UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Can eating and body attitudes affect physiological health outcomes in premenopausal women? Prospective.. Bedford, Jennifer Lynn 2010-12-31

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
[if-you-see-this-DO-NOT-CLICK]
[if-you-see-this-DO-NOT-CLICK]
ubc_2010_spring_bedford_jennifer.pdf [ 3.63MB ]
Metadata
JSON: 1.0070924.json
JSON-LD: 1.0070924+ld.json
RDF/XML (Pretty): 1.0070924.xml
RDF/JSON: 1.0070924+rdf.json
Turtle: 1.0070924+rdf-turtle.txt
N-Triples: 1.0070924+rdf-ntriples.txt
Original Record: 1.0070924 +original-record.json
Full Text
1.0070924.txt
Citation
1.0070924.ris

Full Text

CAN EATING AND BODY ATTITUDES AFFECT PHYSIOLOGICAL HEALTH OUTCOMES IN PREMENOPAUSAL WOMEN? PROSPECTIVE 2-YEAR CHANGES IN BONE, AND RELATIONSHIPS WITH OVULATION, CORTISOL, AND BLOOD PRESSURE    by     Jennifer Lynn Bedford BSNH, Acadia University, 2003       A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY    in     THE FACULTY OF GRADUATE STUDIES   (Human Nutrition)         THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)  April 2010      © Jennifer Lynn Bedford, 2010     ii ABSTRACT Cognitive dietary restraint (CDR) is the perception that one is limiting food intake in an effort to achieve/maintain a perceived ideal body weight. Cross-sectional studies suggest CDR is associated with an increased frequency of subclinical ovulatory disturbances (%SOD; anovulation and luteal phase <10 days long) and lower bone mass, possibly mediated by cortisol, a stress hormone. This research was conducted to prospectively examine relationships among CDR, %SOD, 24-hour urinary free cortisol (UFC) and 2-year areal bone mineral density change (ΔaBMD) in non-obese, regularly-menstruating women, aged 19-35. To monitor %SOD, least-squares quantitative basal temperature (LS-QBT) analysis was used. LS-QBT was first further validated against urinary pregnanediol glucuronide (PdG), an indirect indicator of ovulation (n=40, Chapter 2). Relative to PdG, LS-QBT showed excellent detection of ovulatory cycles (97%) but poor detection of anovulatory cycles (25%). Estimated day of luteal onset was correlated between methods (r=0.8, P<0.001). Chapter 3 presents prospective findings (n=123). Women with higher CDR had higher %SOD (56% versus 34%, P<0.001) and higher UFC (28.0 µg/day versus 24.0 µg/day, P=0.021). ΔaBMD did not differ by CDR level. Women with higher %SOD had less positive lumbar spine (L1-4; 0.7% versus 1.9%, P=0.034) and hip (-0.6% versus 0.9%, P=0.001) ΔaBMD, and higher CDR scores (8.7 versus 7.1, P=0.04). UFC was not associated with %SOD or ΔaBMD. Whether eating/body attitudes (EBA) were associated with 12-hour daytime ambulatory blood pressure (ABP) was explored as a secondary objective (n=120, Chapter 4). Women with negative EBA had higher diastolic ABP and mean arterial pressure, independently of weight loss effort. Finally, at baseline (n=137, Chapter 5), UFC was inversely associated with total body bone mineral content (BMC; r= -0.30, P<0.001) and aBMD (r= -0.27, P=0.003); L1-4 aBMD (r= -0.19, P=0.035) and BMC (r= -0.18, P=0.049); and hip BMC (r= -0.23, P=0.011), after adjustment for potential confounders. In summary, findings suggest CDR and other negative EBA may be associated with adverse health outcomes including higher ABP and %SOD. Furthermore, more frequent SOD, which are not apparent to women, were associated with less positive ΔaBMD. However, cortisol may not be the only or most important mediator of these relationships.      iii TABLE OF CONTENTS ABSTRACT ................................................................................................................................ ii TABLE OF CONTENTS ............................................................................................................ iii LIST OF TABLES ...................................................................................................................... ix LIST OF FIGURES .................................................................................................................... xi LIST OF ABBREVIATIONS ...................................................................................................... xii PREFACE ................................................................................................................................ xiv ACKNOWLEDGEMENTS ......................................................................................................... xv CO-AUTHORSHIP STATEMENT ........................................................................................... xvii Chapter 1: Introduction ............................................................................................................ 1 1.1 Background and rationale ................................................................................................. 2 1.2 Literature review ............................................................................................................... 5 1.2.1 Introduction ................................................................................................................. 5 1.2.2 Cognitive dietary restraint ........................................................................................... 5 1.2.2.1 Background and assessment ............................................................................... 5 1.2.2.2 Behavioural versus perceptual aspects of CDR ................................................... 8 1.2.3 Stress, cortisol and CDR ........................................................................................... 10 1.2.3.1 Chronic stress and neuroendocrine function ...................................................... 10 1.2.3.2 Assessment of cortisol and general stress perception ....................................... 11 1.2.3.3 Cortisol and CDR ............................................................................................... 13 1.2.3.4 Perception of psychosocial stress and CDR ...................................................... 15 1.2.4 Ovulatory function ..................................................................................................... 15 1.2.4.1 Physiology of the menstrual cycle ...................................................................... 15 1.2.4.2 Disturbances in ovulatory function ..................................................................... 16 1.2.4.3 Monitoring ovulatory function ............................................................................. 16 1.2.4.4 Ovulatory function and the HPA axis ................................................................. 18 1.2.4.5 Ovulatory disturbances and CDR ....................................................................... 18 1.2.5 Bone and cortisol ...................................................................................................... 21 1.2.6 Bone and ovulatory function ..................................................................................... 24 1.2.7 Bone and CDR ......................................................................................................... 26 1.3 Gaps in our current understanding .................................................................................. 28 1.4 Study purpose ................................................................................................................. 29 1.4.1 Objectives ................................................................................................................. 30     iv 1.4.1.1 Objectives for Chapter 2 .................................................................................... 30 1.4.1.2 Objectives for Chapter 3 .................................................................................... 30 1.4.1.3 Objectives for Chapter 4 .................................................................................... 31 1.4.1.4 Objectives for Chapter 5 .................................................................................... 31 1.4.2 Hypotheses ............................................................................................................... 32 1.4.2.1 Hypotheses for Chapter 2 .................................................................................. 32 1.4.2.2 Hypotheses for Chapter 3 .................................................................................. 32 1.4.2.3 Hypotheses for Chapter 4 .................................................................................. 33 1.4.2.4 Hypotheses for Chapter 5 .................................................................................. 33 1.5 References ..................................................................................................................... 34 Chapter 2: Detecting evidence of luteal activity by least-squares quantitative           basal temperature analysis against urinary progesterone metabolites          and the effect of wake-time variability ................................................................ 49 2.1 Introduction ..................................................................................................................... 50 2.2 Methods .......................................................................................................................... 51 2.2.1 Participants ............................................................................................................... 51 2.2.2 Procedures ............................................................................................................... 51 2.2.3 Determination of evidence of luteal activity ............................................................... 52 2.2.3.1 Urinalysis ........................................................................................................... 52 2.2.3.2 Basal temperature analysis ................................................................................ 52 2.2.3.3 Statistical analyses ............................................................................................ 53 2.3 Results ............................................................................................................................ 54 2.3.1 Participant characteristics ......................................................................................... 54 2.3.2 Sensitivity, specificity, predictive values and accuracy .............................................. 54 2.3.3 Correlation of luteal onset: sustained PdG rise versus LS-QBT temperature          increase .................................................................................................................... 56 2.4 Discussion ...................................................................................................................... 58 2.5 References ..................................................................................................................... 61 Chapter 3: A prospective exploration of cognitive dietary restraint, subclinical          ovulatory disturbances, cortisol and change in bone density over two          years in healthy young women ........................................................................... 64 3.1 Introduction ..................................................................................................................... 65 3.2 Methods .......................................................................................................................... 66 3.2.1 Participants ............................................................................................................... 66 3.2.2 Data collection .......................................................................................................... 68 3.2.3 Questionnaires ......................................................................................................... 68 3.2.4 Food frequency questionnaire (FFQ) ........................................................................ 69     v 3.2.5 Ovulatory function ..................................................................................................... 69 3.2.6 Urine collection and analyses ................................................................................... 70 3.2.7 Physical measurements ............................................................................................ 70 3.2.8 Statistics ................................................................................................................... 70 3.3 Results ............................................................................................................................ 71 3.3.1 Sample ..................................................................................................................... 71 3.3.2 Questionnaires ......................................................................................................... 72 3.3.3 Urine volume and UFC ............................................................................................. 72 3.3.4 Menstrual cycle and ovulatory function ..................................................................... 73 3.3.5 Physical measurements ............................................................................................ 74 3.3.6 Differences by CDR median split .............................................................................. 75 3.3.7 Differences by subclinical ovulatory disturbances median split ................................. 77 3.4 Discussion ...................................................................................................................... 79 3.5 References ..................................................................................................................... 84 Chapter 4: Negative eating and body attitudes are associated with higher          daytime ambulatory blood pressure in healthy young women ........................ 88 4.1 Introduction ..................................................................................................................... 89 4.2 Methods .......................................................................................................................... 91 4.2.1 Participants ............................................................................................................... 91 4.2.2 Procedure ................................................................................................................. 91 4.2.3 Questionnaires ......................................................................................................... 92 4.2.3.1 Eating and body attitudes .................................................................................. 92 4.2.3.2 General stress ................................................................................................... 93 4.2.3.3 Weight loss effort ............................................................................................... 93 4.2.4 Urine analysis ........................................................................................................... 93 4.2.5 ABP measurement .................................................................................................... 93 4.2.6 Statistical analyses ................................................................................................... 94 4.3 Results ............................................................................................................................ 95 4.3.1 Participant characteristics ......................................................................................... 95 4.3.2 Correlation analyses ................................................................................................. 95 4.3.3 Differences by Eating/Body Attitudes and weight loss effort ...................................... 96 4.4 Discussion ...................................................................................................................... 99 4.5 References ................................................................................................................... 102 Chapter 5: The relationship between 24-hour urinary cortisol and bone in healthy          young women ..................................................................................................... 108     vi 5.1 Introduction ................................................................................................................... 109 5.2 Methods ........................................................................................................................ 109 5.2.1 Participants ............................................................................................................. 109 5.2.2 Questionnaires ....................................................................................................... 110 5.2.3 Dietary intake .......................................................................................................... 111 5.2.4 Urine collection and analysis .................................................................................. 111 5.2.5 Anthropometrics and body composition .................................................................. 111 5.2.6 Statistical analyses ................................................................................................. 112 5.3 Results .......................................................................................................................... 112 5.3.1 Participant characteristics ....................................................................................... 112 5.3.2 Associations with aBMD, BMC and bone area ........................................................ 114 5.3.3 Associations with 24-hour urinary free cortisol ........................................................ 114 5.3.4 Associations with PSS score .................................................................................. 115 5.3.5 Associations between 24-hour urinary free cortisol and aBMD, BMC          and bone area ........................................................................................................ 115 5.4 Discussion .................................................................................................................... 117 5.5 References ................................................................................................................... 121 Chapter 6: Conclusion .......................................................................................................... 126 6.1 General conclusion ....................................................................................................... 127 6.2 General discussion ....................................................................................................... 130 6.3 Strengths and limitations ............................................................................................... 137 6.4 Future directions ........................................................................................................... 142 6.5 References ................................................................................................................... 145 Appendix 1: Recruitment Materials for Quantitative Basal Temperature (QBT)             Validation Study ............................................................................................... 151 Appendix 2: QBT Validation Study Letter of Initial Contact (via email) ............................. 153 Appendix 3: QBT Validation Study Eligibility Phone Script ............................................... 155 Appendix 4: QBT Validation Study Questionnaires ............................................................ 158 Appendix 5: QBT Validation Study Temperature Calendar ................................................ 162 Appendix 6: QBT Validation Study Instructions for Daily Urine Sample Collection ........ 164 Appendix 7: QBT Validation Study Ethics Approval Certificate ........................................ 169 Appendix 8: QBT Validation Study Letter of Consent ........................................................ 170 Appendix 9: QBT Validation Study Transportation Reimbursement Receipt ................... 174     vii Appendix 10: QBT Validation Study Gift Card Receipt....................................................... 175 Appendix 11: QBT Validation Study Individual Results ..................................................... 176 Appendix 12: Sensitivity and specificity of least-squares quantitative basal               temperature analysis (LS-QBT) methods in determining luteal              phase length (LPL) relative to Kassam‟s urinary pregnanediol              glucuronide (PdG) algorithm (n=35) ............................................................. 181 Appendix 13: Recruitment Materials for 2-year Prospective Bone Study ......................... 183 Appendix 14: 2-year Prospective Bone Study Letter of Initial Contact (via email) ........... 185 Appendix 15: 2-year Prospective Bone Study Eligibility Phone Script ............................. 188 Appendix 16: 2-year Prospective Bone Study Letter of Consent ...................................... 192 Appendix 17: 2-year Prospective Bone Study Ethics Approval Certificate ...................... 197 Appendix 18: 2-year Prospective Bone Study Transportation Reimbursement              Receipt ............................................................................................................ 198 Appendix 19: 2-year Prospective Bone Study Gift Card Receipt ....................................... 199 Appendix 20: 2-year Prospective Bone Study 24-hour Urine Collection              Instructions .................................................................................................... 200 Appendix 21: 2-year Prospective Bone Study Temperature Calendar .............................. 202 Appendix 22: 2-year Prospective Bone Study Bone Density Scan Instructions .............. 204 Appendix 23: 2-year Prospective Bone Study Annual Questionnaire Package ............... 206 Appendix 24: List of Validated Questionnaires Included in Annual Questionnaire              Package .......................................................................................................... 236 Appendix 25: 2-year Prospective Bone Study Daily Stress Inventory .............................. 237 Appendix 26: 2-year Prospective Bone Study 12-hour Ambulatory Blood              Pressure Monitoring Instructions and Diary ................................................ 240 Appendix 27: Correlations of Cognitive Dietary Restraint and Subclinical              Ovarian Disturbances with General Stress Questionnaires ....................... 243 Appendix 28: Comparison of Least-squares Basal Temperature Analysis Method              Relative to Other Non-invasive Methods to Detect Ovulation                       Regarding Cost, Participant Acceptability, Ease-of-use and              Accuracy in Detecting the Day of Luteal Onset ........................................... 245 Appendix 29: Partial Correlations of 24-hour Urinary Free Cortisol (UFC) and              Average Percieved Stress Scale (PSS) Scores with Questionnaire              Scores ............................................................................................................. 251     viii Appendix 30: Cognitive Dietary Restraint Score, General Stress Score,              Subclinical Ovulatory Disturbances, 24-hour Urinary Free Cortisol              and 2-year ΔaBMD by Ethnicity (n=123) ....................................................... 254 Appendix 31: Cross-sectional Examination of Differences in 24-hour Urinary              Free Cortisol by Ethnicity and Level of Cognitive Dietary Restraint              (CDR) and the Ethnicity-by-CDR Interaction ................................................ 255 Appendix 32: Pearson‟s Partial Correlations of 12-hour Average Daytime              Ambulatory Blood Pressure (ABP, mm Hg) and Eating and Body              Attitude Questionnaire Scores At First Follow-up (n=120) ......................... 256 Appendix 33: Email Correspondence with Participants of the 2-year Prospective              Bone Study ..................................................................................................... 257 Appendix 32: 2-year Prospective Bone Study Temperature Calendar Individual              Results ............................................................................................................ 266 Appendix 35: 2-year Prospective Bone Study Ambulatory Blood Pressure              Individual Results .......................................................................................... 268 Appendix 36: 2-year Prospective Bone Study Letter Accompanying Bone              Density Results .............................................................................................. 270 Appendix 37: Pearson‟s Correlations Between the Duration of Hormone Use and              2-year ΔaBMD (n=123) ................................................................................... 274 Appendix 38: 24-hour Urinary Free Cortisol at Baseline and Follow-ups and              Level of Significant Difference Between Values by Repeated              Measures General Linear Model (n=116) ...................................................... 275     ix LIST OF TABLES Table 2.1  Descriptive characteristics of the sample (n=40) ................................................... 54 Table 2.2  Sensitivity and specificity of least-squares quantitative basal temperature analysis (LS-QBT) methods in determining evidence of luteal activity (ELA) relative to Kassam‘s urinary pregnanediol glucuronide (PdG) algorithm .............................................................................................................. 55 Table 2.3  Predictive value and accuracy of least-squares quantitative basal temperature analysis (LS-QBT) methods in determining evidence of luteal activity relative to Kassam‘s urinary pregnanediol glucuronide (PdG) algorithm .............................................................................................................. 55 Table 3.1  Mean questionnaire scores and energy intakes, and partial correlation coefficients of the Three Factor Eating Questionnaire subscales and 24-hour urinary free cortisol in healthy premenopausal women (n=123) .................... 72 Table 3.2  Physical measurements at baseline, 2-year follow-up and the 2-year percent change in healthy premenopausal women (n=123) .................................. 74 Table 3.3  Differences between healthy premenopausal women with higher and lower cognitive dietary restraint (by median split) in baseline anthropometrics, Δanthropometrics questionnaire scores, energy intakes, menstrual cycle characteristics, 24-hour urinary free cortisol and 2-year ΔaBMD (n=123) .................................................................................................... 75 Table 3.4  Differences between healthy premenopausal women with higher and lower percentage of cycles with subclinical ovulatory disturbances (median split) in menstrual cycle characteristics, age, anthropometrics, Δanthropometrics, questionnaire scores, 24-hour urinary free cortisol and 2-year ΔaBMD (n=114) ......................................................................................... 78 Table 4.1   Mean age, body mass index, questionnaire scores, energy intakes, 24-h urinary free cortisol (UFC) and 12-h daytime ambulatory blood pressure; and adjusted correlates of Eating/Body Attitude Z-score, General Stress Z-score, and UFC in healthy premenopausal women (n=120) .............................. 96 Table 4.2   Main and interactive effect of Eating/Body Attitude level and weight loss effort on age, body mass index, questionnaire scores, energy intakes, 24-h urinary free cortisol, and 12-h daytime ambulatory blood pressure (n=120). ................................................................................................................ 98 Table 5.1  Physical activity and questionnaire scores, reported nutrient intakes, 24-hour urinary free cortisol excretion, anthropometrics and DXA measurements for all participants and differences by ethnicity ........................... 113 Table 5.2  Correlations of aBMD, BMC and bone area with anthropometrics, perceived stress, physical activity, duration of previous oral contraceptive use, calcium/kcal intake, 24-hour urinary free cortisol excretion and 24-hour urine volume ............................................................................................... 116 Table 5.3  Partial correlation models of the relationship between aBMD, BMC and bone area and 24-hour urinary free cortisol excretion ......................................... 117     x Table 6.1  Summary of results with regard to specific hypotheses ...................................... 127       xi LIST OF FIGURES Figure 1.1 Hypothesis guiding this PhD research programme juxtaposition with the physiological stress response ................................................................................. 4 Figure 2.1  Correlation of the day of LS-QBT temperature rise versus day of sustained PdG rise by Kassam algorithm: All temperatures .................................. 56 Figure 2.2  Correlation of the day of LS-QBT temperature rise versus day of sustained PdG rise by Kassam algorithm: Royston wake-time adjusted ............... 57 Figure 2.3  Correlation of the day of LS-QBT temperature rise versus day of sustained PdG rise by Kassam algorithm: 2-hour average wake time temperatures ........................................................................................................ 57 Figure 2.4  Correlation of the day of LS-QBT temperature rise versus day of sustained PdG rise by Kassam algorithm: Expert reviewed temperatures ............ 58 Figure 3.1  Model driving our hypothesis of cognitive dietary restraint and bone density juxtaposition with the physiological stress response ................................. 66 Figure 3.2  Flow diagram depicting study recruitment, participation and data collection at baseline and first and final follow-up assessments ........................................... 67     xii LIST OF ABBREVIATIONS aBMD: areal bone mineral density (g/cm2) ACTH: adrenocorticotropic hormone ABP: 12-hour daytime mean ambulatory blood pressure (mm Hg) ABP-activity:  continuous score for concurrent activity during ABP (sum of diary codes for each reading ABP [0=sedentary, 1=active] divided by total number of readings per participant) BMC: bone mineral content (g) BMD: bone mineral density (g/cm2) BMI: body mass index (kg/m2) BP: blood pressure BUA: broadband ultrasonic attenuation CDR: cognitive dietary restraint (equivalent to dietary restraint) CRH: corticotropin releasing hormone CVD: cardiovascular disease DEBQ: Dutch Eating Behaviour Questionnaire DEBQ-R: DEBQ Restraint subscale DHQ: Diet History Questionnaire DLT: day of luteal transition DSI: Daily Stress Inventory DXA: dual energy X-ray absorptiometry EBA: eating and body attitudes EDE: Eating Disorder Exam EDI: Eating Disorder Inventory-2 ELA: evidence of luteal activity FFQ: Food Frequency Questionnaire FHA: functional hypothalamic amenorrhea FSH: follicle stimulating hormone GLM: General Linear Model GnRH: gonadotropin-releasing hormone HPA: hypothalamic-pituitary-adrenal HPG: hypothalamic-pituitary-gonadal kcal: kilocalorie L1-4: lumbar spine vertebrae 1 through 4, inclusive LH: luteinizing hormone LPL: luteal phase length LS-QBT: least-squares method of quantitative of basal temperature analysis  mm Hg: millimetres of mercury n: number of participants NPV: negative predictive value p: page P: probability of obtaining a test statistic at least as extreme as the one that was actually observed, assuming that the null hypothesis is true PdG: pregnanediol glucuronide PPV: positive predictive value  PSS: Perceived Stress Scale QBT: quantitative methods of basal temperature analysis QCT: quantitative computed tomography r: Pearson‘s correlation coefficient  REE: resting energy expenditure RS: Restraint Scale SD: standard deviation SOD: subclinical ovulatory disturbances     xiii SOS: speed of sound SPSS: Statistical Package for the Social Sciences TFEQ: Three Factor Eating Questionnaire or Eating Inventory TFEQ-R: TFEQ Restraint subscale (questionnaire to assess CDR) UFC: 24-hour urinary free cortisol (µg/day) VGH: Vancouver General Hospital Z-score: how many standard deviation units away from the mean a particular value of data lies      xiv PREFACE  I prepared this dissertation according to the University of British Columbia Faculty of Graduate Studies requirements for a manuscript-based thesis. Therefore, Chapters 2 to 5 are elaborated versions of manuscripts that have been published, accepted for publication or submitted for publication to scientific journals. Although some overlap may occur, each chapter is designed to stand alone, and these chapters can be read in any order. Chapter 2 includes a validation study that was conducted prior to commencement of the main study. The findings of the main research question (two year prospective data) are presented in Chapter 3. Chapters 4 and 5 include data collected at the first follow-up and baseline, respectively, as secondary objectives. Additional analyses that were not included in the manuscripts due to space constraints are presented as appendices.    xv ACKNOWLEDGEMENTS I could not have completed this research project and the resulting manuscripts and dissertation without the support and hard work of many others. I would like to take this time to formally acknowledge their contributions to the project and my personal experience.  First and foremost, I would like to thank my supervisor, Dr. Susan Barr. Susan, I have no words to adequately express what your unwavering support, guidance and dedication to your students and integrity as a scientist have meant to my professional development. I am the scientist that stands before you today because of you. Your humour, kindness and friendship have made the past six years an enjoyable experience. Beyond the work, your generosity and genuine concern about me as a person has touched me to the very depths of my soul. Thank you Susan. I look forward to a continued friendship and future collaborations.  