Assessing Glycaemic Index Utility: from Bench to Bedside

ASSESSING GLYCAEMIC INDEX UTILITY: FROM BENCH TO BEDSIDE

Shannan Melissa Grant

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Graduate Department of Nutritional Sciences University of Toronto

Assessing Glycaemic Index Utility: From Bench to Bedside Shannan Melissa Grant Doctor of Philosophy Department of Nutritional Sciences University of Toronto 2015 Abstract

Glycaemic index (GI) categorizes carbohydrate-containing food according to postprandial glycaemic effect. The Canadian Diabetes Association supports integration of GI-education into standard care for type 2 diabetes mellitus (T2DM). Notwithstanding, most Canadian RDs identify as “non-users”; citing a need for studies examining GI-mechanism and GI-efficacy and a lack of reliable educational materials as barriers to GI-utility. These barriers highlight practice- based research opportunities spanning bench-to-bedside, such as: The examination of novel metabolic pathways and evaluation of low GI-education/ diet effectiveness using established markers of utility. The overall purpose of this dissertation was to address, experimentally, gaps in the literature, highlighted by educators as barriers to GI-utility.

METHODS: Three studies were completed. Study one examined the effect of slowing carbohydrate absorption on postprandial oxidative stress (novel GI-mechanism) in overweight/obese participants (n=18). Study two involved the development and evaluation of a questionnaire designed to evaluate an evidence-based GI-education platform for DM (n = 29), using a one-armed pre/post-intervention pre-test design. Study three (RCT; n=99) was performed to evaluate the effect of the low GI-education platform on glycaemic control in women with GDM using the questionnaire pre-tested in study two.

RESULTS: Sipping 75g dextrose solution slowly over 3.5h reduced postprandial oxidative stress compared to consuming it over 5min. Studies two and three showed that participants who received GI-education were satisfied with the education, showed a significant increase in GI- knowledge score (36%; p<0.05) and a significant decrease in dietary GI (4-6 units; p≤0.001).

Moreover, low GI-education decreased average postprandial blood glucose compared to standard care (6.02±0.03 vs. 6.10±0.02 mmol/L; p=0.041) in women with GDM.

CONCLUSIONS: The experimental work reported in dissertation provides insight into a novel

GI-mechanism, demonstrates the acceptability and efficacy of the GI education platform and provides evidence that a low GI diet improves glycaemic control in women with GDM.

Therefore, this work has successfully achieved its goal of addressing educators’ perceived barriers to GI-utility. Whether or not these efforts will increase knowledge translation to the end-user remains to be determined.

III

"It is the tension between creativity and skepticism that has produced the stunning and unexpected findings in science." ~ Carl Sagan

Dedications

I dedicate this dissertation to my family and friends; most notably my husband Michael Trumbull and parents Albert and Imelda Grant. I will never be able to thank you enough for the support you have given me during my formal education. I would never have been able to do this without your support.

I dedicate the first study of this research program to Dr Marian Naczk, former faculty of St Francis Xavier University, who did not live to see me become Dr Grant, but continues to inspire me daily.

I also dedicate this dissertation to Canadians living with or caring for someone with diabetes mellitus or cardiovascular disease. We do what we do for you.

Acknowledgements “No man is an Iland, intire of itselfe; every man

is a peece of the Continent, a part of the maine…”*

John Donne, MEDITATION XVII Devotions upon Emergent Occasions

I have been blessed to work with amazing people in various capacities; many of whom my husband (Michael) and I consider our Ontario family. The next six pages will highlight the key players in my Ontario-based academic life from 2009 to 2015. Only the tip of the iceberg, this list is not exhaustive. I could easily write a thesis just of acknowledgments. If you are not included here, please know you are acknowledged and I am grateful for the role you have played in my professional development to date.

First and foremost, I would like to thank my Supervisor, Dr Thomas Wolever. His respect of self-directed learning and my prior learning made the decision to come back to school, to complete my PhD, easy. Over the past six years, I have developed scientific knowledge and skill beyond what I could have imagined when I returned to the Department of Nutritional Sciences in 2009. As a result of my experiences in his laboratory, I have accumulated knowledge on glycaemic index that has made me a nationally recognized “expert” on the topic; with members of national and professional organizations seeking my input on education and knowledge translation (KT). I have improved a number of my professional skills because of the experiences he and our colleagues have offered me. My relationship with Tom has been and is one of the most rewarding of my life.

I had the good fortune of having an excellent committee during my graduate studies at the University of Toronto; including Dr Pauline Darling, Dr Deborah O’Connor, and Dr Robert Josse. They, with Tom, challenged me to speak with conviction about my work and to believe in myself as a scientist, Dietitian and a change-agent. I would like to thank them for the invaluable

* “No man is an island entire of itself; every man is a piece of the continent, a part of the main…” Old English version is provided above. VI

feedback they provided on the scientific and statistical methodology employed during the writing of this thesis. I would also like to thank them for their efforts to highlight the balance between independence and asking for help; key to self-directed learning. Pauline and Tom were the first mentors I had when I arrived in Toronto in 2004 to partake in the Combined Dietetic Internship and Masters of Science offered through St Michael’s Hospital Dietetic Internship and the Department of Nutritional Sciences, University of Toronto. I am thankful for their unfailing and enthusiastic support of me and my ideas and for their openness to innovation. I am very thankful for the opportunity to work with them for so many years. Debbie and Bob have also been a big part of this journey. They have known me for the majority of my time spent contributing to the Ontario research community and have provided me with unquantifiable mentorship and support. I finish this thesis acknowledging that I am honoured and privileged to have had four world-class Clinician Scientists observe and guide me for over a decade.

Although there were a number of unofficial members of my committee (including study collaborators, co-investigators, hospital-based PIs etc.), there are three members that truly stand out in the context of the work included in this dissertation; including Edward Barre, PhD, Department of Health Studies and Emergency Management, Cape Breton University (and his laboratory technician Kaz), Denice Feig, MD, MSc, FRCPC, Diabetes and Endocrinology in Pregnancy Program, Mount Sinai Hospital, and Kevin Thorpe, M.Math, Applied Health Research Centre, St. Michael's Hospital. The training and support provided by each of them, during my degree, has made this document possible.

The nature of the work conducted as part of my PhD brought me to work in a number of clinical settings and often required that I be in multiple locations at once. This ability to transcend time and space was made possible by a group of professionals I like to call my “dream team”; including Alexandra Thompson, RD, IBCLC, MSc, Rebecca Noseworthy, RD, MPH, Andrea Glenn, RD, MSc, and Maxine Seider, MSc. Each research site provided a different learning environment and set of colleagues to share experiences, skills and knowledge with. At GI Labs, I had the pleasure to work with Alexandra Jenkins, RD, PhD, Katherine Corbett, MBA, Janice Campbell, MSc, and Adish Ezatagha, MSc, and Laura Chiavaroli, MSc, PhD(c). I also had the pleasure of working with Koidula Aedna, Clinical Research Assistant, Phil Ciglen, Laboratory Technician, Slobodanka Nojkova, Immunoassay Laboratory Technician, Ana Baniska, VII

Laboratory Supervisor, and the infamous intravenous (IV) team (composed of seasoned Registered Nurses). Each day in the clinic, I worked with one member of the IV team. Although all of the RNs were amazing and have become good friends of mine, I would like to give special thanks to Celine Hanrahan. Her extreme enthusiasm, altruism, grace, passion for experience and professionalism continue to inspire me daily. At the hospital sites, we had six site Principal Investigators; Dr Joel Ray, Dr Thomas Wolever, Dr Meera Luthra, Mary Beth Neibert, RN, Dr Denice Feig and Dr. Julia Lowe. They, their health care teams and administrative staff were an endless source of scientific discussion, logistical information, support and inspiration. Alex, Rebecca, Andrea, Maxine and I felt very much a part of the health care teams at all hospital sites; St Michael’s Hospital, Mount Sinai Hospital, and Sunnybrook Health Sciences Centre, in Toronto, and St Joseph’s Healthcare Hamilton, Hamilton.

Although not directly involved in my research, I would like to acknowledge the following members of the Department of Nutritional Sciences, University of Toronto who certainly played a significant role in my PhD experience: The administrative staff, including Louisa Kung, Emelia D’Souza, Vijay Chetty, Lucile Lo and (most recently) Slavica Jovanovic. The role you play in student life in the department is noteworthy and appreciated. Thank you for your support, guidance, patience and friendship. Dr Anthony Hanley (or Tony) must be acknowledged for his ongoing mentorship and friendship – the work we have done together impacts how I approach every project I am asked to contribute to. I would also like to acknowledge Dr Valerie Tarasuk and Dr Carol Greenwood for their involvement in my defense preparation and for the inspiration they have provided since before my time in the Department of Nutritional Sciences. They and the other women that make up the Department of Nutritional Sciences trail blaze daily; making each day better for women in academia. Dr Vlad Vuksan and Elena Jovanovski, MSc, PhD(c) were invaluable resources during the development of our methodology to assess arterial stiffness during the first study included in this dissertation; a collaboration that has extended beyond the walls of the University of Toronto and St Michael’s Hospital to include my current department at Mount Saint Vincent University. Dr Ann Fox and Dr Debbie Gurfinkel have been a great source of inspiration during the development and evaluation of the education platform discussed in this dissertation, due to their passion for education of nutrition professionals. Dr David Jenkins must also be acknowledged. In addition to playing a key role in the development of the glycaemic index, David’s work ethic, passion for VIII

nutritional science, encouragement, and inquisitive nature were a regular source of creativity and motivation for me during my degree.

The Department of Nutritional Sciences student body must also be acknowledged for their support and feedback. Although many students have played a role in my professional growth, I would like to highlight five current members of the department whom I consider family: Julie Ennis, PhD, Laura Chiavaroli, MSc, PhD(c), Matt Parrott, PhD, Alexander Schwartz, MSc, PhD(c), and Pedro Huot, PhD(c). Former students of the department who are dear to me and were a great source of encouragement during the past six years include: Karen Smith, MSc, PhD, Nishta Saxena, RD, MSc, Andrea Josse Obar, PhD, Julia Wong, RD, PhD, Maria Fernanda Nunez, MSc, Arash Mirrahimi, MSc, MD(c), Russell de Souza, RD, PhD, Brenda Hartman, RD, PhD, and Sohana Shafique, PhD. I would also like to acknowledge the brilliant members of the Wolever Laboratory past, present and future. Specifically, I would like to acknowledge the laboratory members who have become my sisters and brothers; Melissa Kwong, Halah Zawawi, Asmaa Alraefaei, Sari Rosenbloom, Judlyn Fernandes, Kervan Rivera Rufner, Ahmed Aldughpassi and Evan Lewis. I will miss our annual Wolever Laboratory summer retreats at Margaritas on Baldwin Street, Toronto.

I would also like to acknowledge the Department of Nutritional Sciences Alumni Association Mentorship Program and those who work to maintain it. A successful mentoring relationship is a peer to peer relationship in which a more experienced person helps someone with less experience to grow and develop. This confidential exchange should provide a safe place where both people can discuss professional and personal items relevant to objectives they set together. Mentoring is one of the many ways in which a profession maintains its vitality and standards of excellence. It encourages new nutrition professionals to learn from those who are established and challenges those established to stay current. Everyone can obtain benefit from a mentoring experience - the mentee, the mentor and, in turn, our field. The two mentors I had as part of this program have been a great source of guidance as I transitioned from student to my new career; Dr Nick Bellissimo, Ryerson University, and Dr Debbie Gurfinkel, University of Toronto. I would also like to acknowledge my mentee under this program, Laura Chiavaroli. I am a firm believer that each exchange we have is an opportunity to learn and grow personally and professionally. IX

The majority of my time during my PhD was spent in Dr Thomas Wolever’s office within the Risk Factor Modification Centre (RFMC), St. Michael Hospital. St. Michael’s Hospital played a key role in making me the clinician scientist I am today. The training I have obtained from this institution has deeply influenced the research questions I ask and has provided me with the skills to answer them. Under St. Michael’s Hospital, the Keenan Research Centre for Biomedical Science and the Li Ka Shing Knowledge Institute and RFMC, Diabetes in Pregnancy Clinic and Diabetes Comprehensive Care Program are the groups who provided the majority of this support. The following clinicians must be acknowledged for their role (in addition to Dr Pauline Darling [above]): Melinda Glassford, RD, Sherri Storm, RD, Kristi Jane Hurst, RD, Jessica Omand, RD, PhD(c), Katie Southgate, RD, MSc, Julianne Cavanagh, RD, Emily Elliott, RD, Melissa Sobie, RD, Gurita Bhatti, RD, and Lilli Mauer, MSc, Medical Student. I would also like to extend thanks to Dr Philip Connelly, Graham Maguire and Maureen Lee of the J. Alick Little Lipid Research Laboratory, St Michael’s Hospital.

I would like to acknowledge the Natural Sciences and Engineering Research Council of Canada (NSERC) for funding me during my undergraduate, Masters of Science and Doctoral Degree. Without this funding, I would not be where I am today. One of the highlights of my life was receiving the Alexander Graham Bell Canada Graduate Scholarship, as Bell was a significant source of inspiration to me as a child. I would also like to thank the Canadian Institutes for Health Research (CIHR), the Canadian Diabetes Association (CDA) and the Canadian Foundation for Dietetic Research (CFDR) for funding the work completed during my Doctoral degree.

I would like to acknowledge my “KT mentors”. Thank you to Sharon Straus, HBSc, MD, MSc, for reviewing the methodology and first draft of the GIQ with Dr Pauline Darling, Dr Thomas Wolever and I. Your feedback and written resources were invaluable to the start-up of study 2 and the rooting of my interest and education in KT. I would like to extend thanks to Kelly Warmington, MEd PMP, Melanie Barwick, PhD, CPsych, and Donna Lockett, PhD, SE, CTT for their mentorship and training. My participation in the Scientist Knowledge Translation Training™, at The Hospital for Sick Children was a noteworthy milestone in my graduate work. Moreover, your ongoing support, workshops and brilliant job aids (The KT Game) continue to influence my teaching and research. I would also like to thank Sophie Desroches, RD, PhD, X

School of Nutrition, Faculty of Agriculture and Food Sciences, Université Laval and Alison Duncan, RD, PhD, Department of Human Health and Nutritional Sciences at the University of Guelph. Sophie kindly served as the external reviewer on my final PhD defense. Her feedback and kind nature made my defense both a challenging and positive experience. I hope to collaborate with her in the future. Alison and I have worked together in various capacities, but the majority of our communication has been related to our work on the Canadian Nutrition Society and mutual interest in mentorship and KT. Her work ethic, kind nature, and balanced approach to our demanding profession continue to serve as a source of inspiration for me. I would also like to thank Wendy Kirkpatrick and The Kirkpatrick Partners for their training and ongoing mentorship; truly an amazing group of people to work with. I would like to acknowledge the Dietitians of Canada and CDA for their ongoing role in translating and disseminating the lessons learned during my doctoral work; a pleasure to be part of these important activities.

Last and certainly not least, I would like to acknowledge all of the people who have participated in our studies and programming and the community of Sandy Lake First Nation. Sandy Lake First Nation will always be my home away from home. My work in this community confirmed my passion for community-based research and education, my appreciation for Indigenous Knowledge and Worldview and my desire to devote my life to training and mentoring students. Meegwetch.

Published Material Peer Reviewed Articles:

1. Components of chapter 1 and 2 have been previously published: Grant, S. and Wolever, T. (2011) Perceived Barriers to Application of Glycaemic Index: Valid Concerns or Lost in Translation? Nutrients; 3: 330-340. Reprinted, revised and updated with permission of Nutrients Editor (Open Source).

Peer Reviewed Abstracts/ Presentations:

2. Grant. S., Josse, R., Barre, E, Wolever, T. (2014) The effect of continuous sipping of a glucose solution on markers of oxidation in men and women. Canadian Nutrition Society Annual Meeting 2014; St. John's, Newfoundland, Canada (Protocol and Data Presented). ABSTRACT: Applied Physiology, Nutrition, and Metabolism, 2014, 39(5): 605-642 (Published). Reprinted, revised and updated.

Peer Reviewed Client Education Materials

3. Reviewer (S Grant); Dietitians of Canada PEN (Practice-based Evidence in Nutrition) Dietitians of Canada. Diabetes and GI Knowledge Pathway. In: PEN - Practice-based Evidence in Nutrition®. 2010. Available from: http://www.pennutrition.com. Access only by subscription (Published).

4. Grant, S., Noseworthy, R., Eisenbraun, C., Wolever, T. Integrating Glycaemic Index into Practice. Learning on Demand, Dietitians of Canada (In Press).

5. Grant, S., Noseworthy, R., Glenn, A., Chiavaroli, L., Wolever, T. The Glycaemic Index Food Guide and Backgrounder. Canadian Diabetes Association (In Press).

* Client Education Materials #5 are currently under review by Canadian Diabetes Associations Health Education Lifestyle Management Committee.

XII

Table of Contents Chapter Chapter/ Section Page Number Title Number 1.0. INTRODUCTION 1 2.0. LITERATURE REVIEW, DISSERTATION RATIONALE, 6 GOAL AND OBJECTIVES 2.1.0. AN OVERVIEW OF CANADIAN GUIDELINES FOR 7 SCREENING AND TREATMENT OF TYPE 2 DIABETES MELLITUS AND GESTATIONAL DIABETES MELLITUS 2.1.1. Type 2 Diabetes Mellitus: Definitions, Statistics and Guidelines for 7 Screening 2.1.2. Gestational Diabetes Mellitus: Definitions, Statistics, and Guidelines 9 for Screening 2.1.3. Standard Care for Type 2 Diabetes Mellitus and Gestational Diabetes 12 Mellitus: A Focus on Medical Nutrition Therapy 2.2.0 GLYCAEMIC INDEX: AN OVERVIEW 16 2.2.1. Glycaemic Index: Origins in a Desire to Improve Diabetes Medical 16 Nutrition Therapy 2.2.2. Low Glycaemic Index and Gestational Diabetes Mellitus: The State 20 of the Evidence 2.2.3.0. Perceived Barriers to Glycaemic Index Utility: Revealing Pathways 25 to Translation 2.2.3.1. National Institutes for Health Research Translational Roadmap: A 25 Conceptual Framework for Maneuvering the Bench to Bedside Continuum 2.2.3.2. The Role of the Dietitian in Nutritional Science Knowledge 28 Translation 2.2.3.3.0 Perceived Barriers to Glycaemic Index Utility 30 2.2.3.3.1. Perceived Barrier: Glycaemic Index is Too Difficult for Clients to 31 Understand 2.2.3.3.2. Perceived Barrier: Glycaemic Index is too Difficult for Clients to 35 Apply 2.2.3.3.3. Perceived Barrier: Lack of Convincing Evidence 37 2.2.4.0. The Evolution of our Understanding of Glycaemic Index Mechanism: 38 The Importance of Coming Back to the Bench 2.2.4.1. Diabetes Mellitus, Cardiovascular Disease and the Oxidation 38 Hypothesis 2.2.4.2. Antioxidant Capacity and Oxidative Stress Quantification 41 2.2.4.3. Non-invasive Measures of Cardiovascular Disease Risk 43 2.2.4.4. The Effect of Low Glycaemic Index on Postprandial Oxidative 48 Response: A Novel Glycaemic Index Mechanism 2.2.5.0. The Role of Low Glycaemic Index Education Evaluation in 53 Assessing Glycaemic Index Utility in the Context of Clinical Trials 2.2.5.1. Using the Randomized Control Trial to Evaluate Health Education 53 Interventions 2.2.5.2. The Kirkpatrick Model and Glycaemic Index Education Evaluation 56 XIII

Table of Contents Continued Chapter Chapter/ Section Page Number Title Number 2.3.0. LITERATURE REVIEW SUMMARY AND DISSERTATION 62 CONCEPTUAL FRAMEWORK 2.4.0. STUDY-SPECIFIC RESEARCH HYPOTHESES AND PRIMARY 68 OBJECTIVES: AN OVERVIEW 3.0. STUDY #1 THE EFFECT OF CONTINUOUS SIPPING OF A 70 DEXTROSE SOLUTION ON MARKERS OF OXIDATION IN MEN AND WOMEN 3.1. ABSTRACT 71 3.2.0. INTRODUCTION 72 3.2.1. Study Rationale 72 3.2.2. Hypotheses and Primary Objective 72 3.3.0. MATERIALS AND METHODS 73 3.3.1. Study Design 73 3.3.2.0. Sample 73 3.3.2.1. Sample Size Calculation 75 3.3.3. Study Outcomes 75 3.3.4. Study Treatments and Randomization 76 3.3.5.0. Biochemical Outcomes 78 3.3.5.1. Blood Sample Collection and Processing 79 3.3.5.2. Biochemical Analysis 81 3.3.6. Cardiovascular Hemodynamics Outcomes (or “Vitals”) 83 3.3.7. Anthropometric Data 86 3.3.8. Hard Copy Data Collection Tools 86 3.3.9.0. Data Analysis 87 3.3.9.1. Software 87 3.3.9.2. Analysis 87 3.4.0. RESULTS 89 3.4.1. Sample Characteristics 89 3.4.2.0. Biochemical Outcomes 90 3.4.2.1. Plasma Glucose 90 3.4.2.2. Plasma Insulin 91 3.4.2.3. Serum Free Fatty acids 93 3.4.2.4. Serum C - reactive Protein 93 3.4.2.5. Vitamin C 94 3.4.2.6. TRAP (Primary Outcome) 95 3.4.2.7. LDL Oxidation and Baseline Conjugated Dienes 95 3.4.2.8. Lipid Profile 98 3.4.2.9. Vitals 99 3.4.2.10. Protocol Adherence 101 3.5. CONCLUSIONS AND DISCUSSION 102

XIV

Table of Contents Continued

Chapter Chapter/ Section Page Number Title Number 4.0. STUDY#2 EVALUATION OF GLYCAEMIC INDEX 109 EDUCATION IN PEOPLE LIVING WITH TYPE 2 DIABETES MELLITUS: PARTICIPANT SATISFACTION, KNOWLEDGE UPTAKE AND APPLICATION 4.1. ABSTRACT 110 4.2.0. INTRODUCTION 111 4.2.1. Study Rational 111 4.2.2. Hypotheses and Primary Objective 112 4.3.0. MATERIALS AND METHODS 112 4.3.1. Design 112 4.3.2.0. Sample 112 4.3.2.1. Sample Size Calculation 113 4.3.3. Outcomes 113 4.3.4.0. The Study Intervention: Low Glycaemic Index Education 114 4.3.4.1. The Low Glycaemic Index Food Substitution List 115 4.3.4.2. The Low Glycaemic Index PowerPoint Presentation 116 4.3.4.3. Low Glycaemic Index Recipe Book 117 4.3.4.4. Marrying Standard Care and Low Glycaemic Index Education Using 117 Hands-on Activity 4.3.5. Three Day Diet Record: Dietary Intake Data Collection and Analysis 118 Procedures 4.3.6. The Glycaemic Index Questionnaire (GIQ©) 119 4.3.7. The Glycaemic Index Questionnaire Data Entry Handbook 120 4.3.8. Retrospective Medical Chart Review 121 4.3.9. Statistical Analysis 121 4.4.0. RESULTS 123 4.4.1. Sample Characteristics 123 4.4.2.0. Participant Satisfaction (Kirkpatrick Model Level 1 - Reactions) 126 4.4.2.1. Responses to Close-end Questions from GIQ Section 1: How Did You 126 Like the GI Class? 4.4.2.2. Responses to Open-end Questions from GIQ Section 1: How Did You 127 Like the GI Class? 4.4.3. GI Knowledge Score (Kirkpatrick Model Level 2 – Learning) 130 4.4.4. Three Day Diet Record Data (Kirkpatrick Level 3 – Transfer) 131 4.4.5.0. GIQ Section 4a: Is Your Low GI Diet Working For You? 132 (Application and Acceptability) 4.4.5.1. Responses to Close-end Questions from GIQ Section 4a 132 4.4.5.2. Responses to the Open-end Question from GIQ Section 4a 132 4.5 CONCLUSIONS AND DISCUSSION 136

Table of Contents Continued

Chapter Chapter/ Section Page Number Title Number 5.0. STUDY # 3 THE EFFECT OF A LOW GLYCAEMIC INDEX 142 DIET ON MATERNAL AND NEONATAL MARKERS OF GLYCAEMIC CONTROL AND POSTPARTUM DIABETES RISK 5.1. ABSTRACT 143 5.2.0. INTRODUCTION 144 5.2.1. Study Rationale 144 5.2.2. Study Hypotheses 145 5.3.0. MATERIALS AND METHODS 145 5.3.1. Study Design and Randomization 145 5.3.2.0. Sample 145 5.3.2.1. Eligibility Criteria 145 5.3.2.2. Power Analysis/ Sample Calculation 147 5.3.3. Study Outcomes 148 5.3.4.0. The Study Intervention: Low Glycaemic Index Education 149 5.3.4.1. Standard Care Education 149 5.3.4.2. Low Glycaemic Index Education 150 5.3.4.3. The Plate Game 152 5.3.5.0. Data Collection 152 5.3.5.1. Self-Monitored Blood Glucose 152 5.3.5.2. Three Day Diet Record: Dietary Intake Data Collection and Analysis 152 Procedures 5.3.5.3. The Glycaemic Index Questionnaire (GIQ©) 153 5.3.7. Statistical Analysis Procedures 154 5.4.0. RESULTS 157 5.4.1. Sample Characteristics 157 5.4.2. Participant Satisfaction (Kirkpatrick Model Level 1 - Reactions) 161 5.4.3. GI Knowledge Score (Kirkpatrick Model Level 2 – Learning) 162 5.4.4. Three Day Diet Record Data (Kirkpatrick Level 3 – Transfer) 163 5.4.5.0. Self-Monitored Blood Glucose (Kirkpatrick Level 4 – Results) 166 5.4.5.1. Average Postprandial Self-Monitored Blood Glucose 166 5.4.5.2. Percent Postprandial Self-monitored Blood Glucose Values Within 167 Range 5.5. CONCLUSIONS AND DISCUSSION 169 6.0. DISSERTATION SUMMARY AND DISCUSSION 175 6.1. LIMITATIONS 180 6.2. OVERALL CONCLUSIONS 181 6.3. FUTURE DIRECTIONS 183 7.0. REFERENCE LIST 185 8.0. APPENDIX – See page “XX” for full table of contents 217

XVI

List of Tables Table Table title Page Number Number Chapter 2 2.1 Select risk factors for type 2 diabetes mellitus as per the Canadian 8 Diabetes Association Clinical Practice Guidelines 2013 2.2 Select risk factors for gestational diabetes mellitus as per the 11 Canadian Diabetes Association Clinical Practice Guidelines 2013 2.3 Canadian Guidelines for Body Weight Classification in Adults 14 Using Body Mass Index 2.4 Current recommendations for total and rate of weight gain during 14 pregnancy, by pre-pregnancy body mass index 2.5. Acceptable Macronutrient Range; Percent of Energy (Dietary 14 Reference Intakes) Chapter 3 3.1 Study Outcome Collection Schedule at Each Study Visit 76 3.2 Standard Lunch Recipe and Ingredient Mass 77 3.3 Energy and nutrient composition of standard lunch meal 78 3.4 Blood Collection Tube Specifications, Study Outcomes and 79 Destination 3.5. Blood Sample Collection Timeline by Biochemical Outcome 80 3.6. Select Sample Characteristics at Screening 90 3.7 Main effect of supplement administration on total cholesterol and 98 low density lipoprotein 3.8 Main effect of treatment administration on incremental vitals 99 Chapter 4 4.1 Glycaemic Index Questionnaire (GIQ©) Administration Schedule 122 4.2 Glycaemic Index Questionnaire (GIQ©) Qualitative Questions 122 4.3 Glycaemic Index Questionnaire (GIQ©) Qualitative Data Table 122 (Example) 4.4 Glycaemic Index Questionnaire Section 2 Demographic 124 Information: Response Summary to “Which ethnic group do you identify with?” 4.5 Glycaemic Index Questionnaire Section 2 Demographic 125 Information: Questions and responses for 5 to 13. 4.6 Glycaemic Index Questionnaire Section 1: GI Education 126 Satisfaction, statements and responses for question 4. 4.7 Participant Responses Organized under Theme “Standard Care”, 127 Sub-theme Serving Size” 4.8 Participant Responses Organized under Theme “Glycaemic Index” 128 4.9 Participant Responses Falling under Theme “Feedback from 129 Participants” 4.10 Participant Responses Organized under Theme “Standard Care”, 129 Sub-theme Serving Size”

XVII

4.11 Results of three day diet record data analysis 131 4.12 Macronutrients as a (mean) percentage of total daily caloric intake 131 4.13 Participant Responses Organized under Theme “Glycaemic Index” 132 Sub-theme “Glycaemic Index: Education/Comprehension” 4.14 Participant Responses Organized under Theme “Glycaemic Index” 133 Sub-theme “Food Selection” 4.15 Participant Responses Organized under Theme “Glycaemic Index” 133 Sub-theme “GI Self-efficacy” 4.16 Participant Responses Organized under Theme “Feedback from 134 Participants” 4.17 Glycaemic Index Questionnaire Section 4a Is Your Low GI Diet 135 Working For You?: Close-end Questions (Q) and Responses Chapter 5 5.1 Recruitment targets for each hospital site 147 5.2 Study Data Collection Timeline 156 5.3 Glycaemic Index Questionnaire (GIQ©) Administration Timeline 156 5.4 Recruitment totals and targets by Hospital site 157 5.5 Glycaemic Index Questionnaire Section 2 Demographic 160 Information: Response Summary to “Which ethnic group do you identify with?” 5.6 Previous Medical Nutrition Therapy Reported by Sample; 160 Questions and Answers from Glycaemic Index Questionnaire 5.7 Glycaemic Index Questionnaire Section 1: GI Education 161 Satisfaction, statements and responses for question 4. 5.8 Results of three day diet record data analysis 164 5.9 Macronutrients as a percentage of total daily caloric intake 165 5.10 Sample size (counts) attrition in the context of self-monitored 166 blood glucose 5.11 Counts (percent) of postprandial self-monitored blood glucose 168 values within range by group assignment and visit

XVIII

List of Figures Figure Figure title Page Number Number Chapter 2 2.1 Preferred approach for the screening and diagnosis of gestational 10 diabetes mellitus 2.2 Alternative approach for the screening and diagnosis of 11 gestational diabetes mellitus 2.3 National Institutes of Health Research Roadmap 27 2.4 Pulse Wave: Variables that Comprise Augmentation Index 45 2.5 Kirkpatrick Method – The Four Levels 56 2.6 The New World Kirkpatrick Model 58 2.7. Conceptual Framework for Dissertation 67 Chapter 3 3.1 How to Use the Pulse Wave Sensor Unit 85 3.2 Raw plasma glucose versus time 92 3.3 Incremental plasma insulin versus time 92 3.4 Incremental serum free fatty acids versus time 93 3.5 Incremental plasma vitamin C versus time 94 3.6 Incremental plasma TRAP versus time 96 3.7 Incremental LDLox versus time 97 3.8 Incremental conjugated diene concentration versus time 97 3.9 Incremental pulse rate versus time 100 Chapter 4 4.1 The Plate Method 118 4.2 Glycaemic Index Questionnaire Section 2 Demographic 124 Information: Response Summary to “Which ethnic group do you identify with?” (Percent) 4.3 Total Knowledge Score at each administration of Glycaemic 130 Index Questionnaire Section 3: GI Knowledge (by visit) Chapter 5 5.1 Study Outcomes According to Kirkpatrick Method (Four Levels) 148 5.2 Profile of study sample from recruitment to study completion 158 5.3 Glycaemic Index Questionnaire Section 2 Demographic 159 Information: Response Summary to “Which ethnic group do you identify with?” 5.4 Total Knowledge Score at each administration of Glycaemic 162 Index Questionnaire Section 3: GI Knowledge (by visit) 5.5 Mean postprandial self-monitored blood glucose two hours after 167 breakfast, lunch and dinner versus study visit

XIX

List of Appendices (Chapter 8.0. Appendix) Appendix Appendix Title Page Number Number Chapter 2 Appendix 2.1. Glycaemic Index Value Search and Assignment Procedure 218 Chapter 3 Appendix 3.1. Screening Questionnaire (Baseline Lifestyle) 219 3.2. Total Peroxyl Radical Trapping Antioxidant Potential Assay 225 3.3 Baseline Conjugated Diene Quantification (Assay) 230 3.4. Procedure for LDL protein Quantification (Assay) 233 3.5. Methodology: Wako NEFA-HR(2) Microtiter Procedure 236 3.6. Lipid Analysis Procedure: Total Cholesterol, HDL 238 Cholesterol and Triglyceride 3.7. 24 hour Recall 240 3.8. Symptoms Questionnaire 241 Chapter 4 Appendix 4.1. Presentations (Samples): Formative Feedback Initiatives 242 4.2. Three Day Diet Record with Instructions 252 4.3. Glycaemic Index Questionnaire Section 3 259 (with answer key): Glycaemic Index Knowledge 4.4. Glycaemic Index Questionnaire Section 4a: Glycaemic 263 Index Application and Acceptability 4.5. Glycaemic Index Questionnaire Qualitative Data Codebook 267 Chapter 5 Appendix 5.1. Complete list of study outcomes for The effect of a low 268 Glycaemic Index Diet on Maternal and Neonatal Markers of Glycaemic Control and Postpartum Diabetes Risk 5.2. Low Glycaemic Index Food Substitution List 269 5.3. Low Glycaemic Index PowerPoint Presentation 277 5.4. Low Glycaemic Index Recipe Book (Title Page and Table of 281 Contents) 5.5. Glycaemic Index Questionnaire Section 1: Participant 283 Satisfaction 5.6. Glycaemic Index Questionnaire Section 2: Getting to Know 287 You (Demographic Information) XX

Abbreviation Key Acronym/ Abbreviation Expanded/ Full Word A ADA American Dietetic Association AI/ AIx Augmentation Index AOGI Short title for study one ALT Alanine transaminase AMDR Acceptable Macronutrient Distribution Range ApoB-100 ApolipoproteinB-100 AST Aspartate transaminase AT Applanation tonometry ATS Applanation tonometry sensor B BMI Body Mass Index BP Blood Pressure C CCD The Canadian Trial of Carbohydrate and Diabetes CDA Canadian Diabetes Association CD Conjugated diene(s) CHD Coronary heart disease CHIR Canadian Institutes for Health Research CI Confidence Interval CPG Clinical Practice Guidelines Cr Creatinine CRP C Reactive Protein CV Coefficient of variance CVD Cardiovascular Disease CFG Canadian Food Guide D DBP Diastolic blood pressure

XXI

DC Dietitians of Canada DCCP Diabetes Comprehensive Care Program, St Michael’s Hospital DCFH-DA Dichloro-dihydro-fluorescein diacetate DM Diabetes Mellitus DNA Deoxyribonucleic acid DRIs Dietary Reference Intakes E ECG Electrocardiogram ED Endothelial dysfunction EDTA Edetic acid eNOS Endothelial nitric oxide synthase ERF Endothelium relaxing factor ESHA Elizabeth Stewart Hands and Associates; The Food Processor Diet Analysis and Fitness Software F FFA Free fatty acids (also see NEFA) FMD Flow mediated dilatation FPG Fasting plasma glucose

F2IP F2α isoprostanes G GDM Gestational diabetes mellitus GI Glycaemic index/ Glycemic index GIEES Short title for study two GI in GDM Study Short title for study three GIQ© Glycaemic Index Questionnaire GLM General linear model H

HbA1C Hemoglobin A1C HICL Hospital in Commons Laboratory HDL High density lipoprotein

XXII

I iAUC Incremental area under the curve ICDEP The Integrated Competencies for Dietetic Education and Practice IGTP Impaired Glucose Tolerance of Pregnancy IOM Institutes of Medicine ISO International Organization for Standardization IVGTT Intravenous glucose tolerance test K KM Kirkpatrick Model (of education evaluation) KPs Kirkpatrick Partners KT Knowledge Translation L LDL Low density lipoprotein LDLox Oxidized low density lipoprotein L1, L2 Lab 1 (bench), Lab 2 (bedside) M MD Medical Doctor ML Maximum Likelihood MS Microsoft N NIH National Institutes of Health NEFA Non-esterified fatty acids (see FFA) NO Nitric Oxide NWKM New World Kirkpatrick Method O OGTT Oral glucose tolerance test OPDQ Professional des Dietetistes du Quebec OR Odds ratio ORAC Oxygen radical absorbance capacity (assay) P

XXIII

PBS Phosphate buffer solution PEN Practice-based Evidence in Nutrition® PG/FPG Plasma glucose/ Fasting plasma glucose (only used in figures and tables) PP Pulse pressure

PPAI Pulse pressure measured with ATS (see “A”) PR Pulse rate

PRAI Pulse rate measured with ATS (see “A”)

PRBP Please rate measured with blood pressure cuff PUFA Polyunsaturated fatty acid(s) PWV Pulse wave velocity P1 Peak of ejected wave P2 Peak of reflected wave R RCT Randomized control trial RDA Recommended dietary allowance RD Registered Dietitian REB Research Ethics Board RN Registered Nurse ROS Reactive oxygen species RR Relative Risk S SAM Suitability Assessment of Materials Instrument SBP Systolic blood pressure SD Standard deviation SEM Standard error mean T TChol Total cholesterol TG Triglyceride TRAP Total peroxyl radical trapping antioxidant potential (assay/ concentration) TSH Thyroid stimulating hormone

XXIV

T1DM Type 1 diabetes mellitus T2DM Type 2 diabetes mellitus T1, T2, T3 “Translational blocks” (see page 26) U UHP Ultra high purity water US United States of America V V Visit Vit C Vitamin C

XXV

CHAPTER 1.0. INTRODUCTION

1.0. INTRODUCTION

Currently more than 10 million Canadians are living with diabetes mellitus (DM) or pre- diabetes (1). The majority (65% to 80%) of people with DM will die from cardiovascular disease (CVD) and people living with DM are at higher risk of developing CVD than people living without DM (2-4). Moreover, people living with DM are at higher risk of CVD at an earlier age, with women being at higher risk than men (2,5,6). CVD is an umbrella term for a number of conditions adversely affecting the anatomy and physiology of the circulatory system. Coronary artery disease is one example of CVD and develops when a blend of lipid, cholesterol, calcium and scar tissue (plaque) build up in the arteries that supply blood to the heart (7-10). This buildup of atherosclerotic plaque narrows the arteries, impeding passage of blood (and oxygen) to the heart and brain, causing heart attack or stroke (7,8,11). This is relevant to Canadians because every seven minutes in Canada, someone dies from heart disease or stroke (11). Moreover, heart disease and stroke cost Canadians more than $20.9 billion every year in physician services, hospital costs, lost wages and decreased productivity (12).

Two types of DM are discussed in this dissertation; type 2 DM (T2DM) and gestational DM (GDM). T2DM is diagnosed when one experiences a clinically quantifiable decrease in insulin sensitivity or a reduction in insulin production (1,13). GDM has similar risk factors and symptoms to T2DM, but is often defined as transient DM that occurs during pregnancy (14). According to statistics released by the Canadian Diabetes Association (CDA), between three and 20 percent of pregnant women develop GDM, depending on their risk factors (e.g. family history, ethnicity) (13,15-18). GDM has also been highlighted as a model for studying the early events in the natural history of T2DM (prevention). Indeed, 20% of women with prior GDM will develop T2DM (19-22). This acknowledged, it is important to recognize that women living with GDM require a treatment plan that differs from that of people with T2DM (client-focused treatment). Medical nutrition therapy is the cornerstone of prevention and treatment of DM and is examined in this dissertation (14,23-26).

The term “glycaemic index” (GI) first appeared in the literature in the early 1980s. This concept, and associated methodology were proposed as a method to categorize carbohydrate foods according to glycaemic effect during the postprandial or fed period (27-29). Carbohydrate 3 containing foods are typically categorized according low, medium and high GI (29-31). Evidence suggests a diet including low GI foods may be beneficial for the prevention or treatment of a number of chronic conditions; including type 1 DM, T2DM, CVD, and cancer (13,29,32-44). The majority of published GI utility data have been provided by individuals at risk of or living with DM (27,29,36-38,45-61).

The CDA Clinical Practice Guidelines (CPG) provide DM educators with an overview of the current evidence for medical nutrition therapy in prevention and treatment of DM and provides recommendations for practice. The CDA CPG (2013) states that

“Replacing high glycemic index carbohydrates with low glycemic index carbohydrates in mixed meals has a clinically significant benefit for glycemic control in people with type 1 and type 2 diabetes.”

Despite this, data suggests that 61% (n = 642) of Canadian RDs, working with clients diagnosed with DM, do not use GI in practice (62). Also, the CDA CPG (2013) does not take a clear stance on utility of GI education in GDM treatment (13). These findings inspired the following seven questions:

(1.) What are clinicians/ educators perceived barriers to GI utility in DM treatment? (2.) Are the clinicians perceived barriers to GI utility supported by evidence? (3.) If supported by evidence, how can these barriers be addressed? (4.) If not, how can evidence translation be improved? (5.) Is there sufficient data in support of GI utility in GDM treatment? (6.) If yes, how can we integrate GI education into GDM medical nutrition therapy? (7.) If no, how can we supplement the available evidence?

A comprehensive critical review of the literature was completed to obtain answers to these questions and to develop dissertation hypotheses and objectives. During the study development period, part of this critical review was published. This review concluded that many of the published perceived barriers to GI were not supported by existing evidence (63). The authors used an evidence-based approach to address each perceived barrier that was based on 4 misinformation or was lacking scientific merit. Four perceived barriers were identified as having scientific merit and requiring examination. The four perceived barriers identified were:

(1.) Additional data are needed on GI mechanism and utility before professionals can make an informed choice of whether or not to use GI in practice (2.) There is a lack of reliable GI education tools (3.) The GI concept is too difficult for clients to understand (4.) The GI concept is too difficult for clients to incorporate into their diet (to apply) (63)

The overall goal of this dissertation is to address these perceived barriers. That is, to obtain more information about GI mechanism and GI utility, to determine if GI was too difficult for people living with DM to understand and incorporate into their diets and to develop and comprehensively evaluate novel GI education materials and strategies. Critical review of the evidence on GI utility in women with GDM clearly highlighted that this area of research requires additional contribution before GI can be considered for integration into standard medical nutrition therapy for GDM (24,40,64-72).

There is growing interest in the relationship between the effects of reducing postprandial blood glucose on oxidative stress. This relationship has been proposed as a novel mechanism by which low GI foods deliver health benefits (73-78). Despite this, published studies suggesting that low GI foods reduce oxidative stress are confounded by differences in dietary antioxidant and nutrient intake and, hence, do not prove that reducing post-prandial glucose per se is responsible for this effect (74,78). Therefore, the sipping vs. bolus paradigm, originally designed to determine the effect of slowing carbohydrate absorption on glucose and insulin responses, was used to determine the effect of slowing carbohydrate absorption on postprandial oxidative stress (79). An aim of this work was to provide more insight into the mechanism by which GI affects biochemical outcomes in people who have DM risk factors; while excluding common dietary confounders (addressing perceived barrier 1).

An evidence-based GI education platform was developed to address clinicians call for more reliable GI education materials. A questionnaire (Glycaemic Index Questionnaire/ GIQ©) was also developed to facilitate effective evaluation of the GI education platform using the 5

Kirkpatrick Model (KM) (addressing perceived barrier 2). The KM is an evidence-based approach for training and education evaluation that has been adapted for use in health care settings (80-88). It includes the following four levels:

Level 1: Reactions (e.g. client satisfaction) Level 2: Learning (e.g. knowledge uptake, knowledge score) Level 3: Transfer (e.g. behaviour change, change in dietary GI) Level 4: Results (e.g. glycemic control, weight loss, improved β-cell function).

Client education materials and the GIQ were face and content validated and pre-tested in a sample of people living with T2DM. They were then implemented as part of a randomized controlled trial (RCT) designed to determine the effect of a low GI diet on glycaemic control in women with GDM (addressing perceived barrier 1-2). As part of the pre-test and RCT, clients’ understanding (KM Level 2; perceived barrier 3) and ability to apply (KM Level 3; perceived barrier 4) GI was also measured. Details of the development, implementation and evaluation of these education tools are discussed within this dissertation. RCT outcomes (including the primary outcome) related to education evaluation (in the context of KM) were included in this dissertation.

In addition to providing novel data on GI mechanism and utility, this dissertation and the work that has been done to produce it is part of a larger initiative aimed at improving and increasing translation of GI utility findings to clinicians. By involving stakeholders and the end-users (the educators and patients/ clients) in the research process and addressing their questions, we aim to (ultimately) assist educators and clients in making an informed choice of whether or not to use GI in DM medical nutrition therapy (component of integrative knowledge translation) (63,85,89-97). That is, we evaluated evidence-based GI education approaches based on client perspective, comprehension, behaviour change, and effect on clinically relevant outcomes. This work also aims to increase the current state of knowledge regarding GI utility in women living with GDM by building on a pilot study conducted between 2005 and 2007 (40).

CHAPTER 2.0. LITERATURE REVIEW, DISSERTATION RATIONALE, GOAL AND OBJECTIVES

Components of this chapter (2) have been previously published: Grant, S.M.; Wolever, T.M.S. Perceived Barriers to Application of Glycaemic Index: Valid Concerns or Lost in Translation? Nutrients 2011, 3, 330-340. Reprinted, revised and updated with permission of Nutrients Editor (Open Source).

6 7

CHAPTER 2.0. LITERATURE REVIEW, DISSERTATION RATIONALE, GOAL AND OBJECTIVES

2.1.0. AN OVERVIEW OF CANADIAN GUIDELINES FOR SCREENING AND TREATMENT OF TYPE 2 DIABETES MELLITUS AND GESTATIONAL DIABETES MELLITUS

2.1.1. Type 2 Diabetes Mellitus: Definitions, Statistics and Guidelines for Screening

DM is a chronic disease characterized by hyperglycaemia (high blood glucose). People living with DM can experience reduced or undetectable insulin secretion and/ or insulin resistance. Although glucose is a main source of fuel for the human body, sufficient and effective insulin is needed to ensure that it is metabolized (13). Chronic hyperglycaemia damages blood vessels, organs and nerves and has been associated with a number of complications; including CVD, kidney failure and neuropathy. There are three main types of DM: (1.) Type 1 Diabetes Mellitus (T1DM), (2.) T2DM and (3.) GDM. There are also other “specific types” of DM that have been identified, but they are diagnosed secondary to genetic anomaly, existing medical condition/ diagnosis or drug use (1,13). T2DM, originally referred to as adult onset DM, has a number of modifiable risk factors and may be managed with lifestyle modification (diet and physical activity), but may also require pharmacological treatment. Table 2.1. lists select risk factors for T2DM (13).

The Public Health Agency of Canada (2011) reported that from 2008 to 2009 prevalence of T2DM for adults over age 20 years was 8.7% (95% Confidence Interval [CI]: 8.72 to 8.74%); representing one out of 11 Canadians (98). Nine out of 10 individuals living with DM have T2DM and 60,000 new cases are diagnosed each year in Canada (99). Canadians living with DM are three times more likely to be hospitalized with CVD, 12 times more likely to be hospitalized with end-stage renal disease, and approximately 20 times more likely to be hospitalized with lower limb amputations, when compared to Canadians living without DM (100). Annual per capita health care costs have been estimated to be three to four times greater for those living with DM. Although cost estimates available from the Public Health Agency are 8 outdated by 11 years, it is suspected that with the increasing number of Canadians being diagnosed with DM and DM related complications, health care cost is also increasing (100).

Recognizing the relatively low prevalence of T2DM in the general population and the importance of cost effectiveness, it is recommended that clinicians screen individuals with risk factors rather than conduct mass screening (101-103). Fasting plasma glucose (FPG) and/ or 1 glycated hemoglobin (HbA1C) are currently the recommended screening tests, however a 75 g oral glucose tolerance test (OGTT) is indicated when the FPG is 6.1 to 6.9. mmol/ L and/ or

HbA1C is 6.0 to 6.4%. OGTT may also be indicated when FPG is 5.6 to 6.0 mmol/ L and/ or

HbA1C is 5.5 to 5.9% and greater than or equal to one risk factor for T2DM is present (table 2 2.1.). T2DM is diagnosed when FPG ≥ 7.0 mmol/L and/ or HbA1C ≥ 6.5% or when one of the following OGTT results is obtained: ≥ 7.0 mmol/ L or 2 hour post-OGTT value of ≥ 11.1 mmol/L (101). For the complete screening and diagnosis algorithm for T2DM, see the CDA CPG (2013) pages S12 to S15; http://guidelines.diabetes.ca/. Moreover, The Canadian Diabetes Risk Assessment Questionnaire (a validated tool), offered by the Public Health Agency of Canada, is an option for DM risk assessment in the Canadian population; available at: http://healthycanadians.gc.ca/diseases-conditions-maladies-affections/disease-maladie/diabetes- diabete/canrisk/index-eng.php.

Table 2.1. Select risk factors for type 2 diabetes mellitus as per the Canadian Diabetes Association Clinical Practice Guidelines 2013

Risk factors include:  Age ≥ 40 years of age  First-degree relative with type 2 diabetes mellitus  Member of a high risk population (e.g. Aboriginal, African, South Asian)  History of pre-diabetes  History of gestational diabetes mellitus  Overweight; BMI ≥ 25 kg/m2  Abdominal Obesity  Polycystic ovarian syndrome  Psychiatric disorders (e.g. depression, schizophrenia, bipolar disorder)  Human Immunodeficiency Virus (HIV) infection  Glucocorticoids BMI = body mass index; ≥ = greater than or equal to

1 Hemoglobin A1c: A biochemical test that is an indicator of one’s average blood glucose concentration for the past two to three months (101). 2 A repeat confirmatory laboratory test is required in the absence of symptomatic hyperglycemia (101). 9

2.1.2. Gestational Diabetes Mellitus: Definitions, Statistics and Guidelines for Screening

GDM is defined as hyperglycaemia with onset or first recognition during pregnancy (14). In Canada, the prevalence of GDM varies from 4% in the non-Aboriginal population to 8 to 18% in Aboriginal (First Nation, Inuit and Metis) populations (14-18). Untreated GDM or gestational hyperglycaemia has been shown to lead to increased prenatal and perinatal morbidity in mother and offspring, while treatment has been associated with maternal and neonatal outcomes comparable to pregnancy defined by normoglycaemia (14,69,104). GDM was once defined as transient, but any degree of prenatal gestational dysglycaemia has more recently been associated with increased risk of postpartum impaired FPG, impaired glucose tolerance, and T2DM (14,19,22). This is related to impairment of both insulin secretion and action in women with previous diagnosis of GDM. Three to 6 months postpartum, women with prior GDM have a 16 to 20% risk for dysglycemia. After 9 years, 20% of women with prior GDM will develop T2DM (14,19-21,105).

In 2013, the CDA released a revised CPG for management of GDM; discontinuing the diagnosis of impaired glucose tolerance of pregnancy (IGTP) (14). Despite CDA CPG 2013 publication, diagnostic criteria for GDM remain controversial. In light of this controversy and ongoing research, the CDA CPG Expert Committee included a “preferred approach” and “alternate approach” to screening in the most recent CPG (figure 2.1. and 2.2.). Both approaches recommend all pregnant women should be screened for GDM at 24 to 28 weeks gestation, with high risk clients being offered screening at any stage of pregnancy (examples of risk factors are listed in table 2.2.). The preferred approach is to begin with a 50 gram glucose challenge test and, if appropriate, proceed to the 75 gram OGTT (14). Diagnosis of GDM is made using the OGTT if ≥ one value is abnormal (fasting ≥ 5.3. mmol/L, 1 hour ≥ 10.6 mmol/L, 2 hour ≥ 9.0 mmol/L). The alternative approach is a one-step approach where the 75 gram OGTT is administered and GDM diagnosis made if ≥ one value is abnormal (fasting ≥ 5.1. mmol/L, 1 hour ≥ 10.0 mmol/L, 2 hour ≥ 8.5 mmol/L) (14).

Figure 2.1. Preferred approach for the screening and diagnosis of gestational diabetes mellitus

All pregnant women between 24 and 28 weeks gestation

High risk patients should be offered screening at any stage of pregnancy

50 g glucose challenge test with PG 1 hour later

< 7.8 mmol/L 7.8 to 11.0 mmol/L ≥ 11.1 mmol/L

Normoglycemia 75 g OGTT measure FPG, 1hPG, 2hPG

Reassess at FPG ≥ 5.3 mmol/L 24 to 28 weeks if 1hPG ≥ 10.6 mmol/L tested earlier 2hPG ≥ 9.0 mmol/L

If one value is met or

exceeded

Gestational Diabetes Mellitus

Adapted from: Canadian Diabetes Association Clinical Practice Guidelines (2013); Can J Diabetes 37; S168-183. PG = plasma glucose, 1hPG = 1 hour plasma glucose, 2hPG = 2 hour plasma glucose, FPG = fasting plasma glucose, GDM = gestational diabetes mellitus, OGTT oral glucose tolerance test; < = less than, ≥ = greater than or equal to.

Figure 2.2. Alternative approach for the screening and diagnosis of gestational diabetes mellitus

All pregnant women between 24 and 28 weeks gestation High risk patients should be offered screening at any stage of pregnancy

75 g OGTT measure FPG, 1hPG, 2hPG

FPG ≥ 5.1 mmol/L 1hPG ≥ 10.0 mmol/L 2hPG ≥ 8.5 mmol/L

If one value is met or exceeded

Gestational Diabetes Mellitus

Table 2.2. Select risk factors for gestational diabetes mellitus as per the Canadian Diabetes Association Clinical Practice Guidelines 2013 (14) Risk factors include:  Previous diagnosis of gestational diabetes mellitus  Member of a high risk population (e.g. Aboriginal, Hispanic, South Asian, Asian, African)  Age ≥ 35 years  Body Mass Index ≥ 30 kg/m2  Polycystic ovarian syndrome diagnosis  Acanthosis nigricans diagnosis  Corticosteroid use  History of infant with macrosomia (infant birth weight > 4000g) 12

2.1.3. Standard Care for Type 2 Diabetes Mellitus and Gestational Diabetes Mellitus: A Focus on Medical Nutrition Therapy

The 2008 CDA CPG concentrated on client self-management in the context of a supportive inter-professional healthcare team (106). Although these are still key elements of DM management according to the CDA CPG (2013), elements beyond the client and healthcare provider are highlighted in current recommendations (e.g. the role of the health care system, the client’s community, quality improvement strategies) (107). DM self-management education, the process of supporting individuals while they manage their DM, has been considered an integral part of DM standard care since the 1930s. Several meta-analyses have shown that self- management is associated with reductions in HbA1C ranging from 0.36 to 0.81%. Moreover, a number of well-respected educator training programs are based on this approach; such as Stanford University’s Chronic Disease Self-Management Program (Better Choices, Better Health® Workshop) and Diabetes Self-Management Program (108-110). Client self- management education and support includes knowledge and skill building and active client participation using traditional (e.g. lecture) and novel education approaches (e.g. hands-on activities). Self-management education aims to empower the client to engage in his/ her own problem solving, decision making, goal setting and action planning and is facilitated using motivational interviewing techniques and evidence-based approaches to classical behaviour change theory application (110).

