<%BANNER%>

Psychoneuroimmunologic Predictors of Post-Surgical Outcome in Women with Endometrial Cancer

Permanent Link: http://ufdc.ufl.edu/UFE0022599/00001

Material Information

Title: Psychoneuroimmunologic Predictors of Post-Surgical Outcome in Women with Endometrial Cancer
Physical Description: 1 online resource (164 p.)
Language: english
Creator: Jensen, Sally
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: cancer, cortisol, endometrial, gynecologic, psychoneuroimmunology, surgical, women
Clinical and Health Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Endometrial cancer is the most common gynecologic cancer in the United States and is often co-morbid with conditions that confer high risk for post-surgical complications. Among surgical populations, psychosocial distress and maladaptive coping are associated with poorer surgical recovery. Similarly, cortisol, a stress hormone, may serve as an important mediator of the relationship between psychosocial factors and impaired immunity during the peri-operative period. Few studies have examined psychoneuroimmunologic predictors of oncologic surgical. The present study explored the relations among pre-surgical psychosocial stress, pre-surgical emotional support, pre-surgical diurnal cortisol slope, and indices of surgical outcome (e.g., length of hospitalization, severity of post-surgical complications, time to post-surgical ambulation, and post-surgical systemic immune response), among 75 women undergoing total abdominal hysterectomy with bilateral salpingo oophorectomy (TAH-BSO) for suspected endometrial cancer. It was hypothesized that a less complicated surgical recovery would be predicted by less pre-surgical psychosocial stress, more normal diurnal cortisol production, and greater emotional support. Participants completed a pre-surgical psychosocial interview and collected pre-surgical saliva samples for analysis of diurnal cortisol output. Surgical outcome data was abstracted from medical records. Greater perceived emotional support from primary support person was significantly associated with less elevated post-surgical WBC count and was marginally associated with lower post-surgical pain ratings. More elevated pre-surgical indices of cortisol were significantly associated with more elevated post-surgical WBC count and lower post-surgical pain ratings. Contrary to hypothesis, psychosocial stress was not significantly related to indices of surgical recovery or indices of cortisol. Length of hospitalization, severity of post-surgical complications, and time to post-surgical ambulation were not significantly associated with psychoneuroimmunologic predictors. The findings of the present study extend the literature examining psychoneuroimmunologic predictors of surgical outcome by identifying relations among pre-surgical emotional support, pre-surgical indices of cortisol, post-surgical WBC count, and post-surgical pain ratings. They provide preliminary data for future research to examine psychoneuroimmunologic relations during the peri-operative period among women with endometrial cancer.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sally Jensen.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Pereira, Deidre B.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022599:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022599/00001

Material Information

Title: Psychoneuroimmunologic Predictors of Post-Surgical Outcome in Women with Endometrial Cancer
Physical Description: 1 online resource (164 p.)
Language: english
Creator: Jensen, Sally
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: cancer, cortisol, endometrial, gynecologic, psychoneuroimmunology, surgical, women
Clinical and Health Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Endometrial cancer is the most common gynecologic cancer in the United States and is often co-morbid with conditions that confer high risk for post-surgical complications. Among surgical populations, psychosocial distress and maladaptive coping are associated with poorer surgical recovery. Similarly, cortisol, a stress hormone, may serve as an important mediator of the relationship between psychosocial factors and impaired immunity during the peri-operative period. Few studies have examined psychoneuroimmunologic predictors of oncologic surgical. The present study explored the relations among pre-surgical psychosocial stress, pre-surgical emotional support, pre-surgical diurnal cortisol slope, and indices of surgical outcome (e.g., length of hospitalization, severity of post-surgical complications, time to post-surgical ambulation, and post-surgical systemic immune response), among 75 women undergoing total abdominal hysterectomy with bilateral salpingo oophorectomy (TAH-BSO) for suspected endometrial cancer. It was hypothesized that a less complicated surgical recovery would be predicted by less pre-surgical psychosocial stress, more normal diurnal cortisol production, and greater emotional support. Participants completed a pre-surgical psychosocial interview and collected pre-surgical saliva samples for analysis of diurnal cortisol output. Surgical outcome data was abstracted from medical records. Greater perceived emotional support from primary support person was significantly associated with less elevated post-surgical WBC count and was marginally associated with lower post-surgical pain ratings. More elevated pre-surgical indices of cortisol were significantly associated with more elevated post-surgical WBC count and lower post-surgical pain ratings. Contrary to hypothesis, psychosocial stress was not significantly related to indices of surgical recovery or indices of cortisol. Length of hospitalization, severity of post-surgical complications, and time to post-surgical ambulation were not significantly associated with psychoneuroimmunologic predictors. The findings of the present study extend the literature examining psychoneuroimmunologic predictors of surgical outcome by identifying relations among pre-surgical emotional support, pre-surgical indices of cortisol, post-surgical WBC count, and post-surgical pain ratings. They provide preliminary data for future research to examine psychoneuroimmunologic relations during the peri-operative period among women with endometrial cancer.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sally Jensen.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Pereira, Deidre B.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022599:00001


This item has the following downloads:


Full Text

PAGE 1

1 PSYCHONEUROIMMUNOLOGIC PREDICTORS OF POST-SURGICAL OUTCOME IN WOMEN WITH ENDOMETRIAL CANCER By SALLY ELIZABETH JENSEN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Sally Elizabeth Jensen

PAGE 3

3 To Fiona Jensen

PAGE 4

4 ACKNOWLEDGMENTS The dissertation process does not occur in a vacuum ; rather, it is only through the guidance, support, and involvement of others th at such an endeavor is possible. First and foremost, I would like to thank my dissertation chair and mentor, Dr. Deidre Pereira, for her expert guidance, support, and advocacy from this studys inception to its completion. Dr. Pereira graciously provided me with the opportunity to c onduct this dissertation with in the context of her program of research, and allowed me to utiliz e laboratory resources, w ithout which this study would not have been possible. More importantly throughout my years as a graduate student at the University of Florida, Dr. Pereira has been a mentor to me in the truest sense of the word. Her mentorship has not only addr essed issues related to the c onduct of psychosomatic research and the provision of clinical se rvices in the medical setting, bu t has also addressed important areas of professional and personal development. Her work and mentorship have been a true inspiration for me and will always be a valu able influence on me, both professionally and personally. For that, I cannot thank her enough. I would also like to thank my colleagues in the Psycho-Oncology Laboratory at the University of Florida. Among my colleag ues, Ms. Stacy Dodd deserves special acknowledgement, for without her generous assistance and support on occasions too numerous to list, this project would not have succeeded. Ms. Dodd has been an ideal colleague whose keen intellectual contributions and work ethic have served to strengthen the quality of this dissertation. Moreover, I count myself lucky fo r her friendship. I would also like to extend my appreciation to Ms. Lindsey Boegehold, Mr. Timothy Sannes, Mr Patrick Broderick, Ms. Sophie Chrisomalis, Ms. Amber Martin, Ms. Melissa Hosonitz, Ms. Lauren Yuill, Ms. Natalie Cross, Ms. HaNa Kim, Dr. Sarah Rausch, Dr. Evangelina Banou, and Dr. Monica Cortez-Garland, whose work in the Psycho-Oncology laboratory has also contributed to this disserta tion. I would also like to thank

PAGE 5

5 my colleagues in Obstetrics and Gynecology, including Dr. Linda Morgan (University of Florida), Ms. Inslee Baldwin (University of Flor ida), and Dr. Daylene Ripley (North Florida Memorial Hospital), as well as all of the nurses, residents, and medical students at the University of Florida Department of Obstetrics and Gyneco logy for their support of this research. I would also like to thank Dr. Shawn Kneipp, of the Univ ersity of Florida College of Nursing, for her consultation related to salivary co rtisol assays, as well as the Bi obehavioral Research Center in the University of Florida College of Nursing, for generously providing space and resources for storage and preparation of salivar y cortisol. I would also like to acknowledge the members of my dissertation committee, Dr. Mich ael Robinson, Dr. David Janicke, and Dr. Barbara Curbow, for their valuable contributions a nd recommendations which have greatly enhanced the quality of this project. In addition to colleagues and faculty at the University of Florida, I would like to extend my appreciation to my supervisors duri ng my pre-doctoral inte rnship year at the Vanderbilt-VA Internship Consortium, Dr. Mich ele Panucci, Dr. Saundra Saporiti, Dr. Paul Lima, and Dr. Daniel Kearns, for their support of this research, without which it would have been significantly more difficult to complete th is project given the geogr aphic distance between Tennessee and Florida. I would also like to thank th e many loved ones, especially Mom, Dad, and Kate, but also including Barb, Beth, the Gabrys iaks, Tommy, David, Marcia, Elena, Lori, Tonia, Crist, Liz, Andrea, Bonnie, Marie, Michelle, Emily, Fiona (and others) who provi ded invaluable and unconditional support and encouragement throughout this process. These loved ones saw me through all the ups and (sometimes quite devastati ng) downs of the dissertation process. Like the results of my dissertation demonstrated, it is no t the stress associated with a situation, but the

PAGE 6

6 positive support received from loved ones that influences ones outcomea fact which I personally became keenly aware of, and a ppreciative of, during this process. Lastly, I would like to extend my most sincer e and heartfelt apprec iation to the women who participated in this resear ch. Without these women, this di ssertation would not have been possible.

PAGE 7

7 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES .........................................................................................................................10LIST OF FIGURES .......................................................................................................................13ABSTRACT ...................................................................................................................... .............14 CHAP TER 1 INTRODUCTION .................................................................................................................. 16Epidemiology of Endometrial Cancer .................................................................................... 16Pathophysiology of Endometrial Cancer ................................................................................ 16Treatment of Endometrial Cancer .......................................................................................... 17Psychosocial Predictors of Surgical Outcome ........................................................................ 18Post-surgical Pain ............................................................................................................18Post-Surgical Length of Hospitalization .........................................................................20Post-Surgical Complications ........................................................................................... 21Wound Healing Complications ....................................................................................... 22Proposed Mechanisms of the Relationship between Peri-Operative Psychosocial Functioning and Surgical Outcome ..................................................................................... 23Behavioral Mechanisms .................................................................................................. 24Immune Mechanisms .......................................................................................................24Surgical Immune Response a nd Risk for Complications ....................................................... 25Modulators of the Surgical Immune Response ....................................................................... 26Psychosocial Stress and Cortisol ..................................................................................... 27Psychosocial Stress, Cortisol Dysr egulation, and Surgical Outcome ............................. 28Psychosocial Predictors of Peri-Operativ e Neuroendocrine Functioning and Surgical Outcome ....................................................................................................................... .......29Psychosocial Stress ..........................................................................................................29Interpersonal Coping .......................................................................................................29Purpose of Study .....................................................................................................................30Specific Aims ..........................................................................................................................312 METHODS ....................................................................................................................... ......35Design ........................................................................................................................ .............35Participants .................................................................................................................. ...........35Procedures .................................................................................................................... ...........36Psychosocial Assessment ....................................................................................................... .37Demographics .................................................................................................................. 38Medical comorbidity .......................................................................................................38Pre-surgical life stress ...................................................................................................... 38

PAGE 8

8 Emotional support from primary support person ............................................................ 39Pre-surgical perceived stress ........................................................................................... 39Total emotional support from all sources ........................................................................ 40Physiologic Assessment ........................................................................................................ ..40Salivary Cortisol Slope Calculation ........................................................................................ 41Surgical Outcome Assessment ............................................................................................... 43Severity of post-surgical complications .......................................................................... 43Length of post-surgic al hospitalization ........................................................................... 44Systemic immunosuppression ......................................................................................... 44Time to ambulation .......................................................................................................... 44Biobehavioral variables ...................................................................................................44Post-surgical pain ratings ................................................................................................ 45Statistical Procedures ..............................................................................................................45Analyses of Specific Aims ......................................................................................................45Power and Sample Size Considerations ................................................................................. 473 RESULTS ....................................................................................................................... ........57Preliminary Analyses .......................................................................................................... ....57Biological Variables ........................................................................................................ 57Outcome Variables ..........................................................................................................57Control Variables ....................................................................................................................57Relations Among Length of Hospitalization and Control Variables ............................... 58Relations Among Post-surgical Complic ations and Control Variables .......................... 58Relations Among Post-Surgical Ambulation and Control Variables .............................. 59Relations Among WBC Count and Control Variables ....................................................59Relations Among Stress and Control Variables .............................................................. 59Relations Among Emotional Support and Control Variables ......................................... 59Relations Among Diurnal Cortisol Slope and Control Variables ...................................59Descriptive Results ........................................................................................................... ......59Relations Among Stress, Cortis ol, and Surgical Outcome ..................................................... 61Length of Hospitalization ................................................................................................61Post-Surgical Complication Severity ..............................................................................62Post-Surgical Ambulation ............................................................................................... 62WBC Count .....................................................................................................................63Relations Among Emotional Support, Cortisol, and Surgical Outcome ................................ 63Length of Hospitalization ................................................................................................63Post-Surgical Complication Severity ..............................................................................63Post-Surgical Ambulation ............................................................................................... 64WBC Count .....................................................................................................................64Exploratory Analyses .......................................................................................................... ....66Rationale for Exploratory Analyses ................................................................................66Preliminary Analyses ....................................................................................................... 68Length of Hospitalization, Stress, and Cortisol ............................................................... 69Post-surgical Complication Seve rity, Stress, and Cortisol .............................................. 70Time to Ambulation, Stress, and Cortisol ....................................................................... 71WBC Count, Stress, and Cortisol .................................................................................... 72

PAGE 9

9 Length of Hospitalization, Emoti onal Support, and Cortisol ..........................................72Post-surgical Complication Severity, Emotional Support, and Cortisol ......................... 73Time to Ambulation, Emotional Support, and Cortisol .................................................. 74WBC Count, Emotional Support, and Cortisol ............................................................... 74Post-Surgical Pain ...................................................................................................................75Stress, Emotional Support, Cortis ol, and Post-surgical Pain .................................................. 76Moderation Analyses ........................................................................................................... ...774 DISCUSSION .................................................................................................................... ...120Discussion of Results ......................................................................................................... ...120Psychosocial Stress and Surgical Recovery .................................................................. 120Psychosocial Stress and Cortisol ................................................................................... 122Interpersonal Coping and Surgical Recovery ................................................................124Psychoneuroimmunologic Predictors of Post-Surgical Pain .........................................143Implications of Findings ...................................................................................................... .144Study Limitations ............................................................................................................. .....146Conclusions ...........................................................................................................................151LIST OF REFERENCES .............................................................................................................153BIOGRAPHICAL SKETCH .......................................................................................................163

PAGE 10

10 LIST OF TABLES Table page 2-1 Demographic characteristics of participants ......................................................................50 2-2 Surgical recovery effect sizes ............................................................................................ 51 3-1 Correlations between a priori biobehavioral control variables (colum ns) and main predictors and criteria of interest (rows) ............................................................................ 78 3-2 Relationship between intraoperative com plications (colum ns) and criteria of interest (rows) .................................................................................................................................79 3-3 Relationship between use of corticostero id m edication (columns) and criteria of interest (rows) ............................................................................................................... .....80 3-4 Relationship between cortisol collection (colum ns) and outcome variables (rows) .......... 81 3-5 Regression analysis examining length of hospitalization (in log days): Effect of im pact of negative events and diurnal cortisol slope ......................................................... 82 3-6 Regression analysis examining post-surgic al complication severity (in square-root transform ed sum of severity ratings): Effect of impact of negativ e events and diurnal cortisol slope ......................................................................................................................83 3-7 Regression analysis examining post-surgical am bulation (in days): Effect of impact of negative events and di urnal cortisol slope .....................................................................84 3-8 Regression analysis examining post-surgical W BC count: Effect of impact of negative events and diurnal cortisol slope .........................................................................85 3-9 Regression analysis examining length of hospitalization (in log days): Effect of em otional support (primary support pers on) and diurnal cortisol slope ............................ 86 3-10 Regression analysis examining post-surgic al complication severity (in square-root transform ed sum of severity ratings): Eff ect of emotional support (primary support person) and diurnal cortisol slope ...................................................................................... 87 3-11 Regression analysis examining post-surg ical am bulation (in days): Effect of emotional support (primary support pers on) and diurnal cortisol slope ............................ 88 3-12 Regression analysis examining post-su rgical W BC count: Effect of emotional support (primary support person) and diurnal cortisol slope ............................................. 89 3-13 Logistic regression analysis examini ng post-surgical W BC count (normal versus leukocytosis): Effect of emotional suppor t (primary support person) and diurnal cortisol slope ......................................................................................................................90

PAGE 11

11 3-14 Regression analysis examining post-surg ical neutrophil percentage: Effect of em otional support (primary support Pers on) and diurnal cortisol slope ............................ 91 3-15 Regression analysis examining post-surg ical lym phocyte percentage: Effect of emotional support (primary support pers on) and diurnal cortisol slope ............................ 92 3-16 Regression analysis examining post-surg ical W BC count in first four days postsurgery: Effect of emotional support (primary support person) and diurnal cortisol slope ......................................................................................................................... ..........93 3-17 Correlations between a priori biobeha vioral control variables (colum ns) and exploratory predictors (rows) ............................................................................................. 94 3-18 Correlations among alternative measures of stress, em otional support, cortisol and criteria ...................................................................................................................... ..........95 3-19 Regression analysis examining post-surg ical W BC count: Effect of impact of negative events and AUC-G ..............................................................................................96 3-20 Regression analysis examining post-surg ical W BC count : Effect of impact of negative events and mean daily cortisol ............................................................................97 3-21 Regression analysis examining post-surg ical W BC count: Effect of impact of negative events and mean morning cortisol ....................................................................... 98 3-22 Regression analysis examining post-surgic al W BC count: Effect of perceived stress and AUC-G ........................................................................................................................99 3-23 Regression analysis examining post-surgic al W BC count: Effect of perceived stress and mean daily cortisol ....................................................................................................100 3-24 Regression analysis examining post-surgic al W BC count: Effect of perceived stress and mean morning cortisol ............................................................................................... 101 3-25 Regression analysis examining post-su rgical W BC count: Effect of emotional support (primary support person) and AUC-G ................................................................102 3-26 Regression analysis examining post-su rgical W BC count: Effect of emotional support (primary support person) and mean daily cortisol .............................................. 103 3-27 Regression analysis examining post-su rgical W BC count: Effect of emotional support (primary support person) and mean morning cortisol ......................................... 104 3-28 Regression analysis examining post-surgic a l WBC count: Effect of total emotional support and AUC-G ......................................................................................................... 105 3-29 Regression analysis examining post-surgic a l WBC count: Effect of total emotional support and AUC-I ........................................................................................................... 106

PAGE 12

12 3-30 Regression analysis examining post-surgic a l WBC count: Effect of total emotional support and mean daily cortisol .......................................................................................107 3-31 Regression analysis examining post-surgic a l WBC count: Effect of total emotional support and mean morning cortisol.................................................................................. 108 3-32 Regression analysis examining effect of em otional support (primary support person) and diurnal cortisol slope on post-surgical pain ratings ................................................... 109 3-33 Regression analysis examining effect of im pact of negative events and diurnal cortisol slope on post-su rgical pain ratings ...................................................................... 110 3-34 Regression analysis examining effect of im pact of negative events and AUC-I on post-surgical pain ratings ................................................................................................. 111 3-35 Regression analysis examining effect of em otional support (primary support person) and AUC-I on post-surgical pain ratings .........................................................................112 3-36 Regression analysis examining effect of perceived stress and AUC-I on post-surgical pain ratings .......................................................................................................................113 3-37 Regression analysis examining effect of perceived stress and AUC-G on postsurgical pain ratings .........................................................................................................114 3-38 Regression analysis examining the rela tionship between postsurgical length of hospitalization (in log days) and im pact of negative events: Effect of emotional support (primary support person) as a moderator ............................................................115 3-39 Regression analysis examining the relati onship between severity of post-surgical com plications (in square-root transformed sum of severity ratings) and impact of negative events: Effect of emotional support (primary support person) as a moderator 116 3-40 Regression analysis examining the rela tionship between time to post-surgical am bulation (in days) and impact of nega tive events: Effect of emotional support (primary support person) as a moderator ......................................................................... 117 3-41 Regression analysis examining the rela tionship between post-surgical WBC count and im pact of negative events: Effect of emotional support (primary support person) as a moderator ................................................................................................................ ..118

PAGE 13

13 LIST OF FIGURES Figure page 1-1 Wound healing cascade (Adapted from Kiecolt-Glaser et al., 1998). ............................... 33 1-2 Theoretical model. ........................................................................................................ .....34 2-1 Design timeline .................................................................................................................53 2-2 Categories of post-surgical complications. ........................................................................54 2-3 Power Versus Sample Size With Alpha = .05 and Slope = .32 ......................................... 55 2-4 Power Versus Sample Size With Alpha = .05 and Slope = .35 ......................................... 56 3-1 Examples of graphical representations of diurnal cortisol slopes from study participants .................................................................................................................. .....119

PAGE 14

14 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PSYCHONEUROIMMUNOLOGIC PREDICTORS OF POST-SURGICAL OUTCOME IN WOMEN WITH ENDOMETRIAL CANCER By Sally Elizabeth Jensen December 2008 Chair: Deidre B. Pereira Major: Psychology Endometrial cancer is the most common gynecol ogic cancer in the United States and is often co-morbid with conditions that confer high risk for pos t-surgical complications. Among surgical populations, psychosocial distress and maladaptive coping are associated with poorer surgical recovery. Similarly, cort isol, a stress hormone, may serve as an important mediator of the relationship between psychosoc ial factors and impaired immun ity during the peri-operative period. Few studies have examined psychoneuro immunologic predictors of oncologic surgical. The present study explored the re lations among pre-surgical psyc hosocial stress, pre-surgical emotional support, pre-surgical diurnal cortisol slope, and indices of surgical outcome (e.g., length of hospitalization, seve rity of post-surgical complications, time to post-surgical ambulation, and post-surgical systemic imm une response), among 75 women undergoing total abdominal hysterectomy with bilateral sa lpingo oophorectomy (TAH-BSO) for suspected endometrial cancer. It was hypothesized that a le ss complicated surgical recovery would be predicted by less pre-surgical psychosocial stress, more normal diurnal cortisol production, and greater emotional support. Participants completed a pre-surgical psyc hosocial interview and collected pre-surgical saliva samples for analysis of diurnal cortisol output. Surgical outcome data was abstracted from medical records.

PAGE 15

15 Greater perceived emotional support from primary support person was significantly associated with less elevated post-surgical WBC count and was marginally associated with lower post-surgical pain ratings. More elevated pre-surgical indices of cortisol were significantly associated with more elevated post-surgical WBC count and lower postsurgical pain ratings. Contrary to hypothesis, psychosocia l stress was not significantly re lated to indices of surgical recovery or indices of cortisol Length of hospitalization, severity of post-surgical complications, and time to post-surgical ambulation were not significantly associated with psychoneuroimmunologic predictors The findings of the presen t study extend the literature examining psychoneuroimmunologic predictors of surgical outcome by identifying relations among pre-surgical emotional support, pre-surgical indices of cortisol, post-surgical WBC count, and post-surgical pain ratings. They provide preliminary data fo r future research to examine psychoneuroimmunologic relations during the peri-operative period among women with endometrial cancer.

PAGE 16

16 CHAPTER 1 INTRODUCTION Epidemiology of Endometrial Cancer Endom etrial cancer is the most common gynecol ogic cancer in the United States, with approximately 40,880 new cases estimated in 2005 (A merican Cancer Society [ACS], 2005). It is also the second-most-deadly gynecologic cancer in the United States, with approximately 7,310 deaths expected from endometrial cancer in 2005 (ACS, 2005). The five-year survival rate for patients with endometrial cance r is 84% (ACS, 2003). Endometri al cancer primarily affects postmenopausal women, with the peak incidence of onset during the sixth and seventh decades (Dorigo & Goodman, 2003). Pathophysiology of Endometrial Cancer Endom etrial cancer commonly presents as an ov ergrowth of the cells in the endometrium, which is the innermost layer of the body of the uterus. Approximately 90% of endometrial cancers are pathologically classified as endometrial ade nocarcinomas (ACS, 2003). Endometrial cancer is thought to result from a progression fr om endometrial hyperplasi a (pre-malignant cells) to atypical hyperplasia to ma lignant cells (Dorigo & Goodma n, 2003). Endometrial cancer, like breast and ovarian cancer, is an endocrine-medi ated disease. During th e follicular phase of menstruation, estrogen unopposed by progesterone stim ulates the prolifera tion of endometrial cells (Cyr & Skelton, 2003). The luteal phase of menstruation is characterized by the production of progesterone, resulting in the cessation of endometrial proliferation and the eventual sloughing of endometrial cells (Cyr & Skelton, 2003). A shift in the balance between the amount of estrogen and progesterone produc ed by the ovaries, resulting in elevated levels of unopposed estrogen and unchecked proliferatio n of endometrial cells, has been implicated as a risk factor for endometrial cancer (ACS, 2003). Additional known risk fact ors for endometrial cancer

PAGE 17

17 include exogenous use of estrogen (unopposed by pr ogesterone), use of Tamoxifen (a selective estrogen receptor modulator treatment for br east cancer), early mena rche, late menopause, infertility, obesity, and diabetes (ACS, 2003). R ecent evidence suggests that genetics may also play a role in the development of endometrial cancer, given the higher risk for developing endometrial cancer among women with a history of colon, ova rian, or breast cancer (Dorigo & Goodman, 2003). Treatment of Endometrial Cancer The surgical staging of endom etrial cancer is determined by cytology and pathology results from pelvic washings, total abdominal hyste rectomy with bilatera l salpingo oophorectomy (TAH-BSO), and pelvic lymph node biopsies. Se venty-five percent of endometrial cancer diagnoses occur at Stage I, which is defined by the restriction of the original tumor to the endometrium (Dorigo & Goodman, 2003). An addi tional 11% percent of endometrial cancer diagnoses occur at Stage II a nd 11% occur at Stage III (Dor igo & Goodman, 2003). Standard treatment for early-stage endom etrial cancer involves total abdominal hysterectomy with bilateral salpingo oophorectomy (TAH-BSO). Pelv ic irradiation or chemotherapy may be necessary if the pelvic lymph nodes are positive for cancer. Advanced stages of endometrial cancer involve the metastatic spr ead of the tumor to the cervi x, ovaries, abdominal organs, or other sites (ACS, 2003). Despite the favorable survival rate, high rate s of comorbid conditions such as obesity, hypertension, and diabetes, as well as advanced age, confer an increased risk for acute postsurgical complications such as thromboembolism, infection, and wound healing complications among women with endometrial can cer (Tozzi, Malur, Koehler, & Schneider, 2005). These postsurgical complications pose pot ential threats to both physical and psychosocial well-being. Although great advances in medical practice and technology have significantly improved

PAGE 18

18 patients surgical outcomes and recovery, post-surgical complicati ons have not been eliminated entirely, thus suggesting that addi tional risk factors may contribu te to poorer surgical outcome (Kopp et al., 2003). In the search to identify such risk factors, impaired psychosocial functioning has emerged as a potentially important ri sk factor for poorer surgical outcome. Psychosocial Predictors of Surgical Outcome Research ex amining the relationship betw een psychosocial functioning and surgical outcome has been generally heterogeneous in nature. Studies include a va riety of pre-surgical psychosocial predictors, t ypes of surgery, types of anesthesia (general versus local), and followup period (acute versus long-term follow-up). A dditionally, researchers have examined various distinct indices of surg ical outcome. This diversity of outcome variables may reflect the different aspects of surgical recovery, such as restoration of function, quality of life, cost effectiveness, and medical outcome variables. Thus, in order to obtain a comprehensiv e picture of surgical recovery, it may be important to ev aluate a variety of outcomes. Post-surgical Pain Post-surgical pain is an important index of post-surgical outcom e, given its relationship with both psychosocial and physic al well-being. Uncontrolled main is a major predictor of psychosocial distress, including anxiety (Noyes, Holt, & Massie 1998) and depression (Massie & Popkin, 1998). In addition to its potential deleterious effects on quality of life, post-surgical pain may also compromise the immune response to surgery and place individuals at increased risk for post-surgical complications. Moreover, there is evidence that the immunosuppressive effects of pain may be independent of the immu nosuppression associated with surgical tissue trauma. For example, Liebeskind (1991) found that pain without tissue damage was associated with decrements in natural killer (NK) cel l activity, as well as decreased lymphocyte proliferation. Additionally, both acu te and persistent post-surgical pain have been associated

PAGE 19

19 with impaired wound healing (M cGuire et al., 2006). Given th e overlapping relationships between psychosocial distress, pain and immune functioning, distressed i ndividuals may be at risk for not only increased postsurgical pain, but also for im paired post-surgical immune functioning (Kiecolt-Glaser, Page, Marucha, MacCallum, Glaser, 1998). Several p redictors of greater post-surgical pain have been identified in the general surgery literature. Walch et al. (2005) found that individuals exposed to greater sunlight during hospitalization following spinal surgery experienced margina lly less pain, and consumed significantly less analgesic medication per hour th an individuals exposed to lower levels of sunlight. Greater exposure to sunlight during hosp italization was also asso ciated with decreased perceived stress (Walch et al., 2005). Katz et al. (2005) examined predictors of acute postsurgical pain among women undergoing breast cancer surgery and f ound that greater pre-surgical anxiety predicted clinically m eaningful acute pain in the two days following surgery. Several investigations of psychosocial predic tors of gynecologic su rgical outcome have focused on post-operative pain and analgesic use as measures of surgical outcome. Preliminary support for the influence of pre-operative psychosocial functioning on gynecologic surgical pain was demonstrated by Kain, Sevarino, Alexander, Pincus, and Mayes (2000), who found that preoperative anxiety predicted post-operativ e pain among women undergoing abdominal hysterectomy. In an extension of Kain et al.s (2000) findings, Cohen, Fouladi, and Katz (2005) examined pre-operative distress and coping as pred ictors of post-operative pain and morphine consumption. Although pre-operative distress predicted greater post-operativ e pain and morphine consumption, this relationship was no longer significant when preoperative coping was controlled for (Cohen et al., 2005). Moreover, the use of more passive coping strategies was associated with elevated pain and morphine consumption post-operative ly (Cohen et al., 2005).

