Marjan Tabesh

Understanding the Relationship between Obesity/ Fat distribution/Metabolic Profiles and Vitamin D Status in Young Women

(The Safe-D study)

Marjan Tabesh

Supervisors Professor John D. Wark Professor Suzanne M. Garland Associate Professor Alison Nankervis Mrs Alexandra Gorelik

Submitted in total fulfilment of the requirements of the degree of Doctor of Philosophy

March 2018

The University of Melbourne, Faculty of Medicine, Dentistry and Health Sciences, Department of Medicine Royal Melbourne hospital

Table of Contents ABSTRACT ...... 1 PREFACE AND ACKNOWLEDGEMENTS ...... 3 DECLARATION ...... 4 CONTRIBUTION...... 5 THESIS LAYOUT...... 7 ABBREVIATIONS ...... 9 PUBLICATIONS AND PRESENTATIONS ...... 12 GENERAL AIM ...... 14 Chapter 1 ...... 15 INTRODUCTION AND LITERATURE REVIEW ...... 15 1.1 Introduction ...... 16 1.1.1 Cardiovascular disease ...... 16 1.1.2 Obesity ...... 17 1.1.3 Body composition ...... 19 1.1.4 Diabetes ...... 20 1.1.5 Vitamin D ...... 22 1.1.6 Vitamin D, calcium and parathyroid hormone ...... 26 1.1.7 Mobile health technologies and health outcomes ...... 26 1.2 Mechanisms of association of vitamin D and CVD risk factors ...... 28 1.2.1 Mechanisms of association of vitamin D and lipid profiles ...... 28 1.2.2 Mechanisms of association of vitamin D and blood pressure ...... 29 1.2.3 Mechanisms of vitamin D and potential direct effects on atherosclerosis ...... 29 1.2.4 Mechanisms of association between vitamin D and obesity ...... 30 1.2.5 Mechanisms of association of vitamin D and body composition and fat distribution ...... 31 1.2.6 Mechanisms of association of vitamin D and diabetes...... 32 1.3 Literature review ...... 33 1.3.1 Vitamin D and obesity ...... 35 1.3.2 Vitamin D and body composition/fat distribution ...... 42 1.3.3 Vitamin D and vascular health ...... 44 1.3.4 Vitamin D and inflammation ...... 46 1.3.5 Vitamin D and lipid profiles ...... 47 1.3.6 Vitamin D and blood pressure ...... 50

1.3.7 Vitamin D and diabetes mellitus ...... 53 1.3.8 Vitamin D deficiency and pregnancy ...... 57 1.3.9 Effects of behavioural intervention by using mobile health (m-health) on health outcomes ...... 58 1.4 Importance of the study ...... 59 1.5 Rationale...... 60 1.6 Aims ...... 62 1.7 Methodological approach and theoretical framework ...... 63 Chapter 2 ...... 65 MATERIALS AND METHODS ...... 65 2.1 Study design ...... 66 2.2 Subject selection ...... 67 2.2.1 Inclusion criteria ...... 67 2.2.2 Exclusion criteria ...... 68 2.3 Proposed sample size ...... 70 2.4 Recruitment ...... 71 2.5 Verbal, written and electronic consent ...... 75 2.6 Trial interventions ...... 76 2.6.1 Behavioural intervention group ...... 76 2.6.2 Pharmacological intervention group ...... 77 2.6.3 Control group ...... 77 2.7 Randomisation ...... 78 2.8 Blinding ...... 78 2.9 Data collection...... 79 2.9.1 Questionnaire data ...... 84 2.9.2 Site visit assessment and rationale...... 86 2.9.3 Blood collection ...... 87 2.9.4 Physical examination ...... 91 2.9.5 Skin reflectance ...... 93 2.9.6 Pregnancy test ...... 93 2.9.7 Bone density and body composition scans ...... 93 2.9.8 Sun exposure/SunSmart behaviour ...... 94 2.9.9 Calculated variables ...... 96 2.10 Young Female Health Initiative (YFHI) study ...... 98

iii

2.11 Confounding ...... 98 2.12 Statistical analysis ...... 99 2.13 Data handling ...... 101 2.14 Compliance...... 101 2.15 Withdrawal ...... 102 2.16 Ethics and legal consideration ...... 103 2.17 Clinically-significant results ...... 105 2.18 Adverse events ...... 105 Chapter 3 ...... 107 RESULTS ...... 107 3.1 Numbers recruited in Part A and Part B ...... 108 3.2 Sample size re-calculation and power calculation ...... 109 3.3 Unblinding ...... 112 3.4 Protocol deviation ...... 112 3.5 Compliance rate...... 112 3.6 Number of adverse events ...... 114 3.7 General characteristics of participants ...... 115 3.7.1 Comparing general characteristics of participants in Safe-D study with YFHI study ...... 116 3.7.2 Comparing general characteristics of participants in Safe-D Part A (cross-sectional part) with Safe-D Part B baseline (intervention part) ...... 119 3.8 Effects of vitamin D supplementation and behavioural intervention on 25 OHD levels ...... 122 Chapter 4 ...... 123 ASSOCIATIONS BETWEEN VITAMIN D STATUS, ADIPOSITY AND LIPID PROFILES IN YOUNG WOMEN (CROSS-SECTIONAL PART) ...... 123 4.1 Abstract ...... 128 4.2 Introduction ...... 130 4.3 Materials and methods ...... 132 4.4 Results ...... 135 4.5 Discussion ...... 137 Chapter 5 ...... 150 THE EFFECTS OF VITAMIN D SUPPLEMENTATION AND BEHAVIOURAL INTERVENTION ON OBESITY AND METABOLIC PROFILES OVER 4 MONTHS AND ONE YEAR FOLLOW-UPS (RANDOMISED CLINICAL TRIAL PART) ...... 150

5.1 Improving vitamin D status by behavioural & pharmacological interventions; effects on cardiovascular disease risk factors ...... 151 5.1.1 Introduction ...... 156 5.1.2 Material and Methods ...... 157 5.1.3 Results ...... 162 5.1.4 Discussion ...... 164 5.2 The effects of vitamin D supplementation and behavioural intervention on metabolic profiles over one year follow-up ...... 177 Chapter 6 ...... 190 THE EFFECTS OF VITAMIN D SUPPLEMENTATION AND BEHAVIOURAL INTERVENTION ON BODY COMPOSITION OVER ONE YEAR FOLLOW-UP ...... 190 6.1 Introduction and literature review ...... 191 6.2 Aims ...... 193 6.3 Methods and Materials ...... 193 6.4 Results ...... 195 6.5 Discussion ...... 204 Chapter 7 ...... 207 SUN-RELATED BEHAVIOURS AMONG YOUNG AUSTRALIAN WOMEN AND EFFECTS OF USING M-HEALTH ON CHANGE IN SUN PROTECTION BEHAVIOUR ...... 207 7.1 Introduction ...... 208 7.1.1 Skin cancer ...... 208 7.1.2 Sun-related behaviour ...... 210 7.1.3 Sun exposure...... 210 7.1.4 Sun protection behaviour ...... 210 7.1.5 Suntanning behaviours ...... 211 7.1.6 Fake suntan ...... 211 7.1.7 Suntan attitudes...... 212 7.1.8 Balance between sun protection and receiving enough vitamin D ...... 213 7.1.9 Effects of using mobile-based application to improve sun-related behaviour ...... 213 7.2 Literature review ...... 214 7.2.1 Sun-related behaviour ...... 214 7.2.2 Sun-related behaviour and circulating 25 hydroxyvitamin D (25 OHD) levels .... 216 7.2.3 Effects of using mobile-based applications to improve sun-related behaviour ..... 217

7.3 Significance and Aims ...... 220 7.4 Methods and Materials ...... 221 7.5 Statistical analysis ...... 225 7.6 Results ...... 226 7.7 Discussion ...... 231 7.7.1 Sun exposure...... 231 7.7.2 Sun protection behaviour ...... 234 7.7.3 Suntan attitudes...... 235 7.7.4 Associations of 25 OHD levels and sun-related data ...... 236 7.7.5 Effects of using mobile based applications to improve sun-related behaviour ..... 237 7.7.6 Limitations ...... 239 7.7.7 Strengths ...... 240 7.8 Conclusion and future direction ...... 240 Chapter 8 ...... 263 CONCLUSIONS AND FUTURE PERSPECTIVES ...... 263 APPENDICES ...... 274 A) Protocol paper ...... 275 B) Facebook advertisement ...... 303 C) Safe-D Part B brochure ...... 304 D) Participant information sheet & consent form (PICF) ...... 306 E) Photos ...... 319 F) Limesurvey questionnaire ...... 321 G) Clothing chart ...... 339 H) Sun monitoring device log…………………………………………………………....341 BIBLIOGRAPHY ...... 347

List of figures

Figure 1.1 Vitamin D metabolism

Figure 2.1 Study design

Figure 2.2 Study protocol

Figure 3.1 Recruitment flowchart

Figure 6.1 Usual whole body DXA scan

Figure 6.2 Visceral fat analyses on whole body DXA scan

Figure 6.3 Gynoid fat analyses on whole body DXA scan

vii

Legends of tables

------

Table 2.1 Calculated sample size

Table 2.2 Data collected at each time point

Table 3.1 Number of calculated sample size and actual sample size

Table 3.2 Compliance rate in each time point of study

Table 3.3 General characteristics and socio-demographic factors of Safe-D and YFHI studies

Table 3.4 General characteristics and socio-demographic factors of Safe-D Part A and Part B

Table 3.5 Percentage of participants reach the optimal 25 OHD levels (>75 nmol/L)

Table 4.1 Baseline general characteristics of participants stratified by 25 OHD levels

Table 4.2 Summary of 25 OHD levels, metabolic profiles, blood pressure, anthropometric measurements and fat distribution by 25 OHD levels

Table 4.3 Association between 25 OHD levels and metabolic profiles, anthropometric measurements

Table 5.1.1 Baseline general characteristics of participants

Table 5.1.2 Summary of baseline 25 OHD, metabolic profiles, blood pressure and anthropometric measurements

Table 5.1.3 Mean (± standard deviation) change in 25 OHD, metabolic profiles and anthropometric measurements after 4 months of intervention

Table 5.1.4 Paired t-test or Wilcoxon comparing 25 OHD levels and CVD risk factors prior to and following 4 months of intervention

Table 5.1.5 Mean ±SD changes in 25 OHD, metabolic profiles and anthropometric measurements over 4 months intervention (per protocol analysis)

viii

Table 5.1.6 Paired t-test or Wilcoxon, comparing 25 OHD levels and CVD risk factors prior to and following 4 months intervention (per protocol analysis)

Table 5.1.7 Summary of adverse events occurred in each group

Table 5.2.1 The 12 months 25 OHD, metabolic profiles, blood pressure and anthropometric measurements of study participants (intention to treat analysis)

Table 5.2.2 Mean ±SD changes in 25 OHD, metabolic profiles and anthropometric measurements over 12 months intervention (intention to treat analysis)

Table 5.2.3 Paired t-test or Wilcoxon comparing 25 OHD levels and CVD risk factors prior to and following 12 months intervention (intention to treat analysis)

Table 5.2.4 The effects of vitamin D supplementation and behavioural intervention on metabolic profiles/anthropometric measurements (per protocol analysis)

Table 5.2.5 Paired t-test or Wilcoxon, comparing 25 OHD levels and CVD risk factors prior to and following 12 months intervention (per protocol analysis)

Table 6.1 The baseline 25 OHD and body composition of study participants

Table 6.2 Mean ±SD changes in 25 OHD and body composition over 12 months intervention

Table 6.3 Paired t-test or Wilcoxon, comparing 25 OHD levels and body composition prior to and following 12 months intervention

Table 6.4 Association of change in 25 OHD levels and change in body composition over 12 months, independent of group allocation

Table 7.1 Self-reported time spent in the sun in the cross-sectional part

Table 7.2 Self-reported sun protection behaviour in the cross-sectional part

Table 7.3 Self-reported suntan behaviour in the cross-sectional part

Table 7.4 Self-reported suntan attitude in the cross-sectional part

Table 7.5 Sun related behaviour of Safe-D Part A (n=407) and Part B baseline (n=123)

Table 7.6 Time spent in sun- trial part

Table 7.7 Time spent in sun changes over 12 months intervention

Table 7.8 Sun protection behaviour changes over 12 months intervention

Table 7.9 Sun risky behaviour score in each group at baseline, 4 months and 12 months

Table 7.10 Sun risky behaviour score changes over 12 months of intervention

Table 7.11 Suntan behaviour in each group at baseline, 4 months and 12 months

Table 7.12 Suntan protection behaviour score over 12 months of intervention

Table 7.13 Suntan protection behaviour score changes over 12 months of intervention

Table 7.14 Suntan attitude in each group at baseline, 4 months and 12 months

Table 7.15 Suntan attitude score changes over 12 months of intervention

Table 7.16 Association of 25 OHD levels with sun related behaviour change over 12 months of intervention

Table 7.17 Association of objective sun exposure with ethnicity, skin type, physical activity and skin sensitivity to sun exposure

Table 7.18 Association of 25 OHD levels and UV exposure (obtained from dosimeter)

Table 7.19 Association of 25 OHD levels and sun protection behaviour

ABSTRACT

Vitamin D deficiency is highly prevalent and associated with an increased risk of many chronic health conditions including obesity, cardiovascular disease and diabetes. Vitamin D deficiency affects millions of Australians, causing considerable suffering, economic loss and mortality. The aim of this study was to evaluate the associations of vitamin D and obesity/fat distribution/metabolic profiles and examine the effects of behavioural intervention and vitamin D supplementation on weight / fat distribution and metabolic profiles in young women.

This study was run in two parts. In Part A, we evaluated the associations of obesity/fat distribution/metabolic profiles and 25 hydroxyvitamin D (25 OHD) in a cross-sectional study of 407, 16-25 year-old women. In Part B in a controlled randomised clinical trial over a period of 12 months we examined the effects of behavioural intervention and vitamin D supplementation on weight, fat distribution and metabolic profiles in 123 young women with mild-to-moderate vitamin D deficiency (25 OHD between 25-75 nmol/L). Data were collected at baseline, 4 months and 12 months follow-ups using an online survey and when participants attended a site visit in a fasted state. Parameters including blood pressure, anthropometry, metabolic profiles, serum 25 OHD levels, and body composition (using dual energy X ray absorptiometry) were measured.

For the cross-sectional analysis, we combined 407 Safe-D Part A data with 150 participants from another study, the Young Female Health Initiative (YFHI). Cross-sectional analyses showed that after adjustment for covariates, higher 25 OHD levels were associated with greater high density lipoprotein (HDL) levels and greater triglyceride levels. They were also associated with lower body mass index (BMI), total fat percentage, visceral fat percentage, visceral fat to total fat ratio and lower trunk fat to total fat ratio. Although these associations

1 were statistically significant, they were very small in magnitude and of uncertain clinical significance. In the clinical trial part of the project, after four months of intervention, both vitamin D supplementation and the use of a mobile-based application (app), resulted in a significant increase in 25 OHD levels. There were no significant differences between the three groups in lipid profiles, BMI or glucose metabolism after 4 months intervention.

However, vitamin D supplementation resulted in a reduction in BMI, and behavioural intervention resulted in a significant reduction in haemoglobin A1c (HbA1c) when compared to baseline.

Both the per protocol and intention to treat analyses were used to evaluate the effects of vitamin D improvements on cardiovascular risk factors after 12 months follow-up. In total,

102 participants were included in the 12 months follow-up intention to treat analysis. There were no significant differences in glucose metabolism, lipid profiles, systolic and diastolic blood pressure, anthropometric measures, high sensitivity C-reactive protein (hs-CRP) levels and body composition among the three groups after 12 months of follow-up.

Results showed that the improvement in 25 OHD levels by either taking vitamin D supplements or increased safe sun exposure did not affect cardiovascular risk factors in healthy young women.

In chapter 7 we evaluated the effects of behavioural intervention over one year follow-up on sun related behaviours. Time spent in the sun did not change following behavioural intervention but sun risky behaviours decreased significantly in summer time in this group.

Moreover, in this study no significant associations were observed between ultraviolet radiation (UVR) exposure and skin type, ethnicity or skin sensitivity to sun light. However, significant positive association was observed between sun exposure and physical activity.

PREFACE AND ACKNOWLEDGEMENTS

This research project could not have been done without the support and encouragement I have received from a number of individuals. Firstly, I would like to thank my primary supervisor Professor John D. Wark for the continuing support throughout the period of this work, for his motivation and his immense knowledge. Secondly, I would like to thank my co- supervisors, Professor Suzanne M. Garland, Professor Alison Nankervis and Mrs Alexandra

Gorelik for their support and guidance.

I would also like to express my gratitude to all the Safe-D study and YFHI study team members especially Ph.D. student Emma Callegari, A/Professor Nicola Reavley, A/Professor

Marie Pirotta, Professor George Varigos, Professor Kim Bennell, Professor Anthony Jorm,

A/Professor Shanton Chang, Professor Peter Lee, Dr Dale Robinson, Ms Adele Rivers, Ms

Heather Robinson, Ms Anna Scobie, Ms Skye Maclean, Dr Ashwini Kale, Ms Sonia Louise

Romeo, Dr Asvini Subasinghe, Ms Amanda Hawker, Professor Anna-louise Ponsonby,

Professor Robyn Lucas, Dr Ashwin Swaminathan, Ms Rachel Slatyer, Ms Jessica Cargill, Ms

Jemma Christie, , Dr Yasmin Jayasinghe, Dr Catherine Segan, Ms Stefanie Hartley, Ms Elisa

Young, Dr Adrian Bickerstaffe, Ms Maria Bisignano, Ms Alison Brodie, Dr Peter Farlie, Mr

Sin-Hyeong Choi, Mr Peter Gies, Ms Kerryn King, Ms Stefanie Koneski, Ms Jen Makin, Mr

Oktay Tacar, Dr Johannes Willnecker, Professor Steve Howard (deceased). I also would like to thank all participants for their time and effort.

I would also like to thank the Boosted Human members for developing the Safe-D application, Cancer Council Victoria, Melbourne Health Pathology Service, Swisse Wellness

Company for providing the vitamin D supplements for this project and The Australian

Radiation Protection and Nuclear Safety Agency.

Last but not least I would like to thank my family, my parents, my sisters Maryam and

Mahtab and my husband, Amin, for supporting me spiritually throughout doing my Ph.D.

DECLARATION

This is to certify that

(i) This thesis comprises only my original work except where indicated in the text,

(ii) Due acknowledgement has been made in the text to all other material used,

(iii) The thesis is less than 100,000 words in length, exclusive of tables, maps,

bibliographies and appendices.

Marjan Tabesh Department of Medicine, Royal Melbourne Hospital, The University of Melbourne Parkville, Australia

CONTRIBUTION

The whole project was conceived and designed mainly by Professor John D. Wark. Professor

John D. Wark, Professor Suzanne M. Garland, Associate Professor Alison Nankervis and Mrs

Alexandra Gorelik supervised the study.

I have contributed to the conception and execution of my thesis project. My main PhD focus was on Part B, the clinical trial part of the project. I have participated in the data collection, data entry and management, statistical analyses, results interpretation and manuscript drafting for Part A of the project. I have also participated in the data collection, the organization and conduct of study visits, the performance of DXA scans, the handling of samples, data entry and management, statistical analyses, results interpretation and manuscript drafting for Part

My contribution to the work involved the following:

Thesis Publication title Nature and extend of my Co-authors name chapter contribution

John D. Wark 4 Association between 55%. Suzanne M. Garland vitamin D status, adiposity Alison Nankervis and lipid profiles in young data collection, data entry women (cross-sectional and management, Alexandra Gorelik part) cleaning, analysing and Asvini K. Subasinghe interpreting the data, statistical analyses, Emma T. Callegari conceptualisation and drafting of the manuscript

John D. Wark 5 The effects of vitamin D 55% Suzanne M. Garland supplementation and Alison Nankervis behavioural intervention data collection, the on obesity and metabolic organization and conduct Alexandra Gorelik profiles over 4 month and of study visits, the Emma T. Callegari one year follow-ups handling of samples, data (randomised clinical trial) entry and management, statistical analyses, results interpretation and

manuscript drafting

John D. Wark 6 The effects of vitamin D 55% Suzanne M. Garland supplementation and Alison Nankervis behavioural intervention data collection, the on body composition over organization and conduct Alexandra Gorelik one year follow-up of study visits, the Emma T. Callegari performance of DXA scans, the handling of samples, data entry and management, statistical analyses, results interpretation and manuscript drafting

John D. Wark 7 Sun-related behaviours 55% Suzanne M. Garland among young Australian Alison Nankervis women and effects of data collection, the using m-Health on change organization and conduct Alexandra Gorelik in sun protection of study visits, the Emma T. Callegari behaviour handling of samples, data entry and management, statistical analyses, results interpretation and manuscript drafting

THESIS LAYOUT

In the first chapter introduction and literature review on cardiovascular disease, obesity, body composition, diabetes, vitamin D and mobile health technologies and health outcomes are presented. Also the potential mechanisms suggested for the association of vitamin D and cardiovascular risk factors are provided in this chapter followed by the importance of the study, rationale and aims of the project. At the end of this chapter methodological approach and theoretical framework is presented.

Materials and methods for both the cross-sectional and the clinical trial parts are presented in chapter 2 in details. The study protocol was published in Journal of Medical Internet

Research which is presented in the appendix A.

Findings which are not reported in other chapters are presented in chapter 3, including sample size re-calculation and power calculation, number of unblinding happened, results on protocol deviation, compliance rate, number of adverse events and general characteristics of participants.

The cross-sectional findings on the association of vitamin D status and adiposity, lipid profiles and other CVD risk factors are presented in chapter 4. This chapter is a published paper.

Results from the randomised clinical trial part of the project are presented in chapter 5. This chapter is presented in two sections. The first section is prepared as a journal manuscript and this is in the process of publication. Data on improving vitamin D status by behavioural and pharmacological interventions and its effects on cardiovascular disease risk factors over 4 month of intervention are provided in this section. The 12 month follow-up data on the effects of vitamin D supplementation and behavioural intervention on metabolic profiles are presented in the second section of this chapter.

Chapter 6 is prepared as a journal paper to be submitted to The Journal of Clinical

Endocrinology & Metabolism. In this chapter the effects of vitamin D supplementation and behavioural intervention on body composition over one year of intervention are presented.

In chapter 7, the sun-related behaviours data are presented. This chapter starts with a short introduction on skin cancer, sun-related behaviours, sun exposure, sun protection behaviour, fake suntan and suntan attitude followed by a literature review. The method and materials, statistical analysis, results, discussion and conclusion on sun related behaviours from both cross-sectional part and clinical trial part are also presented in this chapter.

Chapter 8 presents final conclusions and future perspectives of the whole project.

ABBREVIATIONS

1,25(OH)2D: 1, 25 dihydroxyvitamin D

25 OHD: 25 hydroxyvitamin D

ABS: Australian Bureau of Statistics

ALT: Alanine Aminotransferase

ANOVA: Analysis of Variance

ANV: Australian Nutrient Values

ANZCTR: Australian New Zealand Clinical Trials Registry

ARPANSA: Australian Radiation Protection and Nuclear Safety Agency

ASSIST: Applied Suicide Intervention Skills Training

BMI: Body Mass Index

CCV: Cancer Council Victoria

CONSORT: Consolidated Standards of Reporting Trials

CI: Confidence Interval

CV: Coefficient of Variation

CMIA: Chemiluminescent Microparticle Immunoassay

CVD: Cardiovascular Disease

DBP: Diastolic Blood Pressure

DQES: Questionnaire for Epidemiological Studies

DXA: Dual-energy X-ray Absorptiometry

EDTA: Ethylenediamine tetraacetic acid eGFR: Estimated Glomerular Filtration Rate

ELISA: Enzyme-linked Immunosorbent Assay

FBE: Full Blood Examination

GGT: Gamma Glutamyl Transferase

GPS: Global Positioning System

HbA1c: Haemoglobin A1c

HDL: High Density Lipoprotein

HOMA-B: Homeostatic Model Assessment of Beta Cell Function

HOMA-IR: Homeostatic Model Assessment of Insulin Resistance

HPLC: High Performance Liquid Chromatography

HREC: Human Research and Ethics committee

Hs-CRP: High Sensitivity C-Reactive Protein

IDF: International Diabetes Federation

IL: Interleukine

ID: Identification

IQR: Inter Quartile Range

ITT: Intention to Treat

IU: International Unit

LC-MS/MS: Liquid Chromatography-tandem Mass Spectrometry

LDL: Low Density Lipoprotein

Log: Logarithm

MAR: Missing at Random

METs: Metabolic Equivalent m-health: Mobile Health

NADPH: Nicotinamide Adenine Dinucleotide Phosphate

NHANES: National Health and Nutrition Examination Survey

NHMRC: National Health and Medical Research Council

OR: Odds Ratio

PICF: Participant Information and Consent Form

PPARS: Peroxisome Proliferator-activated Receptors

PQCT: Peripheral Quantitative Computed Tomography

PTH: Parathyroid Hormone

QUICKI: Quantitative Insulin Sensitivity Check Index

RAAS: Renin Angiotensin System

RMH: Royal Melbourne Hospital

SBP: Systolic Blood Pressure

SD: Standard Deviation

SED: Standard Erythema Dose

SEIFA: Socio-economic Indexes For Areas

SPF: Sun Protection Factor

T2DM: Type 2 Diabetes

TC: Total Cholesterol

TG: Triglyceride

TNF-α: Tumour Necrosis Factor-alpha

TSH: Thyroid Stimulating Hormone

US: United States

UV: Ultraviolet Radiation

VDR: Vitamin D Receptor

VLDL: Very Low Density Lipoprotein

WHO: World Health Organization

WHR: Waist to Hip Ratio

WHtR: Waist to Height Ratio

YFHI: Young Female Health Initiative

PUBLICATIONS AND PRESENTATIONS

1. Marjan Tabesh, Suzanne M. Garland, Nicola Reavley, Stefanie Hartley, Alexandra Gorelik, Ashwini Kale and John D. Wark on behalf of the YFHI and Safe-D study groups. Vitamin D and Metabolic Status in Young Women; poster presented at ANZBMS conference 2014, Queenstown, New Zealand

2. Marjan Tabesh, Suzanne M. Garland, Emma T. Callegari, Alexandra Gorelik, Adele Rivers, Alison Nankervis, Ashwini Kale, Skye Maclean, and John D. Wark on behalf of the YFHI and Safe-D study group. The Link between Cardiovascular Risk Factors and Osteoporosis; poster presented at ANZBMS conference, 2015, Hobart, Australia

3. Marjan Tabesh, Suzanne M. Garland, Nicola Reavley, Stephanie Hartley, Alexandra Gorelik, Alison Nankervis, Ashwini Kale, Emma T. Callegari, John D. Wark and the YFHI/Safe-D study groups. Vitamin D and Cardiovascular Disease Risk in Young Women; poster presented at the 18th vitamin D workshop, 2015, Delft, Netherland

4. Tabesh M, Garland SM, Callegari ET, Gorelik A, Rivers A, Nankervis A, Kale A, and Wark JD on behalf of the YFHI and Safe-D study groups. The Link between Metabolic Syndrome and Osteoporosis; poster presented at the 19th vitamin D workshop, 2016, Boston, United States

5. Tabesh M, Garland SM, Gorelik A, Nankervis A, Maclean S, Callegari ET, Chang S, Heffernan K, Wark JD. Improving Vitamin D Status and Related Health in Young Women: The Safe-D study – Part-B. Journal of Medical Internet Research. 2016 May 10;5(2):e80.

6. Marjan Tabesh, Emma T. Callegari, Alexandra Gorelik, Suzanne M. Garland, Alison Nankervis, Asvini K. Subasinghe, and John D. Wark, on behalf of the YFHI and Safe-D study groups. Associations between 25-hydroxyvitamin D Levels, Body Composition and Metabolic Profiles in Young Women. European Journal of Clinical Nutrition. 2018 Jan 24. doi: 10.1038/s41430-018-0086-1. [Epub ahead of print]

7. Marjan Tabesh, Emma T. Callegari, Alexandra Gorelik, Suzanne M. Garland, Alison Nankervis, and John D. Wark, on behalf of the YFHI and Safe-D study groups. The Effects of Vitamin D Improvements on Cardiovascular Risk Factors in Young

Women: a Randomised Clinical Trial. Submitted to The Journal of Clinical Endocrinology & Metabolism , 8th March 2018

GENERAL AIM

The general aims of this project are:

1) To evaluate the association of serum 25 OHD levels and cardiovascular risk factors

including obesity, lipid profiles, glucose metabolism, blood pressure, fat distribution

and body composition in young women, living in Australia

2) To measure the effectiveness of a behavioural intervention (using mobile application)

and vitamin D supplementation to increase 25 OHD levels and its effects on changes

in obesity, lipid profiles, glucose metabolism, blood pressure, fat distribution and

body composition over 4 months and 12 months of intervention, compared with a

control group in young women

Chapter 1

INTRODUCTION AND LITERATURE REVIEW

1.1 Introduction

This project investigated the association of serum levels of 25 hydroxyvitamin D (25 OHD) and cardiovascular risk factors (obesity, lipid profiles, glucose metabolism, blood pressure, inflammatory biomarkers, fat distribution and body composition), and also investigated the effectiveness of a behavioural and a pharmacological intervention to increase 25 OHD levels over 4 months and 12 months follow-up, compared with a control group of young women aged 16 to 25 years.

1.1.1 Cardiovascular disease

Cardiovascular diseases (CVD) are one of the major health concerns and a main cause of death around the world [WHO, 2014]. The World Health Organization (WHO) reported that

17.5 million people die each year from CVD around the globe. Overall mortality from CVD increased around 41%, between the years 1990 and 2013 [WHO, 2014].

According to the Australian Health Survey in 2012, almost 3.7 million Australians were affected by CVD and on average every 12 minutes one Australian dies due to CVD

[Australian Bureau of Statistics, 2016]. Cardiovascular disease is responsible for around 20% of the burden of disease and it is the second major cause of disease burden in Australia

[Australian Bureau of Statistics, 2016].

CVD is defined as disorders of blood vessels and the heart, and includes heart failure, peripheral vascular disease, coronary heart disease, and stroke. Risk factors for CVD are divided into those which are non-modifiable or modifiable. Non-modifiable risk factors include older age, being male and genetic predisposition; modifiable risk factors include low socioeconomic status, sedentary lifestyle, smoking, unhealthy diet and using excessive amounts of alcohol. Other factors associated with CVD include obesity, high cholesterol, low

16 high density lipoprotein (HDL) levels and hypertension which are potentially modifiable

[Truthmann et al., 2015].