Completion of this project would not have been possible without the 140 women who earnestly contributed their time and effort to my study. It was a pleasure to meet each of you and to be a part of your lives for those two years. Thank you to the Canadian Institute of Health Research who provided both the project operating grant and supported me personally for three years. Thanks also to funders of my other scholarships including the Michael Smith Foundation for Health Research, the National Science and Engineering Research Council, and the University of British Columbia particularly the Faculty of Land and Food Systems.   I am appreciative of the contributions of my supervisory committee members Drs. Jerilynn Prior, Wolfgang Linden and Kathy Keiver. Thank you for providing me the opportunity to learn a small piece of your area of expertise under your guidance and support. As well, thank you to Dr. Christine Hitchcock for your assistance with the Maximina program and the QBT validation paper. You have helped improve my precision while writing and have taught me to think critically about what I say and what I mean to say.   To my mentors Oonagh Holmes, Dr. Shanthi Johnson, Dr. Elizabeth Johnston and Debbie Zibrik. Your guidance, encouragement and support have helped steer me to where I am today. Thank you for sharing your experiences with me.   Thank you to the wonderful women (and their staff) that conducted the various analyses for the project including Ellie Brindle, Romi Chan, Darlene Christopher and Nazneen. Also thanks to the staff and faculty of the Food, Nutrition and Health department particularly Tram Nguyen, Patrick Leung and Karol Traviss.  My deepest gratitude to my research assistant Amandeep Ghuman for her hard work and dedication to the project. Anu, only you could make inputting over 1500 temperature calendars fun. Beyond the project, thank you for being such a sincere friend and for adopting me into your wonderful family.   From the bottom of heart, I thank my extended family including all my Aunts, Uncles and cousins. Your encouragement and love have made it possible for me to get through the past decade of university. Thank you Gramma for all of your cross-country visits and hand written letters. I am so grateful that you taught me to knit and bake as it has allowed me to be more balanced during this process. Deepest thanks to my Aunt Carrol and Uncle Jeff for allowing me to stay with you in Arizona and giving me the space to breathe – it saved my life. Thank you to my Uncle Phil for pointing me in the right direction and for your reassuring emails over the years. It was always nice to hear from someone that truly understands the academic struggle. Uncle David and Michelle, thank you for making the transition to BC easier and for welcoming me and Andrew into your family for those few years. To my friends that are like family, Jessica     xvi Kelly and Brian Milroy. Thank you for your friendship regardless of the distance between us over the past decade.  And last, but certainly not least, my immediate family.   Andrew Nichols, I could not have made it through a single day of this process without you beside me. Thank you for reminding me to breathe; for keeping me warm; for rubbing my back, massaging my head and wiping my tears. Thank you for loving me. Thank you for tolerating more mood swings than I‘m sure you ever thought were humanly possible. Thank you for moving from Nova Scotia to British Columbia so that I could achieve my goals. Thank you for believing in me: my potential and my abilities. Thank you for picking me up from the floor when I could no longer go on and for carrying me through. This degree and dissertation were truly a teamwork effort – I hope that you are proud of our accomplishments. I promise to be saner now that this is finished. I look forward to new adventures together.  Melissa, my sister and best friend. Thank you for allowing me to be who I am, and for understanding that like no one else. Thank you for looking up to me when we were kids- knowing that you were watching my every act, made me want to be the best I could. I am sorry it took me so many years (and fights!) to realize it. Thank you for being silly with me, for truly listening and caring, and for taking on ‗the big sister‘ responsibilities at home when I was unable. Most of all, thank you for moving to Vancouver-it has changed my life. To be able to do all those small sister things so easily is truly the best gift I have ever received.   To my Dad. Thank you for always encouraging my child-like enthusiasm, for wanting me to be happy, for giving me the space to be a silly-carefree-redneck, for bragging about me to near strangers, and for teaching me things beyond the textbook. Your enthusiasm and hard work on my early science projects sparked my passion for research! Thank you for working over-time in the heat and the cold so that I could spend a decade in school without having to eat Kraft Dinner once. Thank you for protecting me. Even though I am Dr. Bedford now, I will always be your little girl.  And finally, to my Mum, for making all of this possible. Every paper I have published, every scholarship I received, every ‗A‘ on my transcript is because of YOU. Because you told me it was better to be smart than pretty when the kids teased me for being a ‗browner‘; because you made up tests for me to complete during the summer when I was little; because you read to me, bought me thousands of books (Baby Sitters Club!) and allowed me to become lost in the stories; because you helped me study for high school tests when you got home from work at 9 o‘clock at night; because you were willing to proof-read everything I ever wrote (including this entire dissertation!); because you drove me everywhere, signed me up for everything and opened every door you could for me; because you taught me how to work hard, be a kind person, manage my time and stay focused. I had the courage and strength to finish this because you encouraged me and you believed in me when I could no longer believe in myself. You are my one and only bosom friend – a true kindred spirit. Although it is far from adequate, I dedicate this dissertation to you with all my love.     xvii CO-AUTHORSHIP STATEMENT  Chapters 2 through 5 are manuscripts that have been published (Chapter 2), are accepted for publication (Chapter 5) or are currently under review (Chapters 3 and 4). For each manuscript, I identified the research question, conceived of the study design, recruited all participants, completed all data collection and management, planned and conducted the data analyses, presented the findings and wrote and edited the manuscript. My co-authors made significant contributions as follows. Dr Susan Barr, my research supervisor, was the Principal Investigator for the Canadian Institutes of Health Research Operating Grant that funded this work. Committee members Drs Linden and Prior were co-applicants. Drs Barr, Keiver (additional committee member), Linden and Prior contributed to study design and implementation. Dr Linden stimulated discussion of results and provided editorial input for Chapter 4. Dr Prior stimulated discussion of results and provided editorial input for Chapters 2 and 3. For Chapter 2, Dr Christine Hitchcock stimulated discussion of results and provided editorial input. For each manuscript, Dr Susan Barr contributed continuously to data collection, management and analysis, results discussion and key editorial input.      1  Chapter 1:    Introduction                           2 1.1 Background and rationale Thinness has become a well established cultural norm for women in North America [1-3]. In response to societal pressure to be thin, many women experience body image dissatisfaction [1-3]. This may lead to unhealthy behaviours such as extreme dieting and exercising [e.g. 4-9] and, in a small number, clinical eating disorders such as anorexia and bulimia [10]. For many women, behaviours do not necessarily change, but rather a disordered relationship and preoccupation with food and weight may develop. One such attitude that has emerged as being experienced by many young women [11-12] is cognitive dietary restraint. Cognitive dietary restraint (CDR) is the perception that one is limiting food intake in an effort to achieve or maintain a perceived ideal body weight [13]. Evidence suggests that CDR is perceptual in nature, reflecting habitual monitoring of food intake and body weight preoccupation, rather than a behaviour such as dieting where food intake is limited in an effort to reduce weight. For example, several studies report no difference between women with higher and lower CDR in energy intakes, relative body mass or weight change over time [12,14-15]. The experience of CDR may negatively influence young women‘s physiological health, possibly mediated by the stress response. With the experience of any psychosocial stress, the hypothalamic-pituitary-adrenal (HPA) axis is activated initiating a sequence of events that results in increased production of the stress hormone cortisol [16]. We and others have hypothesised that the habitual monitoring and preoccupation with food and body weight experienced by women with higher levels of CDR may act as a subtle but chronic stressor that is sufficient to activate the HPA axis. Indeed elevated cortisol and higher CDR are associated in some studies [17-21]. Cortisol at high levels has direct negative affects on bone density by disturbing bone turnover and calcium balance [22]. Indirectly, cortisol may adversely influence bone by disrupting the normal cyclic secretion of the reproductive hormones [23]. Furthermore, elevations in cortisol are also associated with higher blood pressure and greater accumulation of abdominal fat [24]. Whether modestly elevated, yet physiologically normal cortisol levels, such as those occurring as a result of psychosocial stressors are capable of affecting health outcomes is not yet known.  There is some evidence to suggest the possibility. First, an inverse relationship between cortisol and bone density has been observed among healthy older adults [25-28]. Furthermore, higher cortisol levels have been shown to be associated with lower bone density among clinical samples of young women with eating disorders and major depression [29-35], though not consistently [36-37]. Secondly, various life stresses are associated with infertility and evidence suggests it may be related to the physiological stress response [38]. Correspondingly, women with higher levels of CDR are more likely to report menstrual cycle irregularities [12,39-40] and to experience subclinical disturbances in ovulatory function [41-43]. Subclinical ovulatory     3 disturbances (anovulatory cycles and/or cycles with short luteal phase duration) are not apparent to women yet indicate deficiencies or imbalances of the reproductive hormones. In addition to fertility, the reproductive hormones, estradiol and progesterone, are crucial to achieving and maintaining peak bone mass in premenopausal women [44]. There is some evidence to suggest that women who experience more frequent subclinical ovulatory disturbances have reduced bone density [45-49]. However, these findings are not conclusive as others have reported no associations among subclinical ovulatory disturbances and bone [50-51]. Nevertheless, more frequent disturbances in menstrual cycle and ovulatory function represent an additional mechanism by which CDR may negatively affect young women‘s health. In fact, a direct relationship between higher CDR and reduced bone mineral content and/or bone density has been reported in some [40,49,52-54] but not all studies [41,55-56]. At the time the research described herein was proposed, no study had examined these relationships prospectively, and to date, only one prospective study has been published [49]. The cross-sectional studies [12,17,40,43,48-50,52-56] are limited by insufficient power to detect differences in bone density due to small sample sizes and the considerable inter-individual variability in both bone density and menstrual cycle and ovulatory characteristics.  In summary, several cross-sectional studies suggest that women with higher CDR are more likely to experience menstrual cycle and ovulatory disturbances and to have higher levels of cortisol than women with lower CDR. In turn, disturbances in menstrual cycle and ovulatory function and elevated cortisol have the potential to negatively impact bone density. To date, there is only one study that has prospectively examined associations among CDR, subclinical ovulatory disturbances and bone and no one study has prospectively examined these relationships in conjunction with assessment of cortisol.  Increasingly the role of psychosocial characteristics in the development of chronic disease has been recognised; however, the majority of research to date has focused on the health outcomes of middle-aged men [57]. The potential association between CDR-related stress and bone is relevant in regard to future risk of osteoporosis. This condition, characterised by low bone mass and increased bone fragility, is experienced by one in four postmenopausal women in Canada [58]. If fractures occur, osteoporosis is associated with reduced quality of life and considerably increased health care expenditures [58]. A key factor in osteoporosis prevention is thought to be achieving and maintaining peak bone mass during younger years [59]. It is therefore critical that we have a comprehensive understanding of the factors that influence young women‘s bone health.  Thus, the primary objective of this PhD research project, as depicted in Figure 1.1, was to prospectively investigate potential relationships among CDR, cortisol, subclinical ovulatory disturbances, and change in bone density in healthy premenopausal women over two years. In     4 order to conduct this research, a method of monitoring ovulatory function that was inexpensive, accurate and acceptable to women was required. Therefore, a validation study was conducted prior to commencement of the main study to do this. Additionally, associations among eating- and body attitudes, cortisol and blood pressure were explored as secondary objectives. In order to place the current study hypotheses and objectives in context of the current state of knowledge, a review of the literature will be presented that will focus on the primary purpose. For the secondary objectives, the relevant literature will be described in the introduction and discussion sections of the corresponding manuscript. Figure 1.1 Hypothesis guiding this PhD research programme juxtaposition with the physiological stress response   Variables shown as ovals were assessed in my PhD study. The dashed line (1) between CDR and chronic psychosocial stress reflects the hypothesis that CDR acts as a subtle chronic stressor capable of activating the HPA axis. Solid black lines indicate well established mechanisms of the stress response including: (2) increased secretion of cortisol, which has a direct negative effect on bone density, and (3) inhibition of the hypothalamic-pituitary-gonadal (HPG) axis resulting in deficiencies or imbalances of the reproductive hormones and therefore menstrual cycle and ovulatory disturbances. Grey lines represent hypothesised but inconclusive relationships previously reported including: (4) the possibility that subclinical ovulatory disturbances can have detrimental effects on bone density; (5) an association between higher CDR and elevated cortisol; and (6) an association between higher CDR and the occurrence of menstrual cycle and ovulatory disturbances. This leads to the primary hypothesis guiding this project (grey dashed line (7)) that CDR may result in less positive changes in bone density. ACTH, adrenocorticotropic hormone; CRH, corticotropin-releasing hormone; E, estradiol; FSH, follicle stimulating hormone; GnRH, gonadatropin-releasing hormone; P, progesterone.  ↑ cortisol  release  1  2  3  4  5  6  7  HPA-axis  activation  ↓ bone  density Cognitive  Dietary  Restraint  HPG-axis  inhibition ↑ovulatory  disturbances  Chronic  Psychosocial  Stress CRH ACTH GnRH  Pulse LH, FSH, E, P     5 1.2 Literature review 1.2.1 Introduction Several areas of literature will be reviewed in order to place my PhD research project in the context of current knowledge. Gaps in our current understanding of how women‘s eating attitudes may affect physiological health will also be highlighted. First, cognitive dietary restraint (CDR) will be discussed, including the history and assessment of CDR. The relationship between CDR and dieting will also be addressed in order to support the hypothesis that it is the experience of dietary restraint (rather than only behavioural changes) that is associated with negative health outcomes. Broadly, my research was designed to examine whether CDR is associated with cortisol and subclinical ovulatory disturbances, and the association of each of these variables with change in bone density. Therefore, neuroendocrine function will be discussed focusing on the HPA axis, cortisol and psychosocial stress, including the operationalisation of cortisol and perceived stress. The potential relationship between cortisol and CDR will be reviewed. Next, the physiology of the menstrual cycle will be reviewed including disturbances in normal function, monitoring of ovulatory function, the relationship between ovulatory function and the HPA axis and cortisol, and the relationship between CDR and ovulatory function. Lastly, factors affecting bone, specifically CDR, cortisol, and subclinical ovulatory disturbances, will be reviewed. This chapter will conclude by highlighting the gaps in our current understanding of the problem, and the purpose, hypotheses and objectives of this project.  1.2.2 Cognitive dietary restraint 1.2.2.1 Background and assessment Eating behaviour is the result of internalised multidimensional constructs that include behavioural, cognitive and affective elements [60]. The theory of dietary restraint attempts to synthesize these elements in order to understand and assess the complex picture of eating behaviour [61]. Over the past three decades the concept of CDR, its meaning, and our understanding of its potential relationship with physiological health outcomes have evolved from their original presentation. Currently in the literature, there are three unique operational definitions of dietary restraint each with a questionnaire-based assessment tool [13,62-63]. For the purposes of this research project, CDR will be defined as an attitude towards eating and food and a preoccupation with body size, shape and weight, that may or may not result in abnormal eating behaviours [64]. Women with high levels of CDR perceive that they are attempting to limit their food intake in order to achieve or maintain their ideal body weight [13], allowing cognitive processes rather than physiological systems, such as hunger and satiety, to govern eating behaviour [65]. A brief review of the history of this construct is essential to     6 understanding current operational definitions, the definition and assessment tool chosen as appropriate for my PhD research, and how this work will contribute to the current body of knowledge as to whether the experience of CDR affects women‘s physiological health.  The theory of restrained eating originally developed from studies by Schachter [66-67] and Nisbett [68] purporting to describe and explain differences in the eating behaviours of obese and normal-weight persons. Schachter and colleagues performed a series of experiments in which they found that obese individuals were more likely to respond to external than internal food cues compared to normal-weight persons [66-67,69]. When viewed from the current CDR framework, Schachter‘s work demonstrates disinhibition, a tendency to overeat when restraint is removed. Nisbett proposed the set point theory as an alternative explanation [68]. He hypothesised that the number of fat cells in the body is a physiological set point that the body will defend and when individuals fall below this weight, they will be more responsive to external cues [68].   The set point theory was extended to normal-weight persons by Herman and coworkers [62,70] who suggested that a subgroup of normal-weight individuals may have obese set points but restrain their eating to maintain a lower weight. In one of their experiments normal-weight and obese college-aged women completed a questionnaire to measure restraint and were randomly assigned to receive zero, one or two servings of a milkshake preload [70]. As the size of the milkshake preload increased, low restraint eaters consumed less ice cream, while those with high restraint consumed larger quantities of ice cream as the size of the preload increased [70]. Termed ―counter-regulation‖, it was suggested that external cues trigger additional eating when restraint is removed [70]. Subsequent studies confirmed that anxiety [62], depression [71] and alcohol [72-73] could also cause counter-regulation or disinhibition among restrained eaters.  From this work, the 10-item Restraint Scale (RS) was developed consisting of two subscales, Weight Fluctuations and Concern for Dieting [62,70,74]. The RS was the most widely used psychometric tool to operationalise dietary restraint [75]. The current version of the RS has moderate internal consistency with Cronbach alpha scores ≥0.75 [75]. However, the predictive and construct validity of the RS was questioned following repeated observation of an association between RS score and severity of overweight [76-79]. As well, evidence suggested that the factorial composition of the RS differed between obese and normal-weight populations [60]. Subsequently, psychometric studies found that the RS was strongly correlated with weight fluctuation [76,80-83], which may have more to do with obesity than restraint. The RS was also correlated with scores of social desirability scales among obese but not normal-weight individuals [81,84].     7 Over time, it was also observed that not all individuals with high dietary restraint exhibit disinhibition and these constructs are confounded within the RS [85]. Therefore, to truly understand eating behaviour, disinhibition would need to be assessed separately. Stunkard and Messick [13] developed a series of questions based on the RS [62], Pudel‘s Latent Obesity Questionnaire [86] and clinical experience resulting in the 51-item Three Factor Eating Questionnaire (TFEQ) or Eating Inventory. The TFEQ assesses three unique aspects of eating attitudes that may influence eating behaviour: Restraint, which indicates the level of cognitive control of eating behaviour (TFEQ-R); Disinhibition, which assesses the tendency to overeat when restraint is removed; and Hunger, which measures the susceptibility to hunger and food cravings. Since then, many studies support the separation of disinhibition and restraint [60, 86-90]. Around the same time, the 33-item Dutch Eating Behaviour Questionnaire (DEBQ) was developed consisting of three similar subscales: Restraint, External Eating and Emotional Eating [63]. The purpose of the 10-item Restraint subscale (DEBQ-R) is to describe intentions to restrict food intake for weight reasons [63]. The DEBQ-R (Cronbach alpha score ≥0.9) [64] and TFEQ-R (Cronbach alpha score 0.79 to 0.93) [64] have good internal consistency and test-retest reliability. The TFEQ-R however is more widely used to assess restraint in the literature. As well, the validity and reliability of the TFEQ is well established [13,64,75,91]. Despite the good psychometric properties of the TFEQ-R, its factor structure has been investigated [75,92]. Westenhoefer [92] suggested that while restraint was necessary for disinhibition, it did not always result in disturbed eating behaviours as some women with higher restraint do not display disinhibition [93]. Furthermore, there are inconsistencies in the direction of correlations between the TFEQ-R and disinhibition subscales [13,94-95]. Findings from a large sample of overweight and obese persons with higher and lower TFEQ disinhibition scores participating in a weight reduction programme revealed two sources of variation within the TFEQ-R [92]. These were termed Flexible Control, an adaptable and accommodating approach to food and weight that is associated with lower disinhibition, and Rigid Control, an ―all-or-nothing‖ approach associated with higher disinhibition [92,96]. To assess these two distinct factors, additional items were added to the TFEQ-R resulting in the 12-item Flexible Control and 16-item Rigid Control subscales [96]. These subscales have moderate reliability (Cronbach alphas 0.77 to 0.79) and good predictive validity [96]. The negative relationship between Flexible Control and Disinhibition and positive relationship between Rigid Control and Disinhibition have been observed by others [96-97]. As the rigid and flexible control dimensions of CDR were identified in a sample of overweight individuals who were actively dieting, their relevance to those who are not overweight or actively dieting is not certain.  In summary, each of the three assessment tools is based on different operational definitions of CDR and thus measure distinct aspects of the construct. For the purposes of the     8 current research programme, the operational definition of the TFEQ-R is most relevant. Furthermore, the TFEQ-R is frequently used to classify women with higher and lower restraint in similar studies, allowing for more appropriate comparisons of findings.  1.2.2.2 Behavioural versus perceptual aspects of CDR Although the RS, DEBQ and TFEQ are highly inter-correlated [64], it has been suggested the scales do not measure the same behavioural tendencies or actual energy restriction. Heatherton and coworkers [85] suggested that the DEBQ and TFEQ measure successful dieting and the RS is designed to identify dieting. Moreover, factor analysis reveals that although all three scales share a restraint factor, only the TFEQ also assesses behavioural restraint, only the RS also assesses weight fluctuations [75], the RS is a measure of unsuccessful dieting [91], and the DEBQ and TFEQ measure successful dieting behaviour [91]. Additionally, only the RS also assesses binge eating behaviour [98]. As a result of these inconsistencies, the operational definition of CDR in research has become confused with the behavioural aspect of actual energy restriction or dieting. The distinction between CDR and dieting is important when attempting to assess whether it is the perceptual experience of CDR that is associated with health outcomes or the effects of negative energy balance achieved through dieting. Several lines of evidence, described below, establish that CDR should not be considered to be indicative of dieting.   Many women with higher CDR do not self-identify as dieters [11,99-100]. Moreover, while dieting behaviour is sporadic, CDR appears to be a relatively stable perceptual construct [49,64,101]. That is, while women with higher CDR levels are chronically concerned with and aware of the amount and types of foods that they eat in an attempt to control dietary intake, it does not appear that this results in consistently reduced energy intake or lower body weight, indicators of more ―successful dieting‖. There are some studies that have found lower self-reported energy intake among women with higher CDR levels [43,53,91,102-105] or an inverse correlation between self-reported energy intake and TFEQ-R score [91,106-108]. Yet, several studies have observed no difference in energy consumption by level of CDR [41,87,109-110]. In a large population-based study, restrained eaters (assessed by the DEBQ) were more likely to underreport energy intake than those with lower restraint [111]. Thus, self-reported dietary intake may not be an accurate means of examining differences by level of CDR.  More substantial evidence using objective measures of energy intake suggests that TFEQ-R score is not associated with short-, moderate- or longer term energy intake [15,112]. Furthermore, in a sample of 84 physically active university-aged women of normal and stable weight with no history of an eating disorder, there was no difference by level of CDR (TFEQ-R score ≥9 or below 9) in resting energy expenditure (REE) measured by indirect calorimetry, the     9 ratio of predicted REE by Harris Benedict equation (pREE) to measured REE or the proportion of women with an energy deficit (pREE:REE <0.90) [40]. It could be that dietary restraint does not have a consistent effect on energy intake but instead has a highly variable effect. Under certain conditions, CDR may have a large on energy intake yet no effect under other conditions [113].  Further support for the concept that elevated CDR is not synonymous with dieting is provided by studies in which relative body mass or body mass index (BMI; kg/m2) does not differ by level of CDR among normal-weight and obese women [40,42-43,53-54,34,87,91,103,110, 114-116]. In a recent study of over 1000 postmenopausal women, BMI was 1.0 kg/m2 (1 BMI unit) lower in women with higher versus lower TFEQ-R scores, and dieters had BMI that was 4.1 BMI units higher than non-dieters [100]. However, no interaction between dieting status and TFEQ-R score was apparent [100]. The association between BMI and dieting status was much stronger than that between BMI and TFEQ-R score providing further support that CDR and dieting are largely independent. Prospective studies also suggest that change in weight is more strongly associated with dieting than CDR. Among women, a history of dieting but not dietary restraint scores predicted weight gain during the first year of college [14]. Similarly, in a study of 163 middle-aged women, those who identified as dieters had a higher BMI at baseline and gained significantly more weight over six years than non-dieters [117]. On the other hand, CDR was not associated with baseline BMI or weight change [117]. Furthermore, although baseline TFEQ-R score moderated the association between disinhibition and weight, the direction of the relationship was different by dieting status [117]. Similarly, in a study of adolescents and young adults, CDR was not associated with two-year change in BMI, although dieting status was not assessed [101]. On the other hand, in a 6-year study of 283 healthy adults aged 18 to 64, those with higher restraint scores (TFEQ-R >8) gained 1 kg more than those with lower restraint after adjusting for potentially confounding variables [118]. As well, those with higher CDR were 26% more likely to experience a weight gain of ≥5 kg and were 18% more likely to develop obesity [118]. The discrepancy of findings from this study may be related to the inclusion of men and that approximately one half of participants had an obese BMI at baseline. In a prospective study of women with BMI values ranging from 17 to 40 kg/m2, TFEQ-R score was positively associated with BMI [49]. The change in weight by level of CDR was not reported in that study although neither BMI nor percent lean mass changed over time among all participants [49]. Despite the general lack of association between CDR and either BMI or weight change measured longitudinally, several studies have reported an increased frequency of past weight fluctuations in women with higher CDR [12,31,91,97,105,119]. This finding has been used as evidence to confirm that higher CDR scores reflect dieting behaviour. However, as these data     10 were collected retrospectively, the preoccupation and habitual monitoring of weight among women reporting higher CDR scores may heighten their awareness of actual weight changes or even the perception of weight change compared to women with lower CDR. Therefore, prospective studies are warranted to examine weight changes both by level of CDR and by dieting status. In order to distinguish between health outcomes associated with the experience of CDR and those associated dieting behaviours, it would be important to monitor energy intake, weight changes and dieting status in prospective studies.   1.2.3 Stress, cortisol and CDR 1.2.3.1 Chronic stress and neuroendocrine function Stress, whether inflammatory, traumatic or psychosocial, triggers a neuroendocrine response by the central nervous system and its peripheral components [23]. One response is activation of the hypothalamic-pituitary-adrenal (HPA) axis. As this is the focus of my PhD research project, it is the only aspect of the stress response discussed further in this review. The activation of the HPA axis is one of the body‘s main allostatic mediators allowing homeostasis to be maintained during stressful conditions by adaptive responses [120]. As shown in Figure 1.1, HPA axis activation stimulates release of corticotropin releasing hormone (CRH) from the paraventricular nucleus of the hypothalamus and in turn adrenocorticotropic hormone (ACTH) secretion from the pituitary [16]. Cortisol, a glucocorticoid stress hormone, is subsequently released by the adrenal cortex [16]. Cortisol maintains homeostasis during acute stress by activating short-term behavioural and physical changes that improve the chance of survival [23]. Examples include increased alertness, inhibition of hunger and increased respiratory rate and cardiovascular tone [23]. Cortisol excretion is highly variable throughout the day and is associated with the diurnal pattern of circadian rhythm [23]. During nocturnal sleep, cortisol is low and then increases during the second half of the night, peaking shortly after waking (acrophase), and then steadily declining over the course of the day [23]. Under normal circumstances, cortisol levels are regulated via negative feedback: high levels of circulating cortisol inhibit CRH and ACTH, thus decreasing cortisol synthesis [23]. However, dysregulated allostasis can occur, leading to chronically elevated cortisol [120]. McEwen terms this ―wear-and-tear‖ on the body‘s systems allostatic overload [120]. The continuous secretion of CRH and cortisol, as seen with Cushing‘s syndrome, can adversely affect growth and development, thyroid function, reproduction, metabolism, gastrointestinal function and immune function [23].  