2007). Even when sedation endpoints are specified, the actual depth of sedation is often greater than desired (Martin et al., 2006). In a study where one third of nurse observations ranked patients as minimally arousable or unarousable, only 2.6% of nurse ratings indicated the patient was oversedated (Weinert & Calvin, 2007). A similar discrepancy was found between what 75% of nurses reported as the ideal level of sedation (interactive and calm) and what they report as the usual level of sedation of their patients: responsive to pain (26%) or responsive to verbal stimuli (30%) (O'Connor et al., 2010). Table 4. Reported Use of Sedation Protocol, Daily Sedative Interruption, Sedation Scales Citation* Protocol DIS Scale Country Salluhet et al. (2009) 52.7% 51.2% 88.3% Brazil Patel, 2009 71.0% 76.0% 88.0% North America Tanios, 2009 64.0% 40.0% NR US Egerod, 2006 23.0% 31.0% 44.0% Denmark Martin, 2007 46.0% NR 51.0% Germany O'Connor, 2010 54.0% 62.0% 72.0% Australia Mehta , 2006 29.0% 40.0% 49.0% Canada Notes. * all surveys 38 Sedation practice continues to lag behind sedation research findings. Reasons for this may include perceived barriers to protocols and attitudes of health care workers. Reported barriers to use of sedation protocols include a lack of physician order, lack of nursing acceptance, and preference for more control over sedation than allowed within a protocol (Tanios et al., 2009). Barriers to use of DIS that were most commonly reported include the possibility of respiratory compromise, lack of acceptance by nurses, and fear of patient removal of medical devices (e.g. ETT, IV lines) (Tanios et al., 2009). Attitudes of health care workers may also play a role in protocol use. Eighty-one percent of nurses agreed that sedation was necessary for patient comfort (Guttormson et al., 2010) and nurses rated their own experience or intuition as more important than research based knowledge when making decisions about sedation (Randen & Bjork, 2010). Conclusion Sedative medications are commonly administered to patients on the ventilator in the ICU. The use of sedation is a modifiable risk factor for both short and long term complications of critical illness. However, sedation protocols that improve patient outcomes have not been widely adopted and utilized. This may be in part due to a continued belief of health care workers that sedation is necessary for symptom management during mechanical ventilation. Although patients do report multiple physical and psychological stressors during mechanical ventilation, the efficacy of sedation for ameliorating these subjective symptoms has not been systematically evaluated from the patient’s perspective. 39 CHAPTER III Research Methods The purpose of this descriptive, associational study was to (1) describe patients’ memories of ICU; (2) determine the relationship between patients’ recall and evaluation of mechanical ventilation and the trajectory or pattern of sedation during mechanical ventilation; and (3) describe patients’ evaluations of non-pharmacologic interventions for reducing symptoms associated with mechanical ventilation. This chapter will describe the conceptual framework of the study, the specific aims, the study setting and sample, study variables and measurement, and the analysis plan. Conceptual Framework A conceptual model was developed for this study based on the research literature. The patient’s recall and assessment of mechanical ventilation is influenced by multiple factors including severity of illness, age, sedation, duration of mechanical ventilation, and non-pharmacologic interventions. Symptom management, particularly sedation administration, is a factor that can be modified to improve patient outcomes. Figure 1 presents the aspects of the conceptual framework that will be addressed in this study. Patient characteristics affecting patient outcomes. Many factors are hypothesized to contribute to post ICU complications including sedation, hypotension, hypoxemia, metabolic impairments, and age (Hopkins & Jackson, 2006; Pandharipande et al., 2006; Samuelson et al., 2006). Physiologic factors are labeled severity of illness in this conceptual model. Severity of illness and age are included as covariates in the model 40 Sedation has been shown to increase the duration of mechanical ventilation which contributes to complications (De Jonghe et al., 2002; Douglas et al., 2002; Kress et al., 2000). Additionally, a longer duration of mechanical ventilation will potentially impact patient assessment of their experience. Interventions to manage symptoms. Patients receiving mechanical ventilation experience multiple physical and psychological stressors. Sedation is hypothesized to have an impact on the patient symptom experience, recall, and evaluation of mechanical ventilation. Non-pharmacologic interventions—environmental, interpersonal, and physical—have been reported by patients as effective in reducing symptoms during mechanical ventilation (Gries & Fernsler, 1988; Hafsteindottir, 1996; Hupcey & Zimmerman, 2000; Jablonski, 1994; Logan & Jenny, 1997; Wunderlich et al., 1999). 41 Pattern of sedation during mechanical ventilation. Previous studies related to sedation have primarily relied on average or cumulative doses, number of days sedated, or focused on only one medication (De Jonghe et al., 2005; Nelson et al., 2000). Cumulative measures can be confounded by multiple factors including time on the ventilator, individual differences in reaction to medications, and organ dysfunction. With the use of average sedation levels or amount of medication there is a loss of detail regarding a patient’s varying levels of sedation over the course of mechanical ventilation. The pattern of sedation –sedative exposure and level of sedation—provides a more complete picture of a patient’s time on the ventilator and is potentially a better measure of the state of the patient in relation to sedation administration. This premise is an extension of previous research that found the number of days of sedation and proportion of days at high, medium, or low levels of sedation correlated with psychological complications