Many of these risk factors are common to other chronic diseases, particularly diabetes. The interactions between obesity, diabetes, CVD and a number of other chronic conditions are complex and not well understood. However, diabetes dramatically increases the risk of CVD

[Kannel et al., 1979]. Diabetic patients also have up to a five times greater risk of stroke and ten times greater risk of heart attack compared to those without diabetes [Australian Bureau of Statistics, 2017]. Insulin resistance is a condition whereby glucose cannot be readily taken up by cells and insulin is unable to increase uptake and utilization of glucose in the cells.

Although insulin resistance is more common among adults, it is also observed in early childhood, and can continue into adulthood. Insulin resistance is related to many chronic conditions, including metabolic syndrome, diabetes mellitus and CVD [Ginsberg et al.,

2000].

Atherosclerosis is a condition which can start from a young age and progresses with age. In some people, it develops in the 20s and 30s; however, it is more likely to become evident in older age [Hong et al., 2010]. Therefore, it is important to recognise cardiovascular risk factors and associated factors at a younger age to ameliorate the risk of developing CVD in older age. It has been suggested that vitamin D levels may be associated with CVD through affecting body weight, body composition, blood pressure, lipid profiles, insulin resistance and directly affecting vascular smooth muscle [Gouni-Berthold et al., 2009]. In our study, we evaluated the associations of vitamin D with CVD risk factors and tested whether improving vitamin D status can affect any of these risk factors.

1.1.2 Obesity

According to WHO, overweight and obesity in adults are defined as having a body mass index (BMI) between 25 and 30 kg/m2, and more than or equal to 30 kg/m2, respectively.

Obesity in children age 5 to 18 years is defined, as having BMI-for-age, more than 2 standard deviations above the WHO reference [Cole et al., 2000].

The prevalence of obesity is increasing across the world and is one of the biggest global health challenges. In 2014, 39% of the worldwide adult population was overweight and 13% was obese [Hruby et al., 2015]. These numbers have almost doubled between 1980 and 2014.

According to National Institute of Health report, about 30% of 6 to 19-year-old individuals are considered to be overweight or obese and around 17% considered obese. Based on

Australian Bureau of Statistics data, 35% of adults living in Australia are overweight and

28% are obese [Huse et al., 2018]. This means that only one third of Australian adults have a body weight in the normal range. The prevalence of obesity has increased in Australia from

19% in 1995 to 27% in 2012. This increase was greater in women (12%) than in men (8%).

The prevalence of overweight and obesity in Australian children is also high, 18% being overweight and 7% obese, with no significant differences between boys and girls (25% and

26% overweight or obese, respectively) [Hardy et al., 2018]. It has been estimated that the prevalence of obesity will increase from 20.5% to 33.9% % in adults by 2025 [Walls et al.,

2012].

Obesity is associated with an increased risk for many chronic diseases including diabetes mellitus, CVD [Poirier et al., 2006] and some types of malignancies [Pergola, 2013], all of which have contributed to morbidity and mortality over the last decades [Mathers et al.,

2006].

Obesity results from different contributing factors including sedentary lifestyle, unhealthy diet, low socioeconomic status, some medical conditions and genetics. Recently, other factors

18 including lower serum leptin levels and lower serum 25 OHD levels have been reported to be associated with obesity [Vanlint et al., 2013]. Given the high prevalence of obesity, and its association with chronic disease, we have investigated the association of 25 OHD and obesity. Further, we have assessed the effects of increasing vitamin D levels, either by pharmacological supplementation, or by behavioural intervention (using a mobile-based application to increase sun exposure), on obesity.

1.1.3 Body composition

Although BMI is a good estimate of obesity, it does not reflect body composition and cannot distinguish lean mass from fat mass or distribution of fat in the body. Recently, it has been suggested that body composition, including the amount and distribution of fat mass, and amount and distribution of lean mass, can be associated with many health outcomes, including CVD and diabetes [Despre´s et al., 2012]. Adipose tissue is no longer considered an organ just to store energy, as it has been shown that adipose tissue secretes more than 260 different hormones and proteins in the body. Adipose tissue mass is around 5 kg in very lean people and more than 50 kg in obese individuals [Coppack et al., 2012]. Excess adipose tissue has been shown to contribute to many diseases including diabetes, CVD and cancer.

Recent studies have reported that adipose tissue stored in different locations in the body has different impacts on health outcomes. In a study by Hu et al., it has been shown that higher trunk fat is related to an increase incardio-metabolic risk factors. However, higher arm adiposity was significantly associated with increasing cardio-metabolic risk factors among women, but not among men, and leg adiposity was correlated with decreasing risk of cardiometabolic disease [Hu et al., 2011].

Given that body weight and BMI are not reliable indicators of body fat and the possible association of vitamin D levels and total body fat mass, visceral fat mass, trunk fat mass and

19 lean mass, we examined the association of serum 25 OHD levels with body composition.

Moreover, we examined the effects of increased vitamin D levels on body composition and fat distribution, to explore possible causal relationships. Dual energy x-ray absorptiometry

(DXA) is used to measure body composition [Haarbo et al., 1991]. It is an enhanced form of x-ray technology that is used to assess the distribution of lean muscle, fat, water and bone within the body. DXA is a rapid method and associated with a very low dose of radiation exposure so it is safe to be used even in adolescents. DXA scans measure body composition and report lean mass, fat mass and bone mass in each region of the body including left and right arm, trunk, left and right leg, head and total body. To determine the visceral fat area a secondary analysis of the original scan is required.

1.1.4 Diabetes

Obesity, CVD and diabetes often have similar underlying causes, have common risk factors and have similar strategies for prevention, management and treatment [Leon et al., 2015].

Diabetes is the fastest growing chronic disease in the world [Ogurtsova et al., 2017]. In 2013, diabetes caused more than one million deaths globally [Ogurtsova et al., 2017]. The

International Diabetes Federations (IDF) in 2015 estimated that 415 million people have diabetes and almost 12 percent of the health expenses globally relate to diabetes treatment. It has been estimated that 642 million people globally will be affected by diabetes by 2040

[Ogurtsova et al., 2017].

The prevalence of diabetes in Australia is dramatically increasing [Davis et al., 2018]. In

2015, almost 1.7 million people in Australia were diagnosed with diabetes [Davis et al.,

2018]. Every day an additional 280 people are diagnosed with diabetes; this has a significant impact on burden of disease [Davis et al., 2018].

There are two main forms of diabetes. Type 1 diabetes is an auto-immune condition in which the body’s immune system attacks the pancreatic beta cells. Therefore, not enough insulin is produced in the pancreas. Type 1 diabetes is more common at a younger age and there is no known way to prevent it [Atkinsob et al., 2014].

Type 2 diabetes (T2DM) is a metabolic disease, which is defined as having high blood glucose levels, in large part related to insulin resistance [Vijan et al., 2010].

Insulin resistance is defined clinically as the inability of a known quantity of exogenous or endogenous insulin to increase glucose uptake and utilization in an individual as much as it does in a normal population [Lebovitz et al., 2001]. Complications of diabetes include blindness, kidney failure, amputations and heart disease. T2DM is caused by a combination of multiple factors including lifestyle and genetic susceptibility. Some of these factors are controllable, such as diet and obesity [Ripsin et al, 2009]. However, other factors are not controllable, including increasing age, female gender, and genetic predisposition [Ripsin et al,m 2009]. Obesity and sedentary life styles are major risk factors for diabetes [Olokoba et al., 2012]. Currently there is no cure for T2DM, however, it can be managed through lifestyle modifications and medication. WHO criteria define diabetes as having a fasting plasma glucose ≥ 7.0 mmol/L (126 mg/dL), having random plasma glucose of ≥11.1 mmol/L (200 mg/dL) or having a glycated haemoglobin (HbA1c) of 6.5% or more on two separate occasions in asymptomatic individuals [WHO, 2006].

Previously, T2DM was rarely seen in adults aged less than 30 years. However, recently it is being increasingly diagnosed among younger adults and even in children [Reinehr et al.,

2013]. Along with early detection and treatment, finding factors associated with diabetes causation is very important, especially at a young age [Wilmot et al., 2014]. In our study we evaluated the association of 25 OHD levels and glucose metabolism in healthy young

21 women, and additionally assessed the effects of vitamin D supplementation and behavioural intervention on 25 OHD levels and its effects on glucose metabolism.

1.1.5 Vitamin D

The prevalence of vitamin D deficiency is high globally and has become a major health concern. It is estimated that over 1 billion people have vitamin D deficiency around the world. More than fifty percent of the population worldwide have vitamin D levels less than

75 nmol/L. Vitamin D deficiency is also common among young children and adolescents, especially during winter [Rathish et al., 2012].

In Australia, vitamin D deficiency is common and it has been reported that almost one third of Australian adults aged 18 to 34 years have some form of vitamin D deficiency (25 OHD

<50 nmol/L) [Australian Bureau of Statistics, 2012]. Around 4.1% of the population have vitamin D levels less than 25 nmol/L (2.6% of men and 5.6% women). This prevalence increases from 5.6% in women aged 25-34 years, to 12.5% in those aged more than 75 years.

Because of the wide latitude range in Australia, the prevalence of vitamin D deficiency is different in different regions [Daly et al., 2012]. One study has shown that 15% of men and

31% of women living in the northern areas of Australia have 25 OHD levels less than 50 nmol/l. This number increased to 35% in men and 58% in women living in southern areas including Victoria [Gill et al., 2014].

Vitamin D deficiency is reported to be also prevalent among adolescents and appears to increase with age from early in life, with adolescents being the age group in the younger population with the highest prevalence of vitamin D deficiency, especially during winter time

[Rockell, 2005]. The Endocrine Society defined vitamin D deficiency in children as 25 OHD

< 50 ng/mL [Holick et al., 2011]. American Academy of Paediatrics and Institute of

Medicine both defined 25 OHD levels <50 nmol/L as insufficiency in children [Misra et al.,

2008]. In a study in the UK it has been shown that the prevalence of vitamin D deficiency is less than 7% among children aged 1.5 to 10 years. However, prevalence of severe vitamin D deficiency (25 OHD levels <25 nmol/L) was 11–16% among 11–18 years old adolescents

[Cashman et al., 2007]. While most healthy children in Australia and New Zealand receive enough sunlight exposure to maintain adequate vitamin D levels, a significant number develop mild vitamin D deficiency during winter [Munns et al., 2006]. In Tasmania, 8% of 8 years old children and 68% of 16 years old children have serum 25 OHD levels less than 50 nmol/L [Jones et al., 2005].

Vitamin D is a fat-soluble vitamin existing in two forms; 1) vitamin D3 or cholecalciferol, which can be found in animal sources and 2) vitamin D2 or ergocalciferol, which can be found in plant sources. Sun light is the best source of vitamin D, but it also can be obtained from some dietary sources, including fatty fish, egg yolk, beef liver, cheese and fortified food, such as dairy products, cereals and orange juice. 7-dehydrocholesterol is absorbed ultra violet (UV) radiation and converted to pre-vitamin D3. Vitamin D is activated in the liver and metabolised to 25 OHD, then through subsequent hydroxylation in the kidneys is converted to 1,25 dihydroxyvitamin D (1,25(OH)2D). 25 OHD is the major circulating form of vitamin

D and represents the vitamin D status in the body. The hormonal activity of 1,25(OH)2D is through binding to the nuclear vitamin D receptor (VDR) (Figure 1.1).

Normal vitamin D status is defined as having 25 OHD levels above 75 nmol/L or 30 ng/ml, whilst vitamin D insufficiency is defined as 25 OHD between 50 to 75 nmol/L (20 to 30 ng/mL) and vitamin D deficiency is defined as having 25 OHD < 50 nmol/L (20 ng/mL)

[Holick et al., 2011]. The most common factors which can contribute to vitamin D deficiency are low sun exposure due to latitude, clothing, season and sunscreen use; having an unhealthy diet; chronic disease and older age [Holick et al., 2007]. People living at higher latitudes,

23 having a BMI of more than 30 kg/m2, being age over 65 years, having naturally dark skin, or wearing cloths which covers most of body are at high risk of vitamin D deficiency [Mayor et la., 2016].

There is an ongoing controversy about what constitutes a sufficient vitamin D level. There are many different guidelines on optimal 25 OHD levels. The Institute of Medicine recommends 50 nmol/L as adequate on the basis of bone health studies [Ross et al., 2011], which is in contrast to the 75 nmol/L recommended by the US Endocrine Society [Holick et al., 2011]. It has been suggested that there might be no firm cut-off should be advocated, but rather age specific or disease specific vitamin D levels [Spedding et al., 2013]. Currently, the most convincing evidence for vitamin D cut-offs exists for fracture risk reduction [Bischoff-

Ferrari et al., 2012]. A meta-analysis reported that optimal serum 25 OHD concentrations, between 60 nmol/L and 95 nmol/L, are required to reduce the risk of falls by 19% [Bischoff-

Ferrari et al., 2009]. However, more studies are needed before disease-specific levels can be recommended. In our study we used the vitamin D deficiency definition which is provided by

Australian Bureau of Statistics [Ginde et al., 2010].

Vitamin D has a physiological role in maintaining minerals including calcium and phosphate, which are important for bone metabolism. Therefore, vitamin D is associated with bone health and it has been demonstrated that low vitamin D levels can lead to skeletal disease including osteoporosis, rickets, osteomalacia and fracture. Vitamin D receptors have been found in many different cells, which suggests a wide range of biological functions for vitamin

D in the body [Kongsbak et al., 2013]. Therefore, during the last few years, many hypotheses have arisen about the association of vitamin D levels with many chronic conditions including

CVD, diabetes, obesity, malignancies and even autoimmune disease. Recently, these associations have been examined by many researchers, but the results are generally inconsistent and inconclusive for causality.

UV radiation Food or supplement

Skin Intestine

Vitamin D3 or D2

Adipose tissue Liver Hydroxylase

25 OHD3 or 25 OHD2

Kidney Hydroxylase

1,25(OH)2D3 or 1,25(OH)2D2

Blood

Target cells

Figure 1.1 Vitamin D metabolism

1.1.6 Vitamin D, calcium and parathyroid hormone

Vitamin D and parathyroid hormone (PTH) are major regulators of mineral metabolism, especially maintenance of calcium and phosphorus in the body. PTH and Vitamin D form a tightly controlled feedback cycle, PTH being a major stimulator of vitamin D metabolism in the kidney, while vitamin D exerts negative feedback on PTH secretion. When circulating calcium levels are low, secretion of PTH increases. PTH increases uptake of calcium from bones and increases production of 1, 25 (OH)2D in the kidney. Serum 1, 25(OH)2D increases absorption of calcium from the intestine and helps the body to maintain the right calcium balance. Following vitamin D deficiency, absorption of calcium from the intestine decreases and secretion of PTH increases. Hyperparathyroidism and vitamin D deficiency have been implicated in a variety of cardiovascular disorders including hypertension, atherosclerosis and vascular calcification. Both hormones have direct effects on the endothelium, heart, and other vascular structures [Khundmiri et al., 2016].

1.1.7 Mobile health technologies and health outcomes

Mobile health (m-health) is a term for the use of mobile technologies, including mobile phones, smart phones, personal digital assistants and tablets in public health. During the last decade, m-health attracted much research attention, such as use of text messaging, telephone, photos or videos and software applications to provide health support and deliver interventions. It is reported that in developed countries, each person owns 1.1 mobile devices and the number is growing rapidly [Boulos et al., 2011]. The advantages of using m-health include being cost effective; it is easy to use, especially for large population studies; and generalizable to the whole population, as many people use mobile technologies these days.

Delivery of the intervention is also easy, as people carry their phone wherever they go, so checking compliance and being in touch with participants in studies is relatively-

26 straightforward. It also assists participants by being able to access extra assistance, if required. Most previous studies have evaluated the use of m-health for one type of health behavioural change, such as increasing physical activity, following a healthy diet or quitting smoking; or self-management of chronic disease by using text messages or telephone communication.

A study in 2012 [Tay et al., 2017] evaluated the acceptability and usability of a dietary app

(Calci-app) to self-monitor calcium consumption in young women aged 18-25 years and demonstrated acceptable use of Calci-app to self-monitor calcium consumption. The investigators showed relatively high initial compliance with dietary intake recording; however, adherence beyond 3 days was low. Another study in this area [Turner-McGrievy et al., 2013] compared the traditional versus mobile app self-monitoring of physical activity and dietary intake among 96 overweight adults. They demonstrated that over 6 months, app users exercised more frequently and reported greater intentional physical activity than a control group. The frequency of self-monitoring did not differ according to the diet self-monitoring method; however, app users consumed less energy than the control group. Smartphone software programs or applications have stimulated significant attention, as they are cheap and convenient and can provide individualised information and support.

The main source of vitamin D generally is sun light exposure, so it is recommended that adequate sun exposure is acquired to achieve optimal vitamin D levels [Nair et al., 2012]. On the other hand, sun exposure is the major cause of skin cancer [Kennedy et al., 2003]. It is necessary to maintain a balance between the risk of skin cancer from excess sun exposure, with achieving adequate vitamin D levels, especially in Australia where the prevalence of both skin cancer and vitamin D deficiency are high [Lai et al., 2018 and Shrapnel et al.,

2006]. It is hard to have a standard recommendation for the time required to be exposed to sunlight for the population as whole, as it is different for each individual according to their

27 skin type and skin colour [Ginde et al., 2009]. Sun exposure recommendations could even be different for each individual at different times, depending on what they are wearing, how much sun screen they used and the UV index. Therefore, it will be very useful to develop scientifically validated mobile-based applications which can provide individual advice on how much sun is sufficient for each person depending on the UV index, location, season, weather, skin type, skin colour and amount of sun screen used.

1.2 Mechanisms of association of vitamin D and CVD risk factors

1.2.1 Mechanisms of association of vitamin D and lipid profiles

Mechanisms suggested for the association between vitamin D and body weight, body composition and insulin resistance are presented later. The following section discusses possible mechanisms for an association of vitamin D with lipid profiles and blood pressure, as well as a direct effect of vitamin D on atherosclerosis.

There are a number of suggested mechanisms regarding the favourable association of vitamin

D and lipid profiles: 1) vitamin D reduces intestinal absorption and synthesis of lipids by increasing binding of calcium to fatty acids in the intestine; by decreasing the absorption of free fatty acids, serum cholesterol and low density lipoprotein (LDL) levels decrease [Reddy

Vanga al., 2010]; 2) vitamin D can enhance lipolysis by regulating the use of lipolytic substrates for energy metabolism [Shi et al., 2001]; 3) vitamin D also improves insulin production and action [Sung et al., 2012]; insulin plays an important role in lipid metabolism as it reduces lipolysis in adipose tissue and hence decreases fatty acids in plasma, increases formation of very low density lipoprotein (VLDL) in the liver, increases uptake of triglyceride in adipose tissue and increases synthesis of cholesterol in the liver [Sung et al.,

2012]; 4) vitamin D also increases gene expression of lipoprotein lipase in muscles and

28 adipose tissue. Lipoprotein lipase increases the clearance of circulating lipoprotein particles which reduces serum triglyceride and increases serum HDL [Vu et al., 1996].

1.2.2 Mechanisms of association of vitamin D and blood pressure

The potential effects of vitamin D on blood pressure are suggested to be mediated through the renin angiotensin aldosterone system (RAAS) [Ajabshir et al., 2014]. The renin angiotensin system plays an essential role in controlling blood pressure, serum and cell electrolytes, blood volumes and vascular resistance. Angiotensin is a vaso-active peptide which increases arterial blood pressure [Hall et al., 1991]. Vitamin D deficiency is suggested to be associated with increased RAAS activity. It has been shown that inhibition of 1,25(OH)2D synthesis led to an increase in renin gene expression, and 1,25(OH)2D treatment suppressed renin expression

[Hall et al., 1991].

1.2.3 Mechanisms of vitamin D and potential direct effects on atherosclerosis

It has been suggested that vitamin D, as well as affecting cardiovascular disease risk factors, can affect vascular health directly. The mechanisms suggested behind this effect include:

1) vitamin D receptors are found in almost all cell types, including vascular smooth muscle and in cardio myocytes. It is suggested that endothelial cells express vitamin D receptors and contain hydroxylase enzymes, which can affect production of 1,25(OH)2D. Thus, vitamin D deficiency can affect endothelial function [Ni et al., 2014].

2) Inflammatory processes in the endothelium are the main cause of atherosclerosis

[Castellon et al., 2016]. It is suggested that vitamin D, through genomic and non-genomic mechanisms, can reduce inflammation in the endothelium. First, vitamin D protects endothelial cells from H2O2 oxidative stress, by controlling the phospho-active extracellular signal-regulated kinases levels in the body [Uberti et al., 2013]. Second, vitamin D stimulates

29 nitric oxide production in endothelial cells, which affects both intracellular and extracellular kinases [Molinari et al., 2011]. Third, vitamin D suppresses expression of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase [Levy et al., 1991], which plays a major role in increasing inflammation in endothelial cells. Finally, calcitriol has a vascular protective effect by decreasing the concentrations of glucose and decreasing glycation products in endothelial cells [Talmor at al., 2008]. Glycation products can induce the expression of pro-inflammatory biomarkers in endothelial cells [Talmor at al., 2008].

Increased glucose and pro-inflammatory process would also increase the risk of vascular damage.

3) Vitamin D also appears to affect vascular tone via regulation of endothelium-derived contracting factors, vasoconstrictor metabolites of arachidonic acid [Bellien et al., 2008].

4) Vitamin D can reduce calcification of endothelial cells by reducing PTH levels

[Khundmiri et al., 2016].

1.2.4 Mechanisms of association between vitamin D and obesity

The mechanisms that link 25 OHD levels to body weight are not fully understood. However, there are several possible mechanisms suggested by other studies.

1) Obesity can increase expression of vitamin D receptor genes in adipose tissue, which leads to an increase in conversion of 1,25(OH)2D and a decrease in serum 25 OHD levels [Cipriani et al., 2014]. Hence, adipose tissue in obese subjects has a different response to 1,25(OH)2D in people with normal weight, and thus can lead to vitamin D deficiency [Mazahery et al.,

2015].

2) One of the main roles of vitamin D in the body is increasing absorption of calcium in the intestine. In people with vitamin D deficiency, calcium cannot be absorbed as effectively

30 from the intestine, leading to increased PTH levels in the blood. PTH increases calcium resorption from bone and also increases lipogenesis in adipocytes. [Valiña-Tóth et al., 2010].

3) Vitamin D deficiency decreases in vitro leptin secretion, which leads to obesity [Vilarrasa et al., 2010]. Adipocytes, as a part of the endocrine system, secrete several proteins, known as adipokines, into the blood. Leptin is an adipokine, which is involved in processes such as food intake, insulin-related functions, energy balance and glucose and lipid metabolism.

Leptin plays a role in the regulation of energy balance in the long term. It suppresses food intake and thus induces weight reduction [Skowrońska et al., 2005]. However, most people with obesity show high levels of circulating leptin, which is attributed to leptin resistance.

Leptin resistance occurs when leptin receptors in the targeted cells cannot respond to leptin.

This can cause an increase in leptin production and high levels of circulating leptin in the blood [Zhou et al., 2013]. Multiple factors, including inflammation, contribute to leptin resistance [Pan et al., 2018].

4) Outdoor activity and sun light exposure are limited in obese people as they tend to be inside more of the time and cover-up more when they are outside [Florez et al., 2007].

5) The availability of vitamin D is reduced in obesity due to sequestration of vitamin D in adipose tissue [Mutt et al., 2014].

1.2.5 Mechanisms of association of vitamin D and body composition and fat distribution

Previous studies have reported possible mechanisms relating to the association of vitamin D with adipose tissue and muscle cell metabolism [Hivaprakash et al., 2014]. However, it is still unclear whether 25 OHD levels are associated with fat mass stored in specific areas of the body. 1) It is suggested that vitamin D can be catabolised in adipose tissue, which is a source of hydroxylase enzymes. Therefore, increasing adipose tissue can increase catabolism of

31 vitamin D, leading to a decrease in 25 OHD levels [Chang et al., 2017]. 2) Another suggested mechanism is that 1,25(OH)2D binds to vitamin D receptors in muscle and activates muscle growth and improves muscle function, and may influence overall body composition [Ceglia et al., 2009]. 3) Obesity increases inflammation in the body and it is suggested that vitamin D can have anti-inflammatory effects [Calton, et al., 2015]. 4) It has been suggested that muscles cells store 25 OHD and reduce the proportion of the 25 OHD in the blood, by binding 25 OHD to binding protein in the muscle cells. This retention by muscle cells protects the 25 OHD from degradation by the liver [Abboud et al., 2014]. Therefore, with the increase in body fat mass, the 25 OHD level decreases, leading to a decrease in the storage of

25 OHD in the muscle cells. This decrease in the storage of 25 OHD could in turn decrease serum 25 OHD levels that might contribute to unfavourable body composition [Vitezova et al., 2017].

1.2.6 Mechanisms of association of vitamin D and diabetes

Although the underlying mechanism of an association of vitamin D with diabetes is still unclear, there are several plausible suggested mechanisms. 1) Direct effects of vitamin D on beta cell function, increasing gene expression of insulin [Pittas, A.G., Dawson-Hughes, B.,

2010]. 2) Vitamin D increasing insulin sensitivity by enhancing the expression of insulin receptors in skeletal muscle and adipose tissue [Sung et al., 2012]. 3) Activating peroxisome proliferator-activated receptor (PPAR-δ), which affects regulation of fatty acid metabolism in body. PPARs are nuclear receptor proteins which play a role in regulating the expression of different genes in the body [Tyagi et al., 2011].

As presented in this section, cardiovascular diseases are one of the major health concerns around the world. Recently, lower serum 25 OHD levels have been reported to be associated with an increase in CVD risk factors including diabetes and obesity; however, still the results

32 of studies are inconsistent and there is an evidence gap about the best intervention to improve vitamin D levels.

1.3 Literature review

It is becoming clear that vitamin D, other than its classical effects on bone health, has a range of actions in the body including muscle function, immune responses, cell function and cardiovascular homeostasis [Zittermann et al., 2010]. In recent years, increasing attention has been paid to the possible association of vitamin D deficiency and CVD risk factors or the potential effects of vitamin D to improve CVD risk factors. Literature in this area is reviewed and presented in the following section.

First, literature on the associations of vitamin D and obesity is presented followed by clinical trials on the effects of vitamin D intervention on obesity. Almost all of the observational studies in this area showed a negative association between 25 OHD levels and obesity, but the results from the clinical trials are inconsistent.

Other studies, instead of evaluating the association of serum 25 OHD levels and obesity, evaluated the association of vitamin D intake alone or in combination with calcium intake with obesity. These studies are presented next, followed by the clinical trials on the effects of vitamin D supplementation alone or along with calcium on obesity. These studies documented insufficient vitamin D intake among children, and most showed an association between lower vitamin D intake and greater body weight. However, clinical trials could not demonstrate any beneficial effects of vitamin D or calcium intervention on body weight.

Then studies on evaluating the effects of weight loss on vitamin D levels are reviewed.

In the following section, observational studies and clinical trials on vitamin D and body composition/fat distribution are summarised. There are very few studies in this area, and

33 these show inconsistent results. This is probably due to different methods of generating body composition measurements, and different ethnic groups which can affect body composition.

Although it is suggested that the vitamin D concentration is associated with CVD through its effect on CVD risk factors, in recent years many studies have evaluated the direct association of vitamin D levels with cardiovascular health by evaluating cardiac function and endothelial function. Therefore, we reviewed studies which evaluated the direct effects of vitamin D on vascular health, endothelial function and vascular smooth muscle cells. Results from studies in this area are inconsistent, with some studies showing that endothelial function was significantly higher in participants who received vitamin D supplementation. However, other studies could not show any beneficial effects of vitamin D supplementation on endothelial function. More studies in this area, with larger sample size and longer follow-up duration, are needed.

One of the main pathways in the aetiology of CVD is chronic inflammation. There are several studies which evaluated the association of vitamin D and inflammatory biomarkers. Most of the observational studies showed a negative association between vitamin D and inflammation and most of trials showed beneficial effects of vitamin D improvements on inflammation.

However, studies measured different inflammatory biomarkers as inflammation indicators.

In section 1.3.5 studies on vitamin D and lipid profiles are presented. It is reported that people with sufficient vitamin D levels have a more favourable lipid profile, and an increase in 25 OHD levels could have beneficial effects on lipid profiles. There are several studies in this area, but with inconclusive data. Most of studies in this area were designed for other outcomes, not specifically for lipid profiles.

In the following sections associations of vitamin D and blood pressure and associations of vitamin D and diabetes mellitus are discussed. Most of the studies assessing the associations

34 between vitamin D status and blood pressure are cross-sectional in design and provide consistent results. However, results from clinical trials are not convincing regarding the beneficial effects of vitamin D supplementation on blood pressure.

In the last section of the literature review, we present studies which evaluated the effects of behavioural intervention by using mobile health (m-health) on health outcomes. Most of studies in this area are qualitative. There are very few which evaluated the effects of m-health on different lifestyle factors including smoking cessation or increasing physical activity.

More study in this area is required.

1.3.1 Vitamin D and obesity

Association of vitamin D levels and obesity

Over the years, many observational studies have evaluated the association of low vitamin D levels with obesity. Almost all of the observational studies have shown the association between low 25 OHD levels and greater body weight or BMI, but data clarifying the mechanism of this association remain inconclusive. One cross-sectional study of 89 school children has shown that for one unit increase in 25 OHD levels, BMI and body fat mass decreased by 10 and 12 percent, respectively [Wakayo et al., 2016]. Another study on 386

Korean adolescents showed that participants with BMI less than 18.5 kg/m2 had significantly lower 25 OHD concentrations, compared to those with BMI 18.5 to 23 kg/m2 (14.4±4.1 vs.

