INDIVIDUAL AND GROUP FACTORS THAT AFFECT RESILIENCE AND MEDIATE THE RELATIONSHIP BETWEEN RESILIENCE AND THE DEVELOPMENT OF POSTTRAUMATIC STRESS

Transcription

1 INDIVIDUAL AND GROUP FACTORS THAT AFFECT RESILIENCE AND MEDIATE THE RELATIONSHIP BETWEEN RESILIENCE AND THE DEVELOPMENT OF POSTTRAUMATIC STRESS DISORDER IN ICU NURSES by MEREDITH L MEALER A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy College of Nursing 2014

2 ii This thesis for the Doctor of Philosophy degree by Meredith L. Mealer has been approved for the College of Nursing By Joyce Verran, Chair Paula Meek, Advisor Jacqueline Jones Marla Weston Date 4/23/14

3 iii Mealer, Meredith L. (Ph.D., Nursing) Individual and Group Factors that Affect Resilience and Mediate the Relationship between Resilience and the Development of Posttraumatic Stress Disorder in ICU Nurses Thesis directed by Professor Paula Meek ABSTRACT The purpose of this research was to determine the significant individual and group characteristics that affected resilience in intensive care unit (ICU) nurses and whether those significant characteristics had a direct or indirect effect on the development of posttraumatic stress disorder (PTSD) mediated through resilience. An adaptation of the Nurse as Wounded Healer (N-WH) theory was also tested, which added resilience as a concept to facilitate self-healing, transformation and transcendence of trauma experienced in the ICU. This was a secondary database analysis that included 744 ICU nurses from around the United States. Participants were mailed a self-report survey that included demographic, anxiety, depression, PTSD and resilience measures. The factors that significantly affected resilience included: whether the ICU nurse had children, the number of years practicing, type of nursing degree, generational cohort affiliation, and type of unit the nurse was working in. Structural equation

4 iv modeling (SEM) in Mplus was used to model the direct and indirect effects of the significant variables on the development of PTSD mediated through resilience. Children and years practicing as an ICU nurse had direct effects on the development of PTSD. When compared with the medical ICU (MICU), the cardiac ICU, cardiothoracic surgery ICU and other ICUs had significant indirect effects on the development of PTSD mediated through the Personal Competence sub-scale of resilience. ICU nurses with a graduate degree had a significant indirect effect on the development of PTSD mediated through the Leadership sub-scale of resilience, when compared to ICU nurses with a bachelor degree of science in nursing (BSN). The results supported the adapted version of the N- WH theory. Based on the findings of this research, having children, years practicing, type of nursing degree and ICU unit type had significant direct or indirect effects on the development of PTSD mediated through the Personal Competence and Leadership subscales of resilience. Future research is needed to describe the potentially unique characteristics of each unit type and to identify modifiable individual and organizational factors to help inform tailored resilience interventions in the ICU.

5 v The form and content of this abstract are approved. I recommend its publication. Approved: Paula Meek

6 vi ACKNOWLEDGEMENTS I would like to thank my family for inspiring and supporting me through this journey. To my children, Luke and Jessica, thank you for the love and encouragement you shared and the precious time we sacrificed while I pursued my passion for research and professional growth. My father David Walker and grandparents Kenneth and Jean Pettee had an unwavering belief in my ability to succeed and their memory was a motivating force throughout my studies. I also want to thank my mother Marilyn and brother Andrew for always being there to listen and provide support when it was needed. Thank you to my advisor, Paula Meek, for your excellent mentoring and guidance throughout my Ph.D. studies. To my committee members Joyce Verran, Jacqueline Jones and Marla Weston, thank you for your advice, collaboration and intellectual energy. Finally, a special thank you to Marc Moss who introduced me to research and has helped provide me with excellent research opportunities, mentoring and collaboration over the past fifteen years.

7 vii TABLE OF CONTENTS Chapter I. INTRODUCTION 1 Statement of the Problem.4 Background and Significance..5 Theoretical Perspective...14 Summary.16 II. STATE OF THE SCIENCE..17 Metatheory of Resilience...18 The Nurse as Wounded Healer...20 Individual and Group Factors that Influence Resilience..28 Summary.43 III. METHODS.45 Research Design 45 Secondary Dataset.46 Statistical Analyses.55 Summary..63 IV. RESULTS...65 Characteristics of the Sample..65 Research Question One 66 Research Question Two 73 Summary..81

8 viii TABLE OF CONTENTS (CONTINUED) IV. IMPLICATIONS.82 Research Question One...82 Research Question Two...89 Limitations...94 Implications for Future Research.97 Implications for Nursing Practice.98 Summary.99 REFERENCES.101 APPENDIX A. Dummy Coding 111 B. SPSS Descriptives and Recoding 113 C. SPSS Bivariate Correlations.135 D. Mplus Inputs and Outputs.138 E. Human Subject s Approval 151

9 ix LIST OF TABLES Table 1. Summary of Generational Cohorts Participant Demographics CD-RISC Exploratory Factor Analysis Individual and Group Variables Dummy Coding Bivariate Correlations between Individual And Group Factors on the Development of PTSD Direct Effects of Individual and Group Factors on the Development of PTSD Significant Indirect Effects on PTSD through Personal Competence Sub-Scale of Resilience Significant Indirect Effects on PTSD through Leadership Sub-Scale of Resilience...79

10 x LIST OF FIGURES Figure 1. Metatheory of Resilience The Nurse as Wounded Healer First Adaptation The Nurse as Wounded Healer Second Adaptation Confirmatory Factor Analysis 3-Factor Model of CD-RISC Bivariate Correlations Mediation Model for SEM Mediation Model Steps for SEM Individual and Group Variables that Significantly Affect Resilience Sub-Scales Significant Direct Effects on the Development of PTSD Indirect Effects of Unit on the Development of PTSD Mediated through Personal Competence Indirect Effects of Degree on the Development of PTSD Mediated through Leadership Significant Direct and Indirect Effects of Individual and Group Factors on the Development of PTSD through the 3 Sub-Scales of Resilience.80

11 1 CHAPTER I RESILIENCE IN THE ICU NURSE Nurses who work in an intensive care unit (ICU) are under some of the most extreme pressure of any professional in health care. It is not too dramatic to say that an ICU nurse is frequently in positions where their knowledge, skills, judgment and emotions are intertwined in life and death situations. Little thought, information, and research is given on how best to maintain and support these invaluable nurses. The purpose of this research is to explore how individual and group characteristics affect resilience and whether these characteristics have a direct or indirect effect on the development of posttraumatic stress disorder (PTSD) mediated through resilience. The goal of this research is to identify unique characteristics that significantly affect resilience and PTSD. The results may help guide tailored resilience training for nurses working within the intensive care unit environment. This thesis builds upon research conducted by the author over the last decade. These research studies were the first evidence-based publications to address the prevalence of PTSD and resilience, coping mechanisms employed by resilient critical care nurses and how the identified resilient coping mechanisms were used to develop a multi-modal resiliency training program in

12 2 critical care nursing. Critical care nurses and ICU nurses will be used interchangeably to represent the subjects of interest. The first study by the author was conducted at Emory University and found that ICU nurses had a significantly higher prevalence of PTSD symptoms when compared to general medicalsurgical nurses (24% v. 14%, p=0.03). After adjusting for primary hospital, gender, marital status, primary work shift and primary responsibility for household income, being an ICU nurse remained associated with symptoms of PTSD (p=0.02, OR=1.45). To determine whether the results were generalizable, a second crosssectional study was conducted with ICU nurses who lived in the Atlanta-metro area. The nurses who responded represented sixteen different hospitals. Symptoms of PTSD were similar between the Emory ICU nurses and the Atlanta-metro ICU nurses (24% v. 29%, p=0.22). Additionally, symptoms of anxiety and depression were similar in both groups. The third study was conducted at the University of Colorado Hospital and examined the prevalence of PTSD and burnout syndrome (BOS) in ICU nurses. The goal was to determine if BOS and PTSD measures were identifying the same construct. The results were very similar to the first two studies with 29% of ICU nurses having symptoms of PTSD. Essentially all of the ICU nurses who had PTSD also had BOS indicating that PTSD represented a more severe subset of those with BOS. Finally, in a

13 3 large national survey of ICU nurses, 21% were positive for PTSD and 22% were highly resilient. Those with PTSD were significantly more likely to have problems at work and outside of work with relationships and general life satisfaction. When conducting indepth qualitative interviews with the highly resilient ICU nurses, important coping mechanisms that allowed these nurses to remain at the bedside were identified. The coping mechanisms were used to develop a pilot resilience training program. However, the individual and group factors that may help individualize resilience training has not been explored and highlights the importance of this dissertation. The ICU is a stressful environment due to high patient morbidity and mortality, daily confrontations with ethical dilemmas, and a tension-charged atmosphere. ICU nurses are repeatedly exposed to work related stresses including addressing end of life issues, performing cardiopulmonary resuscitation, undertaking postmortem care, and prolonging life by artificial support to their patients (Mealer et al, 2007). ICU nurses have an increased prevalence of psychological disorders such as posttraumatic stress disorder (PTSD), anxiety, depression, burnout syndrome (BOS) (Mealer et al, 2007; Mealer et al, 2009; Mealer et al, 2012), secondary traumatic stress (Beck, 2011; Von Rueden et al., 2010) and compassion fatigue (Maytum & Heiman, 2004; Meadors &

14 4 Lamson, 2007) as a result of working in the critical care environment. Additionally, in a recent meta-analysis, work-related traumatic experiences were positively associated with posttraumatic stress symptoms, anxiety and depression in hospital based healthcare workers (de Boer et al., 2011). However, certain nurses have developed adaptive mechanisms or resilient characteristics that allow them to remain in the difficult critical care environment and successfully work in the ICU for many years. Resilience is a psychological mechanism that can thwart PTSD and is recognized as one of the most important factors for a healthy and pathologic-free adjustment following trauma (Charney, 2004; Davidson, Payne & Connor, 2005; Hoge, Austin & Pollack, 2007; Luthar, Cicchetti & Becker, 2000). ICU nurses with high levels of resilience are significantly less likely to develop PTSD, anxiety, depression and burnout syndrome (Mealer et al, 2012a). Statement of the Problem While there is evidence to support the effects of individual resilience on the psychological outcomes of ICU nurses, the individual and group factors that may significantly contribute to individual resilience and subsequent development of PTSD are relatively unexplored. The purpose of this research is to identify the individual and group characteristics that affect resilience and to determine if the significant individual and group characteristics have