Chronic psychosocial stress is associated with many negative health outcomes including anxiety, depression, infertility, hypertension, obesity, type 2 diabetes mellitus, atherosclerosis, neurovascular degenerative disease, osteoporosis and sleep disorders [16]. Evidence suggests that the relationship between adverse health conditions and chronic psychosocial stress may be     11 related to dysregulation of HPA axis activation [16]. For example, higher cortisol levels have been observed among middle-aged women reporting greater financial strain [121], higher job demands [122], higher job strain [123] and more marital stress [124]. Among young women, studies suggest a relationship between cortisol and eating attitudes, discussed subsequently.  1.2.3.2 Assessment of cortisol and general stress perception Cortisol is a well established biological indicator of HPA axis activation resulting from stress [125]. In the blood, cortisol circulates both in free form and bound to corticosteroid-binding globulin [125]. The Free Hormone Hypothesis assumes that only free cortisol is biologically active and therefore relevant in determining HPA axis activity related to stress. There is debate as to the biological activity of both free and bound cortisol as well as the best method of assessing cortisol levels in relation to chronic stressors [125]. Cortisol levels can be determined using samples of urine, saliva and plasma. Each method measures a unique aspect of the HPA axis response to stress and has its own strengths and limitations.  In plasma, free cortisol levels are calculated from total cortisol and either of corticosteroid-binding globulin binding capacity or corticosteroid-binding globulin, as no kit currently exists to measure free plasma cortisol [125]. There is potential for high variability in cortisol levels using immunoassays, as cortisol is capable of cross-reacting with other steroids [125]. Plasma cortisol is useful in the clinical setting to diagnosis disease states by comparing levels to the normal range. However, the use of plasma cortisol in research has many limitations including the need for medical staff and specialised equipment as well as high costs and subject burden [125]. Furthermore, some participants may find blood sampling stressful, potentially elevating cortisol levels artificially [125]. Due to the diurnal rhythm of cortisol, the timing of plasma sampling is also an important consideration [125]. Multiple measures of plasma cortisol over time are useful for determining the response to stressful stimuli or to determine whether there is dysregulation of the 24-hour rhythm of cortisol production under various conditions [125]. Single assessments are used to examine whether cortisol levels are associated with physiological or affective state characteristics [125].  Salivary cortisol represents free cortisol that has entered the salivary glands by passive diffusion [125].  Cortisol in saliva is assayed using the same kits for total serum cortisol adjusted for sensitivity [125]. Determining cortisol levels in the saliva has become increasingly popular since the 1980s. It is highly correlated with plasma cortisol (r= 0.71-0.96) and has many advantages over plasma assessment, for example, salivary collection is very useful for research outside of the laboratory as sampling is non-invasive and can occur quickly and frequently [125]. Moreover, salivary cortisol is stable at room temperature, does not require specialised staff or equipment and has a lower processing cost [125]. Similarly to plasma assessment of     12 cortisol, the timing of sampling is an important consideration. Other limitations include problems of participant compliance and the potential contaminating effect of food, drink or blood [125]. Salivary cortisol may be a useful diagnostic tool and may be particularly useful for examining the stress response outside the laboratory.  In 1995, a sharp rise in cortisol levels 20-45 minutes after waking was discovered in addition to the previously described diurnal pattern, termed the cortisol awakening response: the difference in salivary cortisol measured immediately upon waking and 30 minutes later [126]. Evidence suggests that it is a distinct occurrence, a reliable indicator of HPA axis activity and is associated with psychiatric, autoimmune and cardiovascular disorders [126]. However, great variability has been documented and potential confounders include age, smoking status, time of wakening, day of the week and participant compliance [126]. Although the physiological role has not been clearly defined, the cortisol awakening response does appear to be associated with various psychosocial stressors including work overload, social stress, lack of social recognition and perceived stress [126]. Based on these and other findings, it is hypothesised that the cortisol awakening response may represent expectation of the demands of the approaching day [126]. Urinary free cortisol excretion is generally measured over 24-hours and is a useful index of 24-hour plasma cortisol [125]. Determination of 24-hour urinary free cortisol (UFC) is currently recognised as the gold standard in diagnosis of hypercortisolism [127]. Traditionally, UFC was determined using immunoassay methods adapted from serum cortisol methods [127]. These methods overestimate UFC as cortisol metabolites, such as cortisone (the inactive form of cortisol), interfere with the immunoassays used [127]. Recently, more specific methods have been developed based on liquid chromatography-tandem mass spectrometry [127]. These methods allow for quantification of cortisone and have reduced interference for cortisol quantification, resulting in increased sensitivity and specificity relative to previous methods [127]. The stress associated with CDR would likely occur mostly when women were awake and actively involved in eating behaviour decisions. Therefore, using salivary samples at various points during the day and/or overnight blood or urine sampling to assess cortisol levels may not capture the persistent activation of the HPA axis associated with chronic stressors such as CDR. Additionally, evidence suggests that cortisol is highly variable and is associated with the occurrence of everyday minor stressful events [128-129]. Therefore, repeated measures of UFC may give a more accurate depiction of ―usual‖ stress-induced HPA axis activation. Thus, multiple 24-hour urine samples assessed for UFC using high-throughput liquid chromatography-tandem mass spectrometry will be used as an indicator of stress induced activation of the HPA axis in the present study.      13 Assessment of the perception of stress is also important in understanding the relationship between physiological and psychosocial health. The most commonly used measure of stress perception in the literature is the 14-item Perceived Stress Scale (PSS) [130]. The PSS measures one‘s feelings of stress in various life situations during the previous month [130]. The PSS is an indicator of the level of stress one feels in various life situations, as opposed to determining the presence or frequency of particular stressful events [131]. It is widely used in the CDR literature and is reported to be both reliable and valid with Cronbach alpha scores ranging from 0.80 to 0.86 [130-132]. In addition to assessing the perception of ―usual‖ stress by employing the PSS, checklists are often used to determine the frequency of stressful events that occurred over a defined period time. One such checklist is the Daily Stress Inventory (DSI) [133] which includes two subscales, Impact and Frequency. The DSI queries the frequency of 58 everyday minor stressful events which may have occurred (Frequency score) as well as a ranking of the intensity of stressful events (Impact score) that occurred on a scale of 1 (―not at all stressful‖) to 7 (―caused me to panic‖) [133]. The DSI has good internal consistency with Cronbach alphas ranging from 0.83 to 0.87 and adequate convergent and discriminant validity [128,133]. Assessment of both usual and acute stressors is likely important when examining the association between a particular stressor (such as CDR) and cortisol levels.   1.2.3.3 Cortisol and CDR We and others have hypothesised that the constant monitoring of food intake and preoccupation with body weight experienced by women with higher CDR may act as a chronic daily stressor that is sufficient to activate the HPA axis (Figure 1.1). The majority of cross-sectional studies investigating this relationship support this hypothesis. Previous work from the Barr lab found significantly higher 24-hour urinary cortisol and cortisol:creatinine ratios among those with a high versus low (TFEQ-R ≥13 versus ≤5 or less) level of CDR [19]. This study included 62 regularly menstruating university-aged women with no previous eating disorders who did not identify as dieters and had normal activity levels and normal BMI values [19]. These results were confirmed in a similar study of 77 healthy postmenopausal women of normal and stable weight [17]. Among 85 premenopausal university students, salivary cortisol was higher in those with higher CDR (TFEQ-R scores ≥8) [18]. Salivary cortisol was also significantly correlated with TFEQ-R score in that study [18]. However, it is noteworthy that saliva samples were collected after participants were weighed and had completed the TFEQ-R and RS, activities that may have acted as a stressor for women with higher CDR and thus acutely elevated cortisol levels in this group.  Conversely, the first study that examined CDR and neuroendocrine function did not observe differences in cortisol by level of CDR in 22 university-aged women using overnight     14 blood samples obtained by venous catheter (every 30 minutes) [103]. The lack of an association may be related to several study design issues including the very small sample size and timing of the cortisol assessment. As described in the Assessment section (1.2.3.2), CDR-related stress may only activate the HPA axis when women are awake and actively involved in eating behaviour decisions. Therefore, the stress associated with CDR may not be captured by overnight assessment of cortisol levels. Two recent studies further indicate that the timing of cortisol assessment in relation to CDR is an important consideration. Serum cortisol was sampled for five hours (during which time breakfast and lunch were provided) in 38 normal-weight women [21]. After the sampling period, a dexamethasone suppression test was performed [21], which assesses the integrity of the HPA axis negative feedback loop. Dexamethasone is a synthetic glucocorticoid which under normal conditions suppresses CRH and thus ACTH and subsequently, cortisol [134]. Higher cortisol levels after a suppression test indicate decreased cortisol feedback functioning, a characteristic that is associated with hypercortisolism [134]. Women with higher CDR (TFEQ-R score ≥9) had higher cortisol levels than women with lower CDR over the 5-hour sampling period and after the suppression test [21]. Furthermore, in a sample of 170 university-aged women, TFEQ-R score was significantly associated with afternoon but not waking salivary cortisol [20]. These three studies suggest that the timing of cortisol assessment is highly important when considering if CDR may influence the HPA axis.  Two other studies did not find associations between CDR and cortisol [56,135]. Among 65 women with characteristics similar to previous studies, no difference was observed in waking salivary cortisol by the median split TFEQ-R score of nine [56]. Furthermore, there was no correlation between TFEQ-R score and waking salivary cortisol collected within 1.5 hours of waking [56]. As wake-time was not specifically accounted for in that study, it is possible that the peak that occurs 30 minutes after waking may have been captured in some but not all of the samples. This may have caused significant variability in cortisol levels; however, salivary cortisol values are not reported [56]. In a small study of 28 women, the cortisol awakening response (averaged from three samples provided over two months) was not correlated with TFEQ-R scores but was inversely associated with other eating attitude questionnaire scores including Rigid Control of restraint, Disinhibition and Hunger [135]. Participants in this study were significantly older (mean 37 years) and had a higher BMI (mean 29 kg/m2) than those studied previously. Though the study included similar eligibility criteria to previous work and a sound experimental procedure, in a sample this small a single outlier would be sufficient to pull correlations one way or the other. Unfortunately, despite a detailed description of their experiment, the authors do not describe whether outliers were examined or how they were dealt with.      15 1.2.3.4 Perception of psychosocial stress and CDR The experience of other psychosocial stresses may be important to consider in the relationship between CDR and cortisol: some women with higher CDR may perceive greater stress in all aspects of their lives. Studies examining the relationship between perceived stress and CDR in women are inconclusive with some reporting no difference in PSS scores by level of CDR [17,43] and others finding higher PSS scores in women with higher versus lower dietary restraint [12,50]. Significant correlations between PSS score and scores on the TFEQ-R [20] and other questionnaire scores reflecting disordered eating attitudes and behaviours, and body dissatisfaction [20,81,136-137] have been reported. As it is not clear if general stress is associated with CDR, it is important to assess and control for this potentially confounding variable  1.2.4 Ovulatory function 1.2.4.1 Physiology of the menstrual cycle The menstrual cycle is the result of the coordinated activity of the following hormones: gonadotropin-releasing hormone (GnRH) secreted from the hypothalamus, luteinizing hormone (LH) and follicle stimulating hormone (FSH) secreted from the pituitary, and estradiol and progesterone secreted from the stimulated follicle within the ovaries [138]. The menstrual cycle occurs on average over 28 days and is divided into two phases:  the follicular phase and the luteal phase. At the onset of menstrual flow, the first day of the cycle, estradiol and progesterone levels are low, allowing for release of GnRH, which then triggers the production and release of LH and FSH [138]. This stimulates the growth and maturation of ovarian follicles, which begin to secrete estradiol that peaks at midcycle and inhibits FSH [138]. The majority of stimulated ovarian follicles are then degraded and resorbed but the most mature and now dominant follicle continues to release large amounts of estradiol [138]. This results in thickening of the endometrial lining of the uterus and stimulation of the LH surge [138]. The surge in LH functions to: (i) inhibit estradiol release by the follicles, (ii) stimulate rupture of the dominant follicle and ovulation by releasing the ovum, and (iii) transform the ruptured follicle into the corpus luteum [138]. The corpus luteum then releases some estradiol and very high levels of progesterone, the predominant hormone in the luteal phase of the cycle [138]. If fertilization does not occur, the corpus luteum regresses and estradiol and progesterone levels decline, resulting in the shedding of the thickened endometrial lining as menstrual flow and thus, the beginning of a new menstrual cycle [138].       16 1.2.4.2 Disturbances in ovulatory function  Disturbances in menstrual cycle length and cycle characteristics as the result of various physiologic and psychosocial stressors may occur during the reproductive years [38]. Some of these disturbances are apparent to women including amenorrhea (the absence of menstrual flow for six or more months) and oligomenorrhea (long cycles of 36 to 180 days in length). It is well established that these disturbances, indicators of insufficient estradiol and progesterone which are important in achieving and maintaining peak bone mass, are associated with reduced bone density [139]. Other cycle disturbances, including anovulation and short luteal phase length (LPL), are not apparent to women yet evidence is mounting that these subclinical disturbances in ovulatory function are also associated with bone loss [45-49]. Subclinical disturbances of the menstrual cycle are difficult to monitor in research studies, as women are unaware of these disturbances. Therefore survey methods cannot be used and a physiological indicator is necessary. Furthermore, the menstrual cycle shows considerable intra-individual variability, particularly in LPL, and therefore long-term monitoring of the menstrual cycle is required to properly identity women with subclinical ovulatory disturbances [140-142]. As most methods to detect ovulatory function are expensive and burdensome procedures, a non-invasive inexpensive method to assess these subclinical ovulatory disturbances over long periods of time and that is acceptable to women, is required.   1.2.4.3 Monitoring ovulatory function The current gold standard for directly determining if and when ovulation has occurred is observation of collapse of the dominant follicle with corpus luteum formation by daily transvaginal ultrasound of the ovaries. During the follicular phase, developing follicles and the dominant follicle can be observed and during the luteal phase, the corpus luteum can be seen [143]. In addition to being costly, and needing a probe in the vagina, this method requires extensive training in its operation and results interpretation [144]. Indirect methods of assessing ovulatory function, referred to as ―evidence of luteal activity‖ (ELA) include: (i) histomorphometric examination of endometrium biopsy to monitor cellular characteristics that change in response to estradiol and progesterone; (ii) repeated blood samples to monitor normal cyclic patterns of estradiol and progesterone levels; (iii) a single blood sample obtained during days 18 to 22 of the cycle to determine if estradiol and progesterone are above baseline values; (iv) urine samples collected between cycle days 12 and 16 for detection of the LH surge or to monitor the change in the ratio of estradiol to progesterone metabolites [145]; (v) measurement of salivary progesterone levels every three to four days; and (vi) monitoring the ratio of daily urinary progesterone metabolites to identify a sustained rise [146]. However, these     17 methods are also costly and burdensome to women that are not motivated to achieve conception or who have infertility.  Analysis of daily basal temperature records can also be used to determine ELA. Progesterone has a thermic effect on the hypothalamus that leads to an increase of approximately 0.3°Celsius from the follicular phase, when progesterone is low, to the luteal phase, when progesterone peaks [147]. Basal temperature records have been used for decades in combination with changes in cervical mucous as an established fertility-awareness based method of contraception to aid conception [148]. In the past, basal temperature was charted by women and ovulatory function was determined using qualitative analysis methods. Qualitative methods include identification of a basal temperature nadir (low point) at the estradiol peak one to two days prior to the LH surge [149] and visual determination of a biphasic (and thus ovulatory) basal temperature graph by reproductive medical experts. Numerous studies have found these qualitative methods to be inaccurate in the documentation of ovulation relative to ultrasound [150-154], the LH surge [155-159] or the ratio of estradiol to progesterone over the cycle [160]. As well, experts do not always agree on whether or not the same plotted temperatures are biphasic [161], even when using uniform analysis criteria [162].  Quantitative methods of basal temperature analysis (QBT) may be more accurate at determining ELA than previous qualitative methods. However, little work has been done to validate them against other established methods of ovulation detection. There are currently three QBT methods described in the literature. The first is the Vollman averaging method, for which the average cycle temperature is computed and then compared with the recorded temperatures to determine at which day the temperature rises higher and is maintained above the average until flow begins [163]. The second method, the cumulative sum method, involves calculating a baseline average, the average temperature of cycle days five to eleven, and then determining which cycle day the recorded temperature is more than 0.35°Celsius above the baseline [164]. The third and most recently developed method is a computerised least-squares analysis of quantitative basal temperature (LS-QBT), the Maximina© programme [165]. LS-QBT detects ELA by dividing the cycle into two phases by least squares criterion and determining if the mean temperature difference between the phases is statistically significant [165]. The day of luteal onset identified by all three QBT methods has been assessed relative to peak serum LH concentration in 24 ovulatory cycles [165]. Both the Vollman averaging (r=0.89, P<0.001) and LS-QBT methods (r=0.88, P<0.001) showed excellent correlation [165]. However, further validation of these methods and against an indicator more clearly and reliably related to ovulation, such as progesterone, is necessary to increase their acceptability.       18 1.2.4.4 Ovulatory function and the HPA axis  Various life stresses have been found to be associated with disturbed ovulatory function among young women [38]. Although the underlying mechanism is not fully understood, evidence suggests that it may be related to the physiological stress response [38]. Stress-induced HPA axis activation triggers the release of CRH from the hypothalamus suppressing pulsatile GnRH release, as shown in Figure 1.1. As described above, GnRH is responsible for the secretion of LH and FSH from the pituitary, which in turn, stimulates the secretion of ovarian estradiol and progesterone. There is a substantial amount of evidence that HPA axis activation is related to disturbed menstrual cycle and ovulatory function [23,38]. For example, in several studies, women with functional hypothalamic amenorrhea (FHA) have been found to have elevated cortisol, lower urinary metabolites of progesterone and reduced 24-hour LH pulse frequency (indicating suppression of GnRH) than ovulatory women or women with organic forms of anovulation [38]. Additionally, in a small longitudinal study, increased cortisol was associated with lower progesterone levels between days four and 10 after ovulation [166].   1.2.4.5 Ovulatory disturbances and CDR Evidence suggests that menstrual cycle and ovulatory disturbances may be more common among women with higher CDR, including both disturbances in cycle length (e.g. irregular cycles, oligomenorrhea or amenorrhea) and of subclinical characteristics (anovulation and short LPL). In an exploratory study of 334 female university students not using oral contraceptives and with no history of eating disorders, 33% of women with higher CDR (TFEQ-R score ≥13) reported irregular menstrual cycles [12]. This was significantly more than the 16% of women with lower levels of restraint (TFEQ-R score ≤5) [12]. Furthermore, TFEQ-R score was the only measured variable that significantly differentiated between women reporting regular versus irregular cycles [12]. Among 38 athletic women, 50% of those with high CDR scores had oligo- or amenorrhea versus 25% of those with lower scores [39]. As well, when divided by the TFEQ-R median score of nine, self-reported oligomenorrhea was higher among those with higher versus lower CDR (50% versus 26%) in a sample of 84 physically active university-aged women of normal and stable weight with no history of an eating disorder [40].  Schweiger and colleagues [43] were the first to observe the relationship between CDR and subclinical disturbances in ovulatory function in a sample of 22 young, normal-weight, normally active, regularly menstruating women [43]. Those with high dietary restraint had shorter menstrual cycle lengths, lower mean luteal phase progesterone levels and shorter LPL than women with lower dietary restraint [43]. Shortened LPL assessed by LS-QBT was also observed among normal-weight, regularly menstruating, ovulatory women participating in various levels of physical activity with higher versus lower dietary restraint [42]. These findings     19 were confirmed in a six-month prospective study, also using LS-QBT, in which normal-weight, regularly menstruating women with higher CDR had more anovulatory cycles and shorter LPL [41]. Furthermore, among 33 female athletes those with menstrual disturbances (determined by salivary progesterone and estradiol levels) reported higher TFEQ-R scores than those with normal menstrual cycles despite similar BMI and exercise levels between groups [167]. Similarly, in a sample of 48 university-aged, non-dieting women of normal and stable weight with no prior history of eating disorders, TFEQ-R scores were significantly higher among exercising amenorrheic women than either exercising or sedentary ovulatory women [168]. Additional support that eating and body stresses can lead to ovulatory disturbances comes from studies that have observed associations using measures other than CDR. Higher scores on the Eating Attitudes Test [169], and the Drive For Thinness and Bulimia subscales of the Eating Disorder Inventory [170] have been reported in women with FHA versus women with organic causes of amenorrhea and/or regularly menstruating women [171-175]. Since my PhD research was proposed, a 2-year prospective study examined the relationship between CDR and ovulatory function [49]. The sample included 189 healthy, regularly menstruating women that were not using oral contraceptives, with a mean age of 32.4 and a mean BMI of 24.3 [49]. Ovulatory function was monitored by salivary progesterone and commercial ovulation kits (mean of 9.8±3.4 cycles were monitored, maximum 12) [49]. Classified by tertiles of TFEQ-R, there was no difference by level of CDR in mean menstrual cycle length, mean LPL, mean luteal salivary progesterone or the percentage of women with three or more cycles with ovulatory disturbances (anovulation or LPL <10 days) [49]. As well, there were no differences by CDR tertiles in serum estradiol or testosterone on cycles days three to five in one of the monitored cycles [49]. There are several possible reasons as to why a null relationship between CDR and ovulatory function was observed in that study. First, the highest CDR tertile included women with TFEQ-R scores >9.4, the median score observed in several other studies. Second, it is unclear why the authors defined more than three cycles with anovulation and/or short LPL as their categorization of higher subclinical ovulatory disturbances. Furthermore, it is not clear why correlations coefficients between the percentage of cycles with subclinical ovulatory disturbances and TFEQ-R scores were not reported.  The authors of that study suggest that the association between subclinical ovulatory disturbances and CDR observed in previous studies among normal- or under-weight women is related to caloric restriction and/or other dieting behaviours such as over exercising. In their study, TFEQ-R scores were associated with higher physical activity levels and women in the highest CDR tertile had higher sport activity levels and BMI than women in the lowest tertile [49]. Based on this, the authors suggest that women with higher CDR were overweight women that were attempting to lose weight by dieting and exercising [49]. However, weight loss     20 attempts, change in weight or BMI, and energy intakes were not reported. The larger sample size of this study would have allowed for examination of the relationship between CDR and subclinical ovulatory disturbances among normal- versus over-weight women to support their hypothesis. However, this analysis was not reported.  It is unlikely that women with higher CDR in the above-cited studies [41-43,49,167-168,171-172,174] were trying to lose weight or were ―successful restrainers/dieters‖ because participants were weight-stable, did not report current dieting and those with current or past eating disorders were excluded. As well, with the exception of three studies [41,173,175], BMI and exercise levels did not differ between groups. Therefore, it is doubtful that an energy deficit in women with higher CDR caused ovulatory disturbances. In fact, in a study of normally menstruating women, LH pulsatility during the follicular phase was not disrupted until energy availability was <30 kilocalories per kg lean mass [176]. Energy intakes that low would be unlikely in samples of healthy, normal weight women. Moreover, literature exists to indicate that physical activity per se does not cause ovulatory disturbances [46,177-179]. Nevertheless, energy intake, physical activity, and changes in anthropometric measurements would be important variables to monitor when assessing disturbances in ovulatory function.  The authors of the only other prospective study also note that their sample had a higher mean age than previous work suggesting that participants would have reached gynaecologic maturity and therefore their cycles would be less likely to be affected by psychosocial stresses [49]. The much lower prevalence of subclinical ovulatory disturbances that was reported in that study (33.3% of women experienced at least one during their study) versus previous work (67 to 80%) provides support for this hypothesis. The criterion for cycles with short LPL used in that study may be an additional reason for the low frequency of subclinical ovulatory disturbances observed. In that study [49], ELA was determined by commercial ovulation kits, which detect the urinary LH surge. The serum LH peak is a well established indirect indicator of ovulation, and occurs 16 to 48 hours (average 24 hours) prior to documentation of follicular collapse (ovulation) by ultrasound [180]. The lag between the serum and urinary LH peak is less than eight hours (average two to three hours) [181]. Therefore, using urinary LH peak as the estimated day of luteal transition, the criterion for short LPL should be <11-12 days rather than <10 days, the cut off used in that study [49]. Ten days or less is used as the criterion for short LPL by LS-QBT analysis as the significant rise in basal temperature occurs approximately 2.4 days following serum peak LH [165]. In summary, although a recent prospective study did not detect an association between CDR and ovulatory function, the majority of research supports such associations. As CDR appears to be associated with elevated cortisol, which can impair ovarian function it could be     21 that the relationship between CDR and ovulatory disturbances is mediated by the physiological stress response.  1.2.5 Bone and cortisol Glucocorticoids such as cortisol have negative effects on bone via direct and indirect mechanisms. Cortisol acts directly to disrupt bone via adverse effects on the bone-forming osteoblast cells, by suppressing their formation and activity, as well as by supporting apoptosis [22]. Synthesis of the bone matrix is inhibited as glucocorticoids decrease the synthesis of type 1 collagen, and alter the expression of messenger ribonucleic acid encoding matrix components including osteopontin, fibronectin, beta-integrin and bone sialoprotein [182]. Although the role of the glucocorticoids on the osteoclasts is less clear, bone resorption is increased via an anti-apoptotic effect resulting in increased osteoclast number [183]. As well, osteoclast activity may increase as glucocorticoids are associated with decreases in serum osteoprotegerin, a cytokine that inhibits osteoclast differentiation [184-185].  Indirectly, glucocorticoids may impact bone density via impaired calcium metabolism by: (i) reducing intestinal calcium absorption possibly by inhibiting active transcellular calcium transport; (ii) decreasing the synthesis of calcium binding protein and/or increasing the rate of degradation of active vitamin D at its mucosal binding site; and (iii) decreasing renal calcium reabsorption as evidenced by increased urinary calcium excretion [22]. Among premenopausal women, the disruption of the normal cyclic patterns of the reproductive hormones is another indirect means by which cortisol may affect bone density. The importance of the reproductive hormones to bone is discussed in the next section of this literature review. Excess endogenous cortisol or hypercortisolism, such as in Cushing‘s syndrome, has long been known to increase the risk of osteoporosis. A recent review indicates that Cushing‘s syndrome patients have reduced bone formation, lower bone density and an increased incidence of osteoporosis and fractures [186]. Subclinical hypercortisolism, as may occur with an adrenal adenoma, shows similar patterns [186]. The effect of subtle increases in cortisol within the physiological normal range on bone is less certain. Currently, there are several lines of evidence, discussed in detail subsequently, that suggest that slight elevations in cortisol may negatively influence bone. Correlations between higher cortisol and lower bone density have been reported in healthy samples of older adults, clinical samples of premenopausal women with anorexia nervosa or major depressive disorder, and in studies of women with higher and lower CDR.  