PAGE 20

20 Post-Surgical Length of Hospitalization Post-surgical length of hospitalization provides an estim ate of speed of recovery following surgery and thus constitutes another important index of surgical outcome. Post-surgical length of hospitalization dependent upon factors related to insurance, institutio nal policies, the availability of home caregiving, and medical status. Despite its multifactorial nature, post-surgical length of hospitalization remains a c linically important measure of physical recovery given its reflection of the physical impact of surgery, physical aspects of recovery, behavioral aspects of recovery, and healthcare costs (Contrada et al., 2003). Longer length of post-s urgical hospitalization may be associated with a number of negative health ou tcomes including post-surg ical complications and infection (Bratzler & Hunt, 2006). McGuire et al. (2006) found that shorter length of hospitalization following gastric bypass surgery wa s marginally related to faster wound healing and that acute post-surgical pain was associat ed with a longer post-su rgical hospitalization. Moreover, recent research suggests that behavioral and psychosocial factors may play a role in post-surgical length of hospitalization. Specifi cally, Oscarsson, Poromaa, Nussler, and Lofgren (2006) compared length of hospitalization for women undergoing minimally invasive surgery (laparascopic supravaginal hysterectomy) with women undergoing standard surgery (abdominal supravaginal hysterectomy) and found no differe nces between the two groups. Thus, the authors concluded that post-surgical length of hospitalization may depend more upon patient expectations and factors during the acute post-surgical period than on the surgical procedure itself (Oscarsson et al., 2006). Recent research has begun to elucidate some of the specific behavioral and psychosocial factors that may impact length of post-surgical hospitalization. Krohne and Slangen (2005) investigated the relations hip between emotional support and duration of post-surgical hospitalization among individu als undergoing maxillofacial surgery and found that patients receiving more emotional support were hospitalized approximately 1.5 fewer days

PAGE 21

21 than individuals receiving less emotional support. Contrada et al. (2004) found additional support for the relationship between interpersonal coping and post-surgical duration of hospitalization in a study that examined religiousness among indivi duals undergoing coronary artery bypass graft (CABG). In addition to its use as a coping st rategy, religiousness has been posited to be associated with social support networks. Contra da et al. (2004) found that individuals with stronger religious beliefs had significantly shorter hospitalization periods following CABG. Post-Surgical Complications The incidence of post-surgical com plications not only contributes to post-surgical length of hospitalization but also serve as an important indicator of surgic al outcome independent of their relationship to duration of hosp italization. Common post-surgical complications include surgical site infections, sepsis, cardiovascular co mplications, respiratory complications, and thromboembolic complications (Bratzler & Hunt, 2006). Post-operative complications are associated with longer post-surgical hospitaliza tions, as well as increased cost. Dimick et al. (2004) examined hospital costs associated with su rgical complications. Increased cost per patient included $1398 for infectious complications $7789 for cardiovascular complications, $52,466 for respiratory complications, and $1810 for th romboembolic complications (Dimick et al., 2004). Although post-surgical mortality has decr eased substantially, th e development of postoperative complications remains a significant index of post-surgical outcome. Among women undergoing hysterectomy, both insurance status, as well as race/ethnicity have been associated with the incidence of post-surgical complications Women on Medicaid were more likely to have a post-surgical complication than privately insu red women and Black women were more likely to have a post-surgical complication than Wh ite women, despite equivalent quality of care (Hakim, Benedict, & Merrick, 2004). There is also emerging evidence that behavioral and psychosocial factors may also be associated with greater risk for post-surgical complications. For

PAGE 22

22 example, Contrada et al (2004) found individuals with strong er religious beliefs had fewer postsurgical complications. Moreover, the relations hip between religious be liefs and length of hospitalization was mediated by th e number of post-operative co mplications (Contrada et al., 2004). Kopp et al. (2003) provided further support for the relati onship between interpersonal coping and surgical outcome, findi ng that life satisfaction and soci al support predicted surgical recovery without complications am ong a general surgery population. Wound Healing Complications Wound healing com plications constitute an important category of post-surgical complications and an important index of r ecovery among individuals undergoing surgery. The wound healing process consists of a series of carefully orches trated phases (See Figure 1-1). Comorbid conditions such as obesity, malnut rition, renal dysfunction, alcoholism, anemia, diabetes, tobacco use, immunosuppression, peripher al vascular disease, cancer, and advanced age may confer risk for wound healing compli cations (Schimp et al., 2004). Surgical wound healing complications may be described by th e following categories: incisional separation, incomplete dehiscence/superficia l separation, complete dehiscence, and evisceration (Schimp et al., 2004). Treatment for wound complications is both time-consuming, expensive, and may require prolonged hospitalizat ion or specialized wound car e (Schimp et al., 2004). Among gynecologic oncology patients, the incidence of incomplete and complete dehiscence ranges from two to five percent (Morrow, 1996). Kiecolt-Glaser and colleagues ha ve accumulated strong eviden ce that stress impairs wound healing, particularly during th e early wound healing cascade (see Kiecolt-Glaser et al., 1998 for review). For example, delayed wound healing has been demonstrated among Alzheimers caregivers (Kiecolt-Glaser, Ma rucha, Malarkey, Mercado, & Gl aser, 1995), among mice subject to restraint stress (Padgett, Marucha, & Sheridan, 1998), am ong dental students during

PAGE 23

23 examination time (Marucha, Kiecolt-Glaser, & Favagehi, 1998), and among couples during a conflict discussion (Kiecolt-Glas er et al., 2005). The relationshi p between psychosocial distress and impaired wound healing is proposed to be me diated by the effects of psychosocial distress on immune functioning. Indeed, Gl aser et al. (1999) found that individuals who had a lower number of cytokines in the wound environment were also likely to have elevated stress levels, negative affect, and number of negative life events Although the majority of wound healing research has been carried out under controlled conditions and involves standardized laboratoryinduced wounds, several studies have examined psychosocial predictors of wound healing and relevant immune variables in the surgical context. Broadbent Petrie, Alley, and Booth (2003) examined psychosocial predictors of wound healing and wound-relevant immune variables among individuals undergoing inguinal hernia surger y and found that greater pre-surgical worry about the surgery was associated with decrease d matrix metalloproteinases 9 (MMP9) at the wound site. Moreover, greater presurgical perceived stress was a ssociated with decreased IL-1 at the wound site (Broadbent et al., 2003). In a sample of in dividuals undergoing gastric bypass surgery, McGuire et al. (2006) f ound that both acute and persiste nt post-surgical pain were associated with slower wound healing of a standardized punch biopsy wound placed at the same time as surgery. Proposed Mechanisms of the Relationship b etween Peri-Operative Psychosocial Functioning and Surgical Outcome The findings mentioned above suggest that psychosocial factors may play an important role in various indices of surgical outcome across diverse surgical populations. Although few studies have specifically examined the potential mechanisms of the relationships between psychosocial functioning and indices of surg ical outcome, both behavioral and immune mechanisms have been proposed to account for these relationships.

PAGE 24

24 Behavioral Mechanisms Engagem ent in health promoting behaviors ma y serve as one modulator of the relationship between psychosocial functioning and surgical outcome. For example, health behaviors may impact surgical outcome via their effect on the type and amount of anesth etic agent used, their effect on the extent of surgery, and via direct effects on immune functioning (Kiecolt-Glaser et al., 1998). Greater psychosocial distress is associated with poorer health behaviors (KiecoltGlaser et al., 1998). Furthermore, adaptive inte rpersonal coping may confer benefits to preoperative health, which may in turn serve as a protective factor during th e peri-operative period (Contrada et al. 2004). Indeed, a nu mber of pre-operative health f actors have been identified as risk factors for impaired post-operative recovery, such as smoking status, hypertension, diabetes, and hypercholesterolemia (Peterson et al., 2002). However, it is unknown whether these indices of pre-operative health mediat e the relationship between ps ychosocial functioning and postsurgical outcome. Additionally, health promoting behaviors are also important in the acute postsurgical period. For example, psychosocial distre ss may be associated with poor compliance with post-surgical recommendations for behaviors relevant to recovery such as proper breathing and ambulation (Kiecolt-Glaser et al., 1998). Immune Mechanisms In addition to behavioral m odulators of the relationship between psychosocial functioning and post-surgical outcome, psychoneuroimmunol ogic (PNI) relations may also modulate this relationship. For example, social support is associated with impr oved immune functioning (Uchino, 2006) as well as improved wound heali ng capabilities (Kiecolt -Glaser et al., 1998). Several studies have also examined PNI rela tions during the peri-o perative period. Greater depression is associated with decreased postsurgical lymphocytes a nd NK cells (Tjemsland, Soreide, Matre, & Malt, 1997). Increased presurgical state anxiet y predicted a blunted

PAGE 25

25 neuroendocrine response to surgery (Pearson, Maddern, & Fitridge, 2005). Thus, psychosocial functioning may influence immune functioning during the peri-ope rative period, which may in turn influence surgical recovery. Surgical Immune Response and Risk for Complications In addition to the psychosocial stress associat ed w ith surgery, the surgical experience is also associated with substantia l physical stress and trauma whic h affect immune functioning. The post-surgical immune response is characte rized by hyperinflammati on and immunosuppression, both of which may contribute to the developmen t of post-surgical complications (Menger & Vollmar, 2004). Typically, the local immune res ponse to surgery involves activation of the inflammatory cascade in order to prevent infect ion and to initiate tissu e rebuilding, whereas the systemic response involves leukocytosis, neutro philia, and lymphocytopenia (Salo, 1992). It should be emphasized that both the activated an d depressed components of immune response to surgery are adaptive physiologic responses that enhance post-surgical surv ival (Salo, 1992). However, the excessive inflammation and immunos uppresson associated with surgical recovery also put individuals at risk for the development of compli cations and wound healing impairments (Salo, 1992). Several aspects of surgery contribute to al tered immune functioni ng during recovery. For example, the role of general anesthesia duri ng surgery is to mainta in homeostasis, thereby decreasing the stress response to su rgery. As a result, general anes thesia is associated with a suppression of cell-mediated immunity (S alo, 1992; Ben-Eliya hu, 2003). Moreover, the suppression of cell-mediated immunity observed with general anesthesia is attenuated in regional anesthesia in which the nociceptive impulses are bl ocked regionally or at th e level of the spinal cord (Ben-Eliyahu, 2003). Blood tr ansfusions also impact peri -operative immune functioning by decreasing cell-mediated immunity (Ben-Eliyah u, 2003; Salo, 1992). Finally, the tissue trauma

PAGE 26

26 associated with the surgical procedure itself has been linked to altered immune functioning during the peri-operative period. Surg ical trauma is related to hypope rfusion of tissue resulting in tissue ischemia, which subsequently alters ne uroendocrine and immune functioning (Salo, 1992). The relationship between surgical tissue trauma and altered immunity is further strengthened by findings that minimally invasive surgical procedures are associated with attenuated cell-mediated immunity suppression and cytokine misbalance (Ben-Eliyahu, 2003; Menger & Vollmar, 2004). Modulators of the Surgical Immune Response Neuroendocrine functioning has been im pli cated in the modulation of the immune response to surgery. One pathway by which neur oendocrine factors appear to modulate immune response to surgery is through sympathetic nerv ous system (SNS) activation and release of catecholamines. Catecholamines, such as epin ephrine and norepinephrine, suppress cellmediated immunity (Ben-Eliyahu, 2003). The imm unosuppressive effects of catecholamines are further supported by evidence that the suppression of cell-mediated immunity is attenuated by administration of beta adrenergic antagonists (Nelson & Lysle, 1998). Activation of the hypothalamic pituitary adrenal (HPA) axis and elevated levels of its endproduct, glucocorticoids, have also demonstr ated strong immunosuppressi ve effects during the peri-operative period. Glucocorticoids, such as cortisol, increase post-surgically (Ben-Eliyahu, 2003; Pearson et al., 2005). Glucocorticoids also act synergistically with catecholamines by increasing beta adrenergic receptor density (Antoni et al., 2006). Glucocorticoids exert immunosuppressive effects by de creasing circulating lymphoc ytes, decrease monocytes functioning, decrease NK cell activity, and decr ease cytokine production (DeKeyser, 2003). Increases in glucocorticoids are also associated with elevated levels of IL-6, a cytokine that functions to downregulate pro-inflammatory immune responses and to enhance antiinflammatory immune responses (Menger & Vollmar, 2004). Specifically, glucocorticoid

PAGE 27

27 dysregulation also negatively impacts levels of IL-1 and TNF, two pro-inflammatory cytokines that are integral in the early wound healing response (Kiecolt-Glaser et al., 1998). Glaser et al. (1999) found elevat ed levels of salivary cor tisol, a glucocorticoid, among individuals who produced low numbers of cytokines in the fluid of standardized blister wounds. Psychosocial Stress and Cortisol Cortisol has increasingly b ecom e a variable of interest in psychoneuroimmunologic (PNI) research, given its relationship with both psychosocial stress and immune functioning. The relationship between psychosocial stress and cort isol results from HPA axis activation. When a situation is perceived as st ressful, the HPA axis becomes activated, causing a cascade of hormones to be produced that may, in turn, negati vely impact immunity and health outcome. The paraventricular nucleus of the hypothalamus releases corticotropin-releasing factor (CRF), which stimulates the pituitary gland. In response to this stimulati on, the pituitary gland releases adrenocorticotropic hormone (ACTH) Finally, the release of ACTH results in the secretion of glucocorticoids from the adrenal cortex. One pa rticular glucocorticoi d, cortisol, follows a circadian rhythm regulated by th e suprachiasmatic nuclei (SCN) in the hypothalamus (Stone et al., 2001). The SCN causes cortisol to peak in concentration in the blood around waking and decline steadily throughout th e day (Stone et al., 2001). Both physical and psychosocial factors can be responsible for the circadian dysregulation of cortisol (Mormont & Levi, 1997). Moreover, both acute and chronic stressors have been linked to dysregulated cortisol Acute stressors such as boot camp (Hellhammer, Buchtal, Gutberlet, & Kirschbaum, 1997), challenging me ntal tasks (Kirschbaum Pruessner, Stone, & Federenko,1995), and lumbar punctu re (Bohnen, Terwel, Twijnstra, & Markerink, 1992) have been associated with higher levels of cortisol. However, altered cortisol production has also been

PAGE 28

28 found among individuals under chroni c life stress, such as indivi duals experiencing interpersonal violence (Seedat, Stein, Kennedy, & Hauger, 2003) and unemployment (Ocken fels, et al., 1995). Under normal conditions, increases in circul ating glucocorticoids act on the hypothalamus to inhibit further secre tion of CRF and, subsequently, the secr etion of ACTH from the pituitary gland. This negative feedback loop causes an in hibition of stress-induced HPA activation when a situation is no longer perceived as stressful. However, high, sustained levels of perceived stress override this negative feedback loop, resulting in dysregulation of circulating cortisol (McEwen, 1998). This sustained dysregulation of cortisol can have significan t implications for health and immunity. Psychosocial Stress, Cortisol Dy sregulation, and Surgical Outcome The surgical experience is characterize d by both psychosocial and physical stress. Consequently, stress m ay be associated with neuroendocrine dysregul ation and subsequent complications among individuals during the peri-operative peri od. Padgett et al. (1998) found that the delayed wound healing ob served among mice under restraint stress was associated with elevated cortisol and that this relationship was attenuated by the blocki ng of glucocorticoid receptors. Additionally, Glaser et al. (1999) found that greater stress was associated with both greater cortisol, as well as lower production of pro-inflammatory cytokines in wound fluid. Pearson et al. (2005) examined state anxiety during the acute pr e-operative period and found that it was associated with a blunted cortisol response during th e intra-operative period. Although neither anxiety nor intra-operativ e cortisol was associated with post-operative complications, these findings provide compelling evidence that PNI relations may be similar in studies of both controlled and standardi zed tissue damage, as well as in st udies of surgical tissue damage.

PAGE 29

29 Psychosocial Predictors of Peri-Operati ve Neuroendocrine Functioning and Sur gical Outcome Psychosocial Stress Given the well-establish ed li nk between psychosocial distress and impaired immunity, numerous studies have examined psychosocia l stress as a predictor of peri-operative neuroendocrine functioning and surgical outcome. One form of psychosocial distress, perceived stress, has been positively associated with cortisol dysregulation follo wing laboratory controlled tissue damage (Glaser et al., 1999). Additionally, perceived stress is associated with an impaired pro-inflammatory response at the wound site among individuals underg oing inguinal hernia surgery (Broadbent et al., 2004) Although cortisol was not exam ined as a mediator of the relationship between perceived st ress and impaired pro-infla mmatory response in this study, glucocorticoids have been iden tified as key modulators of the impaired pro-inflammatory response early in the wound healing cascade. Interpersonal Coping Findings from Cohen et al. (2005) suggest that, am ong wom en undergoing major gynecologic surgery, pre-operative coping strategies may be better predicto rs of post-operative pain outcomes than psychosocial distress. In addi tion, Manyande et al. (1995) report that the use of an active coping imagery inte rvention prior to surgery was a ssociated with lower cortisol, lower heart rate, and less pain during the perioperative period. Furthermore, the association between passive coping and poorer surgical outco me is consistent with previous research findings linking passive coping with poorer health behavior s (Singh et al., 1996), poorer immunity (Goodkin, Fuchs, Feaster, Leeka, & Rishel, 1992), and poorer health outcomes (Leserman et al., 2002) among individuals with HIV. Moreover, passive coping is often associated with maladaptive interpersonal c oping (Pereira et al., 2004). Interpersonal coping

PAGE 30

30 involves the way in which individu als interact with their social environment when faced with stressors. Interpersonal coping has been demonstrat ed to be an influentia l factor in individuals exchanges with the health care system and adherence (Pereira et al., 2004). Specifically, maladaptive interpersonal coping is hypothesized to negatively impact health behaviors and health outcomes via its association with behavi oral disengagement and avoidant coping (Cohen, Gottlieb, & Underwood, 2001). Add itionally, interpersonal coping may be of particular significance to women, given the relational cultural theorys assertion that women commonly identify themselves through their relation to ot hers (Jordan, Kaplan, Mi ller, Stiver, & Surrey 1991). Purpose of Study Although findings am ong general surgery populations suggest that psychosocial functioning is associated with objective indices of poorer post-su rgical outcome, such as longer post-surgical hospitalization and more post-surg ical complications, no published study to-date has examined these variables among wome n undergoing gynecologic oncology surgery. The present study will extend findings from the gene ral surgery population to examine relationships between psychosocial functioning an d indices of post-surgical outco me (post-surgical systemic immune response, post-surgical complications, tim e to ambulation, and post-surgical duration of hospitalization) among women undergoing TAH-BSO for suspect ed endometrial cancer, a population that is considered to be at an elevat ed risk for poor surgical outcome as a result of their high rates of comorbid medical conditions (See Figure 1-2). The findings of the present study are expected to enhance the ability to screen for patients w ho may be at elevated risk for poorer post-surgical outcome based on psychosocia l functioning. The findings are also expected to inform future research, including research examining the potentia l mechanisms of the relationship between pre-surgical stress, pre-surgical interper sonal coping, and post-surgical

PAGE 31

31 outcome, as well as the development of pre-surg ical psychosocial interv entions to improve postsurgical outcome. Specific Aims Aim 1: To exam ine the relations among pre-surgical psychosoc ial stress, pre-surgical diurnal cortisol slope, and indi ces of post-surgical recovery (systemic immunosuppression, postsurgical complications, time to ambulation, and length of post-surgica l hospitalization) among women undergoing total abdominal hysterectomy with bila teral salpingo oophorectomy (TAHBSO) for suspected endometrial cancer. Hypothesis 1a: Women who report greater pre-surg ical psychosocial stress will have a more complicated post-surgical recovery, incl uding greater systemic immunosuppression, greater incidence of post-surgical co mplications, longer time to ambu lation, and longer post-surgical hospitalization. Hypothesis 1b: The relationship between pre-surg ical psychosocial stress and a more complicated post-surgical recovery will be mediated by more abnormal diurnal cortisol slopes. Aim 2: To examine the relations among pre-su rgical emotional s upport, pre-surgical diurnal cortisol slope, and indi ces of post-surgical recovery (systemic immunosuppression, postsurgical complications, time to ambulation, and length of post-surgica l hospitalization) among women undergoing total abdominal hysterectomy with bila teral salpingo oophorectomy (TAHBSO) for suspected endometrial cancer. Hypothesis 2a: Women who report greater pre-surgical emotional support will have a less complicated post-surgical rec overy, including less systemic immunosuppression, lower incidence of post-surgical complications, shorter tim e to ambulation, and s horter post-surgical hospitalization.

PAGE 32

32 Hypothesis 2b: The relationship between pre-surg ical emotional support and a less complicated post-surgical recovery will be mediat ed by a less abnormal diurnal cortisol slope. Aim 3: To examine whether pre-surgical em otional support moderates the potential relations among pre-surgical ps ychosocial stress and indices of post-surgical recovery among women undergoing total abdominal hysterectomy with bila teral salpingo oophorectomy (TAHBSO) for suspected endometrial cancer. Hypothesis 3 : Among women with high pre-surgical psychosocial stress, those who report less emotional support will have a more complicat ed post-surgical recovery, including greater systemic immunosuppression, greater incidence of post-surgical complications, longer time to ambulation, and longer post-surgica l hospitalization, than those who report greater use of perioperative adaptive in terpersonal coping.

PAGE 33

33 Figure 1-1. Wound healing cascade (Adapt ed from Kiecolt-Glaser et al., 1998).

PAGE 34

34 Figure 1-2. Theoretical model.

PAGE 35

35 CHAPTER 2 METHODS Design The design of the present study is prospectiv e. The sam ple consists of 75 women with suspected endometrial cancer sc heduled to undergo TAH-BSO. Participants were recruited from the UF&Shands Gynecologic Onco logy Clinic in Gainesville, FL. Participants completed a psychosocial interview approximately one day pr ior to TAH-BSO surgery and collected saliva samples for the quantitation of presurgical salivary cortisol levels Surgical outcome data (from the time of surgery until discha rge) were abstracted from me dical records. This study was conducted in accordance with the rules and regula tions of the Institutional Review Board (IRB) of the University of Florida and is IRB appr oved (protocol no. 69-2004). Figure 2-1 displays the study design. Participants Inclusion criteria for partic ipants in the present study in cluded: (a) wom en undergoing TAH-BSO with or without pelvic lymph node dissection for either (i) an abnormal endometrial biopsy concerning for endometrial cancer or (i i) a complex adnexal mass without ascites or omental caking concerning for Stage I gynecolo gic malignancy, and (b) fluency in spoken English. Exclusion criteria for participants in the present study in cluded: (a) recurrent endometrial carcinoma, (b) metastasis to the uterine corpus from another site, (c) pre-surgical chemotherapy or radiotherapy, (d) current psychotic disorder, and (e) current suicidal intent/plan. Because the diagnosis of endometrial cancer is based on post-surgical pathology, it was possible that women included in the propos ed study were diagnosed post-sur gically with pre-cancerous or benign disease. Based on preliminary data, it was expected that less than 10% of participants

PAGE 36

36 would eventually be diagnosed with benign di sease. Table 2-1 summarizes the demographic characteristics of participants. Procedures Participants were recruited from the Gyneco logic Oncology Clinic at UF&Shands Medical Plaza. Potentially eligible participants were iden tified at their treatmen t consultation visit through consultation with attending physicians, residents, medical students, and nurse practitioners. Potentially eligible part icipants were notified of the opportuni ty to participate in a research project by one of the previously listed health care providers. If a patient expressed interest in participating, she met with a trained research er who provided an overview of the study and answered any questions the patient had. If a patient stated that she would like to participate in the study, she underwent an informed consent proc ess in which she read and signed the IRBapproved Informed Consent Form. Following inform ed consent, she underwent a brief screening assessment. If the results of the screening assessment were unremarkable, a psychosocial assessment interview was be scheduled for the day of the participant s pre-operative medical appointment at the Gynecologic Oncology Clinic. At the time of the pre-operative appointment, participants underwent a one-hour psychosocial interview in a priv ate room in the Gynecologic Oncology Clinic. Following the psychosocial interv iew, participants were be reimbursed $20 for meal and parking expenses. No psychosocial data was collected on the day of surgery. Post-surgical complication data was obtaine d from inpatient a nd outpatient medical records during the period from the date of surgery until the date of discharge. Biobehavioral data (e.g., age, disease stage, medications, co-morbid illn esses) were abstracted from medical records and were also based on pa rticipants self-report.