and patient memory of ICU (Nelson et al., 2000; Samuelson et al., 2006; Weinert & Sprenkle, 2008). Patient evaluation of mechanical ventilation. The effectiveness of interventions to improve the patient experience is best evaluated by the patient’s own assessment. This hypothesis is founded in pain management and assessment literature where the centrality of patient self report is well established (Barnhouse, Kolodychuk, Pankratz, & Olinger, 1988; Institute for Clinical Systems Improvement, 2006; Jacobi et al., 2002). Specific Aims The specific aims of this study are to: 1. Describe patient memories of the intensive care unit 2. Describe the trajectory or pattern of sedation during mechanical ventilation 42 3. Describe the relationship between sedation trajectories and the mechanically ventilated patient’s evaluation of critical care including awareness of surroundings, recall of experience, frequency of frightening experiences, and satisfaction with care. 4. Identify interpersonal, physical, and environmental interventions patients find beneficial for minimizing physical and emotionally distressful symptoms. Setting and Sample A convenience sample of patients was enrolled over 18 months (May 2009 – October 2010) from a 24 bed medical-surgical ICU in a suburban community hospital in Edina, Minnesota. The unit is staffed 24/7 by university affiliated intensivists. Patient to nurse staffing ratios are 2:1 or 1:1. The unit sedation protocol incorporated both daily sedative interruption and titration of sedative medications to an individualized sedation score target. The targeted sedation level is set daily by physicians and nurses. Adherence to the sedation protocol was not evaluated in this study. Patients were eligible for the study if they were greater than 18 years old, spoke English, had an anticipated duration of MV greater than 24 hours, and had no documented mental incompetence. Patients on a ventilator in a long term unit or at home prior to ICU admission were not eligible. Enrollment was initially limited to non-surgical patients. However, due to low enrollment, this was expanded four months after study start up to include surgical patients remaining on the ventilator for greater than 48 hours post surgery. At the beginning of recruitment, all patients were enrolled while in the ICU. If the initial study consent was obtained from a patient’s proxy (closest family member or power of attorney), the informed consent process was repeated with patients prior to the 43 post ICU interview. During the conduct of the study it was noted that more patients than anticipated did not have proxies available to provide consent. Therefore, in July of 2010, the protocol was expanded to allow enrollment of patients post ICU discharge if they were unable to provide consent while in ICU and a proxy was unavailable. Data Collection Procedure, Study Variables, and Measurement Data was collected from the subject’s medical record while the patient was on the mechanical ventilator. Interviews were primarily conducted after ICU discharge and prior to hospital discharge. Interviews were also conducted at a rehabilitation hospital in St. Paul, Minnesota when enrolled patients transferred directly from the ICU to the rehabilitation hospital’s long term ventilator unit. Post ICU interviews included two measures of patient memories: the Intensive Care Experience Questionnaire and the Intensive Care Unit Memory Scale. Table 5 describes study variables and their measurement. Subject characteristics. Demographic and clinical data were abstracted from the medical record including age, gender, medical history, current diagnoses, reason for ICU admission, length of ICU stay, length of mechanical ventilation, medications received, and sedation assessment scores. Severity of illness. Severity of illness was measured using both the Multiple Organ Dsyfunction Score and the Acute Physiology, Age, and Chronic Health Evaluation III. Multiple Organ Dysfunction Score (MODS). MODS provides an estimate of cumulative organ dysfunction or the level of acuity of the patient during the entire course of critical illness (Buckley, Gomersall, & Ramsay, 2003). The score is calculated by Table 5. Study Variables, Measurement Instruments, and Measurement Frequency Variable Instrument Frequency of Measurement study entry daily during MV post ICU discharge Demographic Data n/a X Age n/a X Severity of Illness MODS calculated with daily values APACHE calculated with ICU admission values Duration mechanical ventilation n/a calculated at extubation Sedative Exposure Sedation Intensity Score X1 Level of Sedation/Arousal Motor Activity Assessment Scale X1 Patient Interview X Satisfaction with Care Intensive Care Experience Questionnaire X Awareness of Surroundings Intensive Care Experience Questionnaire X Frightening Experiences Intensive Care Experience Questionnaire Recall (Global) Intensive Care Experience Questionnaire X Recall (Specific) Intensive Care Unit Memory Tool X Factual Intensive Care Unit Memory Tool Delusional Intensive Care Unit Memory Tool X Feelings Intensive Care Unit Memory Tool X Notes. 1 Abstracted from medical record in four hour blocks during mechanical ventilation; APACHE: acute physiology, age, and chronic health evaluation; MODS: multiple organ dysfunction score 44 45 summing the individual worst value in six organ systems: respiratory, renal, hepatic, cardiovascular, hematologic, and neurological. The highest possible score is 24. Higher scores correlate with an increased ICU mortality ( Marshall et al., 1995). The cardiovascular score requires a central venous pressure (CVP) reading. When CVP values were not available a score of 0 was assigned for the cardiovascular system. Neurological data was taken from Glasgow Coma Scale (GCS) score in the nursing record. The common practice on the unit was to rate the verbal response as 0 (none) for intubated patients regardless of orientation or level of interaction of the patient. This however results in an artificially low GCS score in interactive patients based solely on physical inability to speak. Therefore, orientation status was