15 5 a direct or indirect effect on PTSD mediated through resilience, in a national sample of ICU nurses. This study explores the following research questions: Research Question #1: What are the individual and group characteristics that affect resilience? Research Question #2: Do the significant individual and group characteristics have a direct or indirect effect on PTSD mediated through resilience? The remainder of the chapter outlines the rationale, significance and introduces the theoretical perspective for the current study. Background and Significance The U.S. healthcare system is experiencing a critical nursing shortage (Chan et al., 2013; Jastremski, 2006; Stechmiller, 2002). The reasons for this national crisis are multi-factorial yet one important component is the accelerated departure of nurses from their profession. Across the United States, nursing turnover rates range from 13-20% (Chan et al., 2013; Hauck, Quinn-Griffen & Fitzpatrick, 2011; Oulton, 2006). The growing national nursing shortage is particularly concerning in the ICU. Nursing turnover diminishes nursing productivity, staff morale, and patient quality of care including: increases in medication errors, patient falls, and other measures of patient morbidity (Barlow & Zangaro, 2010;

16 6 Jones, 2008; Waldman et al., 2004). Nursing turnover is also expensive: costing in excess of $65,000 for each newly hired critical care nurse. These estimated costs to replace an ICU nurse are twice as much when compared to medical/surgical inpatient nurses (Kosel & Olivio, 2002). In addition, critical care nurses are more likely to leave their profession due to issues related to their working environment and a desire for a less stressful position (Stone et al., 2006). As the need for ICU beds increases in the future, the already depleted supply of ICU nurses will encounter more work-induced challenges, demands, and stress. As hospitals strive to provide high quality care in the changing U.S. healthcare environment, the nursing supply remains insufficient and is becoming a national emergency (Steinbrook, 2002). Shortages emerge because of an imbalance between supply and demand. Unfortunately, the nursing crisis is related to difficulties on both parts of this equation. The future demand on nurses is expected to increase dramatically as the overall acuity level of hospitalized patients steadily escalates and the general U.S. population continues to age (Chan et al., 2013; Stechmiller, 2002; Steinbrook, 2002). Simultaneously, significant concerns exist regarding the present and future recruitment and retention of nurses. As the nursing workforce ages (approximately 30 percent of nurses are older than 50 years of age, with the average age of

17 7 nurses increasing by almost 2 years in the last decade), fewer young people are choosing nursing as a career (National Center for Health Workforce Analysis, 2013). Additionally, our present nurses express levels of job dissatisfaction that are among the highest in the U.S. workforce, resulting in nurses leaving their profession with increasing frequency. In a survey conducted by the Nursing Executive Center, one third of nurses under 30 years of age plan to leave the nursing profession within 12 months, and more than 55% of nurses in general would not recommend nursing as a career. In another survey of general nurses who had considered leaving the patient-care field for reasons other than retirement, 56% indicated that they wanted a less stressful and less physically demanding job (Hart Research Associates, 2001; Report to the Chairman, 2001). The nursing shortage is most noticeable in specialty areas such as the intensive care unit (ICU). Presently, there are approximately 500,000 critical care nurses in the U.S. who work in the demanding ICU environment. Of U.S. hospitals, 57% reported that critical care nursing positions were the hardest to fill and have the highest vacancy rate (HRSA, 2013; Report to the Chairman, 2001). The turnover rate for ICU nurses can be as high as 60%, with a cost to each individual hospital as much as $2,100,000 each year (HRSA, 2012; Report to the Chairman, 2001). As a result, there will be over 114,000 vacant ICU nursing positions in the U.S.

18 8 by the year 2015 (Hart Research Associates, 2001; HRSA, 2013; Report to the Chairman, 2001). The staffing requirements for ICU nurses are especially vulnerable to turnover as there is a long period of time from the beginning of orientation until independent nursing productivity, and therefore the loss of experienced personnel and hiring replacements adds significant stress to the remaining staff (Aiken et al, 2002). The retention of experienced ICU nurses is important because it is costly (approximately $65,000 to train one ICU nurse), affects patient care and increases job stress, work satisfaction and group cohesion. Job stress and the cumulative exposure to traumatic events experienced in the work environment can lead to psychological distress and the development of disorders such as burnout syndrome (BOS) and posttraumatic stress disorder (PTSD). PTSD is the fourth most common psychiatric disorder and it develops following the exposure to a stressful or traumatic experience that is responded to with fear, horror and helplessness. The three cardinal symptoms associated with PTSD are reexperiencing the event, avoidance, and a state of hyperarousal (Yehuda, 2002). To meet the diagnostic criteria for PTSD, an individual must have at least one symptom of re-experiencing the event, three symptoms of avoidance, and two symptoms of

19 9 hyperarousal. In addition, individuals must have symptoms for at least one month and the symptoms must cause significant impairment in work-related and social functioning. Re-experiencing or intrusions are characterized by images and/or sensory impressions that can include noises, smells, thoughts, flashbacks, tactile sensations and dreams (Steil & Ehlers, 2000). Avoidance symptoms are efforts made by an individual in an attempt to escape or minimize highly emotional states. Avoidance is the result of extreme episodes of emotionality that are triggered by environmental cues and that lead to emotional exhaustion. However, avoidance may actually aggravate the severity of the emotions that are being avoided (Asmundson, Stapleton & Taylor, 2004). The final symptom cluster is hyperarousal, which describes an abnormal state of activation that follows a traumatic event. While hyperarousal is necessary to adapt during normal stress responses, oftentimes, PTSD is characterized by chronic hyperarousal states that alter brain structures, specifically in the hippocampus and results in threat sensitization and overreaction when faced with a new stressor (Kendall-Tackett, 2000). Burnout syndrome (BOS) is a response to emotional and interpersonal stressors that are experienced at work and BOS is characterized by emotional exhaustion, depersonalization, and a lack of personal accomplishment. Emotional exhaustion is the most

20 10 common symptom of BOS and reflects the stress aspect of the work environment. Emotional exhaustion is what leads individuals to distance themselves emotionally and cognitively from work as a coping mechanism. Depersonalization involves ignoring the unique qualities of patients in an attempt to create distance in the nursepatient relationship. The final dimension, a lack of personal accomplishment or inefficacy is common when the chronic, overwhelming demands of the work environment affect an individual s sense of effectiveness (Maslach, Schaufeli & Leiter, 2001). In contrast to PTSD, BOS does not result as a consequence of exposure to a traumatic event and BOS does not cause disruptions in daily functioning outside of the work environment. BOS is common in stressful jobs such as healthcare workers, police officers, and school teachers. PTSD has been identified as a severe condition in a subset of ICU nurses who are also experiencing BOS (Mealer et al., 2009). Approximately 82% of Americans will be exposed to at least one traumatic event in their lifetime but only 8-10% actually develop PTSD (Hoge et al, 2007; Yehuda, 2002). In healthcare professionals, several different concepts describe symptoms of PTSD. The concepts include secondary traumatic stress, compassion fatigue, moral distress, and vicarious traumatization. However, a recent concept analysis has reported that all of these

21 11 terms are measuring symptoms of PTSD and not a unique concept that warrants new terminology (Mealer & Jones, 2013). Because ICU nurses constantly work in a stressful environment, the development of PTSD likely contributes to their higher turnover rate. In two different surveys of over 490 ICU nurses, more than 25% of ICU nurses have symptoms of PTSD related to their work environment (Mealer et al, 2007). In a third follow-up study of 92 ICU nurses, the majority of ICU nurses with symptoms of PTSD also met all of the diagnostic criteria for PTSD. An additional 60% of ICU nurses have BOS, or symptoms of anxiety or depression (Mealer et al, 2009). Finally, in a national survey of over 700 ICU nurses, 18% were positive for symptoms of anxiety and 11% were positive for symptoms of depression. There was also a high rate of BOS with 80% having symptoms in at least one of the three dimensions: emotional exhaustion, depersonalization, and lack of personal accomplishment. Twenty one percent of the ICU nurses met the diagnostic criteria for PTSD. Resilience was also measured in this population and 22% were categorized as being highly resilient. The highly resilient nurses were significantly less likely to have PTSD, symptoms of anxiety or depression and BOS (<0.001 for all comparisons) (Mealer et al, 2012a).

22 12 The development of individual resilience may result in effective coping mechanisms and adaptive responses for ICU nurses. Understanding the individual and group factors that affect resilience may provide important information related to staffing and training needs in intensive care units with the goal of preventing the development of PTSD in the individual ICU nurse. Resilience is defined as the ability to maintain healthy psychological and physical functioning after being exposed to a traumatic event or death of a close family member or friend (Luthar, Cicchetti & Becker, 2000). Resilience has recently been recognized as one of the most important factors when assessing adjustment following trauma and is thought to assist in preventing the development of PTSD symptoms (Davidson, Payne & Connor, 2005; Hoge, Austin & Pollack, 2007). The concept of resilience was first postulated in the early 1970 s. However, the potential effect of individual and group characteristics on resilience and what effects these characteristics have on the development of PTSD in ICU nurses is completely unexplored. Psychologists have identified factors that promote resilience such as individual temperaments, family bonds, and external support systems. Personal qualities are also associated with resilience including the ability to engage the support of others, optimism, faith, the belief that stress can be strengthening, and striving towards personal goals (Davidson, Payne & Connor, 2005;

23 13 Hoge, Austin & Pollack, 2007). The presence of these qualities may allow individuals to effectively cope with stress and even thrive in the face of adversity. A qualitative study of ICU nurses with PTSD and those who were highly resilient identified differences in four major domains: worldview, social network, cognitive flexibility, and self/care and balance. Highly resilient nurses were more likely to use positive coping mechanisms to adapt to the stressful ICU work environment by engaging in supportive social networks, optimism, having a resilient role model and identifying with spirituality and rituals (Mealer, Jones & Moss, 2012b). Again, the individual and group characteristics that describe resilience, in ICU nursing, are relatively unexplored. Researchers suggest that group level resilience is most influenced by having strong leaders in the work environment (Campbell, Campbell & Ness, 2008). It is the leadership behavior that buffers against stressors and promotes mechanisms associated with resilience. Resilient mechanisms include emphasizing work purpose, building and facilitating group cohesion; offering professional support when needed and empowering peers towards a greater level of control in the work environment (Campbell, Campbell & Ness, 2008). However, the specific individual and group factors that affect resilience are unknown. By studying whether individual and group characteristics affect resilience in ICU nurses, there is the potential

24 14 to target recruitment of ICU nurses based on different levels of individual resilience that better fit a particular group. Additionally, this information may help organizational leaders determine where and to what degree of emphasis resilience training is needed and to design the systems to increase resilience. Theoretical Perspective The theoretical perspective that will guide this research has arisen from the Metatheory of Resilience, the original Nurse as Wounded Healer Theory and two adaptations of the Nurse as Wounded Healer Theory. The Metatheory of Resilience (Richardson, 2002) was developed as a consequence of working with survivors of traumatic events that responded to adversity in a positive way and were able to grow as a result of the trauma. The Metatheory of Resilience proposes a three stage model. The first stage is related to the innate characteristics of a resilient individual. The second stage is related to the characteristics of resilience that can be acquired and the mechanisms by which the characteristics are attained. Finally, the third stage of inquiry is related to the motivational forces that are present in all individuals that help to cognitively process life disruptions (Richardson, 2002).