15.5±4.9 ng/mL; p = 0.04) [Kim et al., 2015]. Zhang et al. in 2014 recruited 1488 children aged 7 to 11 years old and showed that vitamin D deficient subjects had significantly higher

BMI (18.4±2.2 versus 16.8±1.7 kg/m2; p < 0.001), body weight (34.1±3.8 vs. 31.5±3.3 kg; p

< 0.001) and body fat percentage 20.2%±2.6% versus 19.1%±2.4%; p < 0.001) [Zhang et al.,

2014]. Moreover, a case-control study of 452 Caucasian children has indicated that low 25

OHD levels are inversely associated with obesity (Odds Ratio (OR): 4.39, 95% CI: 2.19-8.83;

35 p < 0.001) [Pacifico et al., 2011]. Martins et al. found that among more than 15,000

American adults the prevalence of obesity was significantly higher in the lowest quartile compared to the highest quartile of 25 OHD levels after adjustment for age and gender (Odds

Ratio (OR), 2.29, p < 0.001) [Martins et al., 2007]. In the meta-analysis by Parker in 2010 of

28 observational studies including 19 case-control, 3 nested case-control and 6 cohort studies, the overall 25 OHD levels were associated inversely with cardio-metabolic disorders (OR:

0.57, p < 0.001) [Parker et al., 2010]. Another systematic review and meta-analysis on the association of vitamin D and obesity included 15 studies with a total of 3867 subjects with obesity and 9342 normal weight subjects. This study reported a positive association between vitamin D deficiency and obesity (OR: 3.43, p < 0.001) [Yao et al., 2015]. These inconsistent results could be due to differences in participants age range, as some studies were on young population and some on elderly. Further, different methods were used to measure 25 OHD levels. While some studies used ELISA (enzyme-linked immunosorbent assay), others used radio immunoassay and some HPLC (High-performance liquid chromatography) methods.

Moreover the definition of vitamin D deficiency was different among studies; some studies defined vitamin D deficiency as 25 OHD levels less than 50 nmol/L and others used 25 OHD levels less than 37.5 nmo/L.

In 1985, Bell et al. compared vitamin D levels in obese subjects with normal weight subjects and reported that the alteration of vitamin D metabolism in obese people was characterized by secondary hyperparathyroidism. Hyperparathyroidism is associated with increased reabsorption of calcium in the kidney and increased circulating 1,25 (OH)2D in blood [Bell et al., 1985]. One longitudinal study in Amsterdam on 453 participants more than 65 years old, indicated that greater BMI and total body fat are associated with lower 25 OHD levels (p <

0.05) and higher parathyroid hormones (p < 0.05) [Snijder et al., 2005].

Effects of vitamin D intervention on obesity

Although most observational studies confirmed the association of low 25 OHD levels with body weight and BMI, results from the clinical trials are inconsistent although most were not specifically designed to evaluate the effects of vitamin D supplementation on body weight or

BMI.

In a double-blind clinical study on 77 women, 25 mcg per day vitamin D supplementation for

12 weeks resulted in a significant reduction in body fat mass (-2.7±2.1 kg vs. -0.47±2.1 kg; p

< 0.001), but no significant effect on body weight or waist circumference compared to placebo [Salehpour et al., 2012]. Another clinical trial on 52 participants with obesity

(BMI>30 kg/m2) and low vitamin D levels (25 OHD <50 nmol/L) showed that 7000 IU per day vitamin D supplementation for 26 weeks had no significant effects on loss of body fat

[Wamberg et al., 2013]. Moreover, Zittermann et al. in a double blind clinical trial investigated the effects of 83 mcg per day vitamin D supplementation on weight loss and found that weight loss was not changed by vitamin D supplementation [Zittermann et al.,

2009]. Another study with 20,000 IU twice a week or 20,000 IU once a week vitamin D supplementation in 445 overweight or obese subjects did not find that weight reduction occurred with vitamin D supplementation [Sneve et al., 2008]. However, in a randomised clinical trial by Caan et al. it was found that 1000 mg calcium plus 400 IU vitamin D supplementations in women aged 50 to 79 years had a consistently positive effect on weight loss, compared to those receiving placebo (mean differences -0.13 kg, p = 0.001) [Caan et al.,

2007]. A meta-analysis of 26 clinical studies reported that there was no overall evidence for significant effects of vitamin D supplementation on body weight, BMI or fat mass.

Furthermore, no evidence of a dose-response trend was observed [Chandler et al., 2015].

These inconsistent results could be due to age, sex, ethnicity differences, different doses of

37 vitamin D supplementation or different follow-up periods. Also, some of the studies were conducted on overweight people while others were on normal weight people.

Association of vitamin D intake and obesity

The main source of serum 25 OHD is from exposure of the skin to sunlight. However, seasonal changes in sunlight exposure, living at high latitudes, or having dark skin, can impair UV absorption. Therefore, it is important to meet dietary vitamin D recommendations to maintain adequate serum 25 OHD concentrations. It is well known that dietary intake and quality of diet are associated with many chronic diseases and conditions, including obesity.

The Institute of Medicine Dietary Reference Intake committee in 2010 set the recommended calcium intake as 1300 mg per day and 600 IU per day vitamin D intake for children aged 9 to 18 and 1200 mg/day calcium and 600 IU per day vitamin D intake for adults [Ross et al.,

2011]. According to Australian Nutrient Reference Values (ANV), a 5 mcg/day vitamin D intake is required for children or adults. The calcium intake recommendation according to the

ANV is 800 mg/day for children aged 9 to 11 years old, 1050 mg/day for children aged 12 to

18 and 840 mg/day for adults aged 19 to 50 years [Nutrient Reference Values, 2018].

Insufficient dietary intake of vitamin D in children was confirmed by various studies [Dyląg et al., 2014]. Moreover, it has been shown that obese or overweight children have a higher risk of vitamin D deficiency [Bellone et al., 2014]. Keast et al. investigated the associations of dairy consumption with energy, calcium, and vitamin D intakes, and the association of dairy intake with overweight or obesity in children aged 8 to 18 years in the (National Health and Nutrition Examination Survey) (NHANES, 2005–2008). They used two 24-hours recall dietary questionnaires to evaluate micro and macro nutrient intake, and have shown that yogurt and dairy intake was associated with higher intakes of calcium and vitamin D. Also, yogurt intake was associated with lower total body fat [Keast et al., 2015]. Au et al. estimated

38 vitamin D intake from foods and supplements in 3,310 children or adolescents who were examined as part of the 2005-2006 NHANES study. They reported that nearly 75% of US children failed to meet current vitamin D recommendations and overweight or obese children were five times more likely to be at risk for inadequate 25 OHD levels [Au et al., 2013]. In a study on Spanish schoolchildren aged between 7 and 11 years, it was shown that vitamin D intake was lower than the recommended range in almost all children and vitamin D intake was lower in girls, those younger than 7 years and obese children [Anta et al., 2012]. In a study by Wulaningsih et al. 182 nutrition and lifestyles factors were investigated in relation to central obesity in more than 15,000 participants and showed significant associations between serum 25 OHD and abdominal obesity in both men and women [Wulaningsih et al., 2017].

According to studies in this area, vitamin D intake and calcium intake are lower than the recommended range among children. They also have shown that lower intake of vitamin D and calcium could be associated with higher fat mass [Anta et al., 2012; Wulaningsih et al.,

2017].

Association of vitamin D along with calcium in obesity

It is well known that there is an interaction between vitamin D and calcium in the body [Lips et al., 2012]. The reasons for this interaction include 1) Vitamin D status and calcium intake have synergistic effects on calcium absorption. The effect of vitamin D is to increase absorption of calcium in the intestine by mediating transportation of calcium across the intestinal mucosa through both genomic and nongenomic mechanisms [Heaney et al., 2008].

2) Calcium intake can affect serum vitamin D levels by increasing the half-life of serum 25

OHD, which causes a reduction in 1,25(OH)2D synthesis [Mazahery et al., 2015]. Many observational studies evaluated the association of vitamin D alone or in combination with calcium on obesity, but the results are inconsistent [Palacios et al., 2007; Coldberg et al.,

2009]. Palacios et al., in a cross-sectional study on 100 adolescents showed a negative association between total calcium intake and BMI [Palacios et al., 2007]. Several studies in this area evaluated the association of dairy product intake as a source of vitamin D and calcium with obesity. In a prospective cohort study of 18438 women aged more than 45 years old, it was shown that over 11 years of follow-up, dairy intake, assessed by a 131-item food frequency questionnaire, could prevent weight gain, but calcium or vitamin D intake was not associated with being obese or overweight [Rautiainen et al., 2016]. Carruth et al. investigated food intake of 53 preschool subjects and the relation of food intake to body composition. They reported that higher intake of dairy products which contained calcium and vitamin D, was associated with lower total body fat [Carruth et al., 2001]. Azadbakht et al., investigating 827 subject aged 18 to 74, showed that dairy intake was inversely associated with the prevalence of metabolic syndrome and having a high waist circumference; defined as more than 102 cm in men and more than 88 cm in women (p < 0.001) [Azadbakht et al.,

2005]. One systematic review evaluated the association of dairy intake and body weight in nineteen cohort studies and concluded that dairy food intake can contribute to weight loss or prevention of weight gain [Dougkas et al., 2011].

Effects of vitamin D supplementation along with calcium on obesity

The few clinical trials in this area can be categorised into three groups. 1) Trials comparing the effects of vitamin D supplementation with placebo, 2) trials comparing the effects of vitamin D plus calcium, with calcium supplementation as the control group and 3) trials comparing the effects of vitamin D supplementation plus calcium with placebo. However, results of clinical trials in this area are inconsistent. A population-based, double-blind, clinical trial in 2010 by Zhou et al., determined the effects of calcium and vitamin D supplementation on osteoporotic fractures in 1179 postmenopausal women. They randomised

40 participants into three groups. Group 1 received 1500 mg per day calcium supplement; group

2 received 1500 mg/day calcium plus 1100 IU per day vitamin D supplement; and group 3 received placebo. They showed that over 4 years of intervention calcium alone, without vitamin D supplementation, had a favourable effect on obesity and body composition [Zhou et al., 2010]. A clinical trial on 36282 women aged 50 to 79 years old, compared the effects of 1000 mg per day calcium supplementation plus 400 IU per day vitamin D3 supplements with a placebo group over 7 years of intervention and found that calcium plus vitamin D supplementation could prevent or limit weight gain in postmenopausal women [Caan et al.,

2007]. A systematic review and meta-analysis on 26 randomised clinical trials with a total of

42430 participants evaluated the effect of vitamin D supplementation alone or in addition to calcium on adiposity measures. This meta-analysis showed that vitamin D supplementation alone or in combination with calcium supplementation had no beneficial effects on body weight or loss of body fat mass [Chandler et al., 2015].

Effects of weight loss on vitamin D levels

To examine the casual relationship between vitamin D and obesity, some investigators evaluated the effects of weight loss on vitamin D levels and showed that reduction in body fat and body weight achieved by behavioural strategies can improve vitamin D levels. A prospective cohort study in 2012 examined the effects of weight loss on 25 OHD levels in

383 overweight or obese women after participating in a two year weight loss program. They reported that women who lost more than 10% of baseline weight after 2 years, had a significant increase of 2.6 ng/ml in 25 OHD, compared to women who did not lose weight (p

= 0.010 [Rock et al., 2012]. A clinical trial by Mason et al. investigated the effects of one year of weight loss intervention through diet modification or exercise on serum 25 OHD levels. They showed that in 439 postmenopausal women, there was an association between

41 weight loss and increasing vitamin D levels. Women who lost less than 5 percent, 5 to 10 percent, 10 to 15 percent or more than 15 percent of body weight had 2.1, 2.7, 3.3, 7.7 ng/mL of increase in 25 OHD levels respectively (p trend=0.002) [Mason et al., 2011]. Another study on 44 obese postmenopausal women, indicated that 10% weight loss following a 20 week low calorie diet program resulted in a significant increase in 25 OHD levels (p < 0.050)

[Tzotzas et al., 2010].

1.3.2 Vitamin D and body composition/fat distribution

Association of vitamin D and body composition and fat distribution

Several studies have investigated the association between vitamin D and obesity. However, obesity (BMI more than 30 kg/m2) cannot reflect body content, which is affected by age, sex and ethnicity. Very few studies have evaluated the association of vitamin D with body composition, including body fat mass and lean mass and distribution of body fat. Thus, it is still not clear if vitamin D status is associated with fat mass in specific areas in the body and the direction of any association. Defining the association of vitamin D with body composition and fat distribution is important, as it may help to clarify the mechanism of the association of vitamin D and obesity or other chronic diseases. A cross-sectional study on 1697 Korean adults evaluated the association of total fat mass, measured using DXA, BMI and waist circumference with 25 OHD levels. The highest quartile of body fat was associated with higher odds ratio for vitamin D deficiency, irrespective of fat location (abdominal or peripheral). BMI and waist circumferences were not associated with vitamin D levels [Han et al., 2014]. In a population-based study by Vitezova et al., vitamin D deficiency was associated with higher fat percentage, but not with lean mass. Also, an association was found between vitamin D and android fat (fat mainly around the trunk and upper body areas) and android to gynoid fat (fat mainly around the lower body areas) ratio, but these associations

42 became non-significant after adjustment for BMI. So they concluded that these associations are mainly explained by BMI [Vitezova et al., 2017]. However, it seems that this null association after adjustment for BMI could be due to over adjustment in the above mentioned study. In a cross-sectional study on 90 young women (aged 16 to 22 years) in 2009, a strong negative association was reported between 25 OHD levels and visceral fat, measured by computed tomography, and total fat mass, obtained from DXA, (p = 0.009 and p = 0.02, respectively) [Kremer et al., 2009]. Similarly, a longitudinal study on 453 participants aged over 65 years, showed that after adjustment for potential confounders including age, season, smoking, sex and PTH levels, 25 OHD deficiency was strongly associated with higher total body fat mass and weakly associated with BMI and waist circumference (p < 0.05) [Snijder et al., 2005]. Another cross-sectional study in this area, conducted on 1038 subjects (40.3% females), aged 18 to 24 years in 2015, indicated that vitamin D deficiency (25 OHD <20 ng/ml) was associated with higher median body fat (B: -0.06; 95% CI: -0.11 to -0.001; p <

0.05) [Baker et al., 2015]. The association of vitamin D deficiency and body composition may be different according to the different measurement methods of body fat and distribution of body fat, gender and age differences.

Effects of vitamin D supplementation on body composition and fat distribution

Although most of the cross-sectional studies to date report an association of 25 OHD levels with total fat mass and visceral fat mass, results from clinical trials are inconsistent. These inconsistent results could be due to the different dosage of vitamin D supplementation, duration of follow-up, subjects with different underlying health conditions, different age ranges and different methods to measure body composition. Salehpour et al. in a double blind clinical trial randomised 77 women who were overweight or obese into two groups, to receive 25 mcg per day vitamin D supplement or placebo for 12 weeks. They found that fat

43 mass significantly decreased in the supplement group compared to the control group (-

2.7±2.1 kg, -0.47±2.1 kg; p < 0.001). No significant differences were reported for body weight and waist circumferences between the two groups [Salehpour et al., 2012]. Rorenblum et al. investigated the effects of calcium and vitamin D supplemented orange juice on body weight and visceral adipose tissue in 171 overweight or obese adults over a 16-week period.

They reported that the average weight loss was 2.5 kg in both supplement and control groups with no significant differences between two groups. However, in the supplement group the reduction of visceral fat tissue was significantly greater than the control group (-12.7 ± 25.0 cm (2) vs -1.3 ± 13.6 cm (2); p = 0.039) [Rosenblum et al., 2012]. Dong et al. investigated the effects of 400 IU per day or 2000 IU per day vitamin D supplementation on total body fat, measured by DXA, in 49 black adolescents aged 16.3 ± 1.4 years. They showed a significant inverse association between vitamin D, total fat mass at after 4, 8 and 16 week follow-up in the response to 2000 IU per day vitamin D supplementation (r = −0.63; p = 0.001; r = −0.52; p = 0.010 and r = −0.46; p = 0.03, respectively) [Dong et al., 2010]. A randomised clinical trial compared the effects of low energy diet (-2900 kJ/d) combined with 5 mcg vitamin D plus 600 mg per day calcium supplementation with a low energy diet and placebo. Sixty- three overweight or obese women [Tay et al., 2017] with mean age of 43 years old participated in this study. A significant decrease in body weight and fat mass was observed in the group which received calcium and vitamin D supplementation, compared to the placebo group (p < 0.01) [Major et al., 2009].

1.3.3 Vitamin D and vascular health

Vitamin and endothelial function

Although it is suggested that vitamin D concentration is associated with CVD through its effect on CVD risk factors, during recent years many studies evaluated the direct association

44 of vitamin D levels with cardiovascular health by evaluating cardiac function and endothelial function. Witham et al. in 2012 showed that endothelial function was significantly higher in participants who received 100,000 IU vitamin D2 supplementation for 8 weeks (6.9% vs

3.7%, p-value = 0.007), but this was not different after 16 weeks’ intervention [Witham et al.,

2012]. Similarly, in a randomised clinical trial by Harris et al. in 57 adults, a significant improvement in flow-mediated vascular dilation following 60,000 IU per month vitamin D3 supplementation was reported (p = 0.04) [Harris et al., 2011]. However, in another study in

62 patients with vitamin D deficiency (25 OHD < 30 ng/ml), supplementation with 100,000

IU vitamin D3 for one month had no effect on endothelial function [Stricker et al., 2012].

Kharlamov et al. in 2012 examined the effects of 1200 to 1800 units of vitamin D supplementation for 6 months on 120 vitamin D deficient subjects and found that cardiac function (measured by using fluorescence activated cell scanning) improved following replenishing vitamin D levels [Kharlamov et al., 2012]. Wang et al. in 2008 in 1739 participants showed an increased risk of CVD across categories of vitamin D levels (p =

0.01) [Wang et al., 2008]. Another cross-sectional study on 5559 Korean adults revealed that participants in the lowest category of vitamin D levels (25 OHD <25 nmol/L) had a higher prevalence of CVD [Park et al., 2012].

Vitamin D and vascular smooth muscle cells

Many experimental studies have indicated that 25 OHD levels are associated with atherosclerosis by affecting inflammation, formation of foam cells, dysfunction of endothelial cells and proliferation of vascular smooth muscle cells. Vitamin D can increase influx of calcium into the cells and increase calcification of endothelium smooth muscle cells.

Wakasugi et al. indicated that synthesis of prostaglandin I2, which plays an important role in reducing thrombogenicity, increased significantly in the presence of vitamin D. They

45 concluded that vitamin D is a vasoactive agent which plays a protective role in the development of atherosclerosis [Wakasugi et al., 1991]. Chen et al. in 2010 showed that vitamin D affects vascular smooth muscle cell proliferation, through inhibiting cycling dependent kinase 2 [Chen et al., 2010]. Xiang et al. in an experimental study evaluated the association of vitamin D and endothelial cell and vascular smooth muscle cells in aortas in mice, and concluded that treatment with 1,25(OH)2D3 could have a protective effect against the development of atherosclerosis [Xiang et al., 2017].

1.3.4 Vitamin D and inflammation

One of the main biological pathways implicated in the aetiology of CVD is chronic inflammation. Chronic inflammation is evaluated by measuring inflammatory biomarkers such as interleukin 6 (IL-6), TNF-alpha (tumour necrosis factor), and high-sensitivity C- reactive protein (hs-CRP). Several recent studies have demonstrated the role of vitamin D in the reduction of inflammation, through affecting the gene expression of inflammatory biomarkers. Most of the previous observational studies have shown that vitamin D deficiency is associated with inflammatory biomarkers in obese subjects, in people with T2DM, metabolic syndrome and CVD. For example, one pilot cross-sectional study on 50 obese subjects showed that hs-CRP was significantly negatively associated with serum 25 OHD levels (B=-0.43, p < 0.001) [Ilinčić et al., 2017]. However, results from clinical trials are inconsistent. Dosage and duration of vitamin D supplementation, age range and condition of participants were different across the studies. Zittermann et al. investigated the effects of

3332 IU vitamin D supplementations per day for 52 weeks on IL-6 and TNF-α levels in 165 overweight or obese subjects and found that TNF-α significantly decreased in the intervention group compared to placebo (7.84±3.15 pg/mL, 7.04±2.25 pg/mL, p = 0.049), but no significant differences were observed in IL-6 levels between the two groups [Zittermann

46 et al., 2009]. A clinical trial study on 413 healthy subjects found that 200, 400 and 600 IU per day vitamin D supplementation for 22 weeks had no significant effect on inflammatory biomarkers including hs-CRP, IL-6, IL-5, IL10 and TNF-α [Barnes et al., 2011]. Asemi et al. investigated the effects of vitamin D supplementation on hs-CRP level on 54 healthy pregnant women and showed significant improvement in hs-CRP levels after receiving 400

IU per day vitamin D3 supplement for 9 weeks compared to the placebo group (4.53±0.64 mg/mL, 3.12±0.46 mg/mL, (p = 0.01) [Asemi et al., 2013]. Another systematic review in

2017 summarised 21 clinical trials, to evaluate the effects of vitamin D supplementation on inflammation and endothelial function. In that systematic review, they reported that they could conclude that vitamin D supplementation leads to an improvement in inflammation.

Thirteen of the studies included in the systematic review reported no significant effects of vitamin D supplementation on inflammation. However, 8 of the studies reported significant improvements in the biomarkers/parameters measured. [Agbalalah et al., 2017]. Results are inconsistent due to different types and doses of vitamin D supplementation, different duration and different baseline levels of 25 OHD. Moreover, different studies measured different inflammatory biomarkers to evaluate systemic inflammation.

1.3.5 Vitamin D and lipid profiles

Association of vitamin D and lipid profiles

It is reported that people with sufficient vitamin D levels have a more favourable lipid profile, a finding confirmed by many observational studies. An Indian study in 2013, on 150 subjects, showed that vitamin D deficiency is significantly associated with dyslipidaemia (p = 0.0001)

[Chaudhuri et al., 2013]. A large cross-sectional study and retrospective cohort study of more than 4 million subjects in the US compared lipid profiles in vitamin D deficient participants with those with optimal levels and evaluated the effects of changes in 25 OHD levels on lipid

47 profiles. They found that patients with optimal vitamin D compere to vitamin D deficient patients had lower total cholesterol, lower LDL cholesterol levels and lower triglyceride, and higher HDL levels (p = 0.0001). They further showed that raising vitamin D levels from 20 to

30 ng/mL, increased total cholesterol (p = 0.01) and HDL cholesterol (p = 0.02) with no significant changes in LDL cholesterol levels (p = 0.06) and triglycerides (p = 0.97) [Ponda et al., 2012]. In another cross-sectional study in Finland, 909 men aged 45 to 70 were recruited to evaluate the association of vitamin D and lipid profiles. They showed a significant negative association between 25 OHD and total cholesterol, LDL and triglyceride (p <

0.001), but no significant association with HDL cholesterol levels [Karhapää et al., 2010].

Jorde et al. in a cross-sectional study on more than 10,000 subjects reported a significant increase in total cholesterol, HDL and LDL levels and significant decrease in LDL to HDL cholesterol ratio across increasing serum 25 OHD quartiles after adjustment for potential confounders (p < 0.001) [Jorde et al., 2010]. In 2015, a cross-sectional study of more than

1000 children 1 to 5 years old indicated that a 10 nmol/L increase in 25 OHD levels was associated with a decrease in total cholesterol and triglyceride (-1.08 mg/dl and -2.34 mg/dl, respectively) [Birken et al., 2015]. A systematic review and meta-analysis in the paediatric age group in 2014 found a weak negative association of 25 OHD levels and triglyceride, total cholesterol and LDL cholesterol and positive association with HDL cholesterol levels

[Kelishadi et al., 2014]. These inconclusive data could be due to different ethnicity, adjustment for different covariates, different age range and different methods of 25 OHD levels measurements. Moreover, some of the studies were on children and others on adult populations.

Effects of vitamin D supplementation on lipid profiles

Although data from observational studies showed an association of vitamin D with favourable lipid profiles, clinical trials failed to indicate causality for this association. Only a few clinical trials demonstrated a beneficial effect of vitamin D supplementation on lipid profiles. Kane et al. investigated the effects of taking 1000 to 3000 IU per day vitamin D supplement on lipid profiles in 49 adults with vitamin D insufficiency and found that vitamin

D supplementation had no effect on total cholesterol, HDL or LDL levels. However, a trend for decreases in cholesterol and LDL levels in subjects who were already taking atorvastatin was observed (p = 0.08, p = 0.05, respectively) [Kane et al., 2013]. Liunghall et al. were amongst the earliest investigators to assess the effects of 30 IU per day vitamin D supplement on lipid profiles; this led to no significant effects over 3 months in middle aged men

[Ljunghall et al., 1987]. Withman et al. evaluated the effect of a single dose of 100,000 IU vitamin D on 50 adults and revealed no significant differences between treatment and control groups in terms of lipid profiles [Witham et al., 2013]. Similarly, a randomised clinical trial on 151 patient age 18 to 85 years, with vitamin D deficiency investigated the effects of

50,000 IU vitamin D supplementations per week over 8 weeks and found no significant effects on lipid profiles [Ponda et al., 2012]. Zhou et al. compared the effects of 50,000 IU per week vitamin D supplement in obese and normal weight participants and found that vitamin D supplementation has no effect on serum lipid levels in obese or normal weight participants [Zhou et al., 2013]. A population-based prospective study on 464 postmenopausal women compared an intervention group (receiving 300 IU per day vitamin D supplement) with a control group (receiving placebo) over 3 years of follow-up. This found that LDL increased and the HDL to LDL ratio , and HDL decreased significantly in the vitamin D group compared to control group (p = 0.03, p = 0.001, p < 0.001) [Heikkinen et al.,

1997]. Another clinical trial failed to show any effects of 5000 IU per day vitamin D

49 supplement on total cholesterol, triglyceride, LDL and HDL in 100 diabetic patients [Yiu et al., 2013]. Furthermore, a Scottish study reported no significant differences in total cholesterol in 61 diabetics who received one dose of either 100,000 IU or 200,000 IU vitamin

D supplement compared with a control group [Witham et al., 2010]. These inconsistent results could be due to different doses of vitamin D supplements, different duration of intervention or different populations. Further, most of studies in this area were designed for other outcome purposes.

1.3.6 Vitamin D and blood pressure

Association of vitamin D and blood pressure

Most of the studies accessing the associations between vitamin D status and blood pressure are cross-sectional in design and provide consistent results. However, most of the studies in this area were not specifically designed to evaluate the association of vitamin D and blood pressure.

A cross-sectional study on 701 adolescents has shown a significant correlation between low

25 OHD levels and high systolic blood pressure (p = 0.02) and high diastolic blood pressure

(p < 0.01) [Parikh et al., 2012]. In another large cross-sectional study in the third NHANES, blood pressures were allocated into 5 categories and showed that lower 25 OHD levels were associated with higher blood pressure (p < 0.001). However, the association became insignificant after adjustment for age [Judd et al., 2008]. Another study on a similar group in

2011, showed that lower 25 OHD levels were associated with pre-hypertension after adjustment for BMI, serum cholesterol, hs-CRP and estimated glomerular filtration rate

[Sabanayagam et al., 2012]. Similarly, another population-based study on 6810 British adults examined the relationship between 25 OHD and metabolic syndrome components and

50 revealed that lower levels of 25 OHD were associated with prevalence of high blood pressure

(p < 0.004) [Hyppönen et al., 2008].

Data from prospective studies are limited in this area. One prospective cohort study in 1484 women aged 32 to 52 years old, showed that women in the lowest quartile of 25 OHD levels had a significantly higher incidence of hypertension compared to the highest quartile (β =

1.66, p = 0.01) [Forman et al., 2008]. Jorde et al. in a prospective study followed 4,125 participants without hypertension for duration of 14 years and reported that those in the lowest quartile of 25 OHD (<16 ng/mL) compared to those in the highest quartile (>25 ng/mL) had a 4 mm Hg mean increase in systolic blood pressure (p-trend <0.05). However, the odds ratio for having hypertension did not differ between groups [Jorde et al., 2010]. A prospective study on postmenopausal women in the Women’s Health Initiative study in the

US did not find that serum vitamin D levels were associated with changes in systolic blood pressure or diastolic blood pressure or incidence of hypertension [Margolis et al., 2012].

There are data to suggest that improvement in high blood pressure with vitamin D supplementation may be seen only among those with low levels of 25 OHD, which could be the main reason for finding conflicting results in this area.

Effects of vitamin D supplementation on blood pressure

Although observational studies suggest an association between lower levels of vitamin D and raised blood pressure, results from clinical trials are not convincing regarding the beneficial effects of vitamin D supplementation on blood pressure. Several clinical trials in this field are on postmenopausal women. In a randomized clinical trial 305 healthy menopausal women were randomly allocated to receive 400 IU or 1000 IU vitamin D supplement or placebo over one year. This showed that there is a seasonal difference in systolic and diastolic blood pressure, independent of vitamin D supplementation. Furthermore, no effect of different

51 doses of vitamin D supplementation were seen on systolic or diastolic blood pressure [Wood et al., 2012]. Similarly, in a double blind randomized clinical trial 36282 women were randomized into two groups to receive 1000 mg of calcium supplement plus 400 IU vitamin

D or placebo. These authors reported that calcium and vitamin D supplementation over 7 years had no effect on either diastolic or systolic blood pressure. Also, the incidence of hypertension was not significantly different between the intervention and placebo groups

[Margolis et al., 2008]. However, other clinical trials have shown different results. A randomized clinical trial on 148 postmenopausal women aged more than 70 years with 25

OHD levels less than 25 ng/mL demonstrated that supplementation with 1200 mg per day calcium plus 800 IU vitamin D over 8 weeks, significantly reduced systolic blood pressure (p

= 0.02), but had no significant effects on diastolic blood pressure [Pfeifer et al., 2001].

Another study reported that hypertension is more prevalent in African-Americans than whites, and evaluated the effects of different dosages of vitamin D supplementation on

African-Americans [Forman et al., 2013]. They randomised 283 adults to receive 100, 2000, or 4000 IU of cholecalciferol per day or placebo, for 3 months. They reported that for each 1 ng/mL increase in serum 25 OHD, there was a small but significant reduction in systolic blood pressure (0.2 mm Hg; p = 0.02), after adjustment for baseline systolic and diastolic blood pressure. However, there was no effect of vitamin D supplementation on diastolic pressure (p = 0.37) [Forman et al., 2013]. There are two systematic reviews and meta- analyses in this area, both of which showed inconsistent results. Witham et al. [Witham et al.,

2009]. performed a meta-analysis on 8 randomized clinical trials and reported a non- significant decrease in systolic blood pressure in vitamin D groups compared to placebo group [-3.6 mmHg, 95% CI, -8.0 to 0.7]. A statistically-significant decrease was observed in diastolic blood pressure (-3.1 mmHg, 95% CI -5.5 to -0.6). It was shown that there was no reduction in systolic and diastolic blood pressure in studies examining patients with normal

52 blood pressure at baseline. Another meta-analysis in 2010 [Pittas et al., 2010], showed no significant effect of vitamin D supplementation on systolic blood pressure or on diastolic blood pressure. There was no difference in change in systolic or diastolic blood pressure in studies that provided vitamin D alone, or with calcium nor were there changes in systolic blood pressure in studies that provided higher (≥1000 IU/day) versus lower (<1000 IU/day) doses of vitamin D supplementation. Within the meta-analysis, a study using higher doses of vitamin D showed a significant effect on diastolic blood pressure. These conflicting results can be attributed to limited sample sizes which did not allow detection of small differences, short follow-up duration, and heterogeneity in study populations.