As part of standard care, the CDA CPG (2013) recommends that all individuals with DM be assessed and followed by a Registered Dietitian (RD) to establish client-centered medical nutrition therapy that aims to achieve euglycaemia, weight management (e.g. weight loss in T2DM, weight gain in GDM) and adequate nutritional intake. Optimal glycaemic control is essential to management of DM (111). Individuals diagnosed with T2DM or GDM are generally referred to a multi-disciplinary outpatient clinic to receive standard care (in the case of GDM, these clinics will be referred to as Diabetes in Pregnancy Clinic [DIP] in this dissertation). There is a three pronged approach to DM medical therapy, including medical nutrition therapy, physical activity, and pharmaceutical intervention. For individuals living with T2DM, self- monitoring of blood glucose is often used as a supplementary tool to other measures of glycaemia (e.g. HbA1C), but is recommended (110,111). Self-monitoring of blood glucose can 13 provide client and health care provider with important feedback that can inform treatment, serve as a way to help the client understand the relationship with these treatments and their blood glucose, and reinforce DM self-management education. CDA recommends that a self- monitoring schedule for a client living with T2DM be individualized. Also, glycaemic targets for T2DM, should be individualized based on age, duration of diabetes, risk of hypoglycaemia, whether or not a given client has been diagnosed with CVD, and his/ her life expectancy (111).

Therapy for most clients living with T2DM should aim to achieve an HbA1C of ≤ 7.0% in order to reduce the risk of microvascular complications. In order to achieve this HbA1C target, clients living with T2DM should be encouraged to keep FPG between 4.0 to 7.0 mmol/L and two hour postprandial blood glucose between 5.0 to 10.0 mmol/L (111). Clients with GDM are provided with glucometers and asked to log their blood glucose, dietary intake and physical activity. They are asked to test their blood glucose (on average) four times per day; including once in the fasting state and once two hours after breakfast, lunch and dinner. Current blood glucose target values as per the CDA CPG are: (1.) FPG < 5.3. mmol/ L, (2.) 1 hour postprandial < 7.8 mmol/L, (3.) 2 hour postprandial < 6.7 mmol/L (14).

For adults with T2DM, attaining and maintaining a healthy weight is a short and long term goal of DM medical therapy. It is recommended that client height and weight be used to calculate body mass index (BMI), which are compared against current Canadian guidelines for body weight classification (table 2.3) (112-114). For clients with GDM, weight gain recommendations are based on pre-pregnancy BMI and the most recent recommendations published by the Institute of Medicine (IOM) (table 2.4.) (115). Medical nutrition therapy is a main intervention used to achieve weight targets with clients with DM. In general, individuals living with DM are asked to follow Eating Well with Canada’s Food Guide (based on the current Dietary Reference Intakes [DRIs]; including the Acceptable Macronutrient Distribution [AMDR; table 2.5.]). Other education available to RDs and their clients include: The Diabetes Food Guide, The Plate Method, and 2 or 3 dimensional food models. Twenty four hour recalls and three day diet records are also often used as education tools in this context (to build rapport) (116,117).

Table 2.3. Canadian Guidelines for Body Weight Classification in Adults Using Body Mass Index (112-114)

Classification BMI Category (kg/m2) Risk of developing Health Problems Underweight < 18.5 Increased risk Normal weight 18.5 to 24.9 Least risk Overweight 25.0 to 29.9 Increased risk Obese Class I 30.0 to 34.9 High risk Obese Class II 35.0 to 39.9 Very high risk Obese Class III ≥ 40.0 Extremely high BMI = Body mass index; < = less than, ≥ = greater than or equal to

Table 2.4. Recommendations for total and rate of weight gain during pregnancy, by pre- pregnancy body mass index (115)

Pre-pregnancy BMI BMI* (kg/ m2) Total Weight Gain Rates of Weight Gain category Range (lbs) 2nd and 3rd Trimester (lbs) (Mean range in lbs/ week)** Underweight < 18.5 28 to 40 1 (1 to 1.3.) Normal weight 18.5 to 24.9 25 to 35 1 (0.8 to 1) Overweight 25 to 29.9 15 to 25 0.6 (0.5 to 0.7) Obese (All classes)*** ≥ 30 11 to 20 0.5 (0.4 to 0.6) *In agreement with Health Canada’s Nonogram; http://www.hc-sc.gc.ca/fn-an/nutrition/weights- poids/guide-ld-adult/bmi_chart_java-graph_imc_java-eng.php. **Calculations assume a 0.5–2 kg (1.1–4.4 lbs) weight gain in the first trimester (based on Siega-Riz et al., 1994; Abrams et al., 1995; Carmichael et al., 1997). ***A narrower range of weight gain may be advised for women with a pre-pregnancy BMI of 35 or greater. Individualized advice is recommended for these women.

Table 2.5. Acceptable Macronutrient Range; Percent of Energy (Dietary Reference Intakes) (117)

Age Range/ Macronutrient Carbohydrate Fat Protein 4 to 18 years 45 to 65 % 25 to 35 % 10 to 30 % 19 years and over 45 to 65 % 20 to 35 % 10 to 35 %

Hypocaloric and low or very low carbohydrate diets are not recommended in treatment of T2DM and GDM, but current guidelines suggest that RDs review meal planning with their clients, emphasizing moderate carbohydrate restriction (14,118-121). Carbohydrate restriction, in the context of GDM standard care, typically translates to distributing a finite carbohydrate intake (45%; DRIs) over three meals and three snacks (one snack at bedtime). RDs typically recommend clients do not consume a diet where carbohydrate makes up less than 45% of their total caloric intake. Hypocaloric and very low carbohydrate diets are especially not recommended in clients with GDM because restrictive dietary patterns may result in weight loss, clinically significant ketogenesis and inadequate nutrient intake (e.g. B vitamins, fibre, protein, calcium) to support a healthy pregnancy, delivery and lactation for mother and baby (14,118-120). A low GI diet has been proposed as an alternative to restriction, because it can be layered onto current dietary recommendations and does not encourage/ reinforce restriction (29,40,122).

CDA CPG (2008, 2013) recommends that GI education be used as a supplement to T1DM and T2DM standard care with use varying by client interest, ability and need. There is no similar section and statement in the CPG on GI in GDM management (13,106). Despite this, existing data on GI utility in this patient population is promising, as a low GI dietary pattern has the potential to improve glycaemic control (both postprandial and fasting), while allowing women to consume adequate energy and nutrients to support a healthy pregnancy, delivery and lactation (24,40,64,66,70-72,104,123-128).

2.2.0. GLYCAEMIC INDEX: AN OVERVIEW

2.2.1. Glycaemic Index: Origins in a Desire to Improve Diabetes Medical Nutrition Therapy

The term “glycaemic index” (GI) first appeared in the literature in the early 1980s. This concept, and associated methodology, was proposed as a method to categorize carbohydrate foods according to glycaemic effect during the postprandial or fed period (27-29). Carbohydrate containing foods are typically categorized according to the following GI ranges: Low GI (≤ 55); medium GI (56 to 69) and high GI (≥ 70) (29-31). A GI value is obtained when the incremental area under the blood glucose response curve (iAUC), after consumption of 50 grams of available carbohydrate (carbohydrate excluding dietary fibre) from a (test) food, is compared with the iAUC obtained after consumption of 50 grams of available carbohydrate from glucose (consumed by the same person on a separate occasion). This value is expressed out of 100 or as a percent (although units are not typically included with GI values in peer reviewed and popular literature). White bread is an accepted reference food for GI testing, as per the International Organization for Standardization (ISO) scientific methodology for GI testing (29,129). Regardless of reference food selected, all GI values must be reported on the glucose scale. A conversion factor of 0.71 is used to convert from the higher bread scale (GI of white bread =100) values to the glucose scale values (GI of white bread = 71) (29,31,129-131).

Two examples of phrases commonly used in GI literature are “GI mechanism” and “GI utility”. “GI mechanism” is typically used to describe the physiological and metabolic effects of a low or high GI food or dietary pattern. The “relevance” of these mechanisms is often described in terms of their role in prevention or treatment of a given condition (e.g. T2DM) (29,42,63,132). This concept is deeply intertwined with another phrase; “GI utility”. GI utility is a general phrase used to describe the effectiveness of GI in a given setting and for a given outcome (perhaps relevance) (28,29,62,63,133). For instance, in a research setting, GI utility might be to predict the relative glycaemic impact of two different meals. On the other hand, GI clinical utility can be defined as the ability of low GI foods to lower postprandial glycaemic response, but also may be defined as a GI education materials’ ability to increase client GI knowledge and/ or facilitate change in client diet GI. Depending on how one defines these terms, scientists may require application of different methodology and, in some cases, study design. We cannot conduct or 17 translate science unless words and phrases are clearly defined and understood by those involved in science generation and utilization (63,134-136).

A large body of evidence suggests that low GI foods or dietary pattern may be beneficial for the prevention or treatment of a number of chronic conditions; including T1DM, T2DM, overweight/ obesity, CVD (e.g. coronary heart disease, dyslipidemia, stroke), and cancer (e.g. colorectal, breast) (13,29,32-44,137-140). However, the majority of GI utility data are from individuals at risk for or living with T1DM and T2DM (29,35-38,45-51,53,54,56-59,61,63,141- 143). This is mainly because when the GI concept was developed, the main application considered was for DM management (T1DM and T2DM); specifically glycaemic control (e.g. postprandial blood glucose, HbA1c) (29). Over the past 34 years, much of the data generated from RCTs, examining the impact of low GI foods substitution on glycaemic control, have been supportive of GI utility in DM treatment (37,43,50,53,61,144,145). This said, not all low GI interventions have shown statistically significant improvements in glycaemic control (59). This inconsistency has fuelled controversy in the literature and, in some cases, led to divergent opinions of respected experts (26,28,63,146-148). Despite this controversy, the evidence in support of using the low GI dietary pattern to manage glycaemic control in people living with DM is considered strong enough for inclusion in the current CDA CPG (T1DM: Grade B, Level 2; T2DM: Grade B, Level 2) and to serve as the basis of nutrition education programs in other countries (e.g. Australia) (31,144,149). Moreover, in studies where the low GI diet did not significantly affected glycaemic control, side effects and symptoms are usually not reported. These data indicate that this dietary pattern is, at least, not harmful to consumers and is comparable to control or standard care (40,53,59,71,124).

As highlighted above, generally clinical trials suggest a positive/ direct relationship between GI and glycaemic control (i.e. lower GI associated with lower blood glucose) in people living with T1DM and T2DM. Many of these studies have been reviewed in the aforementioned CDA CPG (2008, 2013) and in recent systematic reviews and meta-analyses (37,41,43,50,61,144,145). For instance, Brand-Miller et al. (2003) published a meta-analysis that examined 14 clinical trials, including 356 participants (203 with T1DM, 153 with T2DM) and ranged in duration from 12 days to 12 months; an average of 10 weeks in length. This meta-analysis showed that low GI diets reduced HbA1c by 0.43% points (95% CI 0.72 to 0.13%). Moreover, when the authors pooled the HbA1c and fructosamine data and baseline differences were accounted for, glycated 18 proteins were reduced 7.4% (95% CI: 8.8 to 6.0%) more on the low GI diet than on the high GI diet. These findings were in agreement with a subsequent meta-analysis, conducted by Opperman et al. (2004) that included sixteen clinical trials; nine of which include participants diagnosed with T2DM. Analysis of all clinical trials (396 participants with T2DM [n = 228], T1DM [n = 105], CVD [n = 46] and free from chronic disease [n = 17]) showed that lower GI diets significantly reduced fructosamine by -0.1 mmol/L (95% CI: -0.20 to 0.00 mmol/L) and

HbA1c by 0.27% (95% CI: -0.5 to -0.03%).

Similar findings were obtained in a subsequent systematic review and meta-analysis on 13

RCTs examining the effect of a low GI diet on glycaemic control (HbA1c) in participants with T2DM and were in agreement with subsequent clinical trials (145). For instance, Jenkins et al.

(2008) showed that HbA1c decreased by - 0.50% absolute HbA1c units (95% CI: - 0.61 to - 0.39%) when participants consumed a low GI diet for 6 months. A reduction that was significantly lower (p < 0.001) than the reduction seen when participants consumed a high fibre diet (-0.18% absolute HbA1c units [95% CI: - 0.29 to - 0.07%]) over the same study period. Similarly, Wolever et al. (2008) found that disposition index (a marker of beta-cell function) was higher in participants who consumed a low GI diet for 12 months compared with those who consumed a low carbohydrate, high monounsaturated diet for the same length of time.

The low GI dietary pattern has also been shown to decrease the need for anti-hyperglycaemic medications in participants with T2DM presenting with hyperglycaemia (or uncontrolled DM). Ma et al. (2008) compared the effects of a low GI diet to the American nutritional recommendations of the American Diabetes Association (ADA) using a prospective parallel

RCT. Although both diets resulted in similar reductions in HbA1c at six and 12 months, the low GI diet group was less likely to add or increase use of medications (Odds ratio [OR]: 0.26 to 0; p = 0.01). That is, the low GI diet improved glycaemic control to the same extent as standard care, while reducing need for pharmacological intervention. The authors of this study concluded that the low GI diet is a viable alternative to the standard ADA diet and should be evaluated in this context using larger scale RCTs; noteworthy in light of ADAs historical stance on GI (63,150,151).

A meta-analysis of 37 prospective cohort studies (n= 1,950,198) found that a low GI diet is independently associated with reduced risk of CVD (Relative risk [RR]: 1.25; 95% CI: 1.00, 1.56; p = 0.05) (45). Correspondingly, results of clinical trials show that a low GI diet has the capacity to improve the lipid profile of individuals living with or without CVD. The aforementioned meta-analysis, written by Opperman et al. (2004), found that a low GI diet significantly reduced total cholesterol by -0.33 mmol/L (95% CI: -0.47 to -0.18 mmol/L) in all study participants and a subset of study participants with T2DM (p < 0.0001). The low GI diet also significantly reduced low density lipoprotein (LDL) by -0.24 mmol/L (95% CI: -0.45 to - 0.04 mmol/L; p = 0.02) in participants with T2DM; a significant finding that did not extend to analysis of all participants (p = 0.06). Opperman et al. (2004, 2005) also found that a low GI diet did not affect high density lipoprotein (HDL) and triglyceride (TG) in study participants with T2DM; findings in agreement with a recent Cochrane Reviewof 21 RCTs (n = 713), conducted by Kelly et al. (2004). Participants included in this systematic review and meta- analysis had at least one major risk factor for coronary heart disease (CHD; e.g. abnormal lipids, DM, overweight) (41). Conversely, a subsequent six month RCT, conducted by Jenkins et al. (2008), that showed an increase (or improvement) in HDL cholesterol in participants consuming the low GI diet compared with a decrease in HDL in the high fibre group (low GI: 1.7 mg/dL [95% CI: 0.8 to 2.6 mg/dL] vs. high fibre: -0.2 mg/dL [95% CI -0.9 to 0.5 mg/dL]; p = 0.005). Although the CDA CPG highlights increasing plasma HDL as a potential benefit of the low GI diet, the low GI diet is not currently part of standard care aimed at achieving current LDL targets (primary target of treatment for dyslipidaemia) for Canadians living with T2DM (13,152).

A number of limitations of prospective food-based GI intervention studies have been identified by the authors of the above work (and others). Three frequently cited examples include: (1.) Inherent limitations of dietary data collection tools and free-living participants (e.g. reporting bias), (2.) variability of GI assignment at the dietary intake data entry/ analysis stage, and (3.) diet related confounders (e.g. Is it fibre or GI causing the effect on the dependent variable?) (29,37,43,45,79,145,153-157). Standardized techniques for dietary data collection and evidence informed procedures are the main strategies nutritional scientists use to minimize the impacts of limitation 1; approaches that were applied during the studies summarized in this dissertation (40,59,158). Relevant to limitation 2, recent meta-analysis of observational studies examining 20 the relationship between GI and chronic disease risk made note of the apparent subjectivity of GI assignment as a significant limitation of studies reviewed and their analysis. The authors of this analysis note that by ensuring accurate GI values are assigned, scientists are more likely to avoid misclassification of GI; reducing bias toward the null hypothesis and the risk that the observed effect size will be diminished (45). Limitation 2 has been a focus of research and development within our laboratory for a decade (2005 to 2015) and has resulted in establishment of intra-laboratory standard operation procedure for GI value assignment (appendix 2.1.). Finally, dietary confounders are a noteworthy limitation to dietary intervention evaluation (limitation 3). This limitation has been acknowledged for decades in the context of GI research. Efforts have been made to design interventions and methodologies that control diet-based confounders and isolate the effect of lowering GI on clinically relevant outcomes (e.g. glycaemic control, lipid profile) (29,37,40,79,159,160). Scientists have varying success in addressing this limitation to date, but that majority have successfully shown individuals with DM can reduce dietary GI, while maintaining caloric, macronutrient, and fibre intake. Controlling for energy, macronutrients and fibre has become a mainstay of food-based GI trial design/ methodology (29,37).

2.2.2. Low Glycaemic Index and Gestational Diabetes Mellitus: The State of the Evidence

There are limited Canadian data on low GI diet utility in GDM treatment. Our research group has published the results of one parallel prospective RCT (pilot study) to examine GI utility in this population (40). Forty-seven women (at [mean] 28 weeks gestation), with GDM or IGTP, were randomized to either a low GI or standard care diet (stratified by diagnosis). As part of standard care, the study participants were asked to measure their blood glucose using a glucometer and to log one fasting value and three postprandial values each day. As part of the study, participants were asked to provide self-monitored blood glucose values from 28 weeks gestation until delivery. Both dietary patterns improved glycaemic control in this sample, but those consuming the low GI dietary pattern did have more postprandial glucose values within target when compared to those on the standard care dietary pattern (low GI: 58.4% [n = 1891], standard care: 48.7% [n = 1834]; p < 0.001) (40). Dietary intake data, collected using three day diet records, was described in detail in this paper. Dietary analysis revealed that those on the low GI diet had a dietary GI 9 units lower than those in the standard care group post-intervention 21

(low GI: 49 ± 0.8 vs. standard care: 58 ± 0.5; p = 0.001). Moreover, participants consuming the low GI diet consumed 7 g more fiber post-intervention, reported finding the diet acceptable and indicated they would continue the diet after study completion (40). These findings are in agreement with Moses et al (2006) who reported participants found the low GI diet easier to follow in comparison to the standard care diet.

The majority of published data on the effect of low GI foods/ diet on GDM prenatal, perinatal and postpartum outcomes have been generated by Brand-Miller and colleagues at the University of Sydney and University of Wollongong, Australia (64,67,71,72,124,125,128,161). In a recent systematic review, this group examined evidence on the relationship between low or high GI diet and maternal nutrition and pregnancy outcomes. Eight studies were included in this systematic review; including four epidemiological studies and four intervention studies. Of these studies, three looked specifically at women with GDM; two epidemiological studies, with incidence of GDM as the outcome variable, and one intervention study (two armed RCT; low GI versus standard care) with incidence of insulin prescription was the primary outcome (72). The GDM-specific intervention study, conducted by Moses et al. (2009), showed that assignment to a low GI diet reduced insulin prescription in women with GDM without compromising maternal or neonatal care. Of the 31 women randomly assigned to the low GI diet in this study, nine required insulin. Of the 32 women assigned to the higher GI diet, 19 required insulin (p = 0.023). Moreover, nine of these 19 women were able to avoid insulin use by changing to the low GI diet during the study period (128). These findings are noteworthy, as 50% to 60% of women with GDM following standard care are prescribed insulin (40,128,162). Moreover, although insulin therapy has been shown to be effective in reducing incidence of macrosomia, it has also been linked to an increase in maternal weight gain and physical, emotional and financial stress (162,163). Lastly, it does not rectify peripheral insulin resistance and it has been shown that insulin can increase the incidence of small for gestational age3 infants when used “aggressively”. Medical nutrition therapy that can reduce use of pharmacological treatment is of great interest to both health care team members and clients (162-165).

3 Small for gestational age is most commonly defined as an infant mass below the 10th percentile for the gestational age (168). 22

Counter to their previous work in this area, the Australians conducted an RCT that did not show a clinically relevant effect of a low GI diet on prenatal and perinatal outcomes in women with GDM (n = 99). Louie et al. (2011) reported that although a significantly lower dietary GI was achieved in the intervention group (low GI: 47 ± 1 vs. standard care: 53 ± 1; p < 0.001), a significant difference in birth weight, prevalence in macrosomia, insulin treatment or perinatal outcomes was not found between the two study groups. Post-intervention dietary GI for both groups fell within the low GI category post-intervention, but the difference in dietary GI between groups was comparable to other GI interventions that have seen weight loss, improved beta-cell function and insulin sensitivity (71,166,167). Like Grant et al. (2007, 2011), Louie et al. (2011) noted the difficulty of achieving measurable differences in well-controlled or “intensely controlled” patients. Subsequent to this work, Louie et al. (2013) critically evaluated the evidence regarding the effect of dietary GI on pregnancy outcomes in women with GDM. The authors concluded that women with GDM are likely to benefit from following a low GI dietary pattern, but called for more (larger scale) RCTs to provide clinicians and government agencies with the information they need to make an informed decision. As part of the discussion, they noted that a low GI diet should not be taught in isolation of or as an alternative to current standard care, but layered onto it (124). A message reflected by other recent reviews on GI utility (26,63). In an invited editorial, Moses and Brand-Miller (2011) also stressed that science available on GI in GDM is promising, but incomplete and inadequate to justify integration of GI into CPG for GDM management. In this editorial they stated:

“Hopefully the proposed (multi-center trial) study of Grant et al… already underway...will provide the quantum of science that will make these recommendations possible.”(64).

This statement and above sentiment is further supported by a recent systematic review and meta- analysis (Cochrane Review) that included nine RCTs (n=429) and assessed the effect of different dietary patterns (including low GI) on maternal and neonatal pregnancy outcomes (including incidence of caesarean section, macrosomia, fetal or neonatal mortality) in women with GDM (68). The authors reported that no significant effect of diet was detected and called for more medical nutrition therapy evaluation; specifically clinical trials with sufficient power and long term health outcome measures.

On the contrary, Viana et al. (2014) conducted a recent systematic review and meta-analysis (including the work of Grant et al. [2011] and Louie et al. [2011] above) of three dietary interventions that have been evaluated in the context of GDM treatment: (1.) Energy restriction (two studies, 425 participants), (2.) low carbohydrate diet (n = 2, 182 participants) (3.) low GI (n = 4, 257 participants). In general, this analysis showed that the low GI diet was associated with less maternal insulin use and lower birth weights in participant offspring when compared to the control/ standard care diet. That is, the low GI dietary intervention reduced the proportion of participants who used insulin (RR: 0.767; 95% CI 0.597 to 0.986; p = 0.039) and the newborn birth weight (mean difference = -161.9 g; 95% CI -246.4, -77.4; p ≤ 0.0001) (24). Interventions based on energy and carbohydrate restrictions did not have an effect on maternal or neonatal outcomes of interest (e.g. proportion of maternal insulin use and caesarean section, neonatal birth weight, hypoglycaemia) (24). Like many of the studies published on GI utility in GDM, however, this study did not examine blood glucose, stating that proportion of participants who used insulin and infant body weight are “hard(er) outcomes” and implying this may be more relevant to GI utility (24,71,72,128,161). Although we understand the authors’ perspective, we would argue that self-monitoring blood glucose continues to be a key component of GDM medical therapy and to understanding GI mechanism and therefore warrants attention in the context of RCTs, systemic reviews and meta-analyses. This argument was recently supported by a review conducted by Hernandez et al (2013) that include six RCTs (n = 250) and listed exclusion of self-monitored blood glucose from study outcomes as a limitation of this body of literature and called for more practice-based research on GI utility in GDM management (14,40,70).

The use of low GI diets in pregnancy is controversial because any reduction in the rate of large for gestational age4 has the potential to be matched by an increase rate of small for gestational age (164,168). The Camden Study was an epidemiological study that assessed the diets of 1,082 normoglycaemic pregnant women using 24 hour recalls. A resulting original paper published by Scholl et al. (2004) is often cited in the introduction/ discussions of original papers, included in systematic reviews and meta-analysis and is often considered in the development of interventional studies on the effects of GI during pregnancy. The Camden Study results showed

4 Large for gestational age is defined as an infant mass greater than the 90th percentile for gestational age (168). 24 that HbA1c and plasma glucose increased by 0.0006% and 0.013 mmol/L (respectively) per unit increase in dietary GI (p < 0.05). Analysis of these data also showed that women with a dietary GI < 50 (low GI category as per CDA) gave birth to an infant with a significantly lower birth weight (116 g; p < 0.05). Moreover, Camden mothers, consuming a low GI diet, had twofold increased risk of a small-for-gestational-age infant at birth; a source of concern and dialogue for the authors of this paper (168). This all said, clinicians working with women with GDM are typically aiming to prevent macrosomia or large for gestational age infant. Moreover, Scholl et al. (2004) did not provide sufficient information on food database GI value assignment or previous participant exposure to GI education; a limitation also highlighted in the T1DM and T2DM literature review (45, 168).

Also noteworthy, participants in the Camden Study falling in the lowest GI quintile ate more refined sugar than the higher quintiles. This finding must be interpreted with a critical lens. Upon closer examination, it becomes clear that the majority of Camden participants’ intake was rich in sugar, with approximately 50% of dietary carbohydrate coming from sugar. Hence, Camden study design (observational), GI methodology and participants’ overall dietary intake pattern limits finding generalization and data comparison to dietary interventions based on GI education (168). Recommendations related to added sugars are covered as part of standard medical nutrition therapy, which should serve as the basis of any GI education platform/ intervention (GI education layering) (63). Despite all of this, there are not sufficient data on GI utility in GDM, so it is recommended the lessons learned regarding infant birth weight from the Camden Study be kept in mind; especially when working with clients on “aggressive” insulin treatment (which can increase risk for small for gestational age) (162,164,168).

Women with GDM are a population that deserve attention from the clinical and scientific community for a number of reasons. Retnakaran et al. (2010) and others have highlighted GDM as a model for studying the early events in the natural history of T2DM. Nutrition educators interested in DM prevention and treatment also highlight this client group, as they recognize the importance of creating GDM medical nutrition therapy that can be efficacious acutely and prolong latency to maternal T2DM development. Pregnant women are often motivated to make changes for the sake of the fetus and are often gatekeepers to their family and surrounding community (14,105,169-171). This said, there is interest in the scientific community regarding 25 the unique challenges that must be considered when designing education or behaviour change initiatives for these women (examples/ data below). Hui et al (2014) interviewed 30 women receiving standard nutrition therapy for GDM in DIP Clinic in Winnipeg, Manitoba. This qualitative study aimed at collecting data on participants perceived barriers to implementing dietary advice. “Time” (or lack thereof) was highlighted as one of the biggest challenges for women in this study. That is, participants reported feeling that clinicians’ dietary change expectations were unreasonable in the given time frame (two weeks until first follow-up/ ~ 12 weeks until delivery) (172). Winnipeg participants reported that “unreasonable expectations” created “stress” and “anxiety”. Additional challenges and barriers reported by participants include: Their (1.) food preference conflicted with dietary advice, (2.) food environment conflicted with dietary advice (e.g. social meals), (3.) knowledge and skills where not where they needed to be to ensure effective dietary management; despite receiving standard care. These findings are in agreement with other literature, where authors list socioeconomic barriers, lack of cultural diversity in nutrition education, lack of choice when dining out and lack of time when cooking at home as barriers to dietary behaviour change during GDM (66,172-175). Hui et al (2014) also reported that when clients felt restricted, they would express looking forward to eating their favourite foods after the baby was born. Contrary to this findings, we have found that women with GDM, consuming either a standard care or low GI diet intervention, planned to consume their study diet after baby was born (an ethnically and culturally diverse sample of women) (40,122). Despite this, it is clear, more research has to be done to comprehensively assess GI utility, including education application sustainability, in ethnically and culturally diverse samples diagnosed with GDM.

2.2.3.0. Perceived Barriers to Glycaemic Index Utility: Revealing Pathways to Translation

2.2.3.1. National Institutes for Health Research Translational Roadmap: A Conceptual Framework for Maneuvering the Bench to Bedside Continuum

The phrase “bench to bedside” is commonly used in the health science literature, but use (meaning) can vary between papers and/ or authors. For many, this phrase is used interchangeably with “translational research” or “translational medicine”, but it is important to keep in mind that each of these phrases can mean different things to different people (89-92,95- 26

97,176-179). Translational research/ medicine can be defined as medical research that is concerned with facilitating the practical application of scientific discoveries to the development and implementation of new ways to prevent, diagnose, and treat disease (96). The IOM and National Institutes of Health (NIH), have described two distinct definitions of translational research; describing them as two “translational blocks” (often referred to as T1 and T2) (95,177). T1, in this context, is defined as the transfer of new understandings of disease mechanisms gained in the laboratory to new methods for diagnosis, treatment and prevention. This form of research requires trained staff and research settings that can support molecular biology and other basic sciences and can include animal research and preclinical studies (e.g. studies designed to understand basic mechanisms). T2, on the other hand, is the translation of results into everyday clinical practice and decision making (e.g. CDA CPG) (95,177). The laboratory for T2 is the community and health care settings, where population-based interventions and practice-based research can be conducted. This type of research requires transdisciplinary/ interdisciplinary collaboration to be efficient and effective. More recently, T3 has been added to the “NIH Roadmap” (figure 2.3). This additional “laboratory” and third translational step aim to solve problems encountered by physicians as they try to incorporate new discoveries into clinical practice. These definitions and roadmap are one of the theoretical frameworks considered in the development of the studies included in this dissertation; specifically the omnibus (dissertation) rational and objectives (95,177).

Figure 2.3. National Institutes of Health Research Roadmap

Clinical Practice

Bench (Lab 1) T1 Bedside (Lab 2) Efficient care delivery Preclinical Studies Case series Phase 3 clinical trials Controlled observational Identification of new clinical Animal Research Phase 1 & 2 Clinical Trials studies questions and gaps in care T2

T2 Practice-based Research T3

Guideline Development (Lab 3) Dissemination Research Meta-analyses Implementation Research Systematic Review Phase 3 and 4 Clinical Trials Observational Research Survey Research

Adapted from Westfall et al. (2007) (177); Informed by Woolf (2008) (95); T = translational block/ step, Lab = laboratory; The original National Institutes of Health Roadmap included two major laboratories; bench and bedside and two translational steps (T1 and T2). T3 was proposed by Westfall et al. (2007), adding a third laboratory (practice based research) and T3. This addendum aimed at providing researchers and clinicians with a theoretical framework to improve incorporation of research discoveries into day-to-day care. This roadmap is a continuum, with overlap between laboratories and translational steps. This figure includes some examples of research common in each of the three laboratories and each step (T). This figure is not exhaustive. 28

2.2.3.2. The Role of the Dietitian in Nutritional Science Knowledge Translation

Knowledge translation (KT) is a term used to describe the relationship between knowledge creation and application. The translation of knowledge to practice or policy is a fundamental component of the knowledge-to-action process; a set of phases or activities developed by Graham et al (2006) to guide scientists and end-users as they exchange knowledge and improve health services and products (89,90,96,97). Agencies supporting health research continue to expect evidence/ results to be published in peer-reviewed journals (example of end-of-grant translation), but also expect a comprehensive dissemination strategy be included in the original grant proposal (integrative translation) (94,96,97). This evolution of Canadian research directives, inspires the very important question, “who is responsible for the use of evidence?”. More conservative views of KT regard the researcher as the producer of knowledge and the clinician or client as the user of knowledge. More and more often, however, clinical scientists are identifying and filling roles and responsibilities along the knowledge to action pathway (continuum) (25,63,92,93,97,179-183). In fact, it has been recognized that if researchers are willing and able to act as an administrator, planner, consultant, or at least able to oversee these processes, the potential for using research findings to improve clinical practice is increased. As highlighted by Goering and Wasylenki (1993), when researchers can assume multiple roles (e.g. researcher and clinician/ knowledge user), this reduces the gap between science and practice; a key mission of KT.

RDs have a noteworthy capacity to facilitate nutritional science knowledge-to-action. In fact, RDs have long identified a synergy between KT and dietetic practice in the literature (63,91,92,179,181,184). That is, many of the theoretical frameworks upon which KT is built (e.g. classical behaviour change theorems, planned action approaches, client-centered practice) are the basis of RD training and practice in Canada. DC accredited institutions (e.g. Canadian Universities and Internships that have met the professions accreditation standards to train RDs) have programming guided by The Integrated Competencies for Dietetic Education and Practice (ICDEP) (185). The integrated competencies consist of an interconnected set of practice competencies, performance indicators and foundational knowledge specification rooted in evidence. At entry-to-practice, entry-level proficiency in all practice competencies is required. The performance of a practice competency requires application of a combination of knowledge, 29 skills, attitudes and judgments; including the ability to assess scope of practice, develop and apply transferable skills and understand and implement education principals facilitating client- RD exchange. This exchange is reflective of KT and supportive of engagement of stakeholders and knowledge users from knowledge to action (25,90-92,179,181,183,184). From a KT perspective, the success of such exchanges reflects repeating cycles of the knowledge-to-action cycle - where generating practice-based or community-based research leads to evidence-based practice (63,89,90,177,179,186-189).

Practice-based Evidence in Nutrition® (called PEN) is another example of Canadian RDs efforts to bring knowledge to action. Including powerful search tools and a knowledge pathway format, PEN is a dynamic knowledge translation subscription that has been created by RDs (and other healthcare professionals) for RDs. It is a synthesis of evidence that has been evaluated and graded to support RDs as they bring evidence to practice (much like the grading protocol implemented by the CDA to evaluate evidence for their CPG) (149,188-190). In fact, a GI knowledge pathway called “The Diabetes and GI Knowledge Pathway” was developed with our input, in 2010, as part of our ongoing integrative translation efforts. Other existing KT initiatives spearheaded by DC include Learning on Demand, the Tools to Support Practice webpage and the annual DC Conference. Tools to Support Practice includes 10 evidence-based mobile/ web-based applications RDs can use to facilitate knowledge sharing; some of which they can use with their clients as patient education materials (http://www.dietitians.ca/). Canadian RDs have the knowledge and skills necessary to play a key role in facilitating nutritional science translation and education development, implementation and evaluation. They have published reviews and commentaries calling for our profession to effect and influence change in the face of our evolving practical and research environments (25,179,181,183,183, 188,189). By taking an active role in understanding the current research environment, the relationship between practice-based research and evidence-based practice, and meeting new feedback, information and challenges with an open mind, authors believe RDs can contribute to KT. RDs are often translating nutritional science to the end-user or client. It is therefore imperative that RDs’ (and other educators) perceived barriers to GI utility be addressed by GI researchers and they be involved in study/ education development, implementation and evaluation (25,63,91-93,179,181,183,184,191-194).

2.2.3.3.0. Perceived Barriers to Glycaemic Index Utility

As highlighted in previous sections of this dissertation, the CDA CPG (2008, 2013) includes statements supporting that GI education be used as a supplement to T2DM standard care with use varying by client interest, ability and need; leaving the final decision of to use or not to use up to the discretion of the clinician. Two supporting statements from the CPG (2013) include:

“Dietary advice may emphasize choosing carbohydrate food sources with a low glycemic index to help optimize glycemic control.”

Despite the existence of supportive guidelines, the literature clearly highlights that nutrition educators, more specifically RDs, and (some) scientists continue to question the utility of GI in their practice (62,63,137,147,148,195). For instance, a postal cross-sectional survey was conducted to collect information on RDs use and perception of the GI concept. This survey was completed by members of the Dietitians of Canada (DC) and Ordre Professionnel des Dietetistes du Quebec (OPDQ). It was developed to measure how many Canadian RDs identify as GI “users” or “non-users” and their perceived benefits of and barriers to GI utility, GI knowledge, and confidence in teaching the concept to clients (62).

A total of 6060 RDs (DC = 4014; OPDQ = 2046) were originally contacted. Of the total respondents (n = 2857), 40% (n = 724) identified as users, while 60% (n = 1,081) identified as non-users. One thousand and fifty-seven respondents reported treating clients with DM. Of this subset, 39% (n = 415) identified as users and 61% (n = 642) as non-users. Users were more likely to have a greater DM caseload and perceive benefits to and confidence in teaching the GI concept. Only 3% of non-users (n = 19/ 642) were unaware of the GI concept (62). Of members who identified as users, 90% (n = 374 / 415) used GI as a general descriptor of carbohydrate absorption (fast [high GI] versus slow [low GI] and the slow absorption model), 56% reported teaching GI with the aim of facilitating glycaemic control, 49% used it as part of daily meal 31 planning and 49% used it as part of counseling aimed at supporting clients in weight control. Members who identified as non-users and worked with clients living with DM identified the following four barriers to GI utility in their practice (n = 642):

(1.) 51 % lack of convincing scientific evidence (2.) 79 % reported lack of education resources (3.) 77 % felt the GI concept was too complex for clients to understand (4.) 76 % felt the GI concept was too complex for clients to use (62)

Although the most comprehensive survey-based evaluation of Canadian RDs use and perceptions of GI utility, other studies on clinicians/ educators perceptions of GI-utility have since been published that support and/ or reflect these findings/ barriers to utility (26,62,63,133,137,196). In a critical review, published in 2011, we reviewed each of these perceived barriers to assess if they were supported by evidence or if they were based upon misinformation or “mythology”. This style of translation has been shown to be effective in addressing misinformation in scientific literature. That is, data suggests that dissemination efforts may be more successful if creditable researchers create evidence-based resources (e.g. review articles, job aids) aimed at dispelling misconceptions (“myth busting”), while supporting client-centered care and knowledge users’ prior knowledge, skills and informed choice (63,197).

2.2.3.3.1. Perceived Barrier: Glycaemic Index is Too Difficult for Clients to Understand5

Traditionally and consistently, the transfer of research findings to practice has been and is often a slow and inefficient process. KT research, strategies and training has developed out of an unwavering aim of scientists to address this issue. The overarching goal of the KT initiative is to ensure Canadians are not denied treatments and products of proven benefit because of the lag time from bench to bedside or because of translation break-down or inefficiency (96,198). As highlighted above, educators’ and clients’ involvement in the GI knowledge –to-action process is invaluable to avoid “getting lost in translation” and feedback (knowledge exchange) is part of this process (63). The perceived complexity of GI has been repeatedly noted as a barrier to GI utility and translation (perceived barrier 3 and 4) by educators. In support of this barrier, some

5 Section 2.2.3.3.1. also addresses the perceived barrier: lack of education resources. 32 clinicians and scientists highlight that GI education opposes current dietary guidelines, GI terminology is confusing, and there is a shortage of GI education materials and job aids for nutrition educators to efficiency teach the concept (perceived barrier 1 and 2) (62,63,133,137,148,150,151).

Evidence-based GI education focuses on the concept of carbohydrate quality and the effect of this quality on carbohydrate absorption, digestion and metabolism. These principals should be presented as a supplement to dietary recommendations (layered onto standard care); not as an alternative to it (31,40,63,141,144,146,199). GI critics claim that the low GI diet encourages increased use of unhealthy food choices and may promote increased energy consumption (147,148,150,151). The existence of this misinformation is not surprising, when only 61% of Canadian RDs who identify as GI education users and 26% of non-users report being aware of CDAs position on GI (62). In contrast to this perception, a number of studies and critical reviews conducted within and outside of our laboratory have shown low GI foods can be consumed as part of a healthy diet (40,47,63,67,125,200,201). For instance, Frost et al. (1994) facilitated adherence to dietary recommendations using low GI education. Specifically, study participants on the low GI diet consumed less dietary fat and more fibre (47). Moreover, Grant et al. (2011) used low GI education as a supplement to standard care to obtain improved glycaemic control in women with GDM and IGTP. During this study, adding low GI education to standard care did not result in confusion among study participants. In fact, participants’ receiving supplementary low GI education consumed significantly more fibre than participants receiving standard care only (e.g. low GI: 30 ± 1.6 g, standard care: 23 ± 1.0 g; p = 0.001). Moreover, 100% of study participants receiving and applying the low GI diet, in the context of this study, reported planning to use these lessons learned post-study.

Translation of scientific jargon and concepts must be developed with the target audience (end user) in mind; regardless of the topic at hand (63,134-136,202-205). This involves establishing audience-appropriate phraseology; whether the audience is comprised of researchers and clinicians, individual clients or whole populations. Mendes et al. (2006) found that GI was not deemed appropriate for clinical use by 78% of American RDs treating children for obesity. RDs surveyed indicated they felt knowledgeable about GI (77%; n = 92), but felt GI terminology/ concepts were too challenging for their clients to understand. Although this paper is often cited 33 to support the criticism GI terminology is confusing for clients, the authors of this work reported RDs perception of what clients find difficult rather than client satisfaction or knowledge uptake (identical limitation to data reported by Kalergis et al. [2006]; above) (137). The perceptions of GI utility reported by Mendes et al. (2006) may have been rooted in the long held view, of American health agencies, that GI did not provide additional benefit to standard care (63, 150,151).

There is currently insufficient data to support the statement that GI is too difficult for clients to understand. In fact, there is a noteworthy amount of literature to support the contrary (40,46,67,128,196,206,207). For example, Frost et al. (1994), Grant et al. (2011), Moses et al. (2009) and others have showed that participants successfully and significantly lowered diet-GI after receiving GI education. Moreover, Slabber (2005) completed a comprehensive review that provided readers with insight into consumer understanding of the GI and suggestions for practical guidelines for incorporation of low GI foods into diets. This review highlighted that GI terminology need not be any more difficult than teaching other standard medical nutrition therapy concepts. For instance, low and high GI can be explained using terms like, “slow and fast acting carbohydrate”. “Retrogradation” can be explained using the following phrasing: “When cooked (red) potatoes are cooled in the fridge, the starch in them becomes sticky and gel-like.” (196). Finally, international efforts to blend education principals and novel teaching tools with GI education have been successful in communicating GI messages effectively. For example, Miller and Lindberg (2007) created an interactive computer game designed to teach several principles about GI and health (e.g. slow absorption model, the importance of layering GI onto current dietary recommendations). Adults aged 18 to 30 years (n = 65) either completed the GI computer game or reviewed United States (US) Department of Agriculture’s MyPyramidWeb site regarding healthy eating guidelines. The GI computer game group showed greater gains than the control group in knowledge (p < 0.001) (207).

Canadian RDs believe a shortage of GI education materials is a barrier to GI utility (62,63,133). To overcome this barrier to utility, the CDA has published a GI education tool that can be downloaded from the CDA website; called: The Glycemic Index (30). This handout (or patient education material) is a food substitution list including low, medium and high GI food choices frequently used in the Canadian diet. The aforementioned GI categories are included in this tool: 34

Low GI (55 or less; choose most often), medium GI (56 to 69; choose more often) and high GI (70 or more; choose less often). The CDA tool format is based upon evidence and developed with input from clinical scientists, basic scientists and DM educators. This tool also summarizes current dietary recommendations and key take-home messages for DM management; supporting the recommendation that GI be taught as a supplement to standard care (144). Various versions of the GI food substitution list are available in peer-reviewed literature, popular media, and online. Food substitution lists have been used in our laboratory and others to achieve a moderate difference in food choice and to obtain a significant change in dietary GI (typically the lists are used to lower dietary GI) (37,40,54,59,61,161,208).

Despite these efforts, a recent client education material assessment conducted by Southgate and Wolever (2012) supports the belief that there is a shortage of “suitable” GI education materials in Canada. Applying the Suitability Assessment of Materials (SAM) instrument, developed by Doak et al. (1996), Southgate and Wolever evaluated five publicly available GI handouts. When using the SAM instrument, handouts are assessed for suitability using the following six categories: (1.) content, (2.) literacy demand, (3.) graphics, (4.) layout and typography, (5.) learning stimulation and motivation, and (6.) cultural appropriateness. A total of 22 factors are assessed across these categories, with each factor having one of the four following ratings (paired with a numerical ranking): (1.) Not Suitable, (2.) Adequate, (3.) Superior, or (4.) Not Applicable. Three of these 22 factors are considered to be “go-no/go” factors (209). If any of these factors are found to be unsuitable during material assessment, Doak et al (1996) recommends rating a material “Not Suitable” irrespective of the ratings it obtains in the other 19 factors. The three go-no/go factors are: (1.) reading grade level, (2.) match in cultural logic, language and experience, and (3.) cultural image and examples. Of the handouts assessed by Southgate and Wolever (2012), two were Canadian GI education tools that are available to RDs across Canada. The two Canadian education tools reviewed were: 1. CDA’s The Glycemic Index Education Tool and 2. Understanding the Glycemic Index, published by The Canadian Sugar Institute (30,210). Despite both handouts being sourced from reputable groups, evidence-based and developed with the input of nutrition professionals, both were deemed “Not Suitable”, using the SAM instrument, due to at least one “no go” factor (209). Moreover, the Canadian Sugar Institute handout contained/ contains conceptual errors and therefore misinformation. For instance, the following statement is included in this tool: 35

“Fat or protein eaten along with carbohydrate… reduces the GI of the carbohydrate.” (210).

The authors of this tool mistakenly used the term GI when they are describing glycaemic response; highlighting and propagating a hiccup in the translation process. GI is a characteristic of available carbohydrate and is not synonymous with glycaemic response (211). Protein and fat have effects on glycaemic response which are independent of those produced by carbohydrates and occur by different mechanisms (29,211,212). The proper terminology would be, “Adding fat and protein to a carbohydrate reduced the glycaemic response”. This same error exists in a review published by Galvao Candido et al (2013), perpetuating confusion regarding the distinction between GI and glycaemic response. This is despite the authors being supportive of GI utility, GI food substitution lists and citing Grant and Wolever (2011) (a review that distinguished between GI and glycaemic response) (199).

Although this review supports that GI education does not oppose current dietary guidelines and there is not sufficient data to claim that GI is too difficult for clients to understand, GI terminology is confusing to clinicians and educators. Canadian Sugar Institute handout and the review published by Galvao Candido et al (2013) are examples of how confusion, secondary to GI terminology and phraseology, may exist among clinicians and researchers. Moreover, the shortage of reliable GI education materials and job aids available to Canadian nutrition educators is likely affecting their self-efficacy and ability to efficiently and accurately teach the concept. Research assessing clinicians’ comprehension of GI concepts and terminology and examining GI misinformation in academic literature, websites and popular media are warranted.

2.2.3.3.2. Perceived Barrier: Glycaemic Index is too Difficult for Clients to Apply

There have been criticisms cited in the literature to support the perceived barrier that the GI concept is too difficult for clients to apply. Two examples of these criticisms include: (1.) The low GI diet limits food choice and (2.) GI is not accepted by clients (63). Although these examples are noteworthy barriers to dietary change, these barriers are not supported by data specific to low GI dietary patterns (146,172-174,213). It is impractical for anyone to aim to consume low GI foods 100% of the time (35,63,146,196). The literature indicates that GI 36

“users” agree with this statement and educators recognize that flexibility is important for sustainable lifestyle change (26,146,214). Research shows low GI choices need only be consumed 50 to 60% of the time to significantly reduce dietary GI (5 to 9 units) and achieve published health benefits (29,40,47,166,167,200,215). Despite this, the following two barriers to low GI diet compliance were noted by Brekke et al. (2004): Lack of (1.) choice when dining out and (2.) ideas when cooking at home. Once again, these criticisms are not unique to the low GI diet, but are commonly expressed by people participating in nutrition-related behaviour change. In general, it is challenging for one to change lifestyle behaviors from contemplation to maintenance (216-221). To address these concerns, clinical scientists (including RDs) within and outside of our laboratory have designed GI recipe booklets, and modifiable low GI food lists (clients can add their own foods) and tips for dining out. These culturally sensitive patient education materials provide options for low, medium and high GI food/ meal choices (31,40,56,196,207,222,223).

The question of whether or not low GI is accepted by clients has generated a considerable amount of research. For instance, Burani and Longo (2006) assessed the effect of low GI medical nutrition therapy on multi-ethnic American adults living with T1DM and T2DM (n = 21) one year after education. At baseline 90% of the participants reported not understanding GI and 19% reported feeling they could include it in their current lifestyle. Post-intervention, average dietary GI was 45 (low GI CDA category) and a statistically significant decrease in GI was achieved by 95% participants. Moreover, post-intervention, 85% of participants reported they had adequate understanding of GI and 95% felt they possessed enough knowledge to apply it in their lifestyle after the study (38). All study participants accepted GI education as a useful supplement to current dietary recommendations, perceived low GI foods to be healthy and reported beneficial effects on glycaemic control and weight management. Perhaps most important to note, 100% of participants reported planning to include low GI as a permanent lifestyle change; indicative of low GI diet acceptability and sustainability in this sample (38). These findings are in agreement with those of a recent prospective RCT conducted in Australia where pregnant women following a low GI diet were more likely to agree their study diet was easier to follow in comparison to those on a medium-to-high GI diet (161). Analogous results have also been seen in samples comprised of children and young adults. Nansel et al. (2006) looked at low GI food acceptability of standard versus low GI menus in a 37 youth camp for children with T1DM and T2DM in the US, using a cross-over design. Food service staff were provided with low GI food and training aimed at creating a low GI menu. Camp kitchen staff reported low GI foods were acceptable in terms of preparation effort, perceived healthiness and client appeal. A questionnaire (Likert Scale format; 1 = “I didn’t like it at all”, 5 = “I liked it a whole lot”) was provided to the children after every meal and evening snack. Camp attendees (n = 140; age 7 to 16) provided comparable ratings for low GI and standard food served at dinner and snacks (dinner = 3.68 [low GI] vs. 3.79 [standard]; p = 0.30; snack = 3.74 [low GI] vs. 3.79 [standard]; p = 0.60). On the other hand, low GI food at breakfast and lunch were acceptable, but were rated lower than standard food (breakfast = 3.76 [low GI] vs. 4.04 [standard]; p = 0.01; lunch = 3.64 [low GI] vs. 3.88 [standard]; p = 0.01) (56). Similar findings were obtained during a previous long term prospective randomized trial in children living with T1DM. In this study, children on a low GI diet did not reduce dietary quality or choice in comparison to a control group using carbohydrate exchange dietary advice (48,141).

The data reviewed above does not support the statements: 1. The low GI diet limits food choice and 2. GI is not accepted by clients, but clearly supports that novel efforts to better understand Canadian clinicians’ and clients’ perceptions of GI education (and GI utility) are justified.

2.2.3.3.3. Perceived Barrier: Lack of Convincing Evidence

As highlighted above, the CDA CPG includes recommendations that GI education be used as a supplement to standard care for people living with T2DM and that use should vary by client interest, ability and need (144). Despite this, Canadian RDs, and some well-respected nutrition scientists, perceive there to be a lack of convincing evidence to support GI utility as part of DM medical nutrition therapy (62,63,133,147,148). This is important to note because the guidelines also highlight that the decision “to use or not to use” is up to the discretion of the clinician (106,144). Despite the support for GI utility in DM management, there are certainly areas of GI utility research that require further examination. For example, there are currently no explicit recommendations in the CDA CPG (2013) on GI utility in GDM treatment and a limited number of Canadian studies examining use of GI education in GDM treatment (14,40,144). Moreover, Canadian RDs do not feel they have sufficient understanding GI mechanism and evidence-based resources on GI; making it too complex to teach (62). Despite this, it appears that some of the 38 perceived barriers to GI utility may be a result of unsuccessful translation (e.g. jargon/ misinformation) and can be addressed by efforts such as critical evidence-based review, meta- analyses, and involvement of clinicians/ educators (e.g. RDs, Medical Doctors [MD]) at all stages of knowledge to action or from bench to bedside.

2.2.4.0 The Evolution of our Understanding of Glycaemic Index Mechanism: The Importance of Coming Back to the Bench

2.2.4.1. Diabetes Mellitus, Cardiovascular Disease and the Oxidation Hypothesis

Currently more than 10 million Canadians are living with DM or pre-DM (1). The majority (65% to 80%) of people with DM will die from CVD and people living with DM are at higher risk of developing CVD in comparison to people without DM (2-4). Moreover, people living with DM are at higher risk of CVD at an earlier age, with women being at higher risk than men (2,5,6). CVD is an umbrella term for a number of conditions adversely affecting the anatomy and physiology of the circulatory system. Coronary artery disease is one example of CVD and develops when a blend of lipid, cholesterol, calcium and scar tissue (plaque) build up in the arteries that supply blood to the heart. This buildup of atherosclerotic plaque narrows the arteries, impeding passage of blood (and oxygen) to the heart and brain, causing heart attack or stroke (7,8,11). This is relevant to Canadians because every seven minutes in Canada, someone dies from heart disease or stroke (11). Moreover, heart disease and stroke cost Canadians more than $20.9 billion every year in physician services, hospital costs, lost wages and decreased productivity (12). Oxidative stress and arterial stiffness are significantly higher in people living with these conditions and have been identified as measureable outcomes relevant to risk quantification/ prediction and (potentially) medical treatment planning and intervention (73,224- 231).

Oxidation is a normal physiological process of the human body that can be damaging to tissues when antioxidant capacity of the system is surpassed. Reactive oxygen species (ROS) are formed as a natural by-product of the normal metabolism of oxygen and have important roles in cell signaling and homeostasis (232-236). When ROS and free radicals overwhelm body defenses, they can damage deoxyribonucleic acid (DNA), proteins, carbohydrates and lipids constituents and compromise cell function. This process is often referred to as “oxidative stress” 39

(232,233,237,238). The effect of oxidation on human biological materials can be likened to the effect of rust on a vehicle. Rust can be a problem for car owners, since the outer most layer of paint is exposed to air and water. If the outer finish is not protected, the oxygen molecules in the air will eventually start interacting with the metal and cause rust. If the rust reaction is not kept in check by applying protection (e.g. wax) over time, the car body will break down. This is conceptually very similar to what happens to the body tissues if antioxidant capacity cannot keep oxidative stress in check. Oxidative stress has been linked to hypertension, insulin resistance, pancreatic β-cell dysfunction, abnormal serum lipids, coagulopathy, and systemic inflammation (73,77,224,225,227,228,230-233,238,238-246).

Hyperglycaemia, a defining characteristic of DM, causes selective cell damage, which has been linked to DM complications (e.g. CVD, retinopathy, neuropathy, and nephropathy). Endothelial cells, for instance, are at particular risk for damage from hyperglycaemia because their glucose transport rate does not decrease in the hyperglycaemic state (73,77,232,236,241,247,248). The mechanism by which this damage is thought to take place is by overproduction of superoxide anions via the mitochondrial electron transport chain (232). Hyperglycaemia has been associated with oxidative stress since the early 1960s, but review papers (and original research) by Ceriello (1997-onward), Vincent and Taylor (2006) and Brownlee (2005) have been instrumental in thrusting this concept into the limelight.