25 15 The Nurse as Wounded Healer (Conti-O Hare, 2002) is a middle-range theory that identifies nursing as a wounded profession and acknowledges that all nurses have suffered from trauma and the ability to transform and transcend traumatic experiences is necessary to be therapeutic when caring for others. This theory also acknowledges that nurses are either walking wounded or wounded healers. Walking wounded refers to a nurse who has been traumatized, has not acknowledged the trauma, and has not engaged in the self-healing that is necessary to appropriately care for patients. In contrast, wounded healers have acknowledged that they have been traumatized and have been able to transform and transcend their trauma through self-healing techniques. Wounded healers are able to use this growth to compassionately care for patients without being re-traumatized (Conti-O Hare, 2002). The first adaptation of the Nurse as Wounded Healer Theory introduces the concept of nurse PTSD as it relates to the concept of walking wounded (Mealer & Jones, 2013) and the second adaptation of the Nurse as Wounded Healer Theory is a new adaptation that has not been published and introduces the concept of resilience as it relates to the wounded healer and the self-healing process. These four theories will be discussed in more detail in chapter two. These theories provide a foundation for the

26 16 proposed research questions and will guide future investigations and interventions. Summary In summary, ICU nurses are at an increased risk of developing psychological disorders such as PTSD, BOS, anxiety and depression due to the stressful work environment. The stressful work environment is a consequence of indirect exposure to traumatic events such as end of life issues, performing cardiopulmonary resuscitation, undertaking post-mortem care, and prolonging life by artificial support to their patients. Resilience is a learnable psychological characteristic that allows an individual to thrive in the face of adversity and is known to prevent PTSD and facilitate recovery after being exposed to trauma. While individual resilience is known to significantly reduce the prevalence of psychological disorders in ICU nurses, the individual and group characteristics that affect resilience and the direct and indirect effects that these characteristics have on the development of individual PTSD through resilience are unknown. A better understanding of these associations may provide valuable information to assist with enhancing resilience and reducing PTSD in ICU nurses.

27 17 CHAPTER II STATE OF THE SCIENCE The purpose of this chapter is to introduce the state of the science of resilience as it relates to ICU nurses and the theoretical perspective and literature that guided this research. Research related to resilience was not originally grounded in theory but rather followed as a consequence of working with survivors of traumatic experiences who were able to adapt and rebound in the face of adversity, allowing a return to healthy daily functioning. Over the last ten years, research on resilience has received a lot of attention because of its role in mitigating symptoms associated with common psychological disorders such as anxiety, depression and posttraumatic stress disorder (PTSD). The initial conceptual landscape for this study was informed by the Metatheory of Resilience. An adaptation and refinement of Nurse as Wounded Healer Theory is also used as a conceptual guide. The first section of this chapter describes the conceptual framework and research that builds a platform for further testing of the proposed individual and group factors affecting resilience. First, a qualitative inductive study that explored the contextual difference between ICU nurses who were considered to reach PTSD diagnosis criteria and those ICU nurses who scored as highly

28 18 resilient (Mealer, Jones & Moss, 2012b). The second study, a concept analysis that established PTSD as a disciplinary concept of concern within a nursing discourse and its application to The Nurse as Wounded Healer Theory (Mealer & Jones 2013). Third, interpretive adaptation testing of the resilient domains identified in study one, the place of PTSD from the concept analysis and the deductive application to Nurse as Wounded Healer Theory through secondary qualitative analysis (Unpublished manuscript Mealer & Jones). Metatheory of Resilience Inquiry in resilience occurred in three waves. The first wave identified what types of characteristics were present in individuals who were able to thrive in the face of adversity (Richardson, 2002). Some of the protective factors identified during this wave included: easy temperament, communicates effectively, sense of personal worth, assertive, above average social intelligence, having an informal social support network, ability to have close relationships, internal locus of control, flexibility, self-efficacy, interpersonal sensitivity, trust in others, sense of humor, critical thinking skills, hope and high expectations (Anthony, 1974; Bernard, 1991; Garmezy, 1991; Masten, 1994; Rutter, 1985; Werner, 1992). The second wave of inquiry explored how to acquire the characteristics of resilience identified in wave one. Specifically, the

29 19 second wave investigated the internal and external coping mechanisms that can be fostered or enriched to serve as protective factors against life disruptions (Richardson, 2002). These included research examining the psychosocial factors of resilience building and stress tolerance (Pietrzak et al., 2010; Wu et al., 2013). Factors that promote resilience during child rearing include: a loving and supportive environment, positive relationships, supportive and responsible parenting, avoidance of stress and trauma, positive experiences in overcoming life challenges and stress inoculation training through cognitive-behavioral therapy. In adulthood, resilience can be learned through cognitive behavioral therapy that promotes realistic optimism, active coping mechanisms, high cognitive functioning or reframing, strong emotional regulation, secure attachments and trust, strong social network, spiritual beliefs, humor and altruism (Wu et al., 2013). The resilience process, or wave two of resilience theory, will be used to support this research. Finally, the third wave of resilience inquiry focused on the motivational energy that is required to cognitively reframe a life disruption (Richardson, 2002). Acquiring the resilient qualities identified in the first wave of resilience inquiry allows an individual to experience disruption and adversity but also have the ability to access strengths and coping

30 20 mechanisms to grow through the experience. Reintegration refers to the ability to restore to a condition of wholeness. Depending on an individual s ability to access protective, resilient characteristics, one may integrate resiliently, reintegrate back to homeostasis, reintegrate with loss or reintegrate with dysfunction (Figure 1). Figure 1. Metatheory of Resilience (adapted from Richardson, 2002) The Nurse as Wounded Healer The Nurse as Wounded Healer (N-WH) is a middle-range theory that introduces the concepts of walking wounded, wounded healer, self-healing, transformation and transcendence. Walking wounded refers to individuals who have experienced personal and/or professional trauma and have not been able to move past

31 21 the trauma through coping mechanisms and self-healing (Conti- O Hare, 2002) that results in symptoms such as intrusive memories of the traumatic event, avoidance of reminders of the event, and hyperarousal. The first adaptation of the Nurse as Wounded Healer theory was developed to incorporate the concept of PTSD as it relates to the walking wounded. The first published model of the theory, addressed the relationships among the concepts and proposed bidirectional movement between walking wounded and wounded healer depending on the work environment (Mealer & Jones 2013). Oftentimes, if symptoms of intrusive thoughts, avoidance and hyperarousal are not addressed in a timely fashion, the chronicity of the symptoms can lead to changes in worldview such as selfblame, hopelessness and preoccupation with fear or danger; sleep disturbances and relationship problems (Mealer & Jones, 2013) The concept of wounded healer proposes that nurses or other healthcare professionals are able to care for patients after they identify that they have been traumatized and are able to transform and transcend their experience through self-healing. One study identified the perceived importance of both the professional and personal experience of trauma and the ICU nurses subsequent flexible modification or transformation of self and practice in a positive way as a result of such traumatic experiences.

32 22 Preliminary low-inference interpretive descriptive qualitative study. In a sample of highly resilient (n=13) and PTSD diagnosed ICU (n=14) nurses, mechanisms by which resilient nurses thrived in their work environment and the coping skills used to remain at the bedside were identified. Differences were identified in four major domains between resilient ICU nurses and those with a diagnosis of PTSD: worldview; social network; cognitive flexibility and selfcare/balance (Mealer, Jones & Moss 2012b). Highly resilient nurses identified spirituality, a supportive [personal and professional] social network, optimism, and having a resilient role model as characteristics used to cope with stress in their work environment. (2012b:1445). Highly resilient ICU nurses described a worldview (influenced by values, ethics, emotions and past work experience) that allowed acceptance that death is a part of life and that patient outcomes cannot be controlled (2012b: 1447). The experience of trauma [either personal or professional] was used as a bridge to humanity by these highly resilient nurses and viewed as a learning and or growth experience. When they remembered patient encounters outside of work they were viewed as positive experiences and successful encounters, even if they involved patient death (p1449).

33 23 Those ICU nurses with a diagnosis of PTSD possessed several unhealthy characteristics including a poor [personal and professional] social network, lack of identification with a role model, disruptive thoughts, regret and lost optimism. (2012b:1445). Their worldview was described in stark contrast to the highly resilient nurse as aligned with regrets about patient outcomes and the inability to let go of negative experiences or feeling that something more could have been done for a particular patient (2012b:1447). Traumatic personal and professional experiences resulted in disruptive thoughts that were often intertwined with their decision making capacity in the work environment and interfered with their ability to care for patients that were a reminder of the traumatic experiences, resulting in avoidance. These contrasting experiences found in this exploratory study resonated with The Nurse as Wounded Healer Theory (Conti-O Hare, 2002); PTSD ICU nurses as walking wounded and highly resilient ICU nurses as trauma modifiers and wounded healers Concept analysis of nurse PTSD. A subsequent study clarified and refined the concept of nurse PTSD through a concept analysis using the approach suggested by Walker & Avant (2005). A model of nurse PTSD emerges from synthesis of the literature and adoption of the fundamental assumptions of the Nurse as Wounded Healer theory.