PAGE 37

37 Psychosocial Assessment The following psychological/psychiatric m easur es were completed prior to study entry to determine participants eligibility: Beck Scale for Suicide Ideation (BSS; Beck & Steer, 1991-1993): The BSS is a 21-item, self-report measure of the presence and severity of suicidal ideati on. The reliability of the BSS is well-established, with coeffi cient alphas ranging from .87-.90 (Beck & Steer, 1991). The concurrent validity of the BSS is demonstrated by moderate to high co rrelations with other measures of suicidal construc ts (Beck & Steer, 1991). Althou gh little published data exist regarding the use of the BSS as a screening tool among cance r populations, it has been used extensively among inpatient and outpatient psychiatric populations (Pinninti, Steer, Rissmiller, Nelson, & Beck, 2002). Women who re ported current suicidal idea tion, intent, or plan were referred immediately to the Psycho-Oncology Clinic at the Psychology Clinic (under the supervision of Deidre Pereir a, Ph.D., licensed psychologist) as well as Psychiatry. Women reporting current suicidal ideation, in tent, or plan were not eligible for participation in this study (see exclusion criteria noted above). Psychotic Screening Module of the Structu red Clinical Interv iew for DSM-IV for non-clinical populations (SCID-NP; Spitzer, Williams, Gi bbon, & First, 1992). The SCID-NP is a semi-structured interview for making DSM-IV Axis I psychotic diagnoses in non-psychiatric populations. The SCID-NP has been used widely as a brief screening measure of psychotic disorders among patients with medical illness, such as HIV (e.g., Penedo, et al., 2003). Women with current psychotic symptoms were referred immediately to Psychiat ry for evaluation and treatment. Women with current psychotic symptoms were not eligible for participation in this study (see exclusion criteria noted above).

PAGE 38

38 The following psychosocial variables were measured at the presurgical assessment as part of specific aims: Demographics The MacArthur Sociodemographic Questionna ire(MSQ; Ad ler, Epel, Castellazzo, & Ickovics, 2000) is a questionna ire developed by the MacArthur Foundation that assesses subjective and objective social status. To assess subjective social status, participants indicate their perceived standing in the community and the country by marking their standing on a picture of a ladder with ten rungs. A va riety of traditional socioecono mic status questions such as education level, employment status, and income assess objective social status. The MSQ was completed at participants pre-operative visit (approximately one day prior to surgery). Medical comorbidity The Charlson Com orbidity Index (Charlson, Po mpei, Ales, MacKenzie, 1987) was used to assess medical comorbidity. The Charlson Comorbidit y Index is a weighted index that takes into account the number, as well as seve rity of comorbid diseases. Pre-surgical life stress An abbreviated m easure of the Life Experi ences Survey (LES; Sarason, Johnson, & Siegel, 1978) was used to measure pre-surgical life stre ss. This abbreviated LES measure was devised by Leserman and colleagues at the University of North Carolina for use with chronically ill populations. The version that will be used in the proposed research has been further tailored for use with cancer patients by Pereira and Jensen at the University of Florida by anchoring health items to the experience of bei ng diagnosed with cancer (LES-C ancer, LES-C). This 34-item life stress measure assesses the number of negativ e life events during the past 6 months. Additionally, participants use a 5-point scale ranging from Not Stressful to Extremely Stressful to rate the impact of each negative life event when it occurred.

PAGE 39

39 Emotional support from primary support person The Sources of Social Support Scale (SSSS; Carver, 1999) w as used to assess emotional support. The SSSS is a 50-item self-report questionnaire measuring receipt of both positive and negative emotional and instrumental social support from spouse/partner, friends, adult female family members, other family members, and heal th care providers. For ea ch item, participants rate the frequency with which they receive a specif ic type of social support using a 5-point Likert scale ranging from Not at all to A lot. For the present stud y, self-reported receipt of positive emotional support from spouse/partner was used in the specific aims. For women without a spouse/partner, the self-reporte d receipt of positive emotional support from friends was used. The following psychosocial variables were measured at the presurgical assessment as part of exploratory analyses: Pre-surgical perceived stress The Perceived Stress Scale (PSS; Cohen, Ka m arck, & Mermelstein, 1983) was used to measure pre-surgical perceived stress. The PSS is a 14-item scale measuring the degree to which situations in the past month are perceived as st ressful. The scale was developed to evaluate the degree to which people find their lives unpred ictable, uncontrollable, and overloading. Participants use a 5-point Like rt scale ranging from Never to Very Often to indicate how often they have experienced each item in the pa st month. Higher scores on the PSS indicate higher levels of perceived stre ss. The PSS has coefficient alpha reliability in healthy samples ranging from .84 to .86 (Cohen et al., 1983) and ha s good reliability in studies of individuals with physical illness (coefficient alpha .86-.89; Soderstrom, Dolbier, Leiferman, & Steinhardt, 2000). The PSS is moderately correlat ed with life stress measures ( r =.20-.35) as well as with psychological and medical constructs such as depressive symptomatology ( r =.76) and physical

PAGE 40

40 symptomatology ( r =.52) (Cohen et al., 1983). The valid ity of the PSS has also been demonstrated in gynecologic samples (Kain, et al., 2000). Total emotional support from all sources The Sources of Social Support Scale (SSSS; Carver, 1999) w as used to assess total emotional support from the following sources: s pouse/partner, friends, adult female family members, other family members, and health car e providers. A more comp rehensive description of this measure can be found in the previous section. Physiologic Assessment Salivary cortisol is a reliable reflection of free cortisol leve ls in the blood (Kirschbaum & Hellhamm er, 1994). Participants were asked to collect saliva samples at 8:00 AM, 12:00 PM, 5:00 PM, and 9:00 PM each day for three consecu tive days prior to their pre-operative visit. Participants collected samples at home with Salivette (Sarstedt, Inc., Newton, NC) devices, which consist of a cotton role encased in a pl astic centrifuge tube. Because non-compliance with the sampling schedule can affect the validity of the cortisol levels (Kudielka, Broderick, & Kirschbaum, 2003), participants were encouraged to note the time of sa mple collection if it deviated from the times specified above. Samples were refrigerated or kept in an insulated cooler until participants returned them at their pre-opera tive clinic visit and were transported to the College of Nursing Biobehavioral Research Laboratory for prepara tion and storage. Saliva samples were centrifuged for 15 minutes at 3000 RPM and aliquots were frozen at 70C until assayed. Cortisol assays were conducted at the Behavior al Medicine Research Center Neuroendocrinology Laboratory in th e Department of Psychology at the University of Miami and at Salimetrics Inc. (State College, PA) by tr ained laboratory personnel. Cortisol levels were assessed using a commercially available Enzy me-Linked Immunosorbent Assay (ELISA) kit (Salimetrics, Inc., State College, PA). ELISA is laboratory technique frequently used in studies

PAGE 41

41 involving immune factors such as an tigens and antibodies, as well as hormones. It consists of a surface with an antibody attached to it. The antibody will connect itself to the substance being measured (e.g., the hormone of interest). A mixtur e of a purified version of the substance being measured mixed with an enzyme and the test sample are added to the test system. If there is none of the substance of interest present in the test sample, then only the substance of interest connected with enzyme will bind to the surface. This results in a change in the color of the solution, which can be measured with a plate read er to determine how much of the substance of interest is present in the test sample. ELISA measures the concentrati on of cortisol using a microtiter plate which is coated with rabbit anti bodies to cortisol. Sta ndard and unknown cortisol competes with cortisol linked to horseradish peroxidase for the antibody binding sites. Following a period of incubation, the unbound components were washed away and the bound cortisol was measured by the reaction with the substrate. This reaction produced a blue color which was read on a standard plate reader. ELISA is based upon the principle that the inte nsity of the color is proportional to the concentration of cortisol. A da rker color indicates lo wer concentrations of cortisol. Salivary Cortisol Slope Calculation A growing body of research has exam ined dive rse aspects of cortisol in response to psychosocial factors. Vedhara, Stra, Miles, Sa nderman, & Ranchor (2006) describe four measures of cortisol commonly measured in PNI research: area under the curve with respect to ground (AUCg), area under the curve with respect to increase (AUCi), early morning cortisol peak, and diurnal cortisol slope Although often used interchang eably under the umbrella of cortisol, recent research suggests that these measures of cortisol may reflect varied and distinct processes. For example, diurnal cortisol is a measure of the pattern of cortisol production over time, whereas the early morning peak, which is considered to be a pr oxy measure of diurnal

PAGE 42

42 cortisol, is representative of HP A axis reactivity to waking (Vedha ra et al., 2006). In contrast, the AUC approach is calculated on based upon the tr apezoid formula in orde r determine the overall production of cortisol over time (Vedhara et al., 2006). Thus AUCg reflects the basal HPA activity whereas AUCi reflects HPA reactivity (Vedhara et al., 2006). Based on findings suggesting that diurnal slope is related to health outcome (i.e., severity of disease, mortality) whereas mean cortisol levels or area under th e curve may not be (Abercrombie et al., 2004; Sephton, Sapolsky, Kraemer, & Spiegel, 2000), cortisol diurnal slope was examined as the primary measure of cortisol and AUCg, AUCi, mean daily cortisol, and morning cortisol were examined as secondary measures of cortisol. Di urnal cortisol slope we re be calculated using standard linear regression. This method is consistent with that reported by Abercrombie et al. (2004), Sephton et al., (2000), Gi ese-Davis, Sephton, Abercrombie, Duran, and Spiegel (2004), Giese-Davis, Dimiceli, Sephton, and Spiegel ( 2006), and Vedhara et al. (2006). Because the normal diurnal cortisol slope is characterized by a peak upon waking followed by a subsequent decline throughout the day (Stone et al., 2001), cor tisol concentrations were log transformed in an effort to normalize the distri bution and linearize the change in cortisol over the day (Vedhara et al., 2006). A series of regressions with the time of sample co llection as the outcome variable and the cortisol value for each day (4 timepoint s X 3 days) as the predictor variables were performed individually for each participant to qu antify each persons diurnal cortisol slope using unstandardized beta weights. Based upon prev ious research (e.g., Vedhara et al., 2006), unstandardized beta weights clos er to zero are indicative of more abnormal diurnal slopes (characterized by flattened slopes, abnorma l peaks, and abnormal troughs), whereas unstandardized beta weights further from zero are indicative of more normal diurnal slopes.

PAGE 43

43 Although hierarchical linear modeling (HLM) may also be an appropriate form of cortisol slope estimation (e.g., Stone et al., 2001), the linear regression model described above has demonstrated equivalent results with HLM (Rogosa & Saner, 1995). Additionally, given that among 90% of healthy individuals the diurnal cortisol slope is characterized by a peak upon waking followed by consistent decline (Stone et al., 2001), the use of a linear model is appropriate. Diurnal cortisol slope was identified as the meas ure of cortisol in th e specific aims. Mean daily cortisol, mean morning cortisol, AUCg, and AUCi were designated as the measures of cortisol in exploratory aims. Surgical Outcome Assessment The following surgical outcom e variables were measured during the post-surgical hospitalization period as pa rt of specific aims: Severity of post-surgical complications Data regarding participants pos t-surgical complications were abstracted from medical records. Information was abstracted from ope rative reports, discharg e summaries, diagnostic testing reports, and Gynecologic Oncology and nu rsing notes documented w ithin patients paper and electronic medical charts on the hospital unit as well as at the UF&Shands Gynecologic Oncology Clinic. The determination of whether an event/condition constituted a complication was determined based on the gynecologic surg ery literature (e.g., Nic hols & DeLancey, 1995; Tozzi et al., 2005; Vasilev, 2000). Figure 2-2 lists the events and conditions that will be classified as complications for the proposed st udy. The severity of each complication was rated on a scale from 1 (minimally severe) to 3 (extremely severe) by a board -certified gynecologic oncologist (Daylene Ripley, MD). The we ighted sum of documented post-operative complications was utilized as the outcome of intere st. It should be noted th at the validity of this

PAGE 44

44 method of data collection is dependent upon th e accuracy and consistency of documentation by healthcare providers in the medical record. Thus, in an effort to obtain the most comprehensive information regarding individuals post-surgical recovery, all av ailable sources of information regarding post-surgical recovery in the medical record was review ed. Three additional raters who were blinded to study design and hypotheses revi ewed medical records an d rated post-surgical complications to establish inter-rater reliability with the primary rater ( = .92). Length of post-surgical hospitalization The duration of particip ants pos t-surgical hospitalization was abstracted from discharge summaries documented in participants medical records. Systemic immunosuppression White blood cell (W BC) count is a clinica lly important index of systemic immune response during the acute post-surgical period. Th e post-surgical immune response is typically characterized by leukocytosis, neutrophilia, and lymphocytopenia (Salo, 1992). Participants WBC counts during post-surgical hospitalization, as well as leukocyte subset differential percentages was abstracted from medical r ecords. Less elevated WBC counts during postsurgical hospitalization is considered indicative of systemic immunosuppression. Time to ambulation Inform ation about days to ambulation was abstracted from participants discharge summaries as well as notes documented on patien ts paper medical charts on the hospital unit. Time to ambulation was quantified as the post-op erative day on which ambul ation first occurred. Biobehavioral variables Inform ation about participants age, disease stage, medication use, comorbid illnesses, health behaviors (e.g., use of tobacco, alcohol, illi cit drugs), additional surgical procedures (e.g., panniculectomy or hernia repair at the time of TAH-BSO), intra-operative complications (e.g.,

PAGE 45

45 inadequate hemostasis), intra-operative blood loss anesthetic dose, and post-operative analgesic use were abstracted from medical records. Additionally, the MSQ (see above) was used to measure sociodemographic control variables. These variables have been traditionally examined descriptively and as pot ential confounds in analyses with physiologic and disease outcomes. A priori biobehavioral control variables included age, FIGO stage, BMI, medical comorbidity, post-surgical pain, and length of surgery. The following surgical outcome variables were measured during the post-surgical hospitalization period as part of exploratory analyses: Post-surgical pain ratings Inform ation about participants post-surgical pain ratings was abstra cted from their vital signs as documented in their inpatient medical re cord. Pain ratings were based on a scale from zero to ten, with ten indicati ng the worst pain level. Statistical Procedures The distributions of all variable s were exam ined to confirm that parametric tests could be used. Any variables with non-normal distributions were transformed accordingly. Relations among a priori biobehavioral c ontrol variables and outcome m easures were tested using correlations. Biobehavioral contro l variables significantly related to outcome measures at p .05 were controlled for in subsequent analyses. Analyses of Specific Aims Aim 1 : To exam ine the relations among peri-opera tive psychosocial st ress, pre-surgical diurnal cortisol slope, and indi ces of post-surgical recovery. To examine support for hypothesis 1a, which states that women who report greater presurgical psychosocial stress will have a more complicated post-surgical recovery, a series of hierarchical linear regression models was used. For each hierarchical regression, biobehavioral

PAGE 46

46 control variables were entered as the Block 1 predictor, impact of negative life events (measured by the LES-C) were entered as the Block 2 predicto r, and diurnal cortisol slope was entered as the Block 3 predictor. Separate regressions were planned for each of the following outcome variables: systemic immunosuppression, post-surg ical complications, time to ambulation, and length of post-surgi cal hospitalization. To examine support for hypothesis 1b, which states that pre-surgical diurnal cortisol slope will mediate the relationship between pre-surgical psychosocial stress and indices of surgical recovery, the methods specified by Baron and Kenny (1986) were used. In or der for mediation to be supported, three conditions must be met. First, the hypothesized predictor variable, impact of negative life events, must be related to the outco me variable, index of post-surgical recovery. Next, impact of negative life events must be related to the mediating variable, diurnal cortisol slope. Finally, the relationship between impact of negative life events and index of post-surgical recovery must be weakened (p artial mediation) or eliminated (full mediation) when the mediating variable, diurnal cortis ol slope, is controlled. Throughout this process, the mediator, diurnal cortisol slope, mu st remain significant. Aim 2 : To examine the relations among peri-operative emotional support, pre-surgical diurnal cortisol slope, and indi ces of post-surgical recovery. To examine support for hypothesis 2a, which st ates that women who report greater presurgical emotional support will have a less co mplicated post-surgical recovery, a series of hierarchical linear regression models was us ed. For each regression, biobehavioral control variables were entered as the Block 1 predicto r, pre-surgical positive emotional support was entered as the Block 2 predictor, and diurnal cortisol slope was en tered as the Block 3 predictor. Separate regressions were planned for each of the following outcome variables: systemic

PAGE 47

47 immunosuppression, post-surgical complications, tim e to ambulation, and length of post-surgical hospitalization. To examine support for hypothesis 2b, which states that pre-surgical diurnal cortisol slope will mediate the relationship between pre-surgical emotional support and indices of surgical recovery, the methods specified by Baron and Kenny (1986) were used. In or der for mediation to be supported, three conditions must be met. Firs t, the hypothesized predictor variable, positive emotional support, must be related to the outco me variable, index of post-surgical recovery. Next, positive emotional support must be related to the mediating variable, diurnal cortisol slope. Finally, the relationship between positive emotiona l support and index of pos t-surgical recovery must be weakened (partial mediation) or el iminated (full mediation) when the mediating variable, diurnal cortisol slope is controlled. Throughout this process, the mediator, diurnal cortisol slope, must remain significant. Aim 3 : To examine whether pre-surgical em otional support moderates the potential relations among peri-operative ps ychosocial stress and indice s of post-surgical recovery. To examine support for hypothesis 3, which states that among women with high presurgical psychosocial stress, those who report le ss pre-surgical emotional support will have a more complicated post-surgical recovery, the hi erarchical regression me thods specified by Aiken and West (1997) were planned. Biobehavioral co ntrol variables were en tered as the Block 1 predictor. In the second block, the main effect s of impact of negative life events and positive emotional support were entered. In the third bloc k, the interaction betwee n impact of negative life events and positive emoti onal support were entered. Power and Sample Size Considerations A s mall number of studies have examined ps ychosocial predictors of surgical outcome (e.g., Broadbent et al., 2004; Kopp et al., 2003 ; Krohne & Slangen, 2005). A review of this

PAGE 48

48 literature reveals effect sizes between psychosocial and immune variables in the medium to large range, according to Cohens effect size conventions (Broadbent et al., 2004; Kopp et al., 2003; Krohne & Slangen, 2005). Tabl e 2-2 lists calculated effect sizes for the relationship between psychosocial and immune va riables in the cancer literature. NCSS PASS software was used to perform a power analysis for the proposed study based upon the findings examining psychosocial predic tors of length of hospitalization (Krohne & Slangen, 2005) and of post-operative complications (Contrada et al., 2004). Power analysis revealed that a sample size of 70 provides 80% power to detect a change in slope from 0 under the null hypothesis to .32 under the alternative hypothesis when the standard deviation of the Xs is 1, the standard deviation of Y is 1, and the two-sided significance level is .05 (See Figure 2-3). A small number of studies have examined psychoneuroimmunologi c aspects of women with gynecologic malignancies or those at-ris k for developing gynecol ogic malignancies (e.g. Byrnes et al., 1998; Lutgendorf et al., 2002; Pe reira et al, 2003a; Pere ira et al., 2003b). Among studies that have examined relations among ps ychosocial functioning and cortisol, the effect sizes for the relationship between psychosocial variables and cortisol fall in the medium to large range. Table 2-3 lists calculated effect sizes for the relationship between psychosocial and immune variables in the cancer literature. NCSS PASS software was used to perform a power analysis for the proposed study based upon the findings of Gallagher-Thompson et al. ( 2006). Power analysis rev ealed that a sample size of 60 provides 81% power to detect a change in slope from 0 under the null hypothesis to .35 under the alternative hypothesis when the standard deviation of the Xs is 1, the standard deviation of Y is 1, and the two-sided significan ce level is .05. Given th at the power analysis regarding the aims examining post-surgical recove ry as outcome variable is greater than the

PAGE 49

49 sample size recommended by the power analysis examining cortisol as an outcome variable, the proposed study will use the conservative es timation of N = 70 (See Figure 2-4).

PAGE 50

50 Table 2-1. Demographic charac teristics of participants Total N 75 N with psychosocial data 75 N with cortisol data 62 N with length of hospitalization data 75 N with post-surgical complications data 75 N with ambulation data 43 N with WBC data 75 Age, Mean (SD) 60.71 (9.65) Highest Degree (% of sample) Less than high school 12.3 High School/GED 39.7 Associate degree 13.7 Bachelors degree 17.8 Masters degree 9.6 Professional degree 1.4 Other 5.5 Household Income (% of sample) Less than $5,000 2.7 $5,000 $11,999 5.4 $12,000 $15,999 13.5 $16,000 $24,999 9.5 $25,000 $34,999 13.5 $35,000 $49,999 14.9 $50,000 $74,999 14.9 $75,000 $99,999 4.1 $100,000 or greater 9.5 Dont know 5.4 Race (% of sample) Multiracial 1.3 Black/African American 8.0 White 90.7 Ethnicity (% of sample) Hispanic/Latino 5.3 Not Hispanic/Latino 94.7 Marital Status (% of sample) Married 56.2 Never married 4.1 Separated 2.7 Divorced 17.8 Widowed 19.2 Employment Status (% of sample) Working full-time 32.4 Working part-time 12.2 Unemployed 1.4 Keeping house/Raising children 6.8 Retired 47.3

PAGE 51

51 Table 2-2. Surgical re covery effect sizes Study Population N R2 Broadbent et al., 2003 Inguinal he rnia surgery patients 47 .25 .34 Kopp et al., 2003 General surgery patients 112 .04 .76 Krohne & Slangen, 2005 Maxillofacial surgery patients 84 .25

PAGE 52

52 Table 2-3. Psychosocialcortisol effect sizes Study Population N d R2 Giese-Davis et al, 2004 Metastatic breast cancer 91 .90 Giese-Davis et al., 2006 Metastatic breast cancer 29 .37 Vedhara et al., 2006 Breast cancer 59 .10 Gallagher-Thompson et al., 2006 Dementia caregivers and non-caregivers 83 .40 Turner-Cobb et al., 2004 Metastatic breast cancer 72 .04

PAGE 53

53 Figure 2-1. Design timeline.

PAGE 54

54 Immune Sepsis Hematologic Deep Vein Thrombosis Febrile Thromboembolism Wound complication Venous stasis Infection Low hematocrit Cardiovascular Arrhythmias Transfusion Tachycardia Hypoxemia Myocardial infarction Superficial phlebitis Abnormal blood pressure Renal Low urine output Gastrointestinal Emesis Abnormal creatnine Intestinal Fistula Uretal obstruction Slow return of bowel function Urinary tract infection Laryngospasm Acute renal failure Severe nausea/vomiting Urinary fistula Diarrhea Bladder dysfunction Ileus Acute tubular necrosis Pancreatitis Procedure related Central line hematoma Bowel Obstruction Central line air embolism GI tract Bleeding Venous catheter infection Pulmonary Brochospasm Ve nous catheter thrombosis Pulmonary Embolism Venous catheter obstruction Pneumonia Infection--paracentesis Atelectasis Adverse medication response Pneumothorax Miscellaneous Pain Pulmonary edema Post-surgical ascites Adult respiratory distress Hypothermia Respiratory infection Decubitus ulcer Low O2 saturation Delirium Bronchial inflammation Dizziness Cough with sputum Pelvic hematoma Pleural effusion Aspiration Perihilar airspace disease Lymphocyst Hypoxic event Peritoneal Pneumoperitoneum Respiratory insufficiency Cerebrovascular Stroke Dyspnea Figure 2-2. Categories of pos t-surgical complications.

PAGE 55

55 Figure 2-3. Power Versus Sample Si ze With Alpha = .05 and Slope = .32

PAGE 56

56 Figure 2-4. Power Versus Sample Size With Alpha = .05 and Slope = .35

PAGE 57

57 CHAPTER 3 RESULTS Preliminary Analyses Biological Variables Raw cortisol values were log transform ed pr ior to calculating diurnal slope as per the methodology of Sephton et al. (2000)1. A mean diurnal cortisol slope was created using all cortisol data points for each participant. Mean WBC count during the acute post-surgical hospitalization period was cal culated by computing an average WBC count for each hospitalization day blood comple te blood cell analysis was performed and then averaging the daily WBC count values. Outcome Variables Descriptive analyses exa mined the normality of the distributions of the main outcome variables. Outliers (defined as greater than th ree standard deviations from the mean) were eliminated from analyses. The normal distributio ns of WBC count and po st-surgical ambulation data were confirmed. Length of pos t-surgical hospitalization data were log-transformed to permit the use of parametric analyses. Post-surgical complication data were square root-transformed to permit the use of parametric analyses. Control Variables Pearson bivariate correlations were perform ed to assess for potential relations among a priori continuous biobehavioral c ontrol variables and the main vari ables of interest (Table 3-1). 1 Given that the methodology with regard to the lo g transformation of cortisol during the process of slope calculation is poorly desc ribed in the literature, a random sample of analyses were rerun using diurnal cortisol slope which was calcula ted from the raw cortisol concentrations. A constant of was added to each slope calcula tion and slopes were then log transformed. The use of slopes calculated in this alternative manner failed to produ ce different results than those reported based on cortisol slopes calculated from log-transformed cortisol values.

PAGE 58

58 Spearmans correlations were performed to asse ss for potential relations among a priori ordinal biobehavioral control variables (e .g. FIGO stage) and the main vari ables of interest. Independent samples t-tests were performed to assess for re lations among a priori categorical biobehavioral control variables and the main variables of inte rest (Tables 3-2 and 3-3). Independent samples ttests also examined potential di fferences between participants who collected saliva samples to determine cortisol levels and those who did not (Table 3-4). Ther e were no significant differences in main outcome vari ables between those participants with cortisol data and those without cortisol data; however, participants w ho collected cortisol samples had a marginally shorter length of hospitalization than participants who did not ( p = .06). Relations Among Length of Hospitalization and Control Variables Age was significantly related to length of hospita lization ( r = .27, p < .05), such that older women had a longer length of hospitalizati on. Length of surgery (in minutes) was also significantly related to le ngth of hospitalization (r = .35, p < .01), such that greater length of surgery was associated with longer post-surgica l hospitalization. Conseque ntly, age and length of surgery were controlled for in subsequent analys es examining length of hospitalization. No other a priori control variables were rela ted to length of hospitalization. Relations Among Post-surgical Complications and Control Variables Medica l comorbidity was significantly related to post-surgical complications ( r = .24, p < .05), such that women with great er pre-surgical medical comorbidity experienced more severe post-surgical complications. Thus, medical comorbidity was controlled for in subsequent analyses examining post-surgical complications. No other a priori control variables were related to post-surgical complications.