1.3.7 Vitamin D and diabetes mellitus

Association of vitamin D and diabetes mellitus

Vitamin D may have effects on diabetes by improving beta cell function and insulin sensitivity. Many observational studies have evaluated the association of vitamin D deficiency with impaired glucose metabolism, which is a risk factor for CVD. Results from the literature are inconclusive and inconsistent.

In a Korean population-based cross-sectional study, participant with the lowest 25 OHD levels had a 1.75-fold higher risk for diabetes than those with the highest levels [Rhee et al.,

2012]. Another Korean study by Nam et al. in 2017 showed that the vitamin D deficient group (25 OHD <10 ng/mL) not only had higher BMI, higher total cholesterol and higher blood pressure than those with normal vitamin D levels (25 OHD >30 ng/mL), but also had higher independent risk of diabetes after exclusion of other risk factors [Nam et al., 2017]. In the Tromso study on 160 participants aged 30 to 75 years old, subjects with high 25 OHD levels (mean 85 nmol/L) compared to those with low levels (mean 40 nmol/L) had a significantly higher insulin sensitivity (p = 0.02) index and lower HbA1c (p < 0.001)

[Grimnes et al., 2011]. An analysis of the third National Health and Nutrition Examination data on 6228 adults revealed an inverse association between 25 OHD levels and insulin resistance after adjustment for sex, age, physical activity, BMI and ethnicity [Scragg et al.,

2004]. Knekt et al. pooled data derived from two nested case-control studies with a total of

19518 participants, during a follow-up period of 22 years. They showed that women had lower 25 OHD levels than men and by increasing vitamin D levels the incidence of diabetes in men decreased significantly after adjustment for potential confounders (OR between the highest and lowest quartiles 0.28 (95% CI: 0.10–0.81) in men and 1.14 (0.60 –2.17) in women [Knekt at al., 2008]. Pittas et al. followed 83779 women aged 30 to 55 years old for

20 years and found that vitamin D intake from supplements and not from diet was significantly associated with a lower risk of diabetes mellitus (0.87; p-trend: 0.04) [Pittas at al., 2006]. An inverse but nonlinear association between 25 OHD and HbA1c was observed in the British birth cohort study on 7198 Caucasian subjects [Hyppönen at al., 2006].

Inconsistent results in this area could be due to differences in study populations; some studies were conducted on obese subjects, some on those at risk of diabetes as well as on healthy subjects. There were also differences in age range and adjustment for different covariates.

In summary, most of observational studies in this area showed an association between vitamin D deficiency and diabetes; however, still more investigation is needed for causal effects.

Effects of vitamin D supplementation on diabetes

Many clinical trials have evaluated the effects of vitamin D supplementation on metabolic outcomes including glucose levels, insulin levels, homeostatic model assessment of insulin resistance (HOMA-IR), homeostatic model assessment of beta cell function (HOMA-B),

HbA1c and diabetes incidence to assess the causality of the relationship. Studies were

54 performed on healthy subjects, persons with impaired fasting glucose and patients with

T2DM. Some studies evaluated the effects of vitamin D alone and some with a combination with calcium. A clinical trial on 32 diabetic subjects evaluated the effects of 40,000 IU vitamin D supplementation per week over 6 months, on fasting glucose, insulin and HbA1c levels. They found no significant differences in glycaemic outcomes between the supplement and placebo group, and there were no significant differences in these parameters in each group after 6 months compared to baseline [Jorde at al., 2009]. Another study on diabetic patients was performed by Elkassaby et al. in 2014. They randomised 50 adults with T2DM to receive 6000 IU vitamin D or placebo daily for a period of 6 months. They showed no significant differences in change in HbA1c, fasting plasma glucose or post prandial plasma glucose between the two groups after 6 months of intervention. However, after 3 months, fasting blood glucose and post prandial glucose levels were significantly lower in the vitamin

D group than the placebo group (p = 0.007 and p = 0.005, respectively) [Elkassaby at al.,

2014]. A Korean prospective clinical trial on 129 patients with T2DM and low 25 OHD levels (<20 ng/mL) evaluated the glucose-lowering effects of vitamin D supplementation.

They randomised participants to vitamin D (1000 IU daily or vitamin D plus 100 mg calcium, twice a day) or a placebo group. They showed no significant differences in glucose, HbA1c and HOMA-IR between the two groups [Ryu at al., 2014]. Finally, another study on non- insulin requiring diabetic patients revealed that supplementation of Vigantol oil, providing

1904 IU vitamin D3 per day, for six months had no effect on glucose or insulin levels

[Strobel at al., 2014]. Results from studies performed on healthy or pre-diabetic subjects are inconsistent. In a randomised clinical trial by Pittas et al. 314 adults without diabetes were randomised to two groups, to receive 500 mg calcium plus 700 IU vitamin D supplement or placebo for 3 years. Participants in the supplement group who had an impaired fasting glucose at baseline had a significantly lower increase in fasting plasma glucose (0.02 mmol/L

55 p = 0.042) and HOMA-IR (0.05, p = 0.031) compared to those in the placebo group. There were no significant differences in the change in fasting blood glucose and HOMA-IR between the two groups in subjects who had a normal fasting glucose at baseline [Pittas at al.,

2007]. The Women’s Health Initiative Calcium/Vitamin D Trial investigated the effects of

1000 mg calcium plus 400 IU vitamin D3 supplementation per day for 7 years on 2291 healthy women. They showed that calcium plus vitamin D supplementation had not decreased the risk of diabetes over 7 years [de Boer at al., 2008]. Grimnes et al. performed a population-based study followed by a clinical trial. They randomised participants with low 25

OHD levels to receive 20,000 IU vitamin D or placebo weekly for 6 months. After the follow-up period, there were no significant differences in insulin sensitivity index or HbA1c between the two groups. They concluded that vitamin D supplementation does not have any beneficial effect on insulin secretion or sensitivity in healthy postmenopausal women within the short time of 6 months [Grimnes at al., 2011]. There are a few small trials performed in pre-diabetes; they found no effects of different doses of vitamin D on glucose metabolism or glycaemia. A randomised clinical trial on 255 adults with pre-diabetes investigated the effects of 20,000 IU vitamin D supplementations per week on prevention of progression to T2DM.

In subjects with normal vitamin D levels, vitamin D supplementation was unlikely to prevent progression from pre-diabetes to diabetes [Jorde at al., 2016]. Similarly, in 2014 another study was conducted on 95 adults at risk of T2DM. They randomised participants into two groups, to receive 2000-6000 IU vitamin D plus 1200 mg calcium supplement or placebo, and measured insulin sensitivity, insulin secretion and beta cell function at baseline and after

6 months of intervention. They showed no change in insulin secretion or insulin sensitivity after 6 months of treatment [Gagnon at al., 2014]. There are also a few meta-analyses in this area, which pooled results from different studies. One meta-analysis, which included 35 clinical trials, showed no effect of vitamin D supplementation on glucose metabolism or

56 prevention of diabetes [Seida at al., 2014]. Another earlier meta-analysis, including 11 clinical trials and 8 observational studies also showed no effects of vitamin D supplementation on glycaemic outcomes [Mitri et al., 2011].

1.3.8 Vitamin D deficiency and pregnancy

Vitamin D deficiency during pregnancy not only affects maternal health including causing preeclampsia, preterm delivery and gestational diabetes, it also can cause disturbed skeletal homeostasis, rickets and fractures in the newborn [Ozdemir et al., 2018; Christesen et al.,

2012].

Studies reported from different countries that prevalence of vitamin D deficiency in pregnant women and in infants ranged from 4% to 60% and from 3% to 86%, respectively [Palacios et al., 2014 and Prentice et al., 2008].

A cross-sectional study on 741 pregnant women showed that vitamin D deficiency could be a sign of insulin resistance during pregnancy [Maghbooli et al., 2007]. A systematic review by

Websky et al. showed that vitamin D deficiency during pregnancy was associated with a higher risk of preeclampsia and gestational diabetes [Websky et al., 2017]. Another systematic review and meta-analysis reported that vitamin D deficiency is associated with an increased risk of preeclampsia [Tabesh et al., 2013]. However, in contrast, a retrospective cohort study showed that vitamin D deficiency was not associated with adverse pregnancy outcomes including preeclampsia or gestational diabetes in 310 singleton pregnant women during their first trimester [Flood-Nichols et al., 2015].

These inconsistent results could be due to using different cut-offs for defining vitamin D deficiency, measuring 25 OHD levels in different trimesters of pregnancy and using different methods to measure serum 25 OHD levels.

1.3.9 Effects of behavioural intervention by using mobile health (m-health) on health outcomes

Many qualitative studies have been conducted in different populations to assess the quality and usability of different applications on life style-related health [Duclos et al., 2017; Peng et al., 2016]. However, to our knowledge there are very limited studies using m-health interventions to increase safe UV exposure to achieve sufficient vitamin D levels. Moreover, although there are hundreds of health-related applications available in the market, most of these were not developed by health professionals or academics [Ventola et al., 2014]. The validity and reliability of almost all of the applications were not confirmed. A randomised clinical trial evaluated the effects of using a smart phone application, delivering sun protection advice, on protection behaviour. The study was conducted in New Mexico and provided an android application to be used by 604 adults for 10 weeks. They showed that use of the mobile application was lower than expected, (41%) but was associated with an increase in sun protection behaviour (p = 0.04) [Buller at al., 2015]. A clinical trial conducted from

August 2009 to November 2012, evaluated the effectiveness of using text messages to modify lifestyle to prevent T2DM. They divided 537 participants into two groups, to receive frequent mobile phone text messages providing random lifestyle advice, or to receive general lifestyle modification advice at baseline only. The incidence of T2DM was lower in the intervention group compared to the control group (18% versus 27% p = 0.015) [Ramachandran at al.,

2013]. A meta-analysis evaluated the efficacy of using a text messaging intervention to reduce smoking behaviour. Twenty papers from 10 different countries were included in that meta-analysis. Overall, they indicated that using health promotional text messaging about quitting smoking can reduce cigarette consumption [Scott-Sheldon at al., 2016]. Another meta-analysis on 19 intervention studies evaluated the efficacy of text messaging as a strategy for health promotion and showed that personalization of messages was significantly associated with higher efficacy of the intervention. There were no significant differences

58 between interventions that used only text messages or interventions that included texting plus other components. Also, it has been shown that trials that decreased the frequency of messages over the course of the study and individualized the messages were more successful than those studies that just used fixed frequency messages and not personalized messages.

Trials on smoking cessation or on physical activity were more successful than studies targeting other health outcomes [Head et al, 2013].

During adolescence, young people strive for independence and begin to make decisions that impact them for the rest of their lives. Also future patterns of adult health are established during this age range. It has been suggested that greater attention to adolescent health is needed within each public health domains if global health targets are to be met. Use of smartphones is high among younger demographic groups, with 73% of adolescents aged 13 to 17 years in the US having access to a smartphone and 58% having downloaded apps. A qualitative study by Chan et al. evaluated the prevailing beliefs regarding the use of smartphone applications for managing health among adolescents. They showed that 40% of their participants used health apps for managing medical conditions or achieving a fitness goal [Chan et al., 2017]. In a pilot study, an m-health intervention for management of type 1 diabetes among 20 adolescents aged 12 to 16 years old was assessed. It was shown that m- health diabetes app with the use of gamification incentives led to an improvement in the frequency of blood glucose monitoring in adolescents with type 1 diabetes [Cafazzo et al.,

2012].

1.4 Importance of the study

CVD is one of the major health concerns and a main cause of death around the world. In

Australia, almost 3.7 million people are affected by CVD. Cardiovascular disease is responsible for around 20% of the burden of disease and it is the second major cause of

59 disease burden in Australia [Australian Bureau of Statistics, 2016]. Vitamin D deficiency is also common and it has been reported that almost one third of Australian adults have low vitamin D levels [Daly et al., 2012]. Vitamin D deficiency is an important health risk for young women, particularly during the child-bearing years. Yet few vitamin D studies have focused on young women [Australian Institute of Health and Welfare 2011].

Evidence gaps or limitations in this area are as follows:

1 No conclusive findings on the effect of Ultraviolet B radiation on 25 OHD levels, due to

trials being small and heterogeneous, with few addressing the issue of finding the balance

between sufficient sunlight exposures to maintain healthy vitamin D status, while

controlling the risk of skin cancer .

2 A lack of studies in young people, and lower quality randomized clinical trials in younger

populations.

3 Most previous trials have been limited by the imprecision of the assays used to measure

serum 25 OHD, with a critical need to employ state-of-the-art liquid chromatography-

tandem mass spectrometry (LC-MS/MS) to measure serum 25 OHD concentrations with

much higher precision for both D2 and D3 metabolites.

4 Lack of any scientific validated mobile-based application to provide advice on how much

sun exposure is sufficient

5 Lack of consistent evidence on the effects of vitamin D improvement on body

composition

1.5 Rationale

In our Safe-D study we have tried to address previous studies’ limitations and tried to fill the knowledge gaps in these areas.

Our study was conducted in two parts. Part A was a cross-sectional design and Part B was a randomised clinical trial. In the cross-sectional study, we evaluated the association of 25

OHD levels and cardiovascular risk factors including, hypertension, obesity, impaired lipid profiles, impaired glucose metabolism and body composition in young women age 16 to 25 years old. Those with 25 OHD levels between 25 to 75 nmol/L were invited to participate in the intervention part. In the intervention study, we evaluated the effects of 1000 IU vitamin D supplementation per day, or behavioural intervention using mobile based application, over 4 months and 12 months, on cardiovascular risk factors and body composition in women aged

16 to 25 years old.

Study rationale:

1) We focused on 16 to 25 years old because of the high prevalence of vitamin D

deficiency among this age group and lack of evidence in this age group as most of

previous studies focused on elderly.

2) The importance of this age range as people become more autonomous, and

independent and individual environmental as well as behavioural factors play a

greater role in shaping health patterns that can have long-term health consequences.

3) The high prevalence of cardiovascular disease in Australia and the importance of

recognizing and controlling risk factors from a young age.

4) The popularity in this demographic of communication using mobile technology and

social media.

5) The importance of improving vitamin D levels in young women, as during

childbearing years there is potential harm of vitamin D deficiency on both the mother

and offspring.

6) The lack of any scientifically-validated health application to provide personalised

advice on how much sun exposure is sufficient and appropriate.

7) The high prevalence of skin cancer in Australia and the importance of keeping the

balance of maintaining adequate vitamin D levels with the risk of skin cancer from

excessive sun exposure.

In summary, although it has been shown that vitamin D deficiency is associated with many diseases and health conditions, the potential impact of vitamin D deficiency on young

Australian women has yet to be established. Moreover, there is still an evidence gap about the best intervention to improve vitamin D levels. Most of the previous studies have been limited by imprecise methods used to measure 25 OHD levels, small sample size and short duration of the intervention.

1.6 Aims

Primary aims for the cross-sectional part

Evaluate the association of serum levels of 25 OHD and cardiovascular risk factors (obesity, lipid profiles, glucose metabolism, blood pressure, inflammatory biomarkers, fat distribution and body composition) among young (16 to 25 years old) women, living in Australia

Primary aims for the intervention part

Measure the effectiveness of a behavioural and a pharmacological intervention to increase 25

OHD levels over 4 months, compared with control group.

Secondary aims for the intervention part

1) Measure the effectiveness of a behavioural and pharmacological intervention to increase circulating 25 OHD levels over 12 months

2) Compare the effectiveness of the two interventions to increase circulating 25 OHD levels over 4 months and 12 months of intervention

3) Measure changes in obesity, lipid profiles, glucose metabolism, blood pressure, inflammatory biomarkers, fat distribution and body composition after 4 months and 12 months of intervention

4) Measure compliance of app use and vitamin D supplementation in relation to the primary outcomes for the two intervention groups

5) Measure the impact on UV exposure and “SunSmart behaviour” of a behavioural intervention delivering tailored advice on how to safely meet sun exposure requirements (by comparing with UV exposure in the controls)

Exploratory aims for the intervention part

1) Assessing change in knowledge about safe sun exposure after the intervention

2) Determining the proportion of young women in each arm of the trial in whom recommended vitamin D status is achieved at 4 and 12 months

See Figure 2.1 for study design information.

1.7 Methodological approach and theoretical framework

Vitamin D deficiency is an important health issue postulated to be associated with increased risks for many chronic health conditions including cardiovascular disease [Wang et al.,

2017]. Vitamin D deficiency is leading to considerable suffering and economic loss

[Alshishtawy et al., 2012]. However, many evidence gaps exist in this area, especially about the safest and most effective interventions to improve vitamin D status. Moreover, results from previous studies are inconsistent and it remains unclear whether low serum 25 OHD levels are associated with an increased risk of CVD [Skaaby et al., 2017 and Pilz et al., 2016].

These associations have been little studied in young women.

Use of mobile technology is playing an increasingly important role in shaping health care systems [Thimbleby et al., 2013]. m-Health interventions have the advantages of being widely deployable, adaptable and very cost-effective [Kamel Boulos et al., 2014]. Therefore, they can be used for health interventions with minimum costs and can be deployed to a large number of individuals for long term use [Kamel Boulos et al., 2014]. m-health strategies allow for real time communication and data collection, and are convenient [Kamel Boulos et al., 2014].

In Part A of this project we evaluate the association of serum levels of 25 OHD and cardiovascular risk factors among young (16 to 25 years old) women, living in Australia. In

Part B of the project we measure the effectiveness of a behavioural intervention, using a mobile based application, and a pharmacological intervention to increase 25 OHD levels, compared with control group. Also we evaluate the effects of vitamin D supplementation and behavioural intervention on change in CVD risk factors after 4 months and 12 months of intervention.

Chapter 2

MATERIALS AND METHODS

2.1 Study design

Study design is summarised here and details provided below (study protocol is presented in appendix A). This study was conducted in two parts at the Royal Melbourne Hospital, from

2014 to 2017. In Part A, we evaluated the associations of 25 OHD levels with cardiovascular risk factors including obesity, lipid profiles, glucose metabolism, blood pressure, inflammatory biomarkers and body composition in a cross-sectional study of women aged 16 to 25 years [Callegari et al., 2015]. Relevant covariates were examined to explain any association that was observed. Part B [Tabesh et al., 2016], which formed the basis for my thesis was a randomised controlled clinical trial with period of 12 months; we examined the effects of behavioural intervention, using either a mobile-based application or vitamin D supplementation on cardiovascular risk factors in young women with mild vitamin D status

(25 OHD between 50 and 75 nmol/L) and moderate vitamin D deficiency (25 OHD between

25-50 nmol/L), compared to a control group. In this clinical trial, participants were randomly assigned in to one of the three groups. Group 1, received behavioural intervention which involved mobile health technologies (m-health) to promote adequate and safe sun exposure.

The application provided personalized, real-time advice on the estimated time required to be in the sun to achieve adequate 25 OHD levels according to the UV index profile, skin type, skin colour, type of clothes and amount of sun screen used. As well as safety information to help avoid excessive sun exposure. Group 2 received 1000 IU per day vitamin D3 supplementation for 12 months. The control group (Group 3) received general advice on improving 25 OHD levels in the form of a pamphlet on safe methods to naturally improve 25

OHD levels (Cancer Council Victoria (CCV), “How much sun is enough” brochure)

[www.sunsmart.com.au/downloads/vitamin-d/how-much-sun-is-enough-vitamind.pdf].

Participants were allocated into each group by using block stratified randomisation based on baseline 25 OHD levels (based on baseline serum 25 OHD levels 25 to 50 nmol/L and 50 to

75 nmol/L). Weight, height, body composition, metabolic profiles, blood pressure and other variables were measured at baseline (week 0), after 4 months and then after 12 months of intervention. Data were collected using online questionnaires, during a site visit and by wearing a UV dosimeter. Depending on the timing of participation in Part A, participants were investigated in two different groups. First group: participants who completed the Part A site visit less than 2 weeks previously; this group did not need to complete a further questionnaire or site visit to provide baseline data, as this was derived from the recent Part A data. Second group: participants who finished a Part A site visit more than 2 weeks previously; this group needed to complete the online questionnaire [www.limesurvey.org] and undertake another site visit to establish baseline data, in particular to re-check their 25

OHD levels (Figure 2.1).

2.2 Subject selection

This study recruited women aged 16 to 25 years into Part A, the cross-sectional study, through Facebook advertisement (see appendix B) and then randomised those with 25 OHD between 25 and 75 nmol/L into intervention or control groups. The exclusion and inclusion criteria were designed to reduce any potential harm to participants, while selecting a broad sample to maximise the generalizability of the findings, and also to reduce the effects of confounding on results. If at any time during the study participants met any of the exclusion criteria, they were excluded from the study.

2.2.1 Inclusion criteria

Part A (Cross-sectional study):

1. Being female

2. Residing in Victoria, Australia for the duration of the study

3. Aged 16-25 years

Part B (Intervention study):

1. Completion of Part A site visit of the Safe-D study

2. Serum 25 OHD levels between 25 and 75nmol/L

3. Personal ownership and use of a smartphone.

2.2.2 Exclusion criteria

Part A (Cross-sectional study):

1. Unable to give informed consent

2. Pregnant or breast feeding

3. Chronic illness

Part B (Intervention study):

1. Ever had a melanoma, or has/had a first-degree relative who had a melanoma

2. Any participant who is currently pregnant or breastfeeding or planning to conceive

during next 12 months

3. Currently taking ≥800 IU Vitamin D daily, as supplements

4. Any chronic health condition that may cause safety concerns (eg. sensitivity to light or

UV, malabsorption conditions) or confounding the association of 25 OHD levels with

health outcomes.

5. Planning to move outside of Australia and not able to attend their site visits

Recruitment of eligible participants, using targeted Facebook advertisement

Part A: baseline assessment: survey, UV dosimeter, site visit , blood test to select to part B (N=468)

YFHI: baseline assessment: site visit, survey (N=150)

Randomised into Part B, the clinical trial: those with (Part A) baseline 25 OHD 25-75nmol/L (N=234)

Behavioural intervention Pharmacological intervention Control (Using mobile application, (Vitamin D supplementation delivering individualised (General advice) 1000 IU per day) advice on sun exposure) (N=78) (N=78) (N=78)

4-month follow-up assessment (N=186)

12-month follow-up assessment

(N=144)

Figure 2.1 Study design

Sample sizes presented in this Figure are the calculated sample size.

2.3 Proposed sample size

In a previous meta-analysis, it was determined that each 100 IU per day vitamin D supplementation increased serum 25 OHD levels by 1 to 2 nmol/L [Cranney et al., 2007].

Therefore, we expected changes in 25 OHD levels of approximately 10 to 20 nmol/L following supplementation with 1000 IU vitamin D per day for 4 months. Based on this, we calculated that a sample size of 186 participants at 4 months follow-up, and 144 at 12 months follow-up, were required to achieve 85% power to detect 15 nmol/L differences in 25 OHD levels between the groups. We assumed an attrition rate of 20% by 4 months and 30% at 12 months, based on available data. Based on the above, a sample size of 234 was required for the clinical trial (78 per arm) [Leonard at al., 2014]. According to the prevalence of vitamin

D deficiency in young women in Victoria, we also assumed that 50% of the population have

25 OHD levels between 25 to 75 nmol/L. The Victorian Health Monitor Report reported that over 50% of Victorians have vitamin D deficiency in winter [Cancer Council Victoria, 2018].

To achieve the nominated sample size for the intervention study, it was calculated that a sample size of 468 was required in the cross-sectional study. We calculated there would be 48 participants in each arm at the 12 months follow-up which, provides 80% power (or 85 % power in intention to treat) to detect a 15 nmol/L difference in 25 OHD levels between the groups, with a 5% significance level (Table 2.1).

Table 2.1 Calculated sample size

Part of the study Number

Cross-sectional 468

Intervention 234 (78 per arm)

4 months follow-up 186 (62 per arm)

12 months follow-up 144 (48 per arm)

2.4 Recruitment

Participants were recruited for Part A of this study using Facebook advertising, as previously described [Fenner at al., 2012]. The study advertisement was a targeted pop-up ad, which randomly appeared to women aged 16 to 25 year old living in Victoria, Australia. If they were interested, potential participants could click on the advertisement, to be directed to the study website (www.safedstudy.org) or study Facebook page (www.facebook.com/Safe-D-

Study-Page). Participants could express an interest in participating in the study by providing their name and contact details, in a password-protected, secure website. Participants were contacted by phone by a trained researcher to assess their eligibility, to obtain verbal consent and to book their first site visit at the Royal Melbourne Hospital. If the potential participant failed to answer her phone, a voice message or text message was left, followed by two additional attempts at different times after which they were removed from the expression of interest list. Participants who completed the Part A site visit of the Safe-D study were screened for eligibility for Part B of the study.

Upon receipt of their Part A 25 OHD results, participants with 25 OHD levels between 25 and 75 nmol/L were contacted to be further screened for eligibility, and to be invited into Part

B, the clinical trial (the information brochure which was given to participants is presented in appendix C). If appropriate and willing, they were verbally consented after which they were asked to sign the online written consent form, which was provided to them via LimeSurvey

(Australia, version 2.06) [www.limesurvey.org/]. Sufficient time, around seven days, was allowed for volunteers to review the participant information and consent form (PICF), and to sign and return the written consent. All participants aged less than 18 years had to complete a mature minor assessment form (see the appendix D) [Sanci at al., 2004].

After receiving an expression of interest and the written consent, an unblinded team member contacted the participants to randomise them into one of the three groups. Participants were

71 followed for 12 months. All data were collected at two time points, after 4 months from randomisation, and at 12 months from randomisation (Figure 2.2).

Participants submit an expression of interest

Researcher seeks verbal consent. Verbal consent obtained for cross- sectional study

Yes No: end of the participation

Completion of survey and wearing of dosimeter

Site visit and written consent

Serum 25 OHD levels 25 to 75 nmol/L?

No: end of the study Yes

Professor John Wark reviews pathology results: provides written advice for the participant re any abnormal results

Researcher contacts participant: informs participants of results

eligible?

No: end of study Yes: verbal consent obtained participation for intervention study

Yes No- end of study participation

≤2 weeks from Part A >2 weeks from Part A site visit- all data used site visit- all assessments for baseline, no repeated and modified assessments required questionnaires

Send out PICF and await written consent either electronic or print

Electronically randomised into 1 of the 3 arms and relevant details sent out to participant

Group 1: receive instructions Group 2: receive vitamin Group 3: receive general on how to download and D supplements with information about vitamin utilize mobile app instructions on dosage D

Unblinded study coordinator available via email and phone to answer any study related queries, and record any study related adverse events

4 month study visit to include modified questionnaires, site visit with physical assessment, pathology and Leonardo and review of adverse events, concomitant medications and study compliance by unblinded coordinator

Month 12 study visit, to include repeat of all baseline assessments and scans

END OF STUDY PARTICIPATION

Figure 2.2 Study protocol

2.5 Verbal, written and electronic consent

All participants provided verbal and written consent separately for each part of the study. In

Part A, verbal consent was obtained by phone. During the phone call eligibility was confirmed, and then detailed information about the study, including purpose of the research, what was involved for the participant, the benefits of participation, possible minor risks and details about confidentiality were provided to the participant. If they agreed to participate, the questionnaire link was sent to their email address, and their first site visit was booked. Each of the questionnaires included the PICF (participant information and consent form), which contained an electronic consent. A hard copy of the written consent form was also sent to all participants before the site visit. Prior to the collection of any biological data, the written consent was reviewed at the study visit.

Participants with 25 OHD levels between 25 and 75 nmol/L in Part A visit were then recontacted by phone to screen for further eligibility, and if so were verbally consented for the clinical trial (Part B). During the phone call, detailed information about Part B of the study was provided. If the participant was happy to participate, an electronic consent form, available in a secure link, was sent. Information about the study was displayed on the screen and required the participant to provide their name and click on “I agree” to express that they freely agreed to participate. After the study team received the electronic consent, the participant’s details were sent to an unblinded team member for randomisation. The unblinded team member then contacted the participant to inform her of her group allocation and provided with instructions and detailed information about the intervention. At the beginning of the follow-up visit a written consent was obtained before collecting any biological data.

2.6 Trial interventions

2.6.1 Behavioural intervention group

All participants randomised into this group received an email with the instructions and links to download the Safe-D mobile-based app [Heffernan et al., 2016]. They were asked to install the application and use it every day. This app was developed by Boosted Human and designed for both iOS and Android operating systems [Heffernan et al., 2016]. A web- based survey was used to guide development of the app to determine what young women wanted in a health application. Messages were created by using current information from approved

CCV websites. After development of the Safe-D app, it was beta tested by different team members, with Apple or android devices. All detailed information regarding the development of the Safe-D app was presented elsewhere [Heffernan et al., 2016].

The app delivered advice each morning about how to obtain safe (as per the SunSmart guidelines) and adequate sun exposure during that day, including an estimate of the required sun exposure necessary for adequate vitamin D production [Glanz et al., 2008]. Advice was generated according to each participant’s characteristics, including skin colour, clothing type, use of sun screen and local UV index. The local UV index was obtained from the Bureau of

Meteorology. Location was determined using the phone’s global positioning system (GPS) or by manual entry. Participants started the app timer whenever they were subject to sun exposure, and stopped whenever they were no longer exposed to sunlight. All records were stored in the Safe-D app (See appendix E). According to the recorded sun exposure, the app provided messages to participants to encourage them to stay longer under the sun, or to cover up. Messages were sent as push notifications. Messages encouraged participants to obtain suitable levels of sun exposure and included some general advice on the importance of vitamin D in the body, and consequences of vitamin D deficiency. To convey messages

76 simply, different shapes of sun flowers were displayed in the Safe-D app (sunflower with happy face to show adequate levels of sun exposure, sad face to show inadequate sun exposure and a burnt drooping sunflower to represent excessive sun exposure). Whenever a participant was exposed to too much sunlight, an auto generated safety message was sent as a warning; the application also reported any excess sun exposure to our research team. Use of the application was measured by the number of times app was opened by the participant.