34 24 Nurse PTSD is based on the individual nurse s exposure to traumatic events; the process of self-healing; and the ability to transform and transcend the experience. The refinement of the concept of nurse PTSD is based on Coni-O Hare s (2002) middle range theory. NW-H theory identifies that all nursing professionals experience personal and professional trauma and the way in which they cope with trauma has a direct impact on patient care and determines whether nurses are walking wounded or wounded healers. Walking wounded refers to nurses who remain physically, emotionally, and spiritually bound to past trauma while wounded healers are nurses who are able to use self-reflection and spiritual growth to achieve an expanded consciousness through which the trauma is processed, converted, and healed (Conti-O Hare, 2002). It recognizes that nursing is a wounded profession that requires transforming and transcending traumatic experiences to successfully engage the therapeutic use of self in practice. This model incorporates the contextual richness of the nursing discipline as well as the physical and psychological mechanisms of PTSD (Figure 2, adapted from Mealer & Jones 2013). Interpretive testing of PTSD, self-healing and transformation of trauma in N-WH theory. The second adaptation of the Nurse as Wounded Healer theory is the author s adaptation and has not yet been published or

35 25 Figure 2. The Nurse as Wounded Healer First Adaptation: (Adapted from Mealer & Jones, 2013) empirically tested. Based on a secondary analysis of the qualitative data from the first study, the four domains were re-analyzed through the lens of N-WH theory. This secondary analysis supports the use of N-WH theory as a framework to explore transformative prospects for psychological healing in nursing. It also supports the interrelations between the major themes generated during the primary interviews (worldview, social network, cognitive flexibility, and self-care/ balance) and the major concepts of the framework (walking wounded, wounded healer, self-healing, transformation and transcendence). It offers a window to learn how resilient nurses

36 26 use exposure to trauma in a positive way and how they heal themselves on their journey as nurses. This adaptation introduces resilience and the importance of resilience in facilitating selfhealing. Self-healing or self-care and balance refer to the physical, emotional and psychological strengths that are used to maintain a healthy and balanced lifestyle (Mealer, Jones & Moss, 2012b). Addition of the resilient characteristics of critical care nurses identified in the literature is a method by which self-healing can be initiated. These characteristics include spirituality, social support, having a positive role model, active coping skills and optimism (Mealer, Jones & Moss, 2012) (Figure 3). Past traumas are used to heal and gain insight, allowing a transformation of a traumatic experience into self-acceptance. Selfacceptance and healing of past traumas are necessary before healing can be engendered in others. There is also evidence that resilience serves as a protective mechanism against common psychological disorders and symptoms such as burnout syndrome, anxiety, depression and PTSD. Resilience allows an individual to respond to adversity in a positive way and grow from a traumatic experience. Resilience is also recognized as one of the most important factors for a healthy and pathologic-free adjustment following trauma (Charney, 2004; Davidson, Payne & Connor, 2005; Hoge, Austin & Pollack, 2007;

37 27 Luthar, Cicchetti & Becker, 2000). Evidence suggests that ICU nurses with high levels of resilience are significantly less likely to develop PTSD, anxiety, depression and burnout syndrome (Mealer et al, 2012a). The currently proposed research will be grounded in the second adaptation of the N-WH theory and will test the ability for wounded healers to self-heal by adopting resilient characteristics. There is very little evidence related to individual and group factors that affect resilience and their role in mitigating psychological symptoms. There is evidence to support system characteristics in the nursing work environment that is beneficial for reducing environmental stress. These characteristics include control of nursing practice (Kramer, 2008; Weston, 2008), managerial support (Bennett et al., 2001), nursing empowerment (Armstrong & Laschinger, 2006; Browning, 2013) and nursing satisfaction (Hayes et al., 2006). However, these variables are not available in the database being used for this proposal. The remainder of this chapter will be devoted to evidence in the literature related to resilience as a buffer that mitigates maladaptive psychological symptoms, resilience training in the healthcare setting, and how group membership (type of ICU and generational cohort) effects the development of psychological distress.

38 28 Figure 3. The Nurse as Wounded Healer: Second Adaptation: (Adapted from Mealer & Jones, 2013) Individual and Group Factors that Influence Resilience The next section focuses on the key individual and group factors that may affect resilience. Individual factors. In a study examining protective factors and the role of protective factors when dealing with stress and emotions in later adulthood, psychological resilience was associated with positive emotion. Adults between the ages of 62 and 80 years (n=27), who were randomly chosen from a larger research study, were administered questionnaires on resilience, positive and negative emotions, and stress. Participants collected daily data related to

39 29 emotions and stress for up to 45 days. Using multilevel modeling analyses, trait resilience was found to moderate the relationship between daily stress and negative emotion (y=-2.93, t(925)=-4.68, p <.01); and that the effect of stress on next day s negative emotion was moderated by trait resilience (y=-2.43, t(925)=-3.46, p <.01) (Ong et al, 2006). A second cohort of subjects (n=34) between the ages of who were widows, were chosen from the same prior research study. Each participant completed pre and post interviews and self-report questionnaires as well as questionnaires at 8, 12, 16, 20, 24, 36, and 48 months. The questionnaires were identical to the questionnaires given to the first cohort. Individuals with low levels of trait resilience had an inverse relationship between positive and negative emotion, with a significant difference between high stress (-.32) and low stress (-.15) days, X 2 (1, n=34)=8.96, p <.01. In contrast, highly resilient participants did not have a significant difference in positive and negative emotion and did not differ between high stress (-.09) and low stress (-.06) days, X 2 (1, n=34)=1.59, p >.05 (Ong et al, 2006). Campbell-Sills, Cohan and Stein (2006) investigated the relationship of resilience to personality traits, coping styles, and psychiatric symptoms in a sample of college students. One hundred and thirty two undergraduates from San Diego State University volunteered to participate in this study. Each participant completed

40 30 a resilience questionnaire (CD-RISC), personality measure (NEO Five Factor Inventory), a measure to determine coping (Coping Inventory for Stressful Situations), the Childhood Inventory for Stressful Situations to retrospectively assess childhood trauma, and the Brief Symptom Inventory to assess symptoms of anxiety and depression over the past week. Resilience had a strong inverse relationship with neuroticism and a strong positive relationship with extraversion and conscientiousness (p s <.001). Hierarchical multiple regression was used to test the potential moderating effect of resilience on psychiatric symptoms and trauma. There was a significant main effect between resilience and emotional neglect on psychiatric symptoms (CD-RISC: β= -0.52, t=-6.78, p < 0.001; CD- RISC x CTQ-EN: β=-0.26, t= -3.57, p < 0.001). While most studies related to resilience have examined the concept as a protective factor against reactions to stress and trauma, Vogt and colleagues (2008) examined the impact of stress reactions on resilience. This study included 1,571 marine recruits who were completing training at Paris Island, South Carolina. The sample included men (n=893) and women (n=678) who completed surveys at the beginning of their training and again two days prior to graduation. The surveys included stress reaction, hardiness, perceived social support and negative affectivity measures. All analyses were separated by gender. Structural equation modeling

41 31 (SEM) was used to evaluate hypotheses regarding the relationships between stress reactions and hardiness for men. The fully saturated model provided the best fit with the data after chi-square difference tests were performed. Higher levels of resilience predicted lower stress reactions and high initial stress reactions predicted a reduction in resilience over the course of basic training. For the women participants, SEM produced a good fit model when the cross-lagged paths were deleted from the model. There was no significant relationship between stress reactions and resilience, which suggested that resilience was not a protective factor for women experiencing stress reactions and that stress reactions did not lower resilience levels over time. This difference between genders may suggest that males with resilience use a better combination of coping factors compared to resilient females (Vogt, Rizvi, Shipherd & Resick, 2008). Mealer and colleagues (2012a) examined the relationship between high levels of resilience and the development of psychological distress in a national sample of critical care nurses. A cross-sectional survey was administered to 744 ICU nurses around the United States. The survey included measures of resilience, PTSD, BOS, anxiety and depression. Overall, 22% of the ICU nurses were highly resilient. The presence of resilience was associated with a significantly lower prevalence of a diagnosis of

42 32 PTSD (8%) compared to those who were not highly resilient (25%, p <.001). There were also fewer reported symptoms of anxiety (8% vs. 21%, p=.003) and depression (2% vs. 14%, p<.001). Burnout syndrome was also significantly lower in all three dimensions for the highly resilient ICU nurses (p<.001 for all three dimensions). The association between resilience and the outcome variables of PTSD, BOS, and symptoms of anxiety and depression was tested using individual multivariable logistic regression model, which adjusted for gender, age, the reason for choosing the nursing profession, shift that the nurse worked, and type of ICU. Being highly resilient was independently associated with a negative diagnosis of PTSD (p<.001, OR=.27, 95% CI= ), the absence of depression symptoms (p=.001, OR=.10, 95% CI= ), absence of anxiety symptoms (p=.006, OR=0.26, 95% CI= ) and the absence of BOS (p<.001, OR=0.22, 95% CI= ). Years practicing. Evidence suggests that the fewer number of years practicing as an ICU nurse is significantly associated with a diagnosis of PTSD and BOS. ICU nurses with PTSD have been practicing an average of 8 years compared to 11.7 years in ICU nurses with only BOS and 19.6 years in ICU nurses with neither BOS or a diagnosis of PTSD (p < ) (Mealer et al., 2009). ICU

43 33 nurses working the night shift are more likely to be positive for symptoms of PTSD (p=0.02) (Mealer et al., 2007). Additionally, when ICU nurses were asked to identify the most common traumatic episodes experienced while working in the critical care environment, the cumulative stress related to feeling overextended due to inadequate nurse to patient ratios was commonly reported (Mealer et al., 2007). While individual nursing factors have been associated with PTSD, a gap exists in the literature related to the effect the individual factors have on resilience and the potential mediating roles these factors have on the development of PTSD through resilience. While resilience is largely an individual adaptive response, small measures incorporated by organizations can make a big difference to help healthcare workers cope with their stressful work environment. These measures include using professional networks to help debrief following an emotional incident, providing a comfortable space within the workplace to decompress and escape the often chaotic ICU environment and creating a learning environment to help acquire practices to help buffer work and home stress (Scholes, 2013). Resilience training. Three randomized, controlled trials of resilience training were identified in the literature and a pilot non-randomized trial.