PAGE 59

59 Relations Among Post-Surgical Am bulation and Control Variables There were no significant relati onships between a priori contro l variables and post-surgical am bulation. Relations Among WBC Count and Control Variables There were no significant relati onships between a priori contro l variables and post-surgical WBC count. Relations Among Stress and Control Variables Age was significantly related to both im pact of negative life events ( r = -.26, p < .05) and number of negative life events ( r = -.29, p < .05), such that younger women reported more negative life events and greater impact of negative life events during the past six months. Relations Among Emotional Support and Control Variables Pre-surgical BMI was significantl y related to em otional support (r = -.36, p < .01), such that women with lower BMI perceived greate r emotional support from primary support person prior to surgery. Mean post-surgical pain during hospita lization was also significantly related to emotional support ( r = -.25, p < .05), such that women who pe rceived greater emotional support from primary support person pre-surgically repo rted lower mean post-surgical pain during hospitalization. Relations Among Diurnal Cortisol Slope and Control Variables There were no significant re lationships between a priori control variables and diurnal cortisol slope. Descriptive Results Participan ts ( N = 75) were a mean of 60.71 years old ( SD = 9.65) at study entry. They had a mean BMI of 35.86 ( SD = 11.22). Thirteen participants (1 7.3%) had a BMI within the normal range (18.5 24.9), 14 participants (18.7%) had a BMI within the overweight range (25.0 -29.9),

PAGE 60

60 and 48 participants (64%) had a BMI within th e obese range (>30.0). Participants had a mean medical comorbidity index score of 2.65 ( SD = 1.40). At study entry, 2.7% of participants were taking corticosteroid medications. The mean length of surgery was 177.14 minutes ( SD = 55.75) and 14.7% of participants experienced intra-ope rative complications. Participants reported a mean pain rating of 2.45 ( SD = 1.52) on a scale from zero to ten during the post-surgical hospitalization period. Medical reco rds revealed that common postsurgical analgesics included Morphine PCA, Tylox, Motrin 600, Percocet, an d Dilaudid; however, not all participants received all of these me dications and the present study did not collect data on analgesic doses for each participant. Pathology revealed that 48 partic ipants (64.0%) had Stage I endometrial cancer, 11 (14.7%) had Stage II endometri al cancer, and 10 (13.3% ) had St age III endometrial cancer. Six participants (8%) were identified as having benign or pre-cancerous endometrial disease. According to pathology reports, 62 participants (84.9%) had a histology type of endometrioid adenocarcinoma, three participants (4.1%) had a histology type of complex endometrial hyperplasia with atypia, three participants (4.1 %) had a histology type of papillary serous carcinoma, one participant (1.4%) had a histol ogy type of complex endometrial hyperplasia without atypia, one participan t (1.4%) had a histology type of clear cell carcinoma of the endometrium, one1 participant (1.4%) had a histol ogy type of mixed clear cell and endometrial adenocarcinoma, and one participant (1.4 %) had a histology type of endocervical adenocarcinoma, endometrioid type. Participants reported expe riencing a mean of 3.45 ( SD = 2.87) negative life events during the six months prior to surgery and a mean impact of negative life events of 9.66 ( SD = 11.35). The five most common negative life events duri ng the past six months included major illness related to cancer (64.4% of par ticipants), worsening financial situation (24.3%), major illness

PAGE 61

61 other than cancer (18.9%), worked long hours (18.1%), and change in closeness to a family member (13.5%). Participants reported percei ving a mean emotional support from primary support person of 16.57 (SD = 3.56). The mean diurnal slope was -.04 ( SD = .02). Figure 3-1 displays examples of a selection of st udy participants diurnal cortisol slopes. Participants were hospi talized a mean of 4.38 ( SD = 1.98) days following surgery. Fortysix percent of women experienced post-surgi cal complications duri ng the post-surgical hospitalization period. Participants experienced a mean of .98 ( SD = 1.46) post-surgical complications during hospitalization and their m ean post-surgical complication severity rating was 1.12 ( SD = 1.82). Common post-surgical compli cations included fever (12.2% of participants), blood transfusion (10.8%), atelec tasis (9.5%), low hematocrit (9.5%), pleural effusion (6.8%), urinary tract infection (5.4 %), low oxygen saturation (5.4%), arrhythmias (4.1%), and, aspiration (4.1%). Particip ants first ambulated a mean of 1.28 ( SD = .88) days postsurgery. Their mean WBC count during post-surgical hosp italization was 10.60x109 cells/L ( SD = 2.60x109 cells/L). Differential complete blood ce ll count data revealed that the mean neutrophil percentage was 77.50% ( SD = 8.61%) and the mean lymphocyte percentage was 15.39% ( SD = 7.72%). 49.3% of particip ants had mean WBC counts th at fell within the normal range (5.0-10.0x109 cells/L) and 50.7% of partic ipants had leukocytosis (>10.0x109 cells/L). No participants had mean WBC counts indicating leukopenia (< 5.0x109 cells/L). Relations Among Stress, Cortisol, and Surgical Outcome Length of Hospitalization Based on results of prelim inary analyses, age an d length of surgery were entered as control variables in block one for all analyses examini ng length of hospitalization (in log days). These control variables accounted for 21.3% of the va riance explained in length of hospitalization. Significant effects for length of surgery ( p < .01) indicated that women who underwent longer

PAGE 62

62 surgeries had longer post-surgic al hospital stays. Age emerge d as a marginally significant predictor of post-surgical length of hospitalization, such that there was a trend toward older women experiencing longer postsurgical hospital stays (p = .08). The addition of impact of negative life events in block two, and the addition of pre-surgical diurnal cortisol slope in block three did not significantly increase the vari ance explained in length of post-surgical hospitalization (ps > .50). The main effect analysis for the model is summarize d in Table 3-5. In the full model, only length of surgery emer ged as significant predictor of length of hospitalization, with a trend toward age as a predictor of length of hospitalization. Due to the lack of significant relations among impact of nega tive life events, pre-surg ical diurnal cortisol slope, and length of hosp italization, test for medi ation was not performed. Post-Surgical Complication Severity Based on results of prelim inary analyses, me dical comorbidity score was entered as a control variable in block one, impact of negativ e life events was entered in block two, and presurgical diurnal cortisol slope was entered in block three. However, none of these variables accounted for significant variance in pos t-surgical compli cation severity ( ps > .10). Table 3-6 summarizes the main effects model. Due to the lack of significant relations among impact of negative life events, pre-surgical diurnal cortisol slope, and post-surgica l complication severity, test for mediation was not performed. Post-Surgical Ambulation A m ain effects model was created by entering impact of negative life events in the first block and pre-surgical diurnal cortisol slope in the second block of a hierarchical linear regression. Neither impact of life events nor pre-surgical diurna l cortisol slope significantly accounted for the variance explained in post-surgical time to ambulation ( ps > .10). Table 3-7 summarizes the main effects model. Due to the lack of significant relations among impact of

PAGE 63

63 negative life events, pre-surgical diurnal cortisol slope, and post-surgical time to ambulation, test for mediation was not performed. WBC Count A m ain effects model was created by entering the impact of negative life events in the first block and pre-surgical diurnal cortisol slope in the second block of a hierarchical linear regression. Neither impact of ne gative life events nor pre-surg ical diurnal cortisol slope significantly accounted for the vari ance explained in WBC count ( ps > .50). Table 3-8 summarizes the main effects model. Due to the lack of significant relations among impact of negative life events, pre-surgical diurnal cortisol slope, and po st-surgical WBC count, test for mediation was not performed. Relations Among Emotional Support, Cortisol, and Surgical Outcome Length of Hospitalization Age and length of surgery were entered as control variables in block one, emotional support from prim ary support person was entered in block two, and pre-surgical diurnal cortisol slope was entered into block three. The addition of emotional support from primary support person and pre-surgical diurnal cortisol slope did not significa ntly account for the variance explained in length of post-s urgical hospitalization above a nd beyond the control variables ( p s > .10). Please refer to table 3-9 for the main effect analysis for the model. Due to the lack of significant relations among pre-surgical emo tional support from primary support person, presurgical diurnal cortisol slope, and length of hospitalization, test for mediation was not performed. Post-Surgical Complication Severity Medical co morbidity score was entered as a control variable in block one, emotional support from primary support person was entered in block two, and pre-surgical diurnal cortisol

PAGE 64

64 slope was entered in block three. None of these variables accounted for si gnificant variance in post-surgical complication severity ( ps > .10). Table 3-10 summarizes the main effects model. Due to the lack of significant relations am ong emotional support from primary support person, pre-surgical diurnal cortisol sl ope, and post-surgical complication severity, test for mediation was not performed. Post-Surgical Ambulation A m ain effects model was created by ente ring emotional support from primary support person in the first block and pr e-surgical diurnal cortisol sl ope in the second block of a hierarchical linear regression equation. Neither emotional suppor t from primary support person nor pre-surgical diurnal cortisol slope significantly accounted fo r the variance explained in postsurgical time to ambulation ( ps > .10). Table 3-11 summarizes the main effects model. Due to the lack of significant relations among pre-su rgical emotional support from primary support person, pre-surgical diurnal cortisol slope, a nd time to post-surgical ambulation, test for mediation was not performed. WBC Count A m ain effects model was created by ente ring emotional support from primary support person in the first block and pr e-surgical diurnal cortisol sl ope in the second block of a hierarchical linear regression equation. Emotional support from primary support person emerged as a significant predicto r of post-surgical WBC count, such th at women with grea ter pre-surgical emotional support from primary support person ha d less elevated post-surgical WBC counts ( p < .05). The addition of pre-surgical diurnal cortisol slope in block two of the model did not result in a significant increase in variance e xplained in post-surg ical WBC count (p > .90). The main effect analysis for the model is summarized in Table 3-12. Although a si gnificant relationship emerged between pre-surgical emotional support from primary support person and post-surgical

PAGE 65

65 WBC count, pre-surgical diurnal cortisol sl ope and post-surgical WBC count were not significantly related; thus, test for mediation was not performed. Given that the relationship between emoti onal support from primary support person and post-surgical WBC count was not in the hypothesized direction, a logistic regression equation examined emotional support from primary support pe rson and pre-surgical di urnal cortisol slope as predictors of WBC count clinical categor ies, ranging from 0.0-10.0x109 cells/L and greater than 10.0 x109 cells/L (leukocytosis). The entry of emotional support from primary support person in block one and pre-surgic al diurnal cortisol slope in block two did not significantly account for the variance explained in pres ence of pre-surgical leukocytosis ( ps > .10). Table 313 summarizes the main effects model. Additionally, emotional support from primary support person and pre-surgical diurnal cortisol slope were examined as predictors of differential blood cell count. Neither emotional support from primary support person nor pre-su rgical diurnal cortisol slope significantly accounted for variance explained in either neut rophil percentage or lymphocyte percentage (p s > .60). Table 3-14 summarizes the main effects model for neutrophil percentage and Table 3-15 summarizes the main effects model for lymphocyte percentage. Finally, a main effects model was created in which emotional support from primary support person was entered into block one and presurgical diurnal cortis ol slope was entered into block two to examine mean WBC count during the first four days post-surgery (given that the mean hospitalization length was 4.58 days). Emotional support received from primary support person accounted for a marginally signif icant 5.3% increase in variance explained in WBC count during the first 4 days post-surgery ( p = .08); however, pre-surgical diurnal cortisol slope did not account for a significant amount of variance in WBC count in the first 4 days post-

PAGE 66

66 surgery above and beyond this. The direction of the relationship between emotional support from primary support person and WBC count suggested that greater emotional support from primary support person was marginally asso ciated with less elevated WB C count during the first 4 days post-surgery. Table 3-16 summarizes the main effects model. Exploratory Analyses Rationale for Exploratory Analyses Given the fact that im pact of negative life ev ents and diurnal cortis ol rhythm failed to emerge as predictors of the criteria of intere st, exploratory analyses were pursued to assess whether an additional measure of life stress a nd additional indices of cortisol production were associated with outcomes. Specifically, perceived life stress as measured by the PSS over the past month was examined as an additional meas ure of life stress, while AUC-G, AUC-I, mean daily cortisol, and mean morning cortisol were examined as additional measures of cortisol production. Furthermore, total emotional support was examined as an additional measure of emotional support. Several factors contributed to the decision to include perceived stress, additional cortisol indices, and total emotional support in exploratory analyses of the main outcome variables. First, impact of negative events was not significantly related to any of the outcome variables, nor was it significantly related to cortisol. The lack of a relationship with cortisol is inconsistent with a robust literature linking measures of stress to cortisol production (e .g., Seedat, et al. 2003). Hence, it is possible that the lack of relationship between impact of negative events and cortisol may possibly be attributable to the timeframe c overed by the instrument used to measure impact of negative events, the LES. The LES assesses the incidence and im pact of negative life events over the previous six months. As such, it is more of a measure of chroni c stress than acute, presurgical stress. Consequently, it was determined that the measurement of acute, pre-surgical

PAGE 67

67 stress may be more likely to demonstrate a re lationship with pre-surg ical cortisol. The PSS measures perceived stress during the past mont h, which may better reflect the degree of stress experience by participants undergoing diagnos is and treatment for endometrial cancer. Similarly, it is possible that diurnal cortisol slope may not fully captu re the possibility of abnormal cortisol production during the immediate pre-surgical period. Whereas diurnal cortisol slope is a measure of the pattern of cortisol production over course of a day, other cortisol indices may be more reflective of HPA reactiv ity. For example, the early morning peak in cortisol is considered to be representative of HPA axis reac tivity to waking, whereas AUC-G reflects basal HPA activity and AUC-I reflects post -basal HPA reactivity (Vedhara et al., 2006). Thus, given that the present study examines the acute peri-operative period, it is possible that the cancer diagnosis and surgery are more representa tive of acute stressor s than chronic, and therefore it is possible that co rtisol measures reflective of HPA reactivity may provide more useful information above and beyond diurnal cortisol slope. Finally, operationalizing emo tional support as emotional support received only from a primary support person (spouse, or in the absence of spouse, adult female friend) may have excluded potentially valuable emotional support r eceived from secondary sources. The relational cultural theory posits that women commonly identify themselves thr ough their relation to others (Jordan et al., 1991). This may be of particular importance during a time in which they may be undergoing acute stress, such as undergoing cancer diagnosis/treatme nt, or may be in need of tangible assistance or reassurance from others. Thus, it was determined that examining perceived emotional support from all potential sources (e.g. female friends, healthcare providers, other family members) may provide a more comprehensive measure of emotional support than emotional support from primary support person.

PAGE 68

68 Preliminary Analyses AUC-G and AUC-I for each day of collection were averaged to create a m ean AUC-G and AUC-I for each participant. The daily mean cortis ol and daily morning cortisol values for each day of collection were also averaged to create a mean daily cortisol value and a mean morning cortisol value for each participant. Pearson biva riate correlations were performed to assess for potential relations among a priori continuous biobehavioral c ontrol variables and the exploratory variables of interest (Table 317). Spearmans correlations were pe rformed to assess for potential relations among a priori ordina l biobehavioral control variables (e.g. FIGO stage) and the exploratory variables of intere st. Independent samples t-tests were performed to assess for relations among a priori categorical biobehavioral control variable s and the exploratory variables of interest. Age was marginally related to AUC-I ( r = .23, p = .08), such that older age was associated with greater pre-surgical AUC-I. Le ngth of surgery was significantly related to presurgical AUC-I ( r = -.29, p < .05), such that lower pre-surgical AUC-I was associated with longer surgery. Mean post-surgical pain during hospitalization was margin ally related to AUC-I ( r = -.25, p = .07), such that lower pr e-surgical AUC-I was associated with greater mean postsurgical pain during hospita lization. There was a significan t difference in AUC-I between participants who experienced an intra-operative complication ( M = 1.06, SD = 1.14) and those who did not ( M = .99, SD = .53), t = -.28, p < .05. FIGO stage was significantly related to both AUC-G ( r = -.27, p < .05) and AUC-I (r = -.26, p < .05), such that less advanced tumor stage was associated with greater AUC-G and AUC-I. Women taking corticosteroid medications did not differ significantly from those not taking corticosteroids in AUC-G ( t (57) = .93, p = .39), AUC-I ( t (57) = .29, p = .34), mean daily cortisol ( t (59) =.75, p = .32), or mean morning cortisol ( t (57) = .77, p = .70). Participants had a mean AUC-G of 1.82 nmol/l ( SD = .98) and a mean AUC-I of 1.01 nmol/l ( SD = .65). They had a mean daily cortisol of .15 nmol/l ( SD = .09).

PAGE 69

69 Participants had a mean morning cortisol of .27 nmol/l ( SD = .19). It should be noted that AUCG, AUC-I, mean daily cortisol, and mean morning cortisol were all comput ed from raw cortisol values. Age was marginally related to perceived stress ( r = -.21, p = .08), such that younger participants reported marginally greater percei ved stress. Greater perceived stress was also marginally associated with greater post-surgical pain (r = .22, p = .08). Significant relations also emerged between perceived stress and BMI ( r = .31, p < .01), such that women with greater BMI reported more perceived stress. Greater BMI was also significantly associated with lower total emotional support ( r = -.29, p < .05). Women taking corticosteroid medications did not differ significantly from those not taking corticosteroids in terms of perceived stress ( t (68) = -.54, p = .59) or total emotional support ( t (63) = .14, p = .42). Participants who experienced intraoperative complications did not differ from those who e xperienced no intra-operative complications in perceived stress ( t (68) = -.70, p = .49) or in total emotional support ( t (62) = 1.46, p = .15). Participants re ported a mean perceive d stress score of 22.14 (SD = 8.81) and mean perceived total emotional support of 67.86 ( SD = 18.03). To begin with, relations among impact of nega tive life events, percei ved stress, emotional support from primary support person, total emoti onal support, diurnal cortisol slope, the 4 additional indices of cortisol pr oduction, and the criteria of inte rest were assessed by calculating Pearson bivariate correlations. These data are presented in Table 3-18. Then, a series of regression equations were constructed to exam ine whether these additional predictors were associated with the criteria inte rest, as described below. Length of Hospitalization, Stress, and Cortisol Im pact of negative life events /perceived stress and the additional indices of cortisol production (i.e., AUC-G, AUC-I, mean daily cortisol mean morning cortisol) were examined as

PAGE 70

70 predictors of length of hospitalization with a se ries of hierarchical re gression analyses. Eight individual equations were constructed. In the fi rst set of equations, age and length of surgery were entered as control variables into Block 1, impact of ne gative life events was entered into Block 2, and the four additional indices of cor tisol production were entered individually into Block 3, resulting in a total of 4 regression equa tions. In the second set of equations, perceived life stress was substituted for imp act of negative life events in Bl ock 2, resulting in 4 additional regression equations. As presented earlier, impact of negative life events did not em erge as a predictor of length of hospitalization after controlli ng for age and length of surgery (See Table 3-5). However, none of the 4 additional indices of co rtisol production accounted for si gnificant varian ce in length of hospitalization above and beyond ag e/length of surgery and imp act of negative life events. Perceived stress also failed to emerge as a pred ictor of length of hospitalization after controlling for age and length of surgery; furthermore, the four additional indices of cortisol production failed to predict length of hospitalization above and beyond age/length of surgery and perceived stress. Although neither perceived stress nor mean morning cortisol emerged as predictors of length of hospitalization, a significant relationship in the expected direction emerged between perceived stress and mean morning cortisol ( r = .28, p < .05). Specifically, greater perceived stress was associated with more elevated mean morning cortisol (Table 3-18). Post-surgical Complication Seve rity, Stress, and Cortisol Im pact of negative life events /perceived stress and the additional indices of cortisol production (i.e., AUC-G, AUC-I, mean daily cortisol mean morning cortisol) were examined as predictors of post-surgical complication severity with a series of hierarchical regression analyses. The analyses were identical to those describe d above for length of hospitalization; however,

PAGE 71

71 post-surgical complication severity was substi tuted as the outcome variable and medical comorbidity score was substituted for the control variable. As reported in the analyses for the specific aims, impact of negative life events did not emerge as a predictor of post-surgical complication severity after c ontrolling for medical comorbidity score (See Table 3-6). None of th e four additional indices of cortisol production accounted for significant varian ce in post-surgical complicat ion severity above and beyond medical comorbidity score and impact of negative life events. Perceived stress did not emerge as a significant predictor of post-surgical compli cation severity after c ontrolling for medical comorbidity score. The four additional indices of cortisol production also failed to predict postsurgical complications above and beyond medical comorbidity score and perceived stress. Time to Ambulation, Stress, and Cortisol Im pact of negative life events /perceived stress and the additional indices of cortisol production (i.e., AUC-G, AUC-I, mean daily cortisol mean morning cortisol) were examined as predictors of time to post-surgical ambulation with a series of hierarchi cal regression analyses. The analyses were identical to those descri bed above for other outcome variables, with ambulation substituted for the outcome variable. No control variables we re entered into the regression As reported in the analyses for the specific aims, impact of negative life events did not emerge as a predictor of post-surgical time to ambulation (See Table 3-7). None of the four additional indices of cortisol production accounted for significant variance in time to postsurgical ambulation above and beyond impact of negative life events. Per ceived stress did not emerge as a significant predictor of time to post-surgical ambulation and the four additional indices of cortisol production failed to predict time to pos t-surgical ambulation above and beyond perceived stress.

PAGE 72

72 WBC Count, Stress, and Cortisol Im pact of negative life events /perceived stress and the additional indices of cortisol production (i.e., AUC-G, AUC-I, mean daily cortisol mean morning cortisol) were examined as predictors of post-surgical WBC count with a se ries of hierarchical regression analyses. The analyses were identical to thos e described above for other outco me variables, with WBC count substituted as the outcome variable. No control variables were entered. As reported in the analyses for the primary sp ecific aims, impact of negative life events did not emerge as a predictor of post-surgical WBC count (See Table 3-8). However, when impact of negative events was controlled for, AUC-G (but not AUC-I) accounted for significant incremental variance in post-surgical WBC count ( p < .05). Specifically, greater AUC-G was significantly associated with more elevated pos t-surgical WBC count (see Table 3-19 for main effects model). Additionally, mean daily cortisol and mean morning cortisol emerged as significant predictors of post-surgical WBC count when impact of life events was controlled for ( p < .05). Greater mean daily and morning cortisol were associated with more elevated WBC counts (both ps < .05). Tables 3-19 through 3-21 summari ze these results. Identical results emerged when perceived stress was substituted for impact of negative life events (Tables 3-22 through 3-24). Length of Hospitalization, Emotional Support, and Cortisol Em otional support from primary support person /total emotional support and the additional indices of cortisol production (i.e., AUC-G, AUC -I, mean daily cortisol, mean morning cortisol) were examined as predictors of length of hospita lization with a series of hierarchical regression analyses. As previously describe d in the exploratory analyses examining stress, cortisol and outcome variables, 8 individual equations were constructed. Emotional support from primary

PAGE 73

73 support person/total emotional support were subst ituted as the Block 1 predictors and length of hospitalization was substituted fo r the outcome variable. As presented earlier, emotiona l support from primary support person did not emerge as a predictor of length of hospitalization after cont rolling for age and length of surgery (See Table 39). None of the four additional indices of co rtisol production accounted for significant variance in length of hospitalization wh en either emotional support from primary support person or total emotional support were covaried. Total emotional s upport also failed to emerge as a predictor of length of hospitalization after controlli ng for age and length of surgery. Post-surgical Complication Severity, Emotional Support, and Cortisol Em otional support from primary support person /total emotional support and the additional indices of cortisol production (i.e., AUC-G, AUC -I, mean daily cortisol, mean morning cortisol) were examined as predictors of post-surgical comp lication severity with a series of hierarchical regression analyses. The analyses were identical to thos e described in the pr evious section, with post-surgical complication severity substituted as the outcome variable and medical comorbidity score substituted as the control variable. Consistent with previous analyses of the specific aims, emotional support from primary support person did not emerge as a predictor of post-surgical complication severity after controlling for medical comorbidity score (See Ta ble 3-10). None of the four additional indices of cortisol production accounted for significant variance in post-s urgical complication severity when controlling for medical comorbidity score, and either emotional support from primary support person or total emotional support. Tota l emotional support failed to emerge as a predictor of post-surgical comp lications after controlling for medical comorbidity score.

PAGE 74

74 Time to Ambulation, Emotio nal Support, and Cortisol Em otional support from primary support person /total emotional support and the additional indices of cortisol production (i.e., AUC-G, AUC -I, mean daily cortisol, mean morning cortisol) were examined as predictors of post-surgical tim e to ambulation with a series of hierarchical regression analyses. Analyses were identical to those described in previous sections with time to ambulation substituted as the outcome va riable. No variables were covaried. As reported earlier, emotional support from primary support person did not emerge as a predictor of time to post-surgic al ambulation (See Table 3-11). None of the four additional indices of cortisol production accounted for significant varian ce in time to post-surgical ambulation when medical comorbidity score and either emotional support from primary support person or total emotional support we re covaried. Total emotional support also failed to emerge as a predictor of time to post-surgical ambulation. WBC Count, Emotional Support, and Cortisol Em otional support from primary support person /total emotional support and the additional indices of cortisol production (i.e., AUC-G, AUC -I, mean daily cortisol, mean morning cortisol) were examined as predictors of post-surgical W BC count with a series of hierarchical regression analyses. The analyses were iden tical to those described in ea rlier sections, with WBC count substituted as the outcome variable No variables were covaried. As reported in the analyses for the specific aims, emotional support from primary support person emerged as a predictor of post-surgical WBC count ( p < .05). such that greater emotional support was associated with a less elevated WBC count. Moreover, when emotional support from primary support person was controlled for, AUC-G, mean daily cortisol, and mean morning cortisol emerged as significant predictors of WBC count above and beyond the effect of emotional support from primary support person ( ps < .05). Greater AUC-G, mean daily cortisol,

PAGE 75

75 and mean morning cortisol were all associated with greater WBC count. Tables 3-25 through 327 summarize the full models. AUC-I did not emerge as a significant predictor of post-surgical WBC count above and beyond emotional support from primary support person. Due to the lack of a significant relationship be tween emotional support and indi ces of cortisol production ( r = .00, p> .90), test for mediation was not performed. To tal emotional support marginally explained the variance in post-surgical WBC count (p = .09); however, AUC-G, AUC-I, mean daily cortisol, and mean morning cortisol all emerged as significant predictors of post-surgical WBC count when total emotional support was controlled for ( ps < .05). Greater AUC-G, AUC-I, mean daily cortisol, and mean morning cortisol were associated with more elevated WBC count. Tables 3-28 through 3-31 summarize the main effect models. Post-Surgical Pain Given the finding that em erged in preliminary analyses revealing a significant relationship between greater pre-surgical emotional suppor t from primary support person and lower mean post-surgical pain ratings, as we ll as previous studies which have examined post-surgical pain as an index of post-surgical recovery, a series of exploratory analyses examined potential psychoneuroimmunologic relations to post-surgical pain. Participants pain ratings (on a scale from 0 to 10) were abstracted from medical reco rds and all pain ratings during the post-surgical hospitalization were averaged to create a m ean pain rating. Correlations examined potential relations between the a priori continuous contro l variables and mean post-surgical pain. There was a significant relationship between age and mean post-surgical pain ( r = -.30, p <.05), such that younger women reported more elevated mean pos t-surgical pain. Mean pain ratings did not differ significantly between women taking corticosteroid medications pre-surgically ( M = 2.78, SD = .72) and those not taking corticoste roid medications pre-surgically ( M = 2.43, SD = 1.52), t (69) = -.32, p = .75. Mean pain ratings also did not differ significantly between women who

PAGE 76

76 experienced an intra-operative complication ( M = 2.34, SD = 2.24) and those who did not (M = 2.43, SD = 1.33), t (69) = .19, p = .85. There were no significant differences in mean pain between women who collected cortisol samples (M = 2.45, SD = 1.46) and those who did not ( M = 2.38, SD = 1.71), t (70) = -.16, p = .88. Age was controlled for in all analyses examining postsurgical pain. Stress, Emotional Support, Cortis ol, and Post-surgical Pain Neither of th e pre-surgical indices of stre ss (impact of negative life events, perceived stress) was significantly associat ed with post-surgical pain ratings above and beyond the effects of age. However, emotional support from primary support person (but not total emotional support received) emerged as a marginally signif icant predictor of postsurgical pain ratings above and beyond the effects of age (Table 3-32). Diurnal cortisol slope emerged as a marginally significant predictor of post-surgical pain ratings above and beyond the effects of age and life stress ( p = .08), as well as above and beyond the effects of age and interpersonal coping, ( p = .08). In both instances, more positive (abnormal) diurnal cortisol slopes were marginally associated with greater post-surgical pain ratings (see Tables 3-33 and 3-32). In addition, AUC-I emerged as a marginally significant predictor of post-surgical pain ratings above and beyond the effects of age and impact of life events ( p = .07) (Table 3-34), as well as above and beyond the effects of age and emotional support from primary support person ( p = .09) (Table 3-35). In c ontrast, AUC-I was a significant predictor of post-surgical pain ratings above and beyond the effects of age and perceived stress ( p < .05) (Table 3-36). The direction of all three of these relationships indicated that greater AUC-I was associated with lower post-surgical pain ratings. AUC-G was a marginally significant predictor of post-s urgical pain ratings above a nd beyond the effects of age and perceived stress only (p = .09) (T able 3-37); once again, the di rection of this relationship

PAGE 77

77 indicated that greater AUC-G was marginally asso ciated with lower postsurgical pain ratings. Neither mean daily or mean morning cortisol we re associated with post-surgical pain ratings above and beyond age and life stress or above and beyond age and interpersonal coping. Moderation Analyses Moderation analyses we re performed to examine the possibility that perceived emotional support from primary support person moderated th e relationship betwee n impact of negative events and surgical outcome. Perceived emoti onal support from primary support person did not emerge as a significant moderator of the relations hip between impact of negative events and the four indices of surgical rec overy (e.g., length of hospitaliza tion, severity of post-surgical complications, time to post-surgical ambulation, post-surgical WBC count). Tables 3-38 through 3-41 display the full effects models.