The first line of evidence is available from studies suggesting an inverse relationship between cortisol levels and bone density in healthy older adults. In a sample of 37 healthy men aged 43 to 73, a significant negative correlation was observed between lumbar spine areal bone     22 mineral density (aBMD) and fasting waking cortisol levels (r= -0.33) [25]. As well, backward regression analysis indicated that cortisol (along with testosterone and BMI) was a significant predictor of lumbar spine aBMD [25]. In a sample of 45 healthy, normal-weight, postmenopausal women not using hormone therapy, salivary cortisol level assessed at 11 p.m. was negatively correlated with lumbar spine aBMD (r= -0.20) [27]. However, night salivary cortisol was not associated with aBMD at the femoral neck or trochanter, and salivary cortisol collected at 7 a.m. was not associated with aBMD at any site [27]. That study also included 130 healthy men among whom morning salivary cortisol was negatively correlated with lumbar spine aBMD (r= -0.31) [27]. Unexpectedly, night time cortisol was positively correlated with trochanter aBMD (r=0.18) and radial aBMD (r=0.21). There is no obvious reason for the discrepant findings. The most convincing evidence of the potential for elevated cortisol to negatively impact bone in healthy individuals comes from two prospective studies. The first study included 34 healthy older men who completed measures of aBMD at baseline and four years later, as well as baseline 24-hour serum cortisol assessment [26]. After adjusting for potentially confounding variables, a significant positive correlation was observed between trough cortisol level and the rate of bone loss at the lumbar spine (r=0.38), femoral neck (r=0.47) and trochanteric region (r=0.41) [26]. This suggests that those with higher cortisol levels at the lowest point in the diurnal cycle experienced greater bone loss over four years. The second study involved 151 men and 96 women who had aBMD assessed at baseline and again four years later [28]. Cortisol levels were assessed by 24-hour urinary cortisol as well as a dynamic suppression-stimulation test of the HPA axis [28]. After adjustment for potentially confounding variables, elevated peak plasma cortisol at activation was correlated with lumbar spine bone loss in men (r=0.22) and femoral neck bone loss in women (r=0.24) [28]. However, aBMD change was not associated with 24-hour urinary cortisol or cortisol levels following the suppression test [28]. Findings from this study suggest that HPA axis sensitivity but not dysregulation of the negative feedback mechanism may be related to the rate of bone loss. The effect of cortisol on bone has also been investigated by evaluating fracture risk. A large prospective study evaluated the influence of cortisol on 8-year fracture risk in a sample of 684 generally healthy men and women, aged 70 to 79 [187]. Logistic regression analysis adjusted for confounders found those in the highest quartile of baseline 24-hour urinary free cortisol had a significantly greater risk of fracture (odds ratio 5.38) than those in the lowest quartile [187].    In premenopausal women, cortisol has the potential to negatively influence bone directly and indirectly via associations with reproductive hormone deficiencies or imbalances. As discussed subsequently, the reproductive hormones are crucial to achieving and maintaining bone mass. Yet, few studies have examined the relationship between physiologically normal elevations in cortisol levels and bone in healthy women. Women with anorexia nervosa and     23 major depressive disorder tend to have higher cortisol levels but do not present with hypercortisolism [188-189]. Among women with major depressive disorder, correlations have been reported between waking serum cortisol and aBMD in some [29,31], but not all studies [36-37]. Similarly, findings from studies of anorexia nervosa patients suggest an association between cortisol and bone mineral content (BMC) and/or aBMD [30-31,33]. However, as amenorrhea and very low body weight are part of the diagnostic criteria for anorexia, the independent effect of cortisol on bone is difficult to determine from these studies. Therefore the generalisability of findings from clinical samples to healthy young women is limited due to the presence of other conditions in these disorders (e.g. amenorrhea, muscle atrophy, immune dysfunction) that have significant effects on bone density. Three of the studies examining CDR, cortisol and bone in small samples of healthy young women have also reported on the cross-sectional relationship between cortisol and aBMD. Among 62 regularly menstruating women with either higher or lower dietary restraint, the 24-hour urinary cortisol:creatinine ratio was negatively correlated with total body BMC [53]. However, the 24-hour urinary cortisol:creatinine ratio was not a significant independent predictor of total body BMD in a multiple regression [53]. Salivary cortisol assessed within 1.5 hour of waking was not associated with aBMD measured at any site (total body, lumbar spine, non-dominant hip & forearm) in 65 regularly menstruating university-aged women [56]. However, as mentioned previously, the timing of the cortisol assessment limits the interpretation of these findings. Finally, in a sample of 78 middle-aged obese women with a history of chronic dieting, waking serum cortisol was not related to BMC or aBMD at any site (total body, lumbar spine, right femur) [52]. Furthermore, when women were categorised as those with low and normal bone density (Z-score ≤-1.0 versus >-1.0), no differences in cortisol were apparent [52].  The possible association between cortisol and bone has also been assessed in a study that used broadband ultrasonic attenuation (BUA) and speed of sound (SOS) as indices of bone strength [190]. The cortisol awakening response was positively associated with calcaneal BUA and SOS in a sample of 36 healthy, normally active, regularly menstruating, non-obese women [190]. As well, women with peak cortisol above the median had higher BUA and SOS than women with cortisol levels below the median [190]. The very small sample size and the relevance of ultrasound bone measures to bone strength in premenopausal women [191] limit the significance of these findings. Additional examination of this question, particularly through longitudinal studies, is warranted in order to determine if subtle elevations in cortisol have a small but persistent negative influence on bone change over time.     24 1.2.6 Bone and ovulatory function  Disturbances in menstrual cycle and ovulatory function result in reduced exposure to estradiol and progesterone. It is well established that estradiol deficiency, regardless of the aetiology, results in increased bone loss as estradiol prevents bone resorption [44]. Amenorrhea and oligomenorrhea are menstrual cycle disturbances that are known to result in increased bone loss [139]. It has been suggested that progesterone deficiency may also result in bone loss, as progesterone appears to promote bone formation [44,192]. Progesterone peaks during the luteal phase of the menstrual cycle and thus women who experience luteal phase disturbances (anovulation and short LPL) are exposed to lower levels of progesterone.  Several studies by Prior and colleagues support an association between subclinical ovulatory disturbances and bone. In the first study, 66 normal-weight, regularly menstruating women were screened to be initially normally ovulatory in two consecutive cycles with normal estradiol levels [46]. Significant associations were observed with 1-year change in spinal cancellous BMD and both LPL (r=0.48) and luteal phase progesterone levels (r=0.25) [46]. As well, LPL explained 20% of the variance in annual cancellous BMD change [46]. These data were further analyzed by dividing women into two groups (runners versus normally active women) to evaluate exercise-related effects separately from those of ovulatory function: both exercise and LPL had independent positive effects on spinal cancellous BMD [45]. In a subsample that continued to have regular cycles, 1-year LPL was correlated with 5-year BMD change [47]. However, 1-year LPL was not a predictor of BMD change in regression analyses [47]. Finally, in a 1-year randomised, double-blind, placebo-controlled trial among 61 active women with various menstrual cycle disturbances (amenorrhea, oligomenorrhea, anovulatory cycles or short LPL cycles), supplementation with cyclic medroxyprogesterone (10 mg/day for 10 days/month) resulted in significant gains in lumbar spine aBMD [193]. Support is also available from a nested case-control study of healthy, regularly menstruating women with normal (50 to 75th percentile) versus low (<10th percentile) aBMD who were similar in age of menarche, annual number of periods, smoking history, and calcium and energy intakes [48]. The women with lower aBMD had significantly lower urinary estradiol and progesterone metabolites and a less pronounced LH response than women with normal aBMD [48]. Finally, a prospective study provides convincing evidence of an association between subclinical ovulatory disturbances and less positive changes in aBMD, though perhaps not mediated by progesterone [49]. In this 2-year study, 189 healthy women with regular menstrual cycles, aged 21 to 40, had their cycles monitored two times per year for three consecutive cycles (maximum of 12 cycles, average 9.8±3.4) by daily salivary progesterone and urinary LH surge [49]. Serum estradiol and testosterone for cycle days three to five were assessed during one of the monitored cycles [49]. Subclinical ovulatory disturbances were     25 experienced by 33.3% of participants during the study [49]. Women with three or more disturbed cycles did not differ in physical activity level or lifestyle characteristics than those with fewer than three disturbed cycles [49]. Women with three or more disturbed cycles had significantly less positive changes in lumbar spine aBMD than those with less than three disturbed cycles. Baseline aBMD and neither femoral neck nor total body aBMD change differed by subclinical ovulatory disturbances [49]. Furthermore, in expanded predictive models of 2-year aBMD change, having three or more disturbed cycles resulted in a significantly decreased rate of change (-0.0109 g/cm2) in lumbar spine aBMD but not femoral neck or total body aBMD [49]. As salivary progesterone levels were not associated with change in aBMD in this study [49], the mechanism mediating the relationship between subclinical ovulatory disturbances and bone loss is uncertain. Conversely, others have reported no associations between indicators of subclinical ovulatory disturbances and changes in aBMD among healthy, normal-weight, regularly menstruating women. In a study of 53 sedentary women who collected daily urine samples for a mean of 4.1 cycles (maximum six, minimum not reported) and were then monitored for aBMD over 17.5 months, neither mean LPL nor average urinary progesterone metabolites were associated with baseline aBMD or aBMD change [51]. Similarly, daily urine samples were collected over three menstrual cycles before aBMD was assessed in small samples of sedentary ovulatory women (n=9), active ovulatory women (n=14) and active women with luteal defects (n=10) [50]. Although there were differences in luteal phase function, there were no differences between groups in aBMD or markers of bone turnover [50]. This study included few women for a cross-sectional study of bone density. In a more recent cross-sectional study of 242 women, aged 30-40, with BMIs ranging from underweight to obese, two menstrual cycles were assessed for serum estradiol and progesterone between cycle days 20 to 24 [194]. Neither mean estradiol nor progesterone levels were associated with hip or spine aBMD [194].  Differences in study design may provide some explanation for the contradictory results. First, the length of time that menstrual cycles are monitored is critical as characteristics of the menstrual cycle show considerable intra-individual variation, particularly for LPL [140-142]. Indeed the majority of studies that monitored cycles for a short duration of time (three to six months) did not observe associations between subclinical ovulatory disturbances and aBMD [50-52,194]. On the other hand, studies that monitored more cycles (9.8 to 12 cycles over one to two years) did find that ovulatory disturbances were associated with less positive changes in bone [45-46,49]. Therefore, continuous or long-term monitoring is critical to correctly identify women with luteal phase defects and thus reduced exposure to estradiol and progesterone. Second, some investigated aBMD via dual energy X-ray absorptiometry (DXA) cross-sectionally [50-51,194] and/or assessed aBMD prior to evaluation of ovulatory function [50-51]. On the     26 other hand, others assessed ovulatory function at the same time as changes in BMD using quantitative computed tomography (QCT) [45-47] or 2-year aBMD change by DXA [49]. Examining ovulatory function and bone density at the same time is important as it has been shown that ovulatory characteristics change over time [47,193], and even change cycle by cycle within women [46-47,193]. The method of bone density assessment is also significant to consider. DXA assesses both cancellous and cortical bone and QCT assesses only cancellous bone which turns over more rapidly than cortical bone. This may be important when monitoring bone density over shorter periods of time.   1.2.7 Bone and CDR  Many studies have found that clinical eating disorders, most notably anorexia nervosa, have a substantial negative effect on bone [195]. However, very few studies have examined the effect of non-clinical disordered eating attitudes and behaviours, which are encountered frequently in the general population of young women [1-3]. Given the above associations (between CDR and each of cortisol and ovulatory function, between cortisol and bone, and between ovulatory function and bone), it is logical to ask whether high levels of CDR are associated with lower bone mass or density. The small number of cross-sectional studies that have compared BMC and/or aBMD between women with higher and lower restraint report conflicting results. Due to the large inter-individual variation in bone density and considering the small influence that eating attitudes would exert on bone, cross-sectional studies would require very large sample sizes, and the cross-sectional studies conducted to date are likely insufficiently powered. Barr and coworkers were the first to examine CDR and bone in a cross-sectional study of 27 regularly menstruating, ovulatory women of normal and stable weight categorised to upper and lower tertiles of TFEQ-R score [42]. There was no difference in aBMD of the lumbar spine assessed by either QCT or DXA [42]. Barr and colleagues followed that study by examining CDR and bone in a sample of 62 normally active, regularly menstruating women of normal and stable weight [53]. Women with higher CDR had significantly lower total body BMC than those with lower CDR [53]. As well, in multiple regression analysis, TFEQ-R score was a significant predictor of both total body BMC and aBMD, explaining approximately 5.5% of variation [53]. Dietary restraint did not enter the multiple regression equation for lumbar spine aBMD, although its effect approached significance (P=0.070) [53]. In a larger study (n=185) with greater power to detect differences in bone, premenopausal women with higher restraint levels were found to have lower total body BMC (but not aBMD) than those with lower dietary restraint in three of four body weight groups [54]. The same group found a significant correlation between TFEQ-R score and femoral BMC (r= -0.24) among 78 generally healthy, premenopausal obese women     27 with very high RS scores [52]. However, when grouped as normal versus osteoporotic aBMD, there was no difference in TFEQ-R score between groups [52].  Further evidence of a relationship between CDR and bone comes from a study of women runners [196]. Those with a history of stress fracture had significantly higher TFEQ-R scores than those without stress fractures, yet ran similar distances and were of similar relative weight [196]. Support for the relationship between CDR and bone is also available from studies examining this relationship using other tools to assess different aspects of eating attitudes. Competitive women runners with high levels of weight preoccupation (assessed using the Eating Disorder Inventory-2 subscale scores for Drive For Thinness, Bulimia and Body Dissatisfaction) had lower spinal aBMD than those with normal scores [197]. Adolescent women runners with higher restraint scores (determined by the Eating Disorder Exam (EDE) questionnaire Eating Restraint subscale) had lower lumbar spine BMC and aBMD than runners with elevated Weight or Shape Concern (additional subscales) or normal EDE scores, after controlling for potentially confounding variables [55]. However, findings from that study are difficult to interpret due to inclusion of girls with menstrual cycle irregularities, which were more common among those with high restraint. Lastly, in a study of 100 regularly menstruating, university-aged women, those with DEBQ-R scores above the median and who were not using oral contraceptives had lower tibial SOS by quantitative ultrasound and lower markers of bone formation than women with lower restraint [198]. Other bone turnover markers and SOS at the radius did not differ by CDR level [198].  Three recent cross-sectional studies have examined bone in relation to TFEQ-R scores. The first study included 65 regularly menstruating, university-aged women with normal and stable weight, normal activity levels and no history of eating disorders [56]. While TFEQ-R scores were not associated with BMC or aBMD at any site measured, TFEQ-R score was inversely associated with markers of bone turnover [56]. This suggests a reduced rate of bone turnover, which if maintained throughout adulthood, could potentially impact BMC or aBMD in later life. In a sample of 77 normal-weight post-menopausal women not using hormone therapy, there was no difference in aBMD measurements by level of CDR [17]. However, this study was not powered to detect a difference in bone [17]. The most recent study involved 84 physically active, university-aged women of normal and stable weight with no history of an eating disorder [40]. Lower total body and lumbar spine aBMD was found among women with higher versus lower CDR [40]. Furthermore, an inverse correlation between TFEQ-R scores and aBMD at the total body, lumbar spine, total hip and femoral neck was observed [40]. However, interpretation of these findings is complicated by the inclusion of women with oligomenorrhea, as abnormal cycle length was significantly more prevalent in the high CDR group [40].     28 By examining bone prospectively, a smaller sample size may be required to detect a difference in BMC/aBMD. This is because the inter-individual variability resulting from the many genetic and lifestyle variables that affect bone, is reduced. At the time my research project was proposed, only one study had prospectively examined eating attitudes in relation to bone [199]. In this 2-year study of bone mineral acquisition in 45 healthy premenarcheal girls, those with high scores on the Children‘s Eating Attitude Test Oral Control subscale had lower total body BMC at baseline and lower total body and spinal aBMD at two years, when height, weight and Tanner breast maturation stage were included as covariates [199]. Multiple regression analysis, controlling for the same covariates, found that Children‘s Eating Attitude Test Oral Control score predicted baseline, 2-year and 2-year change in total body and spinal BMC, explaining 0.9% to 7.6% of the variance [199]. Oral control reflects the perception of being able to control or limit food intake. Although not completely synonymous with CDR, the data support an association between eating attitudes and bone health starting at a young age, and before most were menstruating. In contrast, the prospective study examining associations among CDR, ovulatory function and bone described in detail above (Bone & Ovulatory function section) found that CDR had no significant effect on baseline aBMD or 2-year aBMD change in mixed-model analyses after adjustment for BMI and activity levels [49]. Taken together, available data are suggestive of a relationship between CDR and BMC/aBMD in healthy young women but are far from conclusive. Additional prospective studies may provide clarification as to the possibility of a direct association between CDR and bone.   1.3 Gaps in our current understanding   The experience of dietary restraint may act as a subtle but chronic stressor among young women that is capable of activating the physiological stress response including the HPA axis. Elevated yet physiologically normal cortisol levels may have the potential to negatively influence ovulatory function and bone density. Prospective examination of these relationships is required for several reasons. First, cortisol secretion is highly variable and is affected by everyday minor stressful events. Therefore, repeated longitudinal assessment is crucial to correctly capturing individuals‘ ―usual‖ levels. Most of the previous cross-sectional studies to date have used single cortisol assessments (some which are also limited by the timing of collections) and a few used two assessments three to six months apart. Secondly, it is well recognised that there is considerable within-person variability in the characteristics of the menstrual cycle. For that reason, long-term monitoring of ovulatory function is essential before women can be categorised as to their experience of subclinical ovulatory disturbances. The majority of studies to date monitoring ovulatory function concurrently with bone density have observed two to four cycles and only two have assessed a maximum of 12 cycles. As most     29 methods of monitoring ELA are expensive and have high participant burden, a method that is inexpensive, accurate and acceptable to women is required for long-term observation in large numbers of women. Finally, the majority of studies completed to date were cross-sectional in nature and often included a small number of participants. Therefore, these studies were likely not powered to see associations between bone density and either cortisol levels or ovulatory function due to the high inter-individual variability in bone density.  Only one study to date has prospectively examined CDR, ovulatory function and change in bone density [49]. The study included a sample of women with higher relative body weight that was considerably older and more gynaecologically mature than groups studied in previous work–factors which may affect bone and the characteristics of the menstrual cycle. Moreover, this study monitored a maximum of 12 menstrual cycles and did not assess cortisol levels, which is hypothesised to be the mediator in the relationship between both CDR and ovulatory function, and CDR and bone. My PhD study was designed to prospectively examine the associations among CDR, ovulatory function, cortisol and change in bone density over two years. Findings addressed the gaps in our current understanding outlined above, providing additional evidence as to whether or not the experience of CDR influences young women‘s health outcomes. Secondary objectives included examination of the associations among eating and body attitudes, cortisol and blood pressure, which are discussed in more detail in the appropriate chapters.   1.4 Study purpose The primary purpose of this research project was to prospectively explore relationships among CDR, UFC and, subclinical ovulatory disturbances, and the association of each of these variables with change in bone density over two years in healthy premenopausal women. Before conducting the 2-year prospective study, it was necessary to further validate LS-QBT against progesterone, an established indirect indicator of ovulation which may also be associated with bone density. If valid, this method would be used to document ovulatory function in a large number of women in relation to bone density over two years. This validation study is discussed in detail in Chapter 2, prior to presentation of the 2-year study in Chapter 3. As chronic psychosocial stress may also detrimentally affect blood pressure, associations among eating and body attitudes, cortisol and blood pressure were also examined, and are reported in Chapter 4. Baseline findings regarding the relationships among eating attitudes, cortisol and bone density are described in Chapter 5.      30 1.4.1 Objectives 1.4.1.1 Objectives for Chapter 2 1. To compare computerised least-squares analysis of quantitative basal temperature (LS-QBT) to urinary pregnanediol glucuronide (PdG) for detecting evidence of luteal activity, as reflected by menstrual cycles classified as ovulatory versus anovulatory. 2. To compare LS-QBT to PdG in estimating the day of luteal phase onset as reflected by the day of significant temperature rise relative to the day of a sustained PdG rise.  3. To evaluate whether editing temperatures based on wake-time variation prior to LS-QBT improves the performance of the method relative to PdG classification of cycles as ovulatory or anovulatory and estimation of the day of luteal phase onset. 4. To evaluate whether assessment and editing of temperature records by a reproductive expert prior to LS-QBT improves the performance of the method relative to PdG classification of cycles as ovulatory or anovulatory and estimation of the day of luteal phase onset.   1.4.1.2 Objectives for Chapter 3 1. To examine potential relationships among the following study outcome variables in healthy, non-obese, regularly menstruating women: a. CDR score (TFEQ-Restraint score averaged from assessments at baseline and both follow-ups); b. Dietary intake of bone-related nutrients and physical activity averaged from assessments at baseline and both follow-ups; c. 24-hour urinary free cortisol (UFC) averaged from assessments at baseline and both follow-ups;  d. The frequency of subclinical ovulatory disturbances (%SOD, anovulation and/or luteal phase <10 days long) by LS-QBT analyses;  e. Anthropometric measurements and 2-year change in anthropometrics (Δanthropometrics); and f. 2-year change in areal bone mineral density (ΔaBMD; g/cm2) measured at the lumbar spine, both total hips and whole body. 2. To examine whether energy intake, physical activity, General Stress (based on standardised Z-scores of the Perceived Stress Scale and Daily Stress Inventory completed on the days of urine collection), anthropometric measurements, Δanthropometrics, UFC, %SOD, and ΔaBMD differ between healthy, non-obese, regularly menstruating women with higher and lower CDR (median split).      31 3. To examine whether energy intake, physical activity, General Stress, CDR, anthropometric measurements, Δanthropometrics, UFC and ΔaBMD differ between healthy, non-obese, regularly menstruating women with a higher versus lower %SOD (median split).  4. To examine the interactive effect of CDR-by-ethnicity on UFC, %SOD and ΔaBMD, and the interactive effect of %SOD-by-ethnicity on UFC and ΔaBMD.  1.4.1.3 Objectives for Chapter 4 All objectives for Chapter 4 are based on data collected at the first follow-up, approximately six to 12 months (average seven) after the baseline assessment. 1. To examine the main and interactive effects of Eating/Body Attitude level (based on standardised Z-scores of the Three Factor Eating Questionnaire subscales, Body Shape Questionnaire, Beliefs About Appearance Questionnaire, and the Drive For Thinness and Bulimia subscales of the Eating Disorder Inventory-2) and current weight loss effort among young, non-obese, regularly menstruating women on: a. BMI, energy intake, physical activity and General Stress (based on standardised Z-scores of the Perceived Stress Scale and Daily Stress Inventory completed on the days of urine collection and blood pressure monitoring); b. UFC; and c. 12-hour daytime average mean arterial pressure and systolic and diastolic ambulatory blood pressure (ABP). 2. To examine potential associations among Eating/Body Attitudes, General Stress, UFC and 12-hour ABP measures, after adjustment for potentially confounding variables.  1.4.1.4 Objectives for Chapter 5 All objectives for Chapter 5 are based on data collected at the baseline assessment. 1. To examine potential cross-sectional associations among aBMD, bone mineral content (BMC, g), and bone area (cm2) measured at the lumbar spine, both total hips and whole body and the following outcome variables in healthy, non-obese, regularly menstruating women: a. CDR; b. Perceived stress, physical activity, the duration of previous oral contraceptive use, age, age of menarche, reported intake of bone-related nutrients and anthropometric measurements; and  c. UFC, before and after adjustment for potentially confounding covariates.     32 2. To examine potential cross-sectional correlations among UFC and the following outcome variables in healthy, non-obese, regularly menstruating women: a. CDR, perceived stress and physical activity; and b. Anthropometric measurements. 3. To examine potential cross-sectional correlations among perceived stress and the following outcome variables in healthy, non-obese, regularly menstruating women: a. CDR and physical activity; and b. Anthropometric measurements.  1.4.2 Hypotheses Hypotheses are stated in the null form rather than directional and will be tested using two-tailed P-values.  1.4.2.1 Hypotheses for Chapter 2 1. There will be no relationship between LS-QBT and PdG in terms of the proportion of cycles classified as ovulatory versus anovulatory. 2. There will be no relationship between LS-QBT and PdG for the estimated day of luteal phase onset. 3. Editing temperatures based on waking time will have no effect on the performance of LS-QBT relative to PdG in terms of detecting ovulatory versus anovulatory cycles, or in estimation of the day of luteal phase onset. 4. The assessment and editing of temperatures by a reproductive expert will have no effect on the performance of LS-QBT relative to PdG in terms of detecting ovulatory versus anovulatory cycles, or in estimation of the day of luteal phase onset.  1.4.2.2 Hypotheses for Chapter 3 1. There will be no relationships among CDR, intakes of bone-related nutrients or physical activity, UFC, %SOD, Δanthropometrics and ΔaBMD at any measured site and UFC. 2. Women with higher and lower CDR will not differ with regard to energy intake, physical activity, General Stress, anthropometrics, Δanthropometrics, UFC, %SOD and ΔaBMD. 3. Women with higher and lower %SOD will not differ with regard to energy intake, physical activity, General Stress, anthropometrics, Δanthropometrics, UFC, CDR and ΔaBMD. 4. There will be no ethnicity-by-CDR interactive effect on UFC, %SOD or ΔaBMD, or ethnicity-by-%SOD interactive effect on UFC or ΔaBMD.       33 1.4.2.3 Hypotheses for Chapter 4 1. There will be no main or interactive effects of Eating/Body Attitude level or current weight loss effort on BMI, energy intake, physical activity, General Stress, UFC and 12-hour daytime average ABP measures. 2. There will be no cross-sectional relationships among Eating/Body Attitudes, General Stress, UFC and 12-hour daytime average ABP measures after adjustment for potentially confounder variables.  1.4.2.4 Hypotheses for Chapter 5 1. There will be no cross-sectional relationships among aBMD, BMC and bone area measured at any site and the following outcome variables: CDR, perceived stress, physical activity, the duration of previous oral contraceptive use, age, age of menarche, reported intake of bone-related nutrients, anthropometric measurements and UFC. 2. There will be no cross-sectional relationships among UFC and the following outcome variables: CDR, perceived stress, physical activity and anthropometric measurements. 3. There will be no cross-sectional relationships among perceived stress and the following outcome variables: CDR, physical activity and anthropometric measurements.      34 1.5 References [1] Garner DM. The 1997 body image survey results. Psychol Today 1997;30:30-44,75-80,84.  [2] Groesz LM, Levine MP, Murnen SK. The effect of experimental presentation of thin media images on body satisfaction: a meta-analytic review. Int J Eat Disord 2002;31:1-16.  [3] Thompson JK, Heinberg LJ, Altabe MN, Tantleff-Dunn S. Exacting beauty: theory, assessment, and treatment of body image disturbance, 1st edition. Washington DC: American Psychological Association; 1999.  [4] Green KL, Cameron R, Polivy J, Cooper K, Liu L, Leiter L, Heatherton T. Weight dissatisfaction and weight loss attempts among Canadian adults. Canadian Heart Health Surveys Research Group. CMAJ 1997;157:S17-25.  [5] Jankauskiene R, Kardelis K, Pajaujiene S. Body weight satisfaction and weight loss attempts in fitness activity involved women. J Sports Med Phys Fitness 2005;45:537-45.  [6] Lemon SC, Rosal MC, Zapka J, Borg A, Andersen V. Contributions of weight perceptions to weight loss attempts: differences by body mass index and gender. Body Image 2009;6:90-6.  [7] Malinauskas BM, Raedeke TD, Aeby VG, Smith JL, Dallas MB. Dieting practices, weight perceptions, and body composition: a comparison of normal weight, overweight, and obese college females. Nutr J 2006;5:11.  [8] McElhone S, Kearney JM, Giachetti I, Zunft HJ, Martinez JA. Body image perception in relation to recent weight changes and strategies for weight loss in a nationally representative sample in the European Union. Public Health Nutr 1999;2:143-51.  [9] Millstein RA, Carlson SA, Fulton JE, Galuska DA, Zhang J, Blanck HM, Ainsworth BE. Relationships between body size satisfaction and weight control practices among US adults. Medscape J Med 2008;10:119.  [10] Stice E, Shaw HE. Role of body dissatisfaction in the onset and maintenance of eating pathology: a synthesis of research findings. J Psychosom Res 2002;53:985-93.  [11] Alexander JM, Tepper BJ. Use of reduced-calorie/reduced-fat foods by young adults: influence of gender and restraint. Appetite 1995;25:217-30. [12] McLean JA, Barr SI. Cognitive dietary restraint is associated with eating behaviors, lifestyle practices, personality characteristics and menstrual irregularity in college women. Appetite 2003;40:185-92.  [13] Stunkard AJ, Messick S. The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. J Psychosom Res 1985;29:71-83.      35 [14] Lowe MR, Annunziato RA, Markowitz JT, Didie E, Bellace DL, Riddell L, Maille C, McKinney S, Stice E. Multiple types of dieting prospectively predict weight gain during the freshman year of college. Appetite 2006;47:83-90.  [15] Stice E, Cooper JA, Schoeller DA, Tappe K, Lowe MR. Are dietary restraint scales valid measures of moderate- to long-term dietary restriction? Objective biological and behavioral data suggest not. Psychol Assess 2007;19:449-58.  [16] Chrousos GP. Stress and disorders of the stress system. Nat Rev Endocrinol 2009;5:374-81.  [17] Rideout CA, Linden W, Barr SI. High cognitive dietary restraint is associated with increased cortisol excretion in postmenopausal women. J Gerontol A Biol Sci Med Sci 2006;61:628-33.  [18] Anderson DA, Shapiro JR, Lundgren JD, Spataro LE, Frye CA. Self-reported dietary restraint is associated with elevated levels of salivary cortisol. Appetite 2002;38:13-7.  [19] McLean JA, Barr SI, Prior JC. Cognitive dietary restraint is associated with higher urinary cortisol excretion in healthy premenopausal women. Am J Clin Nutr 2001;73:7-12.  [20] Putterman E, Linden W. Cognitive dietary restraint and cortisol: importance of pervasive concerns with appearance. Appetite 2006;47:64-76.  [21] Rutters F, Nieuwenhuizen AG, Lemmens SG, Born JM, Westerterp-Plantenga MS. Hyperactivity of the HPA axis is related to dietary restraint in normal weight women. Physiol Behav 2009;96:315-9. [22] Cooper MS. Sensitivity of bone to glucocorticoids. Clin Sci (Lond) 2004;107:111-23.  [23] Charmandari E, Tsigos C, Chrousos G. Endocrinology of the stress response. Ann Rev Physiol 2005;67:259-84. [24] Black PH. The inflammatory consequences of psychologic stress: relationship to insulin resistance, obesity, atherosclerosis and diabetes mellitus, Type II. Med Hypotheses 2006;67:879-91.  [25] Cetin A, Gokce-Kutsal Y, Celiker R. Predictors of bone mineral density in healthy males. Rheumatol Int 2001;21:85-8.  [26] Dennison E, Hindmarsh P, Fall C, Kellingray S, Barker D, Phillips D, Cooper C. Profiles of endogenous circulating cortisol and bone mineral density in healthy elderly men. J Clin Endocrinol Metab 1999;84:3058-63.  [27] Raff H, Raff JL, Duthie EH, Wilson CR, Sasse EA, Rudman I, Mattson D. Elevated salivary cortisol in the evening in healthy elderly men and women: correlation with bone mineral density. J Gerontol A Biol Sci Med Sci 1999;54:M479-83.      36 [28] Reynolds RM, Dennison EM, Walker BR, Syddall HE, Wood PJ, Andrew R, Phillips DI, Cooper C. Cortisol secretion and rate of bone loss in a population-based cohort of elderly men and women. Calcif Tissue Int 2005;77:134-8.  [29] Altindag O, Altindag A, Asoglu M, Gunes M, Soran N, Deveci Z. Relation of cortisol levels and bone mineral density among premenopausal women with major depression. Int J Clin Pract 2007;61:416-20.  [30] Biller BM, Saxe V, Herzog DB, Rosenthal DI, Holzman S, Klibanski A. Mechanisms of osteoporosis in adult and adolescent women with anorexia nervosa. J Clin Endocrinol Metab 1989;68:548-54.  [31] Kahl KG, Rudolf S, Stoeckelhuber BM, Dibbelt L, Gehl HB, Markhof K, Hohagen F, Schweiger U. Bone mineral density, markers of bone turnover, and cytokines in young women with borderline personality disorder with and without comorbid major depressive disorder. Am J Psychiatry 2005;162:168-74.  [32] Legroux-Gerot I, Vignau J, D'Herbomez M, Collier F, Marchandise X, Duquesnoy B, Cortet B. Evaluation of bone loss and its mechanisms in anorexia nervosa. Calcif Tissue Int 2007;81:174-82.  [33] Misra M, Miller KK, Almazan C, Ramaswamy K, Lapcharoensap W, Worley M, Neubauer G, Herzog DB, Klibanski A. Alterations in cortisol secretory dynamics in adolescent girls with anorexia nervosa and effects on bone metabolism. J Clin Endocrinol Metab 2004;89:4972-80.  [34] Naessen S, Carlstrom K, Glant R, Jacobsson H, Hirschberg AL. Bone mineral density in bulimic women--influence of endocrine factors and previous anorexia. Eur J Endocrinol 2006;155:245-51.  [35] Eskandari F, Martinez PE, Torvik S, Phillips TM, Sternberg EM, Mistry S, Ronsaville D, Wesley R, Toomey C, Sebring NG, Reynolds JC, Blackman MR, Calis KA, Gold PW, Cizza G; Premenopausal, Osteoporosis Women, Alendronate, Depression (POWER) Study Group. Low bone mass in premenopausal women with depression. Arch Intern Med 2007;167:2329-36.  [36] Petronijevic M, Petronijevic N, Ivkovic M, Stefanovic D, Radonjic N, Glisic B, Ristic G, Damjanovic A, Paunovic V. Low bone mineral density and high bone metabolism turnover in premenopausal women with unipolar depression. Bone 2008;42:582-90.  [37] Yazici AE, Bagis S, Tot S, Sahin G, Yazici K, Erdogan C. Bone mineral density in premenopausal women with major depression. Joint Bone Spine 2005;72:540-3.  [38] Berga SL, Loucks TL. Use of cognitive behavior therapy for functional hypothalamic amenorrhea. Ann N Y Acad Sci 2006;1092:114-29.      37 [39] Hontscharuk R, O‘Donnell E, Williams NI, Burke T, De Souza MJ. Dietary cognitive restraint: a marker for altered energy homeostasis and menstrual disturbances in athletic women. Med Sci Sports Exerc 2004;36:abstract 0219.  [40] Vescovi JD, Scheid JL, Hontscharuk R, De Souza MJ. Cognitive dietary restraint: impact on bone, menstrual and metabolic status in young women. Physiol Behav 2008;95:48-55.  [41] Barr SI, Janelle KC, Prior JC. Vegetarian vs nonvegetarian diets, dietary restraint, and subclinical ovulatory disturbances: prospective 6-mo study. Am J Clin Nutr 1994;60:887-94.  [42] Barr SI, Prior JC, Vigna YM. Restrained eating and ovulatory disturbances: possible implications for bone health. Am J Clin Nutr 1994;59:92-7.  [43] Schweiger U, Tuschl RJ, Platte P, Broocks A, Laessle RG, Pirke KM. Everyday eating behavior and menstrual function in young women. Fertil Steril 1992;57:771-5.  [44] Balasch J. Sex steroids and bone: current perspectives. Hum Reprod Update 2003;9:207-22.  [45] Petit MA, Prior JC, Barr SI. Running and ovulation positively change cancellous bone in premenopausal women. Med Sci Sports Exerc 1999;31:780-7.  [46] Prior JC, Vigna YM, Schechter MT, Burgess AE. Spinal bone loss and ovulatory disturbances. N Engl J Med 1990;323:1221-7.  [47] Prior JC, Vigna YM, Barr SI, Kennedy S, Schulzer M, Li DK. Ovulatory premenopausal women lose cancellous spinal bone: a five year prospective study. Bone 1996;18:261-7.  [48] Sowers M, Randolph JF, Crutchfield M, Jannausch ML, Shapiro B, Zhang B, La Pietra M. Urinary ovarian and gonadotropin hormone levels in premenopausal women with low bone mass. J Bone Miner Res 1998;13:1191-202.  [49] Waugh EJ, Polivy J, Ridout R, Hawker GA. A prospective investigation of the relations among cognitive dietary restraint, subclinical ovulatory disturbances, physical activity, and bone mass in healthy young women. Am J Clin Nutr 2007;86:1791-801.  [50] De Souza MJ, Miller BE, Sequenzia LC, Luciano AA, Ulreich S, Stier S, Prestwood K, Lasley BL. Bone health is not affected by luteal phase abnormalities and decreased ovarian progesterone production in female runners. J Clin Endocrinol Metab 1997;82:2867-76.  [51] Waller K, Reim J, Fenster L, Swan SH, Brumback B, Windham GC, Lasley B, Ettinger B, Marcus R. Bone mass and subtle abnormalities in ovulatory function in healthy women. J Clin Endocrinol Metab 1996;81:663-8.      38 [52] Bacon L, Stern JS, Keim NL, Van Loan MD. Low bone mass in premenopausal chronic dieting obese women. Eur J Clin Nutr 2004;58:966-71.  [53] McLean JA, Barr SI, Prior JC. Dietary restraint, exercise, and bone density in young women: are they related? Med Sci Sports Exerc 2001;33:1292-6.  [54] Van Loan MD, Keim NL. Influence of cognitive eating restraint on total-body measurements of bone mineral density and bone mineral content in premenopausal women aged 18-45 y: a cross-sectional study. Am J Clin Nutr 2000;72:837-43.  [55] Barrack MT, Rauh MJ, Barkai HS, Nichols JF. Dietary restraint and low bone mass in female adolescent endurance runners. Am J Clin Nutr 2008;87:36-43.  [56] Nickols-Richardson SM, Beiseigel JM, Gwazdauskas FC. Eating restraint is negatively associated with biomarkers of bone turnover but not measurements of bone mineral density in young women. J Am Diet Assoc 2006;106:1095-101.  [57] Miller G, Chen E, Cole SW. Health psychology: developing biologically plausible models linking the social world and physical health. Ann Rev Psychol 2009;60:501-24.  [58] Osteoporosis Canada, editors. Breaking barriers not bones: 2008 National Report Card on Osteoporosis Care [Internet]. Toronto: Osteoporosis Canada [cited March 2010]. Available from: http://www.osteoporosis.ca/index.php/ci_id/8867/la_id/1.htm.  [59] Bonjour JP, Chevalley T, Ferrari S, Rizzoli R. The importance and relevance of peak bone mass in the prevalence of osteoporosis. Salud Publica Mex 2009;51:S5-17.  [60] Ruderman AJ. Dietary restraint: a theoretical and empirical review. Psychol Bull 1986;99:247-62.  [61] Ruderman AJ. Dysphoric mood and overeating: a test of restraint theory's disinhibition hypothesis. J Abnorm Psychol 1985;94:78-85.  [62] Herman CP, Polivy J. Anxiety, restraint, and eating behavior. J Abnorm Psychol 1975;84:66-72.  [63] Van Strein T, Frijters JER, Berger GPA, Defares PB. The Dutch Eating Behavior Questionnaire (DEBQ) for assessment of restrained, emotional, and external eating behavior. Int J Eat Disord 1986;5:295-315.  [64] Gorman BS, Allison DB. Measures of restrained eating. In: Allison DB, editor. Handbook of assessment methods for eating behaviors and weight-related problems. California: Sage Publications; 1995. p. 208-25.  [65] Herman CP, Polivy J. Restrained eating. In: Stunkard AJ, editor. Obesity. Philadelphia: Saunders; 1980. p. 208-25. [66] Schachter S. Obesity and eating. Internal and external cues differentially affect the eating behavior of obese and normal subjects. Science 1968;161:751-6.      39 [67] Schachter S, Goldman R, Gordon A. Effects of fear, food deprivation, and obesity on eating. J Pers Soc Psychol 1968;10:91-7.  [68] Nisbett RE. Hunger, obesity, and the ventromedial hypothalamus. Psychol Rev 1972;79:433-53.  [69] Schachter S. Some extraordinary facts about obese humans and rats. Am Psychol 1971;26:129-44.  [70] Herman CP, Mack D. Restrained and unrestrained eating. J Pers 1975;43:647-60.  [71] Polivy J, Herman CP. Clinical depression and weight change: a complex relation. J Abnorm Psychol 1976;85:338-40.  [72] Polivy J, Herman CP. Effects of alcohol on eating behavior: influence of mood and perceived intoxication. J Abnorm Psychol 1976;85:601-6.  [73] Polivy J, Schueneman AL, Carlson K. Alcohol and tension reduction: cognitive and physiological effects. J Abnorm Psychol 1976;85:595-600.  [74] Polivy J, Herman CP, Howard KI. Restraint scale: assessment of dieting. In: Bellack AS, Hersen M, editors. Dictionary of behavioral assessment techniques. New York: Pergamon; 1988. p. 377-80.  [75] Allison DB, Kalinsky LB, Gorman BS. A comparison of the psychometric properties of three measures of dietary restraint. Psyc Assess 1992;4:391-8. [76] Drewnowski A, Riskey D, Desor JA. Measures of restraint: separating dieting from overweight. Appetite 1982;3:282.  [77] Hibscher JA, Herman CP. Obesity, dieting, and the expression of "obese" characteristics. J Comp Physiol Psychol 1977;91:374-80.  [78] Klesges RC. Personality and obesity: global versus specific measures? Behav Assess 1984;6:347-56.  [79] Wardle J. Dietary restraint and binge eating. Behav Anal Mod 1980;4:201-9.  [80] Blanchard FA, Frost RO. Two factors of restraint: concern for dieting and weight fluctuation. Behav Res Ther 1983;21:259-67.  [81] Johnson F, Wardle J. Dietary restraint, body dissatisfaction, and psychological distress: a prospective analysis. J Abnorm Psychol 2005;114:119-25.  [82] Johnson WG, Lake L, Mahan JM. Restrained eating: measuring an elusive construct. Addict Behav 1983;8:413-8. [83] Lowe MR. Dietary concern, weight fluctuation and weight status: further explorations of the restraint scale. Behav Res Ther 1984;22:243-8.  [84] Ruderman AJ, Christensen H. Restraint theory and its applicability to overweight individuals. J Abnorm Psychol 1983;92:210-5.      40 [85] Heatherton TF, Herman CP, Polivy J, King GA, McGree ST. The (mis)measurement of restraint: an analysis of conceptual and psychometric issues. J Abnorm Psychol 1988;97:19-28. [86] Pudel V, Metzdorff M, Oetting M. The personality of obese persons in psychological tests with special consideration on latent obesity. Psychosom Med Psychoanal 1975;21:345-61.  [87] Berkowitz R, Stunkard AJ, Stallings VA. Binge-eating disorder in obese adolescent girls. Ann N Y Acad Sci 1993;699:200-6.  [88] Lawson OJ, Williamson DA, Champagne CM, DeLany JP, Brooks ER, Howat PM, Wozniak PJ, Bray GA, Ryan DH. The association of body weight, dietary intake, and energy expenditure with dietary restraint and disinhibition. Obes Res 1995;3:153-61.  [89] Marcus MD, Wing RR, Lamparski DM. Binge eating and dietary restraint in obese patients. Addict Behav 132;10:163-8.  [90] Stunkard AJ, Wadden TA. Restrained eating and human obesity. Nutr Rev 1990;48:78,86,114-31.  [91] Laessle RG, Tuschl RJ, Kotthaus BC, Pirke KM. A comparison of the validity of three scales for the assessment of dietary restraint. J Abnorm Psychol 1989;98:504-7.  [92] Westenhoefer J. Dietary restraint and disinhibition: is restraint a homogeneous construct? Appetite 1991;16:45-55.  [93] Lowe MR, Kleifield EI. Cognitive restraint, weight suppression, and the regulation of eating. Appetite 1988;10:159-68.  [94] Björvell H, Rössner S, Stunkard A. Obesity, weight loss, and dietary restraint. Int J Eat Disord 1986;5:727-34. [95] Ganley RM. Emotional eating and how it relates to dietary restraint, disinhibition, and perceived hunger. Int J Eat Disord 1988;7:635-47. [96] Westenhoefer J, Stunkard AJ, Pudel V. Validation of the flexible and rigid control dimensions of dietary restraint. Int J Eat Disord 1999;26:53-64.  [97] Timko CA, Perone J. Rigid and flexible control of eating behavior in a college population. Eat Behav 2005;6:119-25.  [98] Williamson DA, Martin CK, York-Crowe E, Anton SD, Redman LM, Han H, Ravussin E. Measurement of dietary restraint: validity tests of four questionnaires. Appetite 2007;48:183-92.  [99] Lowe MR, Timko CA. What a difference a diet makes: towards an understanding of differences between restrained dieters and restrained nondieters. Eat Behav 2004;5:199-208.      41 [100] Rideout CA, Barr SI. "Restrained eating" vs "trying to lose weight": how are they associated with body weight and tendency to overeat among postmenopausal women? J Am Diet Assoc 2009;109:890-3.  [101] de Lauzon-Guillain B, Basdevant A, Romon M, Karlsson J, Borys JM, Charles MA; FLVS Study Group. Is restrained eating a risk factor for weight gain in a general population? Am J Clin Nutr 2006;83:132-8.  [102] de Lauzon B, Romon M, Deschamps V, Lafay L, Borys JM, Karlsson J, Ducimetiere P, Charles MA; Fleurbaix Laventie Ville Sante Study Group. The three-factor eating questionnaire-R18 is able to distinguish among different eating patterns in a general population. J Nutr 2004;134:2372-80.  [103] Pirke KM, Tuschl RJ, Spyra B, Laessle RG, Schweiger U, Broocks A, Sambauer S, Zitzelsberger G. Endocrine findings in restrained eaters. Physiol Behav 1990;47:903-6.  [104] Rideout CA, McLean JA, Barr SI. Women with high scores for cognitive dietary restraint choose foods lower in fat and energy. J Am Diet Assoc 2004;104:1154-7.  [105] Tuschl RJ, Laessle RG, Platte P, Pirke KM. Differences in food-choice frequencies between restrained and unrestrained eaters. Appetite 1990;14:9-13.  [106] de Castro JM. The relationship of cognitive restraint to the spontaneous food and fluid intake of free-living humans. Physiol Behav 1995;57:287-95.  [107] McGuire MT, Jeffery RW, French SA, Hannan PJ. The relationship between restraint and weight and weight-related behaviors among individuals in a community weight gain prevention trial. Int J Obes Relat Metab Disord 2001;25:574-80.  [108] Provencher V, Drapeau V, Tremblay A, Despres JP, Lemieux S. Eating behaviors and indexes of body composition in men and women from the Quebec Family Study. Obes Res 2003;11:783-92.  [109] Poehlman ET, Viers HF, Detzer M. Influence of physical activity and dietary restraint on resting energy expenditure in young nonobese females. Can J Physiol Pharmacol 1991;69:320-6.  [110] Tepper BJ, Trail AC, Shaffer SE. Diet and physical activity in restrained eaters. Appetite 1996;27:51-64.  [111] Rennie KL, Siervo M, Jebb SA. Can self-reported dieting and dietary restraint identify underreporters of energy intake in dietary surveys? J Am Diet Assoc 2006;106:1667-72.  [112] Stice E, Fisher M, Lowe MR. Are dietary restraint scales valid measures of acute dietary restriction? Unobtrusive observational data suggest not. Psychol Assess 2004;16:51-9.      42 [113] Legg C, Puri A, Thomas N. Dietary restraint and self-reported meal sizes: diary studies with differentially informed consent. Appetite 2000;34:235-43.  [114] Beiseigel JM, Nickols-Richardson SM. Cognitive eating restraint scores are associated with body fatness but not with other measures of dieting in women. Appetite 2004;43:47-53.  [115] Lautenbacher S, Thomas A, Roscher S, Strian F, Pirke KM, Krieg JC. Body size perception and body satisfaction in restrained and unrestrained eaters. Behav Res Ther 1992;30:243-50.  [116] Wardle J, Steptoe A, Oliver G, Lipsey Z. Stress, dietary restraint and food intake. J Psychosom Res 2000;48:195-202.  [117] Savage JS, Birch LL. Patterns of weight control strategies predict differences in women's 4-year weight gain. Obesity 2009:DOI 10.1038/oby.2009.265. [118] Chaput JP, Leblanc C, Perusse L, Despres JP, Bouchard C, Tremblay A. Risk factors for adult overweight and obesity in the Quebec Family Study: Have we been barking up the wrong tree? Obesity 2009;17:1964-70.  [119] Stewart TM, Williamson DA, White MA. Rigid vs. flexible dieting: association with eating disorder symptoms in nonobese women. Appetite 2002;38:39-44.  [120] McEwen BS, Seeman T. Protective and damaging effects of mediators of stress. Elaborating and testing the concepts of allostasis and allostatic load. Ann N Y Acad Sci 1999;896:30-47.  [121] Grossi G, Perski A, Lundberg U, Soares J. Associations between financial strain and the diurnal salivary cortisol secretion of long-term unemployed individuals. Integr Physiol Behav Sci 2001;36:205-19.  [122] Kunz-Ebrecht SR, Kirschbaum C, Steptoe A. Work stress, socioeconomic status and neuroendocrine activation over the working day. Soc Sci Med 2004;58:1523-30.  [123] Steptoe A. Stress, social support and cardiovascular activity over the working day. Int J Psychophysiol 2000;37:299-308.  [124] Powell LH, Lovallo WR, Matthews KA, Meyer P, Midgley AR, Baum A, Stone AA, Underwood L, McCann JJ, Janikula Herro K, Ory MG. Physiologic markers of chronic stress in premenopausal, middle-aged women. Psychosom Med 2002;64:502-9.  [125] Levine A, Zagoory-Sharon OZ, Feldman R, Lewis JG, Weller A. Measuring cortisol in human psychobiological studies. Physio & Behav 2007;90:43-53. [126] Fries L, Dettenborn L, Kirschbaum C. The cortisol awakening response (CAR): facts and future directions. Int J Psychophysiol 2009;72:67-73.     43 [127] Taylor RL, Machacek D, Singh RJ. Validation of a high-throughput liquid chromatography-tandem mass spectrometry method for urinary cortisol and cortisone. Clin Chem 2002;48:1511-9.  [128] Brantley PJ, Dietz LS, McKnight GT, Jones GN, Tulley R. Convergence between the Daily Stress Inventory and endocrine measures of stress. J Consult Clin Psychol 1988;56:549-51.  [129] van Eck M, Berkhof H, Nicolson N, Sulon J. The effects of perceived stress, traits, mood states, and stressful daily events on salivary cortisol. Psychosom Med 1996;58:447-58.  [130] Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav 1983;24:385-96.  [131] Hewitt PL, Flett GL, Mosher SW. The Perceived Stress Scale: factor structure and relation to depression symptoms in a psychiatric sample. J Psychopath Behav Assess 1992;14:247-57.  [132] Martin RA, Kazarian SS, Breiter HJ. Perceived stress, life events, dysfunctional attitudes, and depression in adolescent psychiatric inpatients. J Psychopath Behav Assess 1995;17:81-95.  [133] Brantley PJ, Waggoner CD, Jones GN, Rappaport NB. A Daily Stress Inventory: development, reliability, and validity. J Behav Med 1987;10:61-74. [134] Makras P, Toloumis G, Papadogias D, Kaltsas GA, Besser M. The diagnosis and differential diagnosis of endogenous Cushing‘s syndrome. Hormones 2006;5:231-50.  [135] Therrien F, Drapeau V, Lupien SJ, Beaulieu S, Dore J, Tremblay A, Richard D. Awakening cortisol response in relation to psychosocial profiles and eating behaviors. Physiol Behav 2008;93:282-8.  [136] Ball K, Lee C. Psychological stress, coping, and symptoms of disordered eating in a community sample of young Australian women. Int J Eat Disord 2002;31:71-81.  [137] Cattanach L, Malley R, Rodin J. Psychologic and physiologic reactivity to stressors in eating disordered individuals. Psychosom Med 1988;50:591-9.  [138] Buffet NC, Bouchard P. The neuroendocrine regulation of the human ovarian cycle. Chronobiol Int 2001;18:893-919.  [139] Gordon CM, Nelson LM. Amenorrhea and bone health in adolescents and young women. Curr Opin Obstet Gynecol 2003;15:377-84.  [140] Cole LA, Ladner DG, Byrn FW. The normal variabilities of the menstrual cycle. Fertil Steril 2009;91:522-7.  [141] Creinin MD, Keverline S, Meyn LA. How regular is regular? An analysis of menstrual cycle regularity. Contraception 2004;70:289-92.      44 [142] Fehring RJ, Schneider M, Raviele K. Variability in the phases of the menstrual cycle. J Obstet Gynecol Neonatal Nurs 2006;35:376-84.  [143] Shoupe D, Mishell DR, Lacarra M, Lobo RA, Horenstein J, d'Ablaing G, Moyer D. Correlation of endometrial maturation with four methods of estimating day of ovulation. Obstet Gynecol 1989;73:88-92.  [144] Khalid A. Irregular or absent periods--what can an ultrasound scan tell you? Best Pract Res Clin Obstet Gynaecol 2004;18:3-11.  [145] Baird DD, Weinberg CR, Wilcox AJ, McConnaughey DR, Musey PI. Using the ratio of urinary oestrogen and progesterone metabolites to estimate day of ovulation. Stat Med 1991;10:255-66. [146] Kassam A, Overstreet JW, Snow-Harter C, De Souza MJ, Gold EB, Lasley BL. Identification of anovulation and transient luteal function using a urinary pregnanediol-3-glucuronide ratio algorithm. Environ Health Perspect 1996;104:408-13.  [147] Moghissi KS. Prediction and detection of ovulation. Fertil Steril 1980;34:89-98.  [148] Frank-Herrmann P, Heil J, Gnoth C, Toledo E, Baur S, Pyper C, Jenetzky E, Strowitzki T, Freundl G. The effectiveness of a fertility awareness based method to avoid pregnancy in relation to a couple's sexual behaviour during the fertile time: a prospective longitudinal study. Hum Reprod 2007;22:1310-9.  [149] Lundy LE, Lee SG, Levy W, Woodruff JD, Wu CH, Abdalla M. The ovulatory cycle. A histologic, thermal, steroid, and gonadotropin correlation. Obstet Gynecol 1974;44:14-25.  [150] Ecochard R, Boehringer H, Rabilloud M, Marret H. Chronological aspects of ultrasonic, hormonal, and other indirect indices of ovulation. Br J Obstet Gynaecol 2001;108:822-9.  [151] Guermandi E, Vegetti W, Bianchi MM, Uglietti A, Ragni G, Crosignani P. Reliability of ovulation tests in infertile women. Obstet Gynecol 2001;97:92-6. [152] Guida M, Tommaselli GA, Palomba S, Pellicano M, Moccia G, Di Carlo C, Nappi C. Efficacy of methods for determining ovulation in a natural family planning program. Fertil Steril 1999;72:900-4.  [153] Luciano AA, Peluso J, Koch EI, Maier D, Kuslis S, Davison E. Temporal relationship and reliability of the clinical, hormonal, and ultrasonographic indices of ovulation in infertile women. Obstet Gynecol 1990;75:412-6.  [154] Wetzels LC, Hoogland HJ, de Haan J. Basal body temperature as a method of ovulation detection: comparison with ultrasonographical findings. Gynecol Obstet Invest 1982;13:235-40.      45 [155] Lenton EA, Weston GA, Cooke ID. Problems in using basal body temperature recordings in an infertility clinic. Br Med J 1977;1:803-5.  [156] Morris N, Underwood L, Easterling W. Temporal relationship between basal body temperature nadir and luteinizing hormone surge in normal women. Fertil Steril 1976;27:780-3. [157] Quagliarello J, Arny M. Inaccuracy of basal body temperature charts in predicting urinary luteinizing hormone surges. Fertil Steril 1986;45:334-7. [158] Templeton AA, Penney GC, Lees MM. Relation between the luteinizing hormone peak, the nadir of the basal body temperature and the cervical mucus score. Br J Obstet Gynaecol 1982;89:985-8. [159] Yong EL, Wong PC, Kumar A, Wong YC, Goh HH, Hagglund L, Ratnam S. Simple office methods to predict ovulation: the clinical usefulness of a new urine luteinizing hormone kit compared to basal body temperature, cervical mucus and ultrasound. Aust N Z J Obstet Gynaecol 1989;29:155-60. [160] Hilgers TW, Bailey AJ. Natural family planning. II. Basal body temperature and estimated time of ovulation. Obstet Gynecol 1980;55:333-9.  [161] Bauman JE. Basal body temperature: unreliable method of ovulation detection. Fertil Steril 1981;36:729-33.  [162] Ayres-de-Campos D, Silva-Carvalho JL, Oliveira C, Martins-da-Silva I, Silva-Carvalho J, Pereira-Leite L. Inter-observer agreement in analysis of basal body temperature graphs from infertile women. Hum Reprod 1995;10:2010-6.  [163] Vollman RF. The menstrual cycle. In: Friedman EA, editor. Major problems in obstetrics & gynecology. Toronto: WB Saunders Company; 1977. p. 11-193. [164] Royston JP, Abrams RM. An objective method for detecting the shift in basal body temperature in women. Biometrics 1980;36:217-24.  [165] Prior JC, Vigna YM, Schulzer M, Hall JE, Bonen A. Determination of luteal phase length by quantitative basal temperature methods: validation against the midcycle LH peak. Clin Invest Med 1990;13:123-31.  [166] Nepomnaschy PA, Welch K, McConnell D, Strassmann BI, England BG. Stress and female reproductive function: a study of daily variations in cortisol, gonadotrophins, and gonadal steroids in a rural Mayan population. Am J Hum Biol 2004;16:523-32.  [167] Lebenstedt M, Platte P, Pirke KM. Reduced resting metabolic rate in athletes with menstrual disorders. Med Sci Sports Exerc 1999;31:1250-6.  [168] Scheid JL, Williams NI, West SL, VanHeest JL, De Souza MJ. Elevated PYY is associated with energy deficiency and indices of subclinical disordered eating in exercising women with hypothalamic amenorrhea. Appetite 2009;52:184-92.      46 [169] Garner DM, Garfinkel PE. The Eating Attitudes Test: an index of the symptoms of anorexia nervosa. Psychol Med 1979;9:273-9.  [170] Garner DM. Eating Disorder Inventory-2 professional manual. Florida: Psychological Assessment Resources; 1991.  [171] Laughlin GA, Dominguez CE, Yen SS. Nutritional and endocrine-metabolic aberrations in women with functional hypothalamic amenorrhea. J Clin Endocrinol Metab 1998;83:25-32.  [172] Marcus MD, Loucks TL, Berga SL. Psychological correlates of functional hypothalamic amenorrhea. Fertil Steril 2001;76:310-6.  [173] Schneider LF, Warren MP. Functional hypothalamic amenorrhea is associated with elevated ghrelin and disordered eating. Fertil Steril 2006;86:1744-9.  [174] Schneider LF, Monaco SE, Warren MP. Elevated ghrelin level in women of normal weight with amenorrhea is related to disordered eating. Fertil Steril 2008;90:121-8.  [175] Warren MP, Voussoughian F, Geer EB, Hyle EP, Adberg CL, Ramos RH. Functional hypothalamic amenorrhea: hypoleptinemia and disordered eating. J Clin Endocrinol Metab 1999;84:873-7.  [176] World Health Organization, Task Force on Methods for the Determination of the Fertile Period, Special Programme of Research, Development and Research Training in Human Reproduction. Temporal relationships between ovulation and defined changes in the concentration of plasma estradiol-17 beta, luteinizing hormone, follicle-stimulating hormone, and progesterone. I. Probit analysis. Am J Obstet Gynecol 1980;138:383-90. [177] Cano A, Aliaga R. Characteristics of urinary luteinizing hormone (LH) during the induction of LH surges of different magnitude in blood. Hum Reprod 1995;10:63-7. [178] Loucks AB, Thuma JR. Luteinizing hormone pulsatility is disrupted at a threshold of energy availability in regularly menstruating women. J Clin Endocrinol Metab 2003;88:297-311.  [179] Keizer HA, Rogol AD. Physical exercise and menstrual cycle alterations. What are the mechanisms? Sports Med 1990;10:218-35.  [180] Bonen A. Recreational exercise does not impair menstrual cycles: a prospective study. Int J Sports Med 1992;13:110-20. [181] Rogol AD, Weltman A, Weltman JY, Serp RI, Snead DB, Levine S, Haskvitz EM, Thompson DL, Schurrer R, Dowling E. Durability of the reproductive axis in eumenorrheic women during 1 yr of endurance training. J Appl Physiol 1992;72:1571-80.     47 [182] Chen MD, O'Byrne KT, Chiappini SE, Hotchkiss J, Knobil E. Hypoglycemic 'stress' and gonadotropin-releasing hormone pulse generator activity in the rhesus monkey: role of the ovary. Neuroendocrinology 1992;56:666-73.  [183] Weinstein RS, Chen JR, Powers CC, Stewart SA, Landes RD, Bellido T, Jilka RL, Parfitt AM, Manolagas SC. Promotion of osteoclast survival and antagonism of bisphosphonate-induced osteoclast apoptosis by glucocorticoids. J Clin Invest 2002;109:1041-8.  [184] Godschalk MF, Downs RW. Effect of short-term glucocorticoids on serum osteocalcin in healthy young men. J Bone Miner Res 1988;3:113-5.  [185] Sasaki N, Kusano E, Ando Y, Nemoto J, Iimura O, Ito C, Takeda S, Yano K, Tsuda E, Asano Y. Changes in osteoprotegerin and markers of bone metabolism during glucocorticoid treatment in patients with chronic glomerulonephritis. Bone 2002;30:853-8.  [186] Chiodini I, Torlontano M, Carnevale V, Trischitta V, Scillitani A. Skeletal involvement in adult patients with endogenous hypercortisolism. J Endocrinol Invest 2008;31:267-76.  [187] Greendale GA, Unger JB, Rowe JW, Seeman TE. The relation between cortisol excretion and fractures in healthy older people: results from The MacArthur Studies. J Am Geriatr Soc 1999;47:799-803.  [188] Krassas GE. Endocrine abnormalities in anorexia nervosa. Pediatr Endocrinol Rev 2003;1:46-54.  [189] Mello AA, Mello MF, Carpenter LL, Price LH. Update on stress and depression: the role of the hypothalamic-pituitary-adrenal (HPA) axis. Rev Bras Psiquiatr 2003;25:231-8.  [190] Brooke-Wavell K, Clow A, Ghazi-Noori S, Evans P, Hucklebridge F. Ultrasound measures of bone and the diurnal free cortisol cycle: a positive association with the awakening cortisol response in healthy premenopausal women. Calcif Tissue Int 2002;70:463-8.  [191] Dane C, Dane B, Cetin A, Erginbas M. The role of quantitative ultrasound in predicting osteoporosis defined by dual-energy X-ray absorptiometry in pre- and postmenopausal women. Climacteric 2008;11:296-303.  [192] Prior JC. Progesterone as a bone-trophic hormone. Endocr Rev 1990;11:386-98.  [193] Prior JC, Vigna YM, Barr SI, Rexworthy C, Lentle BC. Cyclic medroxyprogesterone treatment increases bone density: a controlled trial in active women with menstrual cycle disturbances. Am J Med 1994;96:521-30.  [194] Lu LJ, Nayeem F, Anderson KE, Grady JJ, Nagamani M. Lean body mass, not estrogen or progesterone, predicts peak bone mineral density in premenopausal women. J Nutr 2009;139:250-6.      48 [195] Jayasinghe Y, Grover SR, Zacharin M. Current concepts in bone and reproductive health in adolescents with anorexia nervosa. Br J Obstet Gynaecol 2008;115:304-15.  [196] Guest NS, Barr SI. Cognitive dietary restraint is associated with stress fractures in women runners. Int J Sport Nutr Exerc Metab 2005;15:147-59.  [197] Cobb KL, Bachrach LK, Greendale G, Marcus R, Neer RM, Nieves J, Sowers MF, Brown BW, Gopalakrishnan G, Luetters C, Tanner HK, Ward B, Kelsey JL. Disordered eating, menstrual irregularity, and bone mineral density in female runners. Med Sci Sports Exerc 2003;35:711-9.  [198] Di Giovanni G, Roy BD, Gammage KL, Mack D, Klentrou P. Associations of oral contraceptive use and dietary restraint with bone speed of sound and bone turnover in university-aged women. Appl Physiol Nutr Metab 2008;33:696-705.  [199] Barr SI, Petit MA, Vigna YM, Prior JC. Eating attitudes and habitual calcium intake in peripubertal girls are associated with initial bone mineral content and its change over 2 years. J Bone Miner Res 2001;16:940-7.     49 Chapter 2:   Detecting evidence of luteal activity by least-squares quantitative basal temperature analysis against urinary progesterone metabolites and the effect of wake-time variability1                                                  1 A version of this chapter has been published: Bedford JL, Prior JC, Hitchcock CL, Barr SI. Detecting evidence of luteal activity by least-squares quantitative basal temperature analysis against urinary progesterone metabolites and the effect of wake-time variability. Eur J Obstet Gyencol Reprod Biol 2009;146:76-80. Copyright © Elsevier B.V.     50 2.1 Introduction Variability in ovulation frequency and luteal phase duration are characteristics of ovulatory function related to progesterone that are not apparent to women but may be important to health. Beyond implications for fertility, research suggests that normal ovulation and cyclic progesterone levels may benefit bone [1-4]. However, the ability to examine this relationship is limited by the need for inexpensive and minimally demanding methods to estimate ovulation over long periods. The current ‗gold standard‘, daily transvaginal ultrasound, is costly, has a high subject burden and requires extensive training [5]. Repeated blood or urine samples to assess normal cyclic reproductive hormone patterns can be used to indirectly determine if ovulation has occurred by detecting evidence of luteal activity (ELA). However, these methods are costly for large, long-term studies. Therefore, accurate and reliable methods of determining ovulatory function that are inexpensive, easy-to-use and acceptable to participants are required. For decades basal temperature records have been used in combination with changes in cervical mucous as an established fertility-awareness based method of contraception [6]. Basal temperature is an indirect measure of ovulation as a result of the progesterone-induced temperature increase of approximately 0.3°Celsius from the follicular phase, when progesterone is low, to the luteal phase, when progesterone peaks following ovulation [7]. Previous studies that have examined basal temperature as an ovulation indicator have labelled it an inaccurate method relative to determination by ultrasound [8-11], the luteinizing hormone (LH) surge [12-15], or estradiol:progesterone [16]. However, these studies used non-validated visual methods including identification of a nadir prior to the LH surge and subjective determination of a biphasic basal temperature graph by reproductive experts. As experts do not always agree on visual classification [17], even when using uniform criteria [18] and experience is required for interpretation, these methods are not accessible to all researchers examining the role of ovulatory function in women‘s health. Quantitative methods of basal temperature analysis (QBT) are inexpensive and easy-to-use. QBT methods may also be more accurate at predicting the day of luteal transition (DLT) than previous visually determined methods, though little validation work exists. The three QBT methods used by most computerised systems are the Vollman averaging method [19], Royston‘s cumulative sum method [20] and least-squares analysis (LS-QBT) [21]. Previously we observed excellent correlations between the DLT identified by both LS-QBT and Vollman‘s averaging methods, compared with serum LH peak concentration [21]. Because the progesterone rise is an indicator of ovulatory function that may be important to bone [1-4], validation of LS-QBT against progesterone is needed to confirm it is appropriate for use in scientific studies.     