Oxidation of LDL is thought to play a key role in metabolic processes leading to atherosclerosis and CVD (249-254). The majority of LDL oxidation (LDLox) occurs in the intima layer of the artery (comprised of endothelial cells), but commences in the aqueous phase (plasma). That is, LDL is attacked by lipophilic radicals formed by free radicals that develop in the plasma when the water-soluble antioxidant defenses are overcome. Oxidative damage to LDL may range from slight structural alterations of mildly oxidized LDL to extensive breakdown of lipids and fragmentation of apolipoprotein B (apoB-100) (fully oxidized LDL) (10,250,251,254,255). The oxidative conversion of LDL to LDLox is a noteworthy event, as it initiates and accelerates the development of the atherosclerotic lesions (10,249-251,256,257). It has been recognized for decades that development of atherosclerosis is related to the level or concentration of LDL. This knowledge has been translated to clinical practice via clinical practice guidelines/ standard care practice, resulting in LDL concentration becoming a clinical target (152,250,251,258,259). 40

More recently, however, it has been recognized that the ultimate atherogenic agents are the oxidized form of LDL (230,254,255,260-262).

The oxidization of LDL is accompanied by alterations in its biological properties that accelerate uptake of LDL via “scavenger receptors” on monocyte derived macrophages. This process transforms macrophages to lipid-laden foam cells (a marker of plaque build-up) and does not occur by way of the classic Brown/ Goldstein LDL receptor (10,232,250,251,254-257, 259). The later point is important because by bypassing this receptor, uptake is unregulated by intracellular cholesterol content. LDLox is a chemo-attractant to monocytes and an inhibitor of macrophage motility, thereby promoting retention of macrophages in the arterial wall. LDLox is found in monocyte-derived macrophages in atherosclerotic lesions, but not in healthy arteries. LDLox and modified apoB-100 has, however, been detected in the plasma of both participants living with and without atherosclerosis and therefore can serve as a potential marker of risk (10,245,248-251,254,255). Much like particle size is considered clinically relevant in native LDL, LDLox depends on particle size. Participants with a predominance of small dense LDL have greater risk of coronary artery disease compared with individuals with a predominance of larger and more buoyant LDL. Small dense LDL have been shown to be more susceptible to oxidation (250-252,254,255,263,264).

CVD, its risk factors and complications (e.g. DM, age, hypertension, obesity, dietary factors, hypercholesterolemia and smoking), are associated with increased systemic endothelial dysfunction (ED) (257,265-269). ED has been identified as an early and prominent event in atherosclerosis. Over time, plaque hardens and narrows arteries, negatively impacting the flow of oxygen-rich blood to organs and other parts of the body and contributing to progression of arterial stiffening and ED (7,9,10,29,233,249,257,270). The endothelium, meaning “inner covering”, is the thin and slick (low friction) layer of squamous epithelial cells that lines the interior surface of blood vessels, hollow organs and lymphatic vessels. Released by the endothelium, Nitrogen monoxide (Nitric Oxide or NO) is an important signaling molecule and powerful vasodilator that is involved with many physiological and pathological processes (10,233,257,271-275). NO is the best characterized endothelium relaxing factor (ERF), but NO and ERF should not be used interchangeably. A decline in NO bioavailability may be caused by a decreased expression of post-translational regulation of endothelial NO synthase (eNOS), 41 deficiency of eNOS substrate or cofactors for eNOS activity, or by accelerated NO degradation because of its interaction with ROS or oxygen derived free radicals. A decline in endothelium- derived NO bioavailability (and associated oxidative stress) is a noteworthy factor in the decline of endothelial function and CVD risk and progression (267,269,272-275).

2.2.4.2. Antioxidant Capacity and Oxidative Stress Quantification

In the human model, the body’s first line of antioxidant defense is found in the plasma or extracellular fluids. Extra cellular fluid does not contain antioxidant enzymes, but does contain a number of antioxidant molecules that inhibit the oxidative process. These “inhibitors” are numerous, but the most often studied include vitamin C, vitamin E, uric acid and protein sulphydryl groups (236,276). Efforts have been made to define the relative contribution of each of these (and other) antioxidants to total antioxidant capacity (in vivo), but the results to date have not been conclusive (228,250,251,276-288). Barriers to conclusively identifying and quantifying the plasma antioxidants has been attributed to the likelihood that there are inhibitors that have not yet been identified/ defined and that most antioxidants act synergistically in vivo. For instance, studies have shown that dietary antioxidant supplementation increases overall antioxidant capacity of human serum more than the isolated dietary antioxidant (e.g. Selenium, Beta-carotene) concentration. Recognition of these quantification limitations led to the development of in vitro assays that measure general antioxidant capacity of human materials (e.g. Total peroxyl radical trapping antioxidant potential [TRAP]) (252,280-282,284,287). Moreover, when inhibitors of oxidation are overwhelmed (plasma defenses), peroxidation products have the opportunity to form. Many of the methods used to assess oxidative stress are based on the measurement of the concentration of peroxidation products. These end-products can be put in one of the following three categories: (1.) lipid peroxidation products, (2.) oxidized proteins, and (3.) DNA oxidation biomarkers (228,241,256,276,284,288).

There are various biochemical outcomes that can be measured to assess antioxidant capacity or oxidative stress. An open discussion of the pros and cons of the various antioxidant assays has been introduced in the literature in an attempt to identify and validate bench mark methods that can be broadly applied (228,276,280-282,284-288). To date, a benchmark assay for antioxidant capacity of human biological material (e.g. serum, plasma, breast milk, tissue) does not exist, 42 but some time-tested, widely applied assays do. Technicians/ researchers are encouraged to base choice of assay on their research question and the human materials being assessed for a given outcome (e.g. serum/ plasma, specimen), while weighing variables such as cost, existing resources, assay efficiency, and staff training. For example, assays developed to measure oxidation products and antioxidant potential of LDL cannot be used as a marker of general antioxidant capacity or potential (228,276,285). TRAP, however, can be and measures plasma capacity to neutralize free radicals in the serum. A hydrogen atom transfer reaction, it is very similar in methodology to the Oxygen Radical Absorbance Capacity Assay (ORAC). They both employ a thermolabile azo-radical inhibitor and Trolox standard (228,252,276,280,281,287, 288). Both assays have been frequently used in studies measuring antioxidant capacity of biological fluid of people living with or at risk for T2DM and/ or CVD. TRAP was chosen as our primary end-point because it is designed to answer our research question, our laboratory was able to employ this assay efficiently (existing materials available), our laboratory technician reported comfort with the assay and because of its noteworthy sensitivity to all known chain- breaking antioxidants (78,252,276,280,281,285,287).

Two commonly applied and well established outcomes used to assess LDLox include measurement of apoB-100 molecule oxidation and conjugated diene (CD) formation; a byproduct of polyunsaturated fatty acids (PUFA) oxidation (260,261,263,289). Both of these in vitro tests are regarded highly in the research community, as there have been noteworthy efforts made to make the tests specific, reliable, repeatable and accessible. They are also able to detect oxidation at relatively low levels. Assay kits have been developed to conduct apoB-100 oxidation quantification (e.g. Mercodia Oxidized LDL) (263). The international acceptance and time-tested nature of the diene conjugation method (ex vivo) has led to the development of an assay which measures the amount of baseline diene conjugation in circulating LDL as an indicator of oxidized LDL in vivo. This methodology has been applied in clinical trials as a repeated measure of CD (260,261). CDs are formed during the initial steps of PUFA oxidation - Hydrogen forms a double bond in PUFA, followed by molecular rearrangement, leading to the formation of conjugated double bonds, referred to as CD (250,251,260,261,289). Several independent studies have shown surprisingly strong associations between CD and known atherosclerosis risk factors (e.g. obesity, hypertension, diabetes and arterial function). A number of studies indicate that as an indicator of risk of atherosclerosis, CD exceeds sensitivity and 43 specificity of even the traditional (and widely applied) lipid markers of antherosclerosis (e.g. LDL, TG) (250,251,262,289,290). In terms of the age old question of which assay should be used and why?: Correlations have been established between various assays that measure concentrations of several peroxidation products, including CD, malondialdehyde, F2- Isoprostanes, lipid hydroperoxides, glutathione and protein carbonyls (276). This is promising for those hoping to streamline peroxidation quantification methodology. Scientists should aim to establish inter-lab correlation between assays before drawing comparisons between them or at least note the caveat as a limitation.

2.2.4.3. Non-invasive Measures of Cardiovascular Disease Risk

It has been well established that elevated brachial blood pressure (BP) is associated with an increased risk of CVD and mortality. Brachial BP is a non-invasive outcome, used as part of standard care, and is often factored into clinical decision making (152,291). Despite its widespread clinical application, brachial BP is affected by a number of covariates/ external factors (e.g. caffeine, room temperature) and is, therefore, not completely reflective of central aortic BP. Central aortic BP is a better predictor of cardiovascular events than peripheral BP (292-296). The left ventricle, kidney, and brain are affected by central (not peripheral) pulse pressure (PP) (265,292,293,296,297). PP is the difference between the systolic and diastolic pressure readings. It is measured in millimeters of mercury (mmHg). It represents the force that the heart generates each time it contracts. If resting BP is (systolic/diastolic) 120/80 mmHg, PP is 40. Like BP, however, brachial PP does not always provide an accurate indication of central PP (292-294,296,298). Similar to BP, central PP is a stronger predictor of all-cause mortality in clients with CVD disease when compared to brachial PP and BP. Despite these apparent strengths of central pressure as a marker of CVD risk, it is considered invasive and not accessible in standard practice settings (292,293). For decades, researchers have been exploring non-invasive methods that may improve upon or supplement standard techniques and many proposed non-invasive procedures have been compared against central pressure. To understand and appreciate the differences between these methods (and others that will be discussed in this section), one must possess an understanding of the role of the endothelium and vascular hemodynamics in the development of CVD (265-267,292-294,296,299-301).

Vital to understanding vascular hemodynamics is the concept of pulse wave pattern. A pulse wave is a wave in a blood vessel generated when blood is forced into the aorta by the contraction of the heart. Pressure changes within the blood vessels create wave action called the pressure pulse wave (265,296,298,302-304). The wave consists of two primary elements: (1.) the ejected wave and (2.) reflected wave. The ejected wave is generated by the contraction of the heart, which sends blood throughout the body. The reflected wave is generated by the reflection of the ejected wave from an artery bifurcation or a peripheral artery. The pattern formed by the superimposition of these two wave elements is called a pulse wave pattern (265,266,298,303,305,306). The pulse wave represents a local BP change caused by passage of blood from the left ventricle to the aorta and the wave action through the artery. Systolic BP (SBP) is measured during the period of ventricular contraction (creation of the wave). Diastolic BP (DBP) is measured between periods of ventricular contraction (265,270,307).

Also key to understanding endothelial anatomy and physiology and vascular pathophysiology and hemodynamics, is the concept of arterial stiffness (265,270,298,308-310). The stiffness of an artery is regulated by a number of factors including the following three factors: (1.) Structural elements within the arterial wall (e.g. elastin, collagen), (2.) distention pressure and (3.) vascular smooth muscle tone. Global endothelial function is significantly and inversely correlated with pulse wave velocity (PWV), augmentation index (AI) and PP (265,266,292,293,296,298,302, 306,310-313). Arterial stiffness is an independent predictor of all-cause and cardiovascular mortality in selected patient groups and is commonly measured using PWV and/ or AI. The pulse wave is transmitted through the arterial vessels at a speed inversely related to the viscoelastic properties of the wall; the less elastic the arterial wall the higher the velocity. Aortic PWV (measured using the carotid and femoral arteries) is the current gold standard measure for measuring arterial stiffness (the viscoelastic properties of the aorta) (265,295,296,298,302,304, 308,310,312-315).

To measure the PWV of the aorta, a proximal transducer is placed on the carotid artery and the distal transducer is placed on the femoral artery. PWV and arterial stiffness are often used interchangeably in the literature, as PWV is considered a direct measure of stiffness of an arterial segment (265). As touched on above, when the vessel is elastic, PWV is low. In this case, the reflected pulse wave tends to arrive back at the aortic root during diastole (not an 45 unfavorable phenomenon). In the case of stiff arteries, PWV rises and the reflected wave arrives back at the central arteries earlier, adding to the forward wave and augmenting the SBP (considered an unfavorable outcome) (265). This “augmentation” can also be quantified by AI (265,298,306,310). AI is a measure of arterial wave reflection that has also been identified as a non-invasive method for quantification of arterial stiffness. This estimate of systemic arterial (elastic + muscular) stiffness is defined as the difference between the second and first systolic peaks, expressed as a percentage of the PP (295,298,299,302,305,312). AI is the ratio of the peak of the reflected wave (P2) to the peak of the ejected wave (P1) (figure 2.4.). When a peripheral blood vessel is stiff, the reflected wave becomes larger. Also, when the blood vessels in the pulse wave propagation path are stiff, the time in which the reflected wave returns becomes shorter (298,303,305,310,314,316). As a result, the peak of the reflected wave will be near the peak of the ejected wave. These manifestations increase AI and the heart enhances myocardial contractility to overcome the increased BP. If this occurs chronically, the heart will become stressed and hypertrophy may occur. In short, when the AI is large, the load on the heart is large (266,303,305,314,316,317).

Figure 2.4. Pulse Wave: Variables that Comprise Augmentation Index

Reproduced with permission from: Sa da Fonseca, L. et al. (2014) World Journal of Cardiovascular Disease; 4: 225-235; P1 = ejected wave, P2 = reflected wave, AI = Augmentation Index.

CVD and atherosclerosis can affect many vessels in the human body (7). This said, there are many peripheral factors that have to be considered when deciding where arterial stiffness will be measured in the context of a clinical outcome of interest. Aortic AI and PWV, for instance, have been used as independent markers of increased and total CVD mortality. It is not recommended that radial AI (peripheral) be used to assess risk in this manner (295,296,298,310,312,313). Nevertheless, radial AI is an acceptable marker of peripheral vascular function and a valid option for assessing differences in this arterial stiffness between groups in clinical trials (265,295,296,298,308,310,312,313,316,317). A limitation of both approaches is that they may be difficult to record accurately in participants with obesity or peripheral artery disease. Moreover, PWV and AI increase with age, hypertension, DM, hypercholesterolemia and may be effected by medication. Medication type and dose can effect agreement between AI and PWV and each outcome independently (295,296,304,310,311,313,318). For instance, Lemogoum et al. (2004) found that AI and PP failed as surrogate markers of arterial stiffness during beta- adrenergic stimulation. This medication may be administered for the following conditions: Bradycardia, asthma, chronic obstructive pulmonary disease, and allergic reactions. These factors were considered in during the development of the first study included in this dissertation.

Although a number of devices have been developed to assess pulse wave forms, two devices were considered to answer our research question(s): (1.) The SphygmoCor® System (AtCor Medical Pty. Ltd., West Ryde, Australia) and (2.) the Omron HEM-9000AI (Omron Healthcare Co., Ltd., Kyoto, Japan) (265,305). Both devices implement semi-automated applanation tonometry at the radial artery. SphygmoCor Systems is currently the most popular device used for quantifying arterial stiffness. The SphygmoCor® System includes a tonometer and electrocardiograph. It measures PWV (in duplicate) using the R wave of the QRS complex of the ECG reference point. Central pressure waveform and ascending aorta pressure values are defined by means of a generalized arterial transfer function starting from the radial pulse wave. Omron HEM-9000AI® System (Omron Healthcare Co. Ltd, Kyoto, Japan) was developed by Omron Healthcare as an alternative to the existing SphygmoCor® System for assessment of central BP and AI to ensure accessibility (265,305). SphygmoCor® is a very expensive device and requires a noteworthy time commitment (also expensive) to staff training and quality control. Rather than having a built in transfer function like SphygmoCor® Systems, the Omron HEM-9000AI® System implements an algorithm based on a linear regression model which 47 assesses central SBP starting from the late systolic notch of the radial pulse waveform. The device calibrates the radial pulse waveform using arterial pressure values, measured on the brachial artery, but cannot define carotid-femoral PWV (265,305).

Omron HEM-9000AI® has been used in a number of clinical trials successfully, including published studies from Department of Nutritional Sciences, University of Toronto, and has been noted as an acceptable apparatus for AI measurement in small and large clinical trials (298,305,316,317). Volumes of literature (original articles, reviews, books etc.) are available on the comparison of the available devices to measure these outcomes. Generally, the literature calls for ongoing comparisons/ assessments, highlights the strengths and limitations of each device and the importance of selected apparatus based upon research question and laboratory resources (265,299,301,319). Richardson et al. (2009), for example, showed a strong correlation (r = 0.94; p <0.001) and no difference between mean values of late systolic shoulder of the radial pulse waveform (central SBP can be estimated from this outcome) (0.8 ± 4.8 mmHg; p = 0.4) measured by SphygmoCor® and Omron HEM-9000AI®. Although estimates of central BP produced by Omron HEM-9000AI® were significantly higher than those produced by SphygmoCor® (12.2 ± 4.6 mmHg; p < 0.001), peripheral AI measures were correlated between devices and showed no significant difference between absolute values. Results from both devices were also reproducible with significant correlations and no significant mean differences for values of central SBP and peripheral AI, but the authors called for additional studies that compared the devices to invasive direct measurements (299). This sentiment was echoed by Rezai et al. (2011), who compared central BP and AI between Omron HEM-9000AI®, Arteriograph and SphygmoCor®. In agreement with Richardson et al (2009), Rezai et al. (2011) found Omron HEM-9000AI® estimated the mean central SBP to be significantly higher than did SphygmoCor®. This finding is relevant from a standard care and research perspective, as other outputs from these devices are dependent on/ related to SBP.

2.2.4.4. The Effect of Low Glycaemic Index on Postprandial Oxidative Response: A Novel Glycaemic Index Mechanism

As highlighted above, dietary strategies to reduce postprandial blood glucose have been under examination for decades and have become a key part of medical therapy for DM and CVD (27,51,79,111,140,144,159,311,320-325). Before the term “glycaemic index” was even coined, however, research was underway to better understand the effects of lente or sustained release carbohydrate on markers of cardio-metabolic risk. The four “slow absorption models” that appear in the literature during the 1980s are: (1.) the low GI diet, (2.) nibbling versus gorging (increased meal frequency), (3.) addition of viscous fiber to food, and (4.) pharmacotherapy aimed at digestive enzyme inhibition (e.g. acarbose) (51,52,79,159,320-323,326-329). Although enzyme inhibitors, soluble fibers, and low GI foods all delay nutrient absorption and have been shown to reduce postprandial glycaemic response, each approach also introduces unique extraneous variables (or confounding variables), making it difficult for researchers to obtain a clear understanding of the mechanisms at play. In terms of GI mechanism, a number of original articles were published during the late 1980s and early 1990s that report efforts to isolate rate of carbohydrate delivery as an independent variable and glycemic control (postprandial blood glucose) as the dependent variable (79,142,159,160,320,321). A noteworthy contribution to this work, Jenkins et al. (1990) developed a study design and supporting methodology that was able to permit clear interpretation of the mechanisms at play when slow release carbohydrate is consumed. This methodological paradigm has since come to be referred to as the “sipping-bolus paradigm”. This paradigm and the results of this study are described in detail in the following paragraph, as this work served as the methodological framework of the first study included in this dissertation.

Using a crossover design, nine normoglycaemic male volunteers attended two 4.5 hour clinic visits. At each visit, participants consumed 50 grams of glucose in 700 ml of water. At one visit participants were asked to consume the solution in five minutes (bolus). At the other visit, participants were asked to consume the solution over a 3.5 hour period (sipping) (79). Four hours after participants took their first sip of glucose solution, 100 ml of 5% dextrose was given intravenously into a forearm vein over two minutes (intravenous glucose tolerance test [IVGTT]). Blood samples were collected at 0, 30, 60, 120, 180, and 240 minutes and every 5 49 minutes for the last 30 minutes of the study visit. Outcomes measured in all participants included: Blood glucose, and serum insulin, c-peptide and growth hormone, and free fatty acids (FFA). After bolus, the peak rise in mean blood glucose at 30 min was significantly above the mean blood glucose value after sipping (p < 0.002), whereas the nadir blood glucose after bolus was significantly below sipping at 180 minutes (p < 0.025) (79). After IVGTT, mean blood glucose over the final 10 minutes of the visit was lower after sipping, despite values being alike immediately before the IVGTT (p < 0.01). Serum FFA was suppressed up to two hours after both drinks were administered and this suppression was maintained until four hours after sipping the glucose solution. As expected, there was a rebound effect of FFA between 3 and 4 hours after the bolus treatment. After IVGTT, FFA were reduced on the bolus test but remained above sipping values throughout the IVGTT (second meal6) testing period. Noteworthy reductions in serum insulin (54 ± 10%, p < 0.001) and C-peptide (47 ± 12%, p <0.01) were also seen at the “sipping visit” (79). Bolus also elevated growth hormone levels from 180 min to the end of the study visit; this elevation resulted in significantly higher values compared to sipping (p < 0.05). This study and the authors closing remarks highlighted the need for more research examining the relationship(s) between delayed nutrient absorption, decreased glucose fluctuations, atherogenesis, insulin levels and CVD risk (79).

Since the publication of the sip-bolus paradigm, volumes of literature have been published supplementing our understanding of the physiological and metabolic behaviour of low GI foods in the human body (GI mechanisms) (28,29,42,46,59,132,330-333). As highlighted above, consumption of low GI food has been shown to improve glycaemic control when compared to consumption of medium to high GI foods (29,37,40,43,49,61,142,144,145). Moreover, also highlighted above, postprandial hyperglycaemia and glycaemic fluctuations have been linked to increased oxidative stress and ED (73,76,240,245,248,257,325,334-337). Accordingly, scientists have developed an interest in the effect a low GI diet on markers of antioxidant capacity and endothelial function, which has led to the consideration of novel GI mechanisms that may be influencing the aforementioned benefits of a low GI diet (e.g. reduced chronic disease risk) (78,338,339). Although available data on the relationship between a low GI dietary pattern and

6 Second Meal Effect: A second meal effect is when intake in one meal influences the postprandial response to a subsequent meal. In the context of GI, there have been many theories regarding the mechanisms responsible for how low GI effects second meal response; however the exact cause has yet to be demonstrated (159, 322, 367). 50 oxidative stress and/ or endothelial function are supportive, there are few studies published to date and the aforementioned limitations in study design/ methodology are present. For instance, the available studies are not designed to distinguish if the measured effect on the dependent variable is a result of dietary components, reduced postprandial hyperglycaemia or a combination of both (78,338,339). Moreover, in at least two original papers, it is not clear if the low GI treatment being administered is, in fact, composed of low GI test foods/ diets. Therefore, even in the case of a protective effect being observed, the available results have to be interpreted with caution; with quantified and unquantified extraneous variables in mind.

Botero et al. (2009) conducted a food-based intervention study to evaluate the acute effects of a low GI compared to a high GI diet on oxidative stress and other CVD risk factors. This crossover study included two 10 day inpatient admissions to a clinical research center. For the duration of each admission, participants (12 overweight or obese male participants aged 18 to 35 years) consumed a low or high GI diet. The dietary interventions were presented in random order, with a two to 12 week washout period between admissions (78). On day seven, total antioxidant capacity (plasma ORAC) and oxidative stress status (F2α –isoprostanes [F2IP]) were measured. ORAC measures resistance against oxidative damage and reflects the combined effects of all plasma antioxidants. F2IP, a family of arachidonic acid peroxidation products formed in phospholipids, released into circulation and excreted in the urine, serve as a biomarker of oxidative stress status (78). On day 10, traditional and novel markers of CVD risk were measured; including endothelial function (measured by flow-mediated endothelium- dependent vasodilatation [FMD] of the brachial artery), NO, insulin sensitivity and β-cell function (measured by frequently sampled IVGTT), blood lipids and C-reactive protein (CRP) (78).

Under fasting conditions, total antioxidant capacity was significantly higher in the low GI group; total ORAC (11,736 ± 668 vs. 10,381 ± 612 μmol Trolox equivalents/l [mean ± SEM]; p = 0.002). The area under the glucose and insulin curves (above baseline) were 2.2 and 1.8 times greater (p < 0.001) (respectively) two hours after a high GI breakfast in comparison to the low GI breakfast. Moreover, blood glucose nadir was significantly lower after a high GI breakfast (- 15 ± 2 mg/dl versus -6 ± 2 mg/dl compared to baseline values; p = 0.004) (78). No diet effects were observed for urinary F2IP, endothelial function, or NO under fasting conditions or during 51 the postprandial period. Repeated-measures analysis of postprandial endothelial function showed no diet effect on the response pattern or at any time point (p ≥ 0.18). None of the remaining outcomes were significantly different between diet groups (78). Although the study design and methodology were implemented according to an evidence-based protocol, this study is not without limitations. Although the authors made efforts to control for potentially confounding nutrients, this food based study was still limited by nutrient selection bias (e.g. the authors may be bias to which nutrients they control for). Therefore, it cannot be concluded that the reduced postprandial response is the key factor driving the favourable metabolic effects (i.e. what about the nutrients that were not considered by the authors?) (78). Moreover, it is not clear if and how diet GI was measured for each group. Although efforts were made to create diets that differed in dietary GI, some counterintuitive foods choices were made by the meal design team; including having black beans (GI = 20 to 30; low GI) on the high GI menu and sweet corn (GI = 59 to 62; medium GI) on the low GI menu (30,78).

Another food-based crossover study, published by Lavi et al (2009), reported the acute effects of various GI dietary carbohydrates in normoglycaemic overweight and obese subjects (n=56; ~70% men). Four foods were administered on four separate clinic visits scheduled approximately two weeks apart; placebo, glucose, cornflakes, and high fiber cereal. Serum glucose levels were significantly higher post-administration of the cornflake and glucose meal in comparison to the placebo and high-fibre cereal meal. Serum glucose was also significantly higher in the cornflake group in comparison to the three other groups at 120 min (338). FMD was measured before and after test meals at each visit and differences were calculated and expressed as Δ%FMD. Percent FMD decreased in all groups 2 hours postprandially, but was only significantly different in the glucose and cornflake group (p < 0.01). Although the results of this study were supportive of low GI, closer examination of the methodology showed that the test foods had “estimated GI values” and/ or it was unclear if the authors tested the GI of the test foods (338). Fifty grams of carbohydrate was fed from each of the test foods according to the authors, but it was not specified if this was 50 g of available carbohydrate. Also, the “high-fiber cereal” was not named and there is therefore no way to look up its GI value using publicly available GI databases (31,129-131,338).

Using a parallel randomized design, Philippou et al. (2009) randomized middle-aged men to a high (n=16) or low GI (n=22) diet for six months. All participants in this study were educated on current dietary recommendations. GI education was added to standard education in the intervention group. At six months, dietary GI (high GI: 63 ± 5; low GI: 51 ± 5; p < 0.001) and carbohydrate intake (high GI: 278 ± 7 grams/day, low GI: 224 ± 50 grams/day; p < 0.001) differed between groups (339). There were no other noteworthy differences in dietary composition detected between groups. Although a clinically and statistically significant reduction in energy was seen within groups, there was no significant difference in energy intake between groups during the study. Although, anthropometric measurements were significantly reduced within groups over time (p < 0.001 for weight, BMI, waist, and hip measurements), groups did not differ post-intervention (e.g. high GI: lost 3.0 ± 4.2 kg vs. low GI: lost 2.2 ± 3.6 kg; p = 0.3) (339). Over the six month study period, there was a bigger reduction in total cholesterol in the low GI group compared with the high GI group and also within the low GI group. Plasma LDL cholesterol and TG concentration fell only within the low GI group (p < 0.05). Fasting insulin (high GI: 48.9 ± 21.8 pmol/L vs. low GI: 32.6 ± 13.1 pmol/L; p < 0.01; homeostatic model assessment) and percentage insulin resistance (high GI: 0.88 ± 0.39 vs. low GI group: 0.61 ± 0.24; p = 0.02) were lower in the low GI than the high GI group at six months post-intervention. Glucose and insulin iAUC showed a significant effect of time (p < 0.001), but no differences between groups or visits and no interaction effects (339).

Groups did not differ in brachial SBP, DBP, and PWV at baseline or month 6. Both groups did show reductions in SBP (within in group change ± SD = high GI: −10 ± 10, low GI: −10 ± 10; p ≤ 0.01) and DBP (within in group change ± SD = high GI: −5 ± 7, low GI: −2 ± 9; p ≤ 0.01), whereas PWV fell only in the low GI group (median [interquartile range] = Baseline: 10.3 [10.0 to 10.9] to six months: 9.7 [9.3 to 10.3]; p ≤ .05). Also, twenty-four hour SBP and DBP both decreased within the low GI group and were significantly lower than the high GI group at six months (339). Although this study was conducted by a research group with experience in implementing low GI diet interventions, the GI in the high GI group post-intervention was 63.2 ± 5.6. This GI value is medium GI according to the CDA. Despite this, the post-intervention dietary GI was significantly different between groups (low GI group: 50.6 ± 4.6; p < 0.001) (339). One noteworthy confounder identified by the authors was a significant difference in dietary carbohydrate between groups. Moreover, although the energy reduction observed in each 53 group did not result in differences between groups, one could argue that it may be clinically significant (high GI: 236 ± 632, low GI: 447 ± 499 kcal/d; p = 0.3) (339).

2.2.5.0. The Role of Low Glycaemic Index Education Evaluation in Assessing Glycaemic Index Utility in the Context of Clinical Trials

To decrease a participants’ dietary GI, high and medium GI foods are substituted with low GI foods; typically using a food substitution list (29,30,37,40,54,59,61,63,122,161,208). Examples of lower GI food include: Pulses, pasta, parboiled rice and barley. Examples of higher GI food include: White bread, Corn Flakes, and white short grain rice (31,130,131). This method of substitution, known as the key foods strategy, is an evidence-based approach to establishing differences in dietary GI between study groups and has been used with success within and outside of our laboratory (used in food/ diet-based interventions above). By asking participants to use food substitution lists, scientists and educators ensure that food with a tested GI are used for food substitution (very important at the data analysis stage) (40,54,59,61,63,122,208). A significant decrease in dietary GI has been achieved by replacing 60% of starchy foods with lower GI carbohydrate. A reduction of 5 to 9 GI units has been associated with improvement in clinical outcomes relevant to T2DM prevention and treatment (e.g. glycaemic control, weight loss, improved beta-cell function and insulin sensitivity) (40,47,67,128,166,167,200,340). Education provided to participants and use of the substitution list can, however, vary between studies, with details of education delivery and evaluation often not being reported. This section presents the RCT as a means of evaluating GI utility and the education evaluation model that served as the backbone of the second and third study discussed in this dissertation.

2.2.5.1. Using the Randomized Control Trial to Evaluate Health Education Interventions

RCTs are regarded as a gold standard design in evidence-based medicine; second only to the systematic review and meta-analysis on the hierarchy of evidence (341,342). Despite this, RCTs have been described by some as unable to accommodate the complexity that characterizes programs that serve as the basis of public health interventions and health behaviour change interventions (e.g. nutrition education interventions) (192,214,343-349). Many health scientists have taken great strides to adapt the RCT design so that it can evaluate these complex 54 interventions, while maintaining the scientific integrity of the design (343,348,349). The “randomized cluster trial”, the “process evaluation trial” and the” realist RCT” are all examples of RCT adaptation; appropriate for public health and health behaviour change interventions (192,214,343,344,348,349,349).

In the randomized cluster trial, “clusters” are groups of individuals, such as families, clinics, hospitals, or entire communities. Cluster randomized trials decrease the risk of treatment contamination or contact between participants randomized to different interventions (343,348,349). For instance, clients obtaining standard care within the same clinic and/or participants living in the same home or small town may discuss study interventions. If one participant prefers a fellow participant’s intervention, this can lead to an undesired change in behaviour (from the perspective of the scientist). This form of contamination can result in reducing differences between the groups post-intervention and is often highlighted as a limitation of the RCT in the context of education-based health behaviour change interventions. Cluster randomization has adapted the RCT to overcome this limitation by allowing whole communities or clinics to receive the same intervention or education (343,348,349).

The process evaluation trial includes formative and summative evaluation of the intervention and experimental process as part of the experimental design. In addition to measuring the relationship between the intervention and the primary outcome, this design allows for evaluation of feasibility, methodology, participant satisfaction, and environmental factors (214,343). This approach aims to understand elements that may not have been identified as covariates during the intervention development stage, but may influence the dependent variable. Conceptually similar, the realist RCT also promotes multi-level evaluation of the intervention and the environment in which the intervention is designed and implemented. The realist RCT also puts emphasis on examining mechanisms of change and using both qualitative and quantitative methodology (mixed-methods) (343). Bonell et al. (2012) and others, interested in adapted RCT design, recommend investigating outcomes like knowledge score, attitude, and behaviour change independently and examining their effects on a given primary outcome (or dependent variable). It would be difficult, for some, to differentiate between these adapted RCTs and RCTs being conducted as part of an integrative KT plan (89,90,94,96,198,214,343).

Education-based health behaviour change interventions are considered complex for many reasons. A key contributor to this complexity is the challenge of effectively evaluating the relationship between the intervention and primary outcome (192,214,220,343,347,350). For instance, if an educator teaches a participant the knowledge and skills necessary to follow “X” dietary pattern and the participant does not, is it because the participant: (1) did not understand the education messages, (2) did not have the financial means to buy the food (cannot take action on the education messages), (3) cannot eat the food because of religious beliefs or (4) did not like the taste of the food? Within the context of the RCT, clinical measures (e.g. BP, biochemical outcomes, weight change) and/or changes in morbidity and mortality have been traditionally considered the intervention evaluation. More recently, the importance of including evaluation of participant health behaviour measures (e.g. smoking, diet and exercise), and personal, social and environmental characteristics have become recognized. Out of this recognition, has grown an awareness that the aim of intervention evaluation is not simply to assess if an intervention “works”, but to understand why it works and to examine sustainability so that it can be repeated, refined and disseminated (85,87,187,192,198,343,345,351-353).

There are a number of evaluation models, approaches and tools available to clinical trialists studying health behaviour change (e.g. The National Health Service [United Kingdom] Educations Outcomes Framework, the developmental model for the evaluation of health education programmes as per Nutbeam et al [1990], Kern’s Six Step Guide to Curriculum Design, Kirkpatrick Model of Education Evaluation); some of which have been implemented in the context of the RCT and some that have not (80-83,85,87,343-345,351,352). The Kirkpatrick Model (KM) was selected for evaluation of the education implemented during the education- based intervention studies included in this dissertation. This model was chosen for many reasons; including:

(1) KM has been used in the context of clinical research and high level (evidence hierarchy) research designs (e.g. systematic reviews and meta-analyses), (2) KM can be adapted to a multi-cultural, multi-disciplinary clinical setting in the context of an education-based health behaviour change intervention, (3) KM provides an evidence-based comprehensive theoretical framework for education development AND evaluation, 56

(4) The Kirkpatrick Partners (KPs) provided affordable training, strategies, support and tools to support intervention development, implementation and evaluation, (5) The KM evaluation framework has also been highlighted in the literature as an acceptable methodology for evaluation of health education and integrative KT efforts (80-83,85- 88,193,351,352,354-361).

2.2.5.1. The Kirkpatrick Model and Glycaemic Index Education Evaluation

The KM was originally developed in the mid-1950s by Dr Donald Kirkpatrick to measure the impact of the programs he was teaching at the University of Wisconsin Management Institute. The KM is a goal-oriented approach to training evaluation, traditionally presented as a four level triangle/ pyramid (figure 2.5.) that aims to communicate that higher levels (e.g. transfer and results) are supported by lower levels (e.g. reactions and learning) (80,81).

Figure 2.5. Kirkpatrick Method – The Four Levels

Level 4 (Results): To what degree do targeted outcomes occur as a result of the Results training event? Subsequent reinforcement? Level 3 (Transfer): To what degree do clients/ participants apply what they learned during training, when they are back on the Transfer job? Level 2 (Learning): To what degree do participants acquire the intended Learning knowledge, based on their participation in a training event? Level 1 (Reaction): To what degree do Reactions participants react favorably to the training?

Widely used as a training evaluation model in corporate/ industry settings, the KM has evolved for use in other settings engaged in training and/ or education (e.g. Hospitals, Universities). Decades of research conducted by the KP7 and others has made KM more comprehensive and flexible (80-83,85,86,88,355,358,359). The New World Kirkpatrick Model (NWKM; figure 2.6.) is a revised version of the KM that honours the original four levels. The NWKM has additional components, making it more comprehensive (e.g. Level 1 = engagement, relevance). These components aim to help users better translate information and assessments between levels and stem from user feedback (some of which can be found in peer-reviewed literature) (80- 83,86,351,362). Examples of published feedback, Eseyrel (2002) and Pehrson et al. (2011) wrote that the original KM does not adequately promote comprehensive needs assessment and expressed concern that it does not efficiently guide users to think about how they will use the results. Pehrson et al. (2011) also noted that KM asks “Was the training effective?”, but misses out on the more analytical question, “How can the training be modified in ways that increase its effectiveness?”. Moreover, Bates (2004) reported that KM does not take into account contextual factors such as the established cultures of learning, organizational unit goals, and values. In response to this and other feedback, KP has developed resources and services to assist users with using KM/ NWKM within local contexts (support to monitor and adjust) and has updated their approaches to encouraging needs/ environmental assessment before, during and after a program/ session is implemented. Moreover, NWKM provides users with strategies to modify training to increase effectiveness based on lessons learned in previous implementations (80).

The KM/ NWKM story is an example of the challenges of program planning, the strengths of integrative evaluation and dissemination and the value in comprehensive training and literature review. The NWKM was used as a conceptual framework for intervention development and evaluation for the second and third study included in this dissertation. Due to the familiarly of the four levels, we chose to represent the model using the original KM triangle in this dissertation (81). Results relevant to these four levels will be included for study two and three.

7 A comprehensive list of KP publications can be found at http://www.kirkpatrickpartners.com/. 58

Figure 2.6. The New World Kirkpatrick Model

Level 1 (Reaction) Monitor and Adjust

 Engagement  Relevance  Satisfaction Level 4 (Results) Level 2 (Learning) Level 3  Leading Indicators (Behaviour)  Desired Outcomes  Knowledge  Skills On-the-job  Attitude learning  Confidence  Commitment

KM has been identified as a comprehensive theoretical framework that supports KT philosophy and methodology. For instance, has been used as a model to guide development of eligibility criteria and the data abstraction process during systematic review (a tool used in knowledge synthesis) (84-86,354-360). Boet et al. (2014) used KM levels as eligibility criteria in a systematic review of research examining the use of simulation-based learning for teaching non- technical skills to healthcare professionals (like communication and leadership) on workplace practice and client experience. On the other hand, Leslie et al (2013) conducted a systematic review of research that evaluated faculty development programs in medical education. To facilitate abstraction of the various study outcomes, an adapted version of KM was used (e.g. reaction, attitudes, knowledge, skills, behaviour, organizational benefit); reflective of NWKM. The adapted KM used by Leslie et al. (2013) was built upon a previously published systematic review; also aiming to improve teaching effectiveness in medical education. In the context of systematic reviews, applying KM in this manner is most frequently observed. Another example, Hauer et al. (2012) also used KM as a standard for comparison during systematic review of research evaluating behaviour change counseling curricula aimed at medical students/ trainees. Moreover, Siddiqui and Jonas-Dwyer (2013) used the KM to systematically review research looking at improving efforts to assist health educators in integrating mobile learning into their teaching.

KM has been also adapted to evaluate medical education/ training and client care outcomes in the context of quality improvement and client health education; including medical nutrition therapy (84-86,352,354,355,358). Clinical medical nutrition therapy and nutrition intervention is based on education and behaviour change, but not all trials are designed to effectively evaluate these factors (if they measure them at all) (83,192,198,214,343,344,346,347). As highlighted by KM, without comprehensive intervention evaluation we cannot begin to claim that a given outcome (level 4) can be attributed to a given intervention. KM can offer tools and support to clinical scientists to involve the “end users” (e.g. educators/ clinicians, clients/ clients) in the process of education/ intervention development. This all said, it is important to keep in mind that there a many evaluation models, approaches and tools available and although KM offers very important information, it can only offer correlation, not causation (81,85,86,88,352).

Slater et al. (2012) used all four levels of KM to evaluate The Training and Action for Patient Safety Program; described as an innovative multi-disciplinary, team-based training program that embeds patient safety within quality improvement methods. Participants included medical doctors (various levels of experience), allied healthcare team members, administrators and managers (55 participants, divided into 11 multi-professional teams). Evidence at level 1 was obtained using a questionnaire provided at the end of each workshop (one orientation visit and three workshops, attended over a 20 week period) to evaluate participants overall satisfaction (88). Evaluation at level 2 measured the impact of the program on participants’ patient safety knowledge score (multiple choice quiz evaluation) and skill (on the job performance), pre intervention and at the end of the 20 week period. Evidence at level 3 and 4 was collected to assess changes in practice. Both quantitative and qualitative data was collected during this evaluation. Overall, this program was positively received by participants and a significant improvement was seen in knowledge score (although not all participants completed the post- intervention knowledge evaluation). Eight of the 11 teams showed improvements in patient safety practices (88).

An example of application of the KM to nutrition education evaluation or dietary behaviour change assessment, Shergill et al (2009) evaluated heart heathy dietary behaviours of clients participating in St Michael’s Hospital’s Eat Well, Live Well (unpublished data). Eat Well, Live Well was a 60 minute, RD-facilitated, nutrition group education session, aimed at educating in- 60 clients on heart heathy dietary behaviours. Three levels of KM were used to evaluate the education provided during the session. Participants (n = 43) completed surveys measuring satisfaction (Level 1) post-education, knowledge (Level 2) pre- and post- education, and 10 heart heathy dietary behaviours (Level 3; evaluated against the Transtheoretical Model or Prochaska’s Stages of Change Model) before, 1 month following and 6 month following the session (84,216-218,363). This survey was adapted from Graham et al. (2008) to reflect the National Cholesterol Education Program and Canadian Cardiovascular Society guidelines (and session content) and was face-validated in a professional sample of eight allied health professionals (n = 8) (364). Eighty nine percent (89 %) of Eat Well, Live Well class participants rated the session as good or greater on a Likert scale comprised of the following responses: Poor, satisfactory, no response, good and excellent. Mean knowledge score was measured as self-perceived knowledge before and one month after the Eat Well, Live Well session, rather than a direct measure of knowledge (e.g. testing). Knowledge score was significantly higher at one month post-session in comparison to before the session (mean difference: 1.52; p ≤ 0.001). Change in stage of change score indicated that there was a significant progression along the stages of change continuum in this sample (84).

The recognition that patient engagement, satisfaction, and context should be considered in the development and evaluation of RCT nutrition interventions is a relatively new concept in GI research in Canada. This is especially interesting in light of the fact that KM level 1 (Satisfaction/ Reactions) is the most commonly completed KM level seen in corporate settings (80-83,214,343). As part of this review, GI studies were compared to the KM model and it was established that most GI interventions measure clinical results (KM level 4; e.g. clinically relevant change in HbA1c, body weight). There are very few GI intervention studies that include education evaluation components and those reviewed did not evaluate all KM levels (38,206,222). Notwithstanding, one GI study evaluated KM level 1. Vermeulen and Turnball (2000) collected data on participants “comments and opinions” on a GI education tool using an open-end question format (222). KM level 1 data (reactions) (e.g. comments and opinions) were collected from a professional sample (n = 21) comprised of RDs, Registered Nurses (RNs), and Diabetologists and a patient/client sample (n = 18) living with DM. A direct pre- and (immediately) post- education evaluation of knowledge (KM level 2) was also conducted in both samples; using testing (knowledge score). There was a significant improvement in 61 knowledge score in both samples. KM level 3 and 4 did not appear to be evaluated in this study (or the evaluation details were not included in this publication) (222).

KM level 2 to 4 were evaluated by Burani and Longo (2006). A key limitation of this study was that it was a retroactive chart review; limiting knowledge assessment to post-education assessment and introducing baseline knowledge as a potential confounder. Also a limitation, behaviour change assessment was limited to client reported measures of behaviour change, which were primarily measures of client efficacy or intention to make change rather than direct metrics of change (e.g. smoking cessation, decreased intake of dietary fat). Despite these methodological limitations, participants (n=21) had an average knowledge score of 86.2% (no scores less than 60%) and reported that they believed they would continue to use the GI concept as a permanent lifestyle change (unanimous agreement) (38).

Davis and Miller (2006), used an observational design, focus groups and a 9-item knowledge questionnaire to conduct a comprehensive GI education needs assessment in a group of participants (n = 44), aged 21 to 65, living with T2DM. An example of formative and community-based research, this study was developed to inform subsequent education development initiatives overseen by this research group. Although this exact sample was not followed up to receive education/ intervention and post-intervention knowledge assessment, these data informed a number of GI-education-based interventional studies conducted by Miller (2006-present) (55,143,206,207,365,366). Moreover, this study offers an example of how formative data collection plays a key role in sample needs assessment and education/ trial development. These data often warrant pre-intervention dissemination (often in the form of a methods paper) and are in line with current Canadian KT dissemination guidelines (89, 90).

2.3.0. LITERATURE REVIEW SUMMARY AND DISSERTATION CONCEPTUAL FRAMEWORK

The CDA CPG currently supports the integration of GI into medical nutrition therapy for management of glycaemic control in people living with T1DM and T2DM (144). There is not sufficient evidence available to create similar guidelines for management of GDM (14,24,40,64- 72). Despite the existing support for use of GI in management of T1DM and T2DM, 61% Canadian RDs, working in DM management, are not integrating GI into practice (62). The following barriers to GI utility have been identified by Canadian RDs and other nutrition educators:

By identifying these four perceived barriers, we identified four research gaps (or opportunities) and highlighted the importance of the RD (nutrition educator) in translating GI from bench to bedside. Perceived barrier 1 is supported by the literature. That is, evidence available on GI utility in GDM is not sufficient to support integration of GI education into standard care and the physiological/ metabolic mechanisms by which GI brings about favourable effects (or displays utility) in clients at risk of or living with DM are not fully understood (24,26,40,62- 72,78,79,137). The second perceived barrier is valid. The literature reviewed indicates that there is a deficiency of “reliable” or “suitable” GI education reference materials and job aids in Canada (e.g. handouts, backgrounders). Review and revision of existing GI education tools or development of new GI education tools for clinicians and clients is warranted, but it is clear that the input of clinicians and clients will be invaluable during this process (26,62,63,133,137). Published data do not support perceived barriers three and four. There are not sufficient data to disprove them, however, and they too warrant examination (26,48,55,56,62,63,125,143,173,196, 200,207,222,365,366). Overall, the literature supports that additional evaluation of GI education from the perspective of the educator and the client is needed and there are unmapped facets of 63 the GI mechanism that require attention. Efforts to translate current knowledge into practice, while developing practice-based research that brings us from bench to bedside (and back again) are supportive of the translational paradigms presented in this chapter and will ensure that educators are informed and able to integrate findings into their practice efficiently and confidently.

A proposed GI mechanism of interest to the clinical/ scientific community is the effect of low GI food on postprandial oxidative stress (and antioxidant capacity). Current evidence suggests that a low GI diet may decrease postprandial oxidative stress by decreasing postprandial glycaemic fluctuations. Acute and chronic oxidative stress has been implicated in the development of DM and DM complications (e.g. CVD). Scientists believe that the key physiological mechanism driving the (proposed) GI-oxidative stress relationship is likely related to sustained-release or "lente" carbohydrate (73,78,225,232,247). However, the main study published looking at low GI and postprandial oxidative stress is a food-based study and therefore has inherent dietary confounders (78). The sip-bolus paradigm, developed by Jenkins et al. 1990, offers a design and methodology that isolates the slow absorption model and provides the opportunity to increase our understanding of sustained release carbohydrate on oxidative stress, while addressing educators need for more information on mechanism and GI utility. Examination of this relationship is timely, as there have recently been successful translational efforts exerted by Canadian RDs to support RDs in client education on dietary antioxidants and functional foods that increase antioxidant capacity in the human model (183).

Efforts to develop comprehensive approaches to evaluating education-based and/ or behaviour change-based interventions have led to development of phrases like “realist RCT”, which have become pervasive in clinical trials literature (192,214,343). Using education evaluation frameworks like KM to evaluate nutrition interventions (education), in the context of the RCT, is a comprehensive approach to assessing the effect of diet on the primary outcome. This model provides the opportunity to examine the various covariates that influence the relationship between the intervention and primary outcome (80-83,85,192,214,343). Moreover, nutrition interventions (RCTs) provide the opportunity to evaluate education (e.g. feasibility, participants’ satisfaction) and methodology, while conducting research that spans translational stages or laboratories (figure 2.3.) and disciplines (95,177,192,214,343). 64

Three studies were created to provide a comprehensive assessment of GI utility from bench to bedside, while addressing the gaps (i.e. perceived barriers, published limitations, and questions) identified during the literature review. An overarching research perspective served as the backbone of this work. This perspective (rooted in practice-based research) is that by addressing clinicians’ and scientists’ perceived barriers to GI utility and evaluating an evidence-based GI education platform in two client samples, we will ultimately increase GI knowledge translation. An overarching goal of this work is to address gaps in the literature highlighted by educators; who are often responsible for translating knowledge to the end user (the client). The six objectives of this dissertation, created to achieve this overarching goal are to:

1. Determine the effect of slowing carbohydrate delivery on postprandial oxidative stress 2. Develop an evidence-based GI education platform, designed to be layered onto standard care for people living with type 2 diabetes mellitus or gestational diabetes mellitus. 3. Develop and pre-test an evidence-based GI education evaluation questionnaire (GIQ) in people living with type 2 diabetes mellitus. 4. Use the GIQ and a three day diet record to evaluate the GI education platform in people living with type 2 diabetes mellitus. 5. Use the pre-tested GIQ and a three day diet record to evaluate the GI education platform in people living with gestational diabetes mellitus. 6. Evaluate whether the pre-tested GI education platform improves glycaemic control in people living with gestational diabetes mellitus using a randomized control design.

Figure 2.7. illustrates the unifying conceptual framework upon which to layer the aforementioned goal and objectives and to facilitate understanding of how the individual study protocols support these objectives. Our perspective is that by addressing clinicians perceived barriers to GI utility, we can continue to generate novel data on GI mechanism (study 1) while facilitating translation of knowledge to the client by providing clinicians with sufficient support (job aids) and information upon which to base their professional choice of “to use of not to use” GI in practice (study 2 and 3). Figure 2.7. Illustrates the two laboratories (bench and bedside), described in the NIH Roadmap (figure 2.3) (95,177). The bench laboratory (enclosed by circle 1) is defined by basic/ pre-clinical or mechanism-focused research (lab [L] 1), while the bedside laboratory (circle 2) is interested in intervention evaluation; including effectiveness, side effects, 65 and safety (not an exhaustive list) (lab [L] 2). The overarching aim of L2 is to assess whether or not an intervention should become considered for integration into standard care or practice. One can look at L1 as representing study 1, while L2 represents studies 2 and 3, but there is certainly conceptual overlap between these labs/ studies.

The two-way transparent horizontal arrow upon which the terms “bench” and” bedside” are transcribed, imply that the path from bench (mechanism) to bedside (intervention evaluation) is not one-way. That is, bench research influences bedside research and vice versa. The circles overlap in the centre of the figure, highlighting that GI utility is a unifying concept between mechanism and intervention evaluation. GI mechanisms (L1) are concerned with how GI works (e.g. “the slow absorption model” [physiological mechanism] or the “second meal effect” [metabolic mechanism]) (29,79,132,159,160,322,330,367). GI utility is concerned with whether or not it works (Is it effective in changing or predicting a given outcome?). GI interventions (L2) look at whether or not utility translates to various clinical settings and clients and if it is maintained when delivered as education (28,40,62,63,133,144).

L1 provides the opportunity to examine mechanisms by which GI effects a given outcome (novel markers of utility) in isolation of covariates or common confounders (e.g. dietary antioxidants), as represented by the three small blue circles to the left of the Venn diagram. On the other hand, L2 provides the opportunity to comprehensively examine GI utility in the context of an education platform in various clinical settings and client groups. Rather than isolating confounders, they are measured and considered in assessment of GI utility (e.g. patient satisfaction with the education and food, behaviour change). This is illustrated by the three small blue circles that overlap the right circle (“Clinical Intervention”) of the Venn diagram.

GI utility is bordered by arrows (>), to further stress that L2 research informs L1 research and vice versa and that it is an ongoing process. Important to note, however, is that L1 is concerned with examining novel markers of GI utility or mechanisms that are not yet understood, while the primary outcome of an L2 study is typically an established (evidence-based) marker of GI utility (e.g. glycaemic control). Surrounding the Venn diagram is a broken (line) circle with arrows (>) intersecting it. This circle illustrates that the research included in this dissertation has been guided by a practice-based approach (clinicians’ and scientists’ perceived barriers have guided a 66 client-centred research process) and is an ongoing process where practice informs research and research informs practice. In this dissertation the question of GI utility will be examined, while keeping in mind the importance of coming back to the bench to examine new GI mechanisms and test new perspectives on utility (e.g. ability to reduce oxidative stress or arterial stiffness) and involving the knowledge users in this process.

Figure 2.7. Conceptual Framework for Dissertation

Practice-based Approach

Covariate

Covariate Covariate Bench (L1)  Bedside (L2)

GI Clinical Covariate Covariate Mechanism Utility Intervention

Covariate

L = Laboratory; GI = Glycaemic Index; > = arrows representing ongoing or cyclical effort; Two way arrows indicate exchange. This diagram provides a unifying conceptual framework for the work presented in this dissertation. All studies completed are practice- based. Study 1 examines a novel GI-mechanism and potential marker of GI utility to address educators expressed need for more information on GI mechanism (designed to eliminate common confounders). Study 2 and 3 evaluate a GI education platform based upon an existing understanding of GI mechanism and marker of GI utility used in DM medical therapy (designed to measure common confounders). 68

2.4.0. STUDY-SPECIFIC RESEARCH HYPOTHESES AND PRIMARY OBJECTIVES: AN OVERVIEW

Three studies were designed to address our overarching practice-based research goal. Below are their titles, specific research hypotheses and primary objectives.

STUDY #1: The Effect of Continuous Sipping of a Dextrose Solution on Markers of Oxidation in Men and Women

General hypothesis: Reducing postprandial glucose excursions by reducing the rate of carbohydrate absorption will result in less postprandial oxidative stress.

Specific hypotheses:

1. Primary hypothesis: Sipping a 75 gram dextrose solution slowly over 3.5 hours will result in less oxidative stress than ingesting the same amount of dextrose as a bolus over 5 minutes. 2. Sipping 75 gram dextrose solution will reduce oxidative stress to the same extent as 1 gram of oral vitamin C. 3. The effect of sipping a 75 gram dextrose solution on oxidative stress will occur sooner than that of vitamin C.

Primary objective: To compare the effect of sipping dextrose solution slowly (over 3.5 hours) to consuming it as a bolus (over 5 minutes) on plasma total peroxyl radical trapping antioxidant potential (TRAP) in participants classified as overweight or obese.

STUDY #2: Evaluation of Glycaemic Index Education in People Living with Type 2 Diabetes Mellitus: Participant Satisfaction, Knowledge Uptake and Application

Primary hypothesis: A low GI education platform will reduce dietary GI in men and women living with T2DM (KM level 3).