PAGE 78

78 Table 3-1. Correlations between a priori biobehavioral control variables (columns) and main predictors and criteria of interest (rows) Age (Years) BMI (kg/m2) Comorbidity (Severityweighted sum) Mean Pain (VAS scale) Length of Surgery (Minutes) FIGO Stage Criteria Length of hospitalization (daysa) .27* N = 73 .08 N = 73 .16 N = 72 -.08 N = 63 .35** N = 61 .14 N = 73 Post-surgical complications (severity-weighted sumb) .11 N = 73 -.08 N = 73 .24* N = 72 .-.01 N = 63 .02 N = 61 .16 N = 73 Post-surgical ambulation (days) .11 N = 43 .19 N = 43 -.16 N = 42 -.21 N = 42 .15 N = 40 -.09 N = 43 Post-surgical WBC (109 cells/liter [L]) .10 N = 74 .16 N = 74 -.15 N = 73 .11 N = 64 .19 N = 62 -.14 N = 74 Predictors Impact Negative Life Events (Weighted sum) -.29* N = 71 -.02 N = 71 .05 N = 70 .19 N = 61 .03 N = 59 .05 N = 71 Emotional Support .17 N = 75 -.36** N = 75 .05 N = 74 -.35** N = 65 .01 N = 63 .12 N = 75 Mean cortisol slope (Unstandardized beta) .15 N = 61 .18 N = 61 -.14 N = 60 .21 N = 55 .03 N = 53 -.05 N =61 = p < .05, ** = p < .01, *** = p < .001. a log transformed. b square root transformed.

PAGE 79

79 Table 3-2. Relationship between in tra-operative complications (col umns) and criteria of interest (rows) Complications During Surgery t (df) p Yes No Length of hospitalization (Daysa) M = .69 SD = .21 M = .59 SD = .16 -1.81 (70) .07 Post-surgical complications (severityweighted sumb) M = .66 SD = 1.01 M = .67 SD = .81 .014 (70) .99 Time to Ambulation (Days) M = 1.78 SD = 1.10 M = 1.15 SD = .80 -1.93 (40) .06 WBC (109 cells/L) M = 10.83 SD = 2.39 M = 10.56 SD = 2.65 -.31(72) .76 a log transformed. b square root transformed.

PAGE 80

80 Table 3-3. Relationship between use of corticosteroid medicati on (columns) and criteria of interest (rows) Use of corticosteroid medication t (df) p No Yes Length of hospitalization (Daysa) M = .60 SD = .17 M = .60 SD = .00 .003 (70) 1.0 Post-surgical complications (severityweighted sumb) M = .67 SD = .85 M = .50 SD = .71 .29 (70) .76 Time to Ambulation (Days) M = 1.30 SD = .88 M= .5 SD = .71 1.26 (40) .22 WBC (109 cells/L) M = 10.63 SD = 2.62 M = 9.97 SD = 3.59 .36 (71) .72 a log transformed. b square root transformed.

PAGE 81

81 Table 3-4. Relationship between cortisol collection (columns) and outcome variables (rows) Cortisol Samples Collected t (df) p No Yes Length of hospitalization (Daysa) M = .69 SD = .20 M = .59 SD = .16 1.91 (70) .06 Post-surgical complications (severityweighted sumb) M = .54 SD = .77 M = .69 SD = .85 -.59 (70) .56 Time to Ambulation (Days) M = 1.50 SD = 1.07 M= 1.23 SD = .84 .78 (41) .44 WBC (10.0x103 mm3) M = 11.13 SD = 2.77 M = 10.44 SD = 2.57 .86 (71) .39 a log transformed. b square root transformed.

PAGE 82

82 Table 3-5. Regression analysis examining length of hospitalization (in log days): Effect of impact of negative events and diurnal cortisol slope Step # Variables R2 R2 F of R2 1 Biobehavioral Control Variables: Length of surgery .21 .46*** .21 6.22*** Age .24* 2 Impact of Negative life events .22 .09 .01 .40 3 Pre-surgical diurnal cortisol slope .22 .05 .00 .13 N = 49, Significance of model, F (4,44) = 3.14, p <.05. p < .10. ** p < .05. *** p < .01.

PAGE 83

83 Table 3-6. Regression analysis examining post-surgical complicat ion severity (in square-root transformed sum of severity ratings): Effect of impact of negativ e events and diurnal cortisol slope Step # Variables R2 R2 F of R2 1 Biobehavioral Control Variables: Medical comorbidity .04 .20 .04 2.32 2 Impact of Negative life events .05 -.12 .01 .76 3 Pre-surgical diurnal cortis ol slope .05 -.01 .00 .01 N = 57, Significance of model, F (3,53) = 1.00, p = .40. p < .10. ** p < .05. *** p < .01.

PAGE 84

84 Table 3-7. Regression analysis ex amining post-surgical ambulation (in days): Effect of impact of negative events and di urnal cortisol slope Step # Variables R2 R2 F of R2 1 Impact of Negative life events .05 -.23 .05 1.75 2 Pre-surgical diurnal cortisol slope .09 .20 .04 1.27 N = 33, Significance of model, F (2,30) = 1.52, p = .24. p < .10. ** p < .05. *** p < .01.

PAGE 85

85 Table 3-8. Regression analysis examining post-surgical WBC count: Effect of impact of negative events and diurnal cortisol slope Step # Variables R2 R2 F of R2 1 Impact of Negative life events .01 -.08 .01 .33 2 Pre-surgical diurnal cortisol slope .01 -.01 .00 .01 N = 57, Significance of model, F (2,54) = .16, p = .85. p < .10. ** p < .05. *** p < .01.

PAGE 86

86 Table 3-9. Regression analysis examining length of hospitalization (in log days): Effect of emotional support (primary support pers on) and diurnal co rtisol slope Step # Variables R2 R2 F of R2 1 Biobehavioral Control Variables: Length of surgery .22 .46*** .22 7.00*** Age .26** 2 Emotional support (Primary support person) .25 .17 .03 1.83 3 Pre-surgical diurnal cortisol slope .25 .06 .00 .23 N = 52, Significance of model, F (4,47) = 4.01, p =.07. p < .10. ** p < .05. *** p < .01.

PAGE 87

87 Table 3-10. Regression analysis examining post-surgical complicat ion severity (in square-root transformed sum of severity ratings): Eff ect of emotional support (primary support person) and diurnal cortisol slope Step # Variables R2 R2 F of R2 1 Biobehavioral Control Variables: Medical comorbidity .05 .22 .05 2.77 2 Emotional Support (Primary Support Person) .05 .05 .00 .14 3 Pre-surgical diurnal cortisol slope .05 -.01 .00 .01 N = 59, Significance of model, F (3,55) = .94, p = .43. p < .10. ** p < .05. *** p < .01.

PAGE 88

88 Table 3-11. Regression analysis examining post-surgical ambul ation (in days): Effect of emotional support (primary support pers on) and diurnal cortisol slope Step # Variables R2 R2 F of R2 1 Emotional Support (Primary support person) .02 .15 .02 .80 2 Pre-surgical diurnal cortisol slope .08 .23 .06 1.95 N = 36, Significance of model, F (2,33) = 1.38, p = .27. p < .10. ** p < .05. *** p < .01.

PAGE 89

89 Table 3-12. Regression analysis examining post-surgical WBC count: Effect of emotional support (primary support person) and diurnal cortisol slope Step # Variables R2 R2 F of R2 1 Emotional Support (Primary support person) .07 -.27** .07 4.37** 2 Pre-surgical diurnal cortisol slope .07 .00 .00 .00 N = 60, Significance of model, F (2,57) = 2.16, p = .13. p < .10. ** p < .05. *** p < .01.

PAGE 90

90 Table 3-13. Logistic regression analysis examining post-surgical WBC count (normal versus leukocytosis): Effect of emotional suppor t (primary support person) and diurnal cortisol slope Step # Variables Wald Statistic 95%CI P Value 1 Emotional Support (Primary Support Person) 2.19 .77 .1.04 .14 2 Pre-surgical diurnal cortisol slope .07 .00 -.00 .79 N = 59, Significance of model, F (2,56) = .88, p = .42. p < .10. ** p < .05. *** p < .01.

PAGE 91

91 Table 3-14. Regression analysis examining post-surgical neutrophil percentage: Effect of emotional support (primary support Pers on) and diurnal cortisol slope Step # Variables R2 R2 F of R2 1 Emotional Support (Primary Support Person) .00 -.05 .00 .04 2 Pre-surgical diurnal cortisol slope .01 .10 .01 .16 N = 20, Significance of model, F (2,17) = .10, p = .90. p < .10. ** p < .05. *** p < .01.

PAGE 92

92 Table 3-15. Regression analysis examining post-surgical lymphocyte percentage: Effect of emotional support (primary support pers on) and diurnal cortisol slope Step # Variables R2 R2 F of R2 1 Emotional Support (Primary Support Person) .00 -.01 .00 .00 2 Pre-surgical diurnal cortisol slope .01 -.09 .01 .13 N = 20, Significance of model, F (2,17) = .06, p = .94. p < .10. ** p < .05. *** p < .01.

PAGE 93

93 Table 3-16. Regression analysis examining post-surgical WBC count in first four days postsurgery: Effect of emotional support (primary support person) and diurnal cortisol slope Step # Variables R2 R2 F of R2 1 Emotional Support (Primary Support Person) .05 -.23^ .05 3.17^ 2 Pre-surgical diurnal cortisol slope .05 -.01 .00 .01 N = 59, Significance of model, F (2,56) = 1.57, p = .22. p < .10. ** p < .05. *** p < .01.

PAGE 94

94 Table 3-17. Correlations between a priori bi obehavioral control variables (columns) and exploratory predictors (rows) Age (Years) BMI (kg/m2) Comorbidity (Severity-weighted sum) Mean Pain (VAS scale) Length of Surgery (Minutes) FIGO Stage Predictors Perceived Stress -.21* N = 71 .31*** N = 71 .06 N = 70 .22* N = 62 .20 N =60 -.12 N = 70 Total Emotional Support -.02 N = 65 -.29** N = 65 .03 N = 65 -.09 N = 57 -.31** N = 55 -.07 N = 64 AUC-G (nmol/l) .15 N = 59 -.05 N = 59 .01 N = 59 -.23 N = 53 -.20 N = 59 -.27** N = 59 AUC-I (nmol/l) .23* N = 59 -.04 N = 59 -.07 N = 59 -.26* N = 53 -.29** N = 51 -.26** N = 59 Mean daily cortisol (nmol/l) .18 N = 62 -.06 N = 62 .07 N = 61 -.18 N = 55 -.12 N = 53 -.20 N = 62 Mean morning cortisol (nmol/l) .04 N = 61 -.01 N = 61 -.04 N = 61 -.08 N = 52 -.04 N = 52 -.04 N = 59 p < .10. ** p < .05. *** p < .01. a log transformed. b square root transformed.

PAGE 95

95Table 3-18. Correlations among alternative measures of stress, emotional support, cortisol and criteria 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 Length of Hospitalization (Days) -.49*** N =72 .51*** N =71 .39** N =42 .13 N =72 .07 N =60 .11 N =58 .11 N =58 .11 N =61 .10 N =58 -.01 N =69 .01 N =70 .15 N =73 -.09 N =63 2 Post-surgical complications severity .49*** N =72 -.99*** N =72 .34** N =41 .01 N =72 -.04 N = 60 .00 N =58 .00 N =58 .07 N =61 .05 N =58 -.10 N =70 .15 N =69 .01 N =73 -.04 N =63 3 N umber of pos t surgical complications .51*** N =71 .99*** N =72 -.36** N =40 .01 N =71 -.04 N = 59 .00 N =58 .00 N =58 .07 N =61 .06 N =58 -.10 N =69 .18 N =68 .01 N =72 -.05 N =63 4 Post-surgical ambulation (Days) .39** N =42 .34** N =41 .36** N =40 --.01 N =42 .24 N = 36 .28 N =35 .28 N =35 .23 N =36 .18 N =35 -.18 N =39 .18 N =41 .15 N =43 .20 N =38 5 Post-surgical WBC count .13 N =72 .01 N =72 .01 N =71 -.01 N =42 --.01 N = 60 .31** N =58 .20 N =58 .33** N =61 .31** N =58 -.13 N =70 .05 N =70 -.15 N =74 -.18 N =64 6 Diurnal cortisol slope .07 N = 60 -.04 N = 60 -.04 N =59 .24 N =36 -.01 N = 60 --.13 N = 59 .13 N =59 -.16 N =61 -.37*** N =58 -.01 N =58 .03 N =60 .01 N =61 -.08 N =54 7 AUC-G .11 N =58 .00 N =58 .00 N =58 .28 N =35 .31** N =58 -.13 N = 39 -.84*** N =59 .96*** N =59 .75*** N =47 -.03 N =56 .11 N =58 .00 N =59 .29** N =53 8 AUC-I .11 N =58 .00 N =58 .00 N =58 .28 N =35 .20 N =58 .13 N = 59 .84*** N =58 -.70*** N =59 .28** N =57 -.01 N =56 .02 N =58 .05 N =59 .29** N =53 9 Mean daily cortisol .11 N =61 .07 N =61 .07 N =60 .23 N =36 .33** N =61 -.16 N = 61 .96*** N =59 .70*** N =59 -.85*** N =59 -.10 N =59 .22* N =61 -.02 N =62 .20 N =54 10 Mean morning cortisol .10 N =58 .05 N =58 .06 N =58 .18 N =35 .31** N =58 -.37*** N = 58 .75*** N =57 .28** N =57 .85*** N =59 --.03 N =56 .28** N =58 -.08 N =59 .22 N =52 11 Impact of negative events -.01 N =69 -.10 N =70 -.10 N =69 -.18 N =39 -.13 N =70 -.01 N =58 -.03 N =56 -.01 N =56 -.10 N =59 -.03 N =56 -.30** N =68 -.14 N =71 .07 N =61 12 Perceived stress .01 N =70 .15 N =69 .18 N =68 .18 N =41 .05 N =70 .03 N = 60 .00 N =59 .05 N =59 .22* N =61 .28** N =58 .30** N =68 --.45*** N =71 -.08 N =61 13 Emotional Support (primary support person) .15 N =73 .01 N =73 .01 N =72 .15 N =43 -.15 N =74 .01 N = 61 .00 N =59 .05 N =59 -.02 N =62 -.08 N =59 -.14 N =71 -.45*** N =71 -.23* N =65 14 Emotional support (total) -.09 N =63 -.04 N =63 -.05 N =63 .20 N =38 -.18 N =64 -.08 N = 54 .29** N =53 .29** N =53 .20 N =54 .22 N =52 .07 N =61 -.08 N =61 .23* N =65 -* p < .10. ** p < .05. *** p < .01.

PAGE 96

96 Table 3-19. Regression analysis examining postsurgical WBC count: Effect of impact of negative events and AUC-G Step # Variables R2 R2 F of R2 1 Impact of negative life events .01 -.07 .01 .26 2 AUC-G .10 .31** .10 5.48** N = 55, Significance of model, F (2,53) = 2.88, p = .07. p < .10. ** p < .05. *** p < .01.

PAGE 97

97 Table 3-20. Regression analysis examining post-surgical WBC count : Effect of impact of negative events and mean daily cortisol Step # Variables R2 R2 F of R2 1 Impact of negative life events .01 .08 .01 .39 2 Mean daily cortisol .12 .33*** .11 6.71*** N = 58, Significance of model, F (2,55) = 3.56, p < .05. p < .10 ** p < .05. *** p < .01.

PAGE 98

98 Table 3-21. Regression analysis examining post-surgical WBC count: Effect of impact of negative events and mean morning cortisol Step # Variables R2 R2 F of R2 1 Impact of negative life events .00 .01 .00 .00 2 Mean morning cortisol .11 .33** .11 6.32** N = 55, Significance of model, F (2,52) = 3.16, p =.05. p < .10. ** p < .05. *** p < .01.

PAGE 99

99 Table 3-22. Regression analysis examining post-surgical WBC count: Effect of perceived stress and AUC-G Step # Variables R2 R2 F of R2 1 Perceived Stress .01 .10 .01 .59 2 AUC-G .10 .31** .09 5.60** N = 57, Significance of model, F (2,54) = 3.12, p = .05. p < .10. ** p < .05. *** p < .01.

PAGE 100

100 Table 3-23. Regression analysis examining post-surgical WBC count: Effect of perceived stress and mean daily cortisol Step # Variables R2 R2 F of R2 1 Perceived stress .02 .13 .02 .91 2 Mean daily cortisol .12 .33*** .10 6.52*** N = 60, Significance of model, F ( 2,59) = 3.76, p < .05. p < .10. ** p < .05. *** p < .01.

PAGE 101

101 Table 3-24. Regression analysis examining post-surgical WBC count: Effect of perceived stress and mean morning cortisol Step # Variables R2 R2 F of R2 1 Perceived stress .01 .12 .01 .80 2 Mean morning cortisol .11 .32** .09 5.56** N = 57, Significance of model, F (2,54) = 3.21, p < .05. p < .10. ** p < .05. *** p < .01.

PAGE 102

102 Table 3-25. Regression analysis examining post-surgical WBC count: Effect of emotional support (primary support person) and AUC-G Step # Variables R2 R2 F of R2 1 Emotional support (Primary Support Person) .07 -.26** .07 4.10** 2 AUC-G .16 .31** .10 6.28** N = 58, Significance of model, F (2,55) = 5.37, p < .01. p < .10. ** p < .05. *** p < .01.

PAGE 103

103 Table 3-26. Regression analysis examining post-surgical WBC count: Effect of emotional support (primary support person) and mean daily cortisol Step # Variables R2 R2 F of R2 1 Emotional support (Primary Support Person) .07 -.26** .07 4.31** 2 Mean daily cortisol .17 .32*** .10 7.31*** N = 61, Significance of model, F (2,58) = 6.04, p < .01. p < .10. ** p < .05. *** p < .01.

PAGE 104

104 Table 3-27. Regression analysis examining post-surgical WBC count: Effect of emotional support (primary support person) and mean morning cortisol Step # Variables R2 R2 F of R2 1 Emotional support (Primary Support Person) .05 -.23* .05 3.04* 2 Mean morning cortisol .14 .29** .08 5.27** N = 58, Significance of model, F (2,55) = 4.27, p < .05. p < .10. ** p < .05. *** p < .01.

PAGE 105

105 Table 3-28. Regression analysis examining post-surgical WBC count: Effect of total emotional support and AUC-G Step # Variables R2 R2 F of R2 1 Total emotional support .06 -.24* .06 3.01* 2 AUC-G .25 .46*** .19 12.23*** N = 52, Significance of model, F (2,49) = 7.96, p < .01. p < .10. ** p < .05. *** p < .01.

PAGE 106

106 Table 3-29. Regression analysis examining post-surgical WBC count: Effect of total emotional support and AUC-I Step # Variables R2 R2 F of R2 1 Total emotional support .06 -.24* .06 3.01* 2 AUC-I .17 .35** .11 6.45** N = 52, Significance of model, F ( 2,49) = 4.89, p < .05. p < .10 ** p < .05, *** p < .01.

PAGE 107

107 Table 3-30. Regression analysis examining post-surgical WBC count: Effect of total emotional support and mean daily cortisol Step # Variables R2 R2 F of R2 1 Total emotional support .06 -.24* .06 3.12* 2 Mean daily cortisol .23 .43*** .17 11.17*** N = 53, Significance of model, F (2,50) = 7.46, p < .01. p < .10. ** p < .05. *** p < .01.

PAGE 108

108 Table 3-31. Regression analysis examining post-surgical WBC count: Effect of total emotional support and mean morning cortisol Step # Variables R2 R2 F of R2 1 Total emotional support .05 -.23 .05 2.74 2 Mean morning cortisol .18 .37*** .13 7.70*** N = 51, Significance of model, F (2,48) = 4.41, p < .01. p < .10. ** p < .05. *** p < .01.

PAGE 109

109 Table 3-32. Regression analysis examining effect of emotional support (primary support person) and diurnal cortisol slope on post-surgical pain ratings Step # Variables R2 R2 F of R2 1 Age .03 -.17 .03 1.48 2 Emotional Support (Primary Support Person) .09 -.24* .06 3.32* 3 Diurnal cortisol slope .14 .23* .05 3.16* N = 55, Significance of model, F ( 3,51) = 2.75, p = .05. p < .10. ** p < .05. *** p < .01.

PAGE 110

110 Table 3-33. Regression analysis examining effect of impact of negative events and diurnal cortisol slope on post-su rgical pain ratings Step # Variables R2 R2 F of R2 1 Age .03 -.16 .03 1.31 2 Impact of negative ev ents .05 .15 .02 1.05 3 Diurnal cortisol slope .11 .25* .06 3.26* N = 52, Significance of model, F ( 3,48) = 1.91, p = .14. p < .10. ** p < .05. *** p < .01.

PAGE 111

111 Table 3-34. Regression analysis examining effect of impact of negative events and AUC-I on post-surgical pain ratings Step # Variables R2 R2 F of R2 1 Age .02 -.15 .02 1.17 2 Impact of negative ev ents .05 .16 .02 1.07 3 AUC-I .11 -.26* .07 3.40* N = 50, Significance of model, F ( 3,46) = 1.92, p = .14. p < .10. ** p < .05. *** p < .01.

PAGE 112

112 Table 3-35. Regression analysis examining effect of emotional support (primary support person) and AUC-I on post-surgical pain ratings Step # Variables R2 R2 F of R2 1 Age .03 -.16 .03 1.33 2 Emotional Support (Primary Support Person) .08 -.24* .06 3.05* 3 AUC-I .13 -.24* .05 2.91* N = 53, Significance of model, F ( 3,49) = 2.50, p = .07. p < .10. ** p < .05. *** p < .01.

PAGE 113

113 Table 3-36. Regression analysis examining effect of per ceived stress and AUC-I on postsurgical pain ratings Step # Variables R2 R2 F of R2 1 Age .01 -.12 .01 .70 2 Perceived Stress .04 .15 .02 1.06 3 AUC-I .12 -.31** .08 4.57** N = 52, Significance of model, F ( 3,48) = 2.15, p = .11. p < .10. ** p < .05. *** p < .01.

PAGE 114

114 Table 3-37. Regression analysis examining effect of percei ved stress and AUC-G on postsurgical pain ratings Step # Variables R2 R2 F of R2 1 Age .01 -.12 .01 .70 2 Perceived stress .04 .15 .03 1.07 3 AUC-G .09 -.25* .05 3.03* N = 52, Significance of model, F ( 3,48) = 1.62, p = .2. p < .10. ** p < .05. *** p < .01.

PAGE 115

115 Table 3-38. Regression analysis examining the relationship be tween post-surgical length of hospitalization (in log days) and impact of negative events: Effect of emotional support (primary support pe rson) as a moderator Step # Variables R2 R2 F of R2 1 Age Length of Surgery .23 .33** .42*** .23 7.90*** 2 Impact of Negative Events .23 1.35 .00 .24 3 Perceived Emotional Support (Primary Support Person) .27 .40 .04 2.83* 4 Impact of Negative Events x Perceived Emotional Support Interaction .30 -1.27 03 1.88 N = 53, Significance of model, F ( 3,49) = 2.50, p = .07. p < .10. ** p < .05. *** p < .01.

PAGE 116

116 Table 3-39. Regression analysis examining the relationship between severity of post-surgical complications (in square-root transformed sum of severity ratings) and impact of negative events: Effect of emotional support (primary support person) as a moderator Step # Variables R2 R2 F of R2 1 Medical Comorbidity Score .05 .22* .05 3.61* 2 Impact of Negative Events .06 .61 .01 .82 3 Perceived Emotional Support (Primary Support Person) .06 .15 .00 .05 4 Impact of Negative Events x Perceived Emotional Support Interaction .07 -.71 .01 .59 N = 53, Significance of model, F ( 3,49) = 2.50, p = .07. p < .10. ** p < .05. *** p < .01.

PAGE 117

117 Table 3-40. Regression analysis examining the relationship be tween time to post-surgical ambulation (in days) and impact of negative events: Effect of emotional support (primary support person) as a moderator Step # Variables R2 R2 F of R2 1 Impact of Negative Events .03 -.61 .03 1.21 2 Perceived Emotional Support (Primary Support Person) .03 -.09 .00 .04 3 Impact of Negative Events x Perceived Emotional Support Interaction .03 .41 .00 .05 N = 53, Significance of model, F ( 3,49) = 2.50, p = .07. p < .10. ** p < .05. *** p < .01.