44 34 Each of these trials highlight specific individual factors that are important for bolstering resilience. These trials examined the effects of resilience training on individual level variables. Waite and Richardson (2003) sought to determine the impact of resilience training on the psychological health of employees by assessing levels of self-esteem, purpose in life, locus of control, job satisfaction and interpersonal relations. Twelve work units from a large government organization participated in the trial with a total of 232 individual participants. Work units were randomly assigned to the resilience training (n=6) or control (n=6) groups. The resilience training group participated in a five week educational program that taught participants how to access, nurture and use resilient strategies. Several measures were administered the week prior to training, the week after training and 10 weeks after the training ended. These measures included demographic items, the Resilience and Reintegration survey, Rosenberg Self-Esteem survey, Multidimensional Locus of Control Scale, Purpose in Life Scale, Health-Promoting Lifestyle Profile to measure interpersonal relations and the Employee Satisfaction instrument. A repeated measure ANOVA test showed a significant interaction between time and group assignment (p< 0.05) and paired samples t-tests showed significant positive changes in the resilient group resilience and reintegration scale (t=-3.425, p < 0.007), the self-esteem survey (t=-

45 , p <0.007), the locus of control scale (t=-6.557, p < 0.007, purpose in life scale (t=-7.445, p < 0.007), and the interpersonal relations scale (t=-4.368, p < 0.007). The reintegration scores, selfesteem scores, purpose in life scores and interpersonal relations scores remained significant at each follow-up time point. The second randomized, controlled resilience study was conducted as a pilot study to determine feasibility and acceptability of resilience training in a group of ICU nurses (Mealer et al., 2014). This was a single-center study with eligible nurses including those who worked at least 20 hours per week at the ICU bedside, had no underlying medical condition that would be a contraindication to exercise, and who scored 82 on the Connor-Davidson Resilience Scale (CD-RISC), which indicated low levels of resilience. A total of 33 eligible ICU nurses were identified but four were excluded because their resilience scores were too high. Of the 29 remaining ICU nurses, 14 were randomized to the intervention arm and 15 were randomized to the control arm. Two subjects withdrew prior to initiation of the 12 week training period. The resilience training group participated in a 2-day educational workshop on resilience, event-triggered counseling sessions that used cognitive-behavioral therapy to promote resilience and challenge cognitive distortions, protocolized aerobic exercise that included exercising three times a week for at least 30 minutes, written exposure therapy that included

46 36 writing about traumatic work experiences 30 minutes a week, and mindfulness-based stress reduction techniques that were practiced at least 3 days a week for 30 minutes. The control group did not participate in this program but completed pre and post surveys and recorded daily exercise. The pre and post surveys were administered to both groups prior to the initiation of the 12-week program and within one month of completing the program. The surveys included demographic questions, a resilience measure, and psychological measures of anxiety, depression, posttraumatic stress disorder and burnout syndrome. The program was feasible with 100% of the participants completing the expected writing exercises, 100% attending the 2-day workshop, 88% completing the expected exercise sessions, 66% completing the mindfulness based stress reduction techniques and attending at least two eventtriggered counseling sessions. The interventions were acceptable with high levels of satisfaction reported for each component. While the study was not powered to determine changes in symptom severity or resilience scores, there were significant increases in resilience scores and decreased symptoms of depression and posttraumatic stress disorder. A pilot study (Foureur et al., 2013), examined the ability to enhance the resilience of nurses and midwives by incorporating a mindfulness-based intervention to decrease depression, anxiety

47 37 and stress. Forty nurses participated in a one-day mindfulnessbased stress reduction (MBSR) workshop and were then asked to practice MBSR daily for 20 minutes over an eight week period of time. There were short term benefits related to acknowledging the impact of stress on cognition, behavior and emotion. There was also benefit in engaging in MBSR and developing mindful practices when experiencing stress in the workplace. Group factors. ICU type. A cross-sectional survey was performed by Mealer and colleagues (2007) to determine the prevalence of PTSD, BOS, anxiety and depression in ICU nurses compared to general medical-surgical nurses. Twenty four percent of the ICU nurses were positive for symptoms of PTSD compared to 14% of the general medical-surgical nurses (p=.03). There was no difference between the two groups on symptom scores of anxiety (18% vs. 22%) or depression (34% vs. 35%). However, some evidence suggested that the type of ICU that the nurse worked in significantly affected symptoms of PTSD. In an ICU unit with a dedicated charge nurse who did not have a patient assignment and could help with the work-flow, there were significantly less symptoms of PTSD (p=.02). In contrast, the ICU unit that did not have a dedicated charge nurse and therefore there was not an extra clinician to

48 38 assist with patient care, trended towards higher symptoms of PTSD (p=.07). In a second study, Czaja, Moss & Mealer (2012) also conducted a cross-sectional survey study of pediatric nurses at a tertiary-care children s hospital. Twenty one percent of nurses met the diagnostic criteria for PTSD with seventy-five percent of those nurses experiencing symptoms for at least three months. There was no statistical difference in PTSD scores based on the type of area the nurse worked in (general medical, general surgical, oncology, PICU and ER). There was also a high rate of burnout syndrome, with approximately half of the subjects having symptoms of emotional exhaustion, thirty-eight percent with symptoms of depersonalization and 46% having symptoms of a lack of personal accomplishment. There were grouping factors that were significantly associated with having PTSD and burnout syndrome. If the nurse had confidence in the other nurses he/she worked with, had confidence with the physician that he/she worked with, felt like a part of a team, and believed other nurses helped each other with patient care, he/she was significantly less likely to have PTSD or burnout syndrome. The data were not analyzed by unit type. In a similar study conducted by Mealer and colleagues (2009), a single-center, cross-sectional survey was administered to all inpatient and outpatient nurses. While symptoms of PTSD,

49 39 burnout syndrome, anxiety and depression were high in all classes of nurses, ICU nurses were significantly more likely to have a diagnosis of PTSD (p=.04). Grouping questions similar to the above study were answered but again the data were not analyzed by unit type. If nurses were more likely to help each other with patient care and if the nurses had confidence in doctors, they were significantly less likely to have PTSD or burnout syndrome. Based on the above three studies, a grouping phenomenon may affect resilience and help mitigate the development of psychological disorders. However, structural equation modeling examining the direct and indirect effects is needed to determine the significance of individual factors and grouping factors (unit type) on resilience. Multigenerational nursing workforce. Generational cohorts refer to groups of individuals who share birth years, historical, political and social events that result in a collective personality as a result of their experiences (Sherman, 2006). Currently five generational cohorts of intensive care unit nurses are practicing in the nursing workforce. The generational cohorts include the Postwar, Boomers I (the Leading-Edge Boomers), Boomers II (the Trailing-Edge Boomers), Generation X, and the Millennials. Table 1 defines these four generational cohorts and summarizes the important historical, political and social events that

50 40 Table 1. Summary of Generational Cohorts (Berkowitz & Schewe, 2011; Myers & Sadaghiani, 2010; Sherman, 2006; Smola & Sutton, 2002) Generation Definition Historical, Political & Social Events Personality/Work Characteristics Postwar Cohort Born The end of World War II, Economic growth, social stability Prefer tradition and formality, face-toface communication, loyal, productive, stoic, reserved, respect for authority and hierarchical organizational structure Boomers I Cohort: The Leading- Edge Boomers Born John F. Kennedy, Martin Luther King and Robert Kennedy assassination, Vietnam, civil rights movement, Watergate era Value individualism, self-respect and a sense of accomplishment, drug culture, live to work, senior nursing staff often representing the leadership Boomers II Cohort: The Trailing- Edge Boomers Born Watergate, the energy crisis and the end of the Vietnam War, economic recession Distrust for authority, distrust of medical providers Generation X Born Structure of the American Family changed (increased divorce rates, single parent homes, latchkey children, massive corporate layoffs), AIDS epidemic Cynacism about the future, self-interests, works to live with greater emphasis on work-life balance, prefer flexible work environment, independent The Millennials: Generation Y Born Age of the internet, increased prevalence of terrorism, violence and drugs. Nurturing parents, structured lives, global generational cohort Place value on excitement, fun and enjoyment, team players, accept diversity, fresh perspectives, overly self-confident, lacking in loyalty and work ethic

51 41 have shaped their personality and work ethic. Currently no literature explores the relationship between generational cohorts and psychological disorders such as PTSD, BOS, anxiety and depression or resilience in critical care nursing. However, the American Psychological Association conducts an annual survey on psychological stress and how stress is perceived and managed based on generational affiliation. Both the Millennials and the Generation X cohorts report an average stress level of 5.4 on a 10 point scale, which is higher than the reported stress levels by the Baby Boomers (4.6) and the Post- War cohorts (3.7). Negative consequences of stress are felt by each generation but the Millennials and the Generation X cohorts are more likely to engage in unhealthy coping behaviors such as drinking, smoking, and overeating (APA, 2012). The Millenials also do not get enough sleep when compared to the Post-War (46% Post-War vs. 29% Millennials). As the nursing workforce ages and we recruit new nurses to fill vacant positions, it is important to understand the differences in perceptions of stress and the common coping mechanisms employed to deal with stress, based on generational cohort affiliation (APA, 2012). As evidenced in the literature, generational cohorts differ in psychological stress and it is hypothesized that there are potential differences in psychological stress experienced by ICU nurses

52 42 based on general cohort affiliation. However, how these group characteristics affect resilience are unknown. Also, the direct and indirect effects that individual and group characteristics have on the development of PTSD through resilience are unknown. Resilience has been shown to mitigate psychological distress in a variety of populations, including critical care nursing and preliminary data suggests resilience training is feasible, acceptable and may reduce psychological distress in this population of nurses. However, there is currently no evidence that identifies individual and group characteristics using structural equation modeling (SEM). This information has the potential to guide resilience interventions and concentrate resilience training efforts based on the significant factors and how they are related to individual resilience and the development of PTSD. Research Question #1: What are the individual and group characteristics that affect the three sub-scales of resilience (Personal Competence, Leadership and Perseverance)? Research Question #2: Do the significant individual and group characteristics have a direct or indirect effect on PTSD mediated through resilience? Summary The Metatheory of Resilience and the modified versions of the Nurse as Wounded Healer theory offers a theoretical framework

53 43 that can be used for research inquiry related to the common psychological disorders experienced by critical care nurses such as posttraumatic stress disorder (PTSD), burnout syndrome (BOS), anxiety and depression. The theoretical frameworks of both the Metatheory of Resilience and the Nurse as Wounded Healer also have the potential to help guide research inquiry related to resilience in critical care nursing. The literature provides important evidence related to the prevalence of psychological disorders in critical care nurses, resilience as a protective capacity that can be learned and can mitigate symptoms of PTSD, the feasibility of conducting resilience training in the healthcare setting, and potential links between individual variables and grouping variables such as generation and type of unit. However, there is a gap in the literature in relation to the identification of individual and group factors that may affect a critical care nurses resilience and the potential mediating role these factors play in subsequent development of PTSD.