collected from the nursing assessment and was used by the researcher to assign a verbal response score that was used to modify GCS scores in the MODS calculations. Due to concern regarding unavailable data for cardiovascular and neurological scores, the Acute Physiology, Age, and Chronic Health Evaluation (APACHE III) measure of severity of illness at ICU admission was also calculated. The Acute Physiology, Age, and Chronic Health Evaluation (APACHE III). APACHE III assigns a score based on age, prior health status, and values of twelve physiologic measurements (Knaus et al., 1991). Scores range from 0 to 299 with higher scores indicating greater severity of illness and correlating with an increased incidence of in-hospital mortality (Knaus et al., 1991). Pattern of sedation. The pattern of sedation over the course of mechanical ventilation was measured in two ways: sedative exposure and level of sedation. Sedative 46 exposure is defined as the amount of sedative medications received by the patient and is measured using the sedation intensity score (Weinert & Calvin, 2007). Sedative exposure. A sedation intensity score (SIS) was calculated for all patients based on sedative medications received during mechanical ventilation. Describing overall sedation administration is difficult. Patients receive a number of different medications from different pharmacological classes and no standardized method to compare doses between classes exists. SIS provides a measure of sedative exposure in relation to others in the sample (Weinert & Calvin, 2007). Sedative medications received while on the ventilator, including midazolam, diazepam, lorazepam, haloperidol, propofol, morphine, hydromorphone, dexmedetomidine, and fentanyl were recorded in four hour blocks for each subject. An average daily sedation intensity score was calculated for each subject that completed an interview. First each medication dose in a four hour block was weight adjusted. Propofol and dexmedetomidine did not require weight adjustment since these medications are dosed by weight. Second, each dose within a four hour block was categorized as 1 to 4 based on the quartile of distribution of the drug for the entire sample of interviewed subjects. For each medication, a value of one indicates a sedative exposure in the lowest quartile and four indicates a sedative exposure in the highest quartile for that time period when compared to other subjects (Weinert & Calvin, 2007). The quartile rank values were summed across medications in each time block. For example if during the first four hour time period the subject’s propofol dose fell into the first quartile of the distribution and the subject’s fentanyl dose fell into the third quartile of the distribution the subject’s SIS score for that time block would be 4. Finally, the SIS scores were 47 averaged for each 24 hour period on the ventilator. Only medications received by at least 15% of interviewed subjects were used in the SIS calculations. Morphine and haloperidaol were excluded. Quartile distributions were unable to evenly calculated for medications received by less than 15% of this small sample. Average daily SIS scores were used in the analysis. Level of sedation. Level of sedation or the level of arousal of the patient during mechanical ventilation was assessed using the Motor Activity Assessment Scale (MAAS). The MAAS was collected from the vital sign flowsheet and recorded in four hour blocks. The MAAS ranks sedation or arousal level of patients on a scale of 0 (unresponsive) to 6 (dangerously agitated) with 3 indicating a alert, calm, and cooperative patient (Table 6). MAAS scores are positively associated with changes in heart rate, blood pressure, a visual analog scale rating of level of sedation, and agitation related events such as pulling on catheters or sheets. (Devlin et al., 1999). The scale has demonstrated good interrater agreement (kappa = .83) (Devlin et al., 1999). Although MAAS was selected as the sedation scale for this study because it was already in use on the unit, it was not consistently recorded every four hours on all patients particularly in early stages of the study. MAAS scores were therefore assigned by the researcher based on the patient care nurses’ assessment of level of arousal charted in the neurological assessment. This documentation had 7 options ranging from unresponsive to spontaneous arousal. These responses were then recoded into MAAS scores as follows:  No response recoded as MAAS 0 (unresponsive)  Arousal to pain, vigorous stimulation, or repeated stimulation recoded as MAAS 1 (noxious stimuli) 48  Arousal to touch or voice recoded as MAAS 2 (touch or name)  Spontaneous arousal recoded as MAAS 3 (calm and cooperative)  Spontaneous arousal and documented agitation recoded as MAAS 5 (agitated) Table 6. Motor Activity Assessment Scale Description Definition Score Unresponsive no movement to noxious stimuli2 0 Responsive to Pain/ noxious stimuli movement or eye opening only with noxious stimuli2 1 Responsive to Voice/Touch movement or eye opening when touched or name is spoken loudly 2 Calm and Cooperative spontaneous movement; patient is purposeful and follows commands 3 Restless and Cooperative spontaneous movement; patient is picking at sheets/tubes and follows commands 4 Agitated spontaneous movement; patient attempting to sit up/get out of bed/inconsistently follows commands 5 Dangerously agitated spontaneous movement; patient pulling at tubes/catheters OR thrashing in bed OR trying to climb out of bed; patient does not calm down when reassured/asked 6 Notes. 1 source: Devlin et al., 1999; 2 Noxious stimulus: suctioning OR 5 secs orbital, sternal, or nail bed pressure. 