PAGE 118

118 Table 3-41. Regression analysis examining the relationship be tween post-surgical WBC count and impact of negative events: Effect of emotional support (primary support person) as a moderator Step # Variables R2 R2 F of R2 1 Impact of Negative Events .02 -.88 .02 1.17 2 Perceived Emotional Support (Primary Support Person) .05 -.32 .03 2.65 3 Impact of Negative Events x Perceived Emotional Support Interaction .06 .73 .01 .64 N = 53, Significance of model, F ( 3,49) = 2.50, p = .07. p < .10. **p < .05. *** p < .01.

PAGE 119

119 Figure 3-1. Examples of graphi cal representations of diurnal cortisol slopes from study participants CORTISOL (nmol/l) 1.20 1.00 0.80 0.60 0.40 0.20 0.00 CORTISOL (nmol/l) 1.20 1.00 0.80 0.60 0.40 0.20 0.00 CORTISOL (nmol/l) 1.20 1.00 0.80 0.60 0.40 0.20 0.00 TIME (24-hr clock)21.00 18.00 15.00 12.00 9.00 CORTISOL (nmol/l)1.20 1.00 0.80 0.60 0.40 0.20 0.00 TIME (24-hr clock)21.00 18.00 15.00 12.00 9.00 TIME (24-hr clock)21.00 18.00 15.00 12.00 9.00 ACS033 ACS031 ACS028 ACS021 ACS019 ACS015 ACS011 ACS009 ACS008 ACS007 ACS004 ACS003 3 2 1 DAY

PAGE 120

120 CHAPTER 4 DISCUSSION Discussion of Results The current study is the first to examine psychoeneuroimm unologic predictors of postsurgical recovery among women with endometri al cancer, a group of women who may be at greater risk for a more complicated surgical recovery. The primary hypothesis of this study was that women who reported greater pre-surgi cal psychosocial stress would have a more complicated post-surgical re covery (as evidenced by longe r length of post-surgical hospitalization, greater severity of post-surgical complications longer time to post-surgical ambulation, and less elevated WBC count), and th at the relationship between pre-surgical psychosocial stress and a more complicated post-su rgical recovery would be mediated by more abnormal diurnal cortisol slopes. Similarly, it was expected that women who reported greater pre-surgical adaptive interpersonal coping (as evidenced by great er perceived emotional support from primary support person) would have a less complicated post-surgic al recovery, and the relationship between pre-surgical adaptive inte rpersonal coping and a less complicated postsurgical recovery would be mediated by a less abnormal diurnal cortisol slope. Finally, it was hypothesized that among women with high pre-surg ical psychosocial stress, those who report less use of pre-surgical adaptive interpersonal coping strategies will have a more complicated post-surgical recovery. Although the findings did not fully suppor t the hypotheses, the present study yielded a number of important findings that extend the literat ure examining predictors of surgical recovery. Psychosocial Stress and Surgical Recovery Previous data suggest that el evated pre-surgical psychosocia l distress is associated with less favorable surgical recove ry (e.g., Katz et al., 2005, Kain et al., 2000, Cohen et al., 2005,

PAGE 121

121 Broadbent et al., 2003). The pres ent study explored th e potential relationshi p between two types of psychosocial distress, life ev ent stress and perceive d stress, and four indices of surgical recovery: length of post-surgical hospitalization, severity of post-surgical complications, time to post-surgical ambulation, and post-surgical WBC count. In the pres ent study, no significant relations emerged between measures of psychosocial stress and indi ces of post-surgical recovery. This may reflect a true lack of relationship between psychosocial stress and surgical recovery; however, although the explanation for this lack of relations among pre-surgical psychosocial stress and surgical outcome remains unclear, seve ral plausible explanations exist. First, it is possible that the measures of stress used in th e present study did not capture the potential stress associated specifically with cancer diagnosis and surgery. Previous research examining the effects of stress on surgical outcome has found relations between surgical outcome and psychosocial distress when measuri ng constructs more related to pr e-surgical distress or anxiety (e.g., Kain et al., 2000; Cohen et al ., 2005; Broadbent et al., 2003), as opposed to life events or perceived stress. It is possible th at acute distress related to surgery may be more strongly related to surgical outcome than measures of life events or perceived stress in the months preceding surgery. Indeed, it is possible th at greater life event stress duri ng the preceding months might result in a blunted response to acute stress such that individuals might experience less pronounced stress in response to th e acute pre-surgical period. Ho wever, it should be noted that the present study offers no empirical data to s upport this hypothesis. Second, the absence of a relationship between psychosocial stress and surgical outcome may also be attributable to the possibility that the present population is unique in terms of its experience and/or reporting of negative life events. It might be that women in the present study tended to perceive the negative life events that they experienced as minimally stressful and consequently, examining these

PAGE 122

122 variables in relation to surgical outcome results in no significant relationship. Stawski, Sliwinski, Almeida, and Smyth (2008) found that, compared to younger individuals, older individuals reported decreased exposure to daily stressors, but older individuals (m en and women) did not report different emotional reactivity to daily stre ssors. Although research indicates that women report greater mean stressful life events when compared to men (Bieliauskas, Counte, & Glandon, 1995; Turner & Avison, 2003), there is a pa ucity of research examining the perceived impact of stressful life events among populations similar to the women in the present study (e.g., older, rural women). Similarly, gi ven that the women in the pres ent study experi enced high rates of comorbid medical conditions, it is possible that their experience with previous medical conditions and treatment resulted in less perc eived distress related to their diagnosis of endometrial cancer and subsequent surgical treatment. For exampl e, Wells et al. (2006) found that women with both endometr ial cancer and diabetes evidenced fewer negative coping behaviors, more positive affect, and less anxi ety than their counterparts diagnosed with endometrial cancer who were not diabetic, suggesting that wome n with diabetes who develop endometrial cancer may be buffered from the stress related to the dia gnosis of endometrial cancer through previously devel oped adaptive health-related c oping strategies. Furthermore, Yang, Thornton, Shapiro, and Andersen (2008) f ound that women experien cing a recurrence of breast cancer demonstrated notable resilience, with lower anxiety re ported among women with recurrent cancer, compared to women coping with an initial cancer diagnosis. Thus, the possibility that the level of ps ychosocial stress experienced by the sample in the present study was too minimal to affect surgical response must be considered. Psychosocial Stress and Cortisol The literature is rep lete with evidence for the relationship between psychosocial stress and dysregulated cortisol production. The relationship between psychos ocial stress and dysregulated

PAGE 123

123 cortisol was thought to be of particular sa lience to the potential relationship between psychosocial functioning and surgical outcome, given that neuroendocrine functioning has been hypothesized to modulate the effects of psychosoc ial functioning on surgical recovery ( Ben Eliyahu, 2004). Consequently, it was expected that greater pre-surgical stress would be associated with more dysregulated cortisol in dices. Interestingly, am ong women in the present study, psychosocial stress was not significantly related to any index of cortisol production. Several potential explanations may account for the absence of a relationship between psychosocial stress and cortisol in the present study. First, it is possible that the lack of relationship between stress and cor tisol in this study is due to the way in which stress was measured. Although stress was examined both thr ough the impact of negative life events and perceived stress, it is possible that neither measure of stress accurately captured participants psychosocial stress at the time of surgery. The LES examined the occurrence and impact negative life events during the previous six months and the PSS examined perceived stress during the past month. Neither measure specifica lly assessed the situati on-specific psychosocial stress that participants may e xperience during the acute pre-surg ical period. Consequently, these measures of stress are likely mo re reflective of chronic or cumu lative stress as opposed to the situation-specific stress of cancer diagnosis and surgery. As such, they may not be associated with the acute cortisol response to stress that ma y be evident in the pre-surgical period. Although diurnal slope is an index of ch ronicity of HPA response, it re mains unclear whether cortisol response to acute stress may alte r the diurnal cortisol production in different ways than the chronic stress response. For example, research has found evidence that among individuals with chronic illness, dysregulated inflammatory response may result in a blunted cortisol response to acute stressors (Nijm, Kristenson, Olsson, & Jo nasson, 2007). Second, as discussed above, it is

PAGE 124

124 possible that the level of psychosocial stress perceived by women in the present study was too minimal to significantly affect cortisol production. However, the perceived stress reported by women in the present study ( M = 22.14, SD = 8.81) did not significantly differ from the perceived stress reported by a sample of 17 metastatic breast cancer patients ( M = 21.5, SD = 2.8) in a study by Abercrombie et al. (2004) (t = .30, p > .05). However, it should also be noted that perceived stress was unrelated to diurnal cortisol slope in the Abercrombie et al. (2004) study. Interpersonal Coping and Surgical Recovery Previous research has identified coping as a predictor of post-surgical outcom e (e.g., Cohen et al., 2005; Contrada et al., 2004; Kr ohne et al., 2005; Kopp et al., 2003; Manyande et al., 1995). Interpersonal coping has emerged as a particularly important predictor of health outcomes and post-surgical recovery (Krohne et al., 2005; Kopp et al., 2003; Manyande et al., 1995). Thus, the present study explored potential relations among presurgical perceived emotional support from primary support person and indices of post-surgical recovery. In the present study, no relationships em erged between pre-surgical pe rceived emotional support from primary support person and length of hospitalization, severity of post-surgical complications, or time to post-surgical ambulation. Despite the absence of relationship between perceived emotional support from primary support person and the above-mentioned indices of post-surgical recovery, perhaps the most robust relationship that emerged in the present study was between pre-surgical interpersonal c oping and post-surgical WBC coun t. Women who reported greater perceived emotional support pre-surgically had less elevated WBC counts post-surgically. The direction of this relationship was not consiste nt with the initial hypot hesis that women with greater perceived emotional support would e xperience more elevated WBC count postsurgically. The initial hypothesis was based on th e typical systemic immune response to surgery

PAGE 125

125 marked by leukocytosis, which is considered to be an adaptive physiologic response that favors survival (Salo, 1992). Consequently, it was e xpected that women with more adaptive interpersonal coping (e.g., great er perceived emotional support from primary support person) would have more elevated WBC counts post-surgically. Given th at WBC count is clinically viewed categorically with either low WBC count (leukopenia) or high WBC count (leukocytosis) considered pathologic, several post-hoc exploratory analyses were performed in an effort to elucidate the nature of the relationship between emotional support and WBC count. When WBC count was examined categorically (normal W BC count versus leukocytosis), no significant relationships emerged with emotional support from primary support person. Additionally, because the adaptive increase in WBC count may be more likely to emerge in the acute postsurgical time period when the immune system experiences greater activation, exploratory analysis examined relations between emotiona l support and mean WBC count during the first four days following surgery (given that the average length of stay was 4.58 days). A marginally significant relationship emerged between emo tional support from primary support person and post-surgical WBC count during th e first four days pos t-surgery; however, gr eater pre-surgical emotional support from primary support person con tinued to predict less elevated WBC count. Given that the typical systemic immune response to surgery also involves an increase in the percentage of neurophils (neutrophilia) and a decrease in the percentage of lymphocytes (lymphocytopenia) (Salo, 1992), e xploratory post-hoc analyses examined the relationship between emotional support from primary support person and neutrophil and lymphocyte percentages; however, no significant relations hip emerged between emotional support from primary support person and these indicators of syst emic immune status. It should be noted that only a small portion of the sample had complete blood cell differentials with neutrophil and

PAGE 126

126 lymphocyte percentages documented in their medical records, resulting in a very small sample size that likely lacked sufficient power to de tect a significant relationship. Finally, it is noteworthy that even in the instances in whic h emotional support from primary support person was significantly related to W BC count, it explained a small amount of the overall variance in WBC count. Although post-hoc exploratory an alyses did not clarify the relationship between greater emotional support and less elevated WBC count, a review of the physiologic role of WBC count in surgical recovery yields a pl ausible rationale for the unexpected direction of this relationship. Although the total WBC count c onstitutes a gross measure of the systemic immune and inflammatory changes that may result from surgi cal tissue injury, there are several factors that increase the complexity in its interpretation in relation to surgical outcome. For example, the relationship between leukocytos is and surgical outcome tends to be complex and nonlinear (Coller, 2005; Gurm et al., 2003). To fully addres s the complexities associ ated with interpreting the role of WBC count in surgical outcome, it is necessary to review the physiologic response to tissue injury. The physiologic response to tissue injury is prop osed to consist of three stages: hypodynamic ebb phase (shock), hyperdynamic flow phase, and recuperation phase (Smith & Giannoudis, 1998). The hypodymamic e bb phase is characterized by an attempt to limit the blood loss and to maintain perfusion to the v ital organs. The hyperdynamic flow phase is characterized by increased blood flow to eliminate waste products and to f acilitate the migration of nutrients to the site of injury for repair. Th e recuperation phase, which can last for months, is characterized by attempts to retu rn the human body to its pre-injury level. WBC count plays a critical role in the inflammatory response to tissue injury. This inflamma tory response involves a series of adaptive steps, including increase in blood flow to site of trauma, increased capillary

PAGE 127

127 permeability, and leukocyte migration out of circ ulatory system into surrounding tissue (Clancy 1998). Although this response may be initiated with in minutes of tissue injury, it can persist for weeks (Clancy, 1998). During the inflammatory re sponse to tissue injury, leukocytes serve to remove debris and restore normal tissue structur e and function, as well as to fight infection (Clancy, 1998). The activation of the inflamma tory response is followed by the release of cytokines and the migration of leukocytes into inflamed tissues (Giannoudis, Dinopoulos, Chalidis, & Hall, 2006). Leukocytes are comprise d of a number of ce ll subsets, including neutrophils, lymphocytes, mast cells, eosinophi ls, basophils, and monocytes. Neutrophils typically arrive first at the site of injury and engage in phagocytosis of bacteria in an effort to stave off infection (Clancy, 1998). The mobilization of leukocytes to the s ite of the injury to engage in tissue repair is facili tated by a state of hyperglycemia th at results from a concentration gradient in which there is a dow nhill flow of glucose across the extracellular matrix to the areas of damage where leukocytes amass (Kohl & Deut schman, 2006). This response of leukocytes to the site of injury is also modulated by the production a nd secretion of catecholamines, antidiuretic hormone, cortisol, insulin,gluca gon, growth hormone and cytokines (Kohl & Deutschman, 2006). The adaptive response of leukocyt e migration from the systemic circulatory system into the tissues surrounding the injury raises c oncerns about the inte rpretability of WBC count derived from peripheral blood draw. Sp ecifically, WBC count based on peripheral blood draw may not reflect an increase of leukocytes at the site of tissue in jury and thus, may not reflect the hypothesized elevations in WBC count. Moreover, it coul d be argued that the lack of elevation observed in WBC count from peripheral blood draw may actually reflect the migration of blood cells from the systemic circulatory syst em and into the site of injured tissue. The potential difference in WB C count systemically versus at the si te of tissue injury complicates the

PAGE 128

128 interpretation of research examining WBC count in the context of surgery. It also serves as a potential explanation for the association betw een greater pre-surgical perceived emotional support and less elevated WBC count. Because the methodology of the present study did not allow for an examination of WBC count at the site of tissue inju ry (where the elevation in WBC count would be expected to occur), the initia l hypothesis that interp ersonal coping would be associated with more elevated WBC count di d not account for potential different WBC count response as measured by peripheral blood draw, wh ich would be a measure of WBC count in the systemic circulatory system. Moreover, although elevated WBC count is expe cted as part of the adaptive physiologic response to surgical tissue trauma, recent rese arch examining WBC count and cardiac surgical outcome also suggests that elevated post-surgical WBC count may confer deleterious effects to indices of surgical recovery. For example, Lamm et al. (2006) found that a more pronounced increase in post-surgical WBC count was an independent predictor of the development of postoperative atrial fibrillat ion among individuals undergoing elective cardiac surgery. Additionally, Abdelhadi, Gurm Van Wagoner, and Chung (2004) reported that elevated WBC count independently predicted post -surgical atrial fibrillation in individuals undergoing coronary bypass or cardiac valve surgery. Leukocytosis has al so been established as a significant predictor of post-surgical renal insufficiency among indi viduals undergoing coronary artery bypass graft surgery (Brown et al., 2007). The findings on th e relationship between el evated WBC count and complicated surgical recovery among individu als undergoing cardiac surgery suggest that perhaps elevated WBC count and failure to achieve leukocytosis may each be associated with individual indices of poor surgi cal recovery unique to that dise ase state and surgery. Given the deleterious effects of leukocytos is observed in the cardiac surg ery literature, the relationship

PAGE 129

129 between greater pre-surgical perceived emoti onal support and less elevated WBC count appears to be reflective of a potentially more adaptiv e WBC count response than was specified in the initial hypothesis. In addition, it is possible that the predictive nature of WBC count on surgical outcome may be more related to a change in WBC count fr om preto post-surgery. For example, Amar, Goenka, Zhang, Park, and Thaler (2006) found that the increase in WBC c ount from pre-surgical to post-surgical day one, along with age, significan tly predicted post-surgi cal atrial fibrillation among individuals who underwent general thoracic surgery. Specifically, a two-fold increase in WBC from pre-surgery to post-surgical day one resulted in a 3.3-fold increase in the odds of developing atrial fibrillation (A mar et al., 2006). In the present study, baseline pre-surgical WBC count was unknown for the majority of participants. Consequently, it is possible that mean WBC counts observed post-surgically, although not clini cally elevated, did demonstrate an elevation relative to WBC count pre-surgic ally. This would indicate a w ithin-person increase in WBC count that is not detectable when solely examining post-surgical WBC count. As such, the finding that greater perceived em otional support is associated with less elevated WBC count does not take into account a change from pre-su rgical to post-surgical WBC count, which would be more reflective of the specific immune response to surgic al tissue trauma. Based on evidence of the exis tence of a J-shaped relati onship between WBC count and mortality following percutaneous coronary interv ention (Gurm et al., 2003) it has recently been suggested that overall WBC count may be limited in its predictive ability (Gibson et al., 2007). Consequently, a subset of research has examin ed the predictive ability of leukocyte subsets on surgical outcome. Among indivi duals undergoing coronary arte ry bypass surgery (CABG), an elevated neutrophil count and a depressed lym phocyte count (the normal response to surgery)

PAGE 130

130 were both associated with a worse surgical outco me (Gibson et al., 2007). Specifically, elevated neutrophil count was associated with an increased risk of death, and lower preoperative lymphocyte count was a predictor of mortality (G ibson et al., 2007). Gibs on et al. (2007) also found that an elevated neutropoh il/lymphocyte (N/L) ratio was asso ciated with a poorer survival after CABG, and that this finding was independent of other recognized risk factors. Moreover, elevated N/L ratio independently predicted cardiovascular mortal ity in the subgroup of patients with a normal WBC count undergoing CABG (Gibson et al., 2007). Once again, these findings about the potentially damaging effects of elevat ed neutrophil counts a nd depressed lymphocyte counts following surgery are somewhat inconsistent with that which would be expected, based upon the fact that neutrophilia is an expected and adaptive response to tissue injury. Additionally, it is noteworthy th at the N/L ratio was predicti ve of mortality, even among individuals with normal WBC c ounts. These findings suggest that overall WBC count may have limited predictive ability. Thus, although the relationship between in terpersonal coping and WBC count was in the opposite direction than th at expected in the in itial hypothesis, it is possible that the overall WBC count may not be clinically meani ngful when considered in the absence of N/L ratio. In the pr esent study, analyses examining in terpersonal coping, neutrophil percentage, and lymphocyte pe rcentage yielded no significant relationships. However, the percentage nature of these data, and the lim ited availability of neutrophil/lymphocyte data limited the ability to adequately explore thes e WBC subsets, which according to previous research, may constitute more clinically meaningf ul predictors of surgical outcome than overall WBC count on its own. The quality of overall WBC counts predictive ability of surgical outcome is further questioned by results from research examining the utility of leukodepletion on surgical outcome.

PAGE 131

131 The possibility that less elevated WBC count ma y be an adaptive respon se to surgical tissue injury receives indirect s upport from research examining leukodepletion procedures during cardiac surgery. Due to the observed detrimenta l effects of leukocyte activation on end-organ injury, it has been proposed that leukocyte removal with speci al filters may diminish the deleterious effects of leukocyt e activation during co ronary artery bypass machine procedures (Alexiou et al., 2006). Although leuk ocyte filters do not appear to significantly decrease overall WBC counts, there is evidence that filters act to preferentially reduce activated leukocytes, resulting in improvements in post-surgical pulm onary function and attenu ation of reperfusion injury at the cellular le vel (Warren et al., 2007). Thus, it is pos sible that examining the subset of leukocytes that are activated ma y yield more meaningful predic tive ability than overall WBC count. However, it should be noted that these findings have been found among individuals whose surgery does not involve organ re moval, whereas the individuals in the present study undergo surgery which removes the diseased organ. Once again, this highlights the complex nature of interpreting the relations hip between greater perceived emo tional support and less elevated WBC count. Thus, although the methodology of the present study did not allow for the collection of data on activated leukocytes, this information may help to elucidat e the robust relationship that emerged between interpersonal copi ng and less elevated WBC count. It is also possible that medicat ions administered either preor post-surgically may have resulted in a limited the range of response in WBC counts post-surgicall y. For example, results from the LIPID study revealed th at WBC count predicted cardiova scular mortality in patients with ischemic dis ease randomized to placebo, bu t not in those receiving pravastatin (Stewart et al., 2005), suggesting that statin medications may modulate the eff ects of WBC count on surgical outcome. Because data for the present study did not include information about statin medications

PAGE 132

132 during the acute peri-operative period, it is possible that unmeasur ed statin use may impact WBC count response to surgery. However, it should be noted that there were no significant differences in WBC count between women who took corticosteroid medications pre-surgically and those who did not. The unexpected dire ction of the relationship betw een interpersonal coping and WBC count may also be related to the potential effects of comorbid medical conditions on WBC count response to surgery. Specifically, como rbid conditions, such as chronic obstructive pulmonary disease, renal failure, coronary arte ry disease and diabetes interact with the inflammatory response and may impact an indivi duals ability to mount an appropriate stress response. (Kohl & Deutschma nn, 2006). Given that the wome n in the present study are commonly affected by comorbid me dical conditions, it is possible th at these conditions may have affected the WBC count response to surgery in this population; however, it should be noted that medical comorbidy score was not significantly a ssociated with post-surgical WBC count. In addition to comorbid medical c onditions, reproductive hormones may affect immune response to tissue trauma. Specifically, female reproductive hormones have been found to exhibit immunoprotective properties afte r trauma and severe blood loss (Angele & Chaudry, 2005). Given that most of the participants in the present study were post-menopausal and all underwent a surgical procedure involving the removal of female reproductive organs, it is unlikely that changes in female reproductive hormones would explain affect WBC count in this sample. Finally, genetic variation has been proposed to a ccount for individual differences in the immune response and development of complications following tissue trauma. Research has identified a genetic polymorphism of the neutrophil recepto r for immunoglobulin G (CD16) which may be associated with differences in neutrophil phagoc ytosis (Salmon, Edberg, Brogle, 1992). Thus, it is possible that genetic variation may be a po ssible factor contributi ng to the relationship

PAGE 133

133 between pre-surgical perceived emotional support and less elevated post-surgical WBC count in the present study. In addition to the complex nature of WBC count in the surgical context, it is possible that the unexpected direction of th e relationship between interper sonal coping and post-surgical WBC count may stem from unmeasured psyc hosocial factors between the time emotional support was assessed and the measurement of pos t-surgical WBC count. For example, it is possible that women who perceived greater emotional support from primary support person prior to surgery and hospitalization, experienced a change in their perception of emotional support at the time of surgery or during the acute post-surgical period. Indeed, due to environmental and physical restrictions during the acute post-surgical period, it is possible that women had less access to their primary support person. According to the tend-and-befriend theory of stress response, women are more likely to respond to stress by social affiliation (Taylor et al., 2000). Consequently, perception of isolation or decreas ed affiliation with primary support person during the period of time immediately following surgery may confer deleterious effects for women who previously relied heavily on emotional support from their primary support person. Although the methodology of the present study did not include a measure of perceived emotional support from primary support person during hospitalization, the po ssibility that a change in emotional support from primary support person from preto immedi ately post-surgery resu lted in exploratory analyses examining total perceived emotional support. This measure of interpersonal coping allowed for the inclusion of perceived emotional support from a wider rang e of potential sources, including female friends, healthcare providers, and other family members. However, total emotional support emerged as only a marginally significant predictor of post-surgical WBC count. Although the inclusion of total emoti onal support may have captured perceived support

PAGE 134

134 from a wider range of sources than emotiona l support from primary support person, it is not sensitive to potential changes in support or affiliation that may be inherent in th e post-surgical hospitalization experience. Future research is necessary to clar ify these relations through the investigation of interpersonal coping during the acute post-su rgical hospitalization period. Previous research has identif ied cortisol as a potential ne uroendocrine modulator of the relationship between psychosocial functioning and immune functioning. The findings of the present study revealed a significan t relationship between pre-surgi cal cortisol and post-surgical WBC count; however, like the findings with emo tional support and WBC c ount, the direction of the relationship between cortisol and WBC count was opposite of that originally hypothesized. Specifically, findings revealed th at greater pre-surgical cortis ol indices (e.g., AUG-G, AUC-I, mean daily cortisol, mean morning cortisol) we re associated with more elevated WBC count post-surgically. This is contra ry to what would be expect ed, given the immunosuppressive effects of dysregulated cortisol. The present st udy was unable to empirically examine potential mechanisms that may account for the unexpected relationship between greater pre-surgical cortisol and more elevated WBC counts. However, it is possible that th is relationship emerged due to the timing and nature in which cortisol was measured in the present study. The short duration of measurement (three days) and the fact that cortisol was measured immediately prior to surgery may suggest that it captured the cortisol response to a more acute stressor, namely impending surgery. The literature is replete with evidence that dysregulatio n of cortisol confer deleterious effects to immune functioning (see An toni et al., 2006 for a review). In the psychooncology literature, researchers have approach ed cortisol in several ways. Sephton and colleagues (2000) found that flatte r diurnal cortisol slopes were predictive of s horter survival times among women with metastatic breast can cer; however, marital distress was the only

PAGE 135

135 psychosocial variable associated with diurnal cortisol slope. Abercrombie and colleagues (2004) examined cortisol among women with metastatic breast cancer and healthy controls, finding that no psychosocial variables were associated with flattened cortisol slope s. Additionally, TurnerCobb et al. (2000) examined relations among co rtisol production and social support in women with metastatic breast cancer. A lthough diurnal cortisol slope wa s not significantly associated with social support, although gr eater mean cortisol was associ ated with less social support (Turner-Cobb et al., 2000). Giese-Davis et al. (2004) found that highly anxious women with metastatic breast cancer displayed more flattened diurnal cortisol slopes. Vedhara and colleagues (2006) examined psychosocial predictors of various indices of diurnal cortisol (e.g., diurnal slope, AUC-G, AUC-I, morning peak) among wome n with breast cancer a nd healthy controls, and found that only neuroticism emerged as a signif icant predictor of morn ing peak cortisol. No other psychosocial constructs emerged as signif icant predictors of any other cortisol indices (Vedhara et al., 2006). Thus, the research thus far has been variable in its examination of indices of diurnal cortisol production a nd psychosocial factors. The majority of research thus far has been conducted among women with metastatic breast cancer, a populat ion noted to have unusually high rates of diurnal cortisol dysregu lation. Additionally, none of the aforementioned studies examined women recently diagnosed w ith cancer, or those preparing for oncologic surgery. Consequently, the results of this literatu re may not easily translate to the present study. Additionally, research examini ng the effects of chronic distress and cortisol in nononcology populations has identified blunted cortisol production in response to long-term distress (Ahrens et al., 2008; Yehuda, Teicher, Tres tman, Levengood, & Siever, 1996). Under normal conditions, a negative feedback loop causes an inhibition of stress-elicite d HPA activation when an individual no longer perceives a stressor as stressful. This result s when increases in