2.6.2 Pharmacological intervention group

Participants in this group received 1000 IU per day vitamin D3 supplement orally for the duration of one year. All the supplements were donated by Swisse (Swisse Wellness Pty Ltd,

Victoria, Australia) and had expiration dates that lasted for the duration of the study.

Information on storage conditions, and dose and route of administration of supplement, was provided to all participants in this group. They were advised to take one capsule daily, with a meal, and to store them in the refrigerator, ensuring a temperature of less than 25 degree

Celsius. Weekly text messages were sent to remind participants to take their vitamin D supplement daily and if they missed any, to take the missed dose as soon as possible that day.

Messages were sent weekly for the first month of participation, then fortnightly, and then monthly.

2.6.3 Control group

Participants in the control group received a pamphlet produced by the Cancer Council of

Victoria entitled “How much sun is enough?” [www.sunsmart.com.au/downloads/vitamin- d/how-much-sun-is-enough-vitamin d.pdf]. The pamphlet contains general information about what is vitamin D and why it is important, how much sun we need for healthy 25 OHD levels, who is at risk of vitamin D deficiency, when sun protection is required, and how to

77 achieve enough vitamin D using safe levels of UV exposure. It also provided dietary information and links to the SunSmart Victoria website, Australian and New Zealand Bone and Mineral Society and other relevant websites. The brochure was sent to participants via email. One week after the brochure was sent, participants were contacted to be asked if they received and read the pamphlet and if they had any questions. The pamphlet was also downloadable on the SunSmart website (www.sunsmart.com.au).

2.7 Randomisation

Randomisation was done through the cooperation of a blinded team member, an unblinded team member and the study statistician. After a blinded member received a signed e-consent form, details of the recruited participant, including their name, study identification number and contact numbers, were sent to the study statistician by email. The study statistician allocated the participant into one of the three groups using a prior-generated sequence of group-codes (Cinderella, Ariel and Snow White) and sent the code to the unblinded member

(who generated the key to group allocation) by email. The participants were then contacted by the unblinded team member to inform her about the group allocation. Randomisation was performed using stratified block randomization with vary block sizes of 3, 6, and 9, based on baseline serum 25 OHD levels 25 to 49 nmol/L and 50 to 74 nmol/L.

2.8 Blinding

In this project, all of the researchers involved in recruitment and laboratory measurements, data collection, conduct of site visits, data entry and data analyses were blinded to the participant’s group allocation. Two separate databases were used, one for blinded members and one for unblinded members, to reduce inadvertent unblinding. Only the one unblinded team member, who generated the randomisation, knew the group allocation of the participant.

All participants were blinded to their 25 OHD levels. The vitamin D results were sent to

78 participants only if clinically significant (25 OHD < 25 nmol/L), with a standard letter to follow-up the results with their usual doctor. All blood results, including 25 OHD levels, were provided to participants at the end of the study. All participants were informed that many members of the study team were not allowed to know the group to which they were allocated, and were asked to contact only the unblinded member, in the case of any matter regarding their group allocation. If any unblinding occurred, this was recorded in the database.

2.9 Data collection

A wide range of information was collected in this study. These data were sourced from the self-administered online questionnaire, wearing the sun monitoring watch and site visits.

Collected data included demographics, medical history, use of health care professionals and medications, allergy data, diet and nutrition intake and UV exposure. Examination included blood pressure, waist circumferences, hip circumferences, height, weight, skin reflectance for skin colour (Spectrometer, Konica Minolta Optics Inc, CM-2500d), Investigations comprised lipids, glucose, insulin, PTH, 25 OHD2 and 25 OHD3 levels, bone mineral density, body composition and muscle strength and power.

The extensive data collection in this study allowed for the identification of possible relationships between vitamin D status and a range of other health conditions in the cross- sectional part. Moreover, it allows adjustment for different confounders on the effects of 25

OHD levels improvements and cardiovascular risk factors.

All of the data were collected from all participants at baseline (month 0) and at the end of the study (after 12 months). A range of measures were collected at 4 months, which is the approximate duration it takes for 25 OHD levels to reach a steady state with intervention

[Wootton et al., 2005]. At 4 months follow-up a modified version of the questionnaires was used in that we collected all data except the food frequency questionnaire and some demographic information. DXA scan data were also not collected at 4 months. A repeat baseline visit was conducted for those who completed the Part A visit more than 2 weeks before being randomised into Part B of the study, to check their 25 OHD levels. Those requiring a second baseline assessment were stratified according to the time between the Part

A visit and randomisation into Part B: 1) more than 2 weeks, but less than 8 weeks’ gap 2) more than 8 weeks, but less than 16 weeks’ gap and 3) more than 16 weeks’ gap, from Part A visit. The components of the repeated baseline visit were different for each stratum.

Data collected for the second baseline visit for those with a gap of between 2 and 8 weeks from the Part A visit, included blood pressure, skin reflectance, PTH and 25 OHD levels. For those with time between visits between 8 and 16 weeks a modified version of the questionnaire was undertaken, and blood pressure, waist and hip circumferences, height, weight, skin reflectance, lipids, glucose, insulin, reproductive hormones, PTH and 25 OHD levels were measured. For those in the window of more than 16 weeks gap from Part A visits, all data except the food frequency questionnaire, some demographic information, UV exposure, DXA scan data were collected. Details of the data collected at each time point are presented in the following table (Table 2.2).

Table 2.2 Data collected at each time point

Data collected Rationale for data Baseline- Baseline- if Baseline- if Baseline- if Mid study End of collection Part A data visit > 2 weeks visit > 8 weeks visit > 4 (month 4) study utilized but < 8 weeks but <16 weeks months from (month Part A 12) baseline Questionnaires Module A Demography, medical Yes Modified (use of Yes history, use of health care No Modified Modified health care professionals, use of professionals, medications, allergy data use of medications) Module B Nutrition, weight control Yes No Modified Modified Yes Yes Module C Body image, alcohol use, Yes No Modified Modified Yes Yes tobacco use, illicit drug use Module D Diet, physical activity, Yes No Modified Modified Yes Yes Confidence in doing exercise and stick to a diet, pain, injuries, sun exposure, mental health data (including the PHQ-9 and GAD-7) CCV Questionnaire Diet, portion sizes Yes No No No No Yes Questionnaire Captures changes from Part No No No (modified from A data that may have Yes No No Modules A, B, C and occurred within past 8 D) weeks Site visit components

Blood pressure General health Yes Yes Yes Yes Yes Yes Resting heart rate General health Yes Yes Yes Yes Yes Yes

Waist and girth General health Yes No Yes Yes Yes Yes Height General health Yes No Yes Yes Yes Yes Weight General health Yes No Yes Yes Yes Yes

Skin reflectance Objective measure of Yes Yes Yes Yes Yes Yes melanin density (skin colour) UV Monitoring Measure of daily UV No No No No Yes Yes exposure Measures from Phlebotomy/ bioassays performed

FBE (incl. RCV) General health Yes No Yes Yes Yes Yes

HbA1c General health Yes No Yes Yes Yes Yes Glucose General health Yes No Yes Yes Yes Yes Creatinine General health Yes No Yes Yes Yes Yes eGFR General health Yes No Yes Yes Yes Yes Lipids (TC, HDL, General health Yes No Yes Yes Yes Yes LDL, TG) Insulin General health Yes No Yes Yes Yes Yes ALT General health Yes No Yes Yes Yes Yes GGT General health Yes No Yes Yes Yes Yes TSH General health Yes No Yes Yes Yes Yes hsCRP General health Yes No Yes Yes Yes Yes Calcium General health Yes No Yes Yes Yes Yes Corrected Calcium General health Yes No Yes Yes Yes Yes Albumin General health Yes No Yes Yes Yes Yes PTH General health Yes Yes Yes Yes Yes Yes

Vitamin D 25 OHD levels Yes Yes Yes Yes Yes Yes

Self-collected samples

Urine pregnancy testing (if Yes Yes Yes Yes Yes Yes required) Bone density and muscle strength testing

DXA Bone density testing Yes No No Yes No Yes

2.9.1 Questionnaire data

The primary questionnaire was divided into four modules, which participants were able to complete either altogether or at separate times over a 2-week period, before each site visit.

The questionnaires take around 2 hours to complete and are formulated into 4 modules. The questionnaire links for each module were sent to participants through LimeSurvey

(www.limesurvey.org) which is a password-protected, secure online survey tool, developed by the University of Melbourne (Appendix F). Completion of each module was checked daily, to identify any answers which may have needed any follow-up. Each module starts with the participant information and consent form, followed by different type of questions including multiple choice or free text questions and ends with three questions, how accurate do you think your answers are?, how embarrassing did you find the questions?, and feel free to make any comments. The content of each module and the rationale of collecting these data are presented in detail below.

Module A: Demographic data including age, marital status, income, education level, educational institute, employment, location, ancestry, country of birth, language, religion and being of Aboriginal or Torres Strait Islander descent, how they heard about the study, height and weight, medical conditions, family history of disease, asthma, nasal allergy, skin allergy and medication use. These data were collected to establish health conditions, supplements and medication use which may be associated with 25 OHD levels.

Module B: Beverage intake, food and supplement intake, food exclusion, methods to try to lose or control weight or shape, weight change, snacking, skipping meals, place of eating, food preparation and access and confidence in food preparation. These data were collected to identify changes in body composition and weight, and also to investigate a relationship between obesity and vitamin D.

Module C: Knowledge of food and nutrition, behaviour and attitudes toward eating, trying to change their appearance, how much and how often used alcohol, duration of smoking and number of cigarettes have they smoked in their entire life and during past 30 days, attitude towards smoking and drug use. These data were collected to examine the relationship between smoking, alcohol intake, eating behaviour and vitamin D status in young women.

Cigarette smoking has been associated with vitamin D status in some observational studies, so smoking data were also collected in this module.

Module D: Life events and satisfaction, willingness to do each part of the study including wearing the sun monitoring watch, completing the questionnaires, attending the site visits, effects of their family and friends on improving eating habits, effects of their family and friends on regular exercise, confidence in doing exercise or following diet, barriers in doing exercise or diet adherence, general health and how they would describe their health, physical activity during a typical day, problem due to physical health, problems due to emotional problems, pain, how they feel, mental health, sun exposure, skin type, hair and eye colour, physical activity, sitting, broken bones, back pain, taking time off from work or school, healthcare providers and sport or active recreation injuries. These data were collected to measure dietary intake of vitamin D and physical activity to investigate a relationship between vitamin D and these factors.

Cancer Council Victoria (CCV) food frequency questionnaire:

The CCV food frequency questionnaire (version 2) was used to collect nutritional intake.

This questionnaire was a modified version of the Dietary Questionnaire for Epidemiological

Studies (DQES), which was developed by the Cancer Council Victoria in the 1980s. The validity and reproducibility of the questionnaire were already confirmed [Ireland et al., 1994].

The CCV food frequency questionnaire contains 18 questions about food intake and portion

85 sizes which provided energy, macronutrient and micronutrient and food intake data for each participant over the last 12 months. A specific password, user name and the questionnaire link were generated and emailed to each participant (www.iviewsurveys.com.au). All the results were exported by the CCV team and provided as a spreadsheet.

These data were collected to assess dietary intake as a possible confounder for the relationship between vitamin D and obesity.

2.9.2 Site visit assessment and rationale

All participants were asked to undertake study site visits at baseline, after 4 months of intervention and at the end of the study (12 months), at the Royal Melbourne Hospital,

Parkville, Victoria. The visits took up to 2 hours to complete. A window period of two weeks was allowed for booking the 4 months and 12 months follow-up visits. All visits were scheduled in the morning. Following booking the site visit over the phone, instructions about how long the visit would take, what measurements would be done and how participants could travel to the hospital by car or by public transport were provided. During the phone call participants were asked to fast overnight, not eating or drinking (apart from water), for at least 8 hours before the visit. A study team member met the participant on arrival at the main entrance of the Royal Melbourne Hospital, to guide them to the site visit room and answer any questions. After blood was taken, a light snack and water was provided to participants to break their fast. The physical examination included measurements of height, weight, waist and hip circumferences, blood pressure, skin reflectance and urine test (if pregnancy test required for the DXA scan). Finally, participants were guided to the bone density unit in the

Royal Melbourne Hospital for the whole body DXA scan, lumbar spine scan and hip scan.

A range of blood tests were taken to measure 25 OHD levels, cardiovascular risk factors and other health markers which may have a relationship with 25 OHD levels. Blood pressure and resting heart rate were measured as a general health and cardiovascular risk factor. Waist and hip circumferences, height and weight were measured to calculate body mass index and body size, as obesity and central obesity are CVD risk factors. Skin reflectance was measured for melanin density and skin colour which affects vitamin D production in the skin. Dual Energy

X-ray Absorptiometry (DXA) (QDR 4500A densitometer, Hologic Inc., Bedford, USA) was used for bone density testing, body composition and fat distribution in the body, to investigate the relationship between vitamin D and these measures.

2.9.3 Blood collection

At each visit, 30 mL of fasting bloods were collected from each participant, 20 mL for storage and 10 ml for real time tests for a range of blood analyses. Bloods were collected in 6 different tubes and sent to the Royal Melbourne Hospital pathology laboratory. The laboratory performed blood tests to measure HbA1c, glucose, lipids, insulin, hs-CRP, calcium, corrected calcium, albumin, PTH, and 25 OHD.. After preparation by the laboratory another container of blood was sent to Vivopharm laboratory (Vivopharm laboratory,

Bundoora, Victoria) for further assessment of 25 OHD levels. Vitamin D containers were stored at -40 degree Celsius and sent to Vivopharm in batches through the project. All blood samples were labelled with the participant study number, initials and date of birth.

The RMH Pathology laboratory sent the blood results to our study team by fax and mail.

Some tests were received on the same day of collection and some within 2 weeks of collection. All blood results were reviewed by the study team for any abnormality, and then sent to the study chief investigator (Professor John Wark) for further review and comment on the abnormal results. If any follow-up was required, participants were contacted by the study

87 team and asked to see their usual doctor; a copy of the results were mailed out and emailed to the participant. In this thesis, results of HbA1c, glucose, lipids, insulin, TSH (thyroid stimulating hormone), hs-CRP, calcium, corrected calcium, albumin, and 25 OHD levels are reported.

Sample processing

For real time tests, the blood tubes were mixed well, an aliquot of ~ a 2 mL tube and sent to biochemistry. For storage, the serum and plasma tubes were centrifuged and transferred to specific vials to transfer to the -80C degree freezer.

HbA1c

HbA1c refers to glycated haemoglobin, which indicates the overall average of blood glucose over the preceding 3 months for healthy and diabetic people and is an indicator of diabetes control [Gallagher et al., 2009]. Recently, it has also been reported that HbA1c is associated with cardiovascular disease [Zhao et al., 2014]. The normal range of HbA1c for people without diabetes is between 4 and 5.6%. People with HbA1c between 5.7 and 6.4 % have a higher risk of diabetes and people with HbA1c above 6.5% are considered to have diabetes

[Saudek et al., 2008]. HbA1c was measured in the whole blood collected into an

Ethylenediamine tetraacetic acid (EDTA) tube. HbA1c was measured by a high-performance liquid chromatography method using phenylboronic acid. The major advantage of this technique is the lack of interference between labile HbA1c and other variants of haemoglobin.

Glucose

Testing fasting plasma glucose is a standard method of diagnosis of diabetes. The normal range of fasting plasma glucose recently has been revised and recommended to be between

5.5 to 6.0 nmol/L [WHO, 2017]. In the RMH pathology laboratory, the normal range of glucose is defined as between 3.3 to 7.7 nmol/L. Glucose levels were measured applying

Hexokinase and Glycerol Phosphate Oxidase methods using a reagent kit (Abbott, USA) with a coefficient of variation (CV) of 2.3 %.

Lipids

In our project, lipids were measured in stored serum tubes. In the RMH laboratory, the normal range for cholesterol, HDL, LDL and triglyceride are defined as < 4.0, >1.0, <2.5 and

<1.5nmol/L, respectively. RMH establishes in house reference intervals by running several samples over a number of days and calculating the mean and standard deviation and CV %.

Cholesterol and LDL levels were measured by Abbott ARCHITECT c-Systems utilising an enzymatic method using a reagent kit with 1.48% CV (Abbott, USA). HDL levels were measured directly, using specific detergent with the imprecision of the ultra HDL assay with total CV 3-4%. Triglyceride levels were measured using a specific reagent kit with CV of

2.3%.

Insulin

In our study insulin levels less than 15 mU/L were considered normal. Insulin was assayed using the ARCHITECT, chemiluminescent microparticle immunoassay (CMIA) method with imprecision precision of 3-4% total CV, sensitivity of 1.0 μU/mL and specificity of 10%.

High-Sensitivity C-reactive protein (hs-CRP)

The risk of developing cardiovascular disease is quantified according to the hs-CRP levels as follows: low: hs-CRP level under 1.0 mg/L; average: between 1.0 and 3.0 mg/L; and high: above 3.0 mg/L. The normal concentration in our project is defined as hs-CRP < 5.0 mg/L.

89 hs-CRP was measured using the ARCHITECT c-system with quantitative immunoturbidimetric with precision of <6% and CV of 2.3 to 2.7.

Parathyroid hormone (PTH)

In our project PTH was measured in serum and levels of 1.7 to 7.5 pmol/L considered normal. PTH was measured by CMIA for the quantitative determination of intact parathyroid hormone in human serum with CV of 4.7% at 2.90 pmol/L and 4.6% at 20.0 pmol/L. PTH was measured to recognise those participants with hypoparathyroidism and hyperparathyroidism as these conditions could affects vitamin D metabolism and as vitamin

D deficiency is associated with secondary hyperparathyroidism [Lips et al., 2001].

Calcium and corrected calcium

In our project, total calcium was measured in the serum and corrected calcium was calculated as [total calcium (mmol/L) + 0.02 (40-Albumin level (g/L))]. We measured calcium to recognise persons with hypocalcaemia or hypercalcemia as these conditions could affects vitamin D metabolism and also as vitamin D can affect serum calcium levels [Lee et al.,

2002]. The normal range of calcium was defined as calcium or corrected calcium between

2.10 to 2.60 mmol/L. It was measured by using a specific reagent kit with total CV of 0.83 to

1.2%.

Albumin

Albumin is the most abundant protein in the blood plasma, and carries calcium in the plasma.

To calculate corrected calcium levels, we measured serum albumin levels. The normal range is defined as albumin levels between 35 to 50 g/L. Albumin was measured based on the binding of bromcresol purple specifically with albumin to produce a coloured complex. The

90 absorbance of the complex at 604 nm is directly proportional to the albumin concentration in the sample with total CV of 1.8 to 4.7%.

25 OHD

We measured 25 OHD levels in two laboratories with two different methods at baseline and 4 months follow-up. In the RMH laboratory 25 OHD was measured by using chemiluminescent microparticle immunoassay (CMIA) [Hutchinson et al., 2017] for the quantitative determination of 25-hydroxyvitamin D. We also measured 25 OHD levels by LC-MS/MS method in Vivopharm laboratory as this is the gold standard for measuring 25 OHD levels

[Van den et al., 2013]. This method requires lower sample volume, it is more accurate and precise, and it has specificity to distinguish between mean 25 OH metabolites. Serum 25

OHD3 and serum 25 OHD2 concentrations were measured by a LC-MS/MS method using

Applied Biosystems 4000 Q trap and Agilent LC-MS/MS instruments in VivoPharm laboratories in Bundoora, Victoria, Australia. The intra-assay CV was 5.8% and 4.4% for 25

OHD2 and 25 OHD3 respectively. In the 12 months follow-up, 25 OHD levels measured only by CMIA method were available.

2.9.4 Physical examination

Physical examinations including measurement of blood pressure, heart rate, waist and hip circumferences and height and weight were performed at each visit for general health assessment.

Blood pressure and heart rate

Blood pressure was measured twice by using two different digital blood pressure monitoring devices, A&D CO., LTD, TM-2551, Japan; Omron Healthcare Co., LTD, HEM-7322, Japan.

Blood pressure was recorded as systolic and diastolic blood pressure and is measured after 5

91 minutes of rest, at the upper arm in seated position. If the readings seem unusually high/low the measurement was repeated. According to the Heart Foundation of Australia, blood pressure below 120/80 mmHg is considered normal, between 120/80 mmHg to 140/90 as high normal and more than 140/90 mmHg as hypertension [Lawes et al., 2006]. Blood pressure less than 90/60 mmHg is considered hypotension or low blood pressure.

Waist circumferences

Waist circumference was measured as the abdominal circumference just above the iliac crest, in a standing, straight and relaxed position. Participants were asked to breathe normally and the measurements were done as the participant was breathing out. A flexible normal measuring tape was used and measurement was done to the closest of 0.1 cm. It is reported that waist circumferences more than 102 cm for men and more than 88 cm in women are associated with health problems such as cardiovascular disease [WHO, 2003].

Hip circumferences

Hip circumference was defined as the widest circumference of hip in a standing position and recorded to the closest 0.1 cm by using a flexible tape. Measurements were taken when the participant was breathing out.

Weight

Weight was measured using Model 402KL scales, Continental Scale Corporation Bridgeview

ILL, US. For weight measurements participants were asked to remove their shoes and take off their jacket and any heavy objects in their pockets and then step on to the scale.

Measurements were done with light clothes and recorded to the closest of 0.1 kg.

Height

Height was measured by a wall-mounted stadiometer (Holtain, CRYMYCH, DEFED,

Britain) to the closest 0.1 cm. Participants were asked to remove any hair accessories, take their shoes off and stand steady with feet together and heels against the wall.

2.9.5 Skin reflectance

Skin pigmentation was measured at each visit using a skin reflectance instrument

(Spectrometer, Konica Minolta Optics Inc, CM-2500d), as skin colour is a covariate to be controlled for in assessing vitamin D response. Both a UV-unexposed region (e.g. inner upper arm) and UV-exposed regions (e.g. back left hand and left cheek) were photographed. Three photographs were taken from each region/participant. The instrument provided the skin colour and texture data in the form of melanin index, haemoglobin index, haemoglobin SO2 index (%), and hue with spectral reflectance and colorimetric values in a single operation

(See appendix E).

2.9.6 Pregnancy test

Because DXA involve a low level of ionising radiation exposure, it is recommended not to scan pregnant women. Therefore, a pregnancy test was performed for those participants who were sexually active, not using any kind of hormonal contraception and not between days 1 to day 12 of their menstrual cycle. For a pregnancy test a 15 mL mid-stream urine sample was collected and a pregnancy test kit (Alere Medical Co., Ltd., Japan) was used.

2.9.7 Bone density and body composition scans

Participants attended the Bone Densitometry Unit at the RMH for whole body scan, hip scan and lumbar spine scan. Dual-energy X-ray absorptiometry (DXA; QDR 4500A densitometer,

Hologic Inc., Bedford, USA) method was used to measure bone mineral density and content, as well as whole body and soft tissue composition.

Body composition, including visceral fat and body fat distribution, was also determined by

DXA. T-score, Z-score and bone density of lumbar spine, total hip and femoral neck were also obtained from DXA scanning [Crabtree et al., 2014].

Participants were asked to remove any metallic objects and recline on the DXA machine’s bed, stay steady and breathe normally. All the three scans took around 20 minutes to complete.

To determine the visceral fat area, manual delineation of the whole-body scan was performed. Visceral fat area was determined as the abdominal region from just above the iliac crest to approximately above the second vertebra (L2) and from each side to the muscle of the abdomen [Pritchard et al., 1993].

2.9.8 Sun exposure/SunSmart behaviour

Objective real-time UV exposure was measured in each participant at baseline, 4 months and after 12 months of intervention. UV exposure was measured by using a small, discreet and wearable UV dosimeter, which was worn like a wristwatch for 14 consecutive days before the site visit day.

Participants were also asked to fill out a clothing log and standard questions about sunlight exposure and sun-related behaviour. Questionnaires were used to report clothing worn, sunscreen use, any sunburn, date of starting and finishing wearing the dosimeter and if they took off the sun monitoring watch for any reason.

SunSmart is a promoting health program which is funded by Cancer Council Victoria and

Victoria Health Promotion Foundation to promote a balance between achieving adequate 25

OHD levels and preventing from skin cancer [www.sunsmart.com.au/about/sunsmart- program].

Three weeks before each site visit day the dosimeter, clothing chart (appendix G), sun monitoring log (appendix H) and the instruction about how to wear the dosimeter and how to fill out the questionnaires were posted to the participant’s address, and they were asked to bring the dosimeter back to us at their visit.

Dosimeters were manufactured in Scienterra Ltd, New Zealand and bought from the

Australian Radiation Protection and Nuclear Safety Agency (ARPANSA), Yallambie VIC.

Each dosimeter was configured through the docking cradle, which was connected to a computer through a USB port. After configuration, the dosimeter was removed from the cradle and worn like a wrist watch to record UV exposure. After the exposure period, the dosimeter was returned to the cradle, and data were off-loaded to the computer for processing. Each dosimeter has 2,162,160 bytes of available data storage with 30 second reading intervals. Raw data were analysed to determine the total sun exposure and average sun exposure per day in standard erythema doses (SED) unit.

Each dosimeter was sent to ARPANSA every year for calibration. Calibration was done by comparing the dosimeter data with biometer data which is located in the ARPANSA.

Biometers are instruments using to measure solar UV radiation which are located in the UV monitoring stations across the globe. One of the biometers is in Victoria for continuous monitoring of erythemally weighted solar UV radiation in southern hemisphere [Wood et al.,

2017].

2.9.9 Calculated variables

Body mass index

BMI is a simple index used to indicate whether a person is underweight, overweight, obese or a healthy weight for their height. According to the WHO, having a BMI between 25 and 30 is classified as overweight, and obesity is defined as having a BMI more than 30 kg/m2

[www.who.int/mediacentre/factsheets/fs311/en]. Normal weight is defined as having BMI between 18.5 and 25 kg/m2. BMI is calculated as (weight (kg) / (height (m) 2).

Waist to height ratio

Waist to height ratio or WHtR is an indicator of obesity, which has been shown to be a better indicator of obesity-related disease such as cardiovascular disease than body mass index

[apps.who.int/iris/bitstream/10665/44583/1/]. Waist to height ratio of 0.35 to 0.52 for men and 0.35 to 0.48 for women are considered normal. It is calculated as “waist (cm)/ height

(cm)” [Michelle et al., 2017].

Waist to hip ratio

Waist to hip ratio or WHR is used as a measurement of obesity, which is an indicator of other health conditions. The WHO states that abdominal obesity is defined as a waist to hip ratio above 0.90 for males and above 0.85 for females. It is calculated as “waist circumferences

(cm) / hip circumferences (cm)” [apps.who.int/iris/bitstream/10665/44583/1/].

Homeostatic model assessment of insulin resistance (HOMA-IR)

HOMA-IR is a validated method used to quantify insulin resistance. This method of insulin resistance measurement is non-invasive in nature and cost effective for studies with a large sample size. The normal HOMA-IR values of healthy humans, range are from 1.7 to 2.0.

HOMA-IR is calculated as fasting glucose (mmol/L) × fasting insulin (µU/L)] / 22.5

[Turner et al., 1985].

Homeostatic model assessment of beta cell function (HOMA-B)

HOMA-B is a validated method used to quantify beta cell function. This method of beta cell function assessment is non-invasive in nature and cost effective for studies with a large sample size. HOMA-B is calculated as

20 × fasting insulin (µU/L) / (fasting glucose (mmol/L) − 3.5) [Turner et al., 1985].

Quantitative insulin sensitivity check index (QUICKI)

QUICKI is a useful method of insulin sensitivity measurement which is derived by using the inverse of the sum of the logarithms of the fasting insulin and fasting glucose 1 /

(log (fasting insulin µU/mL) + log (fasting glucose mg/dL) [Katz et al., 2000].

Non-HDL Cholesterol

Recently, non-HDL cholesterol has become a commonly-used marker for a blood lipid pattern associated with increased risk of cardiovascular disease, as the total cholesterol levels cannot distinguish between LDL and HDL cholesterol. It is calculated as “total cholesterol

(nmol/L) – HDL cholesterol (nmol/L)” [Katarzyna et al., 2010].

Total cholesterol to HDL cholesterol ratio

This ratio is also used as a predictor of heart disease. An optimal ratio is less than 3.5 to 1 and a higher ratio means a higher risk of heart disease. It is calculated as: Total cholesterol

(nmol/L) / HDL cholesterol (nmol/L) [Millán et al., 2009].

2.10 Young Female Health Initiative (YFHI) study

For the cross-sectional analysis, we combined our Safe-D Part A data with another study called the Young Female Health Initiative (YFHI) to increase the sample size. The YFHI study is cross-sectional with longitudinal extension, aimed at evaluating the interplay between lifestyle, behavior, and physical and mental health in young Australian women

[Fenner et al., 2012]. YFHI study was approved by the Melbourne Health and the Royal

Women’s Hospital Human Research Ethics Committees. Importantly, Part A of the Safe-D study was based closely on YFHI in design and methodology.

2.11 Confounding

In this project the following variables were considered as confounding factors. Why these variables have been chosen to be controlled and how they were controlled are briefly described. Season, smoking, physical activity, alcohol intake, oral contraceptive pill use, country of birth, multivitamin and vitamin D supplementation, visceral fat and baseline 25

OHD levels were considered as confounding factors. Baseline 25 OHD levels were controlled by group stratification. All other confounding effects were controlled in the statistical analyses. Many studies showed seasonal effects on 25 OHD levels. As the main source of vitamin D is from sunlight, 25 OHD levels tend to be higher after summer time compared to winter [Harris et al., 1998]. It also has been shown that smoking has a significant effect on calcium and vitamin D metabolism [Brot et al., 1999], which is not likely to be explained by other confounding lifestyle factors. The depression of the vitamin D-PTH system seen among smokers may represent another potential mechanism for the effects of smoking on the skeleton, and may contribute to the reported risk of osteoporosis among smokers [Brot et al.,

1999]. Studies also showed that both indoor and outdoor physical activity increases 25 OHD levels [Fernandes et al., 2017] through affecting mental health, body weight reduction,

98 medication intake and quality of life [Fernandes et al., 2017]. Alcohol intake can be associated with 25 OHD levels. However, the direction of the association is not clear as some studies showed positive and others showed negative associations between alcohol intake and vitamin D deficiency [Tardelli et al., 2017]. Studies also showed that oestrogen-containing contraceptive pills increase vitamin D levels [Mayor et al., 2016]]. The mechanism as to how oral contraceptives can increase vitamin D levels is still unknown [Mayor et al., 2016].