49 Patients’ memories of ICU. The subject’s evaluation of mechanical ventilation and critical illness was measured using the Intensive Care Experience Questionnaire (ICEQ) and the ICU Memory Tool (ICUM). Copies of the ICEQ and ICUM and documentation of permission for use are in Appendix C. Instruments were administered during an in-person interview after ICU discharge either on the hospital ward or in a long term ventilator unit in a rehabilitation hospital. Intensive Care Experience Questionnaire (ICEQ). The ICEQ provides a global evaluation of the subject’s experience and consists of 24 items in four domains: awareness of surroundings, frequency of frightening experiences, recall of experience, and satisfaction with care. Items are closed questions with a five point Likert response indicating level of agreement (strongly disagree to strongly agree) or measuring frequency of event (never to all of the time) (Rattray, Johnston, & Wildsmith, 2004). Each item is scored on a 1 to 5 point scale and summed within each subscale (Rattray et al., 2005). Higher scores indicate greater satisfaction, more frequent frightening experiences, greater recall, and greater awareness of surroundings. Reliabilities for subscale scores range from .71 to .93 (Rattray et al., 2004). Validity of the instrument has begun to be established. In one study, frequency of frightening experiences was shown to correlate with higher anxiety (r = .48) and depression (r = .25 to .27) scores at discharge and six months after discharge. Recall of experiences was negatively correlated with avoidance (r = -.35) and intrusion (r = -.26) scores. Awareness of surroundings was negatively correlated with depression (r = -.21) (Rattray et al., 2004). Intensive Care Unit Memory Tool (ICUM). Type and amount of memories of the ICU was measured using one item from the ICUM. This item asks patients to indicate 50 yes or no to a list of twenty factual memories, memories of feelings, and delusional memories. Items are scored as 0 (not remembered) or 1 (remembered) and added to give the number of memories within each domain: feelings, delusional, or factual (Jones, Humphris, & Griffiths, 2000; Jones, Griffiths, Humphris, & Skirrow, 2001). A total memory score was calculated by summing memories from the three subscales. Cronbach’s alpha for the subscales range from 0.73 to 0.86 (Jones et al., 2000). The ICUM does not assess the impact of the memory for the patient. Open-ended interview questions. All interviews were taped and transcribed. In addition to patients explanations of their answers, follow-up questions were asked for clarification of subject’s responses and the following open questions were asked of all individuals during the interview: a. Do you find any of your memories of ICU distressing? b. Is there anything else you would like to share about your experience of being on the ventilator in the ICU? OR Is there anything else you think we should know about your experience? c. Can you describe anything the staff (nurses/physicians/therapists) did or could have done to improve your experience of being on the ventilator? Non-pharmacologic Interventions A checklist of interpersonal, environmental, and physical interventions that may have improved the subject’s experience was developed for this project by the investigator based on a literature review. Items rate both the frequency and the effectiveness of the intervention (not helpful to very helpful). Subjects were asked to rate the effectiveness of the intervention either based on their experience or if they did not recall the intervention, 51 based on their beliefs about the potential effectiveness of the intervention. The nurses caring for the patients were asked to complete a similar checklist once per shift indicating if each intervention was done. Copies of both forms are in Appendix D. Human Subjects Protection and Data Management All aspects of this study were reviewed and approved by the University of Minnesota Institutional Review Board which serves as the Institutional Review Board (IRB) for the primary study site and by the IRB for the acute care rehabilitation hospital where post ICU interviews were also conducted. All information from the medical record was collected by the PI. Interviews were conducted by the PI and a research nurse trained by the PI. Data entry was completed by the PI. Electronic databases are all password protected and subject data is stored in locked file cabinet with limited access. Data Analysis Sample size. At the time of study design, not enough was known about the pattern of sedation over the course of mechanical ventilation for a power or sample size computation which requires that error variance of the parameters be known or estimated. Therefore, sample size guidelines for growth curve modeling were used to provide an estimate of an appropriate sample size. In growth curve modeling, 5- 10 observations are necessary for each parameter in the model (Muthen & Curran, 1987). A cubic model was anticipated based on exploratory analysis of arousal utilizing a similar dataset to the proposed study (Weinert & Calvin, 2007) resulting in a necessary estimated sample size of 60 individuals. The proposed sample size was increased by 40% to account for attrition due to mortality, confusion, or direct transfer to another institution resulting in a planned sample size of 84 subjects. 52 Analysis by specific aim. Data were examined using the appropriate descriptive statistics to assess type of distribution and any patterns of missing data. Descriptive statistics for all variables included means and standard deviations for interval level data that is normally distributed; medians and ranges for non-normally distributed interval data; and frequencies for categorical data. M-Plus version 6.1 was used for estimation of models for arousal and sedative exposure over time. SPSS version 18 was used for all other analyses. Aim One: Describe patient memories of the intensive care unit. Descriptive statistics including frequency of responses and medians with ranges for subscale scores were used to summarize the responses to the two quantitative instruments: ICEQ and ICUM. Interview transcripts were analyzed using a modification of qualitative content analysis: the interpretation of data through systematic identification of patterns or themes (Hsieh & Shannon, 2005). Analysis began with repeated reading of the transcripts (Hsieh & Shannon, 2005; Shields & Twycross, 2008). Transcripts were then open coded using words, phrases, or sections of text to capture meaningful