Supporting Hypotheses: Study participants will:

1. be satisfied with the GI education (KM level 1) 2. show an increase in GI knowledge post-education (KM level 2)

Primary Objective: To evaluate if the low GI education platform can significantly reduce dietary GI in participants with T2DM post-education.

STUDY #3: The Effect of a Low Glycaemic Index Diet on Maternal and Neonatal Markers of Glycaemic Control and Postpartum Diabetes Risk

Primary Hypothesis: A low GI education platform will improve postprandial glycaemic control in women with GDM more than standard care (KM level 4).

Supporting hypotheses: Study participants will: a) be satisfied with the GI education (KM level 1) b) show an increase in GI knowledge post-education (KM level 2) and c) significantly lower their dietary GI post-intervention (KM level 3)

Primary Objective: was to determine whether women with GDM who received low GI education would obtain a higher percentage of postprandial self-monitored blood glucose within the clinically acceptable range than the control group (those receiving standard care medical nutrition therapy) post-education. The CDA CPG (2013) reference ranges were used for analysis (< 6.7 mmol/L).

CHAPTER 3.0. STUDY #1 THE EFFECT OF CONTINUOUS SIPPING OF A DEXTROSE SOLUTION ON MARKERS OF OXIDATION IN MEN AND WOMEN (Short Title: AOGI)

Components of this chapter (3.0.) have been previously presented and published: Grant, S., Josse, R., Barre, E, Wolever, T. (2014) The effect of continuous sipping of a glucose solution on markers of oxidation in men and women. Canadian Nutrition Society Annual Meeting 2014; St. John's, Newfoundland, Canada. ABSTRACT: Applied Physiology, Nutrition, and Metabolism, 2014, 39(5): 605-642.

3.0. STUDY #1 THE EFFECT OF CONTINUOUS SIPPING OF A DEXTROSE SOLUTION ON MARKERS OF OXIDATION IN MEN AND WOMEN

3.1. ABSTRACT

It is thought that low glycaemic index (LGI) foods reduce postprandial oxidative stress (OS) by reducing postprandial glucose excursions. Evidence supporting this hypothesis is limited by lifestyle confounders. To eliminate these confounders and to determine whether reducing postprandial glucose per se reduces postprandial OS, overweight participants (n=18) were administered four treatments: 75g dextrose solution consumed within 5min (Bolus), Bolus plus 1g vitamin-C (BolusC), 75g dextrose sipped slowly over 3hr (Sip), and Sip plus 1g vitamin-C (SipC). Blood was taken at intervals of 360min with a standard lunch at 240 min. Sip significantly flattened the blood glucose and insulin curves and reduced free-fatty acid rebound compared to Bolus (p < 0.05). TRAP (primary outcome) was significantly affected by time, rate of administration, and whether vitamin-C was administered (p < 0.05). There was no significant change in TRAP from baseline after Sip and SipC over 360 min. Mean TRAP increment was significantly higher after SipC and BolusC than B at 300 min (SipC: 81.7 ± 31.1 and BolusC: 58.0 ± 31.1 versus Bolus:-50.9 ± 31.1 µmol/L; p ≤ 0.016). By 360 min, TRAP was significantly lower after Bolus than after BolusC, Sip, and SipC, respectively (-153.1 ± 32.6 versus 10.4 ± 32.6, 7.9 ± 32.6, and 105.9 ± 32.6 µmol/L; p<0.005). Neither rate nor vitamin-C significantly influenced CRP, LDL oxidation and conjugated dienes. Augmentation index was lower after Bolus than Sip (Bolus: -6.31 ± 0.58 versus Sip: -3.48 ± 0.57 %; p=0.001) and blood pressure was lower when vitamin-C was administered. These results support the hypothesis that reducing postprandial glucose reduces postprandial OS, and, thus, provides insight into the mechanism by which LGI diets may influence health outcomes.

Funded by: Canadian Institutes for Health Research Clinicaltrials.gov Identifier: NCT01440790

3.2.0. INTRODUCTION

3.2.1. Study Rationale

As highlighted in the literature review, a link has been established between acute hyper- metabolic responses of the postprandial state (e.g. hyperglycaemia, hyperlipidaemia) and chronic disease development and progression (73,225,231,239,243,247,248,325,368-370). Low GI foods have been associated with decreased risk of chronic disease (e.g. DM, CVD) and favourable treatment outcomes (e.g. improved lipid profile, β-cell function). These favourable effects of low GI food have been attributed, in part, to decreased postprandial blood glucose response, but research is ongoing into the mechanisms at play (29,42,78,132,268,338,339). Recent data suggest that a low GI dietary pattern can decrease markers of oxidative stress and postprandial oxidative burst. The physiological mechanism thought to be responsible for this effect is the delayed/ slowed carbohydrate absorption. Botero et al (2009) examined this potential relationship between GI and oxidation, however, this study was limited by dietary/ lifestyle confounders (did not isolate the specific GI mechanism). AOGI has been developed to address these methodological gaps and to add to understanding of the role low GI/ slowly absorbed carbohydrates may play in oxidative stress and the metabolic cascade that ultimately increases/ decreases risk for DM and CVD.

3.2.2. Hypotheses and Primary Objective

General hypothesis: Reducing postprandial glucose excursions by reducing rate of carbohydrate absorption will result in less postprandial oxidative stress.

Specific hypotheses:

1. Sipping a 75 gram dextrose solution slowly over 3.5 h will result in less oxidative stress than ingesting the same amount of dextrose as a bolus over 5 min.

2. Sipping a 75 gram dextrose solution will reduce oxidative stress to the same extent as 1 gram of oral vitamin C.

3. The effect of sipping a 75 gram dextrose solution on oxidative stress will occur sooner than that of oral vitamin C.

Primary Objective: To compare the effect of sipping dextrose solution slowly (over 3.5 hours) to consuming it as a bolus (over 5 minutes) on plasma total peroxyl radical trapping antioxidant potential (TRAP) in participants classified as overweight or obese.

3.3.0. MATERIALS AND METHODS

3.3.1. Study Design

This study was a randomized cross-over (2*2 factorial) design. The clinical trial portion of this study was conducted at GI Labs, 20 Victoria Street, 3rd Floor, Toronto, Ontario, Canada. The study protocol, including all proposed study procedures, was reviewed and approved by the University of Toronto Research Ethics Board. Trial details were also reviewed, approved and published by clinicaltrials.gov (Identifier: NCT01440790).

3.3.2.0. Sample

Study participants were recruited from GI Labs and the Department of Nutritional Sciences, University of Toronto, using posters and snowball/ referral sampling. Individuals who self- referred to GI Labs were screened by the GI Labs Clinic Manager, after which they were invited to review the study consent form. After written, informed consent was obtained, study participants completed a staff-administered questionnaire (screening questionnaire) and blood sample to ensure all eligibility criteria were met. The eligibility criteria were comprised of the inclusion and exclusion criteria listed below.

Inclusion Criteria 1. Men and women 2. 18 to 65 years of age 3. Overweight or obese class 1 to 2 (BMI = 25 to 39.9) . Participants with a BMI in the overweight and obese categories were selected, as this is a modifiable risk factor for T2DM and CVD (112,113) 4. Willing and able to comply with the study protocol

Exclusion Criteria

Individuals who: 1. had a condition effecting carbohydrate metabolism (e.g. DM, CVD, kidney disease, liver disease etc.) 2. smoked cigarettes 3. exceeded current alcohol intake recommendations > 1 time per week . It is recommended women maintain their average consumption between 0 to 2 standard drinks per day (no more than 10 per week) and for men to maintain their average consumption between 0 to 3 standard drinks per day (no more than 15 per week). (371) 4. consumed antioxidant supplements regularly* 5. lost ≥ 10 pounds one month prior to screening 6. made relevant changes to physical activity regimen one month prior to screening*8

The Screening Lifestyle Questionnaire (appendix 3.1.) was developed to ensure participants met eligibility criteria and to collect baseline data on modifiable lifestyle factors that have been associated with chronic disease risk and progression and antioxidant status (e.g. activity frequency and intensity, medication). This questionnaire was administered to participants the same day as the aforementioned fasting blood sample. This blood sample was used to measure the following biochemical outcomes: Aspartate transaminase (AST), alanine transaminase

8 Those exclusion criteria marked with * were assessed for exclusion on a case-by-case basis by a Physician.

(ALT), Creatinine (Cr), full lipid profile, thyroid stimulating hormone (TSH), insulin and glucose. Analysis of samples was completed in St Michael’s Hospital Diagnostic Laboratories.

3.3.2.1. Sample Size Calculation

This study is novel from a methodological stand-point and a pilot study. Botero et al. (2009) (n = 12) was the study that was ultimately selected for sample size calculation. A sample size of 14 was calculated using the mean within participant standard deviation (1166 μmol TE/l) and minimal detectable difference between mean plasma ORAC (1355 μmol TE/l). The probability was 80% that this crossover study would detect a treatment difference at a two-sided sided 0.05 significance level. To accommodate for 20% dropout, the sample size was set at 18 participants. Participants were recruited and active in the study from October 2010 to May 2011.

3.3.3. Study Outcomes

Study outcomes were measured at each study visit according to a standard data collection/ administration schedule; table 3.1. The primary outcome was Total peroxyl radical trapping antioxidant potential (TRAP) of human plasma. Secondary outcomes included: Plasma conjugated dienes (CD), plasma low density lipoprotein oxidation (LDLox), plasma vitamin C, plasma glucose, plasma insulin, serum free fatty acids (FFA), serum total cholesterol, serum high density lipoprotein (HDL), serum low density lipoprotein (LDL), serum triglycerides (TG), serum C-reactive protein (CRP), blood pressure (BP), Pulse rate (PR)/ Heart rate, Augmentation Index (AI), Pulse Pressure (PP; difference between the systolic and diastolic BP readings), change in dietary intake (and other lifestyle outcomes; e.g. smoking, physical activity), anthropometric data (including mass/ weight, height, waist circumference), symptoms/ side effects, change in medication.

Table 3.1. Study Outcome Collection Schedule at Each Study Visit (4 visits in total) Outcome 0 min 30 min 60 min 120 180 240 270 300 360 min min min min min min TRAP X X X X X X CD X X X X X X LDLox X X X X X X Vitamin C X X X X X X Plasma Glucose X X X X X X X X X Insulin X X X X X X X X X FFA X X X X X X X Lipids X X X X X CRP X X X X Vitals X X X X X X 24 Recall X Symptoms X X X Lifestyle Questionnaire X

TRAP = Total peroxyl radical trapping antioxidant potential; CD = Baseline; Conjugated Diene; LDLox = LDL Oxidation; FFA = free; fatty acid, CRP = C-reactive protein; LQ = Lifestyle Questionnaire.

3.3.4. Study Treatments and Randomization

On four separate occasions, after a 12 hour overnight fast, participants attended a 6.5 hour (390 minute) study visit at GI Labs. At each study visit a different test food/ treatment was administered; visits were identical otherwise. The four treatments included a solution of 75 grams of dextrose dissolved in 250 grams/ml of water (for a total serving size of 325 grams of solution), but varied in terms of rate of consumption and whether or not a 1 gram vitamin C supplement was present. The treatments are listed below:

1. Solution consumed as a bolus (within 5 minutes) 2. Solution consumed as bolus + 1 gram of oral vitamin C 3. Solution consumed as sipping (over 3.5 hours; approximately 93 g/ 1 hour) 4. Solution consumed as sipping + 1 gram of oral vitamin C

Participants were administered the solutions in random order. The order of treatments was created using the “@rand” command in Lotus 1-2-3 1997 edition (Lotus Development Corporation, Cambridge, MA) and assigned to participants in the order they consented. Treatments were administered in 8 fluid ounce glasses. For sipping treatments, the glass was measured and marked with black marker to indicate the volume to be consumed at each sip over the consumption period.

At 240 minutes post-treatment administration, a standard lunch was served to all participants at each visit. This second meal was added so that the effect of treatment on participants’ response to the second meal could be measured. The standard lunch recipe and ingredient weight is outlined below (table 3.2.). All participants were asked to finish this meal by 20 to 25 minutes. The nutrition information for this meal is presented in table 3.3. Tea and coffee were not offered during this study, despite the findings of Aldughpassi and Wolever (2009), since tea and coffee are considered noteworthy sources of dietary antioxidants and have been noted as potential confounders in acute measurement of pulse wave outcomes and postprandial blood glucose. Aldughpassi and Wolever showed tea and coffee does not affect the mean GI value obtained, but may reduce variability and, hence, improve precision (372).

Table 3.2. Standard Lunch Recipe and Ingredient Mass

Recipe Mass (grams) 2 slices of whole wheat Dempsters Bread© Canada Bread Company, 70 Limited 2014 2 tablespoons of Hellmann’s® Real Mayonnaise 28 1.5 leaves of iceberg lettuce 10 to12 1/3 cup of peeled cucumber 40 1 * 8 fluid ounce glass of 2% milk 243 to 245 1 Cheese Strings Ficello™ Parmalat Canada Incorporated 21 I medium granny smith apple with peel 137

Table 3.3. Energy and nutrient composition of standard lunch meal

Food # Wt Cals Carb Fat Protein GI Vit C (g) (kcals) (g) (g) (g) (%) (mg) Meal 554 640 64.0 35.0 21.0 53 8.45 Bread 71 170 32.0 2.0 7.0 71 0.00 Mayonnaise 28 200 0.0 2.2 0.0 NA 0.00 Lettuce 12 1.68 0.4 0.0 0.1 38 0.34 Cucumber 40 5 0.9 0.1 0.2 38 1.27 Milk 244 122 11.0 4.8 8.1 32 0.49 Cheese 21 70 0.0 6.0 5.0 31 0.49 Apple 138 72 19.1 0.2 0.4 37 6.36 Wt = weight, Cals = calories, Carb = carbohydrate, g = grams, GI = glycaemic index, Vit C = vitamin C, mg = milligrams.

3.3.5.0. Biochemical Outcomes

The following is a complete list biochemical outcomes (with units of measure):

1. Plasma TRAP (μmol/L) 2. Plasma CD (μmol/ μmol protein) 3. Plasma LDLox (U/L) 4. Plasma vitamin C (μmol/L) 5. Plasma glucose (mmol/L) 6. Plasma insulin (mU/ml) 7. Serum FFA (mEQ/L) 8. Serum Total Cholesterol (mmol/L) 9. Serum HDL (mmol/L) 10. Serum LDL (mmol/L) (Calculated) 11. Serum TG (mmol/L) 12. Serum CRP (mg/L)

3.3.5.1. Blood Sample Collection and Processing

Biochemical outcomes were measured using blood samples collected by a RN. Blood was drawn from either the right or left arm median cubital vein using a peripheral venous catheter. The cannula was left in the arm from baseline until the end of the visit; saline solution was used to flush the apparatus intermittently to prevent interruption of draw (i.e. clotting/ blockage/ participant dissatisfaction). Participants were given heating pads, with which they wrapped their arm between blood draws, that they were asked to keep on low heat. Decannulation occurred between 360 and 370 minutes. Participants were monitored for 20 to 30 minutes after decannulation by the RN until they were cleared to leave the clinic.

Blood collection procedures were outcome dependent. Three BD Vacutainer® Blood Collection Tubes were used to collect blood samples throughout the study and are listed in table 3.4. Collection procedures were followed as per publically available BD standard operating guidelines. This table also notes the laboratory at which analysis was conducted for each outcome.

Table 3.4. Blood Collection Tube Specifications, Study Outcomes and Destination

Tube Volume Additive Outcome(s) Destination Lab for (ml) Analysis 1. Gold 6 Clot activator/ CRP Diagnostic Laboratories, gel for serum St Michael’s Hospital separation (Core Lab) FFA J. Alick Little Lipid Lipids Research Laboratory, St. Michael's Hospital Blood glucose GI Labs Insulin 2. Green 5 Heparin for Vitamin C Hospital in Common clot prevention Laboratories or HICL and plasma determinations 3. Lavender 10 Spray-coat K2 TRAP Barre Laboratory, Cape EDTA for clot CD Breton University prevention LDLox

FFA = free fatty acid, CRP = C-reactive protein; TRAP = Total peroxyl radical trapping antioxidant potential; CD = Baseline Conjugated Diene; LDLox = LDL Oxidation

Samples collected in the gold top tube were spun at 3000 rpm at 4°C for 15 minutes and lavender top was spun at 2500 rpm at 4°C for 15 min using the Beckman GPR Centrifuge (USA; Serial # OF127) as per respective Laboratory Manager/ Technician specifications. Green top tubes were spun at 2500 rpm at room temperature for 10 min as per Hospital in Common Laboratory (HICL) specifications. Green tops from the first 2 to 4 visits were processed by St Michael’s Hospital (same processing procedure), but after that samples (green top tubes) were processed at GI Labs using Beckman GPR Centrifuge (USA; Serial # OF127) and shipped directly to HICL (now In-Common Laboratories). Serum from gold top tubes was transferred into 2 ml microtubes (specification varying by lab and outcome; e.g. snap top versus screw top, skirted versus conical). Plasma from both the green top and violet top were transferred into 2 ml amber (photo-protective) cryogenic screw-top skirted microtubes. CRP was analyzed fresh at St Michael’s Hospital, while all other outcomes were store at -80 °C until analysis (at locations outlined in table 3.4.). Table 3.5. outlines the blood sample collection timeline, by biochemical outcome for, each of the four visits.

Table 3.5. Blood Sample Collection Timeline by Biochemical Outcome

Assay 0 min 30 min 60 min 120 180 240 270 300 360

min min min min min min

Glucose X X X X X X X X X

Insulin X X X X X X X X X

Lipids X X X X X FFA X X X X X X X CRP X X X X Vitamin C X X X X X X TRAP X X X X X X CD X X X X X X LDLox X X X X X X

FFA = free fatty acid, CRP = C-reactive protein; TRAP = Total peroxyl radical trapping antioxidant potential; CD = Baseline Conjugated Diene; LDLox = LDL Oxidation

3.3.5.2. Biochemical Analysis9

TRAP, LDLox, and CD analysis was conducted at Cape Breton University, Department of Health Sciences and Emergency Management, School of Professional Studies under the supervision of Dr Edward Barre. TRAP is used widely to assess total resistance capacity to oxidation in human plasma/ serum. The TRAP assay used to analyze AOGI samples was based on the spectrophotometric assay developed by Valkonen and Kuusi (1997) to measure TRAP in serum/ EDTA plasma (287). For this study, this assay was adapted for analysis using a microplate reader (Specramax 190, Molecular Devices, Sunnyvale, California). The complete TRAP assay can be found in appendix 3.2. Coefficients of variation (CV) were measured for duplicate lag times of each of plasma and Trolox. A CV less than 10 % was acceptable. If the CV was not met, samples were rerun.

The materials and methods implemented for the LDLox assay, were identical to those outlined in the “Mercodia Oxidized LDL ELISA Directions for Use” (Mercodia Oxidized LDL ELISA kit [Mercodia AB, Sylveniusgatan, BA, Sweden]), save the following three exception(s): (1.) Dilution 1:15 μl of plasma was mixed with 1200 μls sample buffer. Initial dilution = 1/81, (2.) Dilution 2:15 μl of initial dilution was mixed with 1200 μls of sample. Final dilution = 1/6561, (3.) If LDLox values produced by the microplate reader were beyond the range of the standard curve, the plasma was diluted to 1:4 (10 μl of plasma plus 30 μl of Ultra high purity water [UHP]) or 1:6 (10 μls of plasma plus 50 μls of UHP water). The above dilutions were repeated immediately (263).

CDs were measured according to heparin citrate precipitation (referred to as “Baseline Conjugated Dienes”), as described by Ahotupa et al. (1996, 1998) (for protocol/ assay details, see appendix 3.3.) (260,289). Analysis was also conducted to quantify lipid peroxidation, recognizing there is one molecule of apoB-100/ LDL particle (373). This is to say, calculating apoB-100 concentration allows determination of LDL particle number. CD concentrations may increase, decrease or not change. Therefore, it is important to determine the CD concentration/ LDL particle so that it can be determined if changes in CD are due a shift in LDL particle

9 Appendix contains complete assays for biochemical outcomes; Appendix 3.2 to 3.6.

82 number or CD concentration per LDL particle. The complete procedure for apoB- 100 quantification is included in appendix 3.4.

Participant plasma was analyzed for vitamin C concentration at London Health Sciences Centre and managed by HICL (processing details above) using the time-tested adapted methodology of Wagner at al. (1979). Internal lab CV was 4 to 6% during the time of the analysis. The following details were provided by the Team Leader at HICL (as sequential steps):

1. 20% sulphosalicylic acid was added to an aliquote of the specimen (solution)

2. The solution was vortex mixed thoroughly and centrifuged at g-force of 1000g to 1300g (achieved at 2500-2900 rpm at centrifuge radius of 13.5cm) at room temperature for 10 min.

3. An aliquot of the clear supernatant was then mixed with the high performance liquid chromatography (HPLC) mobile phase for analysis. The HPLC was equipped with an electrochemical detector and uses a 15 cm * 4.6 mm C18 column with a mobile phase of 0.8 % metaphophoric acid (374).

Plasma glucose and insulin analysis was conducted at GI Labs. Glucose was measured with the YSI Model 2300 STAT glucose analyzer (based on an enzymatic method); analytical CV: 1.5 to 2.0%. Insulin was measured using an ELISA kit (ALPCO Insulin ELISA; Salem, NH; Catalog Number: 80-INSHU-E01.1, E10.1); analytical CV (4.86 uU/mL [29 pmol/L]): 10.3%.

FFA, total cholesterol, HDL, LDL, TG results were obtained via the J. Alick Little Lipid Research Laboratory, St. Michael's Hospital. FFA were measured non-esterified fatty acids (NEFA) - HR (2) Assay (Wako Diagnostics, Richmond, VA) and the lipid profile was analyzed via Roche/Hitachi Cobas C Systems (Cobas® 6000). Assay Standard Operating Procedures for J. Alick Little Lipid Research Laboratory are found in appendix 3.5. and 3.6.

CRP analysis was conducted at Diagnostic Laboratories, St Michael’s Hospital. High Sensitivity Cardiac CRP reagent is based on the highly sensitive Near Infrared Particle Immunoassay rate methodology. Anti-CRP antibody-coated particle binds to CRP in the sample resulting in the formation of insoluble aggregates causing turbidity. The Beckman Coulter SYNCHRON LX20

System automatically proportions one part sample to 26 parts reagent into a cuvette and monitors change in absorbance at 940 nm. The change in absorbance is proportional to the concentration of CRP based upon a single-point adjusted pre-determined calibration curve. Date-specific details regarding implementation of this assay/ procedure can be obtained via Diagnostic Laboratories, St Michael’s Hospital and the Beckman Coulter SYNCHRON Systems CRPH High Sensitivity C-Reactive Protein Chemistry information sheet A18486 AB, September 2005.

3.3.6. Cardiovascular Hemodynamics Outcomes (or “Vitals”)

The following is a complete list of traditional and non-traditional markers of cardiovascular risk; referred to as “Cardiovascular Hemodynamics Outcomes” or “Vitals” during the study (with units of measure):

1. DBP (mmHg) and SBP (mmHg) 2. AI (%) 3. AI value (%) normalized to a PP of 75 (AIp75)

4. PR (BPM) using applanation tonometry sensor (ATS) = PRAI

5. PR (BPM) using BP cuff = PRBP 6. PP (mmHg)

The Omeron Non-invasive Blood Pressure Monitor with AI (HEM-9000AI, Omeron Healthcare Professional Services) was used to measure participants’ BP and calculate participants’ AI. Adhering to in-lab standard operating procedures based on Jovanovski et al. (2010), the HEM- 9000AI Instruction Manual and Van Bortel et al. (2002), BP was obtained via digital oscillometric method using a BP cuff (295,316). The AI calculation was based on several components of a participants’ pulse wave, which was obtained via applanation tonometry (AT) using a sensor placed against the radial artery. A supplementary transfer function was not applied to the data collected for this study (295,296,300,301,316). Vitals measurements were conducted in a quiet, temperature controlled room by one of two trained clinical scientists at baseline, 60, 120, 240, 300, and 360 minutes.

Participants were seated in an upright position so that they could rest their elbow comfortably on a table set to height just below their sternum after a 15 minute resting period during BP measurement. BP, DBP, PR and PP were obtained, in duplicate, from the left arm. AI was also measured, in duplicate, at baseline, 60, 120, 240, 300, and 360 minutes (± 5 to 10 minutes) immediately after BP measurement was obtained. One location was (marked with black marker using an “X”) used to take each AI measurement (in duplicate). The average value of duplicate measurements was recorded/ reported. Raw and normalized (for a heart rate of 75 beats per min) radial AI is provided by the device. AI measurements were not collected for one participant because of inadequate vessel compression (due to wrist fat). Figure 3.1. illustrates how to apply the pulse wave sensor unit.

Arterial diameter and stiffness show a diurnal variation, resulting in a larger vessel diameter during nighttime (295). As a consequence, arterial stiffness tends to decrease during sleep. With this in mind, participants were kept awake during the 360 min period. Diet, smoking, medications and diagnosis are also factors relevant to the vitals outcomes. Van Bortel et al. (2002) recommend standardizing visits to control for confounders and/ or collecting data on them so that they may be factored into analysis. For instance, it is recommended that nutrients/ calories are provided at the same time at every appointment when measuring vitals outcomes like AI. Moreover, it is recommended that participants not drink beverages containing caffeine three to four hours before assessment (265,295). All of these factors were considered when developing methods (e.g. eligibility criteria, timeline).

Figure 3.1. How to Use the Pulse Wave Sensor Unit

(A) Non-invasive blood pressure monitor with Augmentation Index (AI); (B) Sensor of AI pulse wave measurement unit; (C) Demonstration of palpation of the left radial artery (Note wrist position mark on unit [white circle]; this is the AI measurement point); (D) AI pulse wave sensor in position for measurement. Figure adapted (with permission) from: Sá da Fonseca, L. et al. (2014) World Journal of Cardiovascular Disease; 4: 225-235.

3.3.7. Anthropometric Data

The following anthropometric data were collected: (1.) weight (kg), (2.) height (cm), and (3.) waist circumference (cm). BMI was calculated post-study (kg/m2) using mass and height outcomes. Weight and height were measured using a calibrated SECA Medical Beam Scale. Waist circumference was measured by trained clinical staff. Anthropometric data was collected, in duplicate, at the screening appointment and each study visit; average values were used in analysis. Participants were asked to avoid losing or gaining weight the study period and reminded during each phone call to book study appointments.

3.3.8. Hard Copy Data Collection Tools

There were three hard copy data collection tools implemented during the study; (1.) 24 hour recall, (2.) Lifestyle Questionnaire, and (3.) Symptoms Questionnaire. Dietary intake data was collected using a standardized 24 hour recall (appendix 3.7.) administered to participants by a trained RD. Recalls were administered at ~270 minutes post- test food administration. Standardized questions were asked to facilitate collection data collection. When completed, the recall was reviewed by the RD with the participant to ensure details were not forgotten (e.g. candy snack at 3pm). The Lifestyle Questionnaire was a staff administered mixed-form questionnaire developed to collect data on lifestyle behaviour during the study. This data collection tool was developed as a baseline document (screening appointment) and follow-up document (four visits). Questions were designed to collect data on smoking, coffee/ alcohol consumption and drug use (or change in medication), as all of these outcomes have been linked with human plasma/ serum antioxidant capacity and arterial stiffness. Select demographic information was collected using this questionnaire as well (primarily in the baseline version). The Lifestyle Questionnaire was administered at the beginning of each study appointment. Participants were asked to avoid making changes to their lifestyle choices/ behaviours (e.g. exercise, diet) during the study unless recommended by their health care provider. The Symptoms Questionnaire (adapted from “The Canadian Trial of Carbohydrates in Diabetes” [CCD]; appendix 3.8.) is a participant administered, close-end, horizontal-form, multiple choice questionnaire designed to capture information on potential side effects of treatments/ test food (59). The Symptoms Questionnaire was administered at each study visit at baseline, 120, and 360 min.

3.3.9.0. Data Analysis

3.3.9.1. Software

The majority of the statistical analysis or modeling conducted was done using IBM SPSS version 21 (Copyright © IBM Corporation and other(s) 1989, 2012). Lotus 1-2-3, 1997 edition (Lotus Development Corporation, Cambridge, MA) and Microsoft® (MS) Excel® 2013 (Part of MS Office Professional Plus 2013). Outcomes collected as repeated measurements (e.g. biochemical data, anthropometric data, vitals) were also used for randomization, data organization, and basic descriptive statistics. Dietary Intake Data (24 hour recalls and oral nutritional supplements) and physical activity were entered into ESHA Research’s Food Processor® Nutrition and Fitness (0.14.0.; Salem, Oregon). This program uses its own Nutrition Information Database (http://www.esha.com/nutrition/database-information) and Physical Activity/ Exercise Database. To supplement this database, the Wolever Laboratory has developed an evidence-based GI database (using modified foods from the ESHA database). Appendix 2.1. provides an excerpt from the ESHA Manual of Operations developed to standardize and objectify the process of GI assignment. Dietary intake outcomes of interest included energy, macronutrients, fibre, dietary vitamin C, and dietary GI.

3.3.9.2. Analysis

Analysis of continuous repeated measures were conducted on raw data, data adjusted for baseline as a covariate and incremental outcomes (within participant repeated measures minus baseline value; V2 – V1, V3-V1 etc). Biochemical outcomes have been presented in the results section as either raw or incremental outcomes. Biochemical outcomes (including the primary outcome), mass and change in mass, dietary intake date (24 hour recall output from ESHA) and vitals were analyzed as dependent variables using SPSS Linear Mixed Model (Maximum Likelihood [ML]). The linear mixed model (or linear growth model or individual growth model) is a variation of the standard linear model used within the general linear model (GLM) (375). This procedure models the variance and covariance as well as the overall means, making it a suitable form of analysis for clinic-based human research (typically heterogeneous samples). Moreover, the mixed model procedure is recognized for its ability to handle missing values, which is a reality in clinical trials. Treatment 1/ A and 3/ C were defined as a fixed factor in the

88 model and labeled “rate” (Sip versus bolus), while Treatment 2/ B and 4/ D were defined as a fixed factor in the model and labeled “supplement” (Vitamin C administered versus Vitamin C not administered). The results of these analyses were expressed as mean ± SEM (main effect rate, main effect supplement, and time point means). When a significant time-treatment interaction was detected for rate or supplement, time point testing followed. All time points were deemed relevant as the data collection schedule was developed by clinician scientists to test our aforementioned hypothesis. The Sidak model was used to compare main effects of treatment and for post hoc (time point) analysis. All statistical tests were given a 2-tailed p < 0.05 criterion of significance.

The study questionnaires were designed and collected for the primarily purpose of facilitating clinical assessment and protocol adherence. In combination, the study questionnaires were designed to confirm data collected during the screening procedure, to increase staff-participant communication, to define the sample demographically, and to assist staff/ students in assessment of relevant changes (e.g. change in weight, change in medication) that may affect the primary outcome. Select questionnaire outcomes will be discussed in the results section of this chapter. The questionnaires provided primarily descriptive data; non-continuous variables were described as counts (e.g. percent, ranges) and interval and ratio variables as means ± SD. To analyze effect of treatment on symptoms (or participant burden) we conducted a within symptom aggregation of data (converting ordinal to interval data) to assess overall symptom burden. That is, participants’ responses to each symptom were assigned numbers 0 to 4 for no symptoms to severe symptoms. For each visit/ treatment, these numbers were added/ summed (higher number = higher burden or more symptom) and analyzed within the individual growth model as described above. The statistical analysis plan (for all three studies) was developed with the input of Kevin Thorpe, Biostatistician and Assistant Professor in The Dalla Lana School of Public Health University of Toronto.

3.4.0. RESULTS

3.4.1. Sample Characteristics

Twenty-eight participants expressed interest in the study, 25 were screened, and 18 were eligible to participate. The most common reasons for not being able to participate included ineligibility due to: DM diagnosis, participating in another study and/ or inability to commit to study timeline. Of these 18 participants, all participants completed one visit, 17 participants completed two visits, and 16 participants completed all four visits. The participant who decided to withdraw after the first visit, experienced noteworthy symptoms related to fasting (e.g. dizziness and nausea) and the 75 gram glucose solution (confirmed in Symptoms Questionnaire and clinic chart notes). This participant disclosed that s/he had experienced similar symptoms after fasting in the past during standard clinical care. The Study RN and Principal Investigator (Physician) supported with the participant’s decision to withdraw from the study. A second participant withdrew after the second visit, having developed an aversion to the catheter/ blood draw procedure (participant described experiencing “significant bruising” post-study visit). Select baseline demographics and clinical outcomes are shown in table 3.6. The study sample included 7 males and 11 females aged 25 to 60 years (mean ± SD = 51 ± 8) years. Mean (mean ± SD) BMI was 33 ± 3 kg/m2, as expected as per eligibility criteria. Waist circumference ranged 84 to 129 cm; males having an average waist circumference of 96 ± 12 and females 99 ± 13 cm. The lipid profile (plasma) of the 17 active participants is also included in table 3.6.

Table 3.6. Select Sample Characteristics at Screening

Outcomes Mean ± SD (n = 17) Male (n = 7) Female (n = 11) Age (Years) 51 ± 9 51 ± 7 BMI (kg/m2) 33 ± 3 33 ± 6 Waist Circumference (cm) 96 ± 12 99 ± 13 SBP (mmHg) 138 ± 10 127 ± 14 DBP (mmHg) 79 ± 10 78 ± 10 Total Cholesterol (mmol/L) 5.6 ± 0.9 Triglycerides (mmol/L) 1.3 ± 0.7 HDL (mmol/L) 1.4 ± 0.3 Total Cholesterol/ HDL 4.27 ± 1.0 LDL (mmol/L) 3.6 ± 0.8 SD = standard deviation; BMI = body mass index; SBP = systolic blood pressure, DBP = diastolic BP; HDL = high density lipoprotein; LDL = low density lipoprotein

3.4.2.0. Biochemical Outcomes

3.4.2.1. Plasma Glucose

Reflective of the results obtained by Jenkins et al. (1990), administering 75 g of oral dextrose as a bolus resulted in significantly higher postprandial blood glucose excursions than sipping (figure 3.2.). Although a significant main effect of rate was not detected (bolus: 5.80 ± 0.07 vs. sipping: 5.68 ± 0.07 mmol/L), a time*rate interaction (p ≤ 0.0001) was. As expected, plasma glucose increased sharply during the first hour post-bolus. Plasma glucose was significantly higher 30 min after bolus administration in comparison to sipping (bolus: 8.11 ± 0.27 vs. sipping: 6.68 ± 0.27 mmol/L respectively; p ≤ 0.0001). Plasma glucose rapidly declined below baseline within two hours post-bolus until significantly lower than plasma glucose post-sipping at both 180 (bolus: 4.06 ± 0.20 vs. sipping: 5.30 ± 0.20 mmol/L; p ≤ 0.0001) and 240 min (bolus: 3.94 ± 0.18 vs. sipping: 5.12 ± 0.17 mmol/L; p ≤ 0.0001). The standard lunch (consumed at 240 min) elicited a higher glycaemic response after bolus administration, resulting in significantly higher plasma glucose at 300 min in comparison to post-sipping plasma glucose (bolus: 6.11 ± 0.16 vs. sipping: 5.14 ± 0.16 mmol/L; p ≤ 0.0001). These findings remained

91 significant upon analysis of incremental plasma glucose. A main effect of vitamin C supplementation was not detected, nor was a time*supplement interaction.

3.4.2.2. Plasma Insulin

Plasma insulin responses are shown in figure 3.3. A main effect of rate was detected. Sipping dextrose solution resulted in significantly lower plasma insulin than bolus administration (bolus: 28.58 ± 1.52 vs. sipping: 21.22 ± 1.48 mU/mL, p = 0.001). As expected from the plasma glucose results, a significant time*rate interaction was detected (p ≤ 0.0001). Plasma insulin increased rapidly to 60 minutes post-bolus, after which it declined. Insulin was significantly higher after bolus than sipping at 30 min (bolus: 55. 86 ± 5.28 vs. sipping: 22.45 ± 5.12 mU/ml; p ≤ 0.0001), but not significantly different between treatments at 60 min. At 240 min, plasma insulin was significantly lower after bolus administration in comparison to post-sipping (bolus: 5.32 ± 2.93 vs. sipping: 14.63 ± 2.82 mU/mL; p ≤ 0.0001). From 270 min to 360 min, there was no significant difference in plasma insulin due to rate of dextrose administration (bolus vs. sipping). Whether or not vitamin C was administered did not influence plasma insulin and no time*supplement interaction was detected.

Figure 3.2. Raw plasma glucose (mmol/L) versus time (minutes) (n= 17)

Legend * Solid red line: Bolus Broken red line: Bolus + VitC

Solid blue line: Sip ose (mmol/L) ose * Broken blue line: Sip + VitC * *

Plasma Gluc Plasma

Time (minutes)

Figure 3.3. Incremental plasma insulin (mU/mL) versus time (minutes) (n= 17)

Legend * Solid red line: Bolus

Broken red line: Bolus + VitC

(mU/ml) Insulin Solid blue line: Sip

* Broken blue line: Sip + VitC Plasma

Incremental a

Time (minutes)

For figures 3.2. and 3.3.: Data expressed as mean ± SEM, * = Significant difference between bolus and sipping at specific time point (significant at p < 0.05), Vit C = vitamin C.

3.4.2.3. Serum Free Fatty acids

There was a main effect of rate on serum FFAs. That is to say, administering dextrose solution as a bolus resulted in significantly higher serum FFA concentration than sipping over the six hour study visit (bolus: - 0.21 ± 0.01 vs. sipping: -0.25 ± 0.01 mEQ/L; p = 0.012) (figure 3.4.). Bolus resulted in higher FFA at 240 min when compared to sipping (bolus: 0.02 ± 0.04 vs. sipping: -0.268 ± 0.04 mEQ/L; p ≤ 0.0001). There was no main effect of vitamin C supplementation on FFA, nor was a time*supplement interaction present.

Figure 3.4. Incremental serum free fatty acids (mEQ/L) versus time (minutes) (n= 17)

* Legend

Solid red line: Bolus

Broken red line: Bolus + VitC

Solid blue line: Sip

Broken blue line: Sip + VitC

(mEQ/L) Acids Fatty Free Serum Incremental Time (minutes)

Data expressed as mean ± SEM, * = Significant difference between bolus and sipping at specific time point (significant at p < 0.05), Vit C = vitamin C.

3.4.2.4. Serum C - reactive Protein

Neither rate nor vitamin C supplementation influenced serum CRP. There was no main effect of either treatment on CRP and no time*supplementation or time*rate interaction detected (data not shown).

3.4.2.5. Vitamin C

Rate (bolus vs. sipping) did not influence plasma vitamin C concentration. A main effect of supplementation was detected, when vitamin C was administered with dextrose solution. Overall, mean plasma vitamin C was significantly higher than after dextrose solution(s) without vitamin C (with vitamin C: 23.72 ± 1.43 vs. without vitamin C: -0.41 ± 0.33 μmol/L; p ≤ 0.0001). Also, a time*supplementation interaction was detected (p ≤ 0.0001). Vitamin C supplementation resulted in significantly higher plasma vitamin C concentration at 120 min to 360 min when compared to treatments without vitamin C supplementation (figure 3.5.).

Figure 3.5. Incremental plasma vitamin C (µmol/L) versus time (minutes) (n= 17)

Legend

45 * * Solid red line: Bolus

Broken red line: Bolus + VitC 35 * * Solid blue line: Sip 25 Broken blue line: Sip + VitC *

5 Incremental Plasma Vitamin C (µmol/L) C Vitamin Plasma Incremental -5

Time (minutes)

Data expressed as mean ± SEM, * = Significant difference between treatments given with vitamin C and without vitamin C (significant at p < 0.05), VitC = vitamin C.

3.4.2.6. TRAP (Primary Outcome)

A main effect of vitamin C supplementation on overall mean plasma TRAP concentration was detected (with vitamin C: 48.1 ± 8.3 vs. without vitamin C: 17.2 ± 8.3 μmol/L; p = 0.009), but there was no main effect of rate (bolus: 25. 6 ± 8.4 vs. sipping: 39. 6 ± 8.4 μmol/L; p = 0.24). Nevertheless, a time*supplement and time*rate interaction was observed (p ≤ 0.001). Therefore, all four treatments were compared at each time point (figure 3.6.). TRAP swiftly increased after administration of dextrose solution as bolus, peaking at 180 min. At 180 min, TRAP was significantly lower after sipping in comparison to bolus (sipping: 33.7 ± 31.4 vs. bolus: 133.8 ± 31.4 μmol/L; p = 0.03). Post-bolus TRAP rapidly decreased after 180 min until it was significantly lower than bolus with vitamin C and sipping with vitamin C at 300 min (bolus: - 50.9 ± 31.1 vs. bolus with vitamin C: 58.0 ± 31.1 and sipping with vitamin C: 81.7 ± 31.1 μmol/L; p ≤ 0.016). Post-bolus TRAP was significantly lower than the three other treatments (bolus with vitamin C, sipping, and sipping with vitamin C) at 360 min (-153.1 ± 32.6 vs 10.4 ± 32.6, 7.9 ± 32.6, and 105.9 ± 32.6 μmol/L respectively; p ≤ 0.005). During the 360 min experimental period, vitamin C supplementation attenuated the effect of bolus, but there was no significant difference in TRAP between sipping and sipping with vitamin C.

3.4.2.7. LDL Oxidation and Baseline Conjugated Dienes

There was a main effect of rate on incremental LDLox, however there was no main effect of vitamin C supplementation on LDLox. Administering dextrose solution as a bolus resulted in a less LDLox than sipping (bolus: -1.11 ± 0.39 vs. sipping: 0.62 ± 0.38 U/L; p = 0.001) (figure 3.7.). Moreover, there was no time*rate and no time*supplement interaction during LDLox analysis. There was, however, a main effect of vitamin C supplementation on incremental CD concentration (adjusted for apo-B100), however there was no effect of rate and no time*rate or time*supplementation interaction detected. Administering dextrose solution with vitamin C resulted in a lower CD concentration than administering the dextrose solution without vitamin C (with vitamin C: -0.32 ± 0.52 vs. without vitamin C: 1.82 ± 0.52 μmol/L; p = 0.042) (figure 3.8.).

Figure 3.6. Incremental plasma TRAP (µmol/L) versus time (minutes) (n= 17)

Legend

Solid red line: Bolus

Broken red line: Bolus + VitC

TRAP (µmol/L) TRAP Solid blue line: Sip

Broken blue line: Sip + VitC Plasma Plasma

Incremental Incremental

Time (minutes)

Data expressed as mean ± SEM; significance between treatments have been marked using different letters; significance at p < 0.05; TRAP = total peroxyl radical trapping antioxidant potential, VitC = vitamin C.

Figure 3.7. Incremental LDLox (U/L) versus time (minutes) (n= 17)

Legend

Solid red line: Bolus

Broken red line: Bolus + VitC

Solid blue line: Sip

Broken blue line: Sip + VitC

Time (minutes)

Figure 3.8. Incremental conjugated diene concentration (µmol CD/ µmol of protein [apo- β]) versus time (minutes) (n= 17)

Legend

Solid red line: Bolus

Broken red line: Bolus + VitC

Solid blue line: Sip

Broken blue line: Sip + VitC

(µmol CD/ µmol Protein) µmol CD/ (µmol

Incremental Conjugated Diene Concentration Diene Concentration Conjugated Incremental

Time (minutes) For figures 3.7.and 3.8. Data expressed as mean ± SEM; no significant difference(s) LDLox = Low density lipoprotein oxidation, apo-β = apoB-100/apolipoprotein-B, VitC = vitamin C.

3.4.2.8. Lipid Profile

Neither rate nor supplementation had a main effect on TG and HDL and there was no time*rate or time*supplementation interaction. Total cholesterol and LDL concentrations were significantly influenced by vitamin C, with total cholesterol being 0.08 ± 0.04 mmol/L lower when vitamin C was administered in comparison to when it was not (p = 0.04; table 3.7.). LDL was 0.09 ± 0.04 mmol/L lower when vitamin C was administered in comparison to when it was not (p = 0.02). A time*rate and time*supplementation interaction was not detected for total cholesterol or LDL.

Table 3.7. Main effect of supplement administration on total cholesterol and low density lipoprotein

Outcome Units Treatment Means ± SEM Difference ± SEM p-value

TCholesterol mmol/L VitC+ -0.15 ± 0.03 0.08 ± 0.04 0.04 VitC- -0.07 ± 0.03

LDL mmol/L VitC+ -0.14 ± 0.03 0.09 ± 0.04 0.02 VitC- -0.05 ± 0.03 TCholesterol = total cholesterol; LDL = low density lipoprotein; VitC = Vitamin C; Incremental mean ± SEM; p < 0.05.

3.4.2.9. Vitals

There was no main effect of rate (of dextrose solution administration) or supplementation

(whether or not vitamin C was administered) on PP and PR (measured by AI sensor; PPAI,

PRAI). There was a main effect of rate (p=0.04), but not supplementation, on PR measured using the BP cuff (PRBP) (table 3.8.). PRBP was significantly higher after bolus administration in comparison to sipping. Neither a time*rate nor a time*supplement interaction was detected for

PPAI and PRBP. On the other hand, there was a time*rate interaction (p = 0.034) for PRAI, but no time*supplement interaction. A main effect of rate was detected during AI and AIp75 analysis (p = 0.001) with AI being significantly lower after bolus administration when compared to sipping. No time*rate and no time*supplement interaction was detected for AI and AIp75. A main effect of vitamin C supplementation, but not rate, was detected for SBP and DBP (p ≤ 0.04). Both SBP and DBP were significantly lower when vitamin C was administered in comparison to when it was not administered. No time*rate and no time*supplement interaction was detected for SBP and DBP.

Table 3.8. Main effect of treatment administration on incremental vitals

Outcome Units Treatment Means ± SEM Difference ± SEM p-value Augmentation % Bolus -6.31 ± 0.58 2.83 ± 0.81 0.001 Index (AI) Sip - 3.48 ± 0.57 AIp75 % Bolus -6.12 ± 0.57 2.55 ± 0.79 0.001 Sip -3.56 ± 0.55 Pulse Rate BPM Bolus -1.46 ± 0.32 0.94 ± 0.45 0.04 (BP cuff) Sip - 0.52 ± 0.32 SBP mmHg VitC+ -1.76 ± 0.68 2.61 ± 0.96 0.007 VitC- 0.85 ± 0.70 DBP mmHg VitC+ -2.52 ± 0.48 1.40 ± 0.68 0.04 VitC- -1.12 ± 0.48 Incremental mean ± SEM; p < 0.05. AIp75 = AI value (%) normalized to a pulse rate of 75; SBP = systolic blood pressure; DBP = diastolic blood pressure; BPM = beats per minute. Pulse pressure (mmHg) is not significantly different between treatments.

100

Time point analysis and interpretation of PRAI was completed in the context of a supplement*rate interaction (p ≤ 0.001). This analysis is therefore strictly exploratory and intended for hypothesis generation. As shown in figure 3.9., PRAI increased rapidly and steeply after bolus administration, such that, after bolus, PRAI was significantly higher than all other treatments at 60 min post-administration (bolus: 2.63 ± 0.77, bolus with vitamin C: - 0.37 ± 0.77, sipping: - 2.80 ± 0.77, and sipping with vitamin C: 0.55 ± 0.80 BPM; p < 0.05). In addition, the overall mean PRAI 60 min after bolus (mean of bolus and bolus with vitamin C) was significantly greater than that after sipping (mean of sipping and sipping with vitamin C)

(mean ± SEM difference: 2.85 ± 0.85 BPM; p = 0.001). Overall mean PRAI post-bolus was not different from overall mean PRAI post-sipping at 120 min, but PRAI 120 min after bolus without vitamin C was significantly higher than that after bolus with vitamin C (mean ± SEM difference: 5.11 ± 1.54 BPM; p = 0.009) and higher than sipping without vitamin C (mean difference ± SEM: 5.58 ± 1.49 BPM; p = 0.020) at 120 min. Bolus was not different than any other treatments after 240 minutes.

Figure 3.9. Incremental pulse rate (BPM; AI sensor) versus time (minutes) (n= 17)

Legend

Solid red line: Bolus

Broken red line: Bolus + Vit C

Solid blue line: Sip

Broken blue line: Sip + Vit C

Time (minutes)

Data expressed as mean ± SEM; Significance between treatments have been marked using different letters; significance at p < 0.05, BPM = beats per minute, AI = Augmentation Index, VitC = vitamin C.

101

3.4.2.10. Protocol Adherence

There were no statistically significant differences in anthropometric, dietary intake, symptoms (participant burden) or lifestyle data between visits (data not shown). Responses to the Lifestyle Questionnaire question “Were study participants adhering to our request to not change lifestyle activities/ behaviours during the study period?” was not significantly different among study visits; 29% respondents reported changing their activity between screening and visit 1, 18% between visit 3 and 4, 18% between visit 1 and 2, and 24% between visit 2 and 3. These descriptive data do not indicate direction of change (increase or decrease). The most common activity reported was walking (various intensities). Participants reported walking throughout the year, whereas other activities seemed to be limited by seasonality (or availability). Overall, these data indicate that ~90% of study participants were sedentary to lightly active and maintained this lifestyle behaviour during the study period.

102

3.5. CONCLUSIONS AND DISCUSSION

The design of the AOGI Study was based on a protocol, developed by Jenkins et al. (1990), that showed prolonging the rate of glucose consumption reduces postprandial glucose excursions and enhances insulin economy and glucose disposal. The sipping versus bolus paradigm was used to test the hypothesis that reducing the rate of rise of blood glucose after eating reduces oxidative stress because it isolates the rate of nutrient delivery as the intervention (or independent variable) and eliminates common confounders introduced by food-based interventions (e.g. vitamins and other food components which may have antioxidant activity).

The AOGI results for the primary outcome, TRAP, indicate that sipping 75 g dextrose/ glucose solution over 3.5 hours resulted in less oxidative stress than ingesting the same amount of dextrose as bolus over 5 minutes (figure 3.6.). As highlighted above and in the literature review, these findings are supportive of data published on the effect of low GI foods on markers of oxidative stress and are in agreement with the well-established direct relationship between acute hyperglycemia and postprandial oxidative stress (73,77,232,239,248,262,368,376-379). Sipping dextrose resulted in less TRAP challenge after 180 min in comparison to ingesting the same amount of dextrose as bolus, but there was no main effect of bolus on TRAP. This lack of main effect may be attributed to an initial increase in antioxidant capacity observed between 0 and 180 min, after bolus administration. An increase in plasma antioxidant capacity did not occur when vitamin C (the positive control) was administered with bolus, indicating that an endogenous antioxidant mechanism was at play. That is, adding the antioxidant supplement to the bolus treatment attenuated the body’s acute antioxidant response to the hyperglycaemic insult (262,377-379).

The endogenous antioxidant mechanism is well documented in the literature and is indicative of early stage oxidative stress and systemic acute adaptation or defense (262,377,379). Liu et al. (1999) found acute extreme exercise (e.g. marathon run) significantly increased serum TRAP, despite heavy endurance exercise being linked to increased oxidative stress and ROS generation. In this study, the initial increase in TRAP was attributed to an acute increase in uric acid concentration and the synergistic effects of the various antioxidants present in human plasma. Miglio et al. (2013) showed meal-induced metabolic stress was accompanied by the activation

103 of endogenous antioxidant mechanisms, primarily in the form of increased plasma uric acid and protein thiols, without altering plasma vitamin and carotenoid concentrations. The concept of endogenous antioxidant response is also discussed in subsequent work by this group; published more recently in 2014 and in various peer-reviewed review papers and books (262,377-379). The initial increase in antioxidant capacity may, at first glance, seem counterintuitive, but is an example of the human body’s drive to maintain homeostasis (conceptually similar to milieu intérieur coined by Claude Bernard) and the importance of informed data interpretation.

Enzymatic (e.g. superoxide dismutase, glutathione peroxidase) and non-enzymatic (e.g. dietary vitamin E, β-carotene, vitamin C, thiols, uric acid) defenses work to ensure sufficient antioxidant capacity is available during times of stress. While these enzymes are active in cells, the non-enzymatic antioxidants are at work in the plasma, with uric acid and protein thiol groups being key contributors to overall plasma antioxidant capacity (233,237,276,280,281,284,379, 380). Wayner et al. (1987) showed that urate accounts for 35 to 65% and plasma proteins for 10 to 50 % of the antioxidant activity in samples measured using TRAP. An increase in antioxidant capacity in the plasma or serum, however, is not always “desirable”, as shown by Jackson et al (1995) during their examination of total antioxidant capacity and serum antioxidant concentration in patients with chronic renal failure. In this sample, high antioxidant capacity is related to high serum urate, but secondary to reduced urinary excretion (urate is a salt or ester of uric acid) (381). Moreover, as highlighted by Cadenas and Packer (2005), a decrease in antioxidant capacity in the plasma or serum is not necessarily “undesirable” when the decrease is in the presence of a decrease in ROS. The complexity of interpretation of these assays and metabolic pathways speaks to the importance of conducting multiple assays that measure different antioxidant-oxidant outcomes. A methodological strength of this study was that four biochemical measurements of this type were conducted; one measuring plasma antioxidant capacity, two measuring lipid oxidation/ peroxidation, and one measuring the plasma concentration of a nutrient antioxidant, vitamin C.

The increase in TRAP between baseline and 180 min post-bolus may have accounted for the paradoxical reduction in LDL’s PUFA oxidation. That is, administration of dextrose as a bolus resulted in less LDLox than sipping administration (section 3.4.2.7.). This may be attributed to the aforementioned endogenous antioxidant mechanism, which serves to increase the

104 antioxidant capacity of the plasma and preserve the antioxidant capacity of the lipid (e.g. preservation of vitamin E) (236,262,280,377-379,382,383). Vitamin C supplementation did not affect LDLox (regardless of administration rate); perhaps providing additional support for an endogenous antioxidant response (e.g. protein thiols or uric acid). Conversely, CD, was affected by vitamin C administration, but not rate of administration (section 3.4.2.7.). Strictly speaking, supplementing with vitamin C decreased CD concentration. Vitamin C, a water soluble vitamin and nutrient antioxidant, is a well-known and potent protector of vitamin E. Vitamin E has been recognized for its role in prevention/ reduction of LDL/ lipid oxidation. Vitamin C can even restore vitamin E that has been oxidized by free radicals; allowing it to continue its antioxidant role (among others). In the context of this relationship, both vitamin E and C have been highlighted as potent inhibitors of lipid oxidation (236,278,280,382,384,385).

The second hypothesis was “sipping glucose will reduce oxidative stress to the same extent as 1 gram of oral vitamin C”. Adding vitamin C to bolus administration of dextrose resulted in an attenuation of the acute effect of bolus on TRAP (figure 3.6.). There was no significant difference in plasma TRAP after administration of dextrose as a bolus with vitamin C, compared to sipping alone or sipping with vitamin C. These findings not only indicate that vitamin C reduced postprandial oxidative fluctuations in antioxidant capacity/ concentration, but also indicates that vitamin C supplementation does not provide additional benefit to a system that has sufficient protection (antioxidants) against oxidation (sipping). These findings are in agreement with existing data on the effect of vitamin C on postprandial response to OGTT and other acute nutrient loads (e.g. ultra high fat meal to induce hyperlipidaemia); which have led to vitamin C being regarded as protective (233,246,257,335,369,379).