54 44 CHAPTER III METHODS The purpose of this chapter is to describe the methodological procedures that will be used to answer the following research questions: Research Question #1 What are the individual and group characteristics that affect the three sub-scales of resilience (Personal Competence, Leadership and Perseverance)? Research Question #2 Do the significant individual and group characteristics have a direct or indirect effect on posttraumatic stress disorder (PTSD) mediated through resilience? Research Design This investigation is a secondary analysis of an established dataset of ICU nurses. The original study was a cross-sectional descriptive correlation design that used a mailed survey to a national sample of ICU nurses to determine the prevalence of PTSD, Burnout syndrome (BOS), anxiety, depression and resilience. Secondary Data Set Original procedures. The original sample was drawn from the membership of selected critical care nurses who were members of the American Association of Critical-Care Nurses (AACN). The AACN is the

55 45 largest specialty organization representing the interests of over 500,000 critical care nurses. AACN provided a mailing list for 3500 randomly selected nurses in the United States and a survey was mailed in an anonymous fashion and according to standard survey principles. A reminder card was sent 2 weeks and 1 month after the initial mailing to improve the response rate (Dilman,1978). A total of 1202 nurses completed and returned the survey. Four hundred and fifty eight surveys were excluded because respondents were not currently working as an ICU nurse, which left a final sample of 744. Data were collected in paper format and then entered into a SPSS statistical database. For the purpose of this secondary analysis the final sample of 744 will be used. Participants. The majority of the sample were women (91%) and the mean age was 43.6 years (SD=11.0). Race was self-reported as Caucasian (83%), Asian (10%), African American (5%), American Indian (1%) and mixed ethnicity/other (1%). The full results of this study have been published elsewhere (Mealer et al, 2012a). See Table 2 for participant demographics. Human subject protection. This study was approved by the University of Colorado Multiple Institutional Review Board. The full results of this study have been published elsewhere (Mealer et al, 2012a).Secondary

56 46 analysis of de-identified data was determined to be not human subjects research. Measures available. The investigation used four validated measures for anxiety and depression, burnout syndrome, posttraumatic stress disorder (PTSD), and resilience. PTSD is a common psychiatric disorder that develops following exposure to a stressful or traumatic experience, which is responded to with fear, helplessness and horror. The three symptom clusters associated with PTSD include re-experiencing the event, which may include nightmares and flashbacks; avoidance of reminders of the event; and a state of hyperarousal or anxiety about normal daily events (Yehuda, 2002). Burnout syndrome (BOS) is a response to emotional and interpersonal stressors that are experienced at work and BOS is characterized by emotional exhaustion, depersonalization, and a lack of personal accomplishment. Resilience allows an individual to appropriately respond to adversity and find new meaning related to the traumatic experience. Resilience is important following trauma and provides the necessary resources to adequate move beyond the experience (Charney, 2004; Davidson, Payne & Connor, 2005; Hoge, Austin & Pollack, 2007; Luthar, Cicchetti & Becker, 2000). The measures included in the survey are the Hospital Anxiety and Depression Scale (HADS),

57 47 Table 2. Participant demographics Category Gender: % Female Result 91 (674/744) Age (mean ± SD) 43.6 ± 11.0 % Exercise regularly 62 (458/744) Marital status (percent): Single Married Other Type of Unit (percent): Medical Surgical Cardiac Cardiothoracic Pediatric Other Highest degree (percent): Associates Bachelors Other 16 (119/744) 67 (502/744) 17 (123/744) 35 (262/744) 10 (77/744) 11 (79/744) 9 (68/744) 4 (29/744) 31 (229/744) 20 (147/744) 57 (421/744) 23 (176/744) Years practiced (mean ± SD) 17.8 ± 11.7 PDS: % PTSD diagnosis 21 (156/740) HADS: % Anxiety positive Anxiety score (mean ± SD) % Depression positive Depression score (mean ± SD) 18 (136/742) 12.5 ± (84/742) 10.6 ± 1.8

58 48 the Posttraumatic Diagnostic Scale (PDS), the Maslach Burnout inventory (MBI), and Connor-Davidson Resilence Scale (CD-RISC). A description of the instruments follows. Hospital Anxiety and Depression Scale. The Hospital Anxiety and Depression Scale (HADS) is a 14-item self-report screening scale originally developed to indicate the possible presence of anxiety and depression states in the setting of a medical, nonpsychiatric outpatient clinic. HADS consists of a 7-item anxiety subscale and a 7-item depression subscale. A score of > 8 identifies those with a positive history for anxiety and/or depression (Aylard et al, 1987; Bjelland et al, 2002; Mealer et al, 2009). The validity of the HADS has been extensively studied for identifying anxiety and depression disorders in a variety of populations including the general population, general practice and psychiatric patients (Aylard et al, 1987; Bjelland et al, 2002; Mealer et al, 2009; Mealer et al, 2012a). In this analysis, the Cronbach s alpha was 0.86 for the Hospital Anxiety and Depression Scale (HADS). Posttraumatic Diagnostic Scale. The Posttraumatic Diagnostic Scale (PDS) is a validated, self-report tool that yields both a PTSD diagnosis according to the diagnostic and statistical manual of mental disorders fourth edition (DSM-IV) criteria and a measure of PTSD symptom severity. While the gold standard for diagnosing PTSD is the clinician-administered PTSD scale (CAPS),

59 49 the PDS is highly correlated (r=0.01) with the clinician-rated measures diagnosing PTSD (Foa et al, 1997; Mueser, 2001; Sheeran & Zimmerman, 2002). The PDS consists of a screener for Criterion A events, a checklist of twelve traumatic events (including an other category), and Criteria B, C, and D that assess the seventeen DSM-IV symptoms by using a four-point scale (0=not at all or only one time; 3=five or more times a week/almost always). A total score is calculated by: having had a traumatic event, feeling helpless or terrified during the event (criterion A), and at least one episode of a re-experiencing symptom (criterion B), three avoidance symptoms (criterion C), and two arousal symptoms (criterion D). Symptom severity can be calculated by summing scores of the 17 items addressed in Criterion B, C, and D (Mealer et al, 2009) For the purpose of this study, nurses were asked complete Criterion A on traumatic events that were only experienced while working in the ICU environment, and nurses were asked to fill out Criteria B, C, and D only based upon those events. The PDS also has a section that asks how the problems rated in Criterion B, C, and D have interfered with any of the following areas of life functioning within the past month: work, household chores and duties, relationships with friends, fun and leisure activities, schoolwork, relationships with your family, sex life, general satisfaction with life, and overall level of functioning in all

60 50 areas of your life. The PDS is a well-accepted and validated survey instrument to diagnose individuals with PTSD (Foa et al, 1997; Mealer et al, 2009) with high internal consistency reliability (Cronbach s α range ). In addition, test-retest reliability correlation coefficients of the total PDS score demonstrated satisfactory reliability: 0.83 for total symptom severity, 0.77 for reexperiencing, 0.81 for avoidance, and 0.85 for arousal (Foa et al, 1997; Mealer et al, 2009; Mealer et al, 2012a). Maslach Burnout Inventory. The Maslach Burnout Inventory (MBI) is a 22-item self-report questionnaire consisting of three independently scored dimensions (Leiter & Maslach, 1999; Maslach et al, 2001; Mealer et al, 2009). The MBI is copyright protected and permission was obtained prior to mailing the surveys. The questionnaire includes questions regarding the frequency of experiencing certain feelings related to a participant s work environment on a seven-point Likert scale. The emotional exhaustion (EE) scale, consisting of 9 items, identifies those individuals who are emotionally exhausted or overextended at work. The depersonalization (DP) scale, consisting of five items, identifies those individuals who have an impersonal response towards recipients of their efforts. The personal accomplishment (PA) scale, consisting of 8 items, assesses lack of accomplishment and success related to work. This investigation scored subjects as

61 51 having moderate to high levels of burnout syndrome with the following values: EE greater than seventeen, DP greater than seven, and PA above thirty-one (Maslach et al, 1996; Mealer et al, 2009; MBI-Human Services Survey, copyright 1986 by CPP, Inc.). The MBI is a burnout tool that has been tested extensively for reliability and validity and internal consistency. Previous reported values for Cronbach s coefficient alpha have been in the range of (Gil-Monte, 2005; Kalliath et al, 2000; Leiter & Maslach, 1999). In thisanalysis, the Cronbach s alpha for the MBI Human Service Survey was Alpha on each of the 3 dimensions was: EE α = 0.53, DP α = 0.72, and PA α = 0.91(Mealer et al, 2012a). Connor-Davidson Resilence Scale. The Connor-Davidson Resilience Scale (CD-RISC) was utilized to assess resilience (Connor & Davidson, 2003). The CD-RISC was developed as a short self-report assessment to quantify resilience and as a clinical measure to assess treatment response. The CD-RISC is copyright protected and permission was obtained prior to mailing the surveys. It is a twenty-five item self-report scale with total score ranges from zero to one hundred. Higher scores reflect greater resilience. Resilience is defined as a CD-RISC score of >80 with a median score of 82. Highly resilient is defined as one standard deviation greater than the mean and therefore a score of 92 is defined as a positive score for being highly resilient (Campbell et al, 2009;

62 52 Connor & Davidson, 2003). The CD-RISC has been extensively used in community samples, primary care outpatients, general psychiatric outpatients, a clinical trial of generalized anxiety disorder, and two clinical trials of PTSD. The CD-RISC maintains excellent reliability (Cronbach s alpha 0.89) and a test-retest reliability correlation of 0.87 (Connor & Davison, 2003). For this analysis, the Cronbach s alpha for the Connor-Davidson Resilience Scale was 0.92 (Mealer et al, 2012a). For the purpose of this study, an abbreviated version of the CD-RISC will be used based on the results of an exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) that was conducted on this sample of ICU nurses. The CD-RISC was abridged based on statistical and theoretical issues. Statistical issues included excluding questions based on normal distribution, low correlations with the total, significant cross-loadings and potential factors with less than three measured variables. The data set was split into two samples with the first split being used for the EFA and the second split being used for the CFA. The final abridged version of the CD-RISC included sixteen of the original twenty-five items. The EFA identified three factors that had salient loadings and conceptual coherence and those factors were labeled personal competence, perseverance and leadership (Table 3).