units or themes (Elo & Kyngas, 2008; Germaine, 2001; Hsieh & Shannon, 2005). The themes were then grouped or collapsed into categories based on the subscales of the ICUM and ICEQ. A journal was kept starting prior to data collection and continuing during data analysis to allow the researcher to identify any preconceptions, reactions to interviews or transcripts, insights, and study decisions. The journal also contains a log with memos documenting ideas, thoughts, and decisions on categorization of text. Journaling allowed the researcher to be 53 cognizant of any personal biases that could have influenced data collection, the interview process, or data analysis (Ahern, 1999; Germaine, 2001; Rodgers & Cowles, 1993). Aim two and three: (1) Describe the trajectory or pattern of sedation and (2) Describe the relationship between the sedation trajectories and the mechanically ventilated patient’s evaluation of critical care. In separate analyses, latent class growth analysis (LCGA) was used to model trajectories of arousal and sedative exposure. LCGA allows the identification of distinct classes or groups based on their trajectory or growth curve (Li, Duncan, Duncan, & Acock, 2001). Covariates that influence class membership can be incorporated into the model and distal outcome variables can be added to allow examination of the relation between the outcome and the latent trajectory class (Jung & Wickrama, 2008; Li et al., 2001). Using LCGA, subjects were classified based on their temporal pattern of arousal or sedative exposure over the first 5 days of mechanical ventilation. The goals of this analysis are to determine the number of subpopulations or latent trajectory classes for arousal and sedative exposure in the sample, to describe the way members of the subpopulations change, to estimate the impact of covariates (age and APACHE) on class membership, and to estimate the relationship between class membership and the distal outcome (memories). Schematic representation of the LCGA frameworks for the arousal and sedative exposure models are presented in Figures 2 and 3 respectively. Building the LCGA Model. Prior to building the LGCA model, the pattern of change was explored using individual empirical trajectories for sedative exposure and level of arousal over time. Individual plots were used to identify plausible functional forms and number of classes to be used in LCGA model specification. The LCGA 54 models were built in a stepwise fashion. First the functional form of the growth curve model was selected—linear and quadratic models were compared. Within class variances were fixed to zero consistent with the LCGA framework throughout the analysis (Jung & Wickrama, 2008). Second, an unconditional latent class model was specified for k = 1-3 classes using the identified functional form of change. The number of classes was selected based on the following primary and secondary criteria (Duncan, Duncan, & Strycker, 2006; Jung & Wickrama, 2008). Primary criteria for model selection were (a) Vuong-Lo-Mendell-Rubin likelihood ratio test (VLRT) and the Lo-Mendell-Rubin adjusted likelihood ratio test (LRT) both of which compare the nested models with k and k-1 classes; and (b) Bayesian Information Criteria (BIC). Additionally, three other 55 criteria were considered during model selection (a) entropy a measure of how well individuals fit into their class (closer to 1.0 is better); (b) posterior probabilities or probability of class membership (higher is better); and (c) percent of total sample in each class (should be no less than 1%). Third, after selection of an unconditional model, class was regressed on the covariates of age and APACHE to create a conditional latent class model. Gender was not included in the conditional model due to the small sample size and inability to estimate a multiple group model. Fourth, a measure of total ICU memories was added into the model as a continuous distal outcome variable. Within this framework, average memories in each latent class are estimated. 56 LCGA model variables are presented in Table 7 and 8. For the arousal model, average daily arousal values over the first 5 days of mechanical ventilation were entered as continuous outcome variables throughout the analysis. For the sedative exposure model, average daily sedation intensity scores for the first 5 days of mechanical ventilation were entered as continuous outcome variables throughout the analysis. Times of measurement were centered so that the average arousal or sedative exposure at time 0 represents the first 24 hours of mechanical ventilation for each individual. Arousal and sedative exposure trajectories were only modeled for the first 5 days of mechanical ventilation because after 5 days greater than 50% of the sample was no longer on the ventilator. Table 7. Arousal LCGA Model Variables Variable Instrument Measurement Entered in Model Average Daily Arousal MAAS Centered: average arousal at time 0 represents the first 24 hours of MV continuous outcome variables Severity of Illness APACHE n/a covariate affecting class membership Age n/a n/a covariate affecting class membership Memories ICUM total ICUM score (count of all memories) entered as continuous distal outcome Notes. LCGA: latent class growth analysis; MAAS: motor activity assessment score; MV: mechanical ventilation; APACHE: Acute Physiology, Age, and Chronic Health Evaluation; ICUM: Intensive Care Unit Memory Tool 57 Table 8. Sedative Exposure LCGA Model Variables Variable Instrument Measurement Entered in Model SIS Sedation Intensity Score Centered: average SIS at time 0 represents the first 24 hours of MV continuous outcome variables Severity of Illness APACHE n/a covariate affecting class membership Age n/a n/a covariate affecting class membership Memories ICUM total ICUM score (count of all memories) entered as continuous distal outcome Notes. LCGA: latent class growth analysis; MAAS: motor activity assessment score; MV: mechanical ventilation; APACHE: Acute Physiology, Age, and Chronic Health Evaluation; ICUM: Intensive Care Unit Memory Tool Formal LCGA model specification. LCGA extends the conventional