Hypothesis three was “the effect of sipping dextrose on oxidative stress will occur sooner than that of vitamin C”. The results are supportive of this statement, but again reflect the counterintuitive nature of the results and the action of endogenous antioxidant mechanisms (262,377-379). The first effect of sipping was attenuation of TRAP at 180 min, where TRAP was significantly lower post-sipping in comparison to TRAP concentrations after bolus (figure 3.6.). Sipping appears to have blunted the protective antioxidant response seen after administration the bolus treatment. Moreover, sipping dextrose solution resulted in consistent antioxidant plasma levels from180 to 360 min, whereas antioxidant capacity was significantly

105 reduced post-bolus during this same time period. The first effect of vitamin C occurred at 300 min where TRAP was significantly higher after vitamin C administration (with sipping and bolus) in comparison to bolus (without vitamin C supplementation). This difference was maintained until 360 min (end of the study visit timeline). Plasma vitamin C concentrations were significantly higher from 120 to 360 min after vitamin C supplementation in comparison to treatments without vitamin C supplementation (figure 3.5.). These data further support that sipping (slowing carbohydrate absorption) may be a means by which to maintain the antioxidant capacity of the plasma, independent of supplementation.

Low GI meals have been shown to decrease postprandial glycaemic response to a subsequent meal. This phenomena has been coined the “second meal effect” (159,322,367). In the current study, “second meal effect” has been used to describe both blood glucose and TRAP response after the standard lunch (or second meal). Bolus administration of dextrose solution (representative of quick carbohydrate absorption or a high GI meal) resulted in a significantly higher glycaemic response after the second meal in comparison to the response seen after lunch post-sipping (figure 3.2.). This increase in plasma glucose would be expected to result in the decrease in TRAP post second meal. TRAP increased after bolus dextrose solution administration with vitamin C, however, indicating that the TRAP second meal effect seen post- bolus may have been attenuated by vitamin C administration. This second meal effect was not reflected in plasma glucose or plasma TRAP concentrations post-sipping, which may further support the protective effect of sipping. The standard lunch meal was designed to be low in polyphenolics and nutrient antioxidants, but scientists have reported post-meal antioxidant perturbations after various meal compositions (table 3.2. and 3.3.) (378). These data inspire the question, is this post-lunch increase in TRAP a result of preservation of antioxidant capacity and/or a post-lunch endogenous antioxidant response?

Although HDL and TG were not affected by any of the four study treatments, whether or not vitamin C was administered did have a significant main effect on total cholesterol (LDL + HDL + TG/5) and LDL (table 3.7.). Both total cholesterol and LDL were lower when vitamin C was administered. In both animal and human models, ascorbic acid deficiency has been seen to inhibit bile acids; leading to an accumulation of cholesterol in plasma and tissues (251,256,278,279,386,387). Myasnikova (1947) showed that serum cholesterol concentrations

106 could be lowered in participants diagnosed with lipid abnormalities by the administration of ascorbic acid. Moreover, a meta-analysis examining the relationship between vitamin C and LDL, HDL, and TG in people living with hypercholesterolemia between 1970 and 2007, showed supplementation with at least 500 mg of vitamin C per day for about four weeks resulted in a significant decrease in LDL and TG concentrations (278). Generally, the literature reports an inverse relationship between plasma vitamin C and total cholesterol, LDL and TG and direct relationship between vitamin C with HDL (278,279,384-389). In our sample baseline mean plasma vitamin C was 16.5 ± 18.2 μmol/L (mean ± SD). The reference value for vitamin according to HICL is ≥ 25 μmol/L. Moreover, according to current guidelines for screening and treatment of dyslipidemia, the sample, defined as having intermediate risk, would have dyslipidemia (e.g. LDL ≥ 3.5 mmol/L) (152,258).

Ascorbic acid is generally regarded as a potent antioxidant that has been shown to improve endothelial-NO dependent vasodilation in vascular beds of patients with conditions characterized by ED; hypertension, DM, hypercholesterolemia, and coronary heart disease (236,246,257,277,278,335,383,389,390). Administration of vitamin C had a significant main effect on SBP and DBP in our study participants, but BP was unaffected by rate of administration (bolus/ sipping) (table 3.8.). Conversely, post-bolus AI, AIp75 and PRBP were lower (main effect of rate) than post-sipping and unaffected by supplementation (whether or not vitamin C was administered). These results are supportive of the increase in TRAP observed between time 0 and 180 min and may be indicative of the acute endogenous antioxidant response, observed in the plasma, extending to the vascular endothelium. PRAI was significantly higher after bolus administration at time 60 and 120; which ran counter to the aforementioned

PRBP (table 3.8., figure 3.9.). Due to the supplement*rate interaction, PRAI results must be interpreted with caution and not extrapolated beyond the context of this study and hypothesis generation. Some discussion was included in the results section to highlight the challenges posed by this interaction. That is, when this interaction occurs both treatments are effecting how the other influences the dependent variable.

Although selected strengths and limitations of AOGI have been highlighted throughout this section of the dissertation, the next few paragraphs will highlight additional strengths, limitations and/ or methodological considerations relevant to study design/ implementation. We

107 will also highlight some potential future directions. Seventy-five grams of dextrose was added to 250 grams of water rather than 50 grams of glucose added to 700 grams of water as per Jenkins et al. (1990). The 75 gram dosage was selected to be representative of the OGTT; used as part of DM screening and diagnosis (101). The volume difference may limit comparability of the data, but the literature indicates it will likely not (391-394). Sievenpiper et al. (1998, 2000) showed that varying the volume of sugar solutions resulted in significantly different postprandial glycaemic responses. Conversely, Sievenpiper et al. (2001) found that dilution of the 75 g OGTT, due either to the direct dilution of the solution or allowing additional water at the request of subjects, may improve the overall tolerability of the protocol and will likely not affect reproducibility. These later findings were in support of Brouns et al (1995), a frequently cited paper in this area of research, showed that gastric emptying was influenced by carbohydrate concentration or rate of solution delivery rather than osmolarity. Glucose solution administered via sipping was administered in the Jenkins et al. (1990) paper using pre-measured Dixie® Cups. During this study, eight fluid ounce glasses were used to administer the dextrose solution as a sipping treatment. Study staff used a black marker and timer to facilitate participant solution sipping at predetermined times within the 3.5 hour consumption window. This approach did result in occasional overconsumption of solution, which resulted in a more pronounced glycaemic response than predicted between time 30 and 120 min. Although this response did not abolish the significant difference in blood glucose between sipping and bolus, it may have impacted the comparison of sipping with both vitamin C treatment (bolus with vitamin C and sipping with vitamin C).

The TRAP assay as per Valkonen and Kuusi (1997) was modified to accommodate analysis using a microplate reader during TRAP analysis. By adapting this procedure and utilizing an eight channel pipettor, 30 samples were analyzed per run; more than tripling the spectrophotometric capacity reported by others. This is a noteworthy methodological improvement in light of the sensitive nature of antioxidant concentrations in plasma (e.g. photosensitive, heat sensitive) (236,287). Other examples of noteworthy modifications to the assay include the following details related to reagent preparation: Valkonen and Kussi (1997) reported 2,7-dichlorofluroscein-diacetate (DCFH-DA) should be dissolved in Phosphate Buffered Saline (PBS) with no mention of supplementary materials to assist with preparation. Cayman Chemical, however, notes that NaCO3 can be used to dissolve DCFH-DA in PBS. Use

108 of NaCO3, however, was shown to interfere with plasma lag phase determination. To address this issue, DCFH-DA was dissolved in pure ethanol and mixed thoroughly. Also, we observed that dissolving Trolox in ethanol, as per Valkonen and Kussi (1997), also interfered with lag phase induction. Trolox was therefore dissolved in 1X PBS. Repeatability was established in advance of AOGI sample analysis and during analysis, however a formal inter-lab test of validity was not conducted. Moreover, our results were not compared to the results obtained from the original assay. This may limit comparability of our results to those form other laboratories.

This study has provided supplementary data on the relationship between the slow absorption model and post-prandial oxidative burst and arterial stiffness; addressing clinicians’ and researchers’ call for more data on GI and providing more insight into the GI mechanism. With this study, we were able to confirm our three hypotheses using an adapted version of a protocol by Jenkins et al. (1990) to study delayed carbohydrate absorption without common dietary confounders (e.g. nutrients). This study is in agreement with the findings of Botero et al. (2009) where low GI diets resulted in less oxidative stress and/ or conservation of antioxidant capacity in comparison to high GI diets, but has highlighted the potential role of endogenous antioxidants (262,377-379). This study has provided important data for future hypotheses generation and study development. It is recommended that these data be used to as a rationale to further develop the methodology used to test the effect of slowing carbohydrate delivery on postprandial oxidative response. Outcomes including protein thiols and uric acid would provide more insight into the relationship between TRAP, LDL oxidation and acute hyperglycemia. Vitamin E is also another biochemical outcome that may provide more information about the relationship between vitamin C and LDL oxidation. In the future, educators will be consulted on their interest, ability and suggestions for how to incorporate these and future antioxidant-GI research into their practice. This is a promising path intersection, as Canadian RDs are already currently engaged in translational activities related to integrating the concepts functional foods and antioxidants into practice (183).

109

CHAPTER 4.0. STUDY#2 EVALUATION OF GLYCAEMIC INDEX EDUCATION IN PEOPLE LIVING WITH TYPE 2 DIABETES MELLITUS: PARTICIPANT SATISFACTION, KNOWLEDGE UPTAKE AND APPLICATION (Short Titles: GI Education Evaluation Study, GIEES)

Components of this chapter (4.0.) have been previously presented:

(1.) Bhatti, G (intern), Sobie, M (intern), Grant, S. (preceptor and research coordinator), Hurst, KJ, Darling, P (preceptor), Wolever, T. (2011) Phase 1: Developing a questionnaire to evaluate glycaemic index education, acceptability and application in men and women living with type 2 diabetes mellitus; Dietetic Education Leadership Forum of Ontario Research Day 2011; St Michael’s Hospital, Toronto, Ontario.

(2.) Cavanagh, J (intern), Elliott, E (intern). Grant, S (preceptor and research coordinator), Hurst, KJ, Darling, P (preceptor), Wolever, T. (2012) Phase 2 Preliminary Descriptive Analysis: Evaluating glycaemic index education, acceptability and application in men and women living with type 2 diabetes mellitus; Dietetic Education Leadership Forum of Ontario Research Day 2012; St Michael’s Hospital, Toronto, Ontario.

109

110

CHAPTER 4.0. STUDY#2 EVALUATION OF GLYCAEMIC INDEX EDUCATION IN PEOPLE LIVING WITH TYPE 2 DIABETES MELLITUS: PARTICIPANT SATISFACTION, KNOWLEDGE UPTAKE AND APPLICATION

4.1. ABSTRACT

Use of low glycaemic Index (GI) foods is recommended by the Canadian Diabetes Association for managing Type 2 Diabetes Mellitus (T2DM). Notwithstanding, 61% of Canadian Registered Dietitians (RDs), working with clients with DM, do not use GI in practice, highlighting the following barriers to utility: Lack of suitable GI-education tools, a belief that GI is too difficult for clients to understand and apply, and a need for more GI-utility data from diverse client populations. Although the literature supports that available GI-education materials and job aids are unsuitable, there is not enough evidence available to support or refute that GI is too difficult for clients to understand and apply. To address the lack of data available on GI-education evaluation, a mixed-form questionnaire (GIQ) was developed, pre-tested and used to evaluate an evidence-based GI education platform. Participants (n = 29), with T2DM, attended a 35-40 minute GI education session, led by a trained RD. The GIQ was administered pre-education, immediately post-education, and one and four weeks post-education. Three-day-diet-records were administered pre-education and at one and four weeks post-education. The primary outcome, change in dietary GI, was significantly lower at one and four weeks (mean±SEM; both 54±1) compared to baseline (58±1; p≤0.001; 4-5 unit decrease). Most study participants (28/29) were satisfied with the education session. Knowledge score significantly increased from pre- education (53.6±5.1%) to immediately post-education (83.5±3.4%; p≤0.001), one week post- education (87.5±2.6%; p=0.035) and four weeks post-education (87.6±3.8%; p=0.011). Our findings suggest that a statistically significant reduction in dietary GI can be obtained using the GI education platform; supporting clients can understand and apply GI-knowledge and skills. The education and evaluation materials created for this study have addressed the aforementioned perceived barriers to GI utility and can be used in other DM populations for which more GI utility data is required (e.g. gestational diabetes mellitus).

Funded By: Canadian Diabetes Association, Canadian Institutes for Health Research

111

4.2.0. INTRODUCTION

4.2.1. Study Rational

The CDA CPG includes recommendations that GI education be used as a supplement to standard care for people living with T2DM and that use should vary by client interest, ability and need (144). As previously outlined in this dissertation, Canadian RDs have called for additional data on GI utility and more/ reliable job aids10 (e.g. GI education materials), so that they can make informed professional decisions about when and whether to use GI in their professional practice (62,63,133). It follows that there is a need for GI education development and evaluation and development of validated, evidence-based tools to facilitate this evaluation (e.g. questionnaires). Many studies evaluate the effect of low GI education by examining biochemical and/ or anthropometric outcomes (KM level 4 or “results”) using RCT design (40,43,59,140,215). According to the KM, however, in order to achieve level four, one must ensure the bottom three levels are satisfied. Moreover, this multi-level approach to evaluation provides the opportunity to isolate what level requires supplementary attention (development, evaluation etc.) (80-83). Therefore, the GI Education Evaluation Study (GIEES) was developed to include two phases in order to address these practice-based, clinician-identified gaps. Phase 1 included development and face and content validation of the GI Questionnaire (GIQ; data not shown in this dissertation), while phase 2 included GI education evaluation and GIQ pretest. The GIQ is a hybrid evaluation tool (designed to evaluate KM levels 1 to 3). The GI education tools and approaches implemented in this study represent an evidence-based education platform. This platform pulls from education strategies and patient education materials developed and implemented by researchers experienced in development and implementation of GI interventions (e.g. Wolever, Slabber, Brand-Miller) (30,31,38,40,59,122,130,131,143,196,201, 215,223,395). These materials and methods (or job aids) will be discussed in more detail in the methods section.

10 A job aid is a device or tool that allows an individual to quickly access the information needed to perform a task reliably (e.g. patient education materials and backgrounders) (80).

112

4.2.2. Hypotheses and Primary Objective

Patients with T2DM who participate in a GI education session will be satisfied (KM level 1) with the session, show an increase in GI knowledge (KM level 2), and significantly lower dietary GI post-education in comparison to pre-session dietary GI (KM level 3).

Primary Objective: To evaluate if the low GI education platform can significantly reduce dietary GI in participants with T2DM post-education.

4.3.0. MATERIALS AND METHODS

4.3.1. Design

GIEES had a one-armed questionnaire-based pre- and post- education evaluation design with repeated post-intervention measures. The questionnaire (GIQ) was developed and face- and content- validated (phase one) prior to the pre-test phase (phase two) of this study. Phase two, the dietary intervention evaluation, will be the focus of this dissertation chapter.

All components of this study (phase one and two) were reviewed and approved by St Michael’s Hospital Research Ethics Board and the University of Toronto Office of Research Ethics. A steering committee composed of St Michael’s Hospital Dietetic Internship Preceptors, University of Toronto Graduate Students and Faculty, St Michael’s Hospital Dietetic Interns, and RDs reviewed all evaluation materials and methods developed for this study.

4.3.2.0. Sample

Study participants were patients/ clients receiving standard care within the St Michael’s Hospital Diabetes Comprehensive Care Program (DCCP). Patient screening was completed by a DCCP RD and DCCP Clerical Staff offered a letter of information to all eligible patients. If potential participants signed the letter of information, they were approached by a member of the research team to review the consent form. Those who chose to provide written informed consent became study participants. Individuals with the following characteristics (criteria) were eligible for participation in the study. Patients:

113

(1) diagnosed with T2DM, (2) receiving care within St Michael’s Hospital DCCP, (3) over the age of 18, and (4) able to speak and read English.

Individuals who were unable to follow study protocol were ineligible for participation. Exclusion was established on a case-by-case basis, post-screening, and excluded participants were notified (when appropriate) by the Principal Investigator(s).

4.3.2.1. Sample Size Calculation

The sample size calculation showed that 26 participants would be required to detect a 5 unit difference in the primary outcome (dietary GI) at 80% power. The standard deviation of the difference (in GI) seen by Burani and Longo (2006) was 8.7 units (38). A drop out of 10% was estimated, resulting in a final sample size of 29 participants. Study participants were recruited from DCCP from March 14, 2012 and completed September 20, 2012.

4.3.3. Outcomes

The GI Education Evaluation Study primary outcome was change in dietary GI. For diet GI, a reduction of 5 to 9 GI units (KM level 3) has been associated with weight loss, improved beta- cell function and insulin sensitivity (KM level 4) (47,166,167,200,340).

The GIQ was developed to measure the majority of secondary outcomes. Secondary outcomes (and GIQ Sections 1 to 4) were classified under the following four categories:

1. Participant satisfaction with the GI Education Platform 2. GI knowledge uptake/ score 3. Low GI food acceptability 4. GI Education (knowledge and skill) application

114

As part of the assessment of these outcomes, a number of open- and close- end questions were developed to increase our understanding of participants’ self-efficacy, intention to change, perceived barriers/ facilitators to change, and occurrence of barriers/ facilitators. For descriptive purposes, date of T2DM diagnosis was collected from participants’ medical charts.

4.3.4.0. The Study Intervention: Low Glycaemic Index Education

In the context of GIEES, low GI education was delivered in a group setting (5 participants per class). It was facilitated by an RD, MSc, with training in education development, delivery and evaluation, and two Dietetic Interns. Education materials (e.g. food substitution lists, PowerPoint content) were based upon prior GI education initiatives developed in the Wolever Laboratory, feedback obtained on these initiatives from educators and clients, and updated/ informed by publically available peer-reviewed literature, client education materials, and evidence-based education principals (e.g. critical pedagogy) (31,40,63,81,130,131,133,136,143, 186,202,205,222,366,396-398).

Low GI education was layered onto standard care education for people living with T2DM as per the CDA CPG (2008, 2013). Therefore, the education session included a review of current Canadian recommendations for healthy eating (e.g. Canada’s Food Guide, DRI for fibre) and was influenced by behaviour change theory (13,116,117). Two examples of behaviour change theories that informed protocol and education development and implementation are: (1.) Transtheoretical Model (Prochaska’s Stages of Change) and (2.) Bandura’s Social Learning Theory (specifically the self-efficacy construct11) (216-221,363). Goal setting/ self-management strategies (skill building) were included in the baseline education session, but were not revisited at follow-up appointments (399). Four publically available education materials were used to facilitate standard care education:

11 Self-efficiency is regarded by many as the most important construct in social learning theory. It is often described as an individual's belief in his/ her capacity to partake in behaviors necessary to reach his/ her goals. It is thought that personal efficacy determines whether coping behaviour will be initiated, how much effort will be put forth, and how long this effort will be sustainable in the face of obstacles (219).

115

(1.) Eating Well with Canada’s Food Guide: http://www.hc-sc.gc.ca, (2.) Canadian Diabetes Association’s (CDA) The Glycemic Index: http://www.diabetes.ca, (3.) The Plate Method or Diabeters Portion Plate®: www.diabetes.ca; www.Diabeters.com, (4.) Three-dimensional food models: e.g. https://spectrum-nasco.ca (30,116).

Education development was overseen by an education development team with varying experience in DM education (e.g. Physicians, RDs, Dietetic Interns, Medical Students). As part of ongoing evaluation and dissemination efforts, the GI education tools and concepts were presented to researchers, clinicians and hospital staff (at St Michael’s Hospital) to obtain formative feedback. Feedback collation, assessment and integration were overseen by three RDs/ Research Coordinators with an MSc and experience in education development and evaluation. Two examples of presentations delivered during these meetings have been included in appendix 4.1.

The three participant education materials used to facilitate GI-education translation were:

(1.) Low Glycaemic Index Food Substitution List and (2.) Low Glycaemic Index PowerPoint Presentation (3.) Low Glycaemic Index Recipe Book

Details regarding each participant education material are provided below.

4.3.4.1. The Low Glycaemic Index Food Substitution List

The Low Glycaemic Index Food Substitution List was based on the food substitution food list used by Grant et al. (2007, 2011) in a sample of participants receiving care for GDM. This food list included three GI categories, as published in CDA’s The Glycemic Index Tool (low GI = 55 or less, medium GI = 56 to 69, and high GI = 70 or more) (30). The stoplight method was used to distinguish between the three GI categories (Green = Low GI = Choose Most Often, Yellow = Medium GI = Choose Occasionally, Red = High GI = Choose less often). This method for distinguishing GI categories was suggested by Slabber (2005) and has been used by others in nutrition education/ communication (196,398). Using food substitution lists to guide client/ participant intake pattern is the “key foods strategy”. This strategy has been recognized as

116 effective in achieving moderate modifications in food intake (40,54,59,122,208,215). By asking participants to use food substitution lists and work with RDs to personalize their lists, scientists and educators ensure that foods with a tested GI are used for food substitution (decreasing the risk of subjective GI value assignment; discussed in the literature review). Participants were asked to aim to substitute as many carbohydrate containing foods that they could, but assured that one substitution per meal was sufficient to lower dietary GI. In fact, a significant difference in dietary GI (5 to 9 units) has been achieved by replacing 60% of starchy foods (in a diet comprised of 45-65% carbohydrate) with lower GI carbohydrate (40,47,166,200).

Until this study, our lab only included starchy foods in our low GI education and food substitution lists. The practice of including all food groups in GI Education is an approach used in Australian GI education (Brand-Miller and colleagues) (31). This practice was adopted to provide participants with more options and to communicate the importance of layering GI education onto current recommendations. This client education material was intended for provision to study participants after comprehensive review with the education facilitator.

The GIEES iteration of the food substitution list was our first effort to create a food substitution list that includes examples from all Canadian Food Guide food groups; an effort to create a tool that RDs (and other health care providers) can use to teach GI in the context of current dietary recommendations for healthy eating. Moreover, this list was designed to be flexible, as the population St Michael’s Hospital serves is culturally diverse. This iteration of the food substitution list is not included in the dissertation appendix, however the version implemented in study 3 is included in this dissertation (appendix 5.2).

4.3.4.2. The Low Glycaemic Index PowerPoint Presentation

The PowerPoint slide deck was developed for the educator (e.g. RD, RN) as a job aid to facilitate review of the Low Glycaemic Index Food Substitution List and other relevant GI education (e.g. slow absorption model, how low GI foods effect postprandial blood glucose). This presentation walks the client (and the educator) through how to integrate low GI food choices into current CPGs for management of T2DM. The notes section of each PowerPoint side included notes and strategies for content delivery. This slide deck was also created to act as a client education material; developed to be provided as a handout (2-3 slides per page; notes

117 section omitted). This iteration of the presentation is not included in the dissertation appendix, however the version implemented in study 3 is included in this dissertation (appendix 5.3).

4.3.4.3. Low Glycaemic Index Recipe Book

The recipe book is a revised and updated version of the low GI recipe book developed for a previous study within the Wolever Laboratory (122). This participant education material provides study participants with ~ 36 low GI recipes under the headings of breakfast, sides and entrees. It also provides serving recommendations and nutrition information for each recipe. This document includes a section for addition of new recipes; aimed at facilitating communication between participant and RD and to generate new recipes for future iterations. RDs using the first iteration of this job aid reported that it was an excellent way to build rapport regarding traditional foods/ cooking in culturally diverse samples.

4.3.4.4. Marrying Standard Care and Low Glycaemic Index Education Using Hands-on Activity

An example of a GI education platform activity (interactive portion) involved reviewing “The Plate Method” (standard care; figure 4.1.) using three-dimensional food models and a three-way divided dinner plate (30). This activity involved first asking the participants to create a typical meal (breakfast, lunch or dinner) that they would have made or ate prior to the low GI education session. Then, participants were asked to create a plate that reflected current recommendations (without considering GI; focusing on increasing dietary fibre). Finally, participants were asked to create a plate that reflected current general dietary guidelines and included at least one low GI food substitution.

118

Figure 4.1. The Plate Method

Publically available © Copyright 2015; Reproduced with permission from Canadian Diabetes Association; http://www.diabetes.ca/diabetes-and-you/healthy-living-resources/diet- nutrition/basic-meal-planning

4.3.5. Three Day Diet Record: Dietary Intake Data Collection and Analysis Procedures

A three day diet record (with instructions) was used to collect dietary intake data (appendix 4.2.) at baseline, one week post-education and four weeks post-education. Each diet record was reviewed by an RD or Dietetic Intern with each study participant; either in person (baseline visit) or over the telephone (one week and four weeks post-education) to ensure the record was complete. Dietary intake data were entered into ESHA Research’s Food Processor® Nutrition and Fitness (0.14.0.; Salem, Oregon). This program uses its own Nutrition Information Database (http://www.esha.com/nutrition/database-information) which is composed of data from various sources; including the Canadian Nutrient File (400). To supplement this database, the Wolever Laboratory has developed an evidence-based GI database (using modified foods from the ESHA database). Appendix 2.1. provides an excerpt from the ESHA Manual of Operations developed to standardize and objectify the process of GI assignment. When data entry was complete, data was exported to Microsoft® (MS) Excel® 2013 for data cleaning and preparation for statistical analysis using IBM SPSS version 21 (Copyright © IBM Corporation and other(s) 1989, 2012) .

119

Diet records were analyzed for energy, macronutrients (total dietary carbohydrate, fat and protein), fibre, and dietary GI.

4.3.6. The Glycaemic Index Questionnaire (GIQ©)

Based on the KM hybrid evaluation template and peer-reviewed questionnaires/ surveys extracted from published research studies looking at GI education, the GIQ was developed to evaluate the GI education platform (38,40,55,59,81,207,365,366,395,401,402). The following paragraphs (and table 4.1.) provide an overview of the GIQ sections, questions and administration schedule. The GIQ is a mixed-format (open- and closed- end questions) questionnaire with four sections aimed at evaluating GI education using the first three levels of KM. The four sections of the GIEES GIQ are:

Section 1 (Participant Satisfaction): A participant administered mixed-format questionnaire that was developed to assess participants’ satisfaction or “reaction” to the low GI education (e.g. Question 1 = vertical multiple choice, question 2 = vertical true or false, question 3 = open-end) (KM Level 1). Responses for question 4 were presented as a horizontal five point Likert Scale including text and graphic representations/ anchors (i.e. smiley faces) of these options (1 = strongly agree to 5 = strongly disagree). The most recent version of Section 1 will be included in Chapter 5 appendix 5.5.

Section 2 (Demographic Information): Section 2 was an investigator administered mixed format questionnaire designed to collect demographic information from study participants. This section is primarily composed of vertical format multiple choice questions; including an “other” category. Open-ended and “other” responses have been and will be used in development of future iterations of the GIQ. “I am not sure” and “pass” were response options for select questions in this GIQ section. Examples of demographic outcomes examined include: Gender, language, level of education and ethnic group. The most recent version of Section 2 will be included in Chapter 5 appendix 5.6.

Section 3 (GI Knowledge): Section 3 was a participant administered close-end questionnaire intended to measure participants’ knowledge of key GI concepts (KM Level 2: Learning)

120 covered during GI education. Learning was assessed by knowledge score (X/12; Answer Key available in appendix 4.3.). Questions were multiple choice or true or false (dichotomous format) and marked as either correct (score of 1), incorrect (score of 0) or “I do not know the answer” (score of 0). In some cases a second best answer was available (score of 0.5).

Section 4a (Acceptability and Application of GI Education): Section 4a was a participant administered mixed-format questionnaire intended to measure participant satisfaction/ acceptability with/ of the low GI foods included in their food substitution lists (KM level 1) and education transfer (KM level 3). Section 4a measured participants’ perception of application (including perception of degree of food substitution), self-efficacy, symptoms, time and monetary burden, and the effect of their diet change on others in their life. It also measured participant intention to continue with the low GI diet post-study and acceptability of the food included in the food list. Response options were either presented as a vertical format Likert Scale (i.e. Poor, Fair, Good, Very good, and Excellent) or dichotomous true or false questions. GIQ Section 4a can be found in appendix 4.4.

4.3.7. The Glycaemic Index Questionnaire Data Entry Handbook

A GIQ Data Entry Handbook was developed to standardize GIQ data entry and analysis. Coded data, obtained via close-end questions, were entered directly into IBM SPSS version 21 (Copyright © IBM Corporation and other(s) 1989, 2012) from hard copy questionnaires for descriptive and inferential statistical analysis. Open-end GIQ responses (table 4.2.) were reviewed by two RDs, with experience in qualitative data entry and analysis, who indexed/ coded these data according to categories/ themes. A data table excerpt (table 4.3.) and theme codebook (appendix 4.5.) were developed to further prepare these data for sorting and analysis (403-407).

A qualitative data table was created in Microsoft® (MS) Word 2013 (Part of MS Office Professional Plus 2013). Often used for analysis of qualitative data in research settings, MS Word allowed sorting by theme (column heading), without the added cost of qualitative data analysis software (403-408). As shown in table 4.3. chronological sequence numbers were used to ensure a record of original data entry was maintained after the table was revised by theme code using “sort” function/ options. Data were entered using the following information

121 sequence: Participant ID, Visit, GIQ section and question number. Themes and their associated supporting statements (and sub-themes), were analyzed for patterns, interpreted and compared to quantitative data for connections or commonalities. During this review, the two RDs conducted independent review of the data, in duplicate, and kept detailed summary notes/ rationale for each theme/ code identified. When ready, the two RDs came together to compare themes, codes and notes to determine the extent to which similar conclusions were drawn (403,404,406-408). They then separately made additional notes on their interpretation of these themes and created a final (collated) summary document.

4.3.8. Retrospective Medical Chart Review

Date of T2DM diagnosis was collected from participants’ medical charts obtained from St Michael’s Hospital’s Health Records.

4.3.9. Statistical Analysis

Data analysis was conducted using Microsoft Excel® 2013 and IBM SPSS version 21 (Copyright © IBM Corporation and other(s) 1989, 2012). Dietary intake data were analyzed using descriptive statistics using SPSS Linear Mixed Model (ML). The linear mixed model (or linear growth model or individual growth model) is a variation of the standard linear model used within the GLM (375). Date of T2DM diagnosis was used to calculate years with condition and presented as a mean ± SD. GIQ Section 1 to 4a were subjected to descriptive analysis and presented as counts and/ or percent. Within group comparisons between repeated measures (“visits”) were completed for GIQ Section 3 and 4a. Comparisons for GIQ Section 3 (mean score ± SEM at each visit) were conducted by applying a paired t-test, in the context of the close-testing procedure. The close testing procedure is conducted by statisticians to control for type 1 error and is an acceptable means by which to simplify analysis (K Thorpe, personal communication, 2012). GIQ Section 4a data (collected one and four weeks post-education) were analyzed using Wilson Score Method (CI for a binomial proportion). Data collected using the Likert Scale were collapsed into dichotomous values (e.g. > Good = 1, < Good = 0; Likert Scale). All statistical tests were given a 2-tailed p < 0.05 criterion of significance.

122

Table 4.1. Glycaemic Index Questionnaire (GIQ) Administration Schedule Location of Administration Risk Factor Modification Mail  Centre, SMH Baseline Baseline 1 Week Post-Class 4 Weeks Administration Timeline (Visit 1.0) Post-Class (Visit 2) Post-Class  (Visit 1.1.) (Visit 3)

GIQ Section ↓

1 GI Education X Satisfaction 2 Demographic X Information 3 GI Knowledge X X X X 4a Acceptability and X X Application of LGI Foods “X” marks administration according to administration timeline; SMH = St Michael’s Hospital; GI = Glycaemic Index; LGI = Low Glycaemic Index

Table 4.2. Glycaemic Index Questionnaire (GIQ) Qualitative Questions GIQ Section GIQ Question Question Number Number 1 3 What was the most important thing you learned? 1 5 What can we do to make the class better? 4a 3 How would you describe your experience adding low GI foods to your diet?

Table 4.3. Glycaemic Index Questionnaire (GIQ) Qualitative Data Table (Example) Participant GIQ GIQ Visit # Theme Code Response Sequence ID Section # Question # Number Question: What was the most important thing you learned? 123 1 3 1 1000 “The most 1 important thing I learned was how to choose low GI fruits.” 321 1 3 1 2000 “I learned that 2 cold white potatoes are low GI!” Participant identification numbers (ID) and themes are generated examples, not data. The quotes included in the response column (although not exact) are based on of GI Education Evaluation Study lessons learned.

123

4.4.0. RESULTS

4.4.1. Sample Characteristics

Thirty-nine DCCP patients/ clients signed the letter of information, 30 clients consented to participate and 29 attended the baseline visit; including participating in low GI education and completing the visit 1 GIQ sections outlined in table 4.1. One participant withdrew half-way through the baseline visit because of time limitations (called into work). Although the participant expressed interest to return to class upon leaving, he/ she later withdrew from the study, as he/she felt the information covered in the class was “too basic”. Participants completed the study if they completed the baseline appointment (in full) and either visit 2 and 3 or visit 3.

The average number of years since diagnosis of T2DM (average ± SD) was 7 ± 8 years (average year of diagnosis was 2004; years post-diagnosis ranged from 0 [2012] to 25 years). Sixty-nine percent of the sample were male (20/ 29) and 31% were female (9/ 29). The sample was comprised of people from various ethnic categories. As shown in table 4.4. and figure 4.2., however, the largest number of participants were included in the European category (18/29; including Canadian born individuals identifying as Scottish, Irish and/ or English). The remaining counts by category were: 2/29 Aboriginal (Indigenous Canadian), 3/29 African/ Caribbean, 3/29 West Indian, 7/ 29 Indian/ South Asian, 7/29 East Asian, and 7/29 South East Asian/ Filipino. Fifty-two percent (15/29) of study participants were born in Canada, while 48% were born elsewhere (14/ 29). Of participants who were not born in Canada, average year of immigration was 1975 (Range; 1952 to 2008). The main language spoken at home was English (24/29). Of the participants who reported speaking a language other than English when at home, participants spoke French (n = 1), German (n = 1), Spanish (n = 1), Tagalog (n = 1), or Tigrnga (n = 1). As highlighted in table 4.1., the demographic data was collected using GIQ Section 2. Data from this section of the GIQ not discussed in this paragraph has been presented in table 4.5.

124

Figure 4.2. Glycaemic Index Questionnaire Section 2 Demographic Information: Response Summary to “Which ethnic group do you identify with?” (Percent)

Aboriginal

7% 3% 7% European 7% African/ Caribbean 7% West Indian 3%

3% Indian/ South Asian

63% East Asian

South East Asian/ Pilipino

Table 4.4. Glycaemic Index Questionnaire Section 2 Demographic Information: Response Summary to “Which ethnic group do you identify with?”

What ethic group do you identify with? Counts (x) Percent (%; x/n) Aboriginal 2 7 European 18 63 African/ Caribbean 1 3 West Indian 1 3 Indian/ South Asian 2 7 East Asian 2 7 South East Asian/ Filipino 2 7 Other (Identified as “Mixed Heritage”) 1 3 European/ Middle Eastern = 1

125

Table 4.5. Glycaemic Index Questionnaire Section 2 Demographic Information: Questions and responses for 5 to 13.

Question Responses (Counts) Q5. In your home, who purchases the I do (20) food most often? My spouse/ partner (5) Housekeeper (1) Q6. In your home, who makes the meals most often? I do (19) My spouse/ partner (7) My children (1) Housekeeper (1) Q7. What is the highest level of education you have Did not complete elementary school (1) finished?* Did not complete high school (1) High school or high school equivalent (2) College certificate or diploma (10) Undergraduate degree (11) Graduate degree: Masters (5) Q8. Please list all jobs you have held in the past year *Top five responses: for more than one month.* Desk Job (12) Retired (6) No employment/ volunteer (5) Registered Nurse (2) Lawyer (1) Q9. How do you treat or control your diabetes Diet + Exercise (3) today?* Oral Medication (2) Diet+Oral Medication (2) Exercise+Oral Medication (1) Diet+Exercise+Oral Medication (9) Insulin (1) Diet + Exercise+Insulin (1) Oral Medication+Insulin (1) Diet+Exercise+Oral Medication+Insulin (8) Oral Medication+ Victoza (1) Q10. Have you ever met with a Dietitian before to 90% Yes (26) talk about diet? 10% No (3) Q11. Have you ever heard of the glycaemic index? Not sure (1) 14% No (4) 83% Yes (24) Q12. Do you know what the glycaemic index is? 34% No (10) 66% Yes (19) Q13. Have you been taught about the glycaemic 83% No (24) index from a health professional before? 17% Yes (5) * Marks questions that prompted participants to “choose all that apply” (the number of responses may exceed the sample size [n=29]).

126

4.4.2.0. Participant Satisfaction (Kirkpatrick Model Level 1 - Reactions)

4.4.2.1. Responses to Close-end Questions from GIQ Section 1: How Did You Like the GI Class?

All respondents chose “patient” for question 1 (“Who are you?”) and twenty-eight participants (out of 29) responded “yes” to question 2 (“Did you learn anything new at this GI class?”). Question 4 responses have been summarized in table 4.6. The response from participants was overwhelmingly positive in this section; the options disagree and strongly disagree were not selected. These findings support that KM level 2, participant satisfaction, was achieved using the GI education platform.

Table 4.6. Glycaemic Index Questionnaire Section 1: GI Education Satisfaction, statements and responses for question 4. Statement Response Summary Percent (Counts) The class content was easy to understand. 65% Strongly agree (17/26) 35% Agree (9/26) The class content was interesting. 70% Strongly agree (19/27) 30% Agree (8/ 27) The teacher was easy to understand. 81% Strongly agree (22/27) 19% Agree (5/27) The teacher’s assistants were easy to understand. 79% Strongly agree (22/28) 21% Agree (6/28) The hand-outs helped me learn the class content. 63% Strongly agree (17/27) 33% Agree (9/27) 4% Neither agree/ disagree (1/ 27) The presentation helped me learn the class content. 74% Strongly agree (20/27) 26% Agree (7/27) The hands-on activities helped me learn the class 54% Strongly agree (15/28) content. 36% Agree (10/ 28) 10% Neither agree/ disagree (3/28) I think that what I learned today will help me make 67% Strongly agree (18/27) changes to my diet. 30% Agree (8/27) 3% Neither agree/ disagree (1/27)

127

4.4.2.2. Responses to Open-end Questions from GIQ Section 1: How Did You Like the GI Class12?

All 29 participants responded to question three, which was “If you learned something today, what did you learn?”13. All participants reported that the standard care review provided as part of the GI education was very helpful to them; despite the fact that 90% (26/29) of study participants had seen a RD in the past and received standard care medical nutrition therapy. The most notable themes identified in responses to question three include: (1.) Current Standard Care (theme): Serving Size (sub-theme) (6 instances) (2.) GI (theme): Participant perception of glycemic index knowledge (sub-theme) (14 instances) and (3.) GI: GI Categories (11 instances).

Three examples of participant responses (n = 3) included under the theme title “Current Standard Care”, sub-theme “Serving Size”, are provided below (table 4.7.). Other sub-themes of “Current Standard Care” identified in these response examples include “Perception of Food/ Diet Quality” (e.g. “milk is a good choice”) and “Perception of Dietary Liberation” (e.g. “pasta and rice are not bad choices”).

Table 4.7. Participant Responses Organized under Theme “Standard Care”, Sub-theme Serving Size” Question 3: If you learned something today, what did you learn? Response Example 1 “Quantity – especially my carrots – although (they are) good – (I know) don’t overdo it.” Response Example 2 “The most important thing is that I know how much food I should eat and what kind of food.” Response Example 3 “Pasta, rice are not bad choices. Milk is a good choice. (Important to remember) portion control.”

The GI sub-themes “Participant Perception of GI Knowledge” and “GI Categories” were the most commonly observed during analysis; with 14 and 11 statements/ instances being identified; respectively. Nine examples of responses to question 3 (n=9), category of GI, are provided in table 4.8. below.

12 Reminder that the theme codebook can be found in appendix 4.5. 13 Responses are presented without participant identifiers. Round brackets were used to enclose fragments of responses added after responses were reviewed by staff with participants.

128

Table 4.8. Participant Responses Organized under Theme “Glycaemic Index” Question 3: If you learned something today, what did you learn? Response Example 1 “(I learned what the GI was for certain foods) that I was interested in. I also learned that one can once in a while back off a bit – which I do not want to take advantage of too often!” Response Example 2 “Which breads are low(er) glycaemic… how to lower the GI of a meal.” Response Example 3 “Low vs Med vs High GI foods and how to (use this information to) manage food and blood glucose levels.” Response Example 4 “Converted rice and al dente pasta are not devils!” Response Example 5 “In order to lower index – cook pasta less time (~10 min) to remain firm…potato salad (cold) instead of hot.” Response Example 6 “Pasta, rice are not bad choices. Milk is a good choice. Portion control.” Response Example 7 “Potatoes are not bad for us – eat them cold!” Response Example 8 “Food (the same one) can have different GI based on how it is cooked and served.” Response Example 9 “Better understanding of GI as it apples (applies) to carb-fruit-veg- Section of the CFG (Canadian Food Guide).”

In addition to the identification of the aforementioned sub-theme “Participant Perception of Glycaemic Index Knowledge”, another sub-theme related to GI knowledge uptake was identified and called “Education/ Comprehension” (8 instances identified). Responses like examples four, five and seven indicate that participants increased GI knowledge during the class, whereas responses like two, three, and eight show participants’ perception of GI knowledge uptake. These findings are reflective of the quantitative data collected using GIQ Section 3: What do You Know about GI? (Section 4.4.3.; next section).

Eleven participants responded to question five, “What can we do to make the class better?” Responses to this question called for development of a theme called “Feedback from Participants”. Four sub-themes were established under this theme, including: (1.) Scheduling/ timing of class, (2.) Class flow/ message cohesion, (3.) Class content/ information, and (4.) Education delivery strategies (e.g. Props, Games, Interaction). Four examples of responses to question 5 (n = 4), falling under the general category of “Feedback from Participants”, are included in table 4.9. below.

129

Table 4.9. Participant Responses Falling under Theme “Feedback from Participants” What can we do to make the class better? Response 1 “Start @ 8:30 or after 5pm.” Response 2 “Well done and interesting presentation, good use of AV, time went quickly.” Response 3 “The class moved along well and was made interesting with the use of food products (models) to help build a food plate. Visual goes a long was in driving home the point of what it takes to achieve blood sugar at acceptable levels.“ Response 4 “I appreciated the interactive nature of the lesson and being able to ask questions. Thanks!”

Two examples of quotes taken from question five responses, falling under the theme “Current Standard Care”, sub-theme “Serving Size” are included in table 4.10.

Table 4.10. Participant Responses Organized under Theme “Standard Care”, Sub-theme Serving Size” What can we do to make the class better? Response 1 “Don’t make people play with food models. Although they are useful for serving sizes.” Response 2 “The class moved along well and was made interesting with the use of food products (models) to help build a food plate. Visual goes a long was in driving home the point of what it takes to achieve blood sugar at acceptable levels. “

The examples in table 4.10. were also categorized under the sub-theme “Education Delivery Strategies (e.g. Props, Games, Interaction)”, as this feedback is directed at the use of food models and the plate method during The Plate Method activity (outlined in section 4.3.4.4.).

130

Section 4.4.3. GI Knowledge Score (Kirkpatrick Model Level 2 – Learning)

As shown in figure 4.3., study participants scored significantly higher on the GI Knowledge Questionnaire after they completed the GI education session. That is, GI knowledge score was significantly higher at visit 1.1 (83.5 ± 3.4%; p ≤ 0.001), visit 2.0. (87.5 ± 2.6%; p = 0.035) and visit 3.0. (87.6 ± 3.8%; p = 0.011) when compared to baseline or pre-education (53.6 ± 5.1%). There were no significant differences between GI knowledge scores post-education (visits 1.1, 2.0, 3.0 or post-education baseline, one week post-education, four weeks post-education). These findings support that KM level 2, learning, was achieved using the GI education platform.

Figure 4.3. Total Knowledge Score at each administration of Glycaemic Index Questionnaire Section 3: Glycaemic Index (GI) Knowledge (by visit)

* * *

* Significantly different from baseline score; mean (± SEM); p < 0.05.

131

4.4.4. Three Day Diet Record Data (Kirkpatrick Level 3 – Transfer)

Dietary intake data collected using three day diet records at each visit are shown in table 4.11 The primary outcome, dietary GI, was significantly lower at visit 2 and 3 (54 ± 1 at each visit) when compared to baseline dietary GI (58 ± 1) (p ≤ 0.001). Dietary fiber and protein did not change from baseline during the post-education study period. Calories, carbohydrate, and fat were significantly lower post-intervention; with no significant difference between visit 2 and 3. Macronutrient intake, as a percentage of total daily caloric intake(s), was not significantly different between visits and was maintained within the acceptable macronutrient distribution ranges (AMDR) (table 4.12.) (117). These findings support that KM level 3, transfer, was achieved using the GI education platform.

Table 4.11. Results of three day diet record data analysis

Dietary Intake Data Outcome Baseline Visit 2.0 Visit 3.0 Mean (SEM) Mean (SEM) Mean (SEM)

Calories (Kcal) 1965 (67) 1647 (73)* 1631 (74)*

Carbohydrate, total (g) 222 (8) 183 (9)* 187 (9)* Fibre (g) 21 (1) 22 (1) 23 (1)

Carbohydrate, available/ net (g) 201 (8) 161 (8)* 163 (9)* Protein (g) 95 (4) 86 (4) 91 (4)

Fat (g) 76 (4) 63 (4)* 58 (4)*

Glycaemic Index (%) 58 (1) 54 (1)* 54 (1)* Baseline = pre-education, Visit 2 and 3 = post-education; *Significantly different from baseline; p<0.05

Table 4.12. Macronutrients as a (mean) percentage of total daily caloric intake

Macronutrient Baseline Visit 2.0. Visit 3.0. Pre-education Post-education Post-education Carbohydrate, total (% of total energy) 44 45 44 Fat (% of total energy) 36 35 34 Protein (% of total energy) 20 20 22 *Based on mean values in table 4.6.; no statistically or clinically significant differences detected between study visits.

132

4.4.5.0. GIQ Section 4a: Is Your Low GI Diet Working For You? (Application and Acceptability)14

4.4.5.1. Responses to Close-end Questions from GIQ Section 4a

Close-end responses are summarized in table 4.17. as counts and percent. Data collected using the Likert Scale were collapsed into dichotomous values (e.g. > Good = 1, < Good = 0; Likert Scale), incidence of 0 or 1 was not different between visit 2 and 3.

4.4.5.2. Responses to the Open-end Question from GIQ Section 4a

Twenty-one participants responded to question 3, “How would you describe your experience adding low GI foods to your diet?” at week 1 (visit 2.0.) and week 4 (Visit 3.0) (pooled for analysis). The two prominent themes identified in the responses to this question included: (1.) GI and (2.) Feedback from Participants. The most common sub-themes under GI included: (1.) Education/Comprehension (10 instances), (2.) GI Application (sub-theme): Food Selection (secondary sub-theme) (18 instances) and (3.) GI Self-efficacy (27 instances). The below five quotes (table 4.13.) provide examples of responses falling under the sub-theme “Education/ Comprehension” (relevant response segments enclosed by square brackets).

Table 4.13. Participant Responses Organized under Theme “Glycaemic Index” Sub- theme “Glycaemic Index: Education/Comprehension”

How would you describe your experience adding low GI foods to your diet? Response 1 “I immediately [added more yogourt, milk and parboiled rice to my diet and cut out warm potatoes.]” Response 2 “A little different. I almost never have breakfast (pre-study). I do take [100% Bran and low fat milk for the first meal of the day] (now).” Response 3 “By eating [sprouted 3 grain and 12 grain bread for my night snack,] my morning (fasting) sugar levels have dropped to under 4.0…” Response 4 “I have made [low GI cereals – whole grain oatmeal, bran buds,] all bran – for breakfast and [eliminated white bread]…” Response 5 “(I have made) [changes in cook method and temperature of food (e.g. pasta and potato).]”

14 Questions in this section of the Glycaemic Index Questionnaire have been developed to assess whether or not clients perceive glycaemic index education to be too hard to understand and/ or apply.

133

The below two quotes (table 4.14.) provide examples for the sub-theme “GI Application: Food Selection” (relevant response segments enclosed by square brackets). Also, responses 1 to 4 in table 4.13. would also fall under this sub-theme.

Table 4.14. Participant Responses Organized under Theme “Glycaemic Index” Sub- theme “Food Selection”

How would you describe your experience adding low GI foods to your diet? Response 1 “[Slowly making more low GI choices and experimenting with things like quinoa, kasha, more veggies, etc].” Response 2 “[I was already eating mostly Low GI Foods, so there is not a great change].

The below three quotes (table 4.15.) provide examples for the sub-theme “GI Self-efficacy” (relevant response segments enclosed by square brackets).

Table 4.15. Participant Responses Organized under Theme “Glycaemic Index” Sub- theme “GI Self-efficacy”

How would you describe your experience adding low GI foods to your diet? Response 1 “[Somewhat difficult – I am addicted to sweets – high GI – so I have to be very conscious all the time – self cooking no problem eating at home difficult].” Response 2 “[Good. I am trying.] Bread with low GI is hardest to find.” Response 3 “[Relatively easy to follow.] Blood sugar appears to be in acceptable range.”

Responses categorized under the theme “Feedback from Participants” (20 instances identified) showed participants’ feedback was generally positive (11/ 20). Eight statements identified were constructive (or critical) in nature and were included either under the sub-theme(s) “Revision Required: Indicator of Protocol Revision” (6 instances) or “Revision Required: Indicator of Questionnaire Revision” (2 instances). Three response examples, which are comprised of statements falling under “Revision Required: Indicator of Protocol Revision” (n=3), are included below in table 4.16.

134

Table 4.16. Participant Responses Organized under Theme “Feedback from Participants” How would you describe your experience adding low GI foods to your diet? Response 1 “…not enough time has past (passed) to be indicative of substantive change or not].” Response 2 “[This process takes time to incorporate low GI recipes and to grocery shop for new ingredients. e.g. Still on same loaf of white bread (as at baseline)].” Response 3 “[I bought a bag of quinoa, but I haven’t cooked any yet…]”.

Responses falling under the sub-theme “Revision Required: Indicator of Protocol Revision” highlight that participants felt that the length of time between visits and/ or of the study period did not provide them with adequate time to translate their knowledge to action. This was especially noteworthy at the second follow-up, where the participants often had a week or less to make changes to their diet before following-up with study staff to submit their diet record and to complete the GIQ. Despite this, the participants significantly reduced their dietary GI by week 1 post-education (table 4.11).

135

Table 4.17. GIQ Section 4a Is Your Low GI Diet Working For You?: Close-end Questions (Q) and Responses Q / Answer Type Q # Q (# of respondents) Response Options Visit 2.0 Visit 3.0 Count (%) Count (%) True/ False 1 Since the GI class, I have added low GI foods to True 18 (75) 21 (87) Dichotomous my diet. (n = 24) False 6 (25) 3 (13) Multiple Choice 2 Since the GI class, what percentage of your total ≥ 51% 9 (37) 11 (46) intake of starchy food has been low GI? (n = 24) < 51% 15 (63) 13 (54) Likert Scale 4 Please choose the word below that best ≥ Good 19 (79) 17 (71) Multiple Choice describes your ability to choose low GI foods in < Good 5 (21) 7 (29) the supermarket. (n = 24) Likert Scale 5 How would you rate your skill at choosing low GI ≥ Good 13 (54) 15 (63) Multiple Choice foods when eating out of the home? (n = 24) < Good 11 (46) 9 (37) Likert Scale 6 How would you rate your skill at using low GI ≥ Good 16 (67) 18 (75) Multiple Choice foods in meal planning? (n = 24) < Good 8 (33) 6 (25) Likert Scale 7 How would you rate your ability to make ≥ Good 15 (65) 18 (75) Multiple Choice traditional meals with low GI foods? (n = 23) < Good 8 (35) 6 (25) Yes/ No 8 Since the class, have the people you live with Yes 8 (44) 10 (67) Dichotomous been eating low GI foods? (n = 24) No 10 (56)15 5 (33)16 Likert Scale 9 Overall, how would you say your house mates ≥ Good 11 (58) 4 (25) Multiple Choice (i.e., family, partner, friends, etc.) would rate the < Good 8 (42)17 12 (75)18 low GI foods? (n = 24) True/ False 12 Planning low GI meals does not require more True 16 (70) 18 (75) Dichotomous time than planning other meals. (n = 23 at V2; 24 False 7 (30) 6 (25) at V3) True/ False 13 Low GI foods cost the same as other foods. (n = True 17 (74) 18 (75) Dichotomous 23 at V2; 24 at V3) False 6 (26) 6 (25) Yes/ No 14 Do you think you will continue to eat low GI foods Yes 23 (96) 24 (100) Dichotomous after this study is over? (n = 24) No 1 (4) 0 (0) GI = glycemic index; GIQ = Glycaemic Index Questionnaire, Visit 2 and 3 are both post-education; ≥ = greater than or equal to, ≤ = less than or equal to

15 Six participants selected “this question does not apply to me”; remaining percent expressed /18. 16 Nine participants selected “this question does not apply to me”; remaining percent expressed /15. 17 Five participants selected “this question does not apply to me”; remaining percent expressed / 19. 18 Seven participants selected “this question does not apply to me”; remaining percent expressed / 16.

136

4.5. CONCLUSIONS AND DISCUSSION

The GIEES included two phases. Phase 1 included the development of a face and content validated questionnaire; based on the KM hybrid evaluation tool template and designed to evaluate the GI education platform. Phase 2 was designed to pre-test this questionnaire and evaluate the GI education platform using the KM. The GI education platform was developed to provide participants, living with DM, with knowledge, support and skills to lower their dietary GI (change their behaviour). A key strength of this work was the rigour exerted to ensure both phases of the study were rooted in evidence-based approaches for questionnaire and education development and evaluation.

The study sample was representative of the St Michael’s Hospital DCCP patient population in that approximately 50% of our participants were born in another country and immigrated to Canada. St Michael’s Hospital is nationally/ internationally recognized for its service to immigrant/ multi-cultural populations. The average year of immigration ranged from 1952 to 2008. The majority of the sample (63%; 18/ 29) identified as European, however, and 83% (24/29) of participants spoke English as a main language at home; which may be more reflective of the exclusion criteria than any other factor.