63 53 Table 3. CD-RISC Exploratory Factor Analysis Item: Personal Competence (Factor 1) Perseverance (Factor 2) Leadership (Factor 3) CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD The three factors or sub-scales will be measured separately and as a total score. The CFA model found that the three-factor model was statistically superior to the other tested models (Figure 4). The alpha value for the total score was α=.90 indicating good reliability and the alpha values for each subscale were:

64 54 perseverance α=.83, leadership α=.72 and personal competence α=.81. Statistical Analyses All analyses will be conducted using structural equation modeling (SEM) in Mplus and SPSS 20. SEM in Mplus will be used because of the way it deals with dichotomous and continuous variables in the same equation. In addition, Mplus allows simultaneous testing of the effects of characteristics mediated through resilience on PTSD. The individual and group characteristics that are included in the database being used to answer the research questions include both dichotomous and continuous variables and there are several available corrections within Mplus that can be applied based on these research questions. The following describes the analysis techniques will be used to answer the research question: Research question one. The first research question is what are the individual and group characteristics that affect resilience? To answer the research question, bivariate correlations in SPSS will be analyzed to determine the significant individual and group characteristics that affect the three subscale of resilience (personal competence, leadership and perseverance). The measurement variables are

65 55 identified by the type of characteristic they represent in Table 4. The individual characteristics that will be included in the analysis include: whether the ICU nurse has children (yes or no), whether the ICU nurse exercises regularly (yes or no), highest nursing degree (diploma, associates, bachelors, masters, or other), years practicing as a nurse, nurse-patient ratio (number of patients per nurse), and shift worked (12 hour days, 12 hour nights, 8 hour days, 8 hour evenings, 8 hour nights). The group characteristics that will be included in the analysis are type of ICU unit (medical, surgical, cardiac, cardiothoracic, and other ICU category) and generational cohort grouping (Postwar, Boomers I, Boomers II, Generation X and Millennials). Since there are potential predictors with more than two categories, dummy codes will be created. The following variables will be recoded: generational cohort, degree, nurse-patient ratio, shift and unit type. Several variables will be created and a baseline group will be chosen to compare the other groups against. The baseline group will be based on predefined hypotheses and/or the group that represents the majority of people in the sample (Field, 2009). Working night shift was associated with an increased prevalence of PTSD (Mealer et al., 2007) and higher proportions of BSN prepared nurses in a hospital setting has been associated with

66 56 reduced surgical patient mortality (Aiken et al., 2003) and overall patient mortality (Estabrooks et al., 2005). Unit type has not been explored but it is the author s hypothesis that the medical intensive care unit is more stressful because of the patient population. Table 5 defines the baseline or comparator group for each of the categorical variables that will be recoded. If any of the dummy grouping categories are significantly associated with the resilience sub-scales, all of the dummy grouping categories will be retained to answer research question two (i.e. if GC1 is significant, GC1, GC2 and GC3 will be included in the mediation model to determine indirect effects). The sample size is adequate (n=744) as it is recommended to have at least fifty cases per independent variable (Grimm & Yarnold, 1995). A p value of.05 will be considered statistically significant in the comparison of results. Bivariate correlations will be analyzed to determine the individual and group variables that are significantly correlated with each of the three sub-scales of resilience, Personal Competence, Leadership and Perseverance. Figure 5 diagrams the bivariate correlations between the variables and the three sub-scales of resilience (Personal Competence, Leadership and Perseverance).

67 Figure 4. Confirmatory Factor Analysis 3-Factor Model of CD-RISC 57

68 58 Table 4. Individual and Group Characteristics Characteristic Type Measurement Variable Individual Children Exercises Regularly Highest Nursing Degree Years Practicing Nurse-Patient Ratio Shift Group Unit Type Generational Cohort Table 5. Dummy Coding Baseline Group Hypothesis-Driven Or Group Majority Generational Cohort Boomers II Group Majority Degree BSN Hypothesis Driven Nurse-Patient Ratio 1-2 Group Majority Shift Days (7:00-3:00/7:00- Hypothesis Driven 19:00) Unit Type Medical Hypothesis Driven

69 59 Figure 5. Bivariate Correlations GC=generational cohort, CH=children, DE=highest nursing degree, YP=years practicing, RA=nurse-to-patient ratio, SH=shift, UN=type of ICU unit, EX=exercise, PC=Personal Competence, LD=Leadership, PV=Perseverance To analyze this question a meditational model using structural equation modeling (SEM) within Mplus, will be used for each of the individual and group characteristics that significantly affect resilience in research question one (Figure 6). A mediator is a variable that accounts for all or some of the relationship between an independent variable (predictor) and the dependent variable (outcome)(baron & Kenny, 1986). The Mplus direct and indirect effects of SEM are defined for a continuous meditational variable (resilience) and a binary outcome variable (PTSD). Logit regression

70 60 for the development of PTSD will be used because there is the assumption that the binary outcome of PTSD is rare in this population (Muthen, 2011).This model will propose that each significant characteristic has a direct effect on the development of PTSD and that each significant characteristic has an indirect effect on the development of PTSD through the three sub-scales of resilience (Personal Competence, Leadership and Perseverance)(mediator) (Kim, Sandler & Tein, 1997). β weights, standard errors and p values will be evaluated to determine which individual and group variables are significant and to model the direct and indirect paths between the significant variables and the development of PTSD. Odds ratios and 5%-95% confidence intervals will be assessed to determine which of the significant variables have a significant indirect effect. The odds ratio, an indicator of the change in odds, allows for interpretation of the logistic regression analyses. The odds ratio is similar to the β weight and refers to the change in odds from each unit change in the predictor variable (Field, 2009). If the odds ratio value is greater than one, it indicates that as the predictor variable increases, so does the outcome variable. In contrast, if the odds ratio is less than one, as the predictor variable increases, the odds of the outcome occurring decreases (Fields, 2009). Determining the significant indirect effects is a feature that is only available in Mplus.

71 61 Figure 6. Mediation Model for SEM There are 3 steps taken to test the meditational model in Mplus. The first step tests whether the independent variable cause the dependent variable, the second step tests whether the independent variable causes the potential mediator and the third step tests whether the mediator causes the dependent variable controlling for the independent variable (Figure 7). Step 1: Y=i 1 + cx + e 1 Step 2: M=i 2 + ax + e 2

72 62 Step 3: Y=i 3 + c'x + bm + e 3 The significance of the mediated effect will be assessed by determining the Odds Ratio and 5%-95% confidence intervals on the significant variables identified in step 3. Figure 7. Mediation Model Steps for SEM Summary Bivariate correlations in SPSS and SEM within Mplus will be used to answer the research questions. Research question one

73 63 will be answered by determining the significant bivariate correlations between each individual and group factor with the three sub-scales of resilience (Personal Competence, Leadership and Perseverance). Each individual factor and group factor will also be tested for its direct effect on the three sub-scales of resilience. The second research question will be answered using a mediation logit model with a continuous mediator and a binary outcome variable. This mediation model will test the direct and indirect effects that the significant characteristics in research question one have on the development of PTSD through resilience.

74 64 CHAPTER IV RESULTS This chapter presents the results of the data analyses. The primary purposes of this chapter are to present the results related to the evaluation of individual and group variables that affect the three sub-scales of resilience (Personal Competence, Leadership and Perseverance) and to evaluate whether the significant variables have a direct or indirect effect on the development of PTSD mediated through each of the resilience sub-scales. The following sections will include 1) demographic characteristics of the sample; 2) research question one: individual and group characteristics that significantly affect the three sub-scales of resilience and; 3) research question two: assessment of the direct and indirect effects that the significant variables have on the development of PTSD mediated through the three sub-scales of resilience. Characteristics of the Sample The secondary database analysis consisted of 744 ICU nurses in the United States. Overall participants were female (N=674, 91%) with a mean age of 43.6 (SD=11.0) and mean years practiced of 17.8 (SD=11.7). Twenty one percent (N=158) were positive for a diagnosis of PTSD, 18% were positive for anxiety (N=138) and 11% were positive for depression (N=86). See Table 2

75 65 in Chapter Three for a more detailed description of the sample demographics. The remainder of this chapter will answer the research questions proposed in Chapter One. The research questions will be repeated in this chapter and immediately precede the corresponding result section. Research Question One The first research question was what are the individual and group characteristics that affect resilience? To answer the research question, the individual and group characteristics were correlated with the three sub-scales of resilience (Personal Competence, Leadership and Perseverance). As detailed in Chapter Three, dummy codes were assigned to the categorical predictor variables that had more than 2 categories. This included the generational cohort, degree, nurse-to-patient ratio, shift and unit variables (Table 5). Based on the correlations, three dummy grouping variables were chosen for further analysis. All generational cohort dummy variables were coded to allow comparison to the Boomer II group specifically; Generation X (GC1) compared to Boomer II, Millenials (GC2) compared to Boomer II and Postwar/Boomer I (GC3) compared to Boomer II. The educational degree variable was recoded into two dummy grouping variables with the comparison

76 66 between ICU nurses that held a bachelor of science in nursing (BSN) degree and other educational degrees. Associate degree in nursing (ADN) and Diploma degrees (DE1) were compared with BSN and those with a Master s degree and higher (DE2) were compared with BSN. The nurse-to-patient ratio variable was recoded into one dummy variable that was compared with nurses who took care of two or less patients. Nurse-to-patient ratios > 2 were compared with nurse-to-patient ratios of 2. The primary shift worked variable was categorized into those who worked day shift (7:00-19:00 and 07:00-3:00) and those who worked night shift (15:00-23:00 and 19:00-07:00). The comparison group was day shift; therefore night shift was compared to day shift. The final variable that was recoded into dummy codes was the type of unit variable. Four dummy grouping variables were chosen and all contrasts were made with the medical intensive care unit (MICU), serving as the baseline comparison. The cardiac intensive care unit (CICU) was compared with the MICU (UN1), the surgical intensive care unit (SICU) was compared with the MICU (UN2), the cardiothoracic intensive care unit (CTICU) was compared with the MICU (UN3) and all other types of intensive care units was compared with the MICU (UN4). (See Appendix A for additional specifics and definitions of dummy codes). A priori it was decided

77 67 that if one of the dummy codes was significant, they would all be entered into the final model to answer research question two. Significant individual characteristics. The significant individual characteristics that affected resilience included whether the ICU nurse had children, the number of years practicing as an ICU nurse and the type of degree earned by the ICU nurse. All three of the individual characteristics that were significant correlated with the Leadership sub-scale of resilience. Having children was significantly correlated with a correlation of 0.01 (p= 0.01). The number of years practicing as an ICU nurse had a correlation of 0.09 (p= 0.02). Finally, ICU nurses with a master s degree or higher (DE2) had a correlation of 0.15 (p= 0.001) and was the strongest dummy code when compared with the BSN baseline group. As referenced above, since one of the degree dummy variables was significant, both were retained for the final model (DE1=associates degree or diploma and DE2=master s degree or above). Significant grouping characteristics. The significant grouping characteristics that affected resilience included the type of unit that the ICU nurse worked on and generational cohort. The type of unit was significantly

78 68 correlated with the Personal Competence sub-scale of resilience and the generational cohort was significantly correlated with the Leadership sub-scale of resilience. The types of units include the CICU, SICU, CTICU and all other ICU s compared with the baseline MICU group. The CICU (UN1) was the only unit with a significant correlation of (p= 0.04). The remaining units were not significantly correlated: SICU had a correlation of 0.06 (p= 0.12), CTICU correlation was -.02 (p= 0.60) and all other units correlated (p= 0.44) with the Personal Competence sub-scale of resilience. Since the CICU was significantly correlated, all four unit types were retained for the final model to answer question two. The strongest dummy coded group for unit type was the CICU compared with the MICU. Table 6 presents the Pearson correlations and 2-tailed p-values. The three generational cohorts included Generation X (GC1), Millenials (GC2) and Postwar/Boomers I all compared with Boomers II (GC3). The Millenials group was the only variable that was significantly correlated (r=-0.12, p=0.001) with the Leadership sub-scale but all three groups were retained to answer question two: Generation X (GC1) (r= 0.02, p= 0.51), Postwar/Boomers I (GC3) (r= 0.01, p= 0.73) and the Millenials (GC2). There were no significant individual or group characteristics that affected the perseverance sub-scale of resilience.