growth model to incorporate categorical latent classes (k) to illuminate heterogeneity in the pattern of growth within the population (Li et al., 2001). Additionally, LCGA is considered a special case of growth mixture modeling where the within class variances of the intercepts and slopes are fixed at zero (Jung & Wickrama, 2008). Model specification follows that presented by Li and colleagues (2001). The unconditional LCGA model. The unconditional growth mixture for k = 2 classes can be specified as yik = νik + Λkηik + εik , (1) where k = 1,2,3, i = 0, 1, 2..35. The vector yik contains the observed variables for individual i in the kth class; νik is a vector of y intercepts; Λk is a fixed design matrix; ηik contains the intercept and linear and quadratic growth factors for individuals in the kth 58 class; and and εik is the residual for person i in the kth class. LCGA includes the restriction that ηik = αk , (2) so that parameters are the same for all individuals in the same class. The conditional LCGA model. Adding covariates to the unconditional model, yields the conditional model for the log odds of class membership given a value of the covariates. Let πik = P(cik= 1| Xi) and the k dimensional vector πi = (πi1, πi2, …, πik), then the conditional model can be specified as specified as (3) the covariates. logit πi = αc + ΓcXi , where logit πc is a k-1 dimensional vector equal to (log[πi1/ πik], [πi2/ πik],…, [πi,(k-1)/ πik])´ and πi1 = P(cik= 1| Xi); αc is a k-1 vector of the intercepts in the model for the log odds; and Γc is a k-1 by q matrix containing the changes in log odds of class membership per unit change in The probability of class membership can be obtained by first taking the antilogit of equation three specified as Logit-1 (πi) = eπi / 1 + eπi , (4) then solving equation 4 to express the probability of class membership given the covariates specified as πi = e(αc + ΓcXi) / 1 + e(αc + ΓcXi) , (5) Addition of distal outcome. LCGA can be extended to include continuous distal outcomes (Y) specified as Yik= νc , (6) where mean memories are the same for individuals in the same class. 59 Aim Four: Identify interpersonal, physical, and environmental interventions patients find beneficial for minimizing physical and emotionally distressful symptoms. Descriptive statistics were used to summarize subjects’ and nurses responses including frequencies for recall of interventions and medians and ranges for ratings of intervention effectiveness. 60 CHAPTER IV Results Chapter IV presents results by study aim. The specific aims of this study were to: 1. Describe patient memories of the intensive care unit 2. Describe the trajectory or pattern of sedation during mechanical ventilation 3. Describe the relationship between the sedation trajectories and the mechanically ventilated patient’s evaluation of critical care including awareness of surroundings, recall of experience, frequency of frightening experiences, and satisfaction with care. 4. Identify interpersonal, physical, and environmental interventions patients find beneficial for minimizing physical and emotionally distressful symptoms. This chapter begins with a description of recruitment and patient characteristics followed by a description of memories of the intensive care unit (aim one). The pattern of sedation is described in the latent class growth analysis framework for level of arousal and sedative exposure (aim two). The relationship between the pattern of sedation, as measured by level of arousal and sedative exposure, and memories is described (aim three). Finally, a description of interventions that patients believe to be effective for minimizing physically and emotionally distressful symptoms is presented (aim four). Subject Recruitment and Characteristics Sixty-nine subjects were enrolled into the study of whom 35 (50.7%) completed a post ICU interview. Reasons interviews were not completed included transfer to another facility directly from ICU (32.4%), post ICU confusion (29.4%), deceased in ICU 61 (26.5%), hospital discharge prior to interview (5.9%), and ICU-acquired weakness with patient unable to physically sign consent form (2.9%). One subject declined to participate in the interview stating that they had “too much going on.” Eighteen patients met eligibility criteria and the family or the patient was approached for study consent. Reasons that families declined (n=16) included that the patient wouldn’t want to participate (37.5%); there was too much going on (12.5%); the family preferred not to make the decision for the patient (12.5%); the family needed time to consider (12.5%); and the patient had a pre-existing memory loss (6.3%). Eligible patients in ICU were unable to complete the consent process due to patient confusion (50%) and inability to remain awake for the consent (50%). Due to budget and time constraints the enrollment did not meet the targeted goal of 84 subjects. Subject characteristics are presented in Table 9. Subjects were on the ventilator for a median of 6.2 days. Reasons for ICU admission varied but were primarily pulmonary, medical cardiac, or sepsis/severe infection. Individuals able to complete the post ICU interview had a lower severity of illness as measured by MODS (t(67) = 3.140, p = .003), spent less time on the ventilator (U=199, z(68) = -3.184, p = .001) and in the ICU (U = 299, z(68) = -3.553, p < .001). Admission diagnoses differed between groups (χ2(1, N = 69) = 11.26, p = .001). Patients that completed interviews were more likely to have been admitted to ICU with a pulmonary diagnosis and less likely to have been admitted with sepsis/severe infection or shock. A greater percentage of subjects that were not interviewed received continuous infusions of benzodiazepine and opioids. Additionally more patients not interviewed received propofol, midazolam, dexmedetomidine, dilaudid, and haldol at some time during mechanical ventilation. 