Scientists interested in food environment and the relationship between socio-demographics and food choice, write about the concept of the “gate keeper”, or the person who is the primary food purchaser/ preparer. This person is often highlighted in behaviour change research as an important target for nutrition education and behaviour change support (169,170,409-411). This said, 77% (n = 20/26) of GIQ respondents identified as the main purchaser of food in their home and 68% respondents (19/ 28) identified as the main meal planner/ preparer (table 4.5.). Traditionally, women were the main gate keeper; responsible for meal planning, purchasing and preparation (170,412). The GIEES sample, made up of 69% (20/ 29) men, however, contradicts this traditional observation and may be reflective of demographic shifts in traditional gender roles related to food.

137

Overall, participants were satisfied (KM level 1) with the GI education platform. As highlighted in the section 4.4.2.1, the results generated using GIQ Section 1 (Participant Satisfaction; KM level 1) were overwhelmingly positive. For instance, options “disagree” and “strongly disagree” were not chosen by participants in response to any of the statements provided in questions 4 (table 4.6.). Similar data were generated by GIQ Section 4a (Application and Acceptability) and were indicative of diet sustainability. For instance, in response to the question “Do you think you will continue to eat low GI foods after this study is over?”, 96% of study participants (24/29) answered “yes” (one week post-intervention) and 100% participants (24/29) answered “yes” (four weeks post-intervention) (table 4.17). Qualitative data, collected with both of these GIQ sections, were also generally positive. So much so, that a theme “General Positive Feedback” was developed. That said, the qualitative data did provide valuable constructive feedback that was integrated either immediately (e.g. “Start @ 8:30 or after 5pm.”) or post- study (to inform future iterations of the GIQ and education platform).

According to the KM, the reported participant satisfaction (KM level 1) supports the significant increase in knowledge score (KM level 2) observed post-education (pre-education: 53.5 ± 5.1%, post-education [average]: 86.2 ± 3.3%; p < 0.05) (81). Not only was a significant increase in knowledge score observed immediately after the education session (baseline visit), but this knowledge score was maintained until the end of the study (four weeks); indicating sustained knowledge uptake. The pre-education score (~54%) was lower than expected by many participants; illustrated by 83% (24/ 29) of participants reporting hearing about GI before joining the study and 66% (19/ 29) reporting “knowing what GI is”. This was not surprising, however, as 83% (24/ 29) of study participants reported they had never been taught about GI by a health professional before. This is despite 90% of participants reporting that they had met with an RD to talk about diet before joining GIEES; findings supportive of the existing literature on RDs use of GI in DM management (62). In fact, our findings suggest that GI may be used less by RDs in the DCCP in comparison to other Canadian RDs.

138 significantly different between visits. However, the intervention brought carbohydrate and fat (as a percentage of daily total energy) within the AMDR, while protein was within the AMDR from baseline to study-end (117). Moreover, caloric, dietary carbohydrate and fat intake (in grams) decreased post-education (table 4.11.). Fibre intake was maintained throughout the study (no statistically significant difference between visits), however, indicating the dietary intervention successfully controlled for fibre intake. The mean (± SEM) change in dietary GI was 4 ± 1 GI units (table 4.11.); a decrease that resulted in a dietary reclassification from medium GI to low GI; indicative of behaviour change (KM level 3). Baseline dietary GI was 58 ± 1 GI units in this sample; supporting participants’ perception that they may already be consuming a lower GI diet (they were consuming a diet on the lower end of medium). This is illustrated by one participants’ statement related to his/ her experience with adding low GI foods to his/ her diet (table 4.14):

“I was already eating mostly Low GI Foods, so there is not a great change.”

Also highlighted in the literature review, a decrease in 5 to 9 GI units has been associated with improvements in T2DM treatment outcomes (e.g. glycemic control, weight management). This unit decrease in dietary GI has been achieved with substitution of 60% of carbohydrate foods in the diet (40,47,166,167,200,340). Grant et al. (2011) were able to facilitate a 9 unit decrease in dietary GI in women living with GDM and receiving standard care treatment at St Michael’s Hospital. This change in dietary GI resulted in a higher number of self-monitored blood glucose values within target range. The baseline dietary GI of the GIEES sample was 58 ± 1 GI units, which is identical to the pre-education dietary GI observed by Grant et al. (2011). A potential reason for the relatively small decrease in dietary GI observed in the GIEES sample was study length (diet records were completed at within one week and four weeks post-education), which was short compared to other published GI studies (40,46,47,200). That is, it has been reported that four to six weeks of adherence to the low GI diet are required before biochemical changes become apparent (46,200). GIEES participants reported feeling that they did not have sufficient time to make dietary changes; illustrated by the following participant quotes (table 4.16):

“…not enough time has passed to be indicative of substantive change or not.”.

139

“This process takes time to incorporate low GI recipes and to grocery shop for new ingredients. e.g. Still on same loaf of white bread (as at baseline). ”

These findings are further supported by GIQ section 4a (application) data, where 63% of participants reported < 51% of their total intake of starchy food was low GI at one week post- education and 54% reported < 51% of their total intake of starchy food was low GI at three to four weeks post-education.

The GIQ (and GIEES protocol) was also developed to assess if participants found GI too difficult to understand and apply (perceived barriers to GI utility 3 and 4). The achievement of a statistically significant increase in knowledge and behaviour change during the short study period may be reflective of the level of education reported by the sample. That is, 87% (26/29) of respondents reported completing a college certificate or diploma, undergraduate degree, or graduate degree at a Masters level. Moreover, 90% (36/29) of study participants reported receiving medical nutrition therapy (standard care) prior to GIEES. Although establishing pre and post differences in dietary GI in such a well-educated (or well-treated) sample presents some challenges, this sample was a very good group of people to involve in the GIEES (40,71,122). Their prior learning brought a very important perspective to GIQ development/ pre- testing and GI education evaluation.

GIQ Section 4a was designed to measure participants’ perception of their own skill/ application, self-efficacy, time and monetary burden, and participants’ perception of the effect of their diet change on others in their life (Section 4.4.5.0. for data). This section also measured participant intention to continue with the low GI diet post-study and participants’ acceptability of the food included in the food list. Although this questionnaire section did/ does not offer a direct measure of behaviour change, these data provide insight into the factors influencing participants move from knowledge uptake (KM level 2) to behaviour change (KM level 3) (81,214,219,220). Moreover, there were not significant changes in response between one and four weeks post- education.

140

According to GIQ section 4a, study participants were able to apply the education received and found the low GI foods satisfactory (table 4.17). By four weeks post-education, 75% of respondents reported that low GI foods did not take more time to prepare and did not cost more. These findings were in agreement with an Australian study that showed women were able to follow low GI dietary advice during pregnancy with no significant increase in the daily costs (daily cost of diet pre-education [AUD; $] 9.1 ± 2.7, post-education [AUD] 9.5 ± 2.1; p = 0.52)19 (395). Moreover, by week one 75% (18/24) of respondents reported adding low GI foods to their diet and by week four 88% (21/24) did. By week one, 79% (19/24) of respondents reported being good, very good, or excellent (≥ Good) at choosing low GI foods in the Supermarket. Moreover, 70% of respondents rated their ability to make traditional meals with low GI foods as good, very good or excellent. These data were supported by the aforementioned four unit difference in dietary GI seen post-intervention in comparison to baseline. On the other hand, participants reported experiencing difficulty choosing low GI foods when out of the home; 54% (13/24) at week 1 and 63% (15/24) at week 4. This finding was supported by qualitative data analysis and is illustrated by the following quote:

“Good experience and not difficult. More difficult when travelling.”

When at home, the majority of participants (71%; n = 23); felt their skill at using low GI foods in meal planning was good, very good or excellent. Despite this, there were participants that found decreasing intake of higher GI foods a challenge, as illustrated by the following quote:

“…Somewhat difficult – I am addicted to sweets – high GI – so I have to be very conscious all the time – self cooking no problem....”

Two questions were asked to collect information on the participants’ immediate community (table 4.17); those who lived alone answered not applicable to this question. By one week post- education, 44% (8/18) of participants reported that people they lived with were eating low GI foods. Although not a statistically significant increase, at four weeks post-education 67% (10/ 15) reported that people they lived with were eating low GI foods. Interesting to note, as the

19 Canadian currency ($): Daily cost of diet pre-education: 8.98 ± 1.98, post-education: 8.6 ± 2.7; p = 0.52)

141 number of housemates consuming low GI foods increased, the number of participants reporting that their housemates would rate low GI foods as good, very good or excellent decreased (although not significantly) (week 1: 11/19, week four: 4/ 16).

To conclude, GIEES participants receiving the GI education platform in a group setting at St Michael’s Hospital were satisfied with the GI-education (KM level 1), showed an increase in GI-knowledge (KM level 2), and demonstrated increased consumption of low GI foods post- intervention/ education (KM level 3). Moreover, the results of the GIQ did not support the statement “GI is too hard for clients to understand and apply” (perceived barriers 3 and 4) (62,63). In fact, GIEES data show the opposite; that participants were able to increase their knowledge score and reduce their dietary GI post-education.

GIEES has evaluated an evidence-based GI education platform designed to be layered onto standard care medical nutrition therapy for DM and has addressed educators need for reliable education materials (62,63,133). This study has also pre-tested a questionnaire designed to evaluate this GI education platform (GIQ). As discussed in the literature review, the CDA supports the layering of low GI education onto standard care in T2DM (144). However, such guidelines do not currently exist for GDM standard care and there are limited data on GI utility in the GDM client population (14,40,64,67,71,72,128,161). Therefore, Study 3 (an RCT building on GIEES) was developed to comprehensively evaluate if the low GI education platform improved glycaemic control (main standard care target; KM level 4 - Result) in women with GDM receiving standard care. KM level 1 to 3 were evaluated using the GIQ pre-tested in GIEES and three day diet records. As part of the face/content validation and pre-test that occurred during GIEES, data were collected to improve materials for use in Study 3. For example, GIEES participants’ concerns related to not having sufficient time in which to make dietary change informed the length of the RCT post-education period, which was ~18 weeks long. Moreover, the GI education platform was supplemented with strategies for choosing lower GI foods when dining out and more recipe ideas.

142

CHAPTER 5.0. STUDY # 3 THE EFFECT OF A LOW GLYCAEMIC INDEX DIET ON MATERNAL AND NEONATAL MARKERS OF GLYCAEMIC CONTROL AND POSTPARTUM DIABETES RISK (Short Title: GI in GDM Study)

142

143

CHAPTER 5.0. STUDY # 3 THE EFFECT OF A LOW GLYCAEMIC INDEX DIET ON MATERNAL AND NEONATAL MARKERS OF GLYCAEMIC CONTROL AND POSTPARTUM DIABETES RISK

5.1. ABSTRACT

Low glycaemic index (GI) education is recommended by the Canadian Diabetes Association for management of type 1 and 2 diabetes mellitus (DM), but not for management of gestational diabetes mellitus (GDM). Evidence supports use of low GI in GDM treatment, but more data are needed to influence standard care. An RCT was developed to assess the effect of a low GI diet on maternal postprandial self-monitored blood glucose (PP-SMBG) in women with GDM. In addition to PP-SMBG, intervention evaluation included administration of a pre-tested mixed- form questionnaire (Glycaemic Index Questionnaire/ GIQ©) and three-day-diet-record (DR). Seventy-four women, receiving standard care for GDM, were randomized to either standard care (n = 40) or a low GI diet (n = 34). Participants attended three prenatal and one post-natal visit, at which they provided GIQs and DRs. PP-SMBG were provided during pregnancy. Ninety-nine percent of study respondents (73/74) were satisfied with GI-education. Baseline GI-knowledge and dietary-GI were not different between groups. Knowledge score (mean±SEM) significantly increased from pre-education (47±3%) to immediately post-education (88±3%; p ≤ 0.0001) in the low GI group; this score was maintained until study-end. Dietary-GI significantly decreased from pre-education (57±0.6) to two weeks post-education (51±0.6; p ≤ 0.001); a difference that was maintained until study-end. Average PP-SMBG was lower on low GI (low GI: 6.02 ± 0.03 vs. standard care: 6.10 ± 0.02; p = 0.041), but diet did not influence the percentage of PP-SMBG ≤ 6.7 mmol/L. The GI education platform satisfied participants and supported an increase in GI- knowledge, a reduction in dietary GI, and an improvement in glycaemic control in women living with GDM. These data provide support for use of low GI foods in GDM management and this study has evaluated a GI education platform (and GIQ) that can be used in subsequent studies and in clinical practice.

Funded By: Canadian Diabetes Association, Canadian Institutes for Health Research, Canadian Foundation for Dietetic Research Clinicaltrials.gov Identifier: NCT01589757

144

5.2.0. INTRODUCTION

5.2.1. Study Rationale

Management of GDM focuses on improving glycaemic control with lifestyle modification, but therapy can include medication (often insulin). This approach has been shown to be effective in achieving glycaemic control and reducing GDM complications (14). Despite this, pregnant women are sometimes not able or willing to participate in physical activity and in some clients use of insulin has been linked to physical, emotional and financial stress. Moreover, insulin does not rectify peripheral insulin resistance or address behaviour change (important to long term maternal health) (14,162-165,413-415). The CDA CPG (2013) includes low GI education as part of medical nutrition therapy of type 1 and 2 DM, but not as part of medical nutrition therapy for GDM (144). In general, low GI foods allow postprandial blood glucose to be reduced without reducing carbohydrate intake. This approach is not only consistent with sound nutritional principles for the management of GDM, but also has the potential to address one of the main therapeutic objectives - reducing postprandial glucose (14,29,117). There is, however, limited data on GI utility in this client population; as highlighted in the literature review (40,64- 66,68,70). Moreover, clinicians are looking for more data on GI utility and GI mechanism in diverse client populations with or at risk for DM (26,62,63,66,133,148). To address these gaps, this RCT was developed to investigate whether low GI education may be able to improve glycaemic control in women living with GDM.

As highlighted in the literature review, Canadian RDs are interested in suitable GI education materials (62,63,133). Traditionally, GI clinical utility has been evaluated in the context of an RCT using clinically relevant outcomes, like glycaemic control, without comprehensive evaluation of the GI education (40,59-61,166,416). Education-based and behaviour-based interventions have been recognized for their complexity and methods have been proposed for comprehensive intervention evaluation in the context of the RCT design (80,81,83,85,192,214, 343,344,352). A theoretical framework that can provide a comprehensive evaluation of education-based interventions is the KM. The KM includes evaluation of: (1.) participant satisfaction, (2.) knowledge uptake/ score, (3.) behaviour change, and (4.) results (e.g. glycaemic control) (80-83). During GIEES, the GIQ and GI education platform were developed

145 and pre-tested. The GIQ was developed to evaluate the GI education platform in people living with DM (pre-tested in people living with T2DM). In this study, this questionnaire will be used to evaluate (and improve) the GI education platform in women living with GDM.

5.2.2. Study Hypotheses

Hypothesis: A low GI education platform will improve postprandial glycaemic control in women with GDM in comparison to those receiving standard care (KM level 4).

Supporting (or secondary hypotheses): Study participants will: 1. Be satisfied with the GI education (KM level 1) 2. Show an increase in GI knowledge post-education (KM level 2) 3. Significantly lower their dietary GI post-intervention (KM level 3)

5.3.0. MATERIALS AND METHODS

5.3.1. Study Design and Randomization

The GI in GDM Study was a prospective RCT including two arms (parallel): (1.) standard care group and (2.) low GI group. Participants, stratified by centre, were randomly assigned the low GI or control (medium GI) diet using blocks of various sizes to enhance allocation concealment (417,418). All aspects of this study were reviewed and approved by the University of Toronto, Office of Research Ethics, St Michael’s Hospital Research Ethics Board, Sunnybrook Health Sciences Centre Research Ethics Board, Mount Sinai Hospital Research Ethics Board, and St Joseph’s Healthcare Hamilton Research Ethics Board. Trial details were also reviewed, approved and published by clinicaltrials.gov (Identifier: NCT01589757).

5.3.2.0. Sample

5.3.2.1. Eligibility Criteria

All study participants were receiving treatment for GDM at one of the following four DIP Clinics in Southern Ontario: (1.) St. Michael’s Hospital, Toronto, (2.) Mount Sinai Hospital,

146

Toronto, (3.) Sunnybrook Health Sciences Centre, Toronto, (4.) St. Joseph’s Healthcare Hamilton, Hamilton. Patient screening was completed by a DIP clinic-based RD at the first standard care appointment. At this appointment, patients attend a group class on GDM and are introduced to the logistics of their treatment path, which typically concludes at 6 to 8 weeks postpartum. Eligible patients were invited, by the clinic RD, to stay after class to meet with the Study Coordinator/ Study Staff, who reviewed the letter of information in efforts to confirm eligibility and patient interest before introducing the consent form. Recruitment for GI in GDM started on October, 2011, and completed on May, 2014. Due to delays in REB approval, at our main study site (St Michael’s Hospital), only 15 participants were recruited from November, 2011, to January, 2013. The eligibility criteria are outlined below:

Inclusion criteria Women: (1.) ≥ 18 years of age (2.) diagnosed with GDM or IGTP according to CDA criteria20 (14,106) (3.) being followed within one of four pre-selected DIP Clinics (4.) willing and able to give informed consent (5.) willing and able to comply with the study protocol

Exclusion criteria Women: (1.) with acute or chronic illness other than GDM or IGTP or use of drug (other than insulin) which may affect carbohydrate metabolism, gastrointestinal function or carbohydrate digestion (e.g. crohn’s disease, liver disease, kidney disease etc.). (2.) known to have type 1 or type 2 DM prior to pregnancy (3.) known to have multi-fetal pregnancy at enrolment (4.) ≥ 33 weeks gestation (5.) prescribed oral anti-hyperglycaemic medication (6.) with insurmountable language barriers

20 Until the 2013 CDA CPG was released, both women with GDM and IGTP were recruited. Unlike Grant et al. (2011), participants were not stratified by diagnosis.

147

5.3.2.2. Power Analysis/ Sample Calculation

The pilot study showed the mean percent-on-target self-monitored blood glucose was (mean ± SD) 45.5 ± 18.5% for the 22 participants on the medium/ high GI diet and 54.2 ± 21.7% for the 19 participants on the low GI diet (40). These data yield a pooled standard deviation of 24.4 and a mean difference of 8.7%. A difference of 8.7 in percent-on-target self-monitored blood glucose was considered clinically significant because it was 19% higher than the percentage of self-monitored blood glucose values on target with standard therapy (control group). To detect a difference of 8.7, the number needed per group/ arm for 80% power and a two-tailed p-value < 0.05 was derived as follows: n = (16s²)/d²; therefore n = 16×(24.4)²/(8.7)² = 126 (ie. 252 women in total). With respect to the secondary endpoints, this sample provided 80% power to detect a 0.25 mmol/L difference in postprandial self-monitored blood glucose (mean ± SD; difference in the pilot study was 0.08 ± 0.72 mmol/L). In the pilot study the drop-out rate was 12%. Assuming a more conservative 20% drop-out rate, we aimed to randomize 300 participants in total (ie. approximately 150 per group). Table 5.1. lists the target sample for each study site. Site-specific target samples were calculated based upon DIP attendance statistics, ongoing and up and coming studies that would be competing for participants from the same recruitment pool, and an estimated 1 to 1.5 year recruitment period.

Table 5.1. Recruitment targets for each hospital site

Site Target Sample St Michael’s Hospital 95 Mount Sinai Hospital 60 St Joseph’s Hospital 65 Sunnybrook Health Sciences Centre 80 Total Participant Count 300

148

5.3.3. Study Outcomes

The primary objective (outcome) was to determine whether women with GDM who received low GI education would obtain a higher percentage of postprandial self-monitored blood glucose within the clinically acceptable range than the control group (those receiving standard care medical nutrition therapy) post-education. The CDA CPG (2013) reference ranges were used for analysis (< 6.7 mmol/L) (14).

Greater than 40 secondary outcomes were collected for this multi-centered trial. A complete list of maternal and neonatal study outcomes can be found in appendix 5.1. For this dissertation, analysis has been conducted on the following maternal secondary outcomes:

(1.) Average postprandial self-monitored blood glucose (2.) Dietary energy, macronutrients, fibre, and GI (3.) Participant satisfaction with the GI education platform, (4.) GI knowledge score pre and post GI education.

Demographic information was also collected. The study outcomes have been organized according to KM in figure 5.1. and data will be presented in the results section this order (80,81). This theoretical framework for study development and evaluation was discussed in detail in Chapter 2 of this dissertation.

Figure 5.1. Study Outcomes According to Kirkpatrick Method (Four Levels)

Postprandial Glycaemic Level 4 – Results Control

Change in Dietary Level 3 – Transfer Glycaemic Index

Glycaemic Index Level 2 - Learning Knowledge Score

Participants' Satisfaction with Level 1 - Reactions Glycaemic Index Education Platform

149

5.3.4.0. The Study Intervention: Low Glycaemic Index Education

Two education platforms were developed, implemented and evaluated: (1.) Standard care education for women living with GDM and (2.) Low GI education. Both were informed by GIEES (Chapter 4 of this dissertation) and the GI in GDM pilot study results (40). Regardless of group assignment, all participants attended a 30 minute education session facilitated by one of four trained RDs. The education platform was developed so that it could be delivered in a group (maximum 5 participants) or one-on-one setting. The four RDs delivering GI education were involved with education development and evaluation. One was the RD who oversaw delivery and evaluation of GIEES and the pilot study; she was responsible for training, support and assessment of the other three RDs during the GI in GDM Study.

5.3.4.1. Standard Care Education

Those in the standard care group received education that reviewed information and messages that participants would have received at their first DIP standard care appointment. Therefore, the education session included a review of current Canadian recommendations for healthy eating and were guided by the CDA CPG (2008, 2013) (14,106). The following publically available education tools were used to facilitate this education:

(1.) Eating Well with Canada’s Food Guide21: http://www.hc-sc.gc.ca (2.) The Plate Method or Diabeters Portion Plate®: www.diabetes.ca; www.Diabeters.com (3.) Three-dimensional food models: e.g. https://spectrum-nasco.ca

Members of this study group also reviewed and received a high fiber (medium to high GI) food substitution list. Review of this list was facilitated by an RD using The Standard Care PowerPoint Presentation. Participants were given hard copies of the slide deck for reference. They were also given a forty-one page high fibre recipe book that was reviewed at the baseline visit and participated in “The Plate Game”; described in more detail below (section 5.3.4.3.).

21 The Diabetes Food Guide was used in concert with Eating Well with Canada’s Food Guide; differences in serving sized between tools were highlighted.

150

5.3.4.2. Low Glycaemic Index Education

Low GI education was layered onto standard care education as per CDA CPG (2008, 2013). The three participant education materials that were used to facilitate education translation were:

(1.) Low Glycaemic Index Food Substitution List (2.) Low Glycaemic Index PowerPoint Presentation (3.) Low Glycaemic Index Recipe Book

(1.) Low Glycaemic Index Food Substitution List

The GI in GDM iteration of the Low Glycaemic Index Food Substitution List was based on the food substitution food list used by Grant et al. (2007, 2011) in a sample of participants receiving care for GDM and informed by GIEES results. This food list included three GI categories, as published in CDA’s The Glycemic Index Tool (low GI = 55 or less, medium GI = 56 to 69, and high GI = 70 or more) (30). The stoplight method was used to distinguish between the three GI categories (Green = Low GI = Choose Most Often, Yellow = Medium GI = Choose Occasionally, Red = High GI = Choose less often). This method for distinguishing GI categories was suggested by Slabber (2005) and has been used by others in nutrition education/ communication (196,398). Using food substitution lists to guide client/ participant intake pattern is the “key foods strategy”. This strategy has been recognized as effective in achieving moderate modifications in food intake. By asking participants to use food substitution lists and work with RDs to personalize their lists, scientists and educators ensure that foods with a tested GI are used for food substitution (decreasing the risk of subjective GI value assignment; discussed in the literature review) (40,54,59,61,208,215). Participants were asked to aim to substitute as many carbohydrate-containing foods as they could, but were assured that one substitution per meal was sufficient to lower dietary GI. In fact, a significant decrease/ difference in dietary GI has been achieved by replacing 60% of starchy foods (in a diet made up of 45 to 65% carbohydrate) with lower GI carbohydrate (40,47,166,167,200,340).

151

Until GIEES, our laboratory only included starchy foods in our low GI education and food substitution lists. The practice of including all food groups in GI Education is an approach used in Australian GI education (Brand-Miller and colleagues) (31). This practice was adopted to provide participants with more options, to communicate the importance of layering GI education onto current recommendations and to create a study intervention more likely to create differences in dietary GI. This client education material was intended for provision to study participants after comprehensive review with the education facilitator. The GI in GDM version of the Low Glycaemic Index Food Substitution List can be found in appendix 5.2.

(2.) Low Glycaemic Index PowerPoint Presentation

The PowerPoint slide deck was developed for the educator (e.g. RD, RN) as a job aid to facilitate review of the Low Glycaemic Index Food Substitution List and other relevant GI education (e.g. slow absorption model). This presentation walks the client (and the educator) through how to integrate low GI food choices into current CPGs for management of T2DM. The notes section of each PowerPoint slide includes noes and strategies for content delivery. The slide deck was also created to act as a client education material; developed to be provided as a handout (2 to 3 slides per page; notes section omitted). The GI in GDM version of the Low Glycaemic Index PowerPoint Presentation can be found in appendix 5.3.

(3.) Low Glycaemic Index Recipe Book

The recipe book is based on the recipe book developed for the pilot study. It has been updated using feedback provided by pilot and GIEES participants. This participant education material provided participants with ~ 36 low GI recipes under the headings of breakfast, sides and entrees (appendix 5.4.). It also provides serving recommendations and nutrition information for each recipe and a section for addition of new recipes. This job aid was developed to facilitate communication between participant and RD and to generating new recipes for future iterations. RDs using this job aid during the pilot and GIEES both report that it was particularly useful in building rapport with clients surrounding traditional foods/ recipes and food preferences.

152

5.3.4.3. The Plate Game

The Plate Game is an activity that was used to facilitate participant learning; regardless of diet assignment. Based on “The Plate Method” (standard care; figure 4.1.), RDs used three- dimensional food models and a three-way divided dinner plate to cultivate problem solving and meal planning skills in participants (30). This activity involved first asking participants to create a typical meal (breakfast, lunch, dinner) that they would have made or ate prior to their diagnosis and initial standard care education session. Then, participants were asked to create a plate that reflected current recommendations (high fibre substitutions). If on the low GI diet, participants were then asked to create a plate that reflected current general dietary guidelines and included at least one low GI food substitution. The activity was facilitated by the RD, whom was trained on various meal planning combinations that she could use as examples during the activity.

5.3.5.0. Data Collection

Table 5.2. illustrates the data collection timeline. The following paragraphs outline the data collection procedures for each of the outcomes of interest.

5.3.5.1. Self-Monitored Blood Glucose

Study participants were asked to self-monitor their blood glucose, during the prenatal period, as part of standard GDM treatment. Study participants kept a written record of their blood glucose results for study staff to photocopy at each study appointment. Self-monitored blood glucose results were cross-referenced with the most recent values in participants’ glucometer, to ensure reporting accuracy.

5.3.5.2. Three Day Diet Record: Dietary Intake Data Collection and Analysis Procedures

A three day diet record (with instructions) was used to collect dietary intake data (appendix 4.2) at baseline and the three follow-up visits (table 5.2.). Each diet record was reviewed, in person, by an RD, with the study participant, to ensure the record was complete. Diet intake data were

153 entered into ESHA Research’s Food Processor® Nutrition and Fitness (0.14.0.; Salem, Oregon). This program uses its own Nutrition Information Database (http://www.esha.com/nutrition/ database-information) which is composed of data from various sources; including the Canadian Nutrient File (400). To supplement this database, the Wolever Laboratory has developed an evidence-based GI database (Appendix 2.1 provides and excerpt from the ESHA Manual of Operations, developed to standardize the process of GI assignment.). When data entry was complete, data was exported to Microsoft® (MS) Excel® 2013 for data cleaning and preparation for statistical analysis using IBM SPSS version 21 (Copyright © IBM Corporation and other(s) 1989, 2012) . Diet records were analyzed for energy, macronutrients (total dietary carbohydrate, fat and protein), fibre, and dietary GI.

5.3.5.3. The Glycaemic Index Questionnaire (GIQ©)

Based on the KM hybrid evaluation template and peer-reviewed questionnaires/ surveys extracted from published research studies looking at GI education, the GIQ was developed to evaluate the GI education platform (38,40,55,59,63,80,81,207,365,366,395,401,402). It was face/ content validated and pre-tested as part of GIEES. The following paragraphs and table 5.3. provide an overview of the GIQ sections, questions and administration schedule used to collect data analyzed as part of this dissertation. The GIQ is a mixed-format (open- and closed- end questions) questionnaire with four sections aimed at evaluating GI education using the first three levels of KM. Only quantitative data from GIQ section 1 to 3 will be presented and discussed in this dissertation chapter. GIQ sections 1 to 3 are outlined below.

Section 1 (Participant Satisfaction): A participant-administered mixed-format questionnaire developed to assess participants’ satisfaction or “Level 1: Reaction” to the low GI or standard care education (appendix 5.5.). Responses for question 4 (data included in the results section of this chapter) were presented as a horizontal five point Likert Scale, including text and graphic representations/ anchors of these options (1 = strongly agree to 5 = strongly disagree).

154

Section 2 (Demographic Information): An investigator-administered mixed format questionnaire designed to collect demographic information from all study participants (appendix 5.6.). This section is primarily comprised of vertical format multiple choice questions, including an “other” category. Open-ended and “other” responses have been and will be used in development of future iterations of the GIQ. “I am not sure” and “pass” were response options for select questions in this GIQ section. Examples of demographic outcomes examined during the GI in GDM Study include: Language spoken at home and ethnic group.

Section 3 (GI Knowledge): A participant administered close-end questionnaire intended to measure participants’ knowledge of key GI concepts (Level 2: Learning) covered during GI education. Learning was assessed by knowledge score (X/12; answers included in appendix 4.3.). Questions were multiple choice or true or false (dichotomous format) and marked as either correct (score of 1), incorrect (score of 0) or “I do not know the answer” (score of 0). In some cases a second best answer was available (score of 0.5). Section 3 was administered to participants in the low GI group at baseline and each follow-up visit, while it was only administered once (at baseline) to those in the standard care group.

5.3.7. Statistical Analysis Procedures

Data analysis was conducted using MS Excel® 2013 and IBM SPSS version 21 (Copyright © IBM Corporation and other(s) 1989, 2012). Descriptive statistics were used to analyse demographic and participant satisfaction data. These data will be presented as counts/ percents. Analysis of the effect of dietary assignment on percent postprandial self-monitored blood glucose within the clinically acceptable range (< 6.7 mmol/L 2 hours postprandial) was conducted using logistic regression (dependent outcome is dichotomous; Yes/ No). Average postprandial self-monitored blood glucose and dietary intake data were analyzed using SPSS Linear Mixed Model (ML). The linear mixed model (or linear growth model or individual growth model) is a variation of the standard linear model used within the GLM (375). Pre- pregnancy BMI, gestational age, and baseline blood glucose values were included as covariates in blood glucose data analysis. The Sidak model was used to compare main effects of treatment and for post hoc (time point) analysis. Postprandial self-monitored blood glucose (percentage within range and average) values, analyzed for this dissertation, were an average of blood

155 glucose values obtained two hours after breakfast, lunch and dinner. GI knowledge score (mean±SEM) was compared between groups at baseline (unpaired t-test). Within group GI knowledge score comparisons were conducted between repeated measures (“visits”) for the low GI group. These comparisons were conducted by applying a paired t-test, in the context of the close-testing procedure. All statistical tests were given a 2-tailed p < 0.05 criterion of significance.

156

Table 5.2. Study Data Collection Timeline

Weeks Gestation 28 29 30 31 32 33 34 35 36 37 38 39 40 2 4 6 8 Study Week 0 B 1 2 3 4 5 6 7 8 9 10 11 13 15 17 19 Study Visits 1 2 3 4 Self-monitored blood glucose X X X X X X X X X X X X Glycaemic Index Questionnaire (GIQ) X X X X Three day diet record X X X X B= baseline, Weeks 28-40 are prenatal while weeks 1-8 are postpartum; X = time when outcome was collected for all study participants

Table 5.3. Glycaemic Index Questionnaire (GIQ) Administration Timeline Study Period Questionnaire Purpose/ Outcome 2 Weeks 4 to 6 Weeks 6 to 8 Weeks Section Baseline Baseline (Kirkpatrick’s Model) Post-Edu Post-Edu Post-Edu Pre-Edu Post-Edu (Visit 2) (Visit 3) (Visit 4) To assess participants’ GI Education “reactions” or impact of 1 X*2 Satisfaction the learning environment (level 1) Demographic To collect 2 X*2 Information demographic data To assess GI knowledge (learning) before and 3 GI Knowledge X*2 X X X X after the session (level 2) GI = glycemic index, Pre-edu = pre-education/ intervention; X = time when outcome was collected for participants in low GI group; X*2 = time when outcome was collected for all study participants

157

5.4.0. RESULTS

5.4.1. Sample Characteristics

Ninety nine participants (n = 99) were recruited from the four hospital sites; 33% of target sample size. Outside of not meeting eligibility criteria, the most common reasons potential participants gave for declining study participation was “I’m not interested”, “I’m too busy”, or “I’m participating in another study”. The recruitment totals and targets for each site are included in table 5.4. below.

Table 5.4. Recruitment totals and targets by Hospital site

Site Sample Target Sample St Michael’s Hospital 33 95 Mount Sinai Hospital 23 60 St Joseph’s Hospital 29 65 Sunnybrook Health Sciences Centre 14 80 Total Participant Count 99 300

Figure 5.2. is a flow diagram that represents the study sample composition from recruitment to study completion. For the purposes of this dissertation and due to the small sample size, the sample will not be discussed in terms of hospital site stratification. Of the 1, 167 patients screened, 99 were recruited. Of the 99 women recruited, 25 withdrew from the study before the baseline visit. The most common reason given for withdrawal before the baseline appointment was “I’m too busy”. Of those who attended the baseline appointment (n = 74), five withdrawals were considered true drop outs as per the withdrawal criteria. Participants were considered closed-out or completed if they attended three of the four study visits; including the baseline visit. Of the 74 women recruited, 34 women were randomized to the low GI diet arm and 40 were randomized to the standard care arm. Of the five withdrawals, three were from the low GI arm and two were from the standard care arm. The most common reported reason for withdrawal was “I’m too busy”; difficulty with study diet was not specified as a reason for withdrawal. At study activity completion, 65 participants completed or closed-out the study and 4 participants were lost to follow-up.

158

Figure 5.2. Profile of study sample from recruitment to study completion

Recruited n = 99

Withdrawal before V1 n = 25

Randomized n = 74

Low GI Standard Care n = 34 n = 40

Drop Outs

n = 5 Low GI Standard Care n = 31 n = 38

Lost to Follow-up n = 4 Low GI Standard Care

n = 29 n = 36

Complete/ Close Out

n = 65

n = sample count, V1 = visit 1, Low GI = low glycaemic index

159

All questionnaire respondents were women (n=73). The sample was heterogeneous in terms of ethnicity (figure 5.3., table 5.5.). Twenty seven percent (20/ 73) of the sample chose the “other” category and described their ethnic identity as either a combination of ethnicity categories (e.g. European and Portuguese and German/ Dutch, European and Latin American) or using terms unavailable in the questionnaire (e.g. Canadian, American, Mixed Heritage). Of those who did not choose “other”, the largest categories identified were European (n = 16/ 73), East Asian (n = 12/ 73) and South East Asian (n = 11/ 73). Seven respondents identified as Latin American and four as African/ Caribbean. Two respondents selected Indian (or South Asian) and the Aboriginal response was selected by one participant. Both participant counts and percentages are shown below in table 5.5. Similar to the GIEES sample, 56% (40/ 71) of study respondents were born outside of Canada and immigrated to Canada over a wide range of years (1972 to 2012). The most common language spoken at home was English (64%; 47/73). The three non- English languages that were the most commonly selected as the language spoken at home include: (1.) 8% Chinese (75% Mandarin), (2.) 4% Spanish, and (3.) 4% Russian.

Figure 5.3. Glycaemic Index Questionnaire Section 2 Demographic Information: Response Summary to “Which ethnic group do you identify with?” (Percent)

22% Aboriginal 27% European African/ Carribean Latin American 6% Indian East Asian 15% 10% South East Asian Other 3% 16%

160

Table 5.5. Glycaemic Index Questionnaire Section 2 Demographic Information: Response Summary to “Which ethnic group do you identify with?” (Counts and Percent)

What ethic group do you Counts (x) Percent (%; x/n) identify with? Aboriginal 1 1 European 16 22 African/ Caribbean 4 6 Latin American 7 10 Indian 2 3 East Asian 12 16 South East Asian 11 15 Other 20 27

Also generated by GIQ Section 2, the below data provide insight into the sample’s previous medical nutrition therapy (table 5.6.). Seventy-nine percent (58/ 73) of the sample reported meeting with a RD before to talk about diet, 76% (55/ 72) reported hearing of GI before the baseline visit, and 47% (34/ 73) reported they knew what GI is.

Table 5.6. Previous Medical Nutrition Therapy Reported by Sample; Questions and Answers from Glycaemic Index Questionnaire Question Response Summary Count (Percent) Have you met with a dietitian before to talk Yes = 58 (79) about diet? No = 15 (21) Have you ever heard of the glycaemic index Yes = 55 (76) before today? No = 17 (24) Do you know what the glycaemic index is? Yes = 34 (47) No = 39 (53)

161

5.4.2. Participant Satisfaction (Kirkpatrick Model Level 1 - Reactions)

Seventy three participants completed GIQ Section 1: How Did You Like the Class?. Of this sample, 70 reported learning something new in the class and three reported not learning anything new. The three who reported not learning anything new were in the standard care group. Question 4 was made up of questions, to which participants were asked to select a response. The statements and response summary for question 4 can be found in table 5.7. There were no responses that reflected disagreement with the statements provided; the options disagree and strongly disagree were not selected. These findings support that KM level 2, participant satisfaction, was achieved using the GI education platform.

Table 5.7. Glycaemic Index Questionnaire Section 1: GI Education Satisfaction, statements and responses for question 4. Statement Response Summary Percent (Counts) The class content was easy to understand. 73% Strongly agree (52/ 71) 27% Agree (19/ 71) The class content was interesting. 63% Strongly Agree (46/ 73) 37% Agree (27/ 73) The teacher was easy to understand. 79% Strongly Agree (58/ 73) 21% Agree (15/ 73) The teacher’s assistants were easy to understand. 70% Strongly Agree (31/ 44) 25% Agree (11/ 44) 5% Neither Agree nor Disagree (2/ 44) The hand-outs helped me learn the class content. 56% Strongly Agree (40/ 72) 43% Agree (31/ 72) 1% Neither Agree nor Disagree (1/ 72) The presentation helped me learn the class content. 61% Strongly Agree (44/ 72) 39% Agree (28/ 72) The hands-on activities helped me learn the class 54% Strongly Agree (37/ 68) content. 41% Agree (28/ 68) 5% Neither Agree nor Disagree (3/ 68) I think that what I learned today will help me make 67% Strongly Agree (48/ 72) changes to my diet. 30% Agree (22/ 72) 3% Neither Agree nor Disagree (2/ 72)

162

5.4.3. GI Knowledge Score (Kirkpatrick Model Level 2 – Learning)

Knowledge score was not significantly different between groups at visit 1 or baseline (low GI: 47 ± 4 vs. standard care: 46 ± 3%; p = 0.774). As shown in figure 5.4., study participants scored significantly higher on the GI Knowledge Quiz (or GIQ Section 3: What do you know about GI?) immediately post-education or at visit 1.1. (88 ± 3%; p ≤ 0.0001), at visit 2 (83 ± 3%; p ≤ 0.0001), visit 3 (86 ± 3%; p ≤ 0.0001) and visit 4 (85 ± 3%; p ≤ 0.0001) when compared to baseline or pre-education (47±3%). There were no significant differences between GI knowledge scores post-education. These findings support that KM level 2, learning, was achieved using the GI education platform.

Figure 5.4. Total Knowledge Score at each administration of Glycaemic Index Questionnaire Section 3: Glycaemic Index (GI) Knowledge (by visit)

* * * *

n = 34 n = 30 n = 24 n = 23

Prenatal Prenatal Prenatal Postpartum

Low GI: n = 34 SC: n = 40

Mean (± SEM); * = significantly different from baseline score; p ≤ 0.0001; SC = standard care

163

5.4.4. Three Day Diet Record Data (Kirkpatrick Level 3 – Transfer)

Dietary intake data are shown in table 5.8 and 5.9. There were no significant differences in dietary energy, protein and fat within and between groups during the study period. There was, however, a main effect of diet assignment on dietary carbohydrate (low GI: 197 ± 8 vs. standard care: 223 ± 7 g; p = 0.016) and GI (low GI: 53 ± 0.3 vs. standard care: 56 ± 0.3%; p ≤ 0.001). A visit-diet interaction was not detected for carbohydrate (p = 0.938), but a visit-diet interaction was detected for GI (p ≤ 0.0001) and fibre (p = 0.023). There was, however, no main effect of diet on fibre (p = 0.932). Fibre intake was significantly lower in the low GI group at baseline (or visit 1) in comparison to the standard care group (low GI: 21 ± 1 vs. standard care: 24 ± 1 g; p = 0.008). There was no significant difference in fibre detected between groups at any other visit. Notwithstanding, fibre intake between groups approached significant difference at visit 3 (p = 0.05). Moreover, the low GI group experienced a significant increase in dietary fibre intake (within group) with significantly higher fibre at visit 3 in comparison to baseline (baseline: 21 ± 1 vs. visit 3: 25 ± 1 g; p = 0.035). At baseline, study participants were consuming a medium GI diet (56 to 69 %). In fact, participants joined the study at the low end of the medium range (low GI: 57 ± 0.6 vs. standard care: 56 ± 0.5 %), as per the CDA (30). The low GI intervention (GI education platform) resulted in a significantly lower diet GI than standard care at all post- intervention visits (table 5.8.). At visit 2 and 3 this difference between groups was ~ 5 units (p ≤ 0.001). Moreover, those on the low GI diet experienced a significant decrease in GI post- intervention. That is, the dietary GI of those on the low GI diet intervention was significantly lower at visit 2, 3 and 4 in comparison to baseline (visit 1). Conversely, those who received only standard care did not display a change in dietary GI during the study period. These findings support that KM level 3, transfer, was achieved using the GI education platform. Table 5.9. includes a summary of participants macronutrient composition. This analysis was done to assess if both diets stayed within the current AMDR (117). Data is presented as a percentage of total daily caloric intake. No significant changes were detected within groups during the study period.

164

Table 5.8. Results of three day diet record data analysis

Dietary Intake Data Outcome Day Standard Care Low GI (Mean ± SEM) (Mean ± SEM)

Calories (Kcal) Visit 1 1834 ± 71 1848 ± 61 Visit 2 1887 ± 74 1773 ± 71

Visit 3 1830 ± 82 1882 ± 78 Visit 4 1842 ± 91 1774 ± 86

Overall 1849 ± 37 1819 ± 40 Carbohydrate, total (g) Visit 1 212 ± 12 197 ±14 Visit 2 217 ± 14 187 ± 15

Visit 3 241 ± 15 210 ± 16 Visit 4 221 ± 17 194 ± 18

Overall 223 ± 7 196 ± 8 * Fibre (g) Visit 1 24 ± 1 21 ± 1 * Visit 2 24 ± 1 24 ± 1

Visit 3 22 ± 1 25 ± 1 ^ Visit 4 21 ± 1 22 ± 1

Overall 22 ± 0.6 23 ± 0.6 Protein (g) Visit 1 94 ± 9 98 ± 10 Visit 2 94 ± 10 103 ± 11

Visit 3 126 ± 11 103 ± 12 Visit 4 85 ± 12 105 ± 13

Overall 100 ± 5 103 ± 6 Fat (g) Visit 1 72 ± 6 79 ± 7 Visit 2 76 ± 7 72 ± 7

Visit 3 91 ± 8 74 ± 8 Visit 4 71 ± 9 67 ± 9 Overall 78 ± 4 73 ± 4 Glycaemic Index (%) Visit 1 56 ± 0.5 57 ± 0.6

Visit 2 56 ± 0.6 51 ± 0.6 *^

Visit 3 56 ± 0.7 51 ± 0.7 *^

Visit 4 56 ± 0.8 53 ± 0.8 *^ Overall 56 ± 0.3 53 ± 0.3 * Mean ± SEM, p < 0.05; * Significance between groups; ^ Significance within groups compared to visit 1 (baseline); visit 2 and 3 = prenatal period; visit 4 = postnatal period.

165

Table 5.9. Macronutrients as a percentage of total daily caloric intake (19 years of age and over)

Macronutrient Diet Group Pre-education Post-education AMDR Baseline Prenatal Postpartum (Visit 1) (Average visit 2 and 3) (Visit 4) Carbohydrate, total (% of total energy) Standard Care 45 46 46 45 to 65* Low GI 41 42 46 Fat (% of total energy) Standard Care 34 35 34 20 to 35 Low GI 37 35 33 Protein (% of total energy) Standard Care 21 19 20 10 to 35 Low GI 22 23 21 *AMDR for total carbohydrate as a percentage to total daily energy (117); Based on mean values in table 5.8.; no statistically or clinically significant differences detected between study visits; AMDR = acceptable macronutrient distribution range (dietary reference intakes); GI = glycaemic index; % = percent

166

5.4.5.0. Self-Monitored Blood Glucose (Kirkpatrick Level 4 – Results)

Self-monitored blood glucose values were obtained for 69 participants at baseline, but were increasingly difficult to obtain as pregnancy progressed or delivery occurred. Sixty-five participants provided self-monitored blood glucose at visit 2 (> visit 1 to visit 2), 43 participants provided self-monitored blood glucose at visit 3 (> visit 2 to visit 3), and 22 provided self- monitored blood glucose readings from visit 3 to delivery (provided during the postpartum appointment; visit 4). These data are further broken down by group assignment in table 5.10.

Table 5.10. Sample size (counts) attrition in the context of self-monitored blood glucose

Visit Category Standard Care Low Glycaemic Index Baseline (≤ visit 1) 37 32 > V1 to V2 34 31 > V2 to V3 22 21 > V3 to Delivery 13 9 V = visit, > greater than, ≤ = less than or equal to

5.4.5.1. Average Postprandial Self-Monitored Blood Glucose

There was a main effect of diet assignment on (raw) average postprandial blood glucose (low GI: 6.02 ± 0.03 vs. standard care: 6.10 ± 0.02 mmol/L; p = 0.041), but a visit-diet interaction was not detected (p = 0.191). This significant difference between groups was maintained after step-wise adjustment using (1.) pre-pregnancy BMI and (2.) gestational age. Moreover, a significant visit-diet interaction was detected when pre-pregnancy BMI was included in the model as a covariate (p ≤ 0.001). Further analysis (time-point testing) showed that there was no significant difference in postprandial blood glucose between groups at baseline (low GI: 6.20 ± 0.04 vs. standard care: 6.18 ± 0.04 mmol/L; p = 0.79), visit 3 (low GI: 6.03 ± 0.04 vs. standard care: 6.11 ± 0.04; p = 0.166) and visit 4 (low GI: 5.93 ± 0.06 vs. standard care: 6.04 ± 0.05; p = 0.159). There was, however, a significant difference between groups at visit 2 (low GI: 5.92 ± 0.04 vs. standard care: 6.08 ± 0.03; p = 0.001). The raw means ± SEM versus visit is shown in figure 5.5.

167

Figure 5.5. Mean postprandial self-monitored blood glucose two hours after breakfast, lunch and dinner versus study visit

Mean ± SEM; GI = glycaemic index; V = visit; n = sample count; > = greater than; ≤ = less than or equal to

5.4.5.2. Percent Postprandial Self-monitored Blood Glucose Values Within Range

Percent average postprandial self-monitored blood glucose values within range (< 6.7 mmol/L; primary outcome) was not impacted by diet assignment at baseline (Wald = 0.062; p = 0.910; OR = 0.985; 95% CI: 0.21 to 3.23). This did not change post-intervention (Wald = 0.001; p = 0.973; OR = 0.952; 95% CI: 0.06 to 16.28). Table 5.11. includes a break-down of the number of postprandial self-monitored blood glucose values within and above range (counts [percent]). All values were in range at “visit 4” (low GI: 8 vs. standard care: 13) and are not included in the table below.

168

Table 5.11. Counts (percent) of postprandial self-monitored blood glucose values within range by group assignment and visit

Baseline Low GI Standard Care Total Counts (Percent, %) Counts (Percent, %) ≥ 6.7 mmol/L 5 (16) 5 (14) 10 < 6.7 mmol/ L 27 (84) 32 (86) 59 Total 32 37 69 Visit 2 Low GI Standard Care Total Counts (Percent, %) Counts (Percent, %) ≥ 6.7 mmol/L 3 (10) 3 (9) 6 < 6.7 mmol/ L 28 (90) 31 (91) 59 Total 31 34 65 Visit 3 Low GI Standard Care Total Counts (Percent, %) Counts (Percent, %) ≥ 6.7 mmol/L 1 (5) 1 (5) 2 < 6.7 mmol/ L 20 (95) 21 (95) 41 Total 21 22 43 ≥ = greater than or equal to, < less than

169

5.5. CONCLUSIONS AND DISCUSSION

The GI in GDM Study was a prospective RCT including two diet arms; (1.) low GI and (2.) standard care. The aim of this study was to comprehensively measure GI utility in women living with GDM. The GIQ, face/ content validated and pre-tested during GIEES, and three day diet records were used to measure the first three levels of KM; (1.) participant satisfaction, (2.) change in GI knowledge score, and (3.) change in dietary GI during the prenatal and postpartum period. Participant-measured postprandial self-monitored blood glucose was used to calculate two markers of glycaemic control; (1.) percent postprandial blood glucose below 6.7 mmol/L (primary outcome) and (2.) average postprandial blood glucose (secondary outcome). The original power calculation indicated 300 participants would provide an adequate sample size to detect an effect of diet on both markers of glycaemic control, if present. Together, these outcomes represent KM level 4 – Results (40,80,81).

The study sample was ethnically heterogeneous (figure 5.3, table 5.5.); addressing the call for more research on GDM medical nutrition therapy in ethnically diverse samples (highlighted in the literature review) (66,172,419). The three largest response categories were “Other” (20/73), “European” (16/ 73) and “East Asian” (12/73). Participants also identified as South East Asian (11/73), Latin American (7/73), African/ Caribbean (4/ 73), Indian (2/73), and Aboriginal (1/73). Important to note from a questionnaire development standpoint, twenty-seven percent of respondents selected “other”, providing general descriptors like “Canadian” and “Mixed Heritage” and specific mixed heritage examples like German, Dutch, European, and Latin American. Despite being trained on administration of the GIQ, investigators/ questionnaire administrators expressed difficulty with administration of this question; reporting resistance when encouraging participants to choose a number of specific ethnic categories (e.g. East Asian, European) rather than choosing “other” and giving a general descriptor like “mixed heritage” or “Canadian”. The investigators expressed interest in standardized closed-end geographically- based ethnicity questions for future iterations of the GIQ. Data collected on questionnaire administration during GIEES and GI in GDM and the recent publication by Omand (Darling) et al. (2014) will be considered during development of the next iteration of the GIQ. Omand et al. (2014) reported on the standardization of a closed-end geographically-based ethnicity question.

170

Data from this study were not available during the development of the GI in GDM iteration of GIQ, as it ran from December 2008 to June 2012 (420).

Data generated by the GIQ supports that participants consuming a low GI diet were satisfied (KM level 1) with the GI education platform (e.g. class content, instructors, and patient education materials) and increased their GI knowledge during the GI education session (KM level 2). Those on the standard care diet were also satisfied with the education provided. As highlighted in section 5.4.2., the results generated using GIQ Section 1 (Participant Satisfaction) were overwhelmingly positive. For instance, options “disagree” and “strongly disagree” were not chosen by participants in response to any of the statements provided in question 4 (table 5.7.). As the KM model would predict based upon study participant satisfaction, a significant increase in knowledge score (KM level 2) was observed in the low GI group post-education (pre-education: 47 ± 4%, post-education [average]: 86 ± 3%; p ≤ 0.001) (80,81). Not only was a significant increase in knowledge score observed immediately after the education session, but this knowledge score was maintained until the end of the study (6 to 8 weeks postpartum), spanning the prenatal and postnatal study period. As expected, knowledge score was not significantly different between groups at baseline (low GI: 47 ± 4 vs. standard care: 46 ± 3%, p = 0.774).

Three day diet record analysis showed that participants on the low GI diet reduced their dietary GI (KM level 3) by 6 units in less than two weeks (visit 1: 57 ± 0.6 to visit 2: 51 ± 0.6; p ≤ 0.001) and maintained this reduction until the end of the prenatal period (table 5.8.). At baseline, participants’ dietary GI fell within the medium GI category (56 to 69%) as per the CDA “The Glycemic Index”. Post-education, participants’ on the low GI diet had a dietary GI that fell within the low GI category (≤ 55%), while those on standard care maintained a medium dietary GI. Although the reduction in dietary GI observed in the low GI group was 3 units less than the 9 unit reduction observed in the pilot study, a reduction of 5 to 9 GI units has been associated with clinically relevant change in other samples (e.g. markers of glycaemic control; examples of level 4 outcomes). Important to highlight is the finding that study participants were able to achieve this 6 unit reduction in GI within two weeks post-education. As highlighted in the literature review, women with GDM have a tight timeline (between diagnosis, education, and delivery) in which to achieve behaviour change (aimed at achieving glycaemic and weight gain

171 targets) (14,172). As highlighted in the literature review, Hui et al. (2014) reported that Canadians who received standard care for GDM felt that dietary recommendations presented to them were “unrealistic” given the short period (up to 10 weeks on average) in which they were expected to make dietary behaviour change. Additionally, current standard care includes introduction of insulin treatment if glycaemic targets are not met for two consecutive weeks (14). This is noteworthy, as this treatment approach being linked to an increase in maternal weight gain and physical, emotional and financial stress and does not rectify peripheral insulin resistance (40,162,163,165). These data highlight that layering low GI education onto standard care dietary recommendations may be suitable for an environment in which behaviour change is expected to occur in a short period of time (two to 10 weeks) (supporting GI utility in this unique client population).

As highlighted in the literature review, a common criticism of the low GI diet is that it opposes current dietary guidelines (63,148,150,151). This criticism has not been supported by the literature to date and has been further disputed by GIEES and GI in GDM (40,47,48,67,125,141, 200,201). Specifically (table 5.8.), the low GI did not result in a statistically significant change in protein and fat intake; maintaining a non-significant difference in these macronutrients between groups during the study period. The low GI diet did, however, have a main effect on carbohydrate intake; although a visit-diet interaction was not detected. This main effect translated to a lower intake of carbohydrate in the low GI group in comparison to the standard care group (low GI: 196 ± 8 vs. 223 ± 7 g; p = 0.016). Despite this, the low GI group experienced a statistically significant increase in fibre intake from baseline to visit 3 (~ 4 grams; p = 0.035; during the prenatal period). In this case, an increase was desired, as the low GI group had a significantly lower fibre intake than the standard care group at baseline (low GI: 21 ± 1 vs. standard care: 24 ± 4 g; p = 0.008). By visit 3, those consuming the low GI diet met the DRI (Adequate Intake) for fibre (117). Post-intervention, participants maintained a non-significant difference in fibre intake between groups (eliminating a recognized nutrient confounder); an aim not achieved in the pilot study. As highlighted in the literature review, this is a common aim of studies examining change in dietary GI and was achieved during the GI in GDM study period (29,37,40,122).