PAGE 136

136 circulating glucocorticoids act on the hypothalamus to inhibit further secretion of CRF and, subsequently, the secretion of ACTH from the pi tuitary gland. However, chronically high levels of perceived stress may override this negativ e feedback loop, resulti ng in dysregulation of circulating cortisol and dysregulation in the ove rall cortisol stress respon se (McEwen, 1998). It is plausible that coping with ongoing li fe stressors resulted in a blunt ed cortisol response in which lower levels of cortisol can not necessarily be interpreted as adap tive; however, if this were the case, one might expect a stronger relationship between chronic life event stress (e.g., as was measured by the LES) and cortisol, which was not the case in the present study. Moreover, the methodology of the present study did not allow fo r experimental examination of cortisol response to a stressor; consequent ly, the findings yield no empirical evidence that participants in the present study demonstrated a blunted cortisol stress response. Although this pattern leading to blunted cortisol production may be a plausible explanation for the lack of relationship between measures of stress and cortisol indices, it does not explain why elevated cortisol levels we re associated with more elev ated WBC counts. As discussed above, it is likely that complexities in the WB C count surgical response raise questions about what constitutes the adaptive WBC count respon se to surgery. However, greater indices of cortisol were associated not only with a potentially adaptive systemic immune response postsurgically (i.e., greater elevatio ns in WBC count), but were also associated with lower pain ratings post-surgically. The robust nature of thes e findings suggests the possi bility that, contrary to hypothesis, elevated pre-surgical cortisol c onfers benefits for post-su rgical recovery. Although elevated cortisol is typically discussed as a maladaptive response to chronic stress, research also indicates that in response to an acute, short-term stressor, increa sed cortisol can be an adaptive response. Hans Selye (1936) originally described the adaptive response to stress in his General

PAGE 137

137 Adaptation Syndrome theory. More recent interpreta tions of this theory posit that when faced with a stressor, physiologic mediators of stress, such as gluc ocorticoids, are released and promote adaptation to the stressor (McEwen, 2005) Consequently, the re sults of the present study may be interpreted as an adaptive, short-te rm increase in cortisol related to a potentially adaptive post-surgical systemic immune response. Although this post-hoc hypothesis cannot be examined directly by the present study, the possib ility that cortisol le vels among participants were more reflective of acute stress response, as opposed to chronic dysregulation, may explain the unexpected nature of the relationship between greater cortisol and more elevated WBC count. Additionally, given evidence that the surgical immune response is characterized by both hyperinflammation and immunosuppressi on, and that the local immune response differs from the systemic response in the context of surgery, it is possible that the effect s of cortisol on postsurgical immune response may differ depending on whether one examines local versus systemic immune response. Given that the present study did not examine the local immune response, it is not possible to determine whethe r cortisol confers unique effect s systemically, as opposed to locally. It should also be noted that decline in the es trogen secretion at menopause is associated with increased HPA axis activity (Van Caut er, Leproult, & Kupfer, 1996). Given that postmenopausal status is a significant risk factor for endometrial cancer as well as the advanced age of the sample in the present study, the effects of female re productive hormones on cortisol production cannot be ruled out. Finally, resear ch examining cortisol among women with metastatic breast cancer has iden tified high rates of dysregulated diurnal cortisol production. Indeed, research indicates that among women with metastatic breast cancer, approximately 70% display dysregulated diurnal cort isol slopes, characterized by fl attened slopes, high levels of cortisol, or erratic fluctuati ons (Touitou, Bogdan, Levi, Benavi des, & Auzeby, 1996). Given that

PAGE 138

138 both breast and endometrial cancer s are endocrine-mediated, it is possible that disease processes related to cancer account for dysre gulation in cortisol and its une xpected relations with surgical outcomes and psychosocial predictors. It is noteworthy that no significant relations emerged between indices of cortisol and interpersonal coping. Previous research has identified relationships between social support and cortisol among women with cance r (e.g.,Turner-Cobb et al., 2000), although as Vedhara et al. (2006) point out, there is a growing scarcity in the reliab ility of relationships between indices of cortisol and psychosocial factor s in the literature. This may reflect a genuine lack of a relationship between cortisol i ndices and aspects of psychosocia l functioning, or may reflect methodologic challenges in this line of researc h. The lack of significant relationship between psychosocial functioning and cortisol among part icipants in the presen t study raises several questions. First, it may be that underlying diseas e process related to endo metrial cancer affects cortisol level to a greater degree than do psychosocial factors. I ndeed, endometrial cancer is an endocrine mediated cancer and although relations between psychosocial f unctioning and cortisol have emerged in research examining other endo crine mediated cancers (e.g., Turner-Cobb et al., 2000 ), very little research has examined psychon euroendocrine relations specifically in women with endometrial cancer. Therefore, diseas e impact on neuroendocri ne functioning and its response to psychological factors is largely unknown in this populati on. Second, it is possible that other comorbid conditions such as diabetes or obesity, which are known to impact cortisol (Tomlinson, Finney, Hughes, & Stewart, 2008), may infl uence cortisol levels to a greater degree than psychosocial factors. A lthough the comorbidity index, whic h includes diabetes, was not significantly associated with m easures of cortisol, the present study did not examine the degree to which participants diabetes was well-controll ed, which may serve as a better predictor of

PAGE 139

139 cortisol production than simply the diagnosis of diabetes. BMI was also not significantly related to indices of cortisol; however, more specific measures, such as abdominal adiposity, may be better predictors of cortisol production. Alt hough the relationship between disease process, comorbid conditions, and obesity were not directly examined in the present study, they cannot be ruled out as potential explanations for the lack of relationship between psychosocial factors and cortisol. Furthermore, the present study found no evidence that interpersonal coping moderates the potential relationship between ps ychological stress and indices of post-surgical recovery. The findings of the present study were also somewhat surprising, given the lack of significant relations between psychosocial and endoc rine predictors and th ree of the four main outcome variables: length of hospitalizati on, post-surgical complications, and time to ambulation. These variables were selected in an effort to examine surgical outcome from a diverse and clinically meaningf ul perspective. Several previo us studies have identified significant relationships between psychosocial predictors and length of hospitalization. For example, Krohne et al (2005) found that indivi duals receiving more emotional support were hospitalized approximately 1.5 fewer days post ma xillofacial surgery th an those individuals receiving less emotional support. Moreover, Cont rada et al. (2004) found that individuals with stronger religious beliefs had significantly sh orter hospitalization periods following CABG. In the present study, only a priori biobehavioral control variable s significantly predicted postsurgical length of hospitalization. Thus, in the present study, a stronger relationship emerged between medical variables, such as length of su rgery, and post-surgical hospitalization than the relationship between psychosocial/endocrine variables and lengt h of hospitalization. Additionally, given the multifactoria l nature of post-surgical hospi talization, it is possible that

PAGE 140

140 non-measured factors other than the a priori biob ehavioral control variables contributed to postsurgical hospitalization. Given the high costs asso ciated with inpatient hospitalizations, it is possible that economic factors such as health insurance played a role in the length of hospitalization postsurgery. Additionally, given the poten tial susceptibility to infection and illness in the hospital, it may be that healthcare providers took a more proactive approach to early discharge. Finally, although previous research has identi fied significant psychosocial predictors of post-surgical length of hospitalizati on, it should be noted that those studies examined surgical populations that are potentially quite disparate fr om participants in the present study. Krohne et al. (2005) investigated individuals undergoing maxillofacial surgery and Contrada et al. (2004) investig ated individuals undergoing CABG Both of these procedures likely involve different levels of tissue trauma and sizes of incision sites. Thus, it is possible that among women undergoing an invasive procedure such as TAH-BSO, the effect of psychosocial factors on length of hospitalizati on is minimal when compared to clinical medical factors. Contrary to the present study, previous studies have identified si gnificant relationships between psychosocial predictors and post-surgical complications. For example, in a population of individuals undergoing CABG surgery, Contrada et al. (2004) reported that individuals with stronger religious beliefs had fewer post-surgical complications. Moreover, in a general surgery population, Kopp et al. (2003) found th at life satisfaction and social support predicted surgical recovery without complications. The present st udy did not yield eviden ce of a relationship between psychosocial and endocrine predictors and post-surgical compli cations. Moreover, the present study examined not only incidence of post-surgical comp lications, but also severityweighted incidence of complications as rated by a board-certified gynecologic oncologist. It was expected that examining severity of post-surgical complications would yield more clinically

PAGE 141

141 relevant results than past studies that ha ve only examined incidence of post-surgical complications. It should be not ed that post-hoc analyses revealed a significant relationship between severity of post-surgi cal complications and post-surgi cal length of hospitalization ( r = .49, p < .001), such that women with more seve re complications had longer length of hospitalization. Due to the correlational, cros s-sectional nature of this data, it is not possible to determine causality from this finding. However, in research examining psychosocial predictors of post-surgical complications, Contrada et al (2004) found that post-surgical complications mediated the relationship between psychosoc ial predictor and length of hospitalization. However, in the present study, post-surgical co mplications and length of hospitalization were treated as independent outcomes of post-surgical outcome, given th e inability to establish causal direction between the two variable s. It should also be noted that unlike the findings related to length of hospitalization, no biobe havioral control variables emer ged as significant independent predictors of post-surgical complications. Severa l factors may account for the lack of support for relations between psychosocial and endocrine functioning and severity of post-surgical complications. First, data about post-surgical co mplications was abstract ed from participants official discharge notes, which summarize the events of their entire hospitalization. Through the course of the study, raters observed a wide range of variation in th e level of detail or content of information in the discharge notes, suggesting that the validity of the da ta is dependent upon the healthcare providers documentation of complications in the discharge notes. Thus, it is possible that minor complications that would be part of the participants paper inpatient medical chart may not have been uniformly documented in th eir electronic discharg e summaries. Although the methods of the present study limit further investigation of this possibility, anecdotal evidence from raters of post-surgical complications sugge sts that this is a plausible concern. Second, the

PAGE 142

142 present study was limited to the examination of complications that arose during post-surgical hospitalization. As such, analyses were not ab le to examine incidence or severity of complications that may have arisen later in post-surgical recovery after hospitalization. It is likely that certain types of post-surgical co mplications (e.g., wound comp lications) may not be identified until participants present to the outpa tient gynecologic oncology clinic for removal of surgical staples. Thus, the data in the present study fails to capture the full spectrum of potential post-surgical complications that participants ma y have experienced during their recovery, which may account for the lack of significant findings between psychosocial functioning and postsurgical complications. Psychosocial and endocrine variables also did no t emerge as significant predictors of time to post-surgical ambulation in this study. Time to post-surgical ambulation was included as an outcome variable in this study both because it re presents a more functional outcome of surgery, as well as an outcome that may be more related to behavioral factors, such as adherence to medical recommendations. Similar to observed inconsistencies in the documentation of postsurgical complications, day of post-surgical ambul ation was infrequently clearly documented in participants medical records, leading to a sizeable amount of missing data. Additionally, review of participants medical records revealed that the recommendation that hospital staff encourage ambulation shortly after the participants return from the recovery room was a standard entry in the acute post-surgical physicians orders. T hus, it is possible that time to post-surgical ambulation emerged more as a proxy measure of compliance with medical recommendations than a functional index of post-surgical recovery. Nevertheless, it is likely that the small number of participants with complete post-surgica l ambulation data limited the power to detect meaningful relationships with this outcome variable.

PAGE 143

143 Overall, only post-surgical WBC count was significantly related to psychosocial functioning and cortisol. It is noteworthy that, whereas length of hospitali zation, post-surgical complications, and time to ambulation may be consid ered more indirect indices of the potential effect of stress, coping, and cortisol dysregulation on surgical outcome, given the previously established relationship between psychosocial, endocrine factors, and immune functioning, WBC count appears to be a more directly rela ted outcome measure of psychoneuroimmunologic predictors. Psychoneuroimmunologic Predicto rs of Post-Surgical Pain In addition to the findings related to the designated outcom e variables discussed above, exploratory post-hoc analyses yi elded interesting findings relate d to psychosocial and endocrine predictors of post-surgical pain. Specifically, preliminary analysis identified a significant relationship between greater pre-surgical perc eived emotional support and lower post-surgical pain ratings. When perceived stress and cortisol were examined as predictors in a model with post-surgical pain as the outcome variable AUC-I emerged as a significant independent predictor of post-surgical pain, such that greater pre-surgical AUC-I was associated with lower post-surgical pain ratings. Similarly, in a m odel examining emotional support and cortisol as predictors of post-surgical pain, AUC-I emerged as a marginally significant predictor of postsurgical pain, with a trend toward lower AUC-I being related to greate r post-surgical pain. Emotional support from primary support person also emerged as a marginally significant predictor of post-surgical pain, with a trend toward a relationship between lower pre-surgical perceived emotional support and greater post-surgic al pain. These findings are consistent with previous research examining psyc hosocial predictors of post-surg ical pain. Kain et al. (2000) also found support for the relationship between pre-surgical psychosocial factors and postsurgical pain, with greater pr e-surgical anxiety related to greater post-operative pain among

PAGE 144

144 women undergoing abdominal hyste rectomy. Moreover, Cohen et al. (2005) examined both preoperative distress and coping as predictors of post-operative pain and morphine consumption. The findings revealed that greater pre-operative dist ress predicted greater post-operative pain and morphine consumption; however, this relationshi p was no longer significant when pre-operative coping was controlled for (Cohen et al., 2005). Cohen et al (2005) also found that the use of more passive coping strategies was associated with elevated pain and morphine consumption post-operatively (Cohen et al., 2005). The lack of support for a relationship between pre-surgical stress and post-surgical pain may be attributable to the measurement of stress in the current study, as discussed earlier. It is possible that measures of stress such as distress, as was used by Cohen et al., as opposed to life stress or perceived stress, are more closely related to the experience or report of post-surgical pain. However, the present study extends the findi ngs of Cohen et al. ( 2005) and Kain et al. (2000) via the finding that grea ter pre-surgical AUC-I was a si gnificant predictor of less postsurgical pain. It should be noted that due to the exploratory nature of this analys is the present study was not able to examine analgesic cons umption during the period of time in which participants were rating their pain levels. As su ch, the possibility exists that participants who perceived greater pre-surgical emotional support were also mo re likely to request pain medications when needed, resulti ng in better managed pain and lo wer pain ratings. Nevertheless, these exploratory findings provide converging evidence to support the relationship between coping and post-surgical pain. Implications of Findings The present study is am ong the first to ex amine psychoneuroimmunologic predictors of clinical surgical outcome in the context of oncologic surgery. As such, the findings of the present study extend the literature on predictors of pos t-surgical recovery, which has generally been

PAGE 145

145 carried out in the context of general surgery populations. Although the present study yielded some unexpected findings, it presents important implications for future research. First, the findings of the present study revealed that among pa rticipants in this sample, more elevated presurgical cortisol indices pred icted post-surgical systemic immune response and lower postsurgical pain ratings. Thus, future research may clarify these relationships through the inclusion of repeated measures of cortis ol indices, as opposed to solely measuring cortisol immediately prior to surgery. Such research may elucidate whether the relationship between more elevated pre-surgical cortisol indices a nd improved surgical outcome is re flective of a blunted cortisol response to stress, or simply reflective of an adaptive physiologic response to an acute stressor (e.g., surgery). Future research may also cons ider employing behavioral stress paradigms to directly and empirically examine the cort isol stress response in this population. Second, the finding that greater pre-surgical perceived emotional support (both from primary support person as well as multiple sources) was predictive of less elevated WBC counts, highlights the importance of ex amining interpersona l coping during the immediate post-surgical (e.g., hospitalization) period. Future research incorporating measur es of psychosocial functioning and interpersonal coping while re covering in the hospital may provi de important insight into the relationship between psychosocial functioning and surgical recovery. Such research may also yield important clinical implicat ions for potential psychosocial interventions for individuals hospitalized during post-surgical recovery. Third, the present study focused on psychoneur oimmunologic predictors of post-surgical recovery restricted to the acute post-surgical hospitalization period. As su ch, longer-term aspects of both physical and functional re covery were not included in the analyses of the present study. Future research in this area should consider i nvestigation of longer-term post-surgical recovery

PAGE 146

146 in this population, part icularly given the often lengthy r ecovery from TAH-BSO. A longer follow-up period would also provide the opportu nity to examine aspects of psychosocial functioning once women have been discharged from the hospital and may elucidate the potential psychosocial changes as women progress from diagnosis, to preparation for surgery, to hospitalization, and to lo nger term recovery and adaptation to cancer. Finally, the present study attempted to look at multiple clinically m eaningful indices of post-surgical recovery. However, recovery from invasive surger y presents many more challenges than those captured in the out come variables in the present study. Specifically, given their high rate of comorbid conditions, such as obesity and diabetes, as well as the experience of significant surgical incision and tissue removal, women with endometrial cancer ar e at greater risk for wound healing complications. The psychoneuroimm unologic literature is replete with evidence that psychosocial factors such as distress may impede healthy wound healing. Consequently, future research may wish to examine psychoene uroimmunologic aspects of wound healing in this population. Such research would extend laboratory research on PN I factors in wound healing and may yield important clinical im plications in a population at risk for wound complications. Study Limitations Despite the important im plications discussed above, the results of the present study should be interpreted with caution, gi ven a number of important lim itations. First, although the prospective nature of this design suggests te mporal directionality in relations between presurgical and post-surgical variables, the corr elational nature of th e research precludes interpreting causality. In particular, the present study did not assess or control for numerous psychological and medical variables that may account for relations that emerged in the findings of the present study. Future research is necessary to clarify the complex and likely multifactorial nature of the relationships observed in the present study.

PAGE 147

147 Second, the measurement of the variables of in terest presents anot her potential limitation for the present study. As discussed ea rlier, it is possible that the selection of impact of negative events during the past six months and perceived stress during the past mo nth, did not specifically capture stress participants expe rienced related to their cancer diagnosis and upcoming surgery. As such, whereas the measures of stress were likely more reflective of long-term or more chronic stress, the mediating (indices of cortisol) vari ables, as well as the outcome variables may be more susceptible to the effects of acute stress, distress, or anxiety during the peri-operative period. Additionally, as discussed above, the lack of information about stress or interpersonal coping during the acute post-surgical hospitalization period presents a significant limitation in the interpretation of the observed findings. As such, it remains unclear whether psychosocial functioning immediately following surgery may have had a greater impact on surgical recovery than psychosocial functioning during the pre-surg ical period. The utiliz ation of overall WBC count from peripheral blood draw constitutes anot her significant limitation of the present study. Examination of immune response at the site of tis sue injury (as opposed to systemically), as well as inclusion of data related to activation of leukocytes and leukocyte subsets may yield more interpretable and clinically meaningful findings Moreover, the present study is also limited by its restricted focus on surgical recovery dur ing post-surgical hospita lization. Although the postsurgical hospitalization period represents a critical point in the peri-operative period, recovery from TAH-BSO frequently extends well beyond th e time of discharge and there may be unique challenges to surgical recovery that arise af ter women return to their homes. Consequently, future research is needed to examine the entire spectrum of the peri-operative period in order to clarify the psychoneuroimmunologic re lations and their association with clinical indices of both short-term and long-term post-surgical recovery.

PAGE 148

148 Third, the lack of information about particip ants compliance with the saliva collection protocol presents another salient limitation, partic ularly given the unexpected relationships that emerged between more elevated indices of co rtisol and improved pos t-surgical recovery. Participants were asked to collect four samples each day, for thr ee consecutive days. This sample collection schedule may have been burdensome to pa rticipants, particularly those at an advanced age, due to the need for partic ipants to carry their collection supplies with them throughout the day. Compliance with saliva sampling is essential to the accurate measurement and interpretation of cortisol (Kudielka, Broderick, & Kirschbaum 2003). In fact, diurnal cortisol slope can demonstrate a 100% change within 30 minutes (Kirschbaum & Hellhammer, 1994). In an effort to ensure more accurate cortisol data, participants were instruct ed to label salivettes with the actual time the sample was collected, even if it devi ated from the collection times specified in the protocol. Indices of cortisol were computed based on the time participants indicated their samples were collected, as opposed to the designated collection times. However, the methodology of this study provided no means to a ssess whether samples were actually collected when participants reported. Although it is generally assumed that particip ants would record the actual time of sample collection, it is possible that demand charac teristics may have led some participants to report collecting their samples at the designated time, even if they did not do so. Future research should consider the use of compliance monitoring devices to ensure more reliable diurnal cortisol data. It should also be noted that the present studys saliva collection protocol may raise limitations when interpreting the findings. The cu rrent study used four saliva samples per day (8AM, 12PM, 5PM, 9PM) to plot the diurnal co rtisol slope and calcula te AUC-G, AUC-I, mean daily cortisol and mean morning cortisol. Althou gh previous studies exam ining diurnal cortisol

PAGE 149

149 slope among people with cancer have typically used four daily samp les to plot the diurnal slope (e.g., Abercrombie et al., 2004; Sephton et al., 2000; Turner-Cobb et al ., 2000), many studies examining diurnal cortisol among non-oncologic populations have collected more than four daily samples (Kirschbaum & Hellhammer, 1994; Ki rschbaum, Kudielka, Gaab, Schommer, & Hellhammer, 1999). Specifically, some researchers ha ve raised concern that a small number of potentially unreliable samples could bias the estimation of di urnal slope (Stone et al., 2001). Consequently, it should be noted that the four samples collecte d in the current study may not have provided adequate or reliab le data from which to calculate indices of diurnal cortisol. Moreover, some research has focused on the partic ipants actual time of waking, as opposed to a designated morning collection time, for the first sa liva collection of the da y. It is possible that using individual participants actual time of waking may produce a more accurate approximation of their diurnal cortisol slope, including the peak at waking. As suc h, the presents studys designation of 8:00 AM as the first saliva collection time ma y result in less accurate determination of indices of diurnal cortisol. Fourth, a considerable lim itation of the present study involves missing data across variables, resulting in smaller than desired sample size in a number of analyses. For example, it was not possible to obtain cortis ol samples from a substantial portion of participants in the present study, as a resu lt of participants declining to collect samples, non-compliance with collection, insufficient quantity of saliva for assay, and timing of recruitment. This missing cortisol data resulted in smaller than projected sample sizes and likely inadequate power to detect significant results. Similarly, inconsistencies in documentation in participants medical records resulted in substantially more missing data for time to post-surgical ambulation than the other main outcome variables. Again, the sample size in analyzes involving post-surgical ambulation

PAGE 150

150 likely lacked the power to detect significant relations with pr edictor variables. In addition, inconsistencies in the content and level of detail of documentation of postsurgical complications raises some questions about the accuracy of the post-surgical co mplication data included in the present study. Although these types of problems are inherent to c onducting clinical research in an applied medical setting (as opposed to laborato ry), future research should consider employing methods to ensure more rigorous and consistent documentation of variables of interest and should over-estimate sample size calculation in a conservative attempt to circumvent often inevitable problems due to missing data. Fifth, the multiple comparisons and analyses performed in the present study present a notable limitation, given the poten tial for increased lik elihood of Type I error. Type I error involves rejecting the null hypothesis when it is actually true, leadi ng to falsely positive results. Although the number of analyses performed in th e present study may intr oduce Type I error, given that very few analyses yielded significant findings it is unlikely that the Type I error would have impacted the pattern of results in the present study. A final limitation involves the clinical relevanc e of examining psychoneuroimmunologic predictors of post-surgical re covery. It has been argued th at the identification of new psychosocial predictors of medical outcome carries little weig ht within the field of medicine unless these predictors are indepe ndent of already established (e .g., medical/biologic) predictors (Freedland, 2004). In other words, it could be argued that the findings of the present study have little real-world signif icance if they do not pr edict surgical outcome above and beyond other well-established predictors. Although the findings of the pr esent study are not likely to significantly influence gynecologi c surgery practice, they provide important information about risk factors for complicated su rgical recovery among women with endometrial cancer. Moreover,

PAGE 151

151 they highlight risk factors that unlike certain medical/surgical factors, may be amenable to modification via psychosocial intervention. Conclusions The present study is am ong the first studi es to examine psychoneuroimmunologic predictors of post-surgical rec overy. Moreover, it extends previ ous findings about relationships between psychosocial factors and surgical outcome in the general surgery population by examining these relationships in the unique context of oncologic surgery. The present study also extends the previous literature by examining these potential relationships specifically in a population of women who ar e thought to be at greater risk fo r more complicated post-surgical recovery. The findings suggest significan t relationships between interp ersonal coping and indices of post-surgical recovery, including post-surgical systemic immune response and post-surgical pain ratings. The present study also found relations among pre-surgical indices of cortisol production and post-surgical systemic immune response and pos t-surgical pain ratings. Contrary to initial hypothesis, greater pre-surgical emotional support was associated with less elevated postsurgical WBC count and more elevated pre-surgi cal indices of cortisol were associated with more elevated post-surgical WBC count. The fi ndings suggest a trend toward a relationship between greater pre-surgical perceived emo tional support from primary support person and lower post-surgical pain ratings. The present study also found that more elevated pre-surgical indices of cortisol were associat ed with lower post-surgical pain ratings. Contrary to hypothesis, psychosocial stress was not related to any index of post-surgical outcome, nor was it related to any index of cortisol production. Post-surgical WBC count and post-surgical pain ratings were the only indices of post-surg ical recovery that were significantly related to psychoneuroimmunologic predictors above and beyond the effects of biobehavioral control

PAGE 152

152 variables. The present study highlights both the importance of and methodologic challenges related to conducting psychoneur oimmunologic research in the applied medical setting. It provides preliminary data for future resear ch to examine psychoneuroimmunologic relations during the peri-operative period among women with endometrial cancer. .