Ethnicity and country of birth also have been shown to influence 25 OHD levels, as those born in countries other than Australia are more likely to have lower 25 OHD levels [Gill et al., 2014]. Ethnicity has also been associated with 25 OHD levels and has been linked to time spent outside, clothing and skin colour [Daly et al., 2012]. Multivitamin and vitamin D supplementation influences total vitamin D intake. Higher visceral fat increases the risk of vitamin D deficiency [Zhang et al., 2015] which is explained in details in chapter 6.

2.12 Statistical analysis

Statistical analyses were conducted using the Statistical Package for Social Science version

22 [SPSS Inc., Chicago, Illinois, USA].

In the cross-sectional part, Kolmogorov-Smirnov test and histograms were used to assess the normality of continuous variables. Not normally distributed variables were log-transformed.

Baseline general characteristics were examined using a two-way ANOVA (analysis of variance) or Kruskal-Wallis for continuous variables and Chi-square for categorical variables.

Bonferroni correction was used to counteract the problem of multiple comparisons.

In the intervention part, the main analyses were conducted on an intention-to-treat (ITT) basis to compare the intervention groups against the control group at 4 and 12 months.

Due to the possible protocol violations (eg. non-compliance with the prescribed treatment), secondary per protocol analyses designed to adjust for non-compliance were undertaken at 4 and 12 months and results were compared to the intention to treat analysis. Participants in the control and behavioural intervention groups who started taking vitamin D supplements during the trial and non-compliers with the interventions were excluded from the per protocol analysis.

For intention to treat purposes all participants were included in the analysis with missing at random (MAR) assumption. Missing values were handled based on the multiple imputation method. Multiple imputations is a method which allows the replacement of missing values by a set of variables generated from the posterior predictive distribution of missing data.

Percentages of missing data were different for each variable and were between 5 and 10%.

Non-compliance in the behavioural intervention group was defined as compliance less than

5% and in the supplement group defined as compliance less than 10%.

To determine the effects of supplementation and behavioural intervention over 12 months on all cardiovascular risk factors ANOVA or Kruskal-Wallis tests were used.

A linear regression analysis was used to identify the association between changes in 25 OHD levels and changes in each biomarker after 12 months. P-values less than 0.05 were considered statistically significant. The baseline 25 OHD levels and other potential confounders were adjusted. Baseline 25 OHD levels were adjusted by block stratification and the rest of the confounders were adjusted in the statistical analyses.

Paired t-tests or Wilcoxon sign rank tests were used to compare 12 months 25 OHD levels, physical activity, UV exposure and knowledge about safe sun exposure with the baseline in each group. Also 12-month data were compared with the reference data.

100

The trial was reported in accordance with the CONSORT (consolidated standards of reporting trials) criteria [www.consort-statement.org/].

2.13 Data handling

All electronic data were entered manually and stored in a password-protected database that was located in the Department of Medicine (RMH), University of Melbourne. Only approved members of the Safe-D study team were able to access the study database.

All hard copy data were stored in locked filing cabinets within the study office, located in the

Department of Medicine (Royal Melbourne Hospital).

All biological samples were de-identified (using the participants’ unique study identification number) and stored in locked freezers in the Department of Medicine (Royal Melbourne

Hospital), The University of Melbourne.

As this study was a clinical trial, all data related to the study will be stored for a period of 15 years, in accordance with Melbourne Health research guidelines.

2.14 Compliance

Compliance checks were different between the three groups.

Pharmacological intervention group

Compliance in this group was assessed by manual count of the supplements which remained in the supplement containers at 4 months and 12 months follow-up visits. All participants in this group were asked by the unblinded team member to bring their remaining supplements to their follow-up visits. Participants who forgot to bring their supplements were contacted by phone after the visit.

101

Also, two weeks after randomisation, participants were contacted to be asked if they were taking supplements daily and how they stored the supplement.

Behavioural intervention group

The app contains built-in timers and page access data that allowed the study team to determine how often the participants in the behavioural intervention group accessed the app.

Participants in this group were contacted 2 weeks after the randomisation, to be asked if they used the application daily, if their behaviour towards sun exposure had changed since using app, if they noticed any changes in their skin, if they were receiving push notifications.

Improvement of 25 OHD levels in this group could also indicate compliance of the participants in this group.

Control group

All participants in this group were contacted 2 weeks after randomisation, to see if they received and read the pamphlet. No other compliance measurements were undertaken for the control group.

2.15 Withdrawal

Any participant who failed to meet any of the eligibility criteria at any time during the study was asked to stop following the interventions. However, they were asked and encouraged to complete the study visits for intention to treat analyses. For example, participants who were not in the pharmacological group and started taking more than 800 IU vitamin D supplement per day were withdrawn from the intervention, but asked to attend their site visits.

Non-compliance to the intervention affect ongoing eligibility for participation.

102

Any participant who withdrew their consent at any stage of the study was withdrawn from the study. Any data collected from them continued to be used, unless the participant stated specifically that they wish to have their data removed from our database.

The date and reason for withdrawal were entered on the case report forms and into the study database.

All reasonable attempts including resending emails, text messages and making phone calls and sending a letter to the participant’s home address were made to communicate with the participant. If no contact could be made with the participant for a period of 28 days, the participant was withdrawn from the study and a letter sent to them to inform them of this decision.

2.16 Ethics and legal consideration

This study has been registered with the Australian New Zealand Clinical Trials Registry

(ANZCTR) on 2 September 2013 with trial number ACTRN 12613000972729. The study was approved by the Royal Melbourne Hospital Human Research and Ethics committee

(HREC) with project identification (ID) of 2013.215. The Facebook advertisement also was approved by the Royal Melbourne Hospital HREC. Any minor changes during the study were reported as a form of project amendment to the ethics committee. This trial was conducted according to the principles and rules laid down in the Declaration of Helsinki and its subsequent amendments.

This study was carried out according to the revised National Statement on Ethical Conduct in

Research Involving Humans (2007) produced by the National Health and Medical Research

Council of Australia [www.nhmrc.gov.au/guidelines-publications/e72]. Mandatory reporting is required for physical or sexual abuse and an algorithm of participant management

103 followed. Participants are aware of the obligations of the study team for reporting some of the information. Information regarding illegal drug use may be disclosed to relevant authorities if required by law.

All of the participants’ information was kept confidential and only the study team were able to access the data except if required by law.

All participants were compensated for their time and effort with a $30 Coles/Myer voucher at each visit. Light snacks and water were also provided for all of the participants during their site visits. For the pharmacological group, all the vitamin D supplements were provided free of charge.

All of the possible risks of attending this research project were explained to the participants at the beginning of participation. Minor risks included increasing sun exposure, discomfort, redness, and pain and fainting during blood collection. Some of the survey questions might be disturbing or embarrassing to some. However, they can skip any question which sounds embarrassing or disturbing. Finally, they received a very low dose of ionising radiation during DXA (1-4 microsieverts) and peripheral quantitative computed tomography (pQCT) scans (lower than 0.01 millisieverts) [Wood et al., 2017]; this was below the level at which adverse effects have been detected [Damilakis et al., 2010].

Participants identified by their questionnaire answers as clinically depressed or anxious, or who answered positively to suicidal ideation were contacted in accordance with the Mental

Health Results Protocol. These participants were strongly advised to seek assistance from their general practitioner for follow-up care. They are also sent information booklets

(beyondblue booklet; “What works for anxiety disorders?” [Reavley et al., 2010] and “What works for depression in young people”) [Jorm et al., 2009].

104

Any participants who were identified as suicide positive according to their LimeSurvey questionnaire responses were contacted by an ASSIST (Applied Suicide Intervention Skills

Training)-trained Safe-D team member [www.livingworks.com.au/]. If participants were contacted regarding their suicide positive responses in any other previous site visits and deemed to be safe and under the care of a mental health expert, they were not contacted again in relation to the suicide positive response, unless requesting further assistance. Data regarding mental health and suicide are not presented in this thesis.

All participants were advised that they could withdraw completely from the study project at any time if they needed or they decided not to complete any specific part of the project.

2.17 Clinically-significant results

Participants whose serum 25 OHD levels showed moderate-severe deficiency (<25nmol/L) were contacted by the study team and strongly advised, using a standard algorithm developed by Prof John Wark to obtain standard treatment for this deficiency from their usual treating doctor. Subjects whose levels dropped below 25nmol/L at 4 months were also immediately referred for appropriate treatment; however, they remained in the trial for the purposes of intention-to-treat analyses.

All of the abnormal blood results and abnormal bone scan results were reviewed by a clinician and then sent to the participant for further follow-up with their usual doctor.

2.18 Adverse events

An adverse event is defined as any occurrence that has unfavourable or unintended effects on participants, regardless of severity or being associated with the study. Adverse events may manifest as new findings (signs, symptoms, diagnoses, laboratory results) or alterations in

105 pre-existing conditions. If a participant became distressed, counselling (or alternative appropriate support) were offered.

All adverse events occurring during the study were reported to the senior investigator (JDW) and recorded whether or not they were considered to be serious and/or related to the study.

Documentation of all adverse events was recorded in the study database. Following an adverse event, participants were appropriately counselled. Chief investigators were responsible for overseeing communications with participants regarding any adverse event.

106

Chapter 3

RESULTS

107

Numbers of participants recruited in each part of the project are presented in section 3.1.

Sample size re-calculation and power calculation are discussed in detail in section 3.2, numbers of unblinding occurring during the project are presented in section 3.3, protocol deviation and the percentage of compliance in each group are presented in section 3.4 and 3.5

, respectively. Numbers of adverse events and severe adverse events which occurred during the intervention in each group are presented in section 3.6. General characteristics of participants are presented in details in section 3.7. We also included data from another cross- sectional study (YFHI study) in Part A data. General characteristics of participants in YFHI and Part A are compared in this section. General characteristics of Part A participants and those who were recruited into Part B are compared in the following section. In section 3.8 the effects of vitamin D supplementation and behavioural intervention on 25 OHD levels after 4 months and after 12 month of intervention are presented.

3.1 Numbers recruited in Part A and Part B

Recruitment in the cross-sectional part (Part A) of the project started in April 2014 and finished in November 2015. During this time 824 expressions of interests received through the Safe-D study web page. Participants reported that they have heard about our project through Facebook advertisement (n=397), media (n=20), friends (n=132), Twitter (n=8), friend’s Facebook wall (n=58), Safe-D website (n=9), Safe-D Facebook study page (n=18),

YFHI (Young Female Health Initiative) website (n=5) or from our study team member when they were participating in previous studies (n=153). From 824 expression of interest, 135 were lost to follow-up, 113 declined to participate, 9 had incorrect contact details and 10 entered their details into the website twice. In total, 557 participants verbally consented and

463 completed the survey questionnaires and 407 participants completed their site visit

[Callegari et al., 2015].

108

Two hundred and seventy two participants were potentially eligible for the clinical trial part

(Part B) of the project according to their 25 OHD levels and were contacted for further eligibility assessment. Out of the 272 subjects, 141 were recruited into Part B, 25 were lost to follow-up, 34 declined to participate and 72 were not eligible to participate into the clinical trial part of the project after second screen for eligibility (two did not have smartphone, 20 were taking vitamin D supplements, 24 were moving outside of the Australia during the following year, six had low 25 OHD after second measurements of 25 OHD levels in

Vivopharm laboratory, 12 had a family history of melanoma and 8 had a medical condition).

Out of the 141 recruited participants five were lost to follow-up and 11 withdrew from the study so 123 were randomised into the three groups (n=41 in each group). During the four month follow-up 27 participants were withdrawn from the study (4 start taking supplement, six were busy and not able to attend visit, 11 were unable to contacted, five give no reason, 1 had surgery) and 96 completed their 4 month follow-up visit. In total, 84 participants completed their one year follow-up visit, and 12 were withdrawn (7 busy and not able to attend visit, 4 failed to contact, 1 moved from Victoria, Australia) [Tabesh et al., 2016]

(Figure 3.1).

For the cross-sectional analysis, data from 150 participants from another observational study

(YFHI study) were combined with our Safe-D Part A data (n=407).

3.2 Sample size re-calculation and power calculation

Statistically calculated sample size for Part A was 468 participants, and for Part B was 234

(n=78 in each group)with a requirement for, 186 participants completed the four month follow-up visit and 144 completed the final visit. Due to slow recruitment, power analysis was undertaken on a total of 60 participants at 4 months follow-up visit to see whether the sample size provided sufficient power (85%) to detect 15 nmol/L differences in 25 OHD

109 levels between the three groups. The power calculation revealed that 123 recruited participant in Part B would provide enough power to detect 15 nmol/L 25 OHD level differences among the three groups. As a result, recruitment stopped in Part A with 407 participant and Part B with 123 recruited participants (Table 3.1). This smaller sample size provided enough power for statistical analysis; however, this could be considered as one of our project’s limitations.

During the sample size re-calculation, investigator blinding was maintained.

Table 3.1 Number of calculated sample size and actual sample size

Part of the study Sample size Actual number calculated recruited

Cross-sectional 468 407

Intervention 234 123

4 months follow-up 186 96

12 months follow-up 144 84

110

Participants potentially eligible from Part A (Assessed for eligibility)

N=272  Lost to follow-up N= 25

 Declined (not interested) N=34

 Not eligible N=72  Not having smartphone N=2  Taking vitamin D supplements N=20

 Moving outside of Australia N=24  25 OHD levels N=6  Family history of Melanoma N=12  Medical condition N=8

Recruited N=141

Lost to follow-up N= 5

Withdrew N=11

Randomised N=123

Withdrew N= 27

(4 start taking supplement, 6 busy and not able to attend visit, 11 failed to contact, 5 no reason, 1 had surgery) Completed 4 month follow-up N=96

Withdrew N= 12 (7 busy and not able to attend visit, 4 failed to contact, 1 moved from Victoria) Completed 12 month follow-up N=84

Figure 3.1 Recruitment flowchart

111

3.3 Unblinding

Totally three unblinding events happened during the data collection. Two unblindings events happened during the site visits and blood taking, another happened during booking the site visit. Whenever any unblinding events happened, another study team member who was still blinded continued the data collection.

3.4 Protocol deviation

Visits were booked within one week before or one week after the exact due date of 4 months or 12 months from the baseline. However, for some of the participants a delay in follow-up occurred due to being unable to contact the participants in timely fashion or participants were not able to attend during that time period. Ten participants had a delay in completing their 4 months follow-up visit with a minimum of 18 and maximum 90 days’ delay. Twelve participants had a delay in completing their final visit with a minimum of 16 and a maximum

95 days’ delay.

During baseline visits 2 participants were not fasted for their blood test. At the four months follow-up visit 2 were not fasted and in the final visit 4 participants were not fasted. Non- fasted participants were excluded from the metabolic profile analysis.

3.5 Compliance rate

Compliance with the vitamin D supplementation was assessed by counting the remaining capsules. All participants in this group were asked to bring their remaining capsules to the follow-up visits. An unblinded team member weighed the remaining capsules and calculated the number of capsules that were used according to the following equation.

Number of capsule used = Total numbers of capsules provided – (weight of remaining capsules/weight of one capsule). Compliance rate was calculated by dividing the total number

112 of capsules used by the total number of days. In the 4 months follow-up visit compliance in the supplement group ranged from 64.2% to 100%. Overall compliance at the four months follow-up for this group was 91.4%. For the 12 months follow-up visit, minimum, maximum and average compliance for this group was 34.7%, 100% and 76.8%, respectively.

Compliance in the behavioural intervention group was measured by assessing the number of days participants open the app. Application timer provided a page report in the server that allow the study team to determine how often the participants opened the app. In the initial 4 months of follow-up the app usage ranged between 1 and 94 days. Compliance rate in the app group was calculated by dividing the number of days app opened by the number of days the participant was in the study. The compliance range in the app group was between 0.83% and

78.3%. Overall compliance was 22.3% in the app group. In the 12 months follow-up minimum, maximum and average compliance were 0.27%, 55.6% and 12.8%, respectively.

In addition to the app report, all participants completed a questionnaire and were asked at each follow-up if they were using the app daily and whether the app usage lead to behavioural changes. According to these subjective data at the 4 months follow-up visit, five participants (12.1%) declared that they were using the app daily and 21 (51.2%) declared that their behaviour had changed because of using the app. In the 12 months follow-up the number of daily users was 1 (2.4%) and the number of participants with change in their behavioural was 18 (43.9%).

At the 2 weeks follow-up phone call, all the participants in the control group were asked if they have received and read the brochure sent to them (Table 3.2)

[www.sunsmart.com.au/downloads/vitamin-d/how-much-sun-is-enough-vitamin-d.pdf].

113

Table 3.2: Compliance rate in each time point of study

Compliance rate

4 months 12 months

App 22.3% 12.8%

Supplement 91.4% 76.8%

3.6 Number of adverse events

All adverse events occurred during the study were reported and recorded whether or not they were considered to be serious and/or related to the study. Following an adverse event, participants were appropriately counselled. During the 4 months follow-up, in the control group there were two adverse events: one of abdominal pain and one syncope. In the supplement group, two adverse events occurred one of menorrhagia and one anaphylaxis. In the behavioural intervention group, there was one participant with increased anxiety/depression. All adverse events were grade one except the anaphylaxis and syncope which were grade 3 according to the National Institute of Health (NIH) grading system, version 5 [National Institutes of Health. Common Terminology Criteria for Adverse Events

(CTCAE), 2018]. In addition there were 11 individuals with sunburn during the 4 months follow-up (5 in the control group, 3 in the supplement group and 3 in the behavioural intervention group).

During the 4 months to 12 months follow-up visit, 12 adverse events occurred. In the control group there were two adverse events: one of worsening of allergic reaction and one wound infection. In the supplement group, five adverse events occurred: one tonsillitis, one worsening of allergic reaction, one flu, one peri acetabular osteotomy and one anaphylaxis

114 episode. In the behavioural intervention group, there were five adverse events: one hip pain, one broken hand, two gastrointestinal pain and one wisdom teeth removal. In addition there were four individuals with sunburn during that period (1 in the control group and 3 in the behavioural intervention group). All were reviewed by Professor John Wark and reported to be not related or very unlikely to be related to the study treatment.

3.7 General characteristics of participants

For the cross-sectional analysis, data were also included from a similar population study, the

YFHI study, a study in which the interplay between lifestyle, behaviour, and physical and mental health in young Australian women is examined [Fenner et al., 2012].

Data from 557 women aged 16 to 25 years old were analysed in the cross-sectional part (407 of participants were from Safe-D Part A and 150 from the YFHI study). Mean age, BMI, grams of alcohol intake and physical activity were 22.16±2.54 years, 23.90±4.99 kg/m2,

7.55±6.61 gram, and 1329.3±1286.3 Mets (metabolic equivalent), respectively. Most of participants (71.6%) had annual income less than $26,000 which is lower than the average annual income in Australia [www.abs.gov.au/ausstats/[email protected]/mf/6523.0.]. Totally, 472 of the 557 participants (84.7%) were born in Australia, others were born in a variety of countries including the US, New Zealand, China, Sweden, South Africa. Only 10 participants (1.8%) were living in a farming area at the time of participation in the study. Only 85 participants

(15.2%) had less than 12 years education. Totally 76 participants (13.6%) were in the lowest quartile of the socioeconomic index which is determined by SEIFA (socio-economic indexes for areas) index. Socioeconomic indexes for areas or SEIFA is developed by the Australian

Bureau of Statistics (ABS) in Australia to rank areas according to the relative socioeconomic advantage and disadvantage [www.abs.gov.au/ausstats].Oral contraceptive pills were using by 38.6 percent of participants in our study, which is similar to the percentage reported by

115

ABS (33.6%) of women using an oral contraceptive pill in Australia

[www.abs.gov.au/AUSSTATS]. A total of 72 participants (12.9%) were current smokers.

3.7.1 Comparing general characteristics of participants in Safe-D study with YFHI study

The general characteristics and socio-demographics factors of Safe-D and YFHI participants are presented in Table 3.3.

Independent t-test for continuous variables and Fisher’s exact test for categorical variables were used to compare general characteristics of participants between YFHI and Safe-D study.

As shown in Table 3.3 no significant differences were observed in terms of annual income, country of birth, living on farm, age, BMI, and being current smokers between Safe-D and

YFHI study participants. However, the Safe-D study participants were less educated than

YFHI participants which could be related to the fact that Safe-D participants were younger than YFHI participants (p < 0.001). The socioeconomic index in Safe-D participants was significantly lower than YFHI participants (p = 0.03). The percentage of participants using an oral contraceptive pill was significantly higher among Safe-D participants than YFHI participants (p < 0.001). However, alcohol intake in YFHI participants was higher than Safe-

D (p < 0.001). Safe-D participant are significantly more active than YFHI participants (p =

0.004). As most of the general characteristics of participants were not significantly different between the two studies we combined the data from both studies to increase the sample size.

116

Table 3.3 General characteristics and socio-demographic factors of Safe-D and YFHI studies 1 (n=557)

Variable Total number 2 YFHI Safe-D P-value 3

n=557 n= 150 n= 407

Annual income, n (%)

<26000 dollar 399 (71.6%) 107 (71.3%) 293 (72.0%) 0.86

Born in Australia, n (%)

Yes 472 (84.7%) 122 (81.3%) 343 (84.3%) 0.30

Living on farm, n (%)

Yes 10 (1.8%) 2 (1.3%) 10 (2.4%) 0.20

Education, n (%)

Socio-economics index, n (%) 4

Lowest quartile 76 (13.6%) 16 (10.6%) 63 (15.5%) 0.03

Age (years) 22.16±2.54 22.26±2.05 22.09±2.83 0.38

BMI (kg/m2) 23.90±4.99 23.53±4.65 24.08±5.15 0.19

Using oral contraception, n (%)

Yes 215 (38.6%) 39 (26.0%) 176 (43.2%) <0.001

Current smoker, n (%)

Yes 72 (12.9%) 23 (15.3%) 46 (11.3%) 0.07

Alcohol intake (gram) 7.55±6.61 9.79±11.33 6.64±2.65 <0.001

Physical activity (Mets) 1329.3±1286.3 1161.9±1139.6 1437.2±1362.2 0.004

1 Data are number (percent) or mean (SD)

117

2 Numbers may not add up to total due to missing data 3 P-value obtained from the independent t-test for continues variables and Fisher’s exact test for categorical variables 4 Determined by Socio-Economic Indexes for Areas (SEIFA) 2011 developed by the Australian Bureau of Statistics. Ranks areas in Australia according to relative socioeconomic advantage and disadvantage based on five-yearly census data.

118

3.7.2 Comparing general characteristics of participants in Safe-D Part A (cross- sectional part) with Safe-D Part B baseline (intervention part)

The general characteristics and socio-demographics factors of participants who attended in the cross-sectional part of the Safe-D study and those randomised into the clinical trial part of the Safe-D project are presented in Table 3.4. No significant differences were observed between the participants who attended the cross-sectional part of the project and those who attended the clinical trial part in terms of annual income, country of birth, living on farm, education, socio-economics status, age, BMI, oral contraceptive use, smoking status, alcohol intake and physical activity.

General characteristics of participants who were randomised to the Par-B of the project are also presented in Table 3. Around 66 % of the participants in that part had an annual income of less than $ 26,000 , 86.2% were born in Australia, and only one person was living in a farm area (0.8%). Most of the participants had more than year 12 school education (89.4%),

17 (13.8%) were in the lowest quartile of socio-economics status. Mean and standard deviation of age, BMI, alcohol intake and physical activity were 22.28±2.56 years,

25.01±5.73 kg/m2, 7.30±7.50 gram and 1416.80±1399.14 Mets, respectively. Out of the 123 participants 64 (39.3%) were using an oral contraceptive pill and 12 participants (9.9%) were current smokers (Table 3.4).

119

Table 3.4 General characteristics and socio-demographic factors of Safe-D Part A

(n=407) and Part B (n=123) 1

Variable Part A Part B baseline P-value 3

(n= 407) (n= 123)

Annual income, n (%)

<26000 dollar 333 (71.9%) 77 (65.9%) 0.28

Born in Australia, n (%)

Yes 391 (84.4%) 106 (86.2%) 0.77

Living on farm, n (%)

Yes 11 (2.4%) 122 (99.2%) 0.47

Education, n (%)

Socio-economics index, n (%) 4

Lowest quartile 72 (15.6%) 17 (13.8%) 0.67

Age (years) 22.09±2.83 22.28±2.56 0.49

BMI (kg/m2) 24.08±5.15 25.01±5.73 0.08

Using oral contraception, n (%)

Yes 176 (43.2%) 46 (39.3%) 0.46

120

Current smoker, n (%)

Yes 52 (11.4%) 12 (9.9%) 0.74

Alcohol intake (gram) 6.64±2.65 7.30±7.50 0.15

Physical activity (Mets) 1437.2±1362.9 1416.8±1399.1 0.88

1 Data are number (percent) or mean (SD) 2 Numbers may not add up to total due to missing data 3 P-value obtained from the independent t-test (2-tailed, equal variance assumed) for continues variables and Fisher’s exact test (2-sided) for categorical variables 4 Determined by Socio-Economic Indexes for Areas (SEIFA) 2011 developed by the Australian Bureau of Statistics. Ranks areas in Australia according to relative socioeconomic advantage and disadvantage based on five-yearly census data.

121

3.8 Effects of vitamin D supplementation and behavioural intervention on 25 OHD levels

In this section, the effects of vitamin D supplementation and behavioural intervention on 25

OHD levels over 4 month and 12 month of intervention are provided.

At baseline there were no significant differences in 25 OHD levels among the three groups (p

= 0.78). Over 4 months and 12 months intervention significant differences were observed in

25 OHD levels among the three groups (p < 0.001). Over the 4 months follow-up 25 OHD

levels were significantly higher in the supplement group compared to the control and

behavioural groups. Serum 25 OHD levels were also significantly higher in the behavioural

intervention group when compared to the control group. Same results were observed over 12

months of follow-up.

Supplementation with vitamin D and behavioural intervention both resulted in a significant

improvement of 25 OHD levels after 4 months and after 12 months of intervention. However,

these changes were not statistically significant in the control group.

At 4 months follow-up in the supplement group 54% and in the behavioural intervention

group 40% reached the optimal vitamin D levels. At 12 months follow-up in the supplement

group 53% and in the behavioural intervention group 14% reached the optimal vitamin D

levels. In the control group 26% and 9% reached the optimal levels of 25 OHD after 4 month

and 12 month follow-up, respectively (Table 3.5).

Table 3.5 Percentage of participants reached the optimal 25 OHD levels (>75 nmol/L)

4 months follow-up 12 months follow-up Behavioural group 16/40 (40%) 4/30 (14%) Supplement group 22/41 (54%) 20/38 (53%) Control group 11/42 (26%) 3/34 (9%)

122

Chapter 4

ASSOCIATIONS BETWEEN VITAMIN D STATUS, ADIPOSITY AND LIPID PROFILES IN YOUNG WOMEN (CROSS-SECTIONAL PART)

123

124

125

Associations between 25-hydroxyvitamin D levels, body composition and metabolic profiles in young women

Marjan Tabesh 1, Emma T. Callegari1, Alexandra Gorelik1, 2, Suzanne M. Garland3-5, Alison Nankervis1,6, Asvini K. Subasinghe 3,4, and John D. Wark1,7, on behalf of the YFHI and Safe- D study groups

1 University of Melbourne, Department of Medicine, Royal Melbourne Hospital, Parkville VIC 3050 2Institute for Health and Aging Australian Catholic University, Melbourne, VIC, Australia 3Infection and Immunity Theme, Murdoch Childrens Research Institute, VIC, Australia 4Women’s Centre for Infectious Diseases, Royal Women’s Hospital, Melbourne, VIC, Australia 5 University of Melbourne, Department of Obstetrics and Gynaecology, VIC, Australia 6Diabetes Service, Royal Women's Hospital, Parkville, VIC, Australia 7Bone and Mineral Medicine, Royal Melbourne Hospital, Parkville, VIC, Australia

Running title: Vitamin D, obesity and metabolic profiles in young women

Keywords: 25-hydroxyvitamin D, Obesity, Body composition, Metabolic profiles, Young women

Corresponding author: John D Wark, Professor of Medicine University of Melbourne Department of Medicine (Royal Melbourne Hospital) Parkville, Victoria 3052 Australia Phone: (+61) 8344 3258 Fax: (+61) 9348 2254 E mail: [email protected]

Declarations: The Safe-D study (Part B) has received in-kind support from Swisse Wellness.

Swisse Wellness did not play a role in study design, the implementation of these studies, nor the interpretation of the findings.

Registration: This study is registered in Australian New Zealand Clinical Trials Registry

(ANZCTR) with registration number of ACTRN12617000927325.

Funding: This study was supported by two grants from NHMRC (National Health and

Medical Research Council). Grant number #1049065 for the Safe-D study.

126

Author contributions:

The study was conceived and designed mainly by JD Wark and A Gorelik, with contributions by the other authors. JD Wark supervised the study. M Tabesh and ET Callegari contributed to the data collection. All authors contributed to data interpretation, and drafting of the manuscript. M Tabesh conducted the statistical analysis with supervision by A Gorelik and

AK Subasinghe. All authors approved the final manuscript for submission.

127

4.1 Abstract

Background: Cardiovascular disease (CVD) is a major cause of mortality and morbidity globally. Results from previous studies are inconsistent and it remains unclear whether low serum 25 OHD levels are associated with an increased risk of CVD. These associations have been little studied in young women.

Objective: The aim of this study was to assess the relationship between serum 25 OHD and obesity, body composition, metabolic profiles and blood pressure in young women.

Design: Women aged 16-25 years living in Victoria, Australia, were recruited through

Facebook advertising in this cross-sectional study. Participants completed an online survey and attended a site visit in a fasted state, where parameters including blood pressure, anthropometry, metabolic profiles, serum 25 OHD levels, and body composition (using dual energy X ray absorptiometry) were measured.