172

Participants’ macronutrient distribution did not change significantly over the study period (within groups; table 5.9.) (117). This said, all participants’ protein intake met the AMDR for protein from baseline to study end. Carbohydrate intake, as a percent of total energy, was within the AMDR for the standard care group from baseline to study end. The low GI group did not have a carbohydrate intake reflective of the AMDR at baseline, but it increased gradually during the study period until it was within range by visit 4. The standard care group had a carbohydrate intake that fell within the AMDR. The low GI group had a fat intake that exceed the AMDR pre-intervention, but was within the AMDR post-intervention.

GI in GDM is the second Canadian study examining GI utility in the GDM client population; the first of which was the GI in GDM pilot study (40). To date, studies examining GI utility in this client population have measured the effect of a low GI dietary pattern on insulin prescription (e.g. incidence, dosage), maternal and infant birthweight, incidence of macrosomia and neonatal hypoglycaemia. Often cited in support of using a low GI diet during GDM, this work has shown a low GI diet can reduce insulin prescription during the prenatal period and reduce infant birth weight, without compromising maternal or neonatal care/ health (data/ results reviewed in literature review) (24,64,67,70-72,124,125,128,161,171). Notwithstanding, current standard care includes, and is often influenced by, self-monitored blood glucose values provided by the client (14). Hernanez et al. (2013) highlighted that by not examining self-monitored blood glucose as a primary or secondary outcome, we have created a gap in the literature examining GI utility in GDM that may impede inclusion of a low GI dietary pattern in CPG. This study has aimed to address this gap by making postprandial self-monitored blood glucose our KM level 4 result and postprandial self-monitored blood glucose values within range our primary outcome.

Despite the significant decrease in dietary GI observed in the low GI group and the significant difference in GI detected between the low GI and standard care group post-education, a significant difference in percent postprandial self-monitored blood glucose values within range was not detected between diet groups (Section 5.4.5.2.). In fact, there was no significant difference between groups at baseline. Moreover, self-monitored blood glucose was primarily below 6.7 mmol/L post-education (table 5.11); with only three values being above range in each group at visit 2 and one value being above range at visit three. The challenge of detecting

173 differences in clinically relevant outcomes in well treated participant populations is documented in the literature (40,71). The low GI diet did, however, result in lower average postprandial blood glucose when compared to standard care (low GI: 6.02 ± 0.03 vs. standard care: 6.10 ± 0.02 mmol/L; p = 0.041). Moreover, when adjusted for pre-pregnancy BMI, a visit-diet interaction was detected (p ≤ 0.001). Upon further analysis, it became clear that there was a significant difference in postprandial blood glucose between groups at visit 2 (low GI: 5.92 ± 0.04 vs. standard care: 6.08 ± 0.03 mmol/L; p = 0.001), but not beyond this time point. These findings are in agreement with the significant reduction in dietary GI achieved by the low GI group by week two, that resulted in a significantly lower dietary GI in the low GI group in comparison to the standard care group. As highlighted and discussed above, this swift reduction in dietary GI and reduction in self-monitored blood glucose is favourable in this client population, as there is limited time for behaviour change before pharmacotherapy is introduced (~2 weeks). It is not clear, from this primary analysis, why the difference in average self- monitored blood glucose was not maintained for the rest of the prenatal period, however, the results may have been impacted by the sample attrition noted in figure 5.5. Subsequent analysis may offer some more insight into these findings.

In conclusion, this dissertation chapter includes primary evaluation of select outcomes from the GI in GDM Study according to KM. According to analysis completed to date, the GI education platform appears to have satisfied all four levels of the KM. All participants were satisfied with the education received (KM level 1), those who received low GI education increased their GI knowledge score (KM level 2), and those who received low GI education reduced their dietary GI post-education (KM level 3). Moreover, although percent postprandial self-monitored blood glucose within range did not differ between groups (primary outcome), there was a significant main effect of diet intervention on average postprandial blood glucose (lower on low GI dietary pattern in comparison to standard care; KM level 4). Although these primary analyses are promising, final conclusions regarding the effect of the dietary intervention on the primary outcome and KM level 4 must be held in reserve. For instance, postprandial self-monitored blood glucose values for breakfast, lunch and dinner were pooled for this dissertation (use for analysis of percent values within range and mean self-monitored blood glucose). Future analysis will analyze each meal category separately.

174

Only 33% of the target sample size (n = 300) was obtained during the recruitment period and must be acknowledged as a limitation of this RCT. Undershooting recruitment targets likely impacted our ability to detect an effect of dietary intervention on the primary outcome. This said, the pilot study was designed to detect a difference of 0.6 mmol/L in average postprandial blood glucose between the low GI and standard care group (n = 50), which the GI in GDM Study sample size (n = 65) satisfied (40). Another limitation of this study is that the sample was a very tightly controlled sample (highlighted above). As shown in figure 5.5 and table 5.11, self- monitored blood glucose levels were, for the most part, within range (< 6.7 mmol/L) during the study period; making it difficult to detect an effect of intervention. This limitation has been noted in the literature by other scientists examining the effect of low GI dietary intervention in this client population (40,71).

Subsequent analysis and interpretation of data provided by this RCT will include analysis and interpretation of GIQ Section 1 and 4a. These data will provide more insight into participant perception of the GI education platform and dietary pattern. Although data included in this chapter support that participants can understand and apply GI education, the qualitative data collected will provide more insight into participant perceptions of the education platform and dietary pattern and provide us with additional feedback on the GI education platform and GIQ.

175

CHAPTER 6.0. DISSERTATION SUMMARY AND DISCUSSION

175

176

6.0. DISSERTATION SUMMARY AND DISCUSSION

The overarching goal of this dissertation was to address educators’ perceived barriers to GI clinical utility. That is, to address knowledge and resource gaps identified by members of the practical and research community in efforts to facilitate GI knowledge translation to the end user. Six objectives were created to achieve this goal. The objectives of this dissertation were to:

The following paragraphs will provide a synopsis of the three studies conducted to achieve the overarching goal and supporting objectives.

It is well established that (acute and chronic) hyperglycaemia increases oxidative stress and that this stress is related to the development and progression of DM (and CVD) (73,226,231,232, 239,247,248,370). Also established, is the knowledge that meals composed of low GI foods reduce postprandial glycaemic variation (an established marker of GI utility) (29,37,41,43,50,61,144,145,268). More recently, a low GI diet has been identified as a potential means by which to reduce acute and chronic oxidative stress and improve endothelial function; introducing novel GI mechanisms and perhaps a novel marker of GI utility (74,78,268,338,339). The AOGI Study (Study 1, Chapter 3) was developed to isolate the effect of delayed carbohydrate absorption (from other dietary factors) on postprandial oxidative stress in participants classified as overweight or obese (n = 18). This study was developed to test the

177 following hypotheses: (1.) Sipping dextrose slowly over 3.5 h will result in less oxidative stress (measured using TRAP) than ingesting the same amount of dextrose as a bolus over 5 min, (2.) Sipping dextrose will reduce oxidative stress to the same extent as 1g of oral vitamin C and (3.) The effect of sipping dextrose on oxidative stress will occur sooner than that of vitamin C. The results of the AOGI Study supported the three hypotheses.

Generally, sipping dextrose solution resulted in less oxidative stress than the bolus treatment, as measured by plasma TRAP (primary outcome). Moreover, the study results support that slowing carbohydrate absorption may be a means by which to maintain antioxidant capacity of the plasma, during the postprandial period, without antioxidant supplements (sipping reduced oxidative stress to the same extend of vitamin C). These findings also support that slowing carbohydrate absorption maintains antioxidant capacity during the postprandial period after the second meal. Subsequent work will be required to assess if this acute protection translates into chronic protection and if slowing carbohydrate absorption is, in fact, the main mechanism at play in published food-based studies showing a low GI diet is associated with less oxidative stress than a medium-high GI diet. Moreover, measurement of plasma-based inhibitors liked vitamin E, protein thiols and uric acid would provide more insight into the relationship between TRAP, LDLox, acute hyperglycaemia and endogenous antioxidant stress response; it is recommended these outcomes be considered for subsequent work.

In agreement with published food-based studies looking at comparable hypotheses, this study offers additional insight into the relationship between low GI foods and markers of chronic disease treatment and disease risk (74,78,268). These findings support the proposed novel GI mechanism(s) and highlight a potential marker of GI utility; addressing educators’ and scientists’ desire for additional insight into GI mechanism and utility (62,63,74,78,268). The slow absorption model has been the basis of low GI education for decades; it is a physiological mechanism that RDs and clients readily understand in the context of glycaemic control (29,79,144,159,196,320,323,328,329). This study offers confirmation that slowing carbohydrate absorption can offer utility beyond that of glycaemic control and may impact outcomes of CVD risk. This said, additional examination of this mechanism is warranted before it is considered for implementation into practice or education messages. For instance, additional studies examining the relationship between slowing carbohydrate absorption (or consuming low GI foods) and

178 cardiovascular hemodynamics are warranted. Moreover, examining whether or not milk and milk products (lactose is low GI) or fruits (fructose is low GI) reduce postprandial oxidative stress to the same extent as slowing carbohydrate absorption would provide important insight into the GI mechanisms at play (31).

The second and third studies included, in this dissertation, shifted from a focus on novel mechanism to that of evaluation of a low GI intervention using glycaemic control (an established marker of GI utility in DM) and novel strategies for comprehensive education evaluation.

GIEES (Study 2, Chapter 4), was a two-phase one-armed pre/post-education evaluation design aimed at evaluating the effectiveness of an evidence-based GI education platform in men and women living with T2DM. Phase two was the focus of Chapter 4 and aimed to test the hypothesis that a low GI education platform would significantly reduce participant dietary GI post-intervention. Secondary outcomes included: (1.) participant satisfaction with the GI education platform, (2.) low GI food acceptability, (3.) GI knowledge uptake, and (4.) education application. This study satisfied KM levels 1 to 3, addressed educators’ request for reliable education materials and showed that participants did not find GI too hard to understand and apply (perceived barriers 2 to 4) (80,81). More specifically, participants (n = 29) reported being satisfied with the GI education platform and low GI diet (KM level 1). They also obtained a significantly higher GI-knowledge score immediately post-education in comparison to baseline, and were able to maintain this score until the end of the study period (KM level 2). Additionally, participants were able to reduce their dietary GI post-education (KM level 3). Our participant satisfaction, knowledge score and dietary behaviour change results were reflective of the findings reported in the literature and represent the first GI intervention study where all of these outcomes have been comprehensively quantified in a prospective study based on an evidence- based education evaluation framework (38,48,56,141,206,222).

179

GIEES pre-tested the GIQ; a novel evidence-based questionnaire developed to evaluate the GI education platform. Data collected during GIEES was used to improve the GI education platform and GIQ for implementation in Study 3 (Chapter 5). Study 3, the GI in GDM Study, was a prospective multi-centered RCT with two diet arms: (1.) Standard care and (2.) Low GI. This RCT was designed to comprehensively evaluate the effect of the low GI diet intervention (the GI education platform) on glycaemic control. The primary hypothesis tested in the context of this dissertation was: The low GI education platform will improve postprandial glycaemic control in women with GDM in comparison to those receiving standard care (KM level 4). The primary outcome was percent postprandial self-monitored blood glucose < 6.7 mmol/L. The study was also powered to detect a 0.25 mmol/L difference in mean postprandial self-monitored blood glucose; the primary outcome for the pilot study and the clinical outcome used for guiding medical therapy (40). The other secondary outcomes examined as part of this dissertation include: (1.) (change in) dietary GI, (2.) participant satisfaction with the GI education platform, (4.) (change in) participant GI knowledge score.

The GI in GDM Study satisfied KM levels 1 to 4, provided novel data on GI utility, addressed (perceived barriers 1 to 4) educators’ request for reliable education materials and showed that women with GDM did not find GI too hard to understand and apply. More specifically, study participants were satisfied with the education (KM level 1) (80,81). They obtained a significantly higher GI-knowledge score immediately post-education in comparison to baseline, and were able to maintain this score until the end of the study period (prenatal and postnatal period) (KM level 2). Moreover, participants on the low GI diet were able to significantly reduce their dietary GI and obtain a significantly lower dietary GI than the standard care group in less than two weeks post-education (KM level 3). Despite this, a significant difference in percentage postprandial self-monitored blood glucose values < 6.7 mmol/L was not detected between groups (primary outcome). However, the low GI diet resulted in lower average self- monitored blood glucose than the standard care diet in this well treated group of participants; the primary outcome of the pilot study (KM level 4) (40). Currently, outside of the pilot study, there are no published data from GDM samples against which to compare our self-monitored blood glucose results in the context of low GI intervention evaluation. This acknowledged, our findings are supportive of published data that show a low GI diet reduces insulin prescription and infant birth weight (24,70-72,104,128,161). We are hopeful this work will inspire repeat

180 evaluation of GI utility in other GDM samples; including postprandial blood glucose as an outcome. Self-monitored blood glucose is, after all, a central treatment target as per the CDA CPG and within the DIPs involved in the GI in GDM Study.

In summary, by addressing educators’ four perceived barriers to GI utility (perceived translational roadblocks) and by fulfilling the six dissertation objectives, we were able to achieve the overarching goal of this dissertation. That is, to address knowledge and resource gaps identified by members of the practical and research community that span bench to bedside. This said, assessment of whether or not these efforts actually increase GI knowledge translation to the end user remains to be determined. Future directions are outlined in section 6.3.

6.1. LIMITATIONS

All three studies implemented dietary intake data collection and analysis tools. A 24 hour recall was used in AOGI, while a three day diet record was used in GIEES and GI in GDM. As previously discussed in this dissertation, there are inherent limitations to dietary data collection tools (e.g. underreporting, social desirability bias) and data analysis software (153-157). In the context of GI research, however, there is an added methodological concern highlighted in the literature time and time again. That is, a common criticism of food-based GI interventions is the subjectivity of GI assignment to study foods (at the data entry and analysis stage) (45). To address these limitations, we created an evidence-based standard operating procedure for GI value selection and assignment (appendix 2.1.), trained research staff (RDs) in three day diet administration, collection and data entry, and provided all participants with instructions and examples of how to complete a three day diet record (or 24 hour recall). This said, these efforts to not completely eliminate the risk of GI misclassification; making this a limitation of the methodology and results.

In the case of AOGI, the study was designed to eliminate the nutrient/ antioxidant confounders introduced in food-based studies so that the effect of slowly absorbed carbohydrate on postprandial oxidative stress could be studied in isolation. The slow absorption model is one physiological mechanism explaining how low GI foods function in the human body (29). There are others, which limits comparison of these findings to food-based studies including low GI

181 foods that are not subject to the slow absorption model. That is, the results obtained may not be comparable to food-based studies that rely heavily on fruits (fructose is low GI) to lower dietary GI (29,31). However, this study offers confirmation that slowing carbohydrate absorption can offer utility beyond that of glycaemic control and may impact outcomes of CVD risk. This said, additional examination of this mechanism (and novel outcome) is warranted before it is considered for implementation into practice or education messages.

A noteworthy limitation of the GI in GDM Study was the relatively small sample size, while sample size goals were achieved for AOGI and GIEES. During the GI in GDM Study, thirty- three percent of the goal sample (n = 300) was recruited from the four hospital sites. Sixty-four participants did finish the study, however, surpassing the sample size calculated for the pilot study (40). Interestingly, the primary outcome of the pilot, average postprandial self-monitored blood glucose, was significantly lower in those on the low GI diet compared to those on the standard care diet. Despite this, an effect of diet was not detected on the primary outcome; which may be related to the small sample size. Additionally, this lack of effect may be attributed to the sample being “well-controlled” or “intensely monitored” as part of their standard care; a limitation highlighted in other published GI dietary interventions and thought by some to be symptom of recruitment bias. That is, participants that work to achieve standard care targets may be more apt to join intervention studies (40,71). For instance, in GIEES and GI in GDM, baseline dietary GI was at the low end of the medium GI range as per CDA and in GI in GDM average baseline postprandial blood glucose was < 6.7 mmol/L.

6.2. OVERALL CONCLUSIONS

In conclusion, by addressing educators’ four perceived barriers to GI utility and by fulfilling the six dissertation objectives, we were able to achieve the overarching goal of this dissertation. That is, to address knowledge and resource gaps identified by members of the practical and research community that span bench to bedside.

182

Study 1, the AOGI Study, provided insight into a novel mechanism that is suspected to be driving the relationship between the low GI diet and chronic disease, addressed perceived barrier 1 and satisfied objective 1. From the results generated during this study, we are able to conclude that slowing rate of carbohydrate delivery reduced markers of postprandial oxidative stress (specifically TRAP) in participants diagnosed with overweight or obesity. That is, we are able to confirm recent results of a food-base study that showed that a low GI diet is effective (has utility in) in reducing postprandial oxidative stress or preserving postprandial antioxidant capacity (78).

Based upon the results generated by GIEES and GI in GDM, we are able to conclude that the GI education platform (intervention) satisfies KM levels 1 to 4. Its development, implementation and evaluation of these studies also addressed perceived barriers 1 to 4 and fulfilled objectives 2 to 6. Based on our evaluation, we were able to conclude that all participants were satisfied with the GI education platform (KM level 1) increased their GI knowledge immediately after the education session and maintained it to study end (KM level 2), and were able to significantly reduce their dietary GI (KM level 3) within a 1 to 2 week period. Moreover, women who received low GI education had significantly lower average postprandial self-monitored blood glucose than those that received standard care; a marker of improved glycaemic control (satisfying KM level 4).

In summary, this body of work has generated additional data on GI mechanism and utility to assist professionals in making an informed choice of whether or not to use GI in practice. It has also provided GI utility data that may be considered in development of future CDA CPG recommendations for management of GDM. Moreover, the resources developed and evaluated during this study have addressed DM educators’ desire for reliable GI education tools. Finally, our evaluation of the GI education platform counters the perceived barriers “The GI concept is too difficult for clients to understand” and “The GI concept is too difficult for clients to incorporate into their diet (to apply)”.

183

6.3. FUTURE DIRECTIONS

This dissertation includes three studies that are part of a larger initiative aimed at improving and increasing translation of evidence on GI utility to the educator and client to promote informed choice. In the current research environment, publication in academic journals (with high impact factor) and conference abstracts are still deemed “the currency of academics”. This said, scientists (especially in the health sciences) are being encouraged to explore novel and creative publication mediums that can improve translation of findings to the end user (e.g. open source journals, distance education initiatives, job aids) (90-92,96,97,181). Below is a summary of ongoing research and dissemination activities related to the GI education platform and GIQ.

 GI education platform evaluation (and implementation of the GIQ) has continued in the context of a multi-centre RCT called Avoiding Diabetes After Pregnancy Trial - in moms (ADAPT-M); based out of Women’s College Research Institute. This study is examining the effect of the GI education platform on beta-cell function of women during the postpartum period after GDM using KM. A novelty of this work is that the education platform will be delivered by trained Health Coaches. Two of the RDs involved in the development, implementation and evaluation of the GI education platform have joined Dr Lorraine Lipscombe’s Laboratory (Principal Investigator) to train the Health Coaches and facilitate implementation and evaluation.

 The Low GI Food Substitution List, implemented as part of the GI education platform in GIEES and the GI in GDM Study, has been chosen by CDA to serve as a template for adaption of the current CDA GI education tool; The Glycemic Index (30). Moreover, the RDs involved with the development, implementation and evaluation of this client education material have been asked to create a backgrounder/ job aid for Canadian DM educators to reference when using the new GI Food Guide.

 The Dietitians of Canada (DC) have asked the RDs involved with the development, implementation and evaluation of the GI education platform to develop an online GI education course for RDs; to be offered via Learning on Demand (international reach). The content of this course will be heavily based on the GI education platform and the evaluation conducted during GIEES and GI in GDM. For more on Learning on Demand and to locate

184

the online course in the future: http://www.dietitians.ca/Learn/Learning-On-Demand/Online- Courses.aspx

 A full list of outcomes measured during the GI in GDM Study is included in appendix 5.1. Examination of the effect of the low GI diet on maternal plasma antioxidant capacity in women with GDM is ongoing. Although, maternal plasma was collected to measure antioxidant capacity/ oxidative stress during the GI in GDM Study, these outcomes were not included in this dissertation due to time limitations (time to completion). Assays developed and used during the AOGI Study are currently being used to measure plasma antioxidant capacity and lipid oxidation of samples collected during the GI in GDM Study.

In addition to the above work, this dissertation has provided the basis for subsequent examination of GI mechanism and GI education utility. For instance, it has introduced (KM) and re-introduced (sip-bolus paradigm) evaluation frameworks that can be used by other scientists to evaluate interventions and has introduced a novel approach to measuring TRAP (microplate reader). In terms of GI mechanism, areas of interest include examining if foods containing low glycaemic index sugars (e.g. fructose/ fruit) reduce postprandial oxidative stress to the same extent as slowing carbohydrate absorption. Also, examination of the relationship between low GI foods/ dietary pattern and cardiovascular hemodynamics is warranted to better understand the effect low GI has on cardiovascular risk.

Although this work has managed to address educators call for reliable client education materials, research examining educator and client satisfaction with these resources is needed. Also, research examining whether or not availability of these job aids influences educators’ perception of GI utility and/ or decisions regarding whether or not to integrate GI into DM standard care is warranted. Furthermore, assessment of whether or not these efforts actually increase GI knowledge translation to the end user is of great interest. Finally, this research provides a comprehensive evaluation of a low GI education platform in GDM; including additional data on the effect of a low GI diet on postprandial glycaemic control in this client population (e.g. a gap identified in the literature). Although promising, future studies that aim to address the limitations of and build on this work are needed to provide sufficient evidence upon which to base the decision of whether or not to include GI education in CPGs for GDM management.

185

7.0. REFERENCE LIST

(1) Canadian Diabetes Association. Types of Diabetes. 2015; Available at: http://www.diabetes.ca/about-diabetes/types-of-diabetes. Accessed May/09, 2015.

(2) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Screening for the presence of coronary artery disease. Canadian Journal of Diabetes 2013;37:S105-S109.

(3) American Heart Association. Cardiovascular Disease & Diabetes. 2015; Available at: http://www.heart.org/HEARTORG/Conditions/Diabetes/WhyDiabetesMatters/Cardiovascular- Disease-Diabetes_UCM_313865_Article.jsp/. Accessed October/15, 2014.

(4) Heart and Stroke Foundation. Statistics. 2015; Available at: http://www.heartandstroke.com/site/c.ikIQLcMWJtE/b.3483991/k.34A8/Statistics.htm. Accessed September/ 15, 2014.

(5) Booth GL, Kapral MK, Fung K, Tu JV. Relation between age and cardiovascular disease in men and women with diabetes compared with non-diabetic people: a population-based retrospective cohort study. Lancet 2006;368(9529):29-36.

(6) Lee WL, Cheung AM, Cape D, Zinman B. Impact of diabetes on coronary artery disease in women and men: A meta- analysis of prospective studies. Diabetes Care 2000;23(7):962-968.

(7) Heart and Stroke Foundation. Heart Disease. 2015; Available at: http://www.heartandstroke.on.ca/site/c.pvI3IeNWJwE/b.3581739/k.7EFF/Heart_Disease.htm. Accessed October/15, 2014.

(8) Mahan LKM, Escott-Stump S. Krause's Food, Nutrition, & Diet Therapy. 11th ed. Philadelphia, P.A., U.S.A.: Elsevier; 2004.

(9) Stary HC, Bleakley Chandler A, Glagov S, Guyton JR, Instull Jr. W, Rosenfeld ME, et al. A definition of initial, fatty streak, and intermediate lesions of atherosclerosis: A report from the committee on vascular lesions of the council on arteriosclerosis, American heart association. Arterioscler Thromb Vasc Biol 1994;14(5):840-856.

(10) Chait A. Progression of atherosclerosis: The cell biology. Eur Heart J 1987;8(SUPPL. E):15-22.

(11) Statistics Canada as shown by Heart and Stroke Foundation. Mortality, summary list of causes. 2011; Available at: http://www.heartandstroke.com/site/c.ikIQLcMWJtE/b.3483991/k.34A8/Statistics.htm. Accessed April/ 6, 2015.

(12) Browarski S, Stonebridge C, Theriault L. The Canadian Heart Health Strategy: Risk Factors and Future Cost Implications Report. 2010.

(13) Canadian Diabetes Association Clinical Practice Guidelines Expert Committees. 2013 Clinical Practice Guidelines. Canadian Journal of Diabetes 2013;37.

186

(14) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Diabetes in Pregnancy. Canadian Journal of Diabetes 2013;37:S168-S183.

(15) Dyck R, Klomp H, Tan LK, Turnell RW, Boctor MA. A comparison of rates, risk factors, and outcomes of gestational diabetes between aboriginal and non-aboriginal women in the Saskatoon Health District. Diabetes Care 2002;25(3):487-493.

(16) Godwin M, Muirhead M, Huynh J, Helt B, Grimmer J. Prevalence of gestational diabetes mellitus among Swampy Cree women in Moose Factory, James Bay. CMAJ 1999;160(9):1299- 1302.

(17) Harris SB, Caulfield LE, Sugamori ME, Whalen EA, Henning B. The epidemiology of diabetes in pregnant Native Canadians: A risk profile. Diabetes Care 1997;20(9):1422-1425.

(18) Rodrigues S, Robinson E, Gray-Donald K. Prevalence of gestational diabetes mellitus among James Bay Cree women in northern Quebec. CMAJ 1999;160(9):1293-1297.

(19) Retnakaran R, Qi Y, Sermer M, Connelly PW, Hanley AJG, Zinman B. ß-cell function declines within the first year postpartum in women with recent glucose intolerance in pregnancy. Diabetes Care 2010;33(8):1798-1804.

(20) Feig DS, Zinman B, Wang X, Hux JE. Risk of development of diabetes mellitus after diagnosis of gestational diabetes. Obstetrical and Gynecological Survey 2008;63(12):759-761.

(21) Bellamy L, Casas J-, Hingorani AD, Williams D. Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. The Lancet 2009;373(9677):1773-1779.

(22) Harris MK. Gestational diabetes may represent discovery of preexisting glucose intolerance. Diabetes Care 1988;11(5):402-411.

(23) Pozzilli P, Fallucca F. Diet and diabetes: A cornerstone for therapy. Diabetes Metab Res 2014;30(S1):1-3.

(24) Vercoza Viana L, Gross JL, Azevedo MJ. Dietary intervention in patients with gestational diabetes mellitus: A systematic review and meta-analysis of randomized clinical trials on maternal and newborn outcomes. Diabetes Care 2014;37(12):3345-3355.

(25) Desroches S, Lapointe A, Ratté S, Gravel K, Légaré F, Thirsk J. Interventions to enhance adherence to dietary advice for preventing and managing chronic diseases in adults: A study protocol. BMC Public Health 2011;11.

(26) Kalergis M, De Grandpré E, Andersons C. The role of the glycemic index in the prevention and management of diabetes: A review and discussion. Canadian Journal of Diabetes 2005;29(1):27-38.

(27) Jenkins DJA, Wolever TMS, Taylor RH. Glycemic index of foods: A physiological basis for carbohydrate exchange. Am J Clin Nutr 1981;34(3):362-366.

187

(28) Wolever TMS, Jenkins DJA, Jenkins AL, Josse RG. The glycemic index: Methodology and clinical implications. Am J Clin Nutr 1991;54(5):846-854.

(29) Wolever TMS. The glycaemic index: A physiological classification of dietary carbohydrate. Cambridge, MA: CABI; 2006:1-227.

(30) Canadian Diabetes Association (CDA). The Glycemic Index Tool. Available at: http://archive.diabetes.ca/files/GlycemicIndex_08.pdf. Accessed April/ 14, 2015.

(31) The University of Sydney. Glycemic Index. 2014; Available at: http://www.glycemicindex.com/. Accessed February/ 16, 2015.

(32) Augustin LSA, Polesel J, Bosetti C, Kendall CWC, La Vecchia C, Parpinel M, et al. Dietary glycemic index, glycemic load and ovarian cancer risk: A case-control study in Italy. Annals of Oncology 2003;14(1):78-84.

(33) Augustin LSA, Maso LD, Vecchia CL, Parpinel M, Negri E, Vaccarella S, et al. Dietary glycemic index and glycemic load, and breast cancer risk: A case-control study. Annals of Oncology 2001;12(11):1533-1538.

(34) Augustin LSA, Gallus S, Franceschi S, Negri E, Jenkins DJA, Kendall CWC, et al. Glycemic index and load and risk of upper aero-digestive tract neoplasms (Italy). Cancer Causes and Control 2003;14(7):657-662.

(35) Brand Miller JC. Importance of glycemic index in diabetes. Am J Clin Nutr 1994;59(3 SUPPL.):747S-752S.

(36) Brand JC, Colagiuri S, Crossman S, Allen A, Roberts DCK, Truswell AS. Low-glycemic index foods improve long-term glycemic control in NIDDM. Diabetes Care 1991;14(2):95-101.

(37) Brand-Miller J, Hayne S, Petocz P, Colagiuri S. Low-glycemic index diets in the management of diabetes: A meta-analysis of randomized controlled trials. Diabetes Care 2003;26(8):2261-2267.

(38) Burani J, Longo PJ. Low-glycemic index carbohydrates: An effective behavioral change for glycemic control and weight management in patients with type 1 and 2 diabetes. Diabetes Educ 2006;32(1):78-88.

(39) Franceschi S, Maso LD, Augustin L, Negri E, Parpinel M, Boyle P, et al. Dietary glycemic load and colorectal cancer risk. Annals of Oncology 2001;12(2):173-178.

(40) Grant SM, Wolever TMS, O'Connor DL, Nisenbaum R, Josse RG. Effect of a low glycaemic index diet on blood glucose in women with gestational hyperglycaemia. Diabetes Res Clin Pract 2011;91(1):15-22.

(41) Kelly SAM, Frost G, Whittaker V, Summerbell CD. Low glycaemic indx diets for coronary heart disease (Review). Cochrane Database of Systematic Reviews 2004(4):1-67.

188

(42) Ludwig DS. The glycemic index: Physiological mechanisms relating to obesity, diabetes, and cardiovascular disease. J Am Med Assoc 2002;287(18):2414-2423.

(43) Opperman AM, Venter CS, Oosthuizen W, Thompson RL, Vorster HH. Meta-analysis of the health effects of using the glycaemic index in meal-planning. Br J Nutr 2004;92(3):367-381.

(44) Tavani A, Bosetti C, Negri E, Augustin LS, Jenkins DJA, La Vecchia C. Carbohydrates, dietary glycaemic load and glycaemic index, and risk of acute myocardial infarction. Heart 2003;89(7):722-726.

(45) Barclay AW, Petocz P, McMillan-Price J, Flood VM, Prvan T, Mitchell P, et al. Glycemic index, glycemic load, and chronic disease risk - A metaanalysis of observational studies. Am J Clin Nutr 2008;87(3):627-637.

(46) Fontvieille AM, Rizkalla SW, Penfornis A, Acosta M, Bornet FRJ, Slama G. The use of low glycaemic index foods improves metabolic control of diabetic patients over five weeks. Diabetic Med 1992;9(5):444-450.

(47) Frost G, Wilding J, Beecham J. Dietary advice based on the glycaemic index improves dietary profile and metabolic control in type 2 diabetic patients. Diabetic Med 1994;11(4):397- 401.

(48) Gilbertson HR, Thorburn AW, Brand-Miller JC, Chondros P, Werther GA. Effect of low- glycemic-index dietary advice on dietary quality and food choice in children with type 1 diabetes. Am J Clin Nutr 2003;77(1):83-90.

(49) Gutschall MD, Miller CK, Mitchell DC, Lawrence FR. A randomized behavioural trial targeting glycaemic index improves dietary, weight and metabolic outcomes in patients with type 2 diabetes. Public Health Nutr 2009;12(10):1846-1854.

(50) Jenkins DJA, Kendall CWC, McKeown-Eyssen G, Josse RG, Silverberg J, Booth GL, et al. Effect of a low-glycemic index or a high-cereal fiber diet on type 2 diabetes: A randomized trial. JAMA - Journal of the American Medical Association 2008;300(23):2742-2753.

(51) Jenkins DJA, Wolever TMS, Buckley G, Kah Yun Lam, Giudici S, Kalmusky J, et al. Low- glycemic-index starchy foods in the diabetic diet. Am J Clin Nutr 1988;48(2):248-254.

(52) Jenkins DJA, Wolever TMS, Kalmusky J, Guidici S, Giordano C, Patten R, et al. Low- glycemic index diet in hyperlipidemia: Use of traditional starchy foods. Am J Clin Nutr 1987;46(1):66-71.

(53) Ma Y, Olendzki BC, Merriam PA, Chiriboga DE, Culver AL, Li W, et al. A randomized clinical trial comparing low-glycemic index versus ADA dietary education among individuals with type 2 diabetes. Nutrition 2008;24(1):45-56.

(54) Luscombe ND, Noakes M, Clifton PM. Diets high and low in glycemic index versus high monounsaturated fat diets: Effects on glucose and lipid metabolism in NIDDM. Eur J Clin Nutr 1999;53(6):473-478.

189

(55) Miller CK, Gutschall M. A randomized trial about glycemic index and glycemic load improves outcomes among adults with type 2 diabetes. Health Education and Behavior 2009;36(3):615-626.

(56) Nansel TR, Gellar L, Zeitzoff L. Acceptability of Lower Glycemic Index Foods in the Diabetes Camp Setting. Journal of Nutrition Education and Behavior 2006;38(3):143-150.

(57) Saul N, Maryniuk MD. Using the glycemic index in diabetes management. Am J Nurs 2010;110(7):68-69.

(58) Waller B, Tzeng H-. Glycaemic index knowledge and use among African Americans with type 2 diabetes. J Adv Nurs 2011;67(5):1102-1108.

(59) Wolever TMS, Gibbs AL, Mehling C, Chiasson J-, Connelly PW, Josse RG, et al. The Canadian Trial of Carbohydrates in Diabetes (CCD), a 1-y controlled trial of low-glycemic- index dietary carbohydrate in type 2 diabetes: No effect on glycated hemoglobin but reduction in C-reactive protein. Am J Clin Nutr 2008;87(1):114-125.

(60) Wolever TMS, Jenkins DJA, Vuksan V, Jenkins AL, Buckley GC, Wong GS, et al. Beneficial effect of a low glycaemic index diet in Type 2 diabetes. Diabetic Med 1992;9(5):451- 458.

(61) Wolever TMS, Mehling C, Chiasson J-, Josse RG, Leiter LA, Maheux P, et al. Low glycaemic index diet and disposition index in type 2 diabetes (the Canadian trial of Carbohydrates in Diabetes): A randomised controlled trial. Diabetologia 2008;51(9):1607-1615.

(62) Kalergis M, Pytka E, Yale J-, Mayo N, Strychar I. Canadian dietitians' use and perceptions of glycemic index in diabetes management. Canadian Journal of Dietetic Practice and Research 2006;67(1):21-27.

(63) Grant SM, Wolever TMS. Perceived barriers to application of glycaemic index: Valid concerns or lost in translation? Nutrients 2011;3(3):330-340.

(64) Moses RG, Brand-Miller JC. The use of a low glycaemic index diet in pregnancy: An evolving treatment paradigm. Diabetes Res Clin Pract 2011;91(1):13-14.

(65) Farhanah AS, Fatin Nasirah MD, Barakatun Nisak MY, Nor Azlin MI, Zalilah MS. Current dietetic practices in the management of gestational diabetes mellitus: A survey of Malaysian dietitians. Asian Journal of Clinical Nutrition 2014;6(3):67-74.

(66) Mohd Yusof B-, Firouzi S, Mohd Shariff Z, Mustafa N, Mohamed Ismail NA, Kamaruddin NA. Weighing the evidence of low glycemic index dietary intervention for the management of gestational diabetes mellitus: An Asian perspective. Int J Food Sci Nutr 2014;65(2):144-150.

(67) Moses RG, Casey SA, Quinn EG, Cleary JM, Tapsell LC, Milosavljevic M, et al. Pregnancy and Glycemic Index Outcomes study: Effects of low glycemic index compared with conventional dietary advice on selected pregnancy outcomes. Am J Clin Nutr 2014;99(3):517- 523.

190

(68) Han S, Crowther CA, Middleton P, Heatley E. Different types of dietary advice for women with gestational diabetes mellitus. The Cochrane database of systematic reviews 2013;3.

(69) Hartling L, Dryden DM, Guthrie A, Muise M, Vandermeer B, Donovan L. Benefits and harms of treating gestational diabetes mellitus: A systematic review and meta-analysis for the U.S. preventive services task force and the national institutes of health office of medical applications of research. Ann Intern Med 2013;159(2):123-129.

(70) Hernandez TL, Anderson MA, Chartier-Logan C, Friedman JE, Barbour LA. Strategies in the nutritional management of gestational diabetes. Clin Obstet Gynecol 2013;56(4):803-815.

(71) Louie JCY, Markovic TP, Perera N, Foote D, Petocz P, Ross GP, et al. A randomized controlled trial investigating the effects of a low-glycemic index diet on pregnancy outcomes in gestational diabetes mellitus. Diabetes Care 2011;34(11):2341-2346.

(72) Louie JCY, Brand-Miller JC, Markovic TP, Ross GP, Moses RG. Glycemic index and pregnancy: A systematic literature review. Journal of Nutrition and Metabolism 2010;2010.

(73) Ceriello A, Esposito K, Piconi L, Ihnat M, Thorpe J, Testa R, et al. Glucose "peak" and glucose "spike": Impact on endothelial function and oxidative stress. Diabetes Res Clin Pract 2008;82(2):262-267.

(74) Hu Y, Block G, Norkus EP, Morrow JD, Dietrich M, Hudes M. Relations of glycemic index and glycemic load with plasma oxidative stress markers. Am J Clin Nutr 2006;84(1):70- 76.

(75) Jenkins DJA, Kendall CWC, Josse AR, Salvatore S, Brighenti F, Augustin LSA, et al. Almonds decrease postprandial glycemia, insulinemia, and oxidative damage in healthy individuals. J Nutr 2006;136(12):2987-2992.

(76) Monnier L, Mas E, Ginet C, Michel F, Villon L, Cristol J-, et al. Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. J Am Med Assoc 2006;295(14):1681-1687.

(77) Ceriello A. Acute hyperglycaemia and oxidative stress generation. Diabetic Med 1997;14(SUPPL. 3):S45-S49.

(78) Botero D, Ebbeling CB, Blumberg JB, Ribaya-Mercado JD, Creager MA, Swain JF, et al. Acute effects of dietary glycemic index on antioxidant capacity in a nutrient-controlled feeding study. Obesity 2009;17(9):1664-1670.

(79) Jenkins DJA, Wolever TMS, Ocana AM, Vuksan V, Cunnane SC, Jenkins M, et al. Metabolic effects of reducing rate of glucose ingestion by single bolus versus continuous sipping. Diabetes 1990;39(7):775-781.

(80) Kirkpatrick Partners: The One and Only Kirkpatrick. The Official Site of the Kirkpatrick Model. 2009-2014; Available at: http://www.kirkpatrickpartners.com/Home/tabid/38/Default.aspx.

191

(81) Kirkpatrick J. The hidden power of kirkpatrick's four levels. T and D 2007;61(8):34-37+4.

(82) Kirkpatrick J, Hawk L. Curricula and evaluation: Maximizing results. T and D 2006;60(6):61-62+6.

(83) Kirkpatrick DL. Effective supervisory training and development. I. Responsibility, needs, and objectives. Personnel 1984;61(6):25-30.

(84) Shergill M, Russell F, Luu L, Darling P, Nathanson T, Schamehorn S, et al. Evlautation of a group nutrition education session provided to hospitalized patients with respect to change in heart healthy behaviours post-hospital discharge. Final Report; St Michael Hospital Dietetic Internship 2009.

(85) Sargeant J, Borduas F, Sales A, Klein D, Lynn B, Stenerson H. CPD and KT: Models used and opportunities for synergy. J Contin Educ Health Prof 2011;31(3):167-173.

(86) Pehrson C, Sorensen JL, Amer-Wåhlin I. Evaluation and impact of cardiotocography training programmes: A systematic review. BJOG: An International Journal of Obstetrics and Gynaecology 2011;118(8):926-935.

(87) Kvan T. Evaluating learning environments for interprofessional care. Journal of Interprofessional Care 2013;27(S2):31-36.

(88) Slater BL, Lawton R, Armitage G, Bibby J, Wright J. Training and action for patient safety: Embedding interprofessional education for patient safety within an improvement methodology. J Contin Educ Health Prof 2012;32(2):80-89.

(89) Knowledge Translation Canada. KT Clearinghouse. 2015; Available at: http://ktclearinghouse.ca/ktcanada. Accessed 10-Feb-2015, 2015.

(90) Canadian Institutes of Health Research. More about knowledge translation at CIHR. 2014; Available at: http://www.cihr-irsc.gc.ca/e/39033.html. Accessed February 16, 2015.

(91) Cornelissen E, De Forge RT, Urquhart R. What do transdisciplinarity and knowledge translation matter to dietetics and dietitians? Practice 2011.

(92) Cornelissen E, Mitton C, Sheps S. Knowledge translation in the discourse of professional practice. International Journal of Evidence-Based Healthcare 2011;9(2):184-188.

(93) Vogel E, DeBusk R, Ryan-Harshman M. Advancing Knowledge Translation in Nutritional Genomics by Addressing Knowledge, Skills and Confidence Gaps of Registered Dietitians. Nutrition and Genomics 2009:161-186.

(94) Barwick M. Knowledge Translation Planning Template. 2008, 2013.

(95) Woolf SH. The meaning of translational research and why it matters. JAMA - Journal of the American Medical Association 2008;299(2):211-213.

192

(96) Barwick M, Butterill D, Lockett DM, Buckley L, Goering P. Scientist knowledge translation manual. Toronto, Ontario: The Hospital for Sick Children/ Centre for Addictions and Mental Health; 2005.

(97) Graham ID, Logan J, Harrison MB, Straus SE, Tetroe J, Caswell W, et al. Lost in knowledge translation: time for a map? J Contin Educ Health Prof 2006;26(1):13-24.

(98) Public Health Agency of Canada. Unpublished analysis using 2008/09 data from the Canadian Chronic Disease Surveillance System (Public Health Agency of Canada). 2011; Available at: http://www.phac-aspc.gc.ca/cd-mc/publications/diabetes-diabete/facts-figures- faits-chiffres-2011/chap1-eng.php#endnote1. Accessed May/30, 2015.

(99) Government of Canada. Type 2 Diabetes. 2013; Available at: http://healthycanadians.gc.ca/diseases-conditions-maladies-affections/disease-maladie/diabete- eng.php. Accessed June/20, 2015.

(100) Public Health Agency of Canada. Diabetes in Canada: Facts and figures from a public health perspective. 2011; Available at: http://www.phac-aspc.gc.ca/cd-mc/publications/diabetes- diabete/facts-figures-faits-chiffres-2011/highlights-saillants-eng.php. Accessed June/ 21, 2015.

(101) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Screening for Type 1 and Type 2 Diabetes. Canadian Journal of Diabetes 2013;37:S12-S15.

(102) Raikou M, McGuire A. The economics of screening and treatment in type 2 diabetes mellitus. Pharmacoeconomics 2003;21(8):543-564.

(103) Kahn R, Alperin P, Eddy D, Borch-Johnsen K, Buse J, Feigelman J, et al. Age at initiation and frequency of screening to detect type 2 diabetes: a cost-effectiveness analysis. The Lancet 2010;375(9723):1365-1374.

(104) Falavigna M, Schmidt MI, Trujillo J, Alves LF, Wendland ER, Torloni MR, et al. Effectiveness of gestational diabetes treatment: A systematic review with quality of evidence assessment. Diabetes Res Clin Pract 2012;98(3):396-405.

(105) Feig DS, Cleave B, Tomlinson G. Long-term effects of a diabetes and pregnancy program: Does the education last? Diabetes Care 2006;29(3):526-530.

(106) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Clinical Practice Guidelines for the Prevention and Management of Diabetes in Canada. Canadian Journal of Diabetes 2008;32.

(107) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Organization of Diabetes Care. Canadian Journal of Diabetes 2013;37:S20-S25.

(108) Stanford Patient Education Research Center, Stanford University. Chronic Disease Self- managment Program (Better Health Workshop). 2015; Available at: http://patienteducation.stanford.edu/programs/cdsmp.html. Accessed June/ 24, 2015.

193

(109) Stanford Patient Education Research Center, Stanford University. Diabetes Self- Management Program. 2015; Available at: http://patienteducation.stanford.edu/programs/diabeteseng.html. Accessed June/ 25, 2015.

(110) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Self- Management Education. Canadian Journal of Diabetes 2013;37:S26-S30.

(111) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Targets for Glycemic Control. Canadian Journal of Diabetes 2013;37:S31-S34.

(112) Health Canada. Body Mass Index Nonogram. 2012; Available at: http://www.hc- sc.gc.ca/fn-an/nutrition/weights-poids/guide-ld-adult/bmi_chart_java-graph_imc_java-eng.php. Accessed June/ 25, 2015.

(113) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Weight Management in Diabetes. Canadian Journal of Diabetes 2013;37:S82-S86.

(114) Prospective Studies Collaboration. Body-mass index and cause-specific mortality in 900 000 adults: collaborative analyses of 57 prospective studies. The Lancet 2009;373(9669):1083- 1096.

(115) Institutes of Medicine and National Research Council. Weight gain during pregnancy: Reexamining the guidelines. Washington, D.C.: The National Academies Press; 2009.

(116) Her Majesty the Queen in Right of Canada, represented by the Minister of Health Canada. Eating Well with Canada's Food Guide. 2011; Available at: http://www.hc-sc.gc.ca/fn- an/alt_formats/hpfb-dgpsa/pdf/food-guide-aliment/print_eatwell_bienmang-eng.pdf. Accessed April/ 4, 2015.

(117) Institutes of Medicine, Health Canada. Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein and Amino Acids. 2002; Available at: http://www.hc-sc.gc.ca/fn-an/nutrition/reference/table/index-eng.php. Accessed April/ 2015, 2015.

(118) Knopp RH, Magee MS, Raisys V, Benedetti T. Metabolic effects of hypocaloric diets in management of gestational diabetes. Diabetes 1991;40(SUPPL. 2):165-171.

(119) Knopp RH, Magee MS, Raisys V, Benedetti T, Bonet B. Hypocaloric diets and ketogenesis in the management of obese gestational diabetic women. J Am Coll Nutr 1991;10(6):649-667.

(120) Magee MS, Knopp RH, Benedetti TJ. Metabolic effects of 1200-kcal diet in obese pregnant women with gestational diabetes. Diabetes 1990;39(2):234-240.

(121) Laffel L. Ketone bodies: A review of physiology, pathophysiology and application of monitoring to diabetes. Diabetes Metab Res 1999;15(6):412-426.

194

(122) Grant SM. The effect of a low glycemic index diet on blood glucose control in women with gestational hyperglycemia. Master's of Science: University of Toronto; 2007.

(123) Carolan-Olah MC. Educational and intervention programmes for gestational diabetes mellitus (GDM) management: An integrative review. Collegian 2015.

(124) Louie JCY, Brand-Miller JC, Moses RG. Carbohydrates, glycemic index, and pregnancy outcomes in gestational diabetes. Current Diabetes Reports 2013;13(1):6-11.

(125) Louie JCY, Markovic TP, Ross GP, Foote D, Brand-Miller JC. Higher glycemic load diet is associated with poorer nutrient intake in women with gestational diabetes mellitus. Nutr Res 2013;33(4):259-265.

(126) Parker SE, Werler MM, Shaw GM, Anderka M, Yazdy MM. Dietary glycemic index and the risk of birth defects. Am J Epidemiol 2012;176(12):1110-1120.

(127) Walsh JM, McGowan CA, Mahony R, Foley ME, McAuliffe FM. Low glycaemic index diet in pregnancy to prevent macrosomia (ROLO study): Randomised control trial. BMJ (Online) 2012;345(7875).

(128) Moses RG, Barker M, Winter M, Petocz P, Brand-Miller JC. Can a low-glycemic index diet reduce the need for insulin in gestational diabetes mellitus? A randomized trial. Diabetes Care 2009;32(6):996-1000.

(129) International Organization for Standardization. Food products - Determination of the glycaemic index (GI) and recommendation for food classification. 2010;1(1):1-16.

(130) Foster-Powell K, Holt SHA, Brand-Miller JC. International table of gylcemic index and glycemic load values: 2002. Am J Clin Nutr 2002;76(1):5-56.

(131) Atkinson FS, Foster-Powell K, Brand-Miller JC. International tables of glycemic index and glycemic load values: 2008. Diabetes Care 2008;31(12):2281-2283.

(132) Wolever TMS. Physiological mechanisms and observed health impacts related to the glycaemic index: Some observations. Int J Obes 2006;30(SUPPL. 3):S72-S78.

(133) Southgate KM, Wolever TMS. Integrating the glycemic index into clinical practice: Written education materials and perceptions of utility. Canadian Journal of Diabetes 2012;36(2):54-57.

(134) Tokuda Y, Okamoto S, Yoshioka Y, Aizawa M, Tanaka M, Motomura K, et al. The influence of medical jargon mixed with foreign terminology in the Japanese clinical environment. Internal Medicine 2008;47(14):1329-1334.

(135) Deuster L, Christopher S, Donovan J, Farrell M. A method to quantify residents' jargon use during counseling of standardized patients about cancer screening. Journal of General Internal Medicine 2008;23(12):1947-1952.

195

(136) Sudore RL, Landefeld CS, Pérez-Stable EJ, Bibbins-Domingo K, Williams BA, Schillinger D. Unraveling the relationship between literacy, language proficiency, and patient- physician communication. Patient Educ Couns 2009;75(3):398-402.

(137) Mendes K, Iselin J, Edelstein S. Dietitians' use of the glycemic index/glycemic load as a counseling tool for overweight children. Topics in Clinical Nutrition 2006;21(4):300-311.

(138) Tsai C-, Leitzmann MF, Willett WC, Giovannucci EL. Glycemic load, glycemic index, and carbohydrate intake in relation to risk of cholecystectomy in women. Gastroenterology 2005;129(1):105-112.

(139) Tsai C-, Leitzmann MF, Willett WC, Giovannucci EL. Dietary carbohydrates and glycaemic load and the incidence of symptomatic gall stone disease in men. Gut 2005;54(6):823-828.

(140) Krog-Mikkelsen I, Sloth B, Dimitrov D, Tetens I, Björck I, Flint A, et al. A low glycemic index diet does not affect postprandial energy metabolism but decreases postprandial insulinemia and increases fullness ratings in healthy women. J Nutr 2011;141(9):1679-1684.

(141) Gilbertson HR, Brand-Miller JC, Thorburn AW, Evans S, Chondros P, Werther GA. The effect of flexible low glycemic index dietary advice versus measured carbohydrate exchange diets on glycemic control in children with type 1 diabetes. Diabetes Care 2001;24(7):1137-1143.

(142) Collier GR, Giudici S, Kalmusky J, Wolever TMS, Helman G, Wesson V, et al. Low glycaemic index starchy foods improve glucose control and lower serum cholesterol in diabetic children. Diabetes, Nutrition and Metabolism - Clinical and Experimental 1988;1(1):11-19.

(143) Miller CK, Headings A, Peyrot M, Nagaraja H. Goal difficulty and goal commitment affect adoption of a lower glycemic index diet in adults with type 2 diabetes. Patient Educ Couns 2012;86(1):84-90.

(144) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Nutrition Therapy. Canadian Journal of Diabetes 2013;37:S45-S55.

(145) Opperman M, Venter CS, Oosthuizen W, Thompson RL. Some health benefits of low glycaemic index diets - A systematic review. South African Journal of Clinical Nutrition 2005;18(3):214-221.

(146) Brand Miller J, Colagiuri S, Foster-Powell K. The glycemic index is easy and works in practice. Diabetes Care 1997;20(10):1628-1629.

(147) Xavier Pi-Sunyer F. Glycemic index and disease. Am J Clin Nutr 2002;76(1):290S-298S.

(148) Beebe C. Diets with low glycemic index: Not ready for practice yet! Nutrition Today 1999;34(2):82-86.

(149) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Methods. Canadian Journal of Diabetes 2013;37:S4-S7.

196

(150) Franz MJ. In defense of the American Diabetes Association's Recommendations on Glycemic Index. Nutrition Today 1999;34(2):78-81.

(151) Franz MJ, Bantle J, Beebe CA, Brunzell JD, Chiasson J-, Garg A, et al. American Diabetes Association Position Statement: Evidence-based nutrition principles and recommendations for the treatment and prevention of diabetes and related complications. J Am Diet Assoc 2002;102(1):109-118.

(152) Anderson TJ, Grégoire J, Hegele RA, Couture P, Mancini GBJ, McPherson R, et al. 2012 Update of the Canadian Cardiovascular Society Guidelines for the Diagnosis and Treatment of Dyslipidemia for the Prevention of Cardiovascular Disease in the Adult. Can J Cardiol 2013;29(2):151-167.

(153) Shapiro BB, Bross R, Morrison G, Kalantar-Zadeh K, Kopple JD. Self-Reported Interview-Assisted Diet Records Underreport Energy Intake in Maintenance Hemodialysis Patients. Journal of Renal Nutrition 2015;25(4):357-363.

(154) Sharp DB, Allman-Farinelli M. Feasibility and validity of mobile phones to assess dietary intake. Nutrition 2014;30(11-12):1257-1266.

(155) Illner A-, Freisling H, Boeing H, Huybrechts I, Crispim SP, Slimani N. Review and evaluation of innovative technologies for measuring diet in nutritional epidemiology. Int J Epidemiol 2012;41(4):1187-1203.

(156) Lyu L-, Chi I-, Huang C-, Huang L-, Lee W-, Lu H-. Development of nutrient databases for epidemiological studies in Taiwan. Asia Pac J Clin Nutr 2008;17(SUPPL. 1):302-305.

(157) Buzzard IM, Sievert YA. Research priorities and recommendations for dietary assessment methodology. Am J Clin Nutr 1994;59(1 SUPPL.):275S-280S.

(158) Bjorge-Schohl B, Johnston CS, Trier CM, Fleming KR. Agreement in participant-coded and investigator-coded food-record analysis inoverweight research participants: An examinationof interpretation bias. Journal of the Academy of Nutrition and Dietetics 2014;114(5):796-801.

(159) Jenkins DJ, Wolever TM, Taylor RH, Griffiths C, Krzeminska K, Lawrie JA, et al. Slow release dietary carbohydrate improves second meal tolerance. Am J Clin Nutr 1982;35(6):1339- 1346.

(160) Wolever TMS. Metabolic effects of continuous feeding. Metabolism: Clinical and Experimental 1990;39(9):947-951.