51 Another concern with basal temperature methods is the ability to control variables that may alter temperature such as illness and wake-time variation [22]. The LS-QBT method prompts participants to describe illness and sleep disturbances and to record wake-time. Whether a reproductive expert is required to interpret this information has not been examined. Thus, the purpose of the present study was (1) to assess LS-QBT against urinary progesterone metabolites (pregnanediol glucuronide or PdG), and (2) to assess whether LS-QBT is stable to modest wake-time variations.  2.2 Methods 2.2.1 Participants Fifty-three healthy, normal-weight (body mass index 18.5-25 kg/m2) women aged 19 to 34 were recruited from university classes (Appendix 1). Interested women contacted the investigator for details regarding participation (Appendix 2) and were screened for eligibility including: self-reported regular menstrual cycles (menses every 21-35 days), consistent sleep patterns (wake up and go to bed at approximately the same time most days), no use of hormones in the previous six months and no medical conditions that would interfere with study measurements (Appendix 3). All were nulliparous and non-smoking.  2.2.2 Procedures Eligible participants met with an investigator to receive study materials and instructions, complete a demographic questionnaire (Appendix 4) and have anthropometrics measured. Verbal and written instructions were provided for completing daily temperature records (Appendix 5) and collecting a portion of first void urine daily beginning the first day of flow and continuing until menstrual flow began for the next cycle (Appendix 6). The university‘s Clinical Research Ethics Board approved the study protocol (Appendix 7), and all participants provided written informed consent (Appendix 8). Participants were provided travel compensation (Appendix 9), a $20 gift card for their participation (Appendix 10), and their personal results upon completion of study (Appendix 11). Each morning, immediately upon awakening, participants recorded their temperature using a digital thermometer (524052, Becton Dickinson, NJ, USA). They also recorded wake-time, flow status and any illness. Sleep quality was rated on a scale of zero to four. A daily sample of first morning urine was obtained with a labelled sponge vial and stored in the participant‘s freezer. After one full menstrual cycle, participants returned completed materials.        52 2.2.3 Determination of evidence of luteal activity We did not observe ovulation directly by daily ovarian ultrasound and therefore cannot describe cycles as ―ovulatory‖ or ―anovulatory‖. Several indirect measures of ovulation have been proposed and we compared two of these: (1) the increase in progesterone production by the corpus luteum from the follicular to the luteal phase [7], and (2) the resulting increase in basal temperature. As previously reported [23], we use the term ―evidence of luteal activity‖ (ELA) to clarify that both measures are indirect ovulation indicators. Cycles showing ELA are referred to as ELA+ and cycles that do not show ELA are referred to as ELA-.  2.2.3.1 Urinalysis Samples were stored at -20°Celsius until shipment to the University of Washington where aliquots were taken from thawed specimens, preserved with 17 mg/mL boric acid and refrigerated (4°Celsius) until assayed in duplicate for PdG using competitive enzyme immunoassays [24-25]. The inter- and intra-assay coefficients of variation were 10.3% and 9.2% [24]. PdG concentration was estimated from optical density (Dynex MR7000 MicroPlate Reader, test wavelength 405 nm, reference 570 nm) using a four-parameter logistic model in Biolinx 2.0 Software (Dynex Laboratories, Inc., Chantilly, VA) and was corrected for hydration status by specific gravity [26]. Urinary PdG is highly correlated (r=0.98) with previous day serum progesterone [24]. The Kassam algorithm for determining ELA compares a daily 5-day moving average of PdG to a minimum 5-day baseline PdG level [27]. ELA+ cycles are those with PdG for >3 consecutive days at >3-times baseline [28]. Cycles are not analysed when ≥3 consecutive samples in the second half of the cycle are missing. This method of determining ELA has been shown to have 100% sensitivity and specificity versus visual classification by reproductive experts using daily urinary reproductive hormones in 52 menstrual cycles [28]. The LH surge occurs approximately 3 days before the sustained urinary PdG rise [27-28]. Therefore, cycles were classified as having a short luteal phase length (LPL) if the luteal phase was <10 days in duration and normal LPL if ≥10 days.  2.2.3.2 Basal temperature analysis Temperature records were used to determine ELA using computerised LS-QBT, which divides the cycle into two phases using least-squares criterion in a two-step function to maximize the mean difference [21]. A cycle is classified as ELA+ if the mean temperature difference between the phases is statistically significant (P<0.05) [21]. Cycles are not analysed if febrile illness occurs for ≥5 days or at any point mid-cycle, if ≥33% of temperatures are missing, or if ≥3 days are missing at mid-cycle [21]. The LH surge occurs approximately 2.4     53 days before the day of significant temperature rise [21]. Therefore, cycles were classified as having a short LPL if the luteal phase was <10 days in duration and normal LPL if ≥10 days. To determine the impact of wake-time variation and whether a reproductive expert is required to evaluate LS-QBT, the analysis was repeated for each cycle using:  A. All temperatures: all recorded temperatures were included except for febrile illness (temperatures ≥37°Celsius with note of illness). B. Royston-adjusted [22]: all recorded temperatures were adjusted by subtracting 0.1°Celsius/hour from the earliest wake-time (e.g., if earliest wake-time was 6 a.m., temperatures recorded at 8 a.m. would have 0.2°Celsius subtracted). C. 2-hour average wake-time: temperatures recorded >1-hour before or after the determined average wake-time were removed (e.g., for a 7 a.m. average wake-time, records before 6 a.m. or after 8 a.m. were excluded). D. Expert reviewed: temperatures were removed based on interpretation by a reproductive endocrinologist.   2.2.3.3 Statistical analyses Data were coded, verified and entered into SPSS software (SPSS version 15 Inc., 2006, Chicago, IL) and crosschecked for accuracy. Variables were examined for normality. Descriptive statistics were used to characterise the sample. Sensitivity and specificity of LS-QBT for ELA were determined relative to Kassam‘s PdG algorithm. Sensitivity is the number of cycles classified as ELA+ by LS-QBT divided by the number of ELA+ cycles by PdG (true ELA+). Specificity is the number of cycles classified as ELA- by LS-QBT divided by the number of ELA- cycles by PdG (true ELA-). Positive predictive value (PPV) is true ELA+ cycles divided by true and false ELA+ cycles. Negative predictive value (NPV) is true ELA- cycles divided by true and false ELA- cycles. Accuracy is the number of true ELA+ and true ELA- cycles divided by the total number of cycles. Pearson‘s correlations of the first day of sustained PdG rise and the day of a significant temperature increase by the four LS-QBT methods were also calculated. Although the Kassam PdG algorithm was not designed to assess LPL, it does estimate the day of luteal transition which can then be used to estimate LPL. We conducted sensitivity and specificity analyses of LS-QBT in detecting cycles with short versus normal LPL relative to Kassam‘s PdG algorithm (Appendix 12). The significance level for all analyses was p0.05.      54 2.3 Results 2.3.1 Participant characteristics Of the 53 women recruited to the study, 48 returned completed materials and 40 had sufficient data. There were no differences in descriptive characteristics between the 40 participants included in the analysis and the 13 excluded. Table 2.1 presents descriptive characteristics of the 40 participants. The majority were currently students (90%) and single (85%) and 100% had completed some post-secondary education. Most were either Caucasian (n=17) or Chinese (n=13); others were South/West Asian (n=5), Japanese (n=3) or Latin American (n=2). Method C resulted in removal of 0-14 recorded temperatures. For method D, 0-28 recorded temperatures were removed. This resulted in too many missing values for n=4 and n=6, respectively, and thus exclusion from analysis. As shown in Table 2.1, the mean difference between the earliest and latest wake-time was 4.6±1.6 hours and the within-person standard deviation of wake-time variation was 1.1±0.3 hours. Table 2.1 Descriptive characteristics of the sample (n=40)  Age (years)  24.3 ± 3.7 Age of menarche (years)  12.3 ± 1.3 Gynaecologic age (years)  12.0 ± 3.5 Body mass index (kg/m2) 21.7 ± 1.9 Adult amenorrhea   7.5% (3) Previous therapy with progesterone/progestin  2.5% (1) Previous use of oral contraceptives  47.5% (19) Study cycle length (days) 29.2 ± 3.1  Difference between earliest and latest wake-time (hours) 4.6 ± 1.6 Within-person variation in wake-time (hours) 1.1 ± 0.3 Data are presented as mean ± standard deviation for continuous variables or  proportion (% (n) for categorical variables.  2.3.2 Sensitivity, specificity, predictive values and accuracy Table 2.2 shows the ability of LS-QBT to classify cycles as ELA+ relative to our reference standard, Kassam‘s PdG algorithm. The reference method classified 36 of 40 cycles as ELA+ of which LS-QBT detected 35 (methods A and B), 33 of 34 (method C) and 30 of 31 (method D). Of the four cycles classified as ELA- by the reference, methods A and B detected one and methods C and D detected none. Table 2.3 shows the predictive values and accuracy of the LS-QBT methods. PPV ranged from 91% to 92% and accuracy ranged from 88% to 90%. NPV was 50% for methods A and B and 0% for methods C and D.     55 Table 2.2  Sensitivity and specificity of least-squares quantitative basal temperature analysis (LS-QBT) methods in determining evidence of luteal activity (ELA) relative to Kassam‘s urinary pregnanediol glucuronide (PdG) algorithm     LS-QBT method  ELA+ by PdG ELA- by PdG ELA+ by  LS-QBT  (sensitivity)  ELA- by  LS-QBT   (misclassified) ELA+ by  LS-QBT  (misclassified)  ELA- by  LS-QBT  (specificity) All temperaturesa  97% (35) 3% (1) 75% (3) 25% (1) Royston adjustedb 97% (35) 3% (1) 75% (3) 25% (1) 2-hour average wake-timec  97% (33) 3% (1) 100% (3) 0% (0) Expert reviewedd  97% (30) 3% (1) 100% (3) 0% (0) Data are presented as proportion (% (n).  a. All recorded temperatures were included except for febrile illness (n=40). b. All recorded temperatures were adjusted by 0.1°Celsius/hour from earliest wake-time (n=40).  c. Temperatures recorded >1 hour before or after the average wake-time were removed. Three cycles could no longer be analysed because of the number of temperature values removed (n=37) d. Temperatures were removed based on interpretation by a reproductive endocrinologist. Six cycles could no longer be analysed because of the number of temperature values removed (n=34).  Table 2.3  Predictive value and accuracy of least-squares quantitative basal temperature analysis (LS-QBT) methods in determining evidence of luteal activity relative to Kassam‘s urinary pregnanediol glucuronide (PdG) algorithm   LS-QBT method Positive predictive value Negative predictive value  Accuracy  All temperaturesa   92% (35/38) 50% (1/2) 90% (36/40) Royston adjustedb 92% (35/38) 50% (1/2) 90% (36/40) 2-hour average wake-timec  92% (33/36)  0% (0/1) 89% (33/37) Expert reviewedd  91% (30/33) 0% (0/1) 88% (30/34) Data are presented as proportion (% (n).  a. All recorded temperatures were included except for febrile illness (n=40). b. All recorded temperatures were adjusted by 0.1°Celsius/hour from earliest wake-time (n=40).  c. Temperatures recorded >1 hour before or after the average wake-time were removed. Three cycles could no longer be analysed because of the number of temperature values removed (n=37).  d. Temperatures were removed based on interpretation by a reproductive endocrinologist. Six cycles could no longer be analysed because of the number of temperature values removed (n=34).      56 2.3.3 Correlation of luteal onset: sustained PdG rise versus LS-QBT temperature  Increase  Figures 2.1-2.4 show the relationships between the day of a significant temperature increase by LS-QBT relative to the first day of a sustained PdG rise. All correlations were P<0.001. Inspection of Bland-Altman plots revealed good agreement with no problems in terms of proportional error or variation that depends on the magnitude of the measurement (data not shown).  Figure 2.1  Correlation of the day of LS-QBT temperature rise versus day of sustained PdG rise by Kassam algorithm: All temperaturesa                          a All recorded temperatures were included except for febrile illness.   N  =  3 5R  =  0 . 8 0 3p < 0 . 0 0 1810121416182022242628308 10 12 14 16 18 20 22 24 26 28 30P D GLS QBT    57 Figure 2.2  Correlation of the day of LS-QBT temperature rise versus day of sustained PdG rise by Kassam algorithm: Royston wake-time adjusteda                       a All recorded temperatures adjusted by Royston‘s adjustment of 0.1°Celsius/hour from earliest wake time.    Figure 2.3  Correlation of the day of LS-QBT temperature rise versus day of sustained PdG rise by Kassam algorithm: 2-hour average wake time temperaturesa                       a Temperatures recorded >1 hour before or after the average wake time were removed. Two cycles could no longer be analysed because of the number of temperature values removed. N  =  3 3R  =  0 . 6 5 1 p < 0 . 0 0 1810121416182022242628308 10 12 14 16 18 20 22 24 26 28 30P D GLS QBTN  =  3 5R  =  0 . 7 4 1P< 0 . 0 0 1810121416182022242628308 10 12 14 16 18 20 22 24 26 28 30P D GLS QBT    58 Figure 2.4  Correlation of the day of LS-QBT temperature rise versus day of sustained PdG rise by Kassam algorithm: Expert reviewed temperaturesa                       a Temperatures were removed based on interpretation by a reproductive endocrinologist. Four cycles could no longer be analysed because of the number of temperature values removed.  2.4 Discussion Women have recorded basal temperature for decades as part of fertility awareness-based methods for contraception [6]. Basal temperature is also an inexpensive and accessible method of estimating ovulation in scientific research. Qualitative analysis of temperature records is unreliable and inaccurate [8-18]; however, little information exists on the validity of computerised quantitative methods, such as LS-QBT. Relative to PdG, an established indirect marker of ovulation [28], LS-QBT classification of cycles as ELA+ was excellent, but classification as ELA- was poor. This may, at least in part, be explained by the small number of cycles (10%) classified as ELA- by our reference method. Determination of the day of significant temperature increase by LS-QBT correlated well with the day of sustained PdG rise, as was previously observed relative to the LH surge [21]. Our results suggest that LS-QBT can be used to determine ELA and still retain the positive aspects of being inexpensive and noninvasive. Despite efforts to recruit women with consistent sleep patterns, considerable wake-time variability was observed in our sample, which has been reported to affect basal temperature [22]. We assessed whether selected expert or systematic removal of temperatures improved LS-QBT performance. None of these methods improved performance relative to PdG, indicating that LS-QBT is robust to wake-time variation. Accordingly, LS-QBT is an inexpensive and accessible method for all researchers in women‘s health as reproductive experts are not N  =  3 0R  =  0 . 7 4 7p < 0 . 0 0 1810121416182022242628308 10 12 14 16 18 20 22 24 26 28 30P D GLS QBT    59 required. However, our conclusions apply to women with reasonably regular diurnal schedules and may not extend to extreme changes in sleep patterns such as with rotating shift work. We are aware of only two studies that assessed QBT relative to clinical measures of ovulation [8, 21]. Ecochard and coworkers compared the averaging method to ultrasound and urinary LH determination of ovulation [8]. A temperature rise occurred in 98% of cycles classified as ovulatory by ultrasound, although this occurred within one day of the LH surge in only 20% of cycles, leading the authors to suggest that the averaging method is unreliable [8]. However, the progesterone rise that causes temperatures to increase follows the LH peak by 24-48 hours, and temperature increases require some duration of progesterone exposure. Therefore, one would not expect temperature to increase within one day of the urinary LH peak. Data from Prior and coworkers detected temperature rises 2.7 days after the serum LH surge using the averaging method and 2.4 days after the LH surge using LS-QBT [21]. Ecochard and coworkers‘ distribution of lag times between the LH surge and the temperature rise show that most cycles were within the expected two to three days following the LH surge [8]. Similarly, Prior and colleagues found that using either the averaging or LS-QBT method, approximately 75% of cycles had a temperature rise within three days of the LH surge [21]. Taken together, it appears that LS-QBT and the averaging method detect a temperature rise approximately three days following the LH surge. Lack of exact concordance between the days of LH surge and temperature rise does not invalidate these two QBT methods as indices of luteal activity.  A limitation of our study is the absence of a direct measure of ovulation such as sequential transvaginal ultrasound measurements. We chose Kassam‘s PdG algorithm as our indirect reference method because it reflects follicular to luteal change in progesterone, and the increase in basal temperature during the luteal phase results from the thermogenic effects of progesterone [7]. Kassam‘s algorithm was reported to have 100% sensitivity and specificity relative to visual classification of daily urinary reproductive hormones by experts; however, it should be noted that 10% of the cycles were classified as indeterminate and excluded [28]. It is conceivable that some of the misclassified cycles in our study may have been ―indeterminate‖.  Despite lack of direct ovulation observation, our data comparing LS-QBT to PdG, combined with previous work comparing temperature rise with the serum LH surge [21], suggest that LS-QBT is a useful tool that performs reasonably well in detecting ELA+ cycles and estimating the DLT. Longitudinal epidemiology studies examining ovulatory function relative to other aspects of health require accurate, inexpensive methods that are acceptable to study participants. Because temperature recording is noninvasive and as our findings suggest, robust to greater wake-time variability than previously thought, it does not require expert interpretation or other time consuming or expensive adjustments. Subject burden is low and although cervical mucous assessment is more accurate than basal temperature methods in determining DLT     60 [8,10,14,29], it is less acceptable to participants [30] and may be less suitable for long-term use in epidemiological studies. Further validation of LS-QBT, using daily transvaginal ultrasound as the gold standard and in a population with more anovulatory and irregular cycles, is required.      61 2.5 References [1] Prior JC, Vigna YM, Schechter MT, Burgess AE. Spinal bone loss and ovulatory disturbances. New Engl J Med 1990;323:1221-7. [2] Prior JC, Vigna YM, Barr SI, Rexworthy C, Lentle BC. Cyclic medroxyprogesterone treatment increases bone density: a controlled trial in active women with menstrual cycle disturbances. Am J Med 1994;96:521-30. [3] Sowers M, Randolph JF, Crutchfield M, Jannausch ML, Shapiro B, Zhang B, La Pietra M. Urinary ovarian and gonadotropin hormone levels in premenopausal women with low bone mass. J Bone Miner Res 1998;13:1191-202. [4] Waugh EJ, Polivy J, Ridout R, Hawker GA. A prospective investigation of the relations among cognitive dietary restraint, subclinical ovulatory disturbances, physical activity, and bone mass in healthy young women. Am J Clin Nutr 2007;86:1791-801. [5] Khalid A. Irregular or absent periods—what can an ultrasound scan tell you? Best Pract Res Clin Obstet Gynaecol 2004;18:3-11.  [6] Frank-Herrmann P, Heil J, Gnoth C, Toledo E, Baur S, Pyper C, Jenetzky E, Strowitzki T, Freundl G. The effectiveness of a fertility awareness based method to avoid pregnancy in relation to a couple's sexual behaviour during the fertile time: a prospective longitudinal study. Hum Reprod 2007;22:1310-9. [7] Moghissi KS. Prediction and detection of ovulation. Fertil Steril 1980;34:89-98. [8] Ecochard R, Boehringer H, Rabilloud M, Marret H. Chronological aspects of ultrasonic, hormonal, and other indirect indices of ovulation. Br J Obstet Gynaecol 2001;108:822-9.  [9] Guermandi E, Vegetti W, Bianchi MM, Uglietti A, Ragni G, Crosignani P. Reliability of ovulation tests in infertile women. Obstet Gynaecol 2001;97:92-6. [10] Guida M, Tommaselli GA, Palomba S, Pellicano M, Moccia G, Di Carlo C, Nappi C. Efficacy of methods for determining ovulation in a natural family planning program. Fertil Steril 1999;72:900-4. [11] Luciano AA, Peluso J, Koch EI, Maier D, Kuslis S, Davison E. Temporal relationship and reliability of the clinical, hormonal, and ultrasonographic indices of ovulation in infertile women. Obstet Gynaecol 1990;75:412-6. [12] Morris N, Underwood L, Easterling W. Temporal relationship between basal body temperature nadir and luteinizing hormone surge in normal women. Fertil Steril 1976;27:780-3. [13] Quagliarello J, Arny M. Inaccuracy of basal body temperature charts in predicting urinary luteinizing hormone surges. Fertil Steril 1986;45:334-7.     62 [14] Templeton AA, Penney GC, Lees MM. Relation between the luteinizing hormone peak, the nadir of the basal body temperature and the cervical mucus score. Br J Obstet Gynaecol 1982;89:985-8. [15] Yong EL, Wong PC, Kumar A, Wong YC, Goh HH, Hagglund L, Ratnam S. Simple office methods to predict ovulation: the clinical usefulness of a new urine luteinizing hormone kit compared to basal body temperature, cervical mucus and ultrasound. Aust N Z J Obstet Gynaecol 1989;29:155-60. [16] Hilgers TW, Bailey AJ. Natural family planning. II. Basal body temperature and estimated time of ovulation. Obstet Gynaecol 1980;55:333-9. [17] Bauman JE. Basal body temperature: unreliable method of ovulation detection. Fertil Steril 1981;36:729-33. [18] Ayres-de-Campos D, Silva-Carvalho JL, Oliveira C, Martins-da-Silva I, Silva-Carvalho J, Pereira-Leite L. Inter-observer agreement in analysis of basal body temperature graphs from infertile women. Hum Reprod 1995;10:2010-6. [19] Vollman RF. The menstrual cycle. In: Friedman EA, editor. Major problems in obstetrics & gynecology. Toronto: WB Saunders Company; 1977. p. 11-193. [20] Royston JP, Abrams RM. An objective method for detecting the shift in basal body temperature in women. Biometrics 1990;36:217-24. [21] Prior JC, Vigna YM, Schulzer M, Hall JE, Bonen A. Determination of luteal phase length by quantitative basal temperature methods: validation against the midcycle LH peak. Clin Invest Med 1990;13:123-31. [22] Royston JP, Abrams RM, Higgins MP, Flynn AM. The adjustments of basal body temperature measurements to allow for time of waking. Br J Obstet Gynaecol 1980;87:1123-7. [23] Santoro N, Crawford SL, Allsworth JE, Gold EB, Greendale GA, Korenman S, Lasley L, McConnell D, McGaffigan P, Midgely R, Schocken M, Sowers M, Weiss G. Assessing menstrual cycles with urinary hormone assays. Am J Physiol Endocrinol Metab 2003;284:E521-30. [24] O'Connor KA, Brindle E, Holman DJ, Klein NA, Soules MR, Campbell KL, Kohen F, unro CJ, Shofer JB, Lasley BL, Wood JW. Urinary estrone conjugate and pregnanediol 3-glucuronide enzyme immunoassays for population research. Clin Chem 2003;49:1139-48. [25] O'Connor KA, Brindle E, Shofer JB, Miller RC, Klein NA, Soules MR, Campbell KL, Mar C, Handcock MS. Statistical correction for non-parallelism in a urinary enzyme immunoassay. J Immunoassay Immunochem 2004;25:259-78.     63 [26] Miller RC, Brindle E, Holman DJ, Shofer J, Klein NA, Soules MR, O'Connor KA. Comparison of specific gravity and creatinine for normalizing urinary reproductive hormone concentrations. Clin Chem 2004;50: 924-32. [27] Kassam A, Overstreet JW, Snow-Harter C, De Souza MJ, Gold EB, Lasley BL. Identification of anovulation and transient luteal function using a urinary pregnanediol-3-glucuronide ratio algorithm. Environ Health Perspect 1996;104:408-13. [28] O'Connor KA, Brindle E, Miller RC, Shofer JB, Ferrell RJ, Klein NA, Soules MR, Holman DJ, Mansfield PK, Wood JW. Ovulation detection methods for urinary hormones: precision, daily and intermittent sampling and a combined hierarchical method. Hum Reprod 2006;21:1442-52. [29] Alliende ME, Cabezón C, Figueroa H, Kottmann C. Cervicovaginal fluid changes to detect ovulation accurately. Am J Obstet Gynecol 2005;193:71-5. [30] Wright DM, Kesner JS, Schrader SM, Chin NW, Wells VE, Krieg EF. Methods of monitoring menstrual function in field studies: attitudes of working women. Reprod Toxicol 1992;6:401-9.      64 Chapter 3:   A prospective exploration of cognitive dietary restraint, subclinical ovulatory disturbances, cortisol and change in bone density over two years in healthy young women1                                                  1 A version of this chapter has been accepted for publication: Bedford JL, Prior JC, Barr SI. A prospective exploration of cognitive dietary restraint, subclinical ovulatory disturbances, cortisol and change in bone density over two years in healthy young women. Journal of Clinical Endocrinology & Metabolism.  Date of Acceptance: March 2010. Copyright © The Endocrinology Society.      65 3.1 Introduction Cognitive dietary restraint (CDR) is the perception that one is limiting food intake in an effort to achieve or maintain a perceived ideal body weight [1]. Different from dieting, a behaviour where energy intake is limited in an intent to lose weight, CDR is a psychosocial construct reflecting habitual monitoring of food intake and body weight preoccupation. The perceptual nature of CDR is reflected by the lack of clear evidence that energy intake, relative body mass or weight change differs by restraint level among young women [e.g. 2-4].  Evidence suggests that the experience of higher CDR may detrimentally affect physiological health including menstrual cycle and ovulatory function and bone. Young women with higher CDR are more likely to report menstrual cycle irregularities [2,5] and to unknowingly experience subclinical ovulatory disturbances [6-8]. Subclinical ovulatory disturbances, such as short luteal phases and anovulation, indicate reproductive hormone inadequacies and may influence bone mass [9]. It is well established that overt ovarian disturbances such as amenorrhea detrimentally affect bone [10]. Whether subclinical ovulatory disturbances are associated with lower bone mineral density (BMD) or increased bone loss is controversial [11-17]. As well, a direct cross-sectional relationship between higher CDR and reduced BMD and bone mineral content (BMC) has been reported by some [5,18-21] but not all [7,22] studies. In the only prospective study to date, CDR was not associated with subclinical ovulatory disturbances or BMD change; although BMD change was lower in women with more subclinical ovulatory disturbances [17]. The relationship between CDR, ovulatory function and bone may be mediated by the physiological stress response. The constant monitoring and attempts to control food intake may act as a stressor capable of activating the hypothalamic-pituitary-adrenal (HPA) axis. As shown in Figure 3.1, stress activation of the HPA axis triggers a cascade of events resulting in increased cortisol, and concurrent inhibition of the hypothalamic-pituitary-gonadal (HPG) axis, leading to disturbed menstrual cycles and ovulatory function [23]. Reports of higher cortisol levels among women with higher CDR [24-28] suggest that restraint may be a subtle but chronic stressor, resulting in modest but persistent elevations in cortisol within the physiological range. Cortisol has well established direct effects on bone, and clinical hypercortisolism is consistently associated with reduced BMD [23]. Whether cortisol elevations within the normal range influence bone in young healthy women is unclear [5,7,18-22,29]. Therefore we hypothesised that women with higher CDR would have increased 24-hour urinary free cortisol, more frequent subclinical ovulatory disturbances and less positive change in bone density over two years (Figure 3.1).      66 Figure 3.1  Model driving our hypothesis of cognitive dietary restraint and bone density juxtaposition with the physiological stress response                              Variables shown as ovals were measured in the present study. The dashed line (1) between dietary restraint and chronic psychosocial stress reflects our hypothesis that restraint acts as a subtle chronic stressor capable of activating the hypothalamic-pituitary-adrenal (HPA) axis. Solid black lines indicate well established mechanisms of the stress response including: (2) increased secretion of cortisol, which has a direct negative effect on bone density, and (3) inhibition of the hypothalamic-pituitary-gonadal (HPG) axis resulting in deficiencies or imbalances of the reproductive hormones and therefore menstrual cycle and ovulatory disturbances. Grey lines represent hypothesised but inconclusive relationships that we and others have observed including: (4) the possibility that subclinical ovulatory disturbances can have detrimental effects on bone density; (5) an association between higher dietary restraint and elevated cortisol; and (6) an association between higher restraint and the occurrence of menstrual cycle and ovulatory disturbances. This leads to our hypothesis, indicated by the grey dashed line (7), that dietary restraint may result in less positive changes in bone density. ACTH, adrenocorticotropic hormone; CRH, corticotropin-releasing hormone; E, estradiol; FSH, follicle stimulating hormone; GnRH, gonadatropin-releasing hormone; P, progesterone.  3.2 Methods 3.2.1 Participants Participants were recruited from University of British Columbia classes and the Vancouver (British Columbia, Canada) community for a 2-year study on potential correlates of bone density (Appendix 13). No reference was made to eating/body attitudes in recruitment materials. Interested women contacted the investigators for additional study details (Appendix 14). Potential participants were interviewed by phone to determine eligibility (Appendix 15) ↑ c ort i s ol  re l e a s e1234567H P A - a x i s  a c t i v a t i on↓ bone dens i t yC ogni t i v e  D i e t a r y  R e s t ra i nt  H P G - a x i s  i nhi bi t i on↑ ovul a t or y  di s t urbanc e s  C hroni c  P s y c hosoc i a l  S t re s sCRHA C T HGnR HP ul s eLH ,  F S H ,E, Pi ol   t i oni tt i v e  r  i nt   t i ont or y  e s   i a l      67 including: 19-35 years of age, no pregnancy or breastfeeding currently or within 12 months, regular menstrual cycles (self-reported menses every 21-35 days in the previous ≥6 months), non-obese (self-reported body mass index (BMI) 18-30 kg/m2), consistent sleep patterns (wake up and go to bed at approximately the same time most days) and absence of any medical conditions (previous or current diagnosis of hirsutism, eating disorder, polycystic ovarian syndrome, Cushing‘s syndrome, inflammatory conditions, hyperthyroidism) or use of medications (oral contraceptives, progesterone, glucocorticoids currently or within the past six months) that could affect study variables. Of 148 women assessed, 142 were eligible (Figure 3.2). A final convenience sample of 140 provided written informed consent (Appendix 16) and was oriented to the study. Data collection was completed by 137 at baseline, 127 at first follow-up and 123 at final follow-up. Data are reported for these 123 individuals. The study protocol was approved by the university‘s Clinical Research Ethics Board (Appendix 17). Participants were provided with travel compensation (Appendix 18) and $90 in gift cards for their participation (Appendix 19). Figure 3.2  Flow diagram depicting study recruitment, participation and data collection at baseline and first and final follow-up assessments   N = 123 participated in final follow-up (~2 years after baseline): questionnaires, 24-h urine collection, anthropometric measurements, DXA scan.   Excluded   n=6   Ineligible because:      Oral contraceptive user  n=3     Shift work   n=1     BMI <17 kg/m2  n=1    Glucocorticoid user n=1 N=148 assessed for eligibility from August to December 2006  N=142 eligible  N=137 consented to participate and completed the baseline procedures: questionnaires, 24-h urine collection, anthropometric measurements, dual energy X-ray absorptiometry (DXA) scan N =127 participated in first follow-up (~1 year after baseline): questionnaires, 24-h urine collection, anthropometric measurements.    n=2 did not attend orientation    n=3 came for materials and             instructions but did not            complete the procedures.    Losses to follow up (n=14)   Reasons:     Moved   n=4     Did not respond   n=2     No longer interested n=3     Pregnant  n=3     Androgen excess  n=1     Thyroid cancer  n=1  Basal body temperature recorded every day during 2-year study for menstrual cycle and ovulatory monitoring      68 3.2.2 Data collection This 2-year prospective cohort study included data collection at baseline and two follow-ups at 6-12 months (mean 7) and 1.5-2.5 years (mean 2) after baseline (Figure 3.2). At each of the three data collections, participants met with an investigator to complete anthropometric measurements and to be oriented to study procedures (materials and detailed written and oral instructions). At each data collection, participants completed the following procedures at-home: a questionnaire package, a food frequency questionnaire, and 24-hour urine collections (Appendix 20). Every day during the 2-year study, participants were also asked to record their basal temperature in a provided temperature calendar (Appendix 21). Dual energy X-ray absorptiometry (DXA) scans were conducted at Vancouver General Hospital (VGH) at baseline and 2-year follow-up (Appendix 22).   