Results: A total of 557 participants were recruited into this study. Multiple linear regression analysis showed that after adjusting for visceral fat, season, smoking, physical activity, age, alcohol intake, oral contraceptive use, country of birth, taking multivitamins and taking vitamin D supplement, a 10 nmol/L higher 25 OHD levels was associated with 0.65% greater

HDL levels (p = 0.016) and 0.92% greater triglyceride levels (p = 0.003). It was also associated with 0.48% lower BMI (p < 0.001), 0.50% lower total fat percentage (p <0 .001),

0.09% lower visceral fat percentage (p < 0.001), 0.14% lower visceral fat to total fat ratio (p

< 0.001) and 0.36% lower trunk fat to total fat ratio (p < 0.001), after adjustment for season, smoking, physical activity, age, alcohol intake, oral contraceptive use, country of birth, taking multivitamins and taking vitamin D supplements. Although these associations were statistically-significant, they were very small in magnitude and of uncertain clinical significance.

128

Conclusions: These findings may help to explain an association between 25 OHD levels and

CVD risk factors through associations with HDL, BMI, total body and visceral fat mass.

Possible underlying mechanisms warrant further investigation.

295 words

129

4.2 Introduction

Cardiovascular disease (CVD) is one of the main causes of mortality around the world

[WHO, 2014]. More than 17.5 million deaths resulted from CVD in 2012 and this number is estimated to increase to over 23.6 million per year by 2030 [WHO, 2014]. Recently, much effort has been expended to determine modifiable CVD risk factors including hypertension, obesity, hyperlipidaemia, sedentary lifestyle, smoking, alcohol consumption and unhealthy diets, to prevent the development of CVD. In particular, there has been an increasing dialogue on the impact that vitamin D and obesity may have on CVD progression.

It has been shown that low 25 hydroxyvitamin D (25 OHD) levels are associated with adverse serum lipid profiles, which could potentially explain the association with CVD and mortality [Kelishadi et al., 2014]. Data from most cross-sectional studies have shown that high serum 25 OHD is associated with higher high-density lipoprotein cholesterol (HDL-C) and lower low-density lipoprotein cholesterol (LDL-C), or total cholesterol to HDL-C ratio

[Jorde et al., 2010]. Conversely, findings from clinical trials are inconsistent, with some indicating a positive effect and some a negative effect of vitamin D supplementation on these parameters. However, none of the intervention studies were specifically designed to evaluate the relationship between 25 OHD levels and lipids. In one study, conducted in 2009 with 200 participants, a significant effect was found with an 8% increase (0.28 mmol/L) in serum LDL cholesterol and a 16% decrease (0.22 mmol/L) in serum TG in subjects given 83 mcg per day vitamin D supplements, compared with those given placebo (p < 0.001) for 12 months

[Zittermann et al., 2009].

However, in other clinical trials this effect was not statistically significant. A double-blind randomised clinical trial in 251 healthy adults in Norway showed that 10 mcg per day or 25 mcg per day vitamin D supplementation for a period of 16 weeks had no effect on lipid profiles and body mass index (BMI) [Madar et al., 2014]. Another study of 173 Pakistani

130 adults found that one year of supplementation with 10 or 20 mcg per day vitamin D had no effect on any lipids [Andersen et al., 2009].

There is evidence that the distribution of body fat also is a risk factor for disease, including

CVD and diabetes [Wu et al., 2001]. The main function of subcutaneous fat is energy storage.

However, visceral fat which surrounds internal organs has a variety of clinical effects, including affecting lipid profiles and fasting glucose levels [Wu et al., 2001]. Previous studies have shown that adiposity may affect vitamin D metabolism and action [Hypponen et al.,

2006]. Hence, efforts to achieve and maintain healthy 25 OHD levels could mitigate CVD risk.

Previous literature has shown associations between increasing BMI and lower serum 25 OHD concentrations, but the direction of the association has been inconsistent and causality has been uncertain. Moreover, this relationship has not been adequately studied in young women.

Therefore, a better understanding of the relationship between vitamin D, adiposity and CVD is of high importance. This is particularly important for young women during their childbearing years when vitamin D deficiency can affect both mother and her unborn child, and as women might respond to CVD disease differently, have different symptoms and need different diagnosis or treatment than men. This is particularly important in young women, in whom we have shown that a range of adverse lifestyle factors are prevalent including increasing body weight and decreasing physical activity [Christie et al., 2013].

Finding associations between 25 OHD levels and HDL, and measures of body fat in this study could be important as 25 OHD levels potentially can be improved by changes in lifestyle and the use of vitamin D supplementation, and such interventions could reduce risk of many diseases including CVD. Ultimately, such knowledge could lead to improved health outcomes for women. Therefore, the aim of this study was to evaluate the associations

131 between 25 OHD levels and obesity/body composition including fat distribution/metabolic profiles and blood pressure in young women.

4.3 Materials and methods

Study design and participants: Details of the study design have been described previously

[Callegari et al., 2015]. The Safe-D study comprised two parts, Part A which was a cross- sectional study and Part B which was a clinical trial conducted at the Royal Melbourne

Hospital, Victoria, Australia, from 2012 to November 2016 [Tabesh et al., 2016]. In this current paper, data from Part A of the Safe-D study are presented [Christie et al., 2013].

Participants were also recruited from a similar population, the Young Female Health

Initiative (YFHI), a study in which the interplay between lifestyle, behaviour, and physical and mental health in young Australian women was examined [Fenner et al., 2012].

Inclusion criteria: Participants were included in the study if they were female, aged 16-25 years and living in Victoria. Participants were excluded if they were pregnant, breastfeeding or planning to move from Australia within two months of verbal consent.

Ethics and consent: The Safe-D study was approved by the Melbourne Health Human

Research Ethics Committee and the YFHI study was approved by the Melbourne Health and the Royal Women’s Hospital Human Research Ethics Committees. Each participant provided verbal and written informed consent. Participants completed a comprehensive online questionnaire, were asked to wear an ultraviolet-B (UV-B) dosimeter and to attend a study site visit where metabolic profiles, cardiovascular risk factors, anthropometry, body composition [using dual energy X-ray absorptiometry (DXA)] and 25 OHD levels were measured.

132

Survey data: Physical activity was assessed using a modified Active Australia Survey

[Australian Institute of Health and Welfare et al., 2003] and was calculated from time spent each day at specific activity levels, weighted according to the estimated oxygen consumption

(MET) required for each activity [Ainsworth et al., 2000]. Questionnaires including the physical activity survey were sent to participants approximately 2 weeks before their site visit and asked about their activity in the previous week from survey. Daily energy and calcium intake were obtained from the Cancer Council Victoria (CCV) dietary food frequency questionnaire [Ireland et al., 1994]. Established questionnaires on LimeSurvey software and

Survey Monkey® [www.limesurvey.org] were used to collect demographic data, medical history, use of health care services, contraceptive use, medication use, dietary supplementation, vitamin D intake and smoking status.

Anthropometric measures: Height, weight, waist and hip circumference were recorded during the site visit. Height, measured by a wall-mounted stadiometer (Holtain, CRYMYCH,

DEFED, United Kingdom) to the closest 0.1 cm, and weight, measured to the closest 0.1 kg using Model 402KL scales (Continental Scale Corporation Bridgeview ILL, USA), were used to calculate body mass index (BMI) [weight (kg)/(height (m))2]. Waist circumference was measured as the narrowest abdominal circumference between the last rib and the top of the iliac crest in a standing and relaxed position. Hip circumference was defined as the widest circumference at the hips in a standing position.

Body composition, including abdominal visceral fat mass, total fat mass and body fat percentage, was determined by DXA (QDR 4500A densitometer, Hologic Inc., Bedford,

USA). To determine visceral fat area, manual delineation of the whole body scan was performed. Visceral fat area was determined as the abdominal region from just above the iliac crest to above the second lumbar vertebra (L2) and from each side to the muscle of the abdominal [Pritchard et al., 1988]. Visceral fat percentage was calculated as visceral fat mass

133 as a percentage of total body fat mass. Total fat percentage was calculated as total body fat as a percentage of total body weight. Visceral fat to total fat ratio was calculated by dividing the visceral fat mass by the total fat mass. Blood pressure and heart rate were measured twice with two different blood pressure monitors (A&D CO., LTD, TM-2551, Japan; Omron

Healthcare Co., LTD, HEM-7322, Japan) at the upper arm in seated position after 5 minutes of rest, and recorded as systolic and diastolic blood pressures [Callegari et al., 2015].

Blood analyses: Morning blood samples were obtained from each participant after a minimum of 8 hours overnight fasting for measurement of 25 OHD levels, lipid profiles, glucose, insulin, and HbA1c. Serum 25 OHD3 and serum 25 OHD2 concentrations were measured by a liquid chromatography-tandem mass spectrometry (LC-MS/MS) method using

Applied Biosystems 4000 Q trap and Agilent LC-MS/MS instruments at VivoPharm laboratories (Bundoora, Victoria, Australia) [Van den Ouweland et al., 2013]. The intra-assay coefficients of variation (CV) were 5.8% and 4.4% for 25 OHD2 and 25 OHD3, respectively.

Total 25 OHD levels (the sum of 25 OHD3 and 25 OHD2) were used in statistical analyses.

Cholesterol and LDL levels were measured by Abbott ARCHITECT c-Systems utilising an enzymatic method by a reagent kit with 1.5% CV. HDL levels were measured directly, using specific detergent, with total CV 3.2%. Insulin was assayed using the ARCHITECT, chemiluminescent microparticle immunoassay (CMIA) method with imprecision of 5.6% total CV, sensitivity of 1.0 μU/mL and specificity of 10%. Glucose and triglyceride levels were measured applying Hexokinase and Glycerol Phosphate Oxidase methods using a reagent kit with CV of 2.3 and 2.3%, respectively. HbA1c was measured by a high- performance liquid chromatography method using phenylboronic acid [Davis et al., 1978].

Definitions: Cardiovascular risk factors were defined as: central obesity (waist circumference

>80 cm), hypertension (systolic blood pressure of 140 or higher or diastolic pressure of 90 or higher) a history of diabetes, haemoglobin A1c (HbA1c) >6.5%, low HDL cholesterol levels

134

(<1.2 mmol/L), current smoking and a history of CVD [Mendis et al., 2011]. Homeostatic model assessment of insulin resistance (HOMA-IR), homeostatic model assessment of beta cell function (HOMA-B) and QUICKI were calculated according to standard equations

[Turner et al., 1985 and Katz et al., 2000].

Statistical analysis: All analyses were performed using the Statistical Package for Social

Science version 22 (SPSS Inc., Chicago, Illinois, USA). To calculate the sample size for the study it was necessary that the sample size for the cross-sectional component (Part A) of the project provide sufficient power for Part B (clinical trial) recruitment. Therefore, it was estimated that 468 participants were required for the cross-sectional part of the study

[Callegari et al., 2015 and Tabesh et al., 2016]. Kolmogorov-Smirnov test and histograms were used to assess the normality of continuous variables. Study participants were stratified by 25 OHD levels and association between general characteristics of study participants and

25 OHD levels were assessed using Fisher’s exact test for categorical and Kruskal-Wallis for continuous variables. Multiple linear regression analyses were used to evaluate the association of 25 OHD levels with lipid profiles, glucose, insulin, HbA1c, body composition and anthropometric measurements with adjustment for potential confounders. Variables not normally distributed, e.g. triglyceride, HDL and BMI were log transformed and analysed as such. All other continuous variables including 25 OHD levels were normally distributed.

4.4 Results

In total, 557 women (407 from Safe-D study and 150 from YFHI study) aged 16 to 25 years, living in Victoria, Australia, were recruited via Facebook advertisements. No significant differences were found between the Safe-D and YFHI participants in terms of weight, family history of chronic diseases and 25 OHD levels, so data from these two studies were combined

135 for analysis. Twenty-one participants reported that they had not fasted for their blood tests and therefore were excluded from the metabolic profile analysis.

General characteristics of all participants (overall and stratified by vitamin D levels) are presented in Table 4.1. Medians (IQR) of serum 25 OHD levels, BMI and energy intake of all participants were 64 (86−46) nmol/L, 22.9 (25.6−20.9) kg/m2 and 6227.78 (4933−7772) kJ/day. Mean (± SD) age of participants was 22 ± 3 years. Mean (± SD) vitamin D intake was

1.55±2.09 mcg per day. In total, 67 participants (16.7%) took a multivitamin and 35 (8.7%) used vitamin D supplements during last week. Most participants were born in Australia

(85%); 10.1% were current smokers and median (Q3-Q1) alcohol intake was 3.8 (10.13-0.82) grams per day. Thirteen participants (2.3%) had 25 OHD levels <25 nmol/L, 327 (58.7%) participants had 25 OHD levels between 25 and 75 nmol/L and 217 (39%) participants had

25 OHD >75 nmol/L (Table 4.1). Participants with 25 OHD <25 nmol/L were more likely to have their blood taken during winter compared to the other two groups (p < 0.001).

Participants with 25 OHD >75 nmol/L were less obese (p < 0.001) and more of them were born in Australia (p < 0.001). Alcohol intake in those with 25 OHD > 75 nmol/L was significantly higher than the two other groups (p = 0.016). There were no significant differences in age, energy intake, smoking status or physical activity among participants with different 25 OHD levels. Participants with 25 OHD levels <25 nmol/L had significantly higher total fat percent (p = 0.003), visceral fat percent (p < 0.001), visceral fat to total fat ratio (p = 0.046), trunk fat to total fat ratio (p < 0.001) and BMI (p = 0.025), and lower HDL levels (p = 0.012). Moreover, participants with 25 OHD levels more than 75 nmol/L had significantly higher triglyceride levels (p = 0.003, Table 4.2). Although a negative trend was observed between 25 OHD and HOMA-IR, the association between 25 OHD levels and

HOMA-IR was not statistically significant (p = 0.068). Moreover, a positive trend between

25 OHD and total cholesterol levels was observed which approached significance (p =

136

0.057). No significant differences were observed between the three groups in terms of other metabolic profiles (glucose, insulin, HbA1c, HOMA-B, QUICKI, LDL, non-HDL cholesterol, HDL to total cholesterol ratio), blood pressure and WHR (data not shown).

The associations between 25 OHD and HDL (non-transformed B coefficient = 0.0242 per 10 nmol/L increase in 25 OHD, log-transformed B coefficient = 0.0065 mmol/L per 10 nmol/L increase in 25 OHD, p = 0.004) and triglyceride levels (non-transformed B coefficient =

0.0304 mmol/L per 10 nmol/L increase in 25 OHD, log-transformed B coefficient = 0.0092 mmol/L per 10 nmol/L increase in 25 OHD, p = 0.003) were significant after adjusting for visceral fat, season, age, smoking, physical activity, oral contraceptive use, alcohol intake, country of birth, multivitamin use and vitamin D supplement use (Table 4.3). When visceral fat was removed from the model, the association of 25 OHD and HDL was no longer significant (p = 0.076). However, the association of 25 OHD and TG remained statistically- significant (p = 0.002). Negative associations between 25 OHD levels and BMI (non- transformed B coefficient = -0.31541 kg/m2 per 10 nmol/L increase in 25 OHD; log- transformed B coefficient = 0.0048 kg/m2 per 10 nmol/L increase in 25 OHD, p < 0.001), and total fat percentage (B=-0.5098 % per 10 nmol/L increase in 25 OHD, p < 0.001), visceral fat percentage (B=-0.0939 % per 10 nmol/L increase in 25 OHD, p < 0.001), visceral fat to total fat ratio (B=-0.0014 per 10 nmol/L increase in 25 OHD, p < 0.001) and trunk fat to total fat ratio remained significant (B=-0.0036 per 10 nmol/L increase in 25 OHD, p < 0.001), after adjustment for confounding variables.

4.5 Discussion

In this study, we found statistically-significant, negative associations between 25 OHD levels and BMI, total fat percentage, visceral fat percentage, trunk fat to total fat ratio, visceral fat to total fat ratio and positive associations between 25 OHD and triglyceride levels, after

137 adjustment for potential confounders, in young women. We have examined these relationships in a young female cohort, as women might differ in CVD risk factors, respond to CVD disease differently, have different symptoms and need different diagnosis or treatment than men. Moreover, it has been suggested that there is an age-related difference in vitamin D metabolism [Forsythe et al., 2012]. These relationships have not been comprehensively examined in young women until now [Appelman et al., 2015].

Findings from our study are in line with previous studies in other populations, which have found negative associations between 25 OHD and BMI and body fat mass [Cheng et al.,

2010]. In 2015, Kao et al. in a cross-sectional study on 229 obese or overweight children aged 3-18 years reported an inverse association between 25 OHD levels and adiposity [Kao et al., 2015]. Parikh et al. also reported an inverse relationship between 25 OHD levels and BMI and a negative association with body fat mass in healthy adult subjects (r=-0.4, r=-0.41, p <

0.0001) [Parikh et al., 2004]. Evidence from the British Birth Cohort study, which was conducted on around 17,000 men and women aged 45 years old, confirmed the association between low 25 OHD and obesity [Hypponen et al., 2006]. Adipose tissue can affect vitamin

D levels by reducing bioavailability of 25 OHD due to sequestration into adipose tissue and also increase metabolic clearance of vitamin D, possibly with enhanced uptake by adipose tissue [Wortsman et al., 2000]. However, Piccolo and colleagues, who examined subcutaneous white adipose tissue 25 OHD concentrations in a weight loss intervention trial, suggested that sequestration of 25 OHD in subcutaneous adipose tissue may be insufficient to explain reduced circulating 25 OHD levels in overweight/obesity [Piccolo et al., 2013].

Another study in this area evaluated the association of serum 25 OHD and vitamin D concentration in subcutaneous fat in 19 obese individuals. They have shown that serum 25

OHD levels and vitamin D content in subcutaneous fat tissue are positively associated

(r=0.68, p = 0.003) [Blum et al., 2008]. The strengths of these associations reported by other

138 studies were stronger than the significant associations we found in this study. This could be because we adjusted for potential confounders more comprehensively than previous studies.

In this study, alcohol intake in those with 25 OHD > 75 nmol/L was significantly higher than the two other groups. However, alcohol intake in all groups was in the light to moderate range which is unlikely to affect cardiovascular risk factors [Rimm et al., 1999].

No significant differences in age were observed among the three categories of 25 OHD levels

(<25 nmol/L, 25-75 nmol/L, >75 nmol/L). Results from regression analysis also showed no significant association between age and 25 OHD levels in our study (B= -0.082, p = 0.053).

This null association could be due to the narrow age range studied.

Although BMI is a good reflection of whole body fat, it is not a direct measure of body composition and may not be as applicable to younger populations who generally have a higher lean mass than older adults [Maynard et al., 2001]. Moreover, it is important to recognise that even people with normal weight can have excess visceral fat, increasing their risk for many chronic diseases including CVD [Wang et al., 2005]. Previous evidence suggests that serum 25 OHD levels are inversely correlated with percent fat in adolescents and healthy women [Lenders et al., 2009]. This is similar to the inverse association we found between 25 OHD and total fat percentage and visceral fat percentage (p < 0.001). Although waist circumference is considered an anthropometric index of central obesity, it cannot reflect the fat content of the abdominal area and can be influenced by stomach content, gastritis, etc.

It also cannot distinguish between visceral fat and subcutaneous fat.

Several mechanisms may explain the association between 25 OHD levels and adiposity.

Increased adipose tissue may increase the distribution of vitamin D, thereby reducing serum

25 OHD levels [Wortsman et al., 2000]. Moreover, low levels of serum 25 OHD can lead to an increase in lipogenesis and decrease in lipolysis by increasing secretion of parathyroid hormone (PTH) or by suppressing the peroxisome proliferator activated receptor γ (PPAR γ)

139

[Lu et al., 2012]. Moreover, outdoor activity and sun light exposure are limited in obese people as they tend to be inside more and cover up more when they are outside, which can cause vitamin D deficiency [Florez et al., 2007].

In our study, no significant associations were found between 25 OHD levels and glucose, insulin, HOMA-B or QUICKI. Based on our findings, it seems that 25 OHD may not be associated with glucose metabolism in healthy young participants, as most of the previous studies were not on healthy subjects. Kayanyil et al. showed an independent association of 25

OHD with insulin resistance and beta cell function in 712 subjects at risk of T2DM (β = -

0.003, p = 0.007; β = 0.004, p = 0.028, respectively) [Kayaniyil et al., 2010]. However, in line with our study, Marques-Vidal et al. in a prospective population-based study on 3856 healthy adults aged 51.2 ± 10.4 y showed that 25 OHD3 concentrations were not associated with insulin resistance and glucose metabolism [Marques-Vidal et al., 2015]. Moreover, findings from clinical trials are not consistent. In 2013, Belenchia et al. demonstrated that after 6 months of vitamin D supplementation, there were significant reductions in fasting insulin levels and improvement in HOMA-IR and QUICKI, whereas there was no significant change in glucose levels in 35 obese, adolescent patients [Belenchia et al., 2013]. A clinical trial in 227 healthy adults aged over 20 years of age in 2012 found an influence of age on the response to vitamin D supplementation. The above mentioned study showed that with every 5 kg/m2 increase in BMI, 25 OHD levels decreased by 6.5 nmol/l in participants aged over 40 years old; in contrast, no such relationship was seen between 25 OHD levels and BMI in younger adults (aged 20 to 40 years old) [Forsythe et al., 2012]. There is likely to be an age- related difference in lipid profiles, fat distribution and response to vitamin D supplementation. Therefore, further research is needed to evaluate the associations of 25

OHD levels and glucose metabolism, insulin resistance and beta cell function in young

140 populations and also to investigate the effects of improving 25 OHD levels on these parameters.

In our study, a significant positive association was found between 25 OHD levels and HDL and triglyceride levels in the crude model and after adjustment for visceral fat, season, smoking, physical activity, age, alcohol intake, oral contraceptive, country of birth, multivitamin use and vitamin D supplement use. However, the association between 25 OHD and HDL was no longer significant when visceral fat was removed from the model. This finding suggests that the association of 25 OHD and HDL could be mediated through affecting visceral fat mass. It could also be due to the strong association of visceral fat with

HDL levels (r = -0.222, p < 0.001) or smoking (r = -0.180, p < 0.001). Although the association of 25 OHD with TG was statistically significant, it was a very small effect

(difference was only 0.1 mmol/L) and of uncertain biological significance. No significant association was found between 25 OHD and other lipids. A positive trend between 25 OHD and total cholesterol levels was observed which approached significance (p = 0.057). These marginally-significant results could be due to a low number in the 25 OHD <25 nmol/L group (n=13). In a Norwegian study of 10,105 participants conducted in 2008, total cholesterol, HDL and LDL were significantly higher, and serum triglycerides were lower, with decreasing 25 OHD levels across quartiles [Jorde et al., 2010]. However, data from a serial clinical laboratory cohort study on 107,811 patients in 2009 - 2011, revealed that although there was an association between low vitamin D levels and unfavourable lipid profiles, replenishing vitamin D may not necessarily improve lipid profiles [Ponda et al.,

2012]. Williams et al, studying American adolescents (12-19 years old) showed that there was no significant association between 25 OHD and triglyceride levels [Williams et al.,

2011]. A potential explanation for this finding could be the effects of vitamin D on the expression of genes involved in bile acid synthesis [Gonzalez et al., 2014]. These inconsistent

141 results could be due to age, sex and ethnicity differences. Also, most of the studies were not specifically designed to evaluate the association of 25 OHD and lipid profiles. More studies of young and healthy populations are needed to confirm the association of 25 OHD levels and lipid profiles in this demographic.

Prior findings on the association of serum 25 OHD and systolic or diastolic blood pressure are inconsistent. Some researchers have reported a stronger inverse association between 25

OHD levels and systolic blood pressure in participants aged more than 50 years, than in younger individuals (p = 0.02) [Scragg et al., 2007], whereas others have reported no difference [Caro et al., 2012]. In our current study, no association was found between 25

OHD levels and systolic or diastolic blood pressure.

Strengths of the present study include the evaluation of an under studied demographic, young adult women. Another strength was the measurement of 25 OHD levels using the LC-MS/MS method, as well as measurements of both serum 25 OHD2 and 25 OHD3 levels. LC-MS/MS method has the highest sensitivity and selectivity and is currently the most accurate method of 25 OHD measurements [Freeman et al., 2015]. We also used HOMA-IR and HOMA-B, two well-validated methods, to measure insulin resistance and beta-cell function. In addition, this study comprised a sample size of women which provided adequate power to perform multivariable analysis. The extensive collection of questionnaires and biodata also allowed adjustment for many potential confounding factors and relevant covariates. Moreover, this sample is broadly representative of this age group in Victoria [Fenner et al., 2012]. However, participants in this study were more educated and had higher BMI than the general population in Victoria [Fenner et al., 2012]. Finally, direct measurement of adiposity by using DXA in addition to BMI or skin fold measurements to estimate adiposity is another strength of this study.

142

Limitations include the cross-sectional design of the study, which does not allow the assessment of a cause–effect relationship of 25 OHD levels and cardiovascular risk factors.

We were also not able to directly measure insulin resistance and beta cell function, because of their invasive nature and costs in studies with a large sample size. Moreover, blood pressure was only measured twice and not three times as per the World Health Organization guidelines and we have not used 24-h blood pressure monitoring. Another limitation was the low number of participants in the 25 OHD <25 nmol/L group (n=13), which reduced the study’s potential to observe associations with severe vitamin D deficiency. Finally, the results of our study may not be generalizable to men, other ethnic groups or other age groups given that all participants were women, aged 16 to 25 years and most of them were white. In Part B of the

Safe-D study, a randomised clinical trial, investigators are evaluating the effects of improving

25 OHD levels by vitamin D supplementation or by a behavioural intervention to safely increase sun exposure, compared to standard care of vitamin D deficiency.

In conclusion, we demonstrated that the serum concentration of 25 OHD was inversely associated with BMI, total fat mass, trunk fat, and visceral fat and was positively associated with TG levels in young women. However, no significant associations with other lipids, glycaemic profiles or anthropometric measurements were observed. Therefore, prospective studies including clinical trials need to be designed to further investigate the effects of 25

OHD on CVD risk factors in this demographic.

143

Table 4.1: Baseline general characteristics of participants stratified by 25 OHD levels a

25 OHD p-value b

n All <25 nmol/L 25-75 nmol/L >75 nmol/L (n=557) (n=13) (n=327) (n=217)

Age (years) 557 22.11 ± 2.62 22.08 ± 2.81 22.29 ± 2.52 21.84 ± 2.7 0.157 Blood drawn in winter; n (%) 557 231 (41.5) 7 (53.8)* 152 (46.5) 72 (33.2) <0.001

Energy intake (kJ/day) 553 6227 (4933−7772) 6592 (5641−8488) 6268 (5016−7883) 6060 (4880−7470) 0.346 BMI category c; n (%) 557 Underweight 39 (7.1) 2 (15.4) 23 (7.2) 14 (6.5) <0.001 Normal weight 341 (62.2) 4 (30.8) 188 (58.6) 149 (69.7) Overweight 111 (20.3) 3 (23) 69 (21.4) 39 (18.3) Obese 57 (10.4) 4 (30.8) 41 (12.8) 12 (5.5)* Current smoker; n (%) 548 56 (10.1) 1 (7.7) 37 (11.3) 18 (8.3) 0.485

Physical activity (METs) 527 1050 (480−1890) 1320 (720−1560) 1077 (532−1950) 1065 (540−1980) 0.632 Born in Australia; n (%) 551 475 (85.3) 6 (46.2) 268 (82) 201 (92.6)* <0.001

Alcohol intake; (gram/day) 546 3.8 (0.8−10.1) 0.7 (0.0−2.6) 3.7 (0.8−10.0) 4.5 (1.0−10.8)* 0.016

a Data are mean ± SD or median (interquartile range) or number (percent). b p-value is the differences between groups obtained from Fisher’s exact test for categorical and Kruskal-Wallis for continues variables,

144 c Underweight defined as BMI less than 18.5, normal weight defined as BMI 18.5 to 24.9, overweight defined as BMI 25 to 29.9, obese defined as having BMI of 30 kg/m2 or greater *Significantly different compared to other two 25 OHD categories, p < 0.05 considered as statistically significant Abbreviations: 25 OHD = 25 hydroxyvitamin D.