79 69 Consequently the perseverance sub-scale was not used in further analyses. Figure 8 models the significant paths between the individual and group characteristics and the resilience sub-scales. The nonsignificant paths were not included in the figure.three of the unit variables had significant paths to the Personal Competence subscale of resilience. The comparison of the CICU to the MICU had a significant β of (0.03 S.E. and p= 0.02), as did the CTICU, β of (0.03 S.E. and p= 0.005) and the Other ICU s had a β of (0.03 S.E. and p value 0.010) when compared to the MICU. The SICU and the MICU comparison was the only one that did not have a significant path (β= -0.01, 0.03 S.E., p= 0.76). Only one variable had a significant path to the Leadership sub-scale of resilience. Those with a master s degree or higher as compared to those with a BSN had a β of 0.12 (0.03 S.E., p= 0.001). The remaining variables were not significant and included nurses with an ADN or Diploma (DE1) (β 0.04, S.E. 0.03, p=0.21), having children (β 0.06, S.E. 0.03, p= 0.07), years practicing (β- 0.02, S.E. 0.05, p= 0.67), Generation X dummy variable (GC1) (β , S.E..04, p= 0.91), Millenial dummy variable (GC2) (β-0.06, S.E. 0.05, p= 0.25), Postwar/Boomers I dummy variable (GC3) (β 0.006, S.E. 0.03, p= 0.85).

80 70 Table 6. Bivariate Correlations between Individual and Group Factors with Resilience Sub-Scales Corellation Personal Competence Leadership Perseverance CH ** 0.02 YP * 0.03 EX GC GC *** GC DE DE *** 0.06 RA SH UN1-0.08* UN UN UN CH=children, YP=years practicing, EX=exercise, GC1=generation X compared to boomers II, GC2=millenials compared to boomers II, GC3=postwar/boomers I compared to boomers II, DE1=ADN/diploma compared to BSN, DE2=MS or above compared with BSN, RA1=>2 patients per nurse compared to 2 patients per nurse, SH1=night shift compared to day shift, UN1=cardiac ICU compared to medical ICU, UN2=surgical ICU compared to medical ICU, UN3=CT surgery ICU compared to medical ICU, UN4=other ICU s compared to medical ICU Research Question Two The second research question was whether the significant individual and group characteristics have a direct or indirect effect on PTSD mediated through resilience?

81 71 Figure 8. Individual and group variables that significantly affect resilience sub-scales To answer whether there was a direct effect of the significant individual and group variables on PTSD through resilience, bivariate correlations were evaluated. Table 7 presents the Pearson correlations and p-values related to the potential direct effect that each of the individual and group variables have on the development of PTSD. Based on this evaluation, each of the three resilience sub-scales (Personal Competence, Leadership and Perseverance), whether the ICU nurse had children and the number of years

82 72 practicing as an ICU nurse had significant direct effects on the development of PTSD. Table 7 represents the direct effects that each of the individual and group variables had on the development of PTSD. The significant direct effects included the Personal Competence sub-scale of resilience, Leadership sub-scale of resilience, Perseverance sub-scale of resilience, whether the ICU nurse had children and the number of years practicing as an ICU nurse. Each of the significant direct effect had a β indicating an inverse relationship between the variable and the development of PTSD. Based on the odds ratio, ICU nurses with Personal Competence were 27% less likely to develop PTSD than those without Personal Competence. Those with Leadership were 20% more likely to develop PTSD and those with Perseverance were 11% less likely to develop PTSD compared with those who did not have Leadership and Perseverance. Whether the ICU nurse had children was also significant indicating that those who had children were 40% less likely to develop PTSD compared with those who did not have children. Finally, the years practicing as a nurse significantly effects the development of PTSD. As years experience increased, ICU nurses were 3% less likely to develop PTSD.

83 73 Table 7: Direct effects of Individual and Group Factors on the Development of PTSD Variable β/s.e./p value Odds ratio %5, 95% CI PC -0.45/0.06/*** , 0.79 LD 0.22/0.06/*** , 1.34 PV -0.17/0.06** , 0.96 CH -0.11/0.05* , 0.89 YP -0.17/0.09* , 1.00 GC1-0.02/ , 1.51 GC2-0.11/ , 1.14 GC3-0.11/ , 0.96 DE1 0.12/ , 1.63 DE2-0.74/ , 1.11 UN / , 1.73 UN2 0.03/ , 2.22 UN3-0.03/ , 1.53 UN4 0.07/ , 2.03 β=beta weight, S.E.=standard error, CI=confidence interval PC=personal competence, LD=leadership, PV=perseveranceCH=children, YP=years practicing, EX=exercise, GC1=generation X compared to boomers II, GC2=millenials compared to boomers II, GC3=postwar/boomers I compared to boomers II, DE1=ADN/diploma compared to BSN, DE2=MS or above compared with BSN, RA1=>2 patients per nurse compared to 2 patients per nurse, SH1=night shift compared to day shift, UN1=cardiac ICU compared to medical ICU, UN2=surgical ICU compared to medical ICU, UN3=CT surgery ICU compared to medical ICU, UN4=other ICU s compared to medical ICU * p 0.05, ** p 0.006, *** p=0.001 Figure 9 models the significant direct effects. The nonsignificant direct effects paths were not shown in this model. Negative beta scores indicate an inverse relationship between the

84 74 predictor variables and outcome variable. The resilience sub-scale direct effects are presented for reference. Figure 9. Significant Direct effects on the Development of PTSD The Personal Competence and Perseverance resilience sub-scales had negative beta weights indicating an inverse relationship between resilience and the development of PTSD. This suggests that as Personal Competence and Perseverance increased, PTSD decreased. The leadership sub-scale of resilience had a positive beta weight suggesting that as Leadership increased, PTSD increased. Having children also had a negative beta weight suggesting that those with children are less likely to develop PTSD. The

85 75 negative pathway for years practicing as an ICU nurse suggests that as years experience increases, the likelihood of developing PTSD decreases. To answer whether the significant individual and group variables had an indirect effect on the development of PTSD through the resilience sub-scales, a mediation model was tested in Mplus based on the significant variables that were outlined in research question one (Figure 8). Mplus allows the determination of the significance of the full indirect path to PTSD, which is not available in other statistical software package. Three of the unit grouping variables (UN1=cardiac, UN3=CT surgery and UN4=other) had significant indirect effects on PTSD through the personal competence (PC) sub-scale of resilience when compared to the MICU (Figure 10). The beta weight and standard error for the significant units are as follows: UN1 (cardiac) -.07β/.02 S.E., UN3 -.08β/.03 S.E. and UN4 -.08β/.03 S.E. The odds ratios (5%, 95% (CI)) were -.65 (-1.11,-.19) for UN1, -.82 (-1.31, -.34) for UN3 and -.50 (-.82, -.18) for UN4 (Table 8). Therefore, the type of unit that an ICU nurse works has a significant effect on the development of PTSD through the Personal Competence sub-scale of resilience. ICU nurses who work in the CTICU are 18% less likely to develop PTSD, CICU are 35% less likely to develop PTSD and those who work in Other ICU s are 50% less likely to develop PTSD. The

86 76 MICU represents a grouping characteristic that significantly increases the risk of the development of PTSD in ICU nurses. Figure 10. Indirect Effect of Unit on the development of PTSD Mediated through the Personal Competence Sub-Scale of Resilience The only significant indirect variable through the leadership sub-scale was DE2 (MS/other) with.12 β/.03 S.E (Figure 11). The odds ratio (CI; 5%, 95%) was.82 (.50, 1.13). The positive β indicates a positive relationship between DE2 and the development of PTSD (Table 9). ICU nurses with a master s degree or higher were 18% more likely to develop PTSD than those with a BSN.

87 77 Figure 12 shows the full model with the significant direct and indirect effects. Figure 11. Indirect effect of Degree on the Development of PTSD Mediated through the Leadership Sub-Scale of Resilience Table 8. Significant Indirect Effects on PTSD through Personal Competence Sub-Scale of Resilience Personal Competence (PC) Estimate UN1 CICU compared with MICU UN2 SICU compared with MICU UN3 CTSICU compared with MICU UN4 Other compared with MICU -0.07/0.03* -0.01/ /0.03** -0.08/0.03** (β)/s.e. Odds Ratio n/a (5%, (-1.1, -.12) (-1.3, -0.34) (-0.82, -0.18) 95% CI) β=beta weight, S.E.=standard error, CI=confidence interval *p value 0.05, **p value 0.01

88 78 Table 9. Significant Indirect Effects on PTSD through Leadership Sub-Scale of Resilience Leadership (LD) Estimate CH YP GC1 GC2 GC3 DE1 DE2 0.06/ -0.02/ / -0.06/ 0.006/ 0.04/ 0.12/0.0 (β)/s.e * Odds Ratio (5%, 95% CI) N/A N/A N/A N/A N/A N/A 0.82 (0.50, 1.1) β=beta weight, S.E.=standard error, CI=confidence intervals * p 0.05 Summary In summary, the analysis included a sample size of 744 ICU nurses from the United States. The three sub-scales of resilience (Personal Competence, Leadership and Perseverance), whether the ICU nurse had children and the number of years the nurse had been practicing in the ICU had a direct effect on the development of PTSD. The mediator model identified the type of unit as having an indirect effect on the development of PTSD through the Personal Competence sub-scale of resilience, and having children, years practicing, generational cohort and type of degree as having an indirect effect on the development of PTSD through the Leadership sub-scale of resilience. When the significance of the indirect effects were measured, unit type remained significant as an indirect effect on Personal Competence and degree remained significant as an indirect effect on Leadership.

89 Figure 12. Significant Direct and Indirect Effects of Individual and Group Factors on the Development of PTSD through the 3 Sub- Scales of Resilience 79