62 Interviews were completed a median of 4.8 days (range 1 to 28) after extubation. Two subjects participated in the interview while still receiving intermittent mechanical ventilation. Table 9. Characteristics of Study Sample Total Interviewed Not Interviewed Test Statistic p value Age [mean(SD)] 66.0 (12.7) 66.0 (12.9) 66.1 (12.6) .010a .992 Female [n(%)] 39 (59.5) 18 (51.4%) 21 (61.8%) .750b .387 APACHE III [mean(SD)] 74.7 (26.0) 72.03 (23.8) 77.4 (28.2) .857a .395 MODS[mean(SD)] 8.7 (3.9) 7.4 (3.4) 10.1 (3.9) 3.14a .003 Ventilator Days [median (IQR)] 6.2 (8.1) 4.5 (6.8) 10.2 (7.9) -3.18c .001 ICU Days [median (IQR)] 11.4 (12.8) 7.4 (9) 16.0 (12.1) -3.553c <.001 Mortality [n(%)] 9 (13.0) 0 9 (26.4%) 10.65b .001 Reason for ICU Admission [n(%)] 11.26b .001 pulmonary 29 (42.0) 20 (57.1) 9 (26.5) cardiac-medical 13 (18.8) 8 (22.9) 5 (14.7) cardiac-surgical 4 (5.8) 2 (5.7) 2 (5.9) sepsis/infection 10 (14.5) 3 (8.6) 7 (20.6) other surgical 4 (5.8) 2 (2.9) 3 (8.8) neuro/neuromuscular 4 (5.8) 1 (2.9) 3 (8.8) shock/hypotension 2 (2.9) 0 2 (5.9) otherd 3 (4.3) 0 3 (8.8) Hispanic 0 0 0 n/a Race [n(%)] 1.11b .293 White 65 (94.2) 34(97.1) 31(91.2) African American 4 (5.8) 1 (2.9) 3 (8.8) Notes. a t-value; b chi-square; c z-value; d nose bleed, lower extremity ischemia, renal failure; APACHE: Acute Physiology and Chronic Health Evaluation; MODS: Multiple Organ Dysfunction Score; IQR: interquartile range; SD standard deviations 63 Table 10. Intravenous Sedative and Analgesic Medications Received Medication (%) Total Interviewed Not Interviewed propofol 78.3 65.7 91.2 midazolam 66.7 60.0 73.5 lorazepam 62.3 62.9 61.8 dexmedetomidine 24.6 17.1 32.4 fentanyl 84.1 82.9 85.3 hydromorphone 50.7 45.7 55.9 morphine 10.1 11.4 8.8 haldol 20.3 14.3 26.5 continuous benzodiazepine 40.6 37.1 44.1 continuous opioid 58.0 54.3 61.8 Aim One: Description of Patients Memories of ICU Scale reliabilities. Reliabilities (Cronbach’s alpha) of the ICUM subscales in the present study were: Delusional Memories (.64); Memories of Feelings (.78); and Factual Memories (.81). For the ICEQ instrument only two subscales in the present study had acceptable reliabilities: Frequency of Frightening Experiences (.72) and Satisfaction (.74). Two items in the Awareness of Surroundings Subscale were negatively correlated with all other items. These items—feeling in control and able to let people know what you wanted—were omitted from the subscale. The revised Awareness subscale has six items with a reliability of .84. Only two items in the Recall subscale had correlations greater than .40; all other items had correlations ranging from -.25 to .25. Omitting negatively correlated items resulted in a subscale with only three items and a reliability of .37. Therefore the Recall subscale was omitted from subsequent analysis. All items omitted from subscales are reported as individual items. Tables 11-13 contains descriptive statistics and reliabilities for subscales of ICUM and ICEQ. 64 Table 11. Participant Responses to the Intensive Care Unit Memory Tool (N = 35) median (range) n(%) Reliability Total ICUM Score 10 (0-18) Factual Memories (11 items) 5 (0-11) .807 family 21 (60) alarms 16 (45.7) voices 24 (68.6) lights 21 (60) faces 21 (60) breathing tube 16 (45.7) suctioning 16 (45.7) darkness 14 (40) clock 12 (34.3) nasogastric tube 12 (34.3) physician rounds 18 (51.4) Memories of Feelings (6 items) 4 (0-6) .779 uncomfortable 21 (60) confusion 26 (74.3) feeling down 18 (51.4) anxious/frightened 20 (57.1) panic 15 (42.9) pain 13 (37.1) Delusional Memories (5 items) 1 (0-4) .637 someone trying to hurt them 5 (14.3) hallucinations 15 (42.9) nightmares 11 (31.4) dreams 12 (34.3) someone trying to kill them 0 Note. 2 people had no memories at all (5.7%); ICUM: Intensive Care Unit Memory Tool Table 12. Participants Responses to the ICU Experience Questionnaire Subscales (N = 35) Awareness of Surroundings [(median(range)] n=22 26.5 (10-33) Possible range: 7-35 Reliability = .842 Disagree (%) Neutral (%) Agree (%) no recollection of being in ICU (n=30) 83.4 3.3 13.3 All/Most of the time Some of the Time Rarely/Never aware of somewhere near to me (n=28) 50.0 21.4 28.6 knew what was happening (n=27) 22.2 22.2 55.6 knew where I was (n=29) 55.2 20.7 24.1 remember relatives (n=28) 50.0 17.9 32.1 recognized relatives (n=25) 80.0 4.0 16.0 felt safe (n=28) 67.9 7.1 25.0 Frequency Frightening Experiences [(median(range)] n=26 17 (7-27) Possible range: 6-30 Reliability = .724 Disagree (%) Neutral (%) Agree (%) did not think I would die (n=29) 41.4 3.4 55.2 All/Most of the time Some of the Time Rarely/Never saw strange things (n=27) 37.0 33.3 29.6 felt helpless (n=28) 67.9 10.7 21.4 seemed to be in pain (n=27) 37.0 11.1 51.9 felt scared (n=28) 28.6 25.0 46.4 seemed to have bad dreams (n=28) 32.1 21.4 46.4 65 66 Table 12 (continued). Satisfaction with care [(median(range)] n=28 14 (6-20) Possible range: 4-20 Reliability = .741 Disagree (%) Neutral (%) Agree (%) my care could have been better (n=29) 65.6 3.4 31.0 it was always too noisy (n=29) 58.6 20.7 20.7 I was constantly disturbed (n = 29) 48.3 13.8 37.9 my care was as good as it could have been (n=28) 21.4 0 78.6 Higher scores indicate higher awareness of surroundings; greater frequency of frightening experiences, and greater satisfaction with care Table 13. Participants Responses to the ICU Experience Questionnaire Individual Items (N = 35) Items (%) All/Most of the time Some of the Time Rarely/Never felt I was in control (n=29) 20.7 0 79.3 Disagree (%) Neutral (%) Agree (%) memories of ICU blurred (n=30) 6.7 10.0 83.3 never knew if day or night (n=27) 33.3 10.0 56.7 seemed to sleep to much (n = 28) 57.1 21.4 21.4 unable to let people know what I wanted (n=27) 66.7 3.7 29.6 wished I had known more about what was happening(n=31) 29 9.7 61.3 wished I remembered more about it (n=31) 51.6 6.5 41.9 67 Memories of facts. Two patients had no memories at all of ICU (5.7%). Subjects remembered a median of 5 (range 0-11) of a possible 11 factual memories. Factual items remembered by the most individuals were family (60%), voices (68.6%), lights (60%), and faces (60%). Least recalled were the clock and the nasogastric tube (34.3%). The breathing tube and suctioning were recalled by just under half of subjects; however this was the most commented upon physical memory. One subject clearly