PAGE 153

153 LIST OF REFERENCES Abdelhadi, R.H., Gur m, H.S., Van Wagoner, D.R., Chung, M.K. (2004). Relation of an exaggerated rise in white bl ood cells after coronary bypass or cardiac valve surgery to development of atrial fi brillation postoperatively. American Journal of Cardiology 93, 1176-1178. Abercrombie, H.C., Giese-Davis, J., Sephon, S ., Epel, E.S., Turner-Cobb, J.M., Spiegel, D. (2004). Flattened cortisol rhythms in metastatic breast cancer patients. Psychoneuroendocrinology, 29, 1082-1092. Adler, N.E., Epel, E.S., Castellazzo, G., Ickovics J.R. (2000). Relationship of subjective and objective social status with psychological and physiological functioning: Preliminary data in healthy, White women. Health Psychology, 19 586-592. Ahrens, T., Deuschle, M., Krumm, B., van de r Pompe, G., den Boer, J..A., Lederbogen, F. (2008). Pituitary-adrenal and sympathetic ner vous system responses to stress in women remitted from recurrent major depression. Psychosomatic Medicine 70, 461-467. Aiken, L.S., West, S.G. (1997). Multiple regression: Testi ng and interpreting interactions Newbury Park, CA: Sage. Alexiou, C., Sheppard, S., Tang, A., Rengaraja n, A., Smith, D., Haw, M., et al. (2006). Leukocytes-depleting filters preferentially re move activated leukocytes and reduce the expression of surface adhesion molecules during the simulated extracorporeal circulation of human blood. American Society for Artificial Internal Organs Journal 52, 438-444. Amar, D., Goenka, A., Zhang, H., Park, B., Thaler, H.T. (2006). Leukocytosis and increased risk of atrial fibrillation afte r general thor acic surgery. Annals of Thoracic Surgery 82, 10571061. American Cancer Society (2003). Detailed Guid e: Endometrial Cancer. Atlanta: American Cancer Society. American Cancer Society (2005). Cancer Facts and Figures 2005. Atlanta: American Cancer Society. Angele, M.K., Chaudry, I.H. (2005). Surgical trauma and immunosuppre ssion: pathophysiology and potential immunom odulatory approaches. Langenbecks Archives of Surgery, 390 330-341. Antoni, M.H., Lutgendorf, S.K., Cole, S.W., Dh abhar, F.S., Sephton, S.E., McDonald, P.G., et al. (2006). The influence of bio-behavioural factors on tu mour biology: pathways and mechanisms. Nature Review Cancer 6 240-248. Baron, R.M., Kenny, D.A. (1986). The moderator-m ediator variable distinction in social psychological research: Con ceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51 1173-1182.

PAGE 154

154 Beck, A.T., Steer, R. A. (1991). Manual for the Beck Scale for Suicide Ideation San Antonio, TX: The Psychological Corporation. Ben-Eliyahu, S. (2003). The promotion of tumor metastasis by surgery and stress: immunological basis and implications for psychoneuroimmunology. Brain, Behavior, and Immunity, 17 S27-S36. Bieliauskas, L.A., Counte, M.A., Glandon, G.L. (1995). Inventorying stressful life events as related to health change in the elderly. Stress Medicine 11, 93-103. Bohnen, N., Terwel, D., Twijnstra, A., Markerink, M. (1992). Effects of ap prehension of lumbar puncture procedure on salivary cortisol, pl asma vasopressin and osmolality in man. Stress Medicine, 8, 253-257. Bratzler, D.W., Hunt, D.R. (2006). The surgical infection preventi on and surgical care improvement projects: national initiatives to improve outcomes for patients having surgery. Clinical Infectious Diseases 43, 322-330. Broadbent, E., Petrie, K.J., Alley, P.G., Booth, R.J. (2003). Psychological stress impairs early wound repair following surgery. Psychosomatic Medicine 65 865-869. Brown, J.R., Cochran, R.P., Leavitt, B.J., Dacey, L. J., Ross, C.S., Mackenzie, T.A., et al. (2007). Multivariable prediction of renal insufficiency developing after cardiac surgery. Circulation 116, 1139-1143. Byrnes, D.M., Antoni, M.H., Goodkin, K., Efantis -Potter, J., Asthana, D., Simon, T., et al. (1998). Stressful events, pessimism, na tural killer cell cy totoxicity, and cytotoxic/suppressor T cells in HIV+ bl ack women at risk for cervical cancer. Psychosomatic Medicine 60, 714-722. Carver, C.S. (1999). The Sources of Social Support Survey. Unpublished Instrument. Charlson, M.E., Pompei, P., Ales, K., MacKenzi e, C.R. (1987). A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. Journal of Chronic Diseases 40, 373-383. Clancy, J. (1998). Basic concepts in immunology : A students survival guide New York: McGraw-Hill. Cohen, L., Fouladi, R.T., Katz, J. (2005). Pre operative coping strategi es and distress predict postoperative pain and morphine consum ption in women undergoing abdominal gynecologic surgery. Journal of Psychosomatic Research 58 201-209. Cohen, S., Kamarck, T., Mermelstein, R. (1983). A global measure of perceived stress. Journal of Social and Health Behavior 24, 385-396. Cohen, S., Gottlieb, B.H., Underwood, L.G. (2001). Social relationships and health: Challenges for measurement and intervention. A dvances in Mind-Body Medicine 17 129-141.

PAGE 155

155 Coller, B.S. (2005). Leukocytosis and ischemic va scular disease morbidity and mortality: is it time to intervene? Arteriosclerosis, Thrombosis, and Vascular Biology 25 658-670. Contrada, R.J., Goyal, T.M., Cather, C., Rafalson, L., Idler, E.L., Krause, T.J. (2004). Psychosocial factors in outcomes of heart su rgery: The impact of religious involvement and depressive symptoms. Health Psychology, 23 227-238. Cordova, M.J., Cunningham, L.L., Carlson, C.R., Andrykowski, M.A. (2001). Social constraints, cognitive processing, and adju stment to breast cancer. Journal of Consulting and Clinical Psychology, 69, 706-711. Croy, C.D., Novins, D.K. (2005). Methods for addressing missing data in psychiatric and developmental research. Journal of the American Acade my of Child and Adolescent Psychiatry, 44, 1230-1240. Cyr, P.R., Skelton, A.K. (2003). Menstrual disorders. In A.K. David, T.A. Johnson, D.M. Phillips, & J.E. Scherger (Eds.), Family Medicine: Principles and Practice-6th Edition (Section 103). New York: Springer-Verlag New York, Inc. DeKeyser, F. (2003). Psychoneuroimmunol ogy in critically ill patients. AACN Clinical Issues ,14, 25-32. Dimick, J.B., Chen, S.L., Taheri, P.A., Henders on, W.G., Khuri, S.F., Campbell, D.A. (2004). Hospital costs associated with surgical comp lications: a report from the private-sector National Surgical Quality Improvement Program. Journal of the American College of Surgeons 199 531-537. Dorigo, O., Goodman, A. (2003). Prem alignant & malignant disorder s of the uterine corpus. In A.H. DeCherney & L. Nathan (Eds.), Current Obstetric & Gynecologic Diagnosis & Treatment-9th edition (Section 48). New York: The McGraw-Hill Companies. Enders, C.K. (2006). A primer on the use of mo dern missing-data methods in psychosomatic medicine research. Psychosomatic Medicine 68, 427-436. Freedland, K.E. (2004). Religious beliefs shor ten hospital stays? Psychology works in mysterious ways: comment on Contrada et al. (2004). Health Psychology, 23, 239-242. Gallagher-Thompson, D., Shurgot, G.R., Rider, K ., Gray, H.L., McKibbin, C.L., Kraemer, H.C., et al. (2006). Ethnicity, stress, and cortisol function in Hi spanic and non-Hispanic white women: A preliminary study of family dementia caregivers and noncaregivers. American Journal of Geriatric Psychiatry 14, 334-342. Giannoudis, P.V., Dinopoulos, H., Chalidis, B., Hall, G.M. (2006). Surgical stress response. Injury, 37, S3-S9. Gibson, P.H., Croal, B.L., Cuthbertson, B.H., Sma ll, G.R., Ifuzelike, A.I., Gibson, G., et al. (2007). Preoperative neutrophil-lymphocyte ratio and outcome from coronary artery bypass grafting. American Heart Journal, 154, 995-1002.

PAGE 156

156 Giese-Davis, J., Sephton, S.E., Abercrombie, H. C., Duran, R.E., Spiegel, D. (2004). Repression and high anxiety are associated with aberrant diurnal cortisol rhythms in women with metastatic breast cancer. Health Psychology, 23, 645-650. Giese-Davis, J., Dimiceli, S., Sephton, S., Spie gel, D. (2006). Emotional expression and diurnal cortisol slope in women with metastatic breast cancer in supportive-expressive group therapy: A preliminary study. Biological Psychology 73, 190-198. Glaser, R., Kiecolt-Glaser, J. K., Marucha, P.T ., MacCallum, R. C., Laskowski, B.F., Malarkey, W. B. (1999). Stress-related changes in proinflammatory cytokine production in wounds. Archives of General Psychiatry 56, 450-456. Goodkin, K., Fuchs, I., Feaster, D., Leeka, J., Rishel, D.D. (1992). Life st ressors and coping style are associated with immune measures in HIV-1 infectiona preliminary report. International Journal of Psychiatry in Medicine 22, 155, 172. Gurm, H.S., Bhatt, D.L., Gupta, R., Ellis, S.G ., Topol, E.J., Lauer, M.S. (2003). Preprocedural white blood cell count and death after pe rcutaneous coronary intervention. American Heart Journal 146 692-698. Hakim, R.B., Benedict, M.B., Merrick, N.J. (2004). Quality of care for women undergoing a hysterectomy: effects of in surance and race/ethnicity. American Journal of Public Health 94, 1399-1405. Hellhammer, K.H., Buchtal, J., Gutberlet, I ., Kirschbaum, C. (1997). Social hierarchy and adrenocortical stress reactivity in men. Psychoneuroendocrinology 22, 643-650. Jordan, J.V., Kaplan, A.G., Miller, J.B., Stiver, I.P., Surrey, J.L.(1991). Womens Growth in Connection New York: Guilford Kain, Z.N., Sevarino, F., Alexande r, G.M., Pincus, S., Mayes, L. C. (2000). Preoperative anxiety and postoperative pain in wo men undergoing hysterectomy: A repeated measures design. Journal of Psychosomatic Research, 49, 417-422. Katz, J., Poleshuck, E.L., Andrus, C.H., Hogan, L.A., Jung, B.F., Kulick, D.I., et al. (2005). Risk factors for acute pain and its persis tence following breast cancer surgery. Pain 119, 16-25. Kiecolt-Glaser, J.K., Marucha, P.T., Malark ey, W.B., Mercado, A.M., Glaser, R. (1995). Slowing of wound healing by psychological stress. Lancet 346 1194-1196. Kiecolt-Glaser, J.K., Page, G.G., Marucha, P.T., MacCallum, R.C., Glaser, R. (1998). Psychological influences on surgic al recovery. Perspectives from psychoneuroimmunology. American Psychologist 53, 1209-1218. Kiecolt-Glaser, J. K., Loving, T. J., Stowell, J. R., Malarkey, W. B., Le meshow, S., Dickinson, S. L., et al. (2005). Hostile marital intera ctions, proinflammator y cytokine production, & wound healing. Archives of General Psychiatry 6, 1377-1384.

PAGE 157

157 Kirschbaum, C., Hellhammer, D.H. (1994). Salivar y cortisol in psychoneuroendocrine research: Recent developments and applications. Psychoneuroendocrinology 19, 313-333. Kirschbaum, C., Pruessner, J.C., Stone, A.A., Federenko, I. (1995). Pers istent high cortisol responses to repeated psychological st ress in a subpopulation of healthy men. Psychosomatic Medicine 57, 468-474. Kirschbaum, C., Kudielka, B.M., Gaab, J., Scho mmer, N.C., Hellhammer, D.H. (1999). Impact of gender, menstrual cycle phase, and oral contraceptives on the activity of the hypothalamus-pituitary-adrenal axis. Psychosomatic Medicine 61, 154-162. Kohl, B.A., Deutschman, C.S. (2006). The infl ammatory response to surgery and trauma. Current Opinion in Critical Care, 12 325-332. Kopp, M., Bonatti, H., Haller, C., Rumpold, G., Sollner, W., Holzner, B., et al. (2003). Life satisfaction and active coping style are importa nt predictors of recovery from surgery. Journal of Psychosomatic Research, 55, 371-377. Krohne, H.W., Slangen, K.E. (2005). Influence of social support on adaptation to surgery. Health Psychology, 24, 101-105. Kudielka, B.M., Broderick, J.E., Kirschbaum, C. (2003). Compliance with saliva sampling protocols: Electronic monitori ng reveals invalid cortisol da ytime profiles in noncompliant subjects. Psychosomatic Medicine, 65 313-319. Lamm, G., Auer, J., Weber, T., Berent, R., N g, C., Eber, B. (2006). Postoperative white blood cell count predicts atrial fi brillation after cardiac surgery. Journal of Cardiothoracic and Vascular Anesthesia 20, 51-56. Lazarus, R.S., Folkman, S. (1984). Stress, appraisal and coping New York: Springer Publishing Company. Leserman, J., Petitto, J.M., Gu, H., Gaynes, B. N., Barroso, J., Golden, R.N., et al. (2002). Progression to AIDS, a clinical AIDS c ondition and mortality: Psychosocial and physiological predictors. Psychological Medicine 23, 1059-1073. Liebeskind, J.C. (1991). Pain can kill. Pain 44, 3-4. Lutgendorf, S.K., Johnsen, E.L., Cooper, B., A nderson, B., Sorosky, J.I., Buller, R.E., et al. (2002). Vascular endothelial growth factor and social suppor t in patients with ovarian carcinoma. Cancer 95, 808-815. Manyande, A., Berg, S., Gettins, D., Stanford, S. C., Mazhero, S., Marks, D.F., et al. (1995). Preoperative rehearsal of active coping imagery influences subjective and hormonal responses to abdominal surgery. Psychosomatic Medicine, 57 ,177-82. Marucha, P.T., Kiecolt-Glaser, J.K., Favagehi, M. (1998).Mucosal wound healing is im paired by examination stress. Psychosomatic Medicine 60 362-365.

PAGE 158

158 Massie, M.J., Popkin, M.K. (1998). Depressive disorders (pp. 518-540). In J.C. Holland, W. Breitbart, P.B. Jac obsen, et al. (Eds.). Psycho-Oncology. Oxford University Press, New York. McEwen, B.S. (1998). Protective and damaging effects of stress mediators, The New England Journal of Medicine 338 171-179. McEwen, B.S. (2005). Stressed or stre ssed out: What is the difference? Journal of Psychiatry and Neuroscience 30, 315-318. McGuire L., Heffner, K., Glaser, R., Needleman, B., Malarkey, W. B., Dickinson, S., et al. (2006). Pain and wound heali ng in surgical patients. Annals of Behavioral Medicine 3,165-172. Menger, M.D., Vollmar, B. (2004). Surgical trauma: hyperinflammation versus immunosuppression? Langenbecks Archives of Surgery, 389, 475-84. Mormont, M.C., Levi, F. (1997). Circadian-syst em alterations during cancer processes: A review. International Jo urnal of Cancer, 70, 241-247. Morrow, P.C. (1996). Incisions and wound healing. pp. 141. In: P.C. Morrow and J.P. Curtin, (Eds.), Gynecologic cancer surgery New York: Churchill Livingstone Nelson, C.J., Lysle, D.T. (1998). Severity, time, and beta-adrenergic receptor involvement in surgery-induced immune alterations. Journal of Surgical Research 80, 115-122. Nichols, D.H., DeLancey, J.O.L. (Eds.). (1995). Clinical problems, injuries, and complications of gynecologic and obstetric surgery Baltimore, MD: Williams & Wilkins. Nijm, J., Kristenson, M., Olsson, A.G., Jonasson, L. (2007). Impaired cortisol response to acute stressors in patients with co ronary disease. Implications for inflammatory activity. Journal of Internal Medicine 262 375-384. Noyes, R., Holt, C.S., Massie, M.J. (1998). Anxi ety Disorders (pp. 548-563). In J.C. Holland, W. Breitbart, P.B. Jac obsen, et al. (Eds.). Psycho-Oncology. Oxford University Press, New York. Ockenfels, M.C., Porter, L., Smyth, J., Kirsc hbaum, C., Hellhammer, D.H., Stone, A.A. (1995). Effect of chronic stress associated with unemp loyment on salivary cortisol: Overall cortisol levels, diurnal rhythm, and acute stress reactivity. Psychosomatic Medicine 57, 460-467. Oscarsson, U., Poromaa, I.S., Nussler, E., Lofg ren, M. (2006). No difference in length of hospital stay between lapa roscopic and abdominal supravaginal hysterectomy--a preliminary study. Acta Obstetricia et Gynecologica Scandinavica, 85, 682-687. Padgett, D.A., Marucha, P.T., Sheridan, J.F. (1998). Restraint stress slows cutaneous wound healing in mice. Brain, Behavior, and Immunity 12 64-73.

PAGE 159

159 Pearson, S., Maddern, G.J., Fitri dge, R. (2005). The role of pre-operative state-anxiety in the determination of intra-operative ne uroendocrine responses and recovery. British Journal of Health Psychology, 10, 299-310. Penedo, F.J., Gonzales, J.S., Dahn, J.R., Antoni, M., Malow, R., Costa, P., et al. (2003). Personality, quality of life and HAART a dherence among men and women living with HIV/AIDS. Journal of Psychosomatic Research, 54, 271-278. Pereira, D.B., Antoni, M.H., Danielson, A., Simo n, T., Efantis-Potter, J., Carver, C.S., et al. (2003a). Stress as a rpedictor of symptomatic genital herpes virus recurrence in women with human immunodeficiency virus. Journal of Psychosomatic Research 54, 237-244. Pereira, D.B., Antoni, M.H., Danielson, A., Simo n, T., Efantis-Potter, J., Carver, C.S., et al. (2003b). Life stress and cervical squamous in traepithelial lesions in women with human papillomavirus and human immunodeficiency virus. Psychosomatic Medicine, 65, 427434. Pereira, D.B., Antoni, M.H., Danielson, A., Si mon, T., Efantis-Potter, J., OSullivan, M.J. (2004). Inhibited interpersonal coping style pr edicts poorer adherence to scheduled clinic visits in human immunodefici ency virus infected women at risk for cervical cancer. Annals of Behavioral Medicine 28, 195-202. Peterson, E.D., Coombs, L.P., Ferguson, T.B., Shroyer, A.L., DeLong, E.R., Grover, F.L., et al. (2002). Hospital variability in length of stay after coronary artery bypass surgery: results from the Society of Thoracic Surgeon's National Cardiac Database. Annals of Thoracic Surgery, 74, 464-473. Pinninti, N., Steer, R.A., Rissmiller, D.J., Nelson, S., Beck, A.T. (2002). Use of the Beck Scale For suicide ideation with ps ychiatric inpatients dia gnosed with schizophrenia, schizoaffective, or bipolar disorders. Behavior Research and Therapy, 40, 1071-1079. Rogosa, D., Saner, H. (1995). Longitudinal data analysis examples with random coefficient models. Journal of Educational and Behavioral Statistics 20 994. Salmon, J.E., Edberg, J.D., Brogle, N.L., Kimberly, R.P. (1992). Allelic polymorphisms of human Fc gamma receptor IIA and Fc gamma receptor IIIB. Independent mechanisms for differences in human phagocyte function. Journal of Clinical Investigation 89, 1274-1281. Salo, M. (1992). Effects of anaesthesia and surgery on the immune response. Acta Anaesthesioligica Scandinavica. 36, 201-20. Sarason, I.G., Johnson, J.H., Siegel, J.M. ( 1978). Assessing the impact of life changes: development of the Life Experiences Survey. Journal of Consulting & Clinical Psychology, 46, 932-946. Schim p, V.L., Worley, C., Brunello, S., Levenback, C.C., Wolf, J.K., Sun, C.C., et al. (2004). Vacuum-assisted closure in the treatm ent of gynecologic oncology wound failures. Gynecologic Oncology 92, 586-591.

PAGE 160

160 Seedat, S., Stein, M.B., Kennedy, C.M., Hauger, R. L. (2003). Plasma cortisol and neuropeptide Y in female victims of intimate partner violence. Psychoneuroendocrinology, 28, 796-808. Selye, H. (1936). Syndrome produced by diverse noxious agents. Nature 138, 32. Sephton, S.E., Sapolsky, R.M., Kraemer, H.C., Spiegel, D. (2000). Diurnal cortisol rhythm as a predictor of breas t cancer survival. Journal of the National Cancer Institute, 29, 994-1000. Singh, N., Squier, C., Sivek, C., et al. (1996). Determinants of complian ce with antiretroviral therapy in patients with hu man immunodeficiency virus: Prospective assessment with implications for enhancing compliance. AIDS Care 8, 261-269. Smith, R.M., Giannoudis, Trauma and the immune response. Journal of the Royal Society of Medicine, 91 417-420. Soderstrom, M., Dolbier, C., Leiferman, J., Steinh ardt, M. (2000). The relationship of hardiness, coping strategies, and perceived stress to symptoms of illness. Journal of Behavioral Medicine, 23 311-328. Spitzer, R.L., Williams, J.B., Gibbon, M., First, M.B. (1992). The Structured Clinical Interview for DSM-III-R (SCID). I: History, rationale, and description. Archives of General Psychiatry, 49, 624-629. Stawski, R.S., Sliwinski, M.J., Almeida, D.M., Smyth, J.M. (2008). Reported exposure and emotional reactivity to daily stressors: the ro les of adult age and global perceived stress. Psychology and Aging 23, 52-61. Stewart, R.A., White, H.D., Kirby, A.C., Heritier, S.R., Simes, R.J., Nestel, P.J. et al. (2005). White blood cell count predicts reduction in coronary he art disease mortality with pravastatin. Circulation 111, 1756-1762. Stone, A.A., Schwartz, J.E., Smyth, J., Kirschbaum C., Cohen, S., Hellhammer, C., et al. (2001). Individual differences in th e diurnal cycle of salivary fr ee cortisol: A replication of flattened cycles for some individuals. Psychoneuroendocrinology 26, 295-306. Taylor, S.E., Klein, L.C., Lewis, B.P., Gruenewalk, T.L., Gurung, R.A., Updegraff, J.A. (2000). Biobehavioral responses to stress in fema les: tend-and-befriend, not fight-or-flight. Psychological Review 107 411-429. Tjemsland, L., Soreide, J.A., Matre, R., Ma lt, U.F. (1997). Pre-operative [correction of Properative] psychological variab les predict immunological status in patients with operable breast cancer. Psychooncology 6, 311-320. Tom linson, J.W., Finney, J., Hughes, B.A., Stew art, P.M. (2008). Reduced glucocorticoid production rate, decreased 5alpha-reducta se activity, and adip ose tissue insulin sensitization after weight loss. Diabetes 57, 1536-1543.

PAGE 161

161 Touitou, Y., Bogdan, A., Levi, F., Benavides, M. Auzeby, A. (1996). Disruption of the circadian patterns of serum cortisol in breast and ovarian cancer patients: relationships with tumour marker antigens. British Journal of Cancer 74, 1248-1252. Tozzi, R., Malur, S., Koehler, C., Schneider, A. (2005). Analysis of morbidity in patients with endometrial cancer: Is there a co mmitment to offer laparoscopy? Gynecologic Oncology 97, 4-9. Turner, R.J., Avison, W.R. (2003). Status variations in stress exposure: Implications for the interpretation of research on race, socioeconomic status, and gender. Journal of Health and Social Behavior 44, 488-505. Turner-Cobb, J.M., Koopman, C., Rabinowitz, J.D., Terr, A.I., Se phton, S.E., Spiegel, D. (2004). The interaction of social network size and stressful life events predict delayed-type hypersensitivity among women with metastatic breast cancer. International Journal of Psychophysiology 54, 241-249. Uchino, B.N. (2006). Social support and health: a review of physio logical processes potentially underlying links to disease outcomes. Journal of Behavioral Medicine 29 377-387. Van Cauter, E., Leproult, R., Kupfer, D.J. (1996) Effects of gender and age on the levels and circadian rhythmicity of plasma cortisol. Journal of Clinical Endocrinology and Metabolism 81, 2468-2473. Vasilev, S.A. (Ed.) (2000). Perioperative and supportive care in gynecologic oncology: Evidence-based management New York, NY: Wiley-Liss. Vedhara, K., Stra, J.T., Miles, J.N., Sanderma n, R., Ranchor A.V. (2006). Psychosocial factors associated with indices of cortisol producti on in women with breast cancer and controls. Psychoneuroendocrinology, 31, 299-311. Walch, J.M., Rabin, B.S., Day, R., Williams, J.N., Choi, K., Kang, J.D. (2005). The effect of sunlight on postoperative analgesic medica tion use: a prospective study of patients undergoing spinal surgery. Psychosomatic Medicine, 67, 156-163. Warren, O., Alexiou, C., Massey, R., Leff, D., Purkay astha, S., Kinross, J. et al. (2007). The effects of various leukocyte filtration strategies in cardiac surgery. European Journal of Cardiothoracic Surgery 31, 665-676. Wells, K., Jensen, S., Dodd, S., Hosonitz, M., Ripley, D., Morgan, L., et al. (2006 April). Association between diabetes, psychosocial status, and post-surgical complications in women with endometrial cancer. Poster pres entation at the Society of Behavioral Medicines 27th Annual Meet ing, San Francisco, CA. Yang, H.C., Thornton, L.M., Shapiro, C.L., A ndersen, B.L. (2008). Surviving recurrence: psychological and quality-of-life recovery. Cancer 112, 1178-1187.

PAGE 162

162 Yehuda, R., Teicher, M.H., Trestman, R.L., Levengood, R.A., Siever, L.J. (1996). Cortisol regulation in posttraumatic st ress disorder and major depr ession: a chronobiological analysis. Biological Psychiatry 40, 79-88.

PAGE 163

163 BIOGRAPHICAL SKETCH Sally Elizabeth Jensen w as born in 1978, in Waupun, Wisconsin. She remained in Wisconsin until her graduation from Waupun High School in 1996. She then enrolled in the University of Minnesota in Minneapolis. In 2000, she graduated summa cum laude from the University of Minnesota, where she received a Bachelor of Arts degree in psychology and spanish. In 2000, she was inducted as a member of the Phi Beta Kappa honor society. Ms. Jensen was then employed for three years as a Community Program Specialist in the University of Minnesota, Division of Epidemiology, where sh e worked on two community-based healthpromotion research studies funded by the Nationa l Institutes of Health (NIH) involving the evaluation of behavioral strategies to prevent osteoporosis in girls and the evaluation of the effect of low-fat food availability in hi gh schools on students food choices. In 2003, Ms. Jensen enrolled in the doctoral pr ogram in the Department of Clinical and Health Psychology at the University of Florida in Gainesville, Fl orida. Under the mentorship of Deidre Pereira, Ph.D., Ms. Jensen worked as a gr aduate research assistan t conducting research at the interface of psycho-oncology, psychoneuroi mmunology, and womens health. In 2005, she received the American Psychosomatic Soci ety Young Scholar Award and was awarded a Citation Poster for her project titled, Cognitive Behavioral Stress Management (CBSM) Effects on Social Support and Positive Affect among HIV+ Women at Risk for Cervical Cancer. In 2006, Ms. Jensen received the University of Florida College of Public Health and Health Professions Outstanding Research Award for he r project entitled, Maladaptive Interpersonal Coping Predicts Poorer Surgical Recovery Among Women with Endometrial Cancer. She received the Clinical Health Psychology Research Award from the Depart ment of Clinical and Health Psychology at the Universi ty of Florida in 2007. In June 2007, Ms. Jensen began her oneyear pre-doctoral internship at the Vanderb ilt-VA Internship Consortium in Nashville,

PAGE 164

164 Tennessee. After completing her pre-doctoral in ternship, Ms. Jensen began a postdoctoral fellowship in psychosocial oncology at NorthSho re University HealthSystem (Kellogg Cancer Care Center and the Center on Outcomes, Research and Educa tion) in Evanston, Illinois.