3.2.3 Questionnaires  The questionnaire package (completed at baseline and both follow-ups, Appendix 23) included validated self-report questionnaires (Appendix 24) as well as questions to elicit demographic information (age, ethnicity, education, employment) and health information (cigarette use, medical/menstrual history and changes).  The well-validated Three Factor Eating Questionnaire (TFEQ) was used to examine stress related to eating and the body [1]. The questionnaire includes three subscales that assess dimensions of eating attitudes that may influence eating behaviour including: Restraint with higher scores indicating higher perceived dietary restraint; Disinhibition with higher scores indicating a greater tendency to overeat when restraint is removed; and Hunger for which higher scores indicate increased susceptibility to hunger and food cravings [1].  To examine the role of general psychosocial stress the Perceived Stress Scale was completed at each data collection to determine stress perception over the previous month [30] and the Daily Stress Inventory was completed after each 24-hour urine collection (Appendix 25) to determine the frequency and impact of stressful events [31].  The Baecke Questionnaire of Habitual Physical Activity [32] was completed at each assessment to measure participants‘ usual activity levels at work, in sport, and during leisure.  At the 2-year follow-up only, any reproductive hormone use was documented and the Eating Disorder Examination (EDE) Questionnaire [33] was completed to confirm absence of clinical eating disorders. The EDE includes four subscales (Restraint, Eating Concern, Weight Concern and Shape Concern) and an average Global score to assess body attitudes that are concurrent with eating disorder pathology over the previous four weeks. Global scores ≥4 (possible range 0-6) are considered ―clinically significant‖ but not diagnostic [34-35].     69 3.2.4 Food frequency questionnaire (FFQ) The Diet History Questionnaire (version 1.0, National Institutes of Health, Applied Research Program, National Cancer Institute, 2002) was completed at baseline and both follow-ups and analysed using a Canadian version of the programme [36]. Energy intakes of <600 or >3500 kcal were deemed biologically implausible [36]. This resulted in removal of four FFQs at baseline and two FFQs at each of the first and second follow-ups. Complete FFQ data were available for 119 at baseline and first follow-up and 120 at the second follow-up. All 123 participants had data from at least one FFQ available for use.   3.2.5 Ovulatory function Ovulatory function can be observed indirectly by determining whether or not basal temperature increases from the follicular to luteal phase of the cycle as a result of increased progesterone production. Every day during the 2 year study, participants were asked to record their temperature immediately upon waking. Temperatures were recorded in provided temperature calendars using a digital thermometer (Becton Dickinson, Franklin Lakes, NJ, product number 524052). Time of waking, menstrual flow status and any illness were recorded and sleep quality was rated from 0-4. Completed temperature calendars were returned to investigators at each follow-up. Temperatures collected during hormonal contraceptive use were not analyzed. Temperatures were entered into a computer programme (Maximina ©) which uses least-squares quantitative basal temperature analysis (LS-QBT) to determine evidence of luteal activity by identifying a statistically significant difference by least squares criterion in temperature values to divide the cycle into two phases [37]. Cycles are not analysed if exogenous hormones were used, if a febrile illness occurs for ≥5 days or at any point mid-cycle, if ≥33% of temperature readings for a cycle are missing, or if ≥3 days are missing at mid-cycle. This method has been validated against established markers of ovulation: the serum peak luteinizing hormone (LH) concentration [37] and the rise in urinary progesterone metabolites [38].  Cycles are classified as having evidence of luteal activity or being ―ovulatory‖, if the maximum mean temperature difference between the phases is statistically significant [37]. If no temperature increase occurs, the cycle is classified as ―anovulatory‖. Luteal phase length (LPL) was calculated as the number of days from the day of significant temperature rise until the day before menstrual flow began [37]. As the LH surge occurs approximately 2.4 days prior to the temperature rise [37], LPL is classified as ―short‖ if <10 days or ―normal‖ if ≥10 days. The percentage of cycles with subclinical ovulatory disturbances was calculated by adding the number of anovulatory and/or short LPL cycles and dividing by the total number of cycles     70 analysed. To examine differences in outcome variables, participants were classified with a higher or lower percentage of disturbed ovulation by median split.   3.2.6 Urine collection and analyses Within several weeks of meeting with investigators at each data collection, participants‘ chose a ―normal day‘ free of any unusual physical or mental stresses to complete the 24-h urine collection. Participants discarded their first urine void, recorded the time this occurred and then collected all subsequent voids for 24 hours including a void at the recorded time the following morning. After their last void, participants completed the Daily Stress Inventory described above [34]. At the VGH Laboratory, urine volume was measured and aliquots were frozen and stored prior to analysis of urinary free cortisol (UFC, µg/24-hour) by high-throughput liquid chromatography and tandem mass spectrometry [39]. Six participants completed two urine collections and 117 completed three.   3.2.7 Physical measurements  Physical measurements were made in duplicate at each data collection point. Weight was measured while wearing light indoor clothing without shoes, to the nearest 0.1 kg using an electronic scale. Using a stadiometer (model 214; Seca, Hamburg, Germany), height without shoes was measured to the nearest 0.1 cm at full inspiration. Body mass index (BMI; kg/m2) was calculated from these data. Measurements were made in duplicate. If differences occurred, a third measurement was made and the two closest measurements were averaged.  At baseline and final follow-up (1.95±0.14 years after baseline), DXA scans of the lumbar spine (L1-4), both total hips and whole body were completed. Total body bone-free lean mass (kg), fat mass (kg), percent body fat and areal bone mineral density (aBMD, g/cm2) were measured on a Lunar Prodigy machine with enCORE software (General Electric Healthcare, Madison, WI). Daily quality assurance tests were conducted using a spine phantom scan and densitometric calibration. Repeat aBMD measurements fall within ±0.01 g/cm2 for L1-4 and ±0.012 g/cm2 for the proximal femur according to the manufacturer. The in-house coefficient of variation for aBMD at L1-4 averaged 0.94% (0.82–1.10%) and the coefficient of variation for total proximal femur averaged 0.70% (0.65-0.76%).   3.2.8 Statistics  Data were coded, verified and entered into SPSS software (version 17, SPSS Inc., 2008, Chicago, IL) and crosschecked for accuracy. Physiologic variables were examined for outliers (mean ± >4 standard deviation (SD)) and none were present.      71 Repeated measures General Linear Model (GLM) with least significant difference post-hoc analysis was used to examine changes over time. Because reported nutrient intakes, questionnaire scores (including CDR), UFC and urine volume did not change, averages were calculated and used in analyses.  A General Stress Z-score was calculated from the average Perceived Stress Scale and Daily Stress Inventory Impact and Frequency scores. Questionnaire Z-scores ([participant score – questionnaire mean]/questionnaire SD) were then summed and divided by three.  The second DXA scan was completed 1.95±0.14 years after the baseline scan. For all physical measurements, the percentage of change over two years was calculated and annualised ([observed percent change * 2-year]/[duration between Time 1 and Time 2]) and is hereafter referred to as change (Δ).  Descriptive statistics were used to characterise the sample. Pearson‘s correlations were conducted to identify potentially confounding covariates of study outcome variables (Restraint score, UFC, ovulatory function, ΔaBMD). Comparisons between groups (lost to follow-up, ethnicity, study hormone use, number of cycles analysed) were examined by Chi-square for categorical data and by independent t-tests or GLM with appropriate covariates for continuous variables. Women were classified by median split for subclinical ovulatory disturbances (≥38.8% versus <38.8% of cycles) and for CDR (Restraint score ≥7.7 versus <7.7). Comparisons were made between groups for study outcome variables using independent t-tests and GLM adjusted for appropriate covariates. Because steroid metabolism may differ between Asians and Caucasians [40], interactions between ethnicity and CDR were examined with regard to UFC, ovulatory disturbances and ΔaBMD. The significance level for all analyses was P0.05 and all cases were excluded pairwise.  3.3 Results  3.3.1 Sample Consistent with the UBC student body, the sample included Asians (63%) and Caucasians, who did not differ in study outcome variables (data not shown). Mean age at baseline was 22.1±3.3 years and gynaecological age was 9.7±3.7 years. Almost all had completed some post-secondary education (96%), were single (92%), non-smokers (98%) and nulliparous (98%). During the study, 18 women took oral contraceptives for 1-22 months (mean 8.2±6.2), two used hormonal intrauterine systems for 8-18 months, and two used progesterone cream for 0.3-3 months. Participants who used hormones (n=22) did not differ in outcome variables versus non-users (data not shown). Thus, all participants were included and the duration of study hormone use was examined as a potential confounder. The EDE subscale and     72 Global scores (data not shown) were lower than published norms [34,41] and only one participant had a clinically significant Global score of 4.5.  3.3.2 Questionnaires  Average questionnaire scores and energy intake, and partial correlation coefficients of the Three Factor Eating Questionnaire subscales are shown in Table 3.1. CDR was not associated with energy intake, General Stress or physical activity. Disinhibition was negatively associated with Baecke total and occupational activity score and was positively with Hunger and General Stress. Hunger was positively associated with General Stress.  Table 3.1  Mean questionnaire scores and energy intakes, and partial correlation coefficients of the Three Factor Eating Questionnaire subscales and 24-hour urinary free cortisol in healthy premenopausal women (n=123)   Mean ± SD Rp  All participants CDRab  Disinhibitionab Hungerab UFCc CDRa 7.9 ± 4.1 --- --- --- 0.06 Disinhibitiona 6.0 ± 2.9 0.24 --- --- 0.18 Hungera 5.3 ± 2.3 -0.09 0.53*** --- 0.10 General Stress Z-scored 0.0 ± 0.8 -0.01 0.24** 0.31*** 0.24** Physical Activitye 7.8 ± 1.3 0.08 -0.22* -0.11 0.01     Occupational  2.4 ± 0.5 0.06 -0.24** -0.13 0.15     Sport  2.5 ± 0.8 0.08 -0.15 -0.09 -0.03     Leisure  2.9 ± 0.6 0.04 -0.10 -0.05 -0.05 Energy intake (kcal) 1556 ± 478 0.15 0.08 0.12 0.17 Data are reported as mean ± standard deviation. Values are reported as averages from assessments at baseline and both follow-ups as values did not change over time by repeated measured General Linear Modelling. CDR, Cognitive dietary restraint; Rp, partial correlation coefficients; UFC, 24-hour urinary free cortisol. Correlation significant at * P<0.05, ** P≤0.01; *** P≤0.001. a. Three Factor Eating Questionnaire subscales scores: CDR (possible score 0-21);    Disinhibition (0-16); and Hunger (0-14). b. Adjusted for body mass index (kg/m2). c. Adjusted for urine volume (L/24-hour). d. Z-score of the Perceived Stress Scale and Daily Stress Inventory Impact and Frequency subscales assessed on the days of urine collection. e. Baecke Habitual Physical Activity Questionnaire, possible scores for subscales 1-5 and total 3-15.  3.3.3 Urine volume and UFC Urine volume (mean±SD 1.8 ± 0.8 L/24-hour) and UFC (mean±SD 25.7 ± 9.5 µg/24-hour) were correlated (r=0.34, P<0.001). This did not change after controlling for height and/or     73 weight (data not shown). Therefore UFC was adjusted for urine volume in all analyses including the partial correlations presented in Table 3.1. UFC was positively correlated with General Stress but was not associated with other questionnaire scores (including CDR), or with baseline or Δ anthropometrics (data not shown). Volume-adjusted UFC did not differ by ethnicity (P=0.858).  3.3.4 Menstrual cycle and ovulatory function 114 women provided 1-28 cycles (mean±SD=13.6±7.0) sufficient for analysis. There were no differences in demographics or outcome variables between participants who provided ≤10 cycles (n=42), ≤5 cycles (n=17) and ≤3 cycles (n=6) versus those that provided more cycles for each cut-off. The number of cycles analysed was not correlated with the percentage of cycles with subclinical ovulatory disturbances. Therefore, all 114 participants were included in further analyses and the number of cycles analysed was examined as a potential confounder. Study cycle length was 30.8±4.1 days with 14 women experiencing oligomenorrhea (cycle lengths of 36-90 days) and one experiencing amenorrhea (>180 days between cycles). Cycle length was inversely associated with age (r= -0.25, P=0.007), gynaecological age (r= -0.19, P=0.042), height (r= -0.20, P=0.035), weight (r= -0.23, P=0.015) and total, leisure and sport (r= -0.25-0.32, P<0.01) activity scores. Cycle length was not associated with BMI, Δanthropometrics, energy intake, the number of cycles analysed or the percentage of subclinical ovulatory disturbances (data not shown). There were no differences in study outcome variables between those with longer versus normal cycle lengths after adjusting for appropriate covariates (data not shown). The mean percentage of cycles with subclinical ovulatory disturbances was 43.7±32.0%. Sixty-one percent of women had ≥1 anovulatory cycles and 82% ≥1 cycles with short LPL. Age (r= -0.25, P=0.008), gynaecological age (r= -0.29, P=0.002) and BMI (r=0.20, P=0.031) were associated with subclinical ovulatory disturbances. Examination of a scatterplot suggested that two women with BMI >29 and high percentage of subclinical disturbances may have driven the correlation, as their removal resulted in the association becoming nonsignificant (r= 0.15, P=0.124). No other anthropometric (baseline or Δ) were associated with subclinical ovulatory disturbances.  After adjustment for baseline gynaecological age and BMI, the percentage of cycles with subclinical ovulatory disturbances was positively associated with CDR score (r=0.22, P=0.018), but not with physical activity, energy intake or UFC (urine volume as additional covariate; data not shown).       74 3.3.5 Physical measurements Table 3.2 describes participants‘ baseline, 2-year and annualised percent change in anthropometrics and aBMD. Height, weight and total body and L1-4 aBMD increased significantly during the study. ΔaBMD did not differ by ethnicity (P=0.311-0.398).  Baseline height, weight and lean mass were not associated with ΔaBMD, and. Hip ΔaBMD was inversely associated with Δfat mass (r= -0.18, P=0.047) and Δ%body fat (r= -0.20, P=0.026), and positively with baseline BMI (r=0.23, p=0.012), fat mass (r=0.23, p=0.01) and %body fat (r=0.19, p=0.038). Total body ΔaBMD was positively associated with Δweight (r=0.21, P=0.018), ΔBMI (r=0.18, P=0.049) and Δlean mass (r=0.18, P=0.048). L1-4 ΔaBMD was not associated with Δanthropometrics (data not shown). No dietary intake variables were associated with ΔaBMD except that calcium/kcal was negatively associated with total hip ΔaBMD (r= -0.19, P=0.036). Examination of a scatterplot revealed a participant with a calcium intake of 1.04 mg/kcal and total hip ΔaBMD of -5.94% and her removal made the association nonsignificant (r= -0.13, P=0.164).  Volume-adjusted UFC was not associated with ΔaBMD at the hip (r=0.099, P=0.279), L1-4 (r=0.008, P=0.933) or total body (r=0.04, P=0.658). Adjusted for Δlean mass, baseline gynaecological age and BMI, only hip ΔaBMD was significantly associated with the percentage of cycles with subclinical ovulatory disturbances (r =-0.29, p=0.002). Table 3.2  Physical measurements at baseline, 2-year follow-up and the 2-year percent change in healthy premenopausal women (n=123)   Baseline 2-year % 2-year changea P  valueb Heightc (cm) 163.0 ± 7.2 163.1 ± 7.2 0.001 ± 0.003 <0.001 Weight (kg) 57.9 ± 8.8 58.4 ± 9.0 1.2 ± 5.5 0.036 Body mass index (kg/m2) 21.8 ± 2.5 21.9 ± 2.6 0.7 ± 5.6 0.198 Bone free lean mass (kg) 37.8 ± 5.0 38.0 ± 5.1 0.7 ± 3.7 0.051 Bone free fat mass (kg) 16.8 ± 5.6 17.3 ± 5.7 4.1 ± 17.7 0.053 Bone free body fat (%) 30.3 ± 6.6 30.7 ± 6.5 2.2 ± 12.8 0.169 Total body aBMD   1.136 ± 0.077 1.147 ± 0.078 1.1 ± 1.7 <0.001 Lumbar spine aBMD  1.183 ± 0.121 1.196 ± 0.122 1.2 ± 2.8 <0.001 Hip aBMD  1.025 ± 0.120 1.027 ± 0.122 0.2 ± 2.2 0.380 Data are presented as mean ± standard deviation. aBMD, areal bone mineral density (g/cm2). a 2-year measurements were conducted 1.95 ±0.14 year after baseline. Measurements before or after the 2-year time point were corrected to 2 year percent change.  b Level of significance of differences between baseline and 2-year values by repeated measures General Linear Model. c Height increased significantly over the 2 year period. This is likely due to measurement error and that many participants were young (36% ≤20 years of age at baseline) and may have still been growing.     75 3.3.6 Differences by CDR median split Differences in study variables are presented in Table 3.3. The following variables did not differ by level of CDR: age, ethnicity, height, lean mass, waist circumference, Δanthropometrics, the number of cycles analysed, and the frequency and duration of study hormone use. Women with higher CDR had higher baseline weight, BMI, fat mass, %body fat, BMI-adjusted energy intakes and Disinhibition scores. Physical activity, Hunger and General Stress did not differ.  After adjusting for baseline BMI and gynaecological age (Table 3.3), subclinical ovulatory disturbances were more frequent in women with higher CDR. The ethnicity effect and the ethnicity-by-CDR interaction were not significant.  Women with higher CDR had significantly higher UFC (Table 3.3). There was no effect of ethnicity, but there was a significant ethnicity-by-CDR interaction: Caucasians but not Asians with higher CDR had higher UFC, and among Caucasians, CDR and UFC tended to correlate (r=0.29, p=0.056). For ΔaBMD, there were no main effects of CDR (F=0.032-1.167, P=0.282-0.859) or ethnicity (F=0.635-1.264, P=0.263-0.427) and no interaction (F=0.029-0.263, P=0.609-0.866).  Table 3.3  Differences between healthy premenopausal women with higher and lower cognitive dietary restraint (by median split) in baseline anthropometrics, Δanthropometrics questionnaire scores, energy intakes, menstrual cycle characteristics, 24-hour urinary free cortisol and 2-year ΔaBMD (n=123)   Higher CDRa (n=60) Lower CDRb (n=63) P valuec Age (years) 21.9 ± 3.3 22.4 ± 3.4 0.399 Ethnicity (%)      Caucasian      Asian  33.3 66.7  41.3 58.7 0.827 Height (cm) 163 ± 1.0 162.8 ± 6.6 0.921 Weight (kg) 59.5 ± 1.2 56.3 ± 1.0 0.041 Body mass index (kg/m2) 22.4 ± 0.3 21.2 ± 0.3 0.008 Waist circumference (cm) 66.4 ± 0.8 64.4 ± 0.7 0.058 Bone free fat mass (kg) 18.0 ± 0.8 15.7 ± 0.6 0.025 Bone free lean mass (kg) 38.1 ± 0.7 37.5 ± 0.6 0.470 Bone free body fat (%) 31.6 ± 0.9 29.1 ± 0.8 0.033 Δ Height (%) 0.0 ± 0.0 0.0 ± 0.0 0.349 Δ Weight (%) 0.7 ± 0.7 1.6 ± 0.7 0.332 Δ BMI (%) 0.2 ± 0.6 1.3 ± 0.8 0.297 Age (years) 21.9 ± 3.3 22.4 ± 3.4 0.399     76  Higher CDRa (n=60) Lower CDRb (n=63) P valuec Δ Waist circumference (%) 2.8 ± 0.6 3.4 ± 0.6 0.473 Δ Bone free fat mass (%) 2.5 ± 2.0 5.6 ± 2.5 0.339 Δ Bone free lean mass (%) 0.9 ± 0.5 0.6 ± 0.5 0.663 Δ Bone free percent body fat (%) 1.0 ± 1.4 3.3 ± 1.8 0.309 Total physical activityd 7.9 ± 0.2 7.7 ± 0.2 0.478 Sport activityd 2.6 ± 0.1 2.5 ± 0.1 0.388 Disinhibitione 6.8 ± 0.4 5.3 ± 0.4 0.006 Hungere 5.2 ± 0.3 5.4 ± 0.3 0.665 General stress Z-scoref 0.04 ± 0.1 -0.05 ± 0.1 0.536 Energy intakeg (kcal) 1676 ± 61 1443 ± 60 0.009 Study hormone users (%) 13.3 15.9 0.690 Duration study hormone use (months) 7.8 ± 2.6 8.2 ± 1.5 0.894 Number of cycles analysed 13.4 ± 0.9 13.8 ± 0.9 0.758 Δ Waist circumference (%) 2.8 ± 0.6 3.4 ± 0.6 0.473 Δ Bone free fat mass (%) 2.5 ± 2.0 5.6 ± 2.5 0.339 Δ Bone free lean mass (%) 0.9 ± 0.5 0.6 ± 0.5 0.663 Δ Bone free percent body fat (%) 1.0 ± 1.4 3.3 ± 1.8 0.309 Total physical activityd 7.9 ± 0.2 7.7 ± 0.2 0.478 Sport activityd 2.6 ± 0.1 2.5 ± 0.1 0.388 Disinhibitione 6.8 ± 0.4 5.3 ± 0.4 0.006 Hungere 5.2 ± 0.3 5.4 ± 0.3 0.665 General stress Z-scoref 0.04 ± 0.1 -0.05 ± 0.1 0.536 Energy intakeg (kcal) 1676 ± 61 1443 ± 60 0.009 Study hormone users (%) 13.3 15.9 0.690 Duration study hormone use (months) 7.8 ± 2.6 8.2 ± 1.5 0.894 Cycle lengthh (days) 31.5 ± 0.5 30.1 ± 0.5 0.060 Subclinical ovulatory disturbancesi (%)    Caucasian       Asian 55.8 ± 4.0 64.5 ± 6.5 47.1 ± 4.7 34.1 ± 3.9 33.3 ± 5.9 34.9 ± 4.9 <0.001 UFCi (µg/24-hour)       Caucasian        Asian 28.0 ± 1.2 32.0 ± 2.1 25.8 ± 1.4 24.0 ± 1.1 22.6 ± 1.7 25.4 ± 1.8 0.021   Total body ΔaBMDg (%) 0.9 ± 0.2 1.2 ± 0.2 0.424 L1-4 ΔaBMDg (%) 1.0 ± 0.4 1.5 ± 0.4 0.323     77  Higher CDRa (n=60) Lower CDRb (n=63) P valuec Hip ΔaBMDg (%) -0.1 ± 0.2 0.4 ± 0.3 0.292 Data are presented as mean ± standard error. Questionnaire scores, energy intakes and UFC values are averages from assessments at baseline, and both follow-ups because values did not change over time by repeat measures General Linear Model. CDR; cognitive dietary restraint; UFC, 24-hour urinary free cortisol; ΔaBMD, annualised 2-year percent change in areal bone mineral density (g/cm2); L1-4, lumbar vertebrae 1 to 4. a. Women with Three Factor Eating Questionnaire Restraint subscale scores higher than or equal to the median (7.7) score. b. Women with Three Factor Eating Questionnaire Restraint subscale scores below the median (7.7) score. c. Level of significance of difference between women with higher and lower CDR by independent t-test or General Linear Model adjusted for covariates. d. Baecke Habitual Physical Activity Questionnaire, possible scores for sport 1-5 and total 3-15. e. Three Factor Eating Questionnaire subscale scores: Disinhibition (0-16); and Hunger (0-14). f. Z-score of the Perceived Stress Scale and Daily Stress Inventory Impact and Frequency subscales assessed on the days of urine collection. g. Adjusted for body mass index (kg/m2). h. N=114; Adjusted for weight (kg) and gynaecological age. i. N=114; Adjusted for baseline gynaecological age and body mass index (kg/m2). Interactive effect of ethnicity-by-CDR: F=3.103, P=0.081. Main effect of ethnicity: F=1.930, P=0.168. j. Adjusted for urine volume (L/24 hour). Interactive effect of ethnicity-by-CDR: F=4.5866, P=0.034. Main effect of ethnicity: F=0.218, P=0.641.  3.3.7 Differences by subclinical ovulatory disturbances median split  Participants were classified by median split as those with higher (≥38.8% of cycles had a short luteal phase or were anovulatory) or lower (<38.8%) subclinical ovulatory disturbances as shown in Table 3.4. There were no differences in energy intake, number of cycles analysed, General Stress score or physical activity level. Women with a higher frequency of subclinical ovulatory disturbances were younger, had lower gynaecological age, more positive Δlean mass, and BMI tended to be higher. Other baseline or Δ anthropometrics, and the frequency or duration of study hormone use did not differ. Findings were consistent when analyses were repeated including only those with normal BMI levels (18.5-24.9 kg/m2) and when those having >10, >5 and >3 cycles of available data were analysed (data not shown).  After adjusting for baseline gynaecological age and BMI (and Δlean mass for ΔaBMD, Table 3.4), women with more frequent ovulatory disturbances reported higher CDR scores and had less positive hip and L1-4 ΔaBMD. Other questionnaire scores, total body ΔaBMD and UFC (urine volume as additional covariate) did not differ. No study outcome variables showed a significant main effect of ethnicity or an ethnicity-by-ovulatory disturbances interaction (data not shown).       78 Table 3.4  Differences between healthy premenopausal women with higher and lower percentage of cycles with subclinical ovulatory disturbances (median split) in menstrual cycle characteristics, age, anthropometrics, Δanthropometrics, questionnaire scores, 24-hour urinary free cortisol and 2-year ΔaBMD (n=114)   Higher Subclinical Ovulatory Disturbancesa (n=57) Lower  Subclinical Ovulatory Disturbancesb (n=57) P valuec Number of cycles analysed 12.8 ± 0.9 14.5 ± 0.9 0.193 Cycle length (days) 30.9 ± 0.5 30.7 ± 0.6 0.754 Study hormone users (%) 17.5 15.8 0.802 Duration study hormone use (months) 8.9 ± 2.2 8.5 ± 2.1 0.889 Age (years) 21.4 ± 0.4 22.9 ± 0.5 0.011 Gynaecological age (years) 8.5 ± 0.4 10.8 ± 0.5 0.001 Ethnicity (%)      Caucasian      Asian  38.6 61.4  36.8 63.2 0.847 Height (cm) 162.5 ± 1.0 163.2 ± 0.8 0.602 Weight (kg) 58.6 ± 1.1 57.1 ± 1.2 0.360 Body mass index (kg/m2) 22.3 ± 0.3 21.4 ± 0.3 0.085 Waist circumference (cm) 65.8 ± 0.8 64.7 ± 0.8 0.342 Bone free fat mass (kg) 17.5 ± 0.8 16.1 ± 0.7 0.195 Bone free lean mass (kg) 37.6 ± 0.6 38.0 ± 0.7 0.725 Bone free percent body fat (%) 31.3 ± 0.9 29.2 ± 0.9 0.092 Δ Height (%) 0.0 ± 0.0 0.0 ± 0.0 0.044 Δ Weight (%) 1.1 ± 0.6 1.3 ± 0.9 0.830 Δ BMI (%) 0.5 ± 0.6 1.0 ± 0.9 0.625 Δ Waist circumference (%) 3.1 ± 0.8 3.0 ± 0.6 0.950 Δ Bone free fat mass (%) 2.9 ± 1.9 5.5 ± 2.8 0.441 Δ Bone free lean mass (%) 1.6 ± 0.5 0.0 ± 0.4 0.018 Δ Bone free percent body fat (%) 0.7 ± 1.5 3.7 ± 2.0 0.227 Total physical activityde 7.9 ± 0.2 7.7 ± 0.2 0.621 Sport activityde 2.5 ± 0.1 2.5 ± 0.1 0.963 Restraintef 8.7 ± 0.5 7.1 ± 0.5 0.040 Disinhibitionef 6.3 ± 0.4 5.8 ± 0.4 0.425 Hungeref 5.5 ± 0.3 5.1 ± 0.3 0.431     79  Higher Subclinical Ovulatory Disturbancesa (n=57) Lower  Subclinical Ovulatory Disturbancesb (n=57) P valuec General stress Z-scoreeg 0.0 ± 0.1 0.0 ± 0.1 0.868 Energy intakee (kcal) 1588 ± 67 1516 ± 67 0.468 UFCh (µg/24-hour) 25.8 ± 1.3 25.6 ± 1.3 0.894 Total body ΔaBMDi (%) 1.0 ± 0.2 1.1 ± 0.2 0.775 L1-4 ΔaBMDi (%) 0.7 ± 0.4 1.9 ± 0.4 0.034 Hip ΔaBMDi (%) -0.6  ± 0.3 0.9 ± 0.3 0.001 Data are presented as mean ± standard error. Questionnaire scores, energy intakes and UFC values are averages from assessments at baseline and both follow-ups because values did not change over time by repeat measures General Linear Model. UFC, 24-hour urinary free cortisol; ΔaBMD, annualised 2-year percent change in areal bone mineral density (g/cm2); L1-4, lumbar vertebrae 1 to 4. a. Menstrual cycles were anovulatory and/or had a luteal phase length <10 days by least squares quantitative basal temperature analysis ≥38.8% of the time. b. Menstrual cycles were anovulatory and/or had a luteal phase length <10 days by least squares quantitative basal temperature analysis <38.8% of the time. c. Level of significance of difference between women with higher and lower percentage of cycles with subclinical ovulatory disturbances by independent t-test or General Linear Model adjusted for covariates. d. Baecke Habitual Physical Activity Questionnaire, possible scores for sport 1-5 and total 3-15. e. Adjusted for baseline gynaecological age and body mass index (kg/m2). f. Three Factor Eating Questionnaire subscales scores: Restraint (possible score 0-21); Disinhibition (0-16); and Hunger (0-14). g. Z-score of the Perceived Stress Scale and Daily Stress Inventory Impact and Frequency subscales assessed on the days of urine collection. h. Adjusted for urine volume, and baseline gynaecological age and body mass index (kg/m2). i. Adjusted for change in lean mass, and baseline gynaecological age and body mass index (kg/m2).  3.4 Discussion The purpose of this study was to examine whether the frequency of subclinical ovulatory disturbances and UFC differed by level of CDR, and if these variables affected change in aBMD over two years in healthy young women. We confirmed previous reports of an increased frequency of subclinical ovulatory disturbances and higher UFC among women with higher CDR (Table 3.3). We also confirmed that less positive aBMD changes occurred in women with more frequency subclinical ovulatory disturbances. However, UFC did not differ by the percentage of subclinical ovulatory disturbances (Table 3.4) and there was no difference in aBMD change by CDR level (Table 3.3). Additionally, UFC was not associated aBMD change. Consequently, whether cortisol mediates the relationship between CDR, ovulatory disturbances and aBMD (Figure 3.1) still remains to be established.    The most noteworthy finding of the current study was the confirmation of less positive aBMD changes at lumbar spine among women with more frequent subclinical     80 ovulatory disturbances [12-15,17]. We also found less positive hip aBMD change in women with more frequent ovulatory disturbances. It is well established that overt menstrual cycle abnormalities lead to bone loss [10]. Yet, whether anovulation and short LPL are associated with bone loss remains controversial [9]. Our findings suggest that they are, although their impact is modest. One potential reason for conflicting findings regarding bone and ovulatory disturbances could be the duration of ovulatory observations. As ovulatory function is highly variable [42], long-term monitoring is critical to correctly identify women with subclinical disturbances. The studies which did not observe associations between ovulatory disturbances and bone monitored two to four cycles [11,16]. In contrast, the current study and most others that did see a relationship monitored nine to 14 cycles [13,15,17].  Our results corroborate that ovulatory disturbances are more common among women reporting higher CDR [6-8,43-44]. The only other prospective study of CDR and ovulatory function did not find a difference by CDR level in the proportion of women with >3 subclinical ovulatory disturbances cycles [17]. The null relationship between CDR and subclinical ovulatory disturbances in that study is most likely due to the very low proportion of women with >3 cycles with subclinical ovulatory disturbances (approximately 7%), making detection of a difference more difficult. The authors did not describe why they classified women on that basis, but the low prevalence of subclinical ovulatory disturbances may be related to their sample‘s greater gynaecological maturity, such that psychosocial stress would be less likely to affect cycles. Furthermore, the definition of short LPL in that study (<10 days by urinary LH surge detection) may underestimate the prevalence of cycles with prevalence. Urine LH peaks before follicular collapse by ultrasound [45], whereas the significant rise in basal temperature detected with LS-QBT occurs ~2 days after the LH peak [37]. To equate the two methods, the criterion for short LPL based on urinary LH would be <11-12 days, rather than <10 DAYS used with LS-QBT. Additional support that eating and body stresses can lead to menstrual cycle and ovulatory disturbances comes from studies using measures other than Restraint. Higher scores on the Eating Attitudes Test, and the Drive For Thinness and Bulimia subscales of the Eating Disorder Inventory have been reported in women with functional hypothalamic amenorrhea (FHA) versus women with organic causes of amenorrhea and/or regularly menstruating women [46-48].  It has been suggested that normal- or under-weight women with higher CDR experience more frequent subclinical ovulatory disturbances due to caloric restriction and other dieting behaviours [17]. However, examination of data from studies, which observed relationships among ovulatory disturbances and CDR or similar eating attitudes [6-8,43-44,46-48], suggest that mechanism is unlikely. In our current sample, for example, physical activity did not differ by     81 CDR level and women with higher CDR actually had higher BMI values and energy intakes (Table 3.3). Moreover, we used the EDE questionnaire to establish that participants did not exhibit clinical eating disorders [34-35,41]. Therefore, it is unlikely that an energy deficit in women with higher CDR caused ovulatory disturbances in the current study. Finally, while various life stresses are associated with anovulation and short LPL cycles [49], among our participants general stress was not associated with subclinical ovulatory disturbances and did not differ by CDR level. In fact the only measured variable that differed by the frequency of ovulatory disturbances in our sample was CDR score and ΔaBMD.  Contrary to our hypothesis, we did not confirm that cortisol plays a role in mediating the relationships among CDR, ovulatory disturbances and change in bone density. While UFC was higher among women with higher CDR, as previously reported [24-28], it was not correlated with CDR score in the entire group, and did not differ by %SOD level. It is generally accepted that stress-induced HPA axis activation is related to menstrual cycle and ovulatory disturbances [49]. Corticotropin-releasing hormone alters pulsatile gonadotropin-releasing hormone (GnRH) release (Figure 3.1), this leads to impaired secretion of reproductive hormones and a spectrum of disturbances of decreasing severity from amenorrhea to oligomenorrhea, to regular cycles with anovulation or short luteal phases [23]. However, it could be that eating and body stress impacts ovulatory function via neuroendocrine pathways that do not involve the HPA axis. The secretion of GnRH can be affected by numerous neurotransmitters and neuropeptides of which several relate to appetite control [49]. This may be relevant to CDR in which to women attempt to override physiological cues to hunger.  