recalled the feeling of the endotracheal tube (ETT) in their mouth. “It felt like a cigar sticking out the side of my mouth or a chunk of wood or something” (Subject 9). Other subjects had only vague memories of the ETT. “I remember something down there. I have a bad sore throat right now, so something did happen you know” (Subject 41). Memories of feelings. Subjects remembered a median of 4 (range 0-6) of 6 possible memories of feelings. Subjects most commonly remembered feeling confused (74.3%) and anxious or frightened (57.1%). For 42.9% of subjects this anxiety reached the level of panic. Although 60% of subjects recalled being uncomfortable only 37.1% recalled experiencing pain. The most often described reason for panic was related to the ETT and ventilator. “So I kept thinking they weren’t giving me enough air. In fact, for half of it I thought it was cutting my air off”(Subject 7). Another individual described feelings of panic related to the ventilator as a “fear of losing my ability to breath” (Subject 45). Some subjects were able to overcome their initial feelings of panic. 68 I remember specifically thinking to myself, okay, just let it go, making a choice. That surprised me, that I’m just able to relax. I don’t know if that’s because they gave me a drug to do that or I just did. I remember panicking about it, but then just relaxing and letting it breathe for me” (Subject 41). Memories of delusions and frightening experiences. Subjects remembered a median of 1 (range 0-4) delusional memories of a possible 5 items. Hallucinations were recalled by the most subjects (42.9%). Subjects reported frightening experiences only some of the time (Mdn 17 range 7-27). The most commonly remembered frightening experience was feeling helpless, reported as occurring most to all of the time by 67.9% of subjects. Seventy-three percent of subjects recall seeing strange things and 54.6% of subjects felt scared at least some of the time. Forty-eight percent of patients recalled experiencing pain. Fifty-five percent of subjects did not have thoughts that they would die while in ICU while 41% did recall having thoughts that they would die. Hallucinations or bad dreams were some of the most frequent and detailed descriptions provided by subjects during interviews. Subjects described hallucinations and bad dreams of being shackled, strangled, caged, and feeling like they were underwater. I was locked up and it was like a big black cage with a black veil over it… and every time I tried to get out, that person would try and stick me with a pitchfork (Subject 22). 69 They were terrible things…I woke up one night and thought the walls were covered with blood and I woke up another night..and I thought I was in a jail cell (Subject 56). You know I’d wake up and …well like..somebody is holding your head underwater or strangulation was a major theme of the scary dreams I had (Subject 39). I dreamed I was in a little car, like a roller coaster car, and I was being pulled..by this engine. And we were shackled down, with our wrists tied, feet tied. And this went on for several days it seemed to me (Subject 3). Although many subjects found the hallucinations and dreams scary others described them as “just there” (Subject 48) or “weird like something from a totally different universe” (Subject 18). Subjects also described their interpretation of hallucinations and dreams while in they were in the ICU. I know it’s a hallucination and part hallucination, part dream, but at the time you’re going through it, to me it was very, very real (Subject 22). You know any dream feels real enough at the moment, you know. When I was totally under, you don’t have that waking experience then to re-establish where you are (Subject 39). Awareness of surroundings and quality of memories. Overall subjects were aware of their surroundings while in ICU some to most of the time (Mdn 26.5, range 10- 70 33). Only 13.3% agreed that they had no recollection of ICU. Subjects were most often aware of having someone near to them (50%) and having relatives at the bedside (50%). Subjects recalled recognizing their relatives 80% of the time. Fifty-five percent of subjects recalled knowing where they were all to most of the time. However, only 22.2% of subjects recalled frequently knowing what was happening to them while they were in the ICU. Sixty-eight percent of subjects felt safe all to most of the time. Eighty-three percent of subjects agreed that their memories of ICU were blurred. While 41.9% of subjects wished they remembered more about ICU, 51.6% of subjects did not want to remember more. One subject described “not knowing what was going to happen” as the worst thing about ICU (Subject 48). Terminology used by subjects to describe their awareness while in ICU included being in a “fog”, being “fuzzy”, or being “kind of hazy” (Subjects 9, 35, 44). One subject described the experience of emerging awareness while in ICU. It was like doing what you’re told, but still trying to understand what’s going on. You know, the brain is trying to process. You’re kind of hazy from being medicated or sedated, but still trying to answer questions and…follow directions or whatever. And then the brain is trying to process, where am I, what is this, and what’s being done to me?…but you do eventually come around to, you know, getting a feel of the picture” (Subject 44). Subjects had mixed responses to the lack of memory of ICU with some wishing they had more, while others preferred not to know. 71 so that I wouldn’t remember. And I don’t, there’s a whole like three weeks in my life that I don’t remember. When asked how this felt she responded: I hate it. I mean I’m glad I don’t remember some of the scary stuff, but I don’t like …feeling like an Alzheimer’s patient (Subject 18). No one wants to remember pain, but it’s overwhelming to know that there’s, [pause] I’ve got the right to the last two and a half, three weeks…and only a small portion of it do I remember (Subject 22). It’s a little disconcerting to know you’ve just lost two days of your life and you don’t recall any of it (Subject 58). I really don’t want to hear about it