(161) Moses RG, Luebcke M, Davis WS, Coleman KJ, Tapsell LC, Petocz P, et al. Effect of a low-glycemic-index diet during pregnancy on obstetric outcomes. Am J Clin Nutr 2006;84(4):807-812.

197

(162) Brankston GN, Mitchell BF, Ryan EA, Okun NB. Resistance exercise decreases the need for insulin in overweight women with gestational diabetes mellitus. Am J Obstet Gynecol 2004;190(1):188-193.

(163) Gunderson EP. Intensive nutrition therapy for gestational diabetes: Rationale and current issues. Diabetes Care 1997;20(2):221-226.

(164) Langer O, Levy J, Brustman L, Anyaegbunam A, Merkatz R, Divon M. Glycemic control in gestational diabetes mellitus-How tight is tight enough: Small for gestational age versus large for gestational age? Obstet Gynecol 1989;161(3):646-653.

(165) Kuhl C. Etiology and pathogenesis of gestational diabetes. Diabetes Care 1998;21(B19- 26).

(166) Wolever TMS, Mehling C. High-carbohydrate-low-glycaemic index dietary advice improves glucose disposition index in subjects with impaired glucose tolerance. Br J Nutr 2002;87(5):477-487.

(167) Larsen TM, Dalskov S-, Van Baak M, Jebb SA, Papadaki A, Pfeiffer AFH, et al. Diets with high or low protein content and glycemic index for weight-loss maintenance. N Engl J Med 2010;363(22):2102-2113.

(168) Scholl TO, Chen X, Khoo CS, Lenders C. The Dietary Glycemic Index during Pregnancy: Influence on Infant Birth Weight, Fetal Growth, and Biomarkers of Carbohydrate Metabolism. Am J Epidemiol 2004;159(5):467-474.

(169) Wansink B. Nutritional Gatekeepers and the 72% Solution. J Am Diet Assoc 2006;106(9):1324-1327.

(170) Jilcott SB, Laraia BA, Evenson KR, Ammerman AS. Perceptions of the community food environment and related influences on food choice among midlife women residing in rural and urban areas: A qualitative analysis. Women and Health 2009;49(2-3):164-180.

(171) Louie JCY, Markovic TP, Ross GP, Foote D, Brand-Miller JC. Effect of a low glycaemic index diet on gestational diabetes mellitus on post-natal outcomes after 3 months of birth: A pilot study. Maternal and Child Nutrition 2013;11(3):409-414.

(172) Hui AL, Sevenhuysen G, Harvey D, Salamon E. Barriers and coping strategies of women with gestational diabetes to follow dietary advice. Women and Birth 2014;27(4):292-297.

(173) Brekke HK, Sunesson Å, Axelsen M, Lenner RA. Attitudes and barriers to dietary advice aimed at reducing risk of type 2 diabetes in first-degree relatives of patients with type 2 diabetes. Journal of Human Nutrition and Dietetics 2004;17(6):513-521.

(174) Thornton PL, Kieffer EC, Salabarría-Peña Y, Odoms-Young A, Willis SK, Kim H, et al. Weight, diet, and physical activity-related beliefs and practices among pregnant and postpartum latino women: The role of social support. Matern Child Health J 2006;10(1):95-104.

198

(175) Tuffery O, Scriven A. Factors influencing antenatal and postnatal diets of primigravid women. Journal of The Royal Society for the Promotion of Health 2005;125(5):227-231.

(176) Drolet BC, Lorenzi NM. Translational research: Understanding the continuum from bench to bedside. Translational Research 2011;157(1):1-5.

(177) Westfall JM, Mold J, Fagnan L. Practice-based research - "Blue highways" on the NIH roadmap. J Am Med Assoc 2007;297(4):403-406.

(178) Westergren A. Action-oriented study circles facilitate efforts in nursing homes to go from feeding to serving: Conceptual perspectives on knowledge translation and workplace learning. Journal of Aging Research 2012;2012.

(179) Gavaravarapu SM. Knowledge translation-some perspectives for nutrition education and communication. Journal of Nutrition Education and Behavior 2013;45(4).

(180) Goering PN, Wasylenki, DA. Promoting the utilization of outcome study results by assuming multiple roles within an organization. Eval Program Plann 1993;16(4):329-334.

(181) Atkinson SA. A nutrition odyssey: Knowledge discovery, translation, and outreach: 2006 Ryley-Teffs Memorial Lecture. Canadian Journal of Dietetic Practice and Research 2006;67(3):150-156.

(182) Hessett C, Moran A, Boulton AJM. An evaluation of diabetes knowledge amongst general practitioners and senior medical students. Diabetic Med 1989;6(4):351-353.

(183) Duncan AM, Dunn HA, Stratton LM, Vella MN. Translating knowledge into dietetic practice: A functional foods for health aging toolkit. Applied Physiology Nutrition and Metabolism 2014;39:600-603.

(184) Cornelissen E, Urquhart R, Chan VWY, DeForge RT, Colquhoun HL, Sibbald S, et al. Creating a knowledge translation trainee collaborative: From conceptualization to lessons learned in the first year. Implementation Science 2011:98.

(185) Partnership for Dietetic Education and Practice. The Integrated Competencies for Dietetic Education and Practice (ICDEP). : Developed by the Partnership for Dietetic Education and Practice; 2013.

(186) Satterfield JM, Spring B, Brownson RC, Mullen EJ, Newhouse RP, Walker BB, et al. Toward a Transdisciplinary Model of Evidence-Based Practice. Milbank Q 2009;87(2):368-390.

(187) Rycroft-Malone J. The PARIHS framework - A framework for guiding the implementation of evidence-based practice. J Nurs Care Qual 2004;19(4):297-304.

(188) Hemming J, Lordly D, Glanville NT, Corby L, Thirsk J. Developing an interview guide to evaluate practice-based evidence in nutrition: Use of the delphi technique. Canadian Journal of Dietetic Practice and Research 2011;72(4):186-190.

199

(189) Bowden FM, Lordly D, Thirsk J, Corby L. Phase II Practice-based Evidence in Nutrition (PEN) evaluation: Interviews with key informants. Canadian Journal of Dietetic Practice and Research 2012;73(1):e233-e240.

(190) Dietitians of Canada. PEN: Practice-based Evidence in Nutrition®. 2015; Available at: http://www.pennutrition.com/index.aspx. Accessed April/ 05, 2015.

(191) Heller T, Maislos M, Shahar D. Physicians' and Nurses' knowledge and attitude towards nutritional therapy in diabetes. Harefuah 2007;146(9):670-674.

(192) Cambon L, Minary L, Ridde V, Alla F. Transferability of interventions in health education: A review. BMC Public Health 2012;12(1).

(193) Le TQN, Goto A, Tran TT, Nguyen QV, Nguyen TK. Capacity building toward evidence- based medicine among healthcare professionals at the university of medicine and pharmacy, Ho Chi Minh City, and its related institutes. Japan Medical Association Journal 2014;57(1):49-55.

(194) Diamant R, Corby L. The challenge of change. One bulletin 1995;14(3).

(195) Palmer S. Finding a place for glycemic index. Today's Dietitian 2009;11(6):8.

(196) Slabber M. Complexities of consumer understanding of the glycaemic index concept and practical guidelines for incorporation in diets. South African Journal of Clinical Nutrition 2005;18(3):252-257.

(197) Wansink B, Shimizu M, Brumberg A. Dispelling myths about a new healthful food can be more motivating than promoting nutritional benefits: The case of Tofu. Eating Behav 2014;15(2):318-320.

(198) Berwick DM. Disseminating Innovations in Health Care. J Am Med Assoc 2003;289(15):1969-1975.

(199) Cândido FG, Pereira EV, Alfenas RCG. Use of the glycemic index in nutrition education. Rev Nutr 2013;26(1):89-96.

(200) Frost G, Keogh B, Smith D, Akinsanya K, Leeds A. The effect of low-glycemic carbohydrate on insulin and glucose response in vivo and in vitro in patients with coronary heart disease. Metabolism: Clinical and Experimental 1996;45(6):669-672.

(201) Brand-Miller J. The glycemic index as a measure of health and nutritional quality: An Australian perspective. Cereal Foods World 2007;52(2):41-44.

(202) Reach G. Linguistic barriers in diabetes care. Diabetologia 2009;52(8):1461-1463.

(203) Singh H, Duerksen DR. Survey of clinical nutrition practices of Canadian gastroenterologists. Canadian Journal of Gastroenterology 2006;20(8):527-530.

200

(204) van Mook WNKA, van Luijk SJ, O'Sullivan H, Wass V, Harm Zwaveling J, Schuwirth LW, et al. The concepts of professionalism and professional behaviour: Conflicts in both definition and learning outcomes. Eur J Intern Med 2009;20(4):e85-e89.

(205) Sudore LR, Schillinger D. Interventions to improve care for patients with limited health literacy. Journal of Clinical Outcomes Management 2009;16(1):20-29.

(206) Davis MS, Miller CK. Educational needs regarding the glycemic index in diabetes management. Topics in Clinical Nutrition 2006;21(1):17-25.

(207) Miller CK, Lindberg DV. Evaluation of a computer-based game about the glycemic index among college-aged students. Topics in Clinical Nutrition 2007;22(3):299-306.

(208) Heilbronn LK, Noakes M, Clifton PM. Effect of energy restriction, weight loss, and diet composition on plasma lipids and glucose in patients with type 2 diabetes. Diabetes Care 1999;22(6):889-895.

(209) Doak C, Doak L, Root J. Teaching Patients with Low Literacy Skills. 2nd ed. Philadelphia: Lipincott; 1996.

(210) Sugar Institute. Clips on sugars: For Up-To-Date Information on Sugars and Healthy Eating: Understanding the Glycemic Index. 2008; Available at: http://www.sugar.ca/SUGAR/media/Sugar-Main/PDFs/Clip9(1).pdf. Accessed 5/2, 2015.

(211) Wolever TMS. Glycemic index versus glycemic response: Nonsynonymous terms [5]. Diabetes Care 1992;15(10):1436-1437.

(212) Wolever TMS, Jenkins DJA, Vuksan V, Katzman L, Jenkins AL, Josse RG. Variation in meal fat content does not affect the relative glycaemic response of spaghetti in subjects with Type II diabetes. Diabetes, Nutrition and Metabolism - Clinical and Experimental 1992;5(3):191-197.

(213) Shai I, Schwarzfuchs D, Henkin Y, Shahar DR, Witkow S, Greenberg I, et al. Weight loss with a low-carbohydrate, Mediterranean, or low-fat diet. N Engl J Med 2008;359(3):229-241.

(214) Tarquinio C, Kivits J, Minary L, Coste J, Alla F. Evaluating complex interventions: Perspectives and issues for health behaviour change interventions. Psychology and Health 2015;30(1):35-51.

(215) Morris V. Can preoperative modification of dietary glycemic index in patients undergoing coronary artery bypass surgery reduce both pre- and post- surgical insulin resistance? Master's of Science: University of Toronto; 2005.

(216) Prochaska JO, Velicer WF. The transtheoretical model of health behavior change. American Journal of Health Promotion 1997;12(1):38-48.

201

(217) Prochaska JO, Velicer WF, Rossi JS, Goldstein MG, Marcus BH, Rakowski W, et al. Stages of Change and Decisional Balance for 12 Problem Behaviors. Health Psychology 1994;13(1):39-46.

(218) Prochaska JO, DiClemente CC. Stages and processes of self-change of smoking: Toward an integrative model of change. J Consult Clin Psychol 1983;51(3):390-395.

(219) Bandura A. Self-efficacy: Toward a unifying theory of behavioral change. Psychol Rev 1977;84(2):191-215.

(220) Bandura A, Adams NE. Analysis of self-efficacy theory of behavioral change. Cognitive Therapy and Research 1977;1(4):287-310.

(221) Bandura A, Adams NE, Beyer J. Cognitive processes mediating behavioral change. J Pers Soc Psychol 1977;35(3):125-139.

(222) Vermeulen A, Turnbull WH. Feasibility of The G.I. guide to increase knowledge about the glycaemic index in practice. Journal of Human Nutrition and Dietetics 2000;13(6):397-405.

(223) Brand-Miller JC, Barclay AW, Irwin T. A new food labelling program for the glycemic index. Asia Pac J Clin Nutr 2001;10(4).

(224) Al-Solaiman Y, Jesri A, Zhao Y, Morrow JD, Egan BM. Low-sodium DASH reduces oxidative stress and improves vascular function in salt-sensitive humans. J Hum Hypertens 2009;23(12):826-835.

(225) Ceriello A. Possible role of oxidative stress in the pathogenesis of hypertension. Diabetes Care 2008;31 Suppl 2:S181-184.

(226) Ceriello A, Lush CW, Darsow T, Piconi L, Corgnali M, Nanayakkara N, et al. Pramlintide reduced markers of oxidative stress in the postprandial period in patients with type 2 diabetes. Diabetes Metab Res 2008;24(2):103-108.

(227) Robertson R, Zhou H, Zhang T, Harmon JS. Chronic oxidative stress as a mechanism for glucose toxicity of the beta cell in type 2 diabetes. Cell Biochem Biophys 2007;48(2-3):139- 146.

(228) Cerne D, Lukac-Bajalo J. Oxidative stress assays for disease risk stratification. Acta Pharmaceutica 2006;56(1):1-17.

(229) Violi F, Cangemi R, Brunelli A. Oxidative stress, antioxidants, and cardiovascular disease. Arterioscler Thromb Vasc Biol 2005;25(4).

(230) Higdon JV, Frei B. Obesity and oxidative stress: A direct link to CVD? Arterioscler Thromb Vasc Biol 2003;23(3):365-367.

202

(231) Giugliano D, Ceriello A, Paolisso G. Diabetes mellitus, hypertension, and cardiovascular disease: Which role for oxidative stress? Metabolism: Clinical and Experimental 1995;44(3):363-368.

(232) Brownlee M. The pathobiology of diabetic complications: A unifying mechanism. Diabetes 2005;54(6):1615-1625.

(233) Kay CD, Holub BJ. The postprandial effects of dietary antioxidants in humans. Curr Atheroscler Rep 2003;5(6):452-458.

(234) Erel O. A novel automated method to measure total antioxidant response against potent free radical reactions. Clin Biochem 2004;37(2):112-119.

(235) Tessier D, Khalil A, Fülöp T. Effects of an oral glucose challenge on free radicals/antioxidants balance in an older population with type II diabetes. Journals of Gerontology - Series A Biological Sciences and Medical Sciences 1999;54(11):M541-M545.

(236) Gropper SS, Smith JL. Advanced Nutrition and Human Metabolism. 6th ed. Belmont, CA: Wasworth Publishing; 2013.

(237) Rice-Evans CA, Diplock AT, Symons MCR. Techniques in Free Radical Research. New York, New Work: Elsevier Science Publishers; 1991.

(238) Diplock AT. Antioxidant nutrients and disease prevention: An overview. Am J Clin Nutr 1991;53(1 SUPPL.):189S-193S.

(239) Ceriello A, Motz E. Is Oxidative Stress the Pathogenic Mechanism Underlying Insulin Resistance, Diabetes, and Cardiovascular Disease? The Common Soil Hypothesis Revisited. Arterioscler Thromb Vasc Biol 2004;24(5):816-823.

(240) Ceriello A, Taboga C, Tonutti L, Quagliaro L, Piconi L, Bais B, et al. Evidence for an independent and cumulative effect of postprandial hypertriglyceridemia and hyperglycemia on endothelial dysfunction and oxidative stress generation: Effects of short- and long-term simvastatin treatment. Circulation 2002;106(10):1211-1218.

(241) Vincent HK, Taylor AG. Biomarkers and potential mechanisms of obesity-induced oxidant stress in humans. Int J Obes 2006;30(3):400-418.

(242) Evans JL, Goldfine ID, Maddux BA, Grodsky GM. Are oxidative stress - Activated signaling pathways mediators of insulin resistance and β-cell dysfunction? Diabetes 2003;52(1):1-8.

(243) Koška J, Blažícek P, Marko M, Grna JD, Kvetnanský R, Vigaš M. Insulin, catecholamines, glucose and antioxidant enzymes in oxidative damage during different loads in healthy humans. Physiological Research 2000;49(SUPPL. 1):S95-S100.

203

(244) Andreeva-Gateva P, Popova D, Orbetzova V. Antioxidant parameters i metabolic syndrome - A dynamic evaluation during oral glucose tolerance test. Internal Diseases 2001;33(2-3):48-53.

(245) Sampson MJ, Davies IR, Brown JC, Ivory K, Hughes DA. Monocyte and neutrophil adhesion molecule expression during acute hyperglycemia and after antioxidant treatment in type 2 diabetes and control patients. Arterioscler Thromb Vasc Biol 2002;22(7):1187-1193.

(246) Mullan BA, Young IS, Fee H, McCance DR. Ascorbic acid reduces blood pressure and arterial stiffness in type 2 diabetes. Hypertension 2002;40(6):804-809.

(247) Ceriello A. Postprandial hyperglycaemia: A new risk factor for cardiovascular disease. Diabetes, Stoffwechsel und Herz 2008;17(5):363-373.

(248) Ceriello A. Cardiovascular effects of acute hyperglycaemia: Pathophysiological underpinnings. Diabetes and Vascular Disease Research 2008;5(4):260-268.

(249) Berliner JA, Navab M, Fogelman AM, Frank JS, Demer LL, Edwards PA, et al. Atherosclerosis: Basic mechanisms: Oxidation, inflammation, and genetics. Circulation 1995;91(9):2488-2496.

(250) Jialal I, Devaraj S. The role of oxidized low density lipoprotein in atherogenesis. J Nutr 1996;126(4 SUPPL.):1053S-1057S.

(251) Jialal I, Devaraj S. Low-density lipoprotein oxidation, antioxidants, and atherosclerosis: A clinical biochemistry perspective. Clin Chem 1996;42(4):498-506.

(252) Tsai EC, Hirsch IB, Brunzell JD, Chait A. Reduced plasma peroxyl radical trapping capacity and increased susceptibility of LDL to oxidation in poorly controlled IDDM. Diabetes 1994;43(8):1010-1014.

(253) Esterbauer H, Striegl G, Puhl H, Rotheneder M. Continuous monitoring of in vztro oxidation of human low density lipoprotein. Free Radic Res 1989;6(1):67-75.

(254) Itabe H, Obama T, Kato R. The dynamics of oxidaized LDL during atherogenesis. Journal of Lipids 2011.

(255) Steinberg D. Low density lipoprotein oxidation and its pathobiological significance. J Biol Chem 1997;272(34):20963-20966.

(256) Heinecke JW. Oxidants and antioxidants in the pathogenesis of atherosclerosis: Implications for the oxidized low density lipoprotein hypothesis. Atherosclerosis 1998;141(1):1- 15.

(257) Price KD, Price CSC, Reynolds RD. Hyperglycemia-induced ascorbic acid deficiency promotes endothelial dysfunction and the development of atherosclerosis. Atherosclerosis 2001;158(1):1-12.

204

(258) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Clinial Practice Guidelines - Dyslipidemia. Canadian Journal of Diabetes 2013;37:S110-S116.

(259) Brown MS, Goldstein JL. Lipoprotein metabolism in the macrophage: implications for cholesterol deposition in atherosclerosis. Annu Rev Biochem 1983;52:223-261.

(260) Ahotupa M, Marniemi J, Lehtimäki T, Talvinen K, Raitakari OT, Vasankari T, et al. Baseline diene conjugation in LDL lipids as a direct measure of in vivo LDL oxidation. Clin Biochem 1998;31(4):257-261.

(261) Ahotupa M, Vasankari TJ. Baseline diene conjugation in LDL lipids: An indicator of circulating oxidized LDL. Free Radical Biology and Medicine 1999;27(11-12):1141-1150.

(262) Liu M-, Bergholm R, Mäkimattila S, Lahdenperä S, Valkonen M, Hilden H, et al. A marathon run increases the susceptibility of LDL to oxidation in vitro and modifies plasma antioxidants. American Journal of Physiology - Endocrinology and Metabolism 1999;276(6 39- 6):E1083-E1091.

(263) Mercodia. Oxidized LDL ELISA Instruction/ Assay Manual. 2011;10-1143-01.

(264) Witztum JL. Susceptibility of low-density lipoprotein to oxidative modification. Am J Med 1993;94(4):347-349.

(265) Salvi P. Pulse Waves: How Vascular Hemodynamics Affects Blood Pressure. New York, New York: Springer; 2012.

(266) Westerhof N, Stergiopulous N, Noble MIM. Snapshots of Hemodynamics: An Aid For Clinical Research and Graduate Education. 2nd ed. New York, New York: Springer; 2012.

(267) Stouffer GA. Cardiovascular Hemodynamics for the Clinician. Malden, Massachusetts, U.S.A.: Blackwell Publishing; 2008.

(268) Buscemi S, Cosentino L, Rosafio G, Morgana M, Mattina A, Sprini D, et al. Effects of hypocaloric diets with different glycemic indexes on endothelial function and glycemic variability in overweight and in obese adult patients at increased cardiovascular risk. Clinical Nutrition 2013;32(3):346-352.

(269) Lanza GA, Spera FR, Villano A, Russo G, Di Franco A, Lamendola P, et al. Effect of smoking on endothelium-independent vasodilatation. Atherosclerosis 2015;240(2):330-332.

(270) Marieb EN. Human Anatomy & Physiology. 5th ed. U.S.A.: Benjamin Cummings; 2001.

(271) O'Donnell VB, Freeman BA. Interactions between nitric oxide and lipid oxidation pathways: Implications for vascular disease. Circ Res 2001;88(1):12-21.

(272) Taddei S, Virdis A, Ghiadoni L, Versari D, Salvetti A. Endothelium, aging, and hypertension. Curr Hypertens Rep 2006;8(1):84-89.

205

(273) Taddei S, Virdis A, Ghiadoni L, Versari D, Salvetti A. Which endothelium-derived factors are really important in humans? Biol Chem 2006;387(2):151-157.

(274) Taddei S, Virdis A, Ghiadoni L, Magagna A, Salvetti A. Vitamin C improves endothelium-dependent vasodilation by restoring nitric oxide activity in essential hypertension. Circulation 1998;97(22):2222-2229.

(275) Taddei S, Virdis A, Mattei P, Ghiadoni L, Fasolo CB, Sudano I, et al. Hypertension causes premature aging of endothelial function in humans. Hypertension 1997;29(3):736-743.

(276) Dotan Y, Lichtenberg D, Pinchuk I. Lipid peroxidation cannot be used as a universal criterion of oxidative stress. Prog Lipid Res 2004;43(3):200-227.

(277) Hernández-Guerra M, García-Pagán JC, Turnes J, Bellot P, Deulofeu R, Abraldes JG, et al. Ascorbic acid improves the intrahepatic endothelial dysfunction of patients with cirrhosis and portal hypertension. Hepatology 2006;43(3):485-491.

(278) McRae MP. Vitamin C supplementation lowers serum low-density lipoprotein cholesterol and triglycerides: a meta-analysis of 13 randomized controlled trials. Journal of Chiropractic Medicine 2008;7(2):48-58.

(279) McRae MP. The efficacy of vitamin C supplementation on reducing total serum cholesterol in human subjects: a review and analysis of 51 experimental trials. Journal of Chiropractic Medicine 2006;5(1):2-12.

(280) Wayner DDM, Burton GW, Ingold KU, Barclay LRC, Locke SJ. The relative contributions of vitamin E, urate, ascorbate and proteins to the total peroxyl radical-trapping antioxidant activity of human blood plasma. BBA - General Subjects 1987;924(3):408-419.

(281) Wayner DDM, Burton GW, Ingold KU, Locke S. Quantitative measurement of the total, peroxyl radical-trapping antioxidant capability of human blood plasma by controlled peroxidation. The important contribution made by plasma proteins. FEBS Lett 1985;187(1):33- 37.

(282) Ghiselli A, Serafini M, Natella F, Scaccini C. Total antioxidant capacity as a tool to assess redox status: Critical view and experimental data. Free Radical Biology and Medicine 2000;29(11):1106-1114.

(283) Malminiemi K, Palomäki A, Malminiemi O. Comparison of LDL trap assay to other tests of antioxidant capacity; effect of vitamin E and lovastatin treatment. Free Radic Res 2000;33(5):581-593.

(284) Cao G, Prior RL. Comparison of different analytical methods for assessing total antioxidant capacity of human serum. Clin Chem 1998;44(6):1309-1315.

(285) Huang D, Boxin OU, Prior RL. The chemistry behind antioxidant capacity assays. J Agric Food Chem 2005;53(6):1841-1856.

206

(286) Valkonen MM, Kuusi T. Vitamin C prevents the acute atherogenic effects of passive smoking. Free Radical Biology and Medicine 2000;28(3):428-436.

(287) Valkonen M, Kuusi T. Spectrophotometric assay for total peroxyl radical-trapping antioxidant potential in human serum. J Lipid Res 1997;38(4):823-833.

(288) Karadag A, Ozcelik B, Saner S. Review of methods to determine antioxidant capacities. Food Analytical Methods 2009;2(1):41-60.

(289) Ahotupa M, Ruutu M, Mäntylä E. Simple methods of quantifying oxidation products and antioxidant potential of low density lipoproteins. Clin Biochem 1996;29(2):139-144.

(290) Ahotupa M, Suomela J-, Vuorimaa T, Vasankari T. Lipoprotein-specific transport of circulating lipid peroxides. Ann Med 2010;42(7):521-529.

(291) Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Treatment of Hypertension. Canadian Journal of Diabetes 2013;37:S17-S18.

(292) Roman MJ, Devereux RB, Kizer JR, Okin PM, Lee ET, Wang W, et al. High Central Pulse Pressure Is Independently Associated With Adverse Cardiovascular Outcome. The Strong Heart Study. J Am Coll Cardiol 2009;54(18):1730-1734.

(293) Roman MJ, Devereux RB, Kizer JR, Lee ET, Galloway JM, Ali T, et al. Central pressure more strongly relates to vascular disease and outcome than does brachial pressure: The strong heart study. Hypertension 2007;50(1):197-203.

(294) Avolio A. Central aortic blood pressure and cardiovascular risk: A paradigm shift? Hypertension 2008;51(6):1470-1471.

(295) Van Bortel LM, Duprez D, Starmans-Kool MJ, Safar ME, Giannattasio C, Cockcroft J, et al. Clinical applications of arterial stiffness, task force III: Recommendations for user procedures. American Journal of Hypertension 2002;15(5):445-452.

(296) Van Bortel LMAB, Struijker-Boudier HAJ, Safar ME. Pulse pressure, arterial stiffness, and drug treatment of hypertension. Hypertension 2001;38(4):914-921.

(297) Millar JA, Lever AF. Implications of pulse pressure as a predictor of cardiac risk in patients with hypertension. Hypertension 2000;36(5):907-911.

(298) McEniery CM, Wallace S, MacKenzie IS, McDonnell B, Newby DE, Cockcroft JR, et al. Endothelial function is associated with pulse pressure, pulse wave velocity, and augmentation index in healthy humans. Hypertension 2006;48(4):602-608.

(299) Richardson CJ, Maki-Petaja KM, McDonnell BJ, Hickson SS, Wilkinson IB, McEniery CM. Comparison of estimates of central systolic blood pressure and peripheral augmentation index obtained from the Omron HEM-9000AI and SphygmoCor systems. Artery Research 2009;3(1):24-31.

207

(300) Chen C-, Nevo E, Fetics B, Pak PH, Yin FCP, Maughan WL, et al. Estimation of Central aortic pressure waveform by mathematical transformation of radial tonometry pressure: Validation of generalized transfer function. Circulation 1997;95(7):1827-1836.

(301) Hope SA, Meredith IT, Cameron JD. Arterial transfer functions and the reconstruction of central aortic waveforms: Myths, controversies and misconceptions. J Hypertens 2008;26(1):4- 7.

(302) Wilkinson IB, Fuchs SA, Jansen IM, Spratt JC, Murray GD, Cockcroft JR, et al. Reproducibility of pulse wave velocity and augmentation index measured by pulse wave analysis. J Hypertens 1998;16(12 SUPPL.):2079-2084.

(303) Saba PS, Roman MJ, Pini R, Spitzer M, Ganau A, Devereux RB. Relation of arterial pressure waveform to left ventricular and carotid anatomy in normotensive subjects. J Am Coll Cardiol 1993;22(7):1873-1880.

(304) Asmar R, Benetos A, Topouchian J, Laurent P, Pannier B, Brisac - AM, et al. Assessment of arterial distensibility by automatic pulse wave velocity measurement: Validation and clinical application studies. Hypertension 1995;26(3):485-490.

(305) OMRON Healthcare Professional Services. OMRON HEM-9000AI: Non-invasive Blood Pressure Monitor with Augmentation Index Instruction Manual. : OMRON Healthcare; NA.

(306) Yasmin, Brown MJ. Similarities and differences between augmentation index and pulse wave velocity in the assessment of arterial stiffness. QJM - Monthly Journal of the Association of Physicians 1999;92(10):595-600.

(307) Little WC, Applegate RJ. Congestive heart failure: Systolic and diastolic function. J Cardiothorac Vasc Anesth 1993;7(4 SUPPL. 2):2-5.

(308) Boutouyrie P, Fliser D, Goldsmith D, Covic A, Wiecek A, Ortiz A, et al. Assessment of arterial stiffness for clinical and epidemiological studies: Methodological considerations for validation and entry into the European Renal and Cardiovascular Medicine registry. Nephrology Dialysis Transplantation 2014;29(2):232-239.

(309) Lukich E, Matas Z, Boaz M, Shargorodsky M. Increasing derangement of glucose homeostasis is associated with increased arterial stiffness in patients with diabetes, impaired fasting glucose and normal controls. Diabetes Metab Res 2010;26(5):365-370.

(310) Laurent S, Cockcroft J, Van Bortel L, Boutouyrie P, Giannattasio C, Hayoz D, et al. Expert consensus document on arterial stiffness: Methodological issues and clinical applications. Eur Heart J 2006;27(21):2588-2605.

(311) Greenfield JR, Samaras K, Chisholm DJ, Campbell LV. Effect of postprandial insulinemia and insulin resistance on measurement of arterial stiffness (augmentation index). Int J Cardiol 2007;114(1):50-56.

208

(312) Nichols WW, Singh BM. Augmentation index as a measure of peripheral vascular disease state. Curr Opin Cardiol 2002;17(5):543-551.

(313) Lemogoum D, Flores G, Van Den Abeele W, Ciarka A, Leeman M, Degaute JP, et al. Validity of pulse pressure and augmentation index as surrogate measures of arterial stiffness during beta-adrenergic stimulation. J Hypertens 2004;22(3):511-517.

(314) Sa da Fonseca LJ, Mota-Gomes MA, Rabelo LA. Radial applanation tonometry as an adjuvant tool in the noninvasive arterial stiffness and blood pressure assessment. World Journal of Cardiovascular Diseases 2014;4:225-235.

(315) Van Den Hurk K, Smulders YM, Alssema M, Nijpels G, Kamp O, Paulus WJ, et al. Independent associations of glucose status and arterial stiffness with left ventricular diastolic dysfunction: An 8-year follow-up of the Hoorn Study. Diabetes Care 2012;35(6):1258-1264.

(316) Jovanovski E, Jenkins A, Dias AG, Peeva V, Sievenpiper J, Arnason JT, et al. Effects of korean red ginseng (Panax ginseng C.A. Mayer) and its isolated ginsenosides and polysaccharides on arterial stiffness in healthy individuals. American Journal of Hypertension 2010;23(5):469-472.

(317) Mucalo I, Jovanovski E, Rahelic D, Božikov V, Romic Z, Vuksan V. Effect of American ginseng (Panax quinquefolius L.) on arterial stiffness in subjects with type-2 diabetes and concomitant hypertension. J Ethnopharmacol 2013;150(1):148-153.

(318) Otsuka T, Kawada T, Ibuki C, Kusama Y. Obesity as an independent influential factor for reduced radial arterial wave reflection in a middle-aged Japanese male population. Hypertension Research 2009;32(5):387-391.

(319) Rezai M-, Goudot G, Winters C, Finn JD, Wu FC, Cruickshank JK. Calibration mode influences central blood pressure differences between SphygmoCor and two newer devices, the Arteriograph and Omron HEM-9000. Hypertension Research 2011;34(9):1046-1051.

(320) Jenkins DJA, Wolever TMS, Nineham R, Sarson DL, Bloom SR, Ahern J, et al. Improved glucose tolerance four hours after taking guar with glucose. Diabetologia 1980;19(1):21-24.

(321) Jenkins DJ, Jenkins AL, Wolever TM, Collier GR, Rao AV, Thompson LU. Starchy foods and fiber: reduced rate of digestion and improved carbohydrate metabolism. Scandinavian Journal of Gastroenterology, Supplement 1987;129:132-141.

(322) Wolever TMS, Jenkins DJA, Ocana AM, Rao VA, Collier GR. Second-meal effect: Low- glycemic-index foods eaten at dinner improve subsequent breakfast glycemic response. Am J Clin Nutr 1988;48(4):1041-1047.

(323) Jenkins DJA, Wolever TMS, Vuksan V, Brighenti F, Cunnane SC, Rao AV, et al. Nibbling versus gorging: Metabolic advantages of increased meal frequency. N Engl J Med 1989;321(14):929-934.

209

(324) Grysman A, Carlson T, Wolever TMS. Effects of sucromalt on postprandial responses in human subjects. Eur J Clin Nutr 2008;62(12):1364-1371.

(325) Ceriello A, Quagliaro L, Piconi L, Assaloni R, Da Ros R, Maier A, et al. Effect of Postprandial Hypertriglyceridemia and Hyperglycemia on Circulating Adhesion Molecules and Oxidative Stress Generation and the Possible Role of Simvastatin Treatment. Diabetes 2004;53(3):701-710.

(326) Jenkins DJA, Khan A, Jenkins AL, Illingworth R, Pappu AS, Wolever TMS, et al. Effect of nibbling versus gorging on cardiovascular risk factors: Serum uric acid and blood lipids. Metab Clin Exp 1995;44(4):549-555.

(327) Clissold SP, Edwards C. Acarbose. A preliminary review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential. Drugs 1988;35(3):214-243.

(328) Dimitriadis G, Tessari P, Go V, Gerich J. Effects of the disaccharidase inhibitor acarbose on meal and intravenous glucose tolerance in normal man. Metabolism: Clinical and Experimental 1982;31(8):841-843.

(329) Gérard J, Luyckx AS, Lefebvre PJ. Improvement of metabolic control in insulin dependent diabetics treated with the a-glucosidase inhibitor Acarbose for two months. Diabetologia 1981;21(5):446-451.

(330) Wolever TMS, Bentum-Williams A, Jenkins DJA. Physiological modulation of plasma free fatty acid concentrations by diet: Metabolic implications in nondiabetic subjects. Diabetes Care 1995;18(7):962-970.

(331) Wolever TMS, Bhaskaran K. Use of glycemic index to estimate mixed-meal glycemic response. American Journal of Clinical Nutrition 2012;95:256-257.

(332) Venn B, Brown R, Dodd H. Reply to TMS Wolever and K Bhaskaran. American Journal of Clinical Nutrition 2012;95:257-258.

(333) Kinnear T, Wolever TMS, Murphy AM, Sullivan JA, Liu Q, Bizimungu B. Effect of preparation method on the glycaemic index of novel potato clones. Food and Function 2011;2(8):438-444.

(334) Rammos G, Peppes V, Zakopoulos N. Transient insulin resistance in normal subjects: Acute hyperglycemia inhibits endothelial-dependent vasodilatation in normal subjects. Metabolic Syndrome and Related Disorders 2008;6(3):159-170.

(335) Mullan BA, Ennis CN, Fee HJP, Young IS, McCance DR. Protective effects of ascorbic acid on arterial hemodynamics during acute hyperglycemia. American Journal of Physiology - Heart and Circulatory Physiology 2004;287(3 56-3):H1262-H1268.

(336) Kawano H, Motoyama T, Hirashima O, Hirai N, Miyao Y, Sakamoto T, et al. Hyperglycemia rapidly suppresses flow-mediated endothelium-dependent vasodilation of brachial artery. J Am Coll Cardiol 1999;34(1):146-154.

210

(337) Cerielo A, Motz E, Cavarape A, Lizzio S, Russo A, Quatraro A, et al. Hyperglycemia counterbalances the antihypertensive effect of glutathione in diabetic patients: Evidence linking hypertension and glycemia through the oxidative stress in diabetes mellitus. J Diabetes Complications 1997;11(4):250-255.

(338) Lavi T, Karasik A, Koren-Morag N, Kanety H, Feinberg MS, Shechter M. The acute effect of various glycemic index dietary carbohydrates on endothelial function in nondiabetic overweight and obese subjects. J Am Coll Cardiol 2009;53(24):2283-2287.

(339) Philippou E, Bovill-Taylor C, Rajkumar C, Vampa ML, Ntatsaki E, Brynes AE, et al. Preliminary report: the effect of a 6-month dietary glycemic index manipulation in addition to healthy eating advice and weight loss on arterial compliance and 24-hour ambulatory blood pressure in men: a pilot study. Metabolism: Clinical and Experimental 2009;58(12):1703-1708.

(340) Jenkins DJA, Kendall CWC, Augustin LSA, Franceschi S, Hamidi M, Marchie A, et al. Glycemic index: Overview of implications in health and disease. Am J Clin Nutr 2002;76(1):266S-273S.

(341) Evans D. Hierarchy of evidence: A framework for ranking evidence evaluating healthcare interventions. J Clin Nurs 2003;12(1):77-84.

(342) Evans KD, Boyce KE. Fostering a hierarchy of evidence within the profession. Journal of Diagnostic Medical Sonography 2008;24(3):183-188.

(343) Bonell C, Fletcher A, Morton M, Lorenc T, Moore L. Realist randomised controlled trials: A new approach to evaluating complex public health interventions. Social Science and Medicine 2012;75(12):2299-2306.

(344) Nutbeam D, Smith C, Catford J. Evaluation in health education. A review of progress, possibilities, and problems. J Epidemiol Community Health 1990;44(2):83-89.

(345) Glasgow RE, Vogt TM, Boles SM. Evaluating the public health impact of health promotion interventions: The RE-AIM framework. Am J Public Health 1999;89(9):1322-1327.

(346) Rychetnik L, Frommer M, Hawe P, Shiell A. Criteria for evaluating evidence on public health interventions. J Epidemiol Community Health 2002;56(2):119-127.

(347) Attree M. Evaluating healthcare education: Issues and methods. Nurse Educ Today 2006;26(8):640-646.

(348) Grant S. An Overview of the Promotion of Breastfeeding Intervention Trial (PROBIT). In-Touch 2009;26(4):1-6.

(349) Kramer MS, Martin RM, Sterne JAC, Shapiro S, Dahhou M, Platt RW. The double jeopardy of clustered measurement and cluster randomisation. BMJ (Online) 2009;339(7719):503-505.

211

(350) Bowen DJ, Meischke H, Tomoyasu N. Preliminary evaluation of the processes of changing to a low-fat diet. Health Educ Res 1994;9(1):85-94.

(351) Bates R. A critical analysis of evaluation practice: The Kirkpatrick model and the principle of beneficence. Eval Program Plann 2004;27(3):341-347.

(352) Anderson ES, Thorpe LN. Students improve patient care and prepare for professional practice: An interprofessional community-based study. Med Teach 2014;36(6):495-504.

(353) Rycroft-Malone J, Kitson A, Harvey G, McCormack B, Seers K, Titchen A, et al. Ingredients for change: Revisiting a conceptual framework. Quality and Safety in Health Care 2002;11(2):174-180.

(354) Boet S, Bould MD, Fung L, Qosa H, Perrier L, Tavares W, et al. Transfer of learning and patient outcome in simulated crisis resource management: A systematic review. Canadian Journal of Anesthesia 2014;61(6):571-582.

(355) Lin S-, Chien L-, Tai C-, Lee C-. Effectiveness of a prenatal education programme on breastfeeding outcomes in Taiwan. J Clin Nurs 2008;17(3):296-303.

(356) Leslie K, Baker L, Egan-Lee E, Esdaile M, Reeves S. Advancing faculty development in medical education: A systematic review. Academic Medicine 2013;88(7):1038-1045.

(357) Siddiqui ZS, Jonas-Dwyer DRD. Mobile learning in health professions education: A systematic review. Pedagogical Applications and Social Effects of Mobile Technology Integration 2013:193-205.

(358) Hauer KE, Carney PA, Chang A, Satterfield J. Behavior change counseling curricula for medical trainees: A systematic review. Academic Medicine 2012;87(7):956-968.

(359) Buckley S, Coleman J, Davison I, Khan KS, Zamora J, Malick S, et al. The educational effects of portfolios on undergraduate student learning: A Best Evidence Medical Education (BEME) systematic review. BEME Guide No. 11. Med Teach 2009;31(4):282-298.

(360) Dhliwayo S, Nyanumba LK. An evaluation of an on the job training program at a UK based public health care company. Problems and Perspectives in Management 2014;12(2):164- 172.

(361) Militello LK, Gance-Cleveland B, Aldrich H, Kamal R. A methodological quality synthesis of systematic reviews on computer-mediated continuing education for healthcare providers. Worldviews on Evidence-Based Nursing 2014;11(3):177-186.

(362) Eseyrel D. Approaches to evaluation of training: theory & practice. Educational Technology & Society 2002;5:93-98.

(363) O'Hea EL, Boudreaux ED, Jeffries SK, Carmack Taylor CL, Scarinci IC, Brantley PJ. Stage of change movement across three health behaviors: The role of self-efficacy. American Journal of Health Promotion 2004;19(2):94-102.

212

(364) Graham M, Slovinec-D'Angelo M, L. Evaluation of dietary changes following heart healthy eating workshop. Journal of Cardiopulmonary Rehabilitation and Prevention 2008;28:342-343.

(365) Miller CK, Gutshcall MD, Mitchell DC. Change in food choices following a glycemic load intervention in adults with type 2 diabetes. J Am Diet Assoc 2009;109(2):319-324.

(366) Miller CK, Gutschall MD, Lawrence F. The development of self-efficacy and outcome expectation measures regarding glycaemic load and the nutritional management of type 2 diabetes. Public Health Nutr 2007;10(6):628-634.

(367) Trinick TR, Laker MF, Johnston DG, Keir M, Buchanan KD, Alberti KG. Effect of guar on second-meal glucose tolerance in normal man. Clin Sci 1986;71(1):49-55.

(368) Ceriello A. The emerging role of post-prandial hyperglycaemic spikes in the pathogenesis of diabetic complications. Diabetic Med 1998;15(3):188-193.

(369) Title LM, Cummings PM, Giddens K, Nassar BA. Oral glucose loading acutely attenuates endothelium-dependent vasodilation in healthy adults without diabetes: An effect prevented by vitamins C and E. J Am Coll Cardiol 2000;36(7):2185-2191.

(370) Mohanty P, Hamouda W, Garg R, Aljada A, Ghanim H, Dandona P. Glucose challenge stimulates reactive oxygen species (ROS) generation by leucocytes. Journal of Clinical Endocrinology and Metabolism 2000;85(8):2970-2973.

(371) Canadian Centre on Substance Abuse. Drinking Guidelines. 2015; Available at: http://www.ccsa.ca/Eng/topics/alcohol/drinking-guidelines/Pages/default.aspx. Accessed May/9, 2015.

(372) Aldughpassi A, Wolever TMS. Effect of coffee and tea on the glycaemic index of foods: No effect on mean but reduced variability. Br J Nutr 2009;101(9):1282-1285.

(373) Peterson GL. A simplification of the protein assay method of Lowry et al. Which is more generally applicable. Anal Biochem 1977;83(2):346-356.

(374) Wagner ES, Lindley B, D. Coffin R. High-performance liquid chromatographic determination of ascorbic acid in urine. Effect on urinary excretion profiles after oral and intravenous administration of vitamin C. Journal of Chromatography B: Biomedical Sciences and Applications 1979;163(2):225-229.

(375) Singer JD. Using SAS PROC MIXED to Fit Multilevel Models, Hierarchical Models, and Individual Growth Models. Journal of Educational and Behavioral Statistics 1998;23(4):323- 355.

(376) Ceriello A, Colagiuri S, Gerich J, Tuomilehto J. Guideline for management of postmeal glucose. Nutrition, Metabolism and Cardiovascular Diseases 2008;18(4):S17-S33.

213

(377) Miglio C, Peluso I, Raguzzini A, Villaño DV, Cesqui E, Catasta G, et al. Fruit juice drinks prevent endogenous antioxidant response to high-fat meal ingestion. Br J Nutr 2014;111(2):294- 300.

(378) Miglio C, Peluso I, Raguzzini A, Villaño DV, Cesqui E, Catasta G, et al. Antioxidant and inflammatory response following high-fat meal consumption in overweight subjects. Eur J Nutr 2013;52(3):1107-1114.

(379) Cadenas E, Packer L. Handbook of Antioxidants. 2nd edition ed. New York: CRC Press; 2002.

(380) Ceriello A, Bortolotti N, Crescentini A, Motz E, Lizzio S, Russo A, et al. Antioxidant defences are reduced during the oral glucose tolerance test in normal and non-insulin-dependent diabetic subjects. Eur J Clin Invest 1998;28(4):329-333.

(381) Jackson P, Loughrey CM, Lightbody JH, McNamee PT, Young IS. Effect of hemodialysis on total antioxidant capacity and serum antioxidants in patients with chronic renal failure. Clin Chem 1995;41(8):1135-1138.

(382) Esterbauer H, Striegl G, Puhl H, Oberreither S, Rotheneder M, El-Saadani M, et al. The role of vitamin E and carotenoids in preventing oxidation of low density lipoproteins. Ann N Y Acad Sci 1989;570:254-267.

(383) Turley SD, West CE, Horton BJ. The role of ascorbic acid in the regulation of cholesterol metabolism and in the pathogenesis of atherosclerosis. Atherosclerosis 1976;24(1-2):1-18.

(384) Hallfrisch J, Singh VN, Muller DC, Baldwin H, Bannon ME, Andres R. High plasma vitamin C associated with high plasma HDL- and HDL2 cholesterol. Am J Clin Nutr 1994;60(1):100-105.

(385) Jacques PF, Sulsky SI, Perrone GE, Jenner J, Schaefer EJ. Effect of vitamin C supplementation on lipoprotein cholesterol, apolipoprotein, and triglyceride concentrations. Ann Epidemiol 1995;5(1):52-59.

(386) Anonymous (Unpublished). Ascorbic acid and the catabolism of cholesterol. Nutr Rev 1973;31(5):154-156.

(387) Myasnikova I. Effect of ascorbic acid, nicotinic acid and thiamine on cholestolemia. Voenno Morski Med Akad Leningrad 1947;8:140.

(388) Anderson RA, Evans LM, Ellis GR, Khan N, Morris K, Jackson SK, et al. Prolonged deterioration of endothelial dysfunction in response to postprandial lipaemia is attenuated by vitamin C in Type 2 diabetes. Diabetic Medicine 2006;23(3):258-264.

(389) Hillstrom RJ, Yacapin-Ammons AK, Lynch SM. Vitamin C inhibits lipid oxidation in human HDL. J Nutr 2003;133(10):3047-3051.

214

(390) Moreau KL, Gavin KM, Plum AE, Seals DR. Ascorbic acid selectively improves large elastic artery compliance in postmenopausal women. Hypertension 2005;45(6):1107-1112.

(391) Sievenpiper JL, Jenkins DJA, Josse RG, Vuksan V. Dilution of the 75-g oral glucose tolerance test improves overall tolerability but not reproducibility in subjects with different body compositions. Diabetes Res Clin Pract 2001;51(2):87-95.

(392) Sievenpiper JL, Jenkins DJA, Josse RG, Vuksan V. Dilution of the 75-g oral glucose tolerance test increases postprandial glycemia: Implications for diagnostic criteria. CMAJ 2000;162(7):993-996.

(393) Sievenpiper JL, Vuksan V, Wong EYY, Mendelson RA, Bruce-Thompson C. Effect of meal dilution on the postprandial glycemic response: Implications for glycemic testing. Diabetes Care 1998;21(5):711-716.

(394) Brouns F, Senden J, Beckers EJ, Saris WHM. Osmolarity does not affect the gastric emptying rate of oral rehydration solutions. J Parenter Enteral Nutr 1995;19(5):403-406.

(395) Cleary J, Casey S, Hofsteede C, Moses RG, Milosavljevic M, Brand-Miller J. Does a low glycaemic index (GI) diet cost more during pregnancy? Nutrients 2012;4(11):1759-1766.

(396) Osborne H. Health Literacy From A to Z: Practical Ways to Communicate Your Health Message. 2nd edition ed. Burlington, Massachusetts, U.S.A.: Joan and Bartlett Learning; 2013.

(397) Wink J. Critical Pedagogy: Notes from the Real World. 4th edition ed. Upper Saddle River, New Jersey: Pearson Education Incorportated; 2011.

(398) Emrich T, Mendoza J, L'Abbe M. Effectiveness of Front-of-pack Nutrition Symbols: A Pilot Study with Consumers. Canadian Journal of Dietetic Practice and Research 2012;73(4):200-203.

(399) Canadian Diabetes Association (CDA). Five Elements that Promote Self-Mangement Infographics: 5 Characteristics of Effective Goal Setting. 2013; Available at: http://guidelines.diabetes.ca/SelfManagementEducation/SME5Infographic. Accessed September/15, 2014.

(400) Health Canada. Canadian Nutrient File - Search By Food. 2012; Available at: http://webprod3.hc-sc.gc.ca/cnf-fce/index-eng.jsp. Accessed April/ 04, 2015.

(401) Peterson RA. Constructing Effective Questionnaires. California, U.S.A.: SAGE Publications Incorporated; 2000.

(402) Sudore RL, Stewart AL, Knight SJ, McMahan RD, Feuz M, Miao Y, et al. Development and Validation of a Questionnaire to Detect Behavior Change in Multiple Advance Care Planning Behaviors. PLoS ONE 2013;8(9).

(403) Lapelle NR. Simplifying Qualitative Data Analysis Using General Purpose Software Tools. Preventative and Behavioural Medicine Publications and Presentations 2004(84):1-18.

215

(404) Taylor-Powell E, Renner M. Analyzing Qualitative Data. Madison, Wisconsin, U.S.A.: Cooperative Extension Publishing Operations; 2003.

(405) MacQueen KM, McLellan E, Kay K, Milstein B. Codebook development for team-based qualitative analysis. Field Methods 1998;10(2):31-36.

(406) Miles MB, Huberman AM. An Expanded Sourcebook: Qualitative Data Analysis. 2nd edition ed. Thousand Oaks, California, U.S.A.: SAGE Publications Incorporated; 1994.

(407) Willms DG, Best JA, Taylor DW, Gilbert JR, Wilson DMC, Lindsay EA, et al. A systematic approach for using qualitative methods in primary prevention research. Medical Anthropology Quarterly 1990;4(4):391-409.

(408) Lieffers JRL, Vance VA, Hanning RM. Use of mobile device applications: In Canadian dietetic practice. Canadian Journal of Dietetic Practice and Research 2014;75(1):41-47.

(409) Wansink B. Empowering Nutrition Gatekeepers: Policy with a Small "p". Journal of Nutrition Education and Behavior 2012;44(2):97.

(410) Wansink B. Empowering nutrition gatekeepers: The products. Journal of Nutrition Education and Behavior 2012;44(1):1.

(411) Wansink B. Profiling nutritional gatekeepers: Three methods for differentiating influential cooks. Food Quality and Preference 2003;14(4):289-297.

(412) De Brún A, McCarthy M, McKenzie K, McGloin A. " Fat is your fault". Gatekeepers to health, attributions of responsibility and the portrayal of gender in the Irish media representation of obesity. Appetite 2013;62:17-26.

(413) Seneviratne SN, McCowan LME, Cutfield WS, Derraik JGB, Hofman PL. Exercise in pregnancies complicated by obesity: Achieving benefits and overcoming barriers. Obstet Gynecol 2015;212(4):442-449.

(414) Downs DS, Ulbrecht JS. Understanding exercise beliefs and behaviors in women with gestational diabetes mellitus. Diabetes Care 2006;29(2):236-240.

(415) Hartmann S, Bung P. Physical exercise during pregnancy - Physiological considerations and recommendations. J Perinat Med 1999;27(3):204-215.

(416) Wolever TMS, Jenkins DJA, Vuksan V, Jenkins AL, Wong GS, Josse RG. Beneficial effect of low-glycemic index diet in overweight NIDDM subjects. Diabetes Care 1992;15(4):562-564.

(417) Schulz KF, Grimes DA. Allocation concealment in randomised trials: Defending against deciphering. Lancet 2002;359(9306):614-8.

(418) Schulz KF, Grimes DA. Unequal group sizes in randomised trials: Guarding against guessing. Lancet 2002;359(9310):966-970.

216

(419) Jones EJ, Roche CC, Appel SJ. A review of the health beliefs and lifestyle behaviors of women with previous gestational diabetes. JOGNN - Journal of Obstetric, Gynecologic, and Neonatal Nursing 2009;38(5):516-526.

(420) Omand JA, Carsley S, Darling PB, Parkin PC, Birken CS, Urquia ML, et al. Evaluating the accuracy of a geographic closed-ended approach to ethnicity measurement, a practical alternative. Ann Epidemiol 2014;24(4):246-253.

217

8.0. APPENDIX

Appendix Appendix Title Page Number Number Chapter 2 Appendix 2.1. Glycaemic Index Value Search and Assignment Procedure 218 Chapter 3 Appendix 3.1. Screening Questionnaire (Baseline Lifestyle) 219 3.2. Total Peroxyl Radical Trapping Antioxidant Potential Assay 225 3.3 Baseline Conjugated Diene Quantification (Assay) 230 3.4. Procedure for LDL protein Quantification (Assay) 233 3.5. Methodology: Wako NEFA-HR(2) Microtiter Procedure 236 3.6. Lipid Analysis Procedure: Total Cholesterol, HDL 238 Cholesterol and Triglyceride 3.7. 24 hour Recall 240 3.8. Symptoms Questionnaire 241 Chapter 4 Appendix 4.1. Presentations (Samples): Formative Feedback Initiatives 242 4.2. Three Day Diet Record with Instructions 252 4.3. Glycaemic Index Questionnaire Section 3 259 (with answer key): Glycaemic Index Knowledge 4.4. Glycaemic Index Questionnaire Section 4a: Glycaemic 263 Index Application and Acceptability 4.5. Glycaemic Index Questionnaire Qualitative Data Codebook 267 Chapter 5 Appendix 5.1. Complete list of study outcomes for The effect of a low 268 Glycaemic Index Diet on Maternal and Neonatal Markers of Glycaemic Control and Postpartum Diabetes Risk 5.2. Low Glycaemic Index Food Substitution List 269 5.3. Low Glycaemic Index PowerPoint Presentation 277 5.4. Low Glycaemic Index Recipe Book (Title Page and Table of 281 Contents) 5.5. Glycaemic Index Questionnaire Section 1: Participant 283 Satisfaction 5.6. Glycaemic Index Questionnaire Section 2: Getting to Know 287 You (Demographic Information)

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293