Furthermore, we found that UFC was elevated in Caucasians with higher CDR (and that UFC and CDR tended to correlate in Caucasians), but UFC did not differ by CDR level among Asians. In a study of young, healthy, regularly menstruating women, Asian women had 6-beta-hydroxycortisol: cortisol ratios that were two to three times lower than Caucasians [40]. This is significant as the 6-beta-hydroxycortisol:cortisol ratio is an indirect indicator of cytochrome P450 3A4 activity, an enzyme that is involved in the metabolism of steroids including cortisol, estradiol and progesterone [40]. However, as has also been reported by others [50-51], we did not see a difference in UFC by ethnicity. Moreover, Asians with higher CDR, despite having similar UFC as Asians with lower CDR, had more frequent subclinical ovulatory disturbances. Taken together, this suggests that cortisol may not mediate the association between CDR and ovulatory function.  That UFC differed by level of CDR among Caucasians but not Asians is an interesting finding. There were no differences in CDR, general stress, the frequency of subclinical ovulatory disturbances, UFC or aBMD change by ethnicity. If cortisol is metabolised more rapidly by Asians [40], we would still expect to see the same pattern of difference by CDR, although lower     82 absolute levels. It could be that despite similar CDR scores, the qualitative experience of eating and body-related stress differs between Asians and Caucasians. The influence of ethnicity on the experience of CDR as a stressor has not yet been explored.   We also did not find an association between UFC and change in aBMD. Although reduced BMD is observed in hypercortisolism [23], it is less clear if this occurs in healthy young women, when cortisol is elevated yet remains within the normal range [5,7,18-22,29]. Estradiol may mediate the relationship between cortisol and bone: women who continue to menstruate, such as our participants, would likely have normal estradiol levels. A major negative effect of cortisol on bone may be prevented by estrogen‘s antiresorptive effects [52]. This may explain why studies including women with oligomenorrhea found an association between higher restraint and lower aBMD [5,19]. It could also be that the association of bone change with elevated cortisol within the normal range is relatively subtle, and difficult to detect over two years. In fact, we observed a modest association between higher UFC and lower aBMD and BMC in this sample at baseline [29]. Although differences in the rate of bone loss between those with slightly elevated versus lower cortisol may be relatively small, over time the accumulated affects could substantially impact aBMD and fracture risk.   This study was not without limitations. Our sample was relatively homogeneous and our findings are generalisable only to those with similar characteristics. We did not account for osteoporosis family history or physical activity during adolescence. Both may be associated with aBMD in healthy premenopausal women. Twenty-two women used hormonal contraceptives during the study, however, these did not influence ΔaBMD. Moreover, in our study women initiated hormone use for contraception, not because of menstrual abnormalities. Although we screened for polycystic ovarian syndrome based on clinical symptoms, androgen levels were not measured. Our method of observing ovulatory function is not as accurate as cyclic determinations of reproductive hormones. However, the quantitative basal temperature method we used, LS-QBT, has been validated [37-38], is inexpensive and is acceptable to women. It should be noted that our method differs from previous qualitative methods in which basal temperature was plotted and each chart was visually inspected to determine if a shift is apparent. We used a computer programme to conduct quantitative analyses where a least squares criterion is used to determine ovulation if the maximum mean temperature difference between the first and latter parts of the cycle is statistically significant [37]. Furthermore, given within-person variability in ovulatory function, especially LPL [42], accurate characterisation requires that cycles be monitored over a long period. This method allowed us to accomplish this.        83  The use of DXA to measure aBMD may also be a limitation. DXA assesses bone mass rather than bone strength. Additional prospective studies examining eating attitudes, ovulatory function and bone would be improved by using quantitative computed tomography, which differentiates between cortical and trabecular bone or other measures that can document bone micro-architecture.  Our study contributes to the emerging field of research linking psychosocial and physiological health, a field that is ―changing what it means to be healthy‖ by defining well-being not only by our behaviours but our attitudes [53]. Interestingly, cognitive behavioural therapy has been shown to improve ovulatory function among women with FHA [53]. Whether cognitive behavioural therapy may help young women with disordered eating attitudes warrants investigation.  In summary, we confirmed that healthy premenopausal women higher CDR experienced more frequent subclinical ovulatory disturbances and that a higher occurrence of these disturbances resulted in less positive changes in bone density over two years. Although the magnitude of the effect on bone was modest, subclinical ovulatory disturbances may have a persistent negative influence on bone in young women [15]. Contrary to our hypothesis, although UFC was higher in women with greater CDR, we did not confirm that cortisol played a role in mediating associations among CDR, subclinical ovulatory disturbances and change in aBMD. Future studies would be improved by examining other potential mechanisms including neuropeptides. The high variability in BMD and ovulatory function indicate the need for a large sample with longer follow-up in order to firmly establish whether young women‘s eating attitudes can affect bone.       84 3.5 References  [1] Stunkard AJ, Messick S. The Three-Factor Eating Questionnaire to measure dietary restraint, disinhibition and hunger. J Psychosom Res 1985;29:71-83. [2] McLean JA, Barr SI. Cognitive dietary restraint is associated with eating behaviors, lifestyle practices, personality characteristics and menstrual irregularity in college women. Appetite 2003;40:185-92.  [3] Lowe MR, Annunziato RA, Markowitz JT, Didie E, Bellace DL, Riddell L, Maille C, McKinney S, Stice E. Multiple types of dieting prospectively predict weight gain during the freshman year of college. Appetite 2006;47:83-90. [4] Stice E, Cooper JA, Schoeller DA, Tappe K, Lowe MR. Are dietary restraint scales valid measures of moderate- to long-term dietary restriction? Objective biological and behavioral data suggest not. Psychol Assess 2007;19:449-58.  [5] Vescovi JD, Scheid JL, Hontscharuk R, De Souza MJ. Cognitive dietary restraint: impact on bone, menstrual and metabolic status in young women. Physiol Behav 2008;95:48-55.  [6] Barr SI, Janelle KC, Prior JC. Vegetarian vs nonvegetarian diets, dietary restraint, and subclinical ovulatory disturbances: prospective 6-mo study. Am J Clin Nutr 1994;60:887-94.  [7] Barr SI, Prior JC, Vigna YM. Restrained eating and ovulatory disturbances: possible implications for bone health. Am J Clin Nutr 1994;59:92-7.  [8] Schweiger U, Tuschl RJ, Platte P, Broocks A, Laessle RG, Pirke KM. Everyday eating behavior and menstrual function in young women. Fertil Steril 1992;57:771-5.  [9] Balasch J. Sex steroids and bone: current perspectives. Hum Reprod Update 2003;9:207-22.  [10] Gordon CM, Nelson LM. Amenorrhea and bone health in adolescents and young women. Curr Opin Obstet Gynecol 2003;15:377-84.  [11] De Souza MJ, Miller BE, Sequenzia LC, Luciano AA, Ulreich S, Stier S, Prestwood K, Lasley BL. Bone health is not affected by luteal phase abnormalities and decreased ovarian progesterone production in female runners. J Clin Endocrinol Metab 1997;82:2867-76.  [12] Sowers M, Randolph JF, Crutchfield M, Jannausch ML, Shapiro B, Zhang B, La Pietra M. Urinary ovarian and gonadotropin hormone levels in premenopausal women with low bone mass. J Bone Miner Res 1998;13:1191-202.  [13] Petit MA, Prior JC, Barr SI. Running and ovulation positively change cancellous bone in premenopausal women. Med Sci Sports Exerc 1999;31:780-7.      85 [14] Prior JC, Vigna YM, Schechter MT, Burgess AE. Spinal bone loss and ovulatory disturbances. N Engl J Med 1990;323:1221-7.  [15] Prior JC, Vigna YM, Barr SI, Kennedy S, Schulzer M, Li DK. Ovulatory premenopausal women lose cancellous spinal bone: a five year prospective study. Bone 1996;18:261-7.  [16] Waller K, Reim J, Fenster L, Swan SH, Brumback B, Windham GC, Lasley B, Ettinger B, Marcus R. Bone mass and subtle abnormalities in ovulatory function in healthy women. J Clin Endocrinol Metab 1996;81:663-8.  [17] Waugh EJ, Polivy J, Ridout R, Hawker GA. A prospective investigation of the relations among cognitive dietary restraint, subclinical ovulatory disturbances, physical activity, and bone mass in healthy young women. Am J Clin Nutr 2007;86:1791-801.  [18] Bacon L, Stern JS, Keim NL, Van Loan MD. Low bone mass in premenopausal chronic dieting obese women. Eur J Clin Nutr 2004;58:966-71.  [19] Barrack MT, Rauh MJ, Barkai HS, Nichols JF. Dietary restraint and low bone mass in female adolescent endurance runners. Am J Clin Nutr 2008;87:36-43.  [20] McLean JA, Barr SI, Prior JC. Dietary restraint, exercise, and bone density in young women: are they related? Med Sci Sports Exerc 2001;33:1292-6.  [21] Van Loan MD, Keim NL. Influence of cognitive eating restraint on total-body measurements of bone mineral density and bone mineral content in premenopausal women aged 18-45 y: a cross-sectional study. Am J Clin Nutr 2000;72:837-43. [22] Nickols-Richardson SM, Beiseigel JM, Gwazdauskas FC. Eating restraint is negatively associated with biomarkers of bone turnover but not measurements of bone mineral density in young women. J Am Diet Assoc 2006;106:1095-101.  [23] Chiodini I, Torlontano M, Carnevale V, Trischitta V, Scillitani A. Skeletal involvement in adult patients with endogenous hypercortisolism. J Endocrinol Invest 2008;31:267-76.  [24] Anderson DA, Shapiro JR, Lundgren JD, Spataro LE, Frye CA. Self-reported dietary restraint is associated with elevated levels of salivary cortisol. Appetite 2002;38:13-7.  [25] McLean JA, Barr SI, Prior JC. Cognitive dietary restraint is associated with higher urinary cortisol excretion in healthy premenopausal women. Am J Clin Nutr 2001;73:7-12.  [26] Putterman E, Linden W. Cognitive dietary restraint and cortisol: importance of pervasive concerns with appearance. Appetite 2006;47:64-76.  [27] Rideout CA, Linden W, Barr SI. High cognitive dietary restraint is associated with increased cortisol excretion in postmenopausal women. J Gerontol A Biol Sci Med Sci 2006;61:628-33.      86 [28] Rutters F, Nieuwenhuizen AG, Lemmens SG, Born JM, Westerterp-Plantenga MS. Hyperactivity of the HPA axis is related to dietary restraint in normal weight women. Physiol Behav 2009;96:315-9. [29] Bedford JL, Barr SI. The relationship between 24-h urinary cortisol and bone in healthy young women. Int J Behav Med 2009 October 3. DOI 10.1007/s12529-009-9064-2. [30] Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav 1983;24:385-96.  [31] Brantley PJ, Waggoner CD, Jones GN, Rappaport NB. A Daily Stress Inventory: development, reliability, and validity. J Behav Med 1987;10:61-74.  [32] Baecke JA, Burema J, Frijters JE. A short questionnaire for the measurement of habitual physical activity in epidemiological studies. Am J Clin Nutr 1982;36:936-42.  [33] Fairburn CG, Beglin SJ. Assessment of eating disorders: interview or self-report questionnaire? Int J Eat Disord 1994;16:363-70.  [34] Mond JM, Hay PJ, Rodgers B, Owen C. Eating Disorder Examination Questionnaire (EDE-Q): norms for young adult women. Behav Res Ther 2006;44:53-62.  [35] Stice E, Peterson CB. Eating disorders. In: Mash EJ, Barkely RA, editors. Assessment of childhood disorders, 4th edition. New York: The Guilford Press; 2007. p. 751-80.  [36] Csizmadi I, Kahle L, Ullman R, Dawe U, Zimmerman TP, Friedenreich CM, Bryant H, Subar AF. Adaptation and evaluation of the National Cancer Institute's Diet History Questionnaire and Nutrient Database for Canadian populations. Public Health Nutr 2007;10:88-96. [37] Prior JC, Vigna YM, Schulzer M, Hall JE, Bonen A. Determination of luteal phase length by quantitative basal temperature methods: validation against the midcycle LH peak. Clin Invest Med 1990;13:123-31.  [38] Bedford JL, Prior JC, Hitchcock CL, Barr SI. Detecting evidence of luteal activity by least-squares quantitative basal temperature analysis against urinary progesterone metabolites and the effect of wake-time variability. Eur J Obstet Gynecol Reprod Biol 2009;146:76-80.  [39] Taylor RL, Machacek D, Singh RJ. Validation of a high-throughput liquid chromatography-tandem mass spectrometry method for urinary cortisol and cortisone. Clin Chem 2002;48:1511-9.  [40] Lin Y, Anderson GD, Kantor E, Ojemann LM, Wilensky AJ. Differences in the urinary excretion of 6-beta-hydroxycortisol/cortisol between Asian and Caucasian women. J Clin Pharmacol 1999;39:578-82. [41] Luce KH, Crowther JH, Pole M. Eating Disorder Examination Questionnaire (EDE-Q): norms for undergraduate women. Int J Eat Disord 2008;41:273-6.      87 [42] Cole LA, Ladner DG, Byrn FW. The normal variabilities of the menstrual cycle. Fertil Steril 2009;91:522-7.  [43] Cano A, Aliaga R. Characteristics of urinary luteinizing hormone (LH) during the induction of LH surges of different magnitude in blood. Hum Reprod 1995;10:63-7. [44] Lebenstedt M, Platte P, Pirke KM. Reduced resting metabolic rate in athletes with menstrual disorders. Med Sci Sports Exerc 1999;31:1250-6.  [45] Scheid JL, Williams NI, West SL, VanHeest JL, De Souza MJ. Elevated PYY is associated with energy deficiency and indices of subclinical disordered eating in exercising women with hypothalamic amenorrhea. Appetite 2009;52:184-92.  [46] Laughlin GA, Dominguez CE, Yen SS. Nutritional and endocrine-metabolic aberrations in women with functional hypothalamic amenorrhea. J Clin Endocrinol Metab 1998;83:25-32.  [47] Marcus MD, Loucks TL, Berga SL. Psychological correlates of functional hypothalamic amenorrhea. Fertil Steril 2001;76:310-6.  [48] Schneider LF, Monaco SE, Warren MP. Elevated ghrelin level in women of normal weight with amenorrhea is related to disordered eating. Fertil Steril 2008;90:121-8.  [49] Meczekalski B, Podfigurna-Stopa A, Warenik-Szymankiewicz A, Genazzani AR. Functional hypothalamic amenorrhea: current view on neuroendocrine aberrations. Gynecol Endocrinol 2008;24:4-11.  [50] Reynolds RM, Fischbacher C, Bhopal R, Byrne CD, White M, Unwin N, Walker BR. Differences in cortisol concentrations in South Asian and European men living in the United Kingdom. Clin Endocrinol (Oxf) 2006;64:530-4. [51] Hsiao HP, Iglesias ML, Keil MF, Boikos S, Robinson-White A, Stratakis CA. Differences in cortisol levels and body mass index between East Asians and Caucasians with Cushing's syndrome: an 'East Asian' phenotype for Cushing syndrome. Clin Endocrinol (Oxf) 2007;66:753-5.  [52] Tauchmanova L, Pivonello R, De Martino MC, Rusciano A, De Leo M, Ruosi C, Mainolfi C, Lombardi G, Salvatore M, Colao A. Effects of sex steroids on bone in women with subclinical or overt endogenous hypercortisolism. Eur J Endocrinol 2007;157:359-66.  [53] Berga SL, Loucks TL. Use of cognitive behavior therapy for functional hypothalamic amenorrhea. Ann N Y Acad Sci 2006;1092:114-29.       88 Chapter 4:   Negative eating and body attitudes are associated with higher daytime ambulatory blood pressure in healthy young women1                                                 1 A version of this chapter has been submitted for publication: Bedford JL, Linden W, Barr SI. Negative eating and body attitudes are associated with elevated daytime ambulatory blood pressure in healthy young women. Date of Submission: April 2010.     89 4.1  Introduction For more than 100 years, cardiovascular disease (CVD) has been the leading cause of death among American adults [1]. High blood pressure (BP) or hypertension (systolic BP >140 or diastolic BP >90 mm Hg) is one of the strongest CVD risk factors [2] and affects nearly one in three Americans [1]. Although the prevalence of hypertension among young adults is low [1], BP while young is associated with BP later in life [3]. Additionally, young adult BP is positively correlated with atherosclerosis [4] and also predicts carotid intima-media thickness [5], another CVD risk marker. Furthermore, young adults with prehypertension (systolic BP 120-139 or diastolic BP 80-89 mm Hg) were shown to have an increased risk of coronary calcium atherosclerosis 15-20 years later, after adjustment for other risk factors including current BP [6]. Thus, BP level even in the young and healthy is related to future CVD risk, accentuating the need to fully understand factors that influence BP in this population. Evidence is accumulating that the subjective experience of stress may influence cardiovascular outcomes, mediated by the physiological stress response [7]. When a stressor is perceived, the central nervous system and peripheral components are activated, seeking to maintain homeostasis via adaptive responses to deal with the perceived threat [8]. Two allostatic mediators of the stress response are activation of the hypothalamic-pituitary-adrenal (HPA) axis, resulting in increased secretion of the stress hormone cortisol, and stimulation of the sympathetic nervous system which causes BP to rise [8]. Increased cortisol and BP are beneficial during acute stress; however, continuous elevations can lead to allostatic overload, causing ―wear-and-tear‖ on body systems [8]. High cortisol levels may also be independently associated with increased BP [9].  Exposure to laboratory stressors clearly elevates BP and cortisol in otherwise healthy adults [10]. It is less clear whether chronic psychosocial stressors of limited salience are sufficient to increase BP and cortisol [7]. When evaluating the physiological effects of chronic stress, the measurement of clinical BP may be of limited relevance. Perceived stress that is encountered over the course of the day, rather than discrete events or laboratory tasks, may not be captured by a single measurement of BP. Therefore, ambulatory BP (ABP) monitoring while performing the activities of daily living may be a more sensitive tool. After controlling for CVD risk factors including clinical BP, ABP is independently associated with cardiovascular morbidity and mortality [11-13].  Occupational stressors have received the most attention in the investigation of stress, cortisol and ABP. The results among middle-aged women are inconsistent such that some [14-18] but not all [15,19-21] studies report higher ABP and cortisol in women reporting higher occupation-related stress. It could be that women experience other stressors that either interact with work stress or are more relevant to stress perception [22-24]. Other chronic stressors     90 associated with ABP and cortisol identified among women include financial stress [25-26], family/marital stress [15,27-29] and lack of social support [30-31]. Given that the majority of women report negative attitudes towards food and body [32-34], we and others have hypothesised that eating/body attitudes may be subtle but chronic daily stressors sufficient to activate the physiological stress response, and potentially lead to negative health outcomes.  A large number of psychometric scales have been developed to assess eating/body attitudes [35], and some studies have detected associations between increasing scores on these scales and cortisol in healthy women [36-41]. Most work in the area of eating/body stress has employed the Three Factor Eating Questionnaire (TFEQ) Restraint subscale to assess cognitive dietary restraint: the perception that one is constantly monitoring and attempting to limit food intake in an effort to achieve or maintain a perceived ideal body weight [42]. Generally, there are no or only minimal differences by level of dietary restraint in women‘s self-reported energy intakes, relative weight, weight changes or dieting behaviour [34,43-44]. This suggests that some negative eating/body attitudes are not necessarily indicative of dieting or disordered eating behaviours. Taken together, these studies also support the idea that the experience of negative eating/body attitudes could be associated with the physiological stress response. It therefore seems reasonable to ask whether or not these attitudes are also associated with BP. The authors are aware of only one study that examined eating attitudes in relation to ABP (Koo-Loeb et al., 2000).[45]. In that study, no differences in 24-hour ABP were observed between healthy university-aged women with very high or very low scores on the Eating Disorder Inventory Bulimia subscale who did not meet the diagnostic criteria for eating disorders; although 24-hour urinary cortisol was higher in women with higher scores [45]. Participants completed the ABP and urine assessments after administration of the diagnostic interview for bulimia nervosa; answering psychosocial questionnaires; and performing the laboratory stress test [45]. Thus, findings do not reflect participants‘ ―usual‖ cortisol and ABP, which was the goal of the current study.  Therefore, given that psychosocial stressors are capable of elevating ABP in healthy middle-aged women [14-16,22-25,27-31], and that negative eating/body attitudes are common among women [32-34], it is reasonable to postulate that eating/body attitudes may be a source of subtle but chronic stress with the potential to elevate BP. This relationship could be most evident among university-aged women, since others stressors (i.e. occupational, family) would be less significant for most. Thus, the objective of this study was to examine whether women with negative versus neutral/positive eating/body attitudes had higher 24-hour urinary free cortisol (UFC) and daytime ABP. To fully conceptualise the experience of eating/body stress, several body image and eating attitude questionnaires, which have previously been associated     91 with cortisol [36-41], were included because we hypothesised that they may also be associated with BP, an additional health outcome of chronic physiological stress.  In order to differentiate stress that is specific to eating and body image from ―general stress‖, chronic perceived stress and stressful events for the days of cortisol and ABP monitoring were assessed. We hypothesised that those with negative eating/body attitudes were not highly stressed people in general, but experienced stress specific to food and weight. Finally, in order to distinguish between the potential effect of cognitive versus behavioural aspects of eating/body attitudes on ABP and cortisol, current weight loss effort was also examined. We hypothesised that cortisol and ABP would not differ by weight loss effort, supporting our hypothesis that it is the subjective experience of stress related to food and weight, rather than behaviours, that are associated with negative health outcomes.  4.2 Methods 4.2.1 Participants Potential participants were recruited between August and December 2006 for a 2-y bone density study from university classes and using poster advertisements (Appendix 13). Eligibility was assessed by telephone interview in 148 interested women (Appendix 15). Criteria included: age 19-35, no pregnancy/breastfeeding currently or within 12 months, regular menstrual cycles (menses every 21-35 days in the previous ≥6 months), non-obese (self-reported body mass index (BMI) 18-30 kg/m2), consistent sleep patterns (arise and retire at approximately the same time most days) and absence of medical conditions (current or previous diagnosis of eating disorder, polycystic ovarian syndrome, Cushing‘s syndrome, inflammatory conditions, hypertension, hyperthyroidism or hirsutism) or use of medications (oral contraceptives, progesterone or glucocorticoids currently or within the past 6 months) that could affect study variables. Of the 142 eligible women screened, 140 were oriented to the study. Data collection for the cross-sectional study presented here occurred 6-12 months following enrollment. During that time interval, seven participants moved, four no longer wanted to participate and two became ineligible. Results are reported for the 120 women with complete data. The study protocol was approved by the university‘s Clinical Research Ethics Board (Appendix 17), and written informed consent was obtained from all participants (Appendix 16). Participants were provided with travel compensation (Appendix 18) and a $30 gift card for their participation (Appendix 19).   4.2.2 Procedure Participants met with an investigator at UBC for study orientation. A questionnaire package (Appendix 23) was given to complete at home, which included a food frequency     92 questionnaire and validated self-report instruments (Appendix 24), as well as questions to elicit demographic information and weight loss effort. Height and weight were measured in duplicate. From these data, BMI was calculated.  Participants were fitted with an ABP monitor including demonstration of cuff placement on the non-dominant arm over the brachial artery. A sample reading was performed to familiarise them with the process. Monitoring for 12-h was completed within three days on a ―normal day‖, avoiding any unusual physical or mental stresses, or heavy physical activity while wearing the monitor. Detailed written instructions similar to those provided verbally were given for review prior to starting the procedure (Appendix 26). The monitor was programmed to take blinded measurements of ABP and heart rate every 30 minutes. Participants were instructed to keep their arm still during readings and, if possible, to be seated. If there was too much movement, the monitor was programmed to abort and re-try one minute later. Immediately after each reading, participants recorded their concurrent activity in a provided diary (Appendix 26). After 12 hours, participants removed the monitor and completed the Daily Stress Inventory [46] (Appendix 25).  Materials and oral and written instructions for home completion of a 24-hour urine collection were reviewed (Appendix 20). Participants were instructed to complete the urine collection within several weeks of the meeting, on a different ―normal day‖, after reviewing written instructions. On the day of collection, participants discarded their first urine void, recorded the time this occurred and then collected all subsequent voids for 24 hours including a void at the recorded time the following morning. After their last void, participants completed the Daily Stress Inventory [46] (Appendix 25).   4.2.3 Questionnaires 4.2.3.1 Eating and body attitudes The TFEQ pertains to three dimensions of eating attitudes: Cognitive Dietary Restraint, the perception that one is constantly monitoring and attempting to limit food intake to achieve a perceived ideal body weight; Disinhibition, which is the tendency to overeat when restraint is removed; and Hunger, which assesses susceptibility to hunger [42]. Two subscales from the Eating Disorders Inventory were included: Drive for Thinness with higher scores indicating extreme concern with weight, dieting and the intense pursuit of thinness, and Bulimia which assesses one‘s tendency to think about and engage in uncontrolled overeating [47]. The shortened Body Shape Questionnaire was used to measure participants‘ body dissatisfaction caused by feelings of being fat [48]. The Beliefs About Appearance Scale assesses the degree of agreement with beliefs about the perceived importance of appearance for relationships, achievement, self-view and feelings [49]. These beliefs are thought to underlie the desire to     93 restrict eating, criticise the body and focus on appearance-related stimuli [49]. From these seven questionnaires/subscales, a single standardised ―Eating/Body Attitude‖ Z-score was calculated.   4.2.3.2 General stress  The Perceived Stress Scale was used to evaluate participants‘ perception of stress during the previous month [50]. To account for everyday minor stressful events, the Daily Stress Inventory [46] (Appendix 25) was completed after ABP and cortisol assessments. Participants indicated the frequency of 58 everyday minor stressful events which may have occurred during the day (Frequency score). They also ranked the intensity (Impact score) on a scale of 1 (―not at all stressful‖) to 7 (―caused me to panic‖). From these three questionnaires/subscales, a single standardised ―General Stress‖ Z-score was calculated.   4.2.3.3 Weight loss effort Participants were asked ―are you currently trying to lose weight?‖ and grouped as those reporting and not reporting current weight loss attempts. To determine energy intake, the Diet History Questionnaire (version 1.0, National Institutes of Health, Applied Research Program, National Cancer Institute, 2002) was completed. Scannable questionnaires were analysed with a Canadian version of the programme [51]. All reported energy intakes were within range considered biologically plausible (600-3500 kcal).  4.2.4 Urine analysis At the Vancouver General Hospital Laboratory, 24-hour urine volume was measured in duplicate, and aliquots were frozen and stored prior to analysis of urinary free cortisol (UFC, µg/24 hour) by high-throughput liquid chromatography and tandem mass spectrometry [52].   4.2.5 ABP measurement  The Spacelabs 90207 ABP monitor (Redmond, WA) measured 12-hour average systolic BP), diastolic BP, mean arterial pressure and heart rate. Monitoring for 12 hours relative to 24 hours avoids discomfort during sleep [53] and provides meaningful data regarding stress during participants‘ typical activities. We have found that 8-hour and 24-hour measurements correlated with r>0.90 (unpublished observation).   Participants‘ data were reviewed following the modified Casadei criteria [54]. Readings were considered artifactual if: systolic BP <70 or >240 mm Hg, diastolic BP <40 or >140 mm Hg, or heart rate <40 or >125 beats per minute. When any of these criteria were met, all data for that time point were excluded. This resulted in 18 participants having one reading excluded and     94 three participants having two readings. Paired t-tests revealed that ABP before and after data editing were not significantly different (data not shown). After editing, the mean ± standard deviation (SD) number of readings per participant was 23.4±1.5, range 18-26. To account for physical activity during ABP monitoring, participants recorded their activity in a diary at the time of each ABP measurement. Each entry was coded as either sedentary and given a score of one (e.g. sitting in class/work, watching television, studying/reading) or active and given a score of two (e.g. laundry, cooking, walking). A continuous score for activity during ABP (ABP-activity) was derived by summing the diary codes and dividing by the number of readings available for each participant. As physical fitness may also be associated with ABP, the Baecke Questionnaire of Habitual Physical Activity was used to determine usual activity levels at work, in sport, and during leisure [55].  4.2.6 Statistical analyses Data were coded, verified, entered into SPSS software (version 17, SPSS Inc., 2008) and crosschecked for accuracy. Physiologic variables were examined for outliers (mean ± >4SD) and none were present. Descriptive statistics were used to characterise the sample.  Single standardised ―Eating/Body Attitude‖ and ―General Stress‖ Z-scores were calculated. For each questionnaire, participants‘ scores were subtracted from the corresponding mean and divided by the SD. The questionnaire Z-scores were then summed and divided by the number of scales/subscales included. Because higher scores on the eating and body attitude questionnaires reflect more negative eating/body attitudes, Z-scores were inverted so that women were classified as having either negative (Z-score <0) or neutral/positive (Z-score ≥0) attitudes towards food and body. Higher General Stress Z-score reflects higher levels of perceived psychosocial stress. Pearson‘s correlations were used to identify correlates that could potentially confound analyses of ABP and UFC. Partial correlations adjusted for potential confounders were conducted between Eating/Body Attitude Z-score, General Stress Z-score, UFC and ABP. Chi-square for categorical variables and independent t-tests and General Linear Modeling (with appropriate covariates) for continuous variables were used to examine differences between women with negative versus neutral/positive Eating/Body Attitudes, and between those reporting and not reporting current weight loss attempts. Interactive effects were also examined in order to differentiate between the cognitive and behavioural aspects of Eating/Body Attitudes. As cortisol metabolism may differ between Asians and Caucasians [56], interactions between ethnicity and Eating/Body Attitudes were examined with regard to UFC. For all analyses, cases were excluded pairwise and the significance level for all analyses was P0.05.       95 4.3 Results 4.3.1 Participant characteristics All women were normotensive (systolic BP ≤135 and diastolic BP ≤85 mm Hg). Most participants were currently students (86%) and single (91%). All had completed some post-secondary education. Similar to the student population of UBC, 62.5% of the sample was Asian and the remainder was Caucasian. Current weight loss attempts were reported by 41%. Six women (5%) started using oral contraceptives between eligibility screening and completing the study procedures. Given this small number and that they had used them for ≤3 months, we did not discard their data. Table 4.1 describes participants‘ mean age, BMI, questionnaire scores, energy intake, UFC and ABP.  4.3.2 Correlation analyses As shown in Table 4.1, more negative Eating/Body Attitudes were associated with higher BMI and General Stress, and lower physical activity level. ABP was not associated with BMI, energy intakes or physical activit