145

Table 4.2: Summary of 25 OHD levels, metabolic profiles, blood pressure, anthropometric measurements and fat distribution by 25 OHD levels a 25 OHD

N <25 nmol/L 25-75 nmol/L >75 nmol/L p-value b

(n=13) (n=327) (n=217)

Metabolic profiles

Glucose (mmol/L) 536 4.6 (4.3−4.9) 4.6 (4.4−4.9) 4.6 (4.4−4.9) 0.451

Insulin (mU/L) 530 9.8 (7.1−16.0) 8.3 (6.1−10.9) 7.8 (5.4−9.7) 0.092

HbA1c (mmol/mol Hb) 530 33.0 (30.0−36.0) 32.0 (30.0−34.0) 32.0 (30.0−34.0) 0.960

HOMA-IR 530 1.9 (1.5−3.2) 1.7 (1.2−2.3) 1.5 (1.1−2.0) 0.068

HOMA-B 530 185.5 (98.9−315.7) 144.2 (106.7−200.5) 140 (110.0−192.3) 0.551

QUICKI 530 1.6 (1.4−1.8) 1.7 (1.6−1.9) 1.7 (1.6−2.0) 0.116

TG (mmol/L) 536 0.8 (0.7−1.3) 0.8 (0.6−1.1) 0.9 (0.7−1.2)* 0.003

LDL (mmol/L) 536 2.7 (2.1−3.0) 2.6 (2.1−3.1) 2.6 (2.2−3.2) 0.690

HDL (mmol/L) 536 1.4 (1.2−1.4)* 1.5 (1.3−1.7) 1.6 (1.3−1.8) 0.012

TC (mmol/L) 536 4. 5 (3.8−4.9) 4.5 (3.9−5.2) 4.7 (4.2−5.3) 0.057

Non-HDL cholesterol 536 3.1 (2.5−3.6) 3.0 (2.5−3.5) 3.1 (2.6−3.7) 0.958 (mmol/L)

146

HDL-C/total-C 536 0.30 (0.24−0.38) 0.33 (0.27−0.38) 0.34 (0.27−0.39) 0.636 Blood pressure

SBP (mmHg) 552 112.0 (97.5−119.5) 111.0 (103.0−119.0) 111.5 (105.5−119.5) 0.860

DBP (mmHg) 552 75.0 (64.0−80.0) 72.0 (65.0−76.7) 72.0 (65.0−77.0) 0.734 Anthropometric measurements WHR 546 0.81 ± 0.06 0.79 ± 0.06 0.78 ± 0.07 0.122

BMI (kg/m2) 551 25.4 (20.9−32.1)* 23.1 (21.0−26.5) 22.5 (20.4−24.8) 0.025 Fat distribution Total fat percent (%) 539 36.80 ± 7.32* 31.92 ± 6.73 30.67 ± 5.93 0.003 Visceral fat percent (%) 539 3.29 ± 0.92* 2.72 ± 0.99 2.40 ± 0.86 <0.001 Visceral fat to total fat 539 0.10 ± 0.01* 0.08 ± 0.02 0.08 ± 0.02 0.046 ratio Trunk fat to total fat 539 0.39 ± 0.05* 0.37 ± 0.04 0.35 ± 0.04 <0.001 ratio

a Data are mean ± SD or median (interquartile range). b p-value is the differences between groups by ANOVA or Kruskal Wallis, *Significantly different compared to other two 25 OHD categories, p < 0.05 considered as statistically significant Abbreviations: 25 OHD: 25 hydroxyvitamin D; DBP: Diastolic blood pressure; HbA1c: haemoglobin A1c; HDL: high density lipoprotein; HOMA-B: homeostatic model assessment of beta cell function; HOMA-IR: homeostatic model assessment of insulin resistance; TC: total cholesterol TG: triglyceride; LDL: low density lipoprotein; QUICKI: Quantitative insulin sensitivity check index; SBP: systolic blood pressure; WHR: waist to hip ratio; BMI: body mass index

147

Table 4.3: Association between 25 OHD levels and metabolic profiles, anthropometric measurements and fat distribution in young women

25 OHD

B a 95% CI p-value b

Metabolic profiles

Log TG Crude c 0.00111 0.00058, 0.00163 <0.001 Model 1 d 0.00097 0.00038, 0.00156 0.002

Model 2 e 0.00092 0.00031, 0.00148 0.003

TG Crude 0.00190 0.00006, 0.00373 Model 1 0.00208 0.00012, 0.00403 Model 2 0.00304 0.00105, 0.00503

Log HDL Crude 0.00062 0.00035, 0.00089 <0.001 Model 1 0.00089 -0.00002, 0.00048 0.076 Model 2 0.00065 0.00007, 0.00060 0.004

HDL Crude 0.00256 0.00155, 0.00358 Model 1 0.00318 -0.00182, 0.00457 Model 2 0.00242 0.00105, 0.00380

Anthropometric measurements

Log BMI Crude -0.00045 -0.00069, -0.00021 <0.001 Model 1 -0.00048 -0.00085, -0.00011 <0.001

BMI Crude -0.029558 -0.04598, -0.01312 Model 1 -0.031541 -0.05437, -0.00871

148

Fat distribution Total fat percentage Crude -0.03799 -0.06160, -0.01439 0.002 Model 1 -0.05098 -0.07886, -0.02311 <0.001

Visceral fat percentage Crude -0.05940 -0.08752, -0.03128 <0.001 Model 1 -0.00939 -0.01365, -0.00514 <0.001

Visceral fat/total fat Crude -0.00011 -0.00018,- 0.00004 0.001 ratio Model 1 -0.00014 -0.00023, -0.00005 <0.001

Trunk fat/total fat ratio Crude -0.00038 -0.00053, -0.00023 <0.001 Model 1 -0.00036 -0.00056, -0.00019 <0.001

a 25 OHD considered as predictor, per every 1 nmol/L increase in 25 OHD b Bonferroni adjusted p-value obtained from linear regression analysis and p < 0.005 considered as statistically significant c Crude: Unadjusted d Model 1: Adjusted for season, smoking, physical activity, age, alcohol intake, oral contraceptive use, country of birth, taking multivitamins and taking vitamin D supplements e Model 2: Adjusted for visceral fat, season, smoking, physical activity, age, alcohol intake, oral contraceptive use, country of birth, taking multivitamins and taking vitamin D supplements Abbreviations: 25 OHD: 25 hydroxy-vitamin D; Log TG: log transformed triglyceride; Log HDL: log transformed high density lipoprotein; Log BMI: log transformed body mass index

149

Chapter 5

THE EFFECTS OF VITAMIN D SUPPLEMENTATION AND BEHAVIOURAL INTERVENTION ON OBESITY AND METABOLIC PROFILES OVER 4 MONTHS AND ONE YEAR FOLLOW-UPS (RANDOMISED CLINICAL TRIAL PART)

150

151

152

5.1 Improving vitamin D status by behavioural & pharmacological interventions; effects on cardiovascular disease risk factors

Marjan Tabesh1, MSc, Alexandra Gorelik1,2, MSc; Suzanne M. Garland3-5, MD; Alison Nankervis1,6, MD; Emma T. Callegari1 BBiomed (Hons), and John D. Wark1,7, PhD; on behalf of the YFHI and Safe-D study groups

1The University of Melbourne, Department of Medicine, Royal Melbourne Hospital, Parkville, VIC, Australia 2Institute for Health and Aging Australian Catholic University, Melbourne, VIC, Australia 3Murdoch Childrens Research Institute, Royal Children’s Hospital, Parkville, VIC, Australia 4Women’s Centre for Infectious Diseases, Royal Women’s Hospital, Melbourne, VIC, Australia 5The University of Melbourne, Department of Obstetrics and Gynaecology, Parkville, VIC, Australia 6Diabetes Service, Royal Women's Hospital, Parkville, VIC, Australia 7Bone and Mineral Medicine, Royal Melbourne Hospital, Parkville, VIC, Australia

Corresponding author: John D Wark, Professor of Medicine The University of Melbourne, Department of Medicine Royal Melbourne Hospital Parkville, Victoria, 3050 Australia Phone: (+61) 8344 3258 Fax: (+61) 9347 1863 E mail: [email protected]

Short title: Improving vitamin D; effects on CVD risk factors (49 characters)

Registration: This study has been registered with the Australian New Zealand Clinical Trials

Registry (ANZCTR) on 2nd September 2013 with trial number ACTRN 12613000972729

Funding: This project was funded by a National Health and Medical Research Council

(NHMRC) project grant, APP1049065.

153

ABSTRACT

Context: Cardiovascular disease (CVD) is a primary cause of death globally. Low 25 hydroxyvitamin D (25 OHD) levels have been reported to be associated with increased CVD risk.

Objective: The aim was to examine the effects of a behavioural (using a mobile application to increase safe sun exposure) and a pharmacological (1000 IU/day vitamin D supplementation) intervention on vitamin D status and their impact on CVD risk factors.

Methods: This was a 3-arm, block-stratified, open-label, single-blind, single-site, parallel- design, randomized-controlled-trial with healthy women aged 16 to 25 years. Participants were randomized into one of three groups (pharmacological, mobile-based intervention or control). Data were collected at baseline and after four months, using online questionnaires and study site visits.

Results: After four months, both vitamin D supplementation and the use of a mobile-based application resulted in a significant increase in seasonally-adjusted 25 OHD levels (mean change 30±28 nmol/L, p < 0.001 and 11±22 nmol/L, p = 0.031, respectively). Serum 25 OHD levels did not change significantly in the control group (p = 0.270). Vitamin D supplementation resulted in a 1.4% reduction in body mass index (BMI) (p = 0.031), whilst the behavioural intervention resulted in a 4.4% reduction in haemoglobin A1c (HbA1c), both compared to baseline (p = 0.001). There were no other significant differences in CVD risk factors between the three groups after four months of intervention.

Conclusion: Both interventions resulted in significant increases in seasonally-adjusted 25

OHD levels after four months. Vitamin D supplementation resulted in a significant reduction in BMI and behavioural intervention resulted in reduction in HbA1c, compared to baseline.

154

250 words

KEY WORDS: Vitamin D supplementation, behavioural intervention, m-health, cardiovascular disease risk factors

Précis

Improving vitamin D levels through using m-health and pharmacological interventions for four months had significant but small effects on cardiovascular disease risk factors on young healthy women.

Disclosure Summary: The Safe-D study (Part B) has received in-kind support from Swisse

Wellness. Swisse Wellness did not play a role in study design, the implementation of this study, nor the interpretation of the findings.

155

5.1.1 Introduction

The global burden of cardiovascular disease (CVD) has increased during the last few years

[Bansilal et al., 2015]. According to the Australian Bureau of Statistics, around fifty thousand deaths were attributable to CVD in Australia in 2013 [Australian Bureau of Statistics, 2016 and AIHW]. Recent evidence suggests that low vitamin D levels may contribute to an increased risk of CVD by affecting obesity, lipid profiles, glucose metabolism and inflammatory biomarkers [Wang et al., 2008]. Nevertheless, it remains unclear whether low

25 hydroxyvitamin D (25 OHD) levels are a causal risk factor for CVD [Wimalawansa et al.,

2018].

Most previous observational studies have suggested an association between vitamin D deficiency and an increase in CVD risk [Kelly et al., 2011 and Giovannucci et al., 2008]. A cross-sectional study of 85 participants aged 4 to 18 years showed a significant association between low 25 OHD levels and higher body mass index (BMI), higher fasting glucose, insulin levels and HOMA-IR (homeostatic model assessment of insulin resistance) [Kelly et al., 2011]. A prospective cohort study of more than 18,000 adults found that the incidence of cardiovascular events in those with low 25 OHD levels (<37 nmol/L) was 2.1 times higher than those with optimal 25 OHD levels (>75 nmol/L) [Giovannucci et al., 2008]. A systematic review and meta-analysis of 25 observational studies reported that the association between vitamin D deficiency and CVD risk factors in children was inconclusive. They reported that current evidence on vitamin D deficiency and CVD in children and young adults was very limited and well-designed studies in this area were scant [Murni et al., 2016]. Most of the earlier trials assessed fracture risk and physical performance, rather than CVD risk and most were conducted in older populations. Results from clinical trials are inconsistent and most could not confirm the effects of vitamin D improvement on CVD risk factors. The

Women’s Health Initiative calcium and vitamin D trial, involving 36,282 postmenopausal

156 women, found no benefits of daily 400 IU (international unit) vitamin D3 with 1000 mg/day calcium supplementation on cardiovascular events [Hsia et al., 2007]. Another double-blind, randomized controlled trial of 52 participants aged 18 to 50 years old evaluated the effect of

7000 IU/day vitamin D3 supplementation on CVD risk factors, compared to a placebo group.

The authors concluded that supplementation did not affect HOMA-IR, blood pressure, plasma lipids or inflammatory biomarkers [Wamberg et al., 2013]. Thus, findings from previous studies are inconsistent; and the effects of vitamin D improvement on CVD risk factors have not been adequately investigated, particularly among young, healthy women.

Mobile-Health (m-health), which is the use of mobile computing and communication technologies in health care and public health, is a rapidly-expanding area of research [Free et al., 2010]. Most of the published studies using m-health for lifestyle interventions have focussed on cigarette smoking cessation [Whittaker et al., 2009], or increasing physical activity [Klasnja et al., 2009]. Using m-Health to improve vitamin D status could be beneficial as it is easy to use, cost-effective and user-friendly for the majority. Studies which have used m-health or any other lifestyle interventions to improve vitamin D status are very limited. Moreover, most of the earlier studies used only vitamin D supplementation to improve vitamin D levels [Goodman et al., 2016].

The Safe-D study aimed to evaluate the effects of improving vitamin D status, through a behavioural intervention or vitamin D supplementation, on CVD risk factors including obesity, glucose metabolism, lipid profiles, blood pressure and inflammatory biomarkers following four months of intervention. Recruitment via Facebook advertising and subsequent enrolment into an intervention trial is novel in this study.

5.1.2 Material and Methods

157

Study Design: The Safe-D study comprised two parts, Part A which was a cross-sectional study [Callegari et al., 2015], and Part B which was a single-blinded randomized controlled trial [Tabesh et al., 2016]; it was conducted at the Royal Melbourne Hospital, Victoria,

Australia. In this paper, data from Part B of the study are presented. The trial was a phase IV clinical trial and registered in the Australian New Zealand Clinical Trials Registry

(ANZCTR) with trial number ACTRN 12613000972729. The study was approved by the

Melbourne Health Human Research and Ethics Committee (HREC 2013.215). The trial was conducted according to the principles and rules laid down in the Declaration of Helsinki

[World Medical Association et al., 2013]. The details of the study protocol for both parts of the study were described elsewhere [Callegari et al., 2015 and Tabesh et al., 2016].

Participants: Facebook advertising was used to recruit participants into the cross-sectional part (Part A) of the study. Part A study recruitment began in April 2014 and ceased in

November 2015. The overall number of participants required for Part A was determined by the sample size required for the randomized controlled trial part (Part B), and adjusted for the expected attrition rates and selection criteria. Based on the sample size calculation, 234 participants (78 per arm) at baseline were needed to provide adequate numbers at four months of follow-up, to give 85% power to detect a difference of 15 nmol/L in 25 OHD levels between groups, with a standard deviation (SD) of 25 nmol/L and total level of significance of 0.05. However, study power was reassessed mid-study (after 123 participants were recruited and 60 had completed their four month follow-up visit) and differences of more than 15 nmol/L in 25 OHD levels were observed among the three groups after four months of intervention. Hence, recruitment was stopped in November 2015 with a total of 123 participants enrolled. Inclusion criteria for the trial were: female sex, age between 16 and 25 years old, residing in Victoria, Australia for the duration of the trial and able to attend study site visits, serum 25 OHD levels between 25 and 75 nmol/L, in addition to ownership and

158 regular use of a smartphone. Exclusion criteria for entry to the trial included having a history of melanoma or having a first-degree relative (parent or sibling) who had a melanoma, current pregnancy or breastfeeding or planning to conceive in the next four months, supplementing with more than 800 IU vitamin D daily, planning to move outside Australia during the course of the study, having any chronic disease or medication with safety concerns or whose use may disturb vitamin D metabolism, or increased sensitivity to sun light or ultraviolet radiation (UVR). All participants provided verbal and written consent. Participants were randomized into one of the three study arms in a 1:1:1 ratio using stratified block randomization, based on baseline 25 OHD levels (stratum 1: 25-50 nmol/L; stratum 2: 50-75 nmol/L). A two-step randomization process was used to minimise selection bias during the randomization. The randomization schedule was created by the study statistician [AG] using group codes, with the key known to an unblinded investigator who was responsible for participants’ recruitment, dealt with intervention problems and assessed intervention compliance at study visits. All other investigators and study team members, including the statistician and pathology laboratory technicians, were blinded to the participants’ group allocation. Participants were randomly allocated into one of the following groups 1) pharmacological intervention, receiving 1000 IU per day vitamin D3 supplement; 2) behavioural intervention, using a mobile-based application designed to obtain safe and effective sun exposure [Heffernan et al., 2014]; and 3) controls receiving general advice in the form of the “How much sun is enough?” brochure produced by Cancer Council Victoria

[How much sun is enough, 2016]. Vitamin D supplements were open-label and provided by

Swisse Wellness Company. Compliance with the vitamin D supplementation was assessed by counting the remaining capsules on return of the supplements as (number of capsule used / number of days in the study) × 100 after four months of intervention.

159

Compliance in the behavioural intervention group was calculated as (number of days application opened / number of days in the study) × 100 after four months of intervention.

Compliance was not assessed for the control group.

Data Collection: Demographic, behavioural and clinical data, including concomitant medications and vitamin D intake were collected using an online LimeSurvey questionnaire

[LimeSurvey project team, 2016]. Energy intake was obtained using the Cancer Council

Victoria food frequency questionnaire [Ireland et al., 1994]. Physical activity was assessed by using a modified version of the Active Australia Survey and was expressed as metabolic equivalent (METs) in hours per day [Ainsworth et al., 2000]. UV exposure was assessed by wearing a portable UV dosimeter (Scienterra Ltd, Otago, New Zealand) for 14 consecutive days preceding each site visit. Participants also completed a clothing chart and log of when they wore the dosimeter and if they had experienced sunburn throughout this period. Sun exposure and skin types were also assessed by using validated questionnaires [Fitzpatrick et al., 1988 and Glanz et al., 2008]. To obtain anthropometric and blood pressure measurements, all participants attended a morning 2-hour site visit at the Royal Melbourne Hospital. Weight was measured to the closest 0.1 kg using Model 402KL scales (Continental Scale Corporation

Bridgeview ILL, USA). BMI was calculated as [weight (kg)/(height (m))2]. Body surface area was calculated as √퐻푒𝑖𝑔ℎ푡 (푐푚) × 푤푒𝑖𝑔ℎ푡 (푘𝑔)/3600 [Mosteller et al., 1987]. To examine serum 25 OHD, metabolic profiles, glucose metabolism, parathyroid hormone

(PTH) and high sensitivity C-reactive protein (hs-CRP), blood samples were taken after a minimum of 8 hours fasting overnight and measured by standard methods. Serum 25 OHD levels were measured at VivoPharm laboratories (Bundoora, Victoria, Australia) by a liquid chromatography-tandem mass spectrometry (LC-MS/MS) method using Applied Biosystems

4000 Q trap and Agilent LC-MS/MS instruments [Van den Ouweland et al., 2013]. The intra- assay coefficients of variation (CV) were 5.8% and 4.4% for 25 OHD2 and 25 OHD3,

160 respectively. Total 25 OHD levels (the sum of 25 OHD3 and 25 OHD2) were used in statistical analyses. HDL levels were measured directly, using specific detergent, with total

CV 3.2%. Insulin and hs-CRP were assayed using the ARCHITECT, chemiluminescent microparticle immunoassay (CMIA) method. Cholesterol and LDL levels were measured by

Abbott ARCHITECT c-Systems utilising an enzymatic method by a reagent kit with 1.5%

CV. Glucose and triglyceride levels were measured applying Hexokinase and Glycerol

Phosphate Oxidase methods using a reagent kit with CV of 2.3 and 2.3%, respectively.

HbA1c was measured by a high performance liquid chromatography method using phenylboronic acid with CV of 5-10% [Davis et al., 1978]. HOMA-IR and HOMA-B

(homeostatic model assessment of beta cell function) were calculated according to the standard equations [Turner et al., 1985]. QUICKI (quantitative insulin sensitivity check index) was calculated according to the following formula [1 / (log(fasting insulin in µU/mL)

+ log(fasting glucose in mg/dL))] [Katz et al., 2000]. All data were collected twice, at baseline and four months after the baseline site visit. The protocol-defined visit window for the site visit was four months ± 2 weeks post-baseline.

An adverse event was defined as any occurrence that had unfavourable and/or unintended effects on participants, regardless of severity or study treatment-relatedness. A serious adverse event was defined as any adverse event which resulted in death, hospitalization, significant disability or congenital abnormality/birth defect [Food and Drug Administration].

All adverse events that occurred during the study were reported to the coordinating principal investigator (JDW) and recorded whether or not they were considered to be serious and/or related to the study treatment.

Statistical Analysis: The Statistical Package for Social Science (IBM SPSS Statistics for

Windows, version 22.0. Armonk, NY: IBM Corp) was used for all statistical analysis [IBM

161

Corp, 2013]. The primary analysis was done based on the intention to treat (ITT) with multiple imputation used to deal with missing data. A per protocol analysis was used to ensure robustness of findings. Kolmogorov-Smirnov tests and histograms were used to examine the distribution of continuous variables. Two-way ANOVA (analysis of variance) or

Kruskal-Wallis for continuous and chi-square test for categorical variables were used to compare baseline characteristics among the three groups and the impact of intervention on the outcome of interest. As there were limited baseline differences across groups, baseline levels were not adjusted in the analysis. To examine the effects of either vitamin D supplementation or behavioural intervention on CVD risk factors, two way ANOVA or

Kruskal-Wallis were used. To compare changes in each intervention groups with the control group post hoc Tukey test was used. Paired t-test or Wilcoxon were used to compare changes in 25 OHD levels and metabolic profiles from baseline to four month follow-up in each group. The t and z statistic are presented which show the ratio of the difference between estimated and hypothesized value to standard error of the tested variable. Sensitivity analyses were conducted on the per protocol dataset, including all participants with no protocol deviations and who completed all measurements for the primary variables. A p-value less than 0.05 was considered statistically significant.

5.1.3 Results

In total, 407 participants were recruited into the cross-sectional part (Part A). Based on the selection criteria for the trial, 200 (49%) were deemed eligible, and 123 (30%) gave written informed consent and were randomized. Of those, 96 (78.0%) participants completed their four month follow-up site visit. This trial started in August 2014, and all the four month follow-up visits were conducted by March 2016. In total, 27 participants withdrew from the study before their four month follow-up visit, leaving 96 participants (78%) with complete

162 four month follow-up visit data. All of the 123 randomized participants were included in the intention to treat analysis. Results from the cross-sectional part of the study have been published elsewhere [Tabesh et al., 2017]. Baseline demographics and general characteristics of Part B participants are shown in Table 5.1.1. There were no significant differences in age, race, season when blood was taken, vitamin D intake, energy intake, obesity, central obesity, sun exposure, physical activity and alcohol intake, between groups. However, a higher proportion of participants in the behavioural group reported drinking daily to weekly compared to the other two groups (p = 0.032) (Table 1). Mean serum 25 OHD levels were 54

± 15 nmol/L at baseline and there were no significant differences in 25 OHD levels across the three groups (p = 0.491). Baseline levels of metabolic profiles, anthropometric measurements and blood pressure were also similar across the three groups (Table 5.1.2). Compliance of all recruited participants with vitamin D supplementation was 91% and based on the remaining capsules in the container. Compliance in the behavioural intervention group was 22% as defined by the number of days participants opened the application. However, sensitivity analysis showed that there were no significant differences in outcomes between the lowest quartile and highest quartile of application compliance. The effects of vitamin D supplementation and behavioural intervention on lipid profiles, glucose metabolism and blood pressure are presented in Table 5.1.3. Changes in BMI, HbA1c (haemoglobin A1c), non-HDL (high density lipoprotein) cholesterol and HDL to total cholesterol ratio were not significantly different between the three groups over four months. Metabolic profiles and 25

OHD level changes from baseline to four months after intervention in each group are presented in Table 5.1.4. Both interventions led to significant increases in seasonally-adjusted

25 OHD levels (30.8 nmol/L in the supplement group: p < 0.001, 11.1 nmol/L in behaviour- change group: p = 0.031), while 25 OHD levels in the control arm did not increase significantly (p = 0.270). There was a significant reduction in BMI in participants who were

163 taking vitamin D supplements (0.32 kg/m2, p = 0.031). Whilst a significant reduction was seen in HbA1c levels in the behavioural intervention group (p = 0.001), no change was observed in the supplement group. HbA1c also decreased significantly in the control group (t

= -2.65, p = 0.008) (Table 5.1.4). No significant changes were observed in the non-HDL or

HDL to total cholesterol ratio in all groups, after four month of intervention. Sensitivity analysis based on the per protocol analysis (supplementary tables 5.1.5 and 5.1.6) showed a similar result: therefore, we reported only the ITT analysis.

In the control group (n=42) there were seven adverse events in the supplement group (n=41), five adverse events and in the behavioural intervention group (n=40) four adverse events occurred during the study. All details are presented in Table 5.1.7. All adverse events were graded according to the National Institute of Health (NIH) grading system, version 5

[National Institutes of Health, 2018]. All adverse events were reviewed by a senior investigator (JDW). No adverse events were deemed to be study treatment-related.

5.1.4 Discussion

In this randomized controlled clinical trial of vitamin D supplementation and behavioural interventions, 25 OHD levels increased significantly after four months of intervention in both groups. Four months of vitamin D supplementation resulted in a reduction in BMI (t=-2.33, p

= 0.031), compared to baseline. Moreover, the behavioural intervention resulted in a significant reduction (t=-3.28, p = 0.001) in HbA1c levels. However, these changes were quantitatively small and of uncertain clinical significance. Moreover, there were no significant differences among the three groups in measures of glucose metabolism, metabolic profiles and blood pressure. There was no evidence of improvement in CVD risk factors with either 1000 IU vitamin D supplementation daily or implementing a behavioural intervention over four months.

164

In a previous systemic review and meta-analysis of twenty-one clinical trials, it was shown that increased 25 OHD levels could be associated with improvement in CVD risk factors including lipid profiles, blood pressure, glucose metabolism and BMI [Ford et al., 2014].

However, evidence to date is not adequate to establish a causative link between 25 OHD levels and CVD risk. Clinical trials that have been conducted in this area have reported inconsistent findings [Pilz et al., 2015 and Al-Dujaili et al., 2016]. Similar to our findings, a randomized controlled trial on 188 hypertensive patients in 2015 evaluated the effects of supplementing with 2800 IU vitamin D3 daily for 8 weeks on blood pressure, lipid profiles and HOMA-IR and showed that vitamin D3 supplementation had no effects on blood pressure and other CVD risk factors, but it was associated with an increase in triglyceride levels [Pilz et al., 2015]. A cohort study in line with our findings, showed that although vitamin D deficiency is associated with unfavourable lipid profiles in cross-sectional analyses, increasing 25 OHD levels from less than 20 to more than 30 ng/mL did not result in lipid profile improvement [Ponda et al., 2012]. However, Al-Dujali et al. conducted a randomized, placebo-controlled, single-blinded, parallel-group trial on 15 healthy male and female participants (aged 19 to 53 years old) and showed that daily 2000 IU vitamin D3 supplementation for 14 days significantly decreased systolic (9.5 mmHg, p = 0.022) and diastolic (6.9 mmHg, p = 0.012) blood pressure [Al-Dujaili et al., 2016].

These inconsistent findings could be attributed to the small sample sizes, short duration of follow-up, differences in age, sex and different dosage of vitamin D supplementation.

Moreover, most previous trials were designed for clinical endpoints other than CVD risk factors and they were largely conducted in elderly, unhealthy participants.

Various mechanisms have been proposed to explain the effects of 25 OHD level improvements on BMI or HbA1c. Improvement in serum 25 OHD levels can cause a reduction in parathyroid hormone (PTH) levels, which can increase calcium absorption,

165 calcium influx into the adipocytes and reduce lipolysis [Khundmiri et al., 2016]. Moreover,

25 OHD levels can suppress the expression of the peroxisome proliferator activated receptor

γ (PPAR γ), which plays a role in lipogenesis and lipolysis [Tyagi et al., 2011]. Vitamin D also improves insulin sensitivity by reducing systemic inflammation, enhancing glucose uptake and production of insulin in beta cells [Al-Shoumer et al., 2015].

A number of qualitative studies have assessed the quality and usability of using m-Health to achieve life style changes in different populations [Free et al., 2010]. However, to our knowledge there are very limited studies using m-health interventions to increase safe UV exposure to achieve sufficient vitamin D levels [Buller et al., 2015]. A clinical trial on healthy young adult males and females aged 18 years or older evaluated the effects of using a smartphone mobile application on sun protection behaviours. They reported that use of the mobile application was lower than expected (41%), but was associated with increased sun protection [Buller et al., 2015]. Goodman et al. in 2016 evaluated the effects of watching an educational video about vitamin D, using a mobile application on vitamin D intake among 90 adults (18 to 25 years old). They showed that vitamin D intake increased significantly more in the intervention group, compared to the control group, after 12 weeks of using the application (p = 0.046) [Goodman et al., 2016]. It seems that the compliance rate for using health-related mobile applications was also lower than expected in previous studies.

Therefore, finding ways to encourage mobile application use are needed [Buller et al., 2015].

As we expected, app use was higher at the beginning of the study to build self-efficacy and then subjects became less dependent on the app as they formed sun safe habits.

To our knowledge, our study is the first clinical trial evaluating the effects of using a mobile- based application on safely increasing UV exposure and improving vitamin D status, as well as its impact on CVD risk factors.

166

Strengths of the present study include its single-blind randomized controlled trial design, measuring vitamin D metabolites using a liquid chromatography-tandem mass spectrometry

(LC-MS/MS) method (currently considered the gold standard method to measure serum 25

OHD levels) [Freeman et al., 2015]. Another strength of the project is high compliance in the supplement group (91.4%). Finally, in this study we measured sun exposure directly by using dosimeters - an objective measure. Most previous studies used a questionnaire to collect sun exposure data.

The present study has some limitations. First, the compliance in the behavioural intervention group was low (22 %). Although the application compliance was poor according to our definition, which was the number of days on which the application was opened, our original concept was that the application would be a learning tool and that users eventually would change their behaviour with respect to sun exposure. Hence, the percentage of participants who changed their behaviour to use more sun protection was 52%, a reasonable outcome for those actively using the application. A second limitation was the relatively short duration of intervention (four months), which could be too short to see some effects of vitamin D supplementation or the behavioural intervention. The complete follow-up of participants in this trial is for a total of 12 months: these results will be reported separately. Third, whilst all study and laboratory staff were blinded to the participant groups, we were not able to blind participants to their group allocation; thus, this study was an open-label design study.

However, it has been shown that open-label design studies are lower in cost and have greater similarity to standard clinical practice, which make results more readily extrapolated to real life [Hansson et al., 1992]. Finally, the results of our study may not be generalizable to men or older individuals as all participants in this project were women aged 16 to 25 years.

In conclusion, vitamin D supplementation resulted in a reduction in BMI, whilst a behavioural application intervention resulted in a significant reduction in HbA1c compared to

167 baseline after four months follow-up. However, these effects were quantitatively small and of uncertain clinical significance. After four months of intervention, we found no clear evidence for improvement in 25 OHD levels reducing CVD risk factors in young women.

168

Table 5.1.1: Baseline general participant characteristics 1

All Behavioural group Supplement group Control group p-value2 (n=123) (n=40) (n=41) (n=42) Age (years) 23 (21-24) 23 (22-25) 23 (20-25) 23 (21-24) 0.761 Race, ethnicity; n (%) - European 84 (68 %) 28 (70%) 25 (69%) 29 (69%) 0.891

- Other 39 (31%) 9 (30%) 11 (30%) 13 (31%) Blood taken in winter; n 46 (37.4%) 14 (35.0%) 15 (41.7%) 15 (35.7%) 0.807

(%) Vitamin D intake (ug/day) 0.91 (0.36-1.60) 0.80 (0.15-1.41) 0.43 (0.27-1.57) 1.05 (0.31-2.02) 0.698 Energy intake (kJ/day) 6015 (4709-7668) 5742 (4406-8626) 5851 (4577-7922) 6616 (4985-7563) 0.873 Obesity3; n (%) 21 (17.1%) 5 (12.5%) 7 (19.4%) 8 (19.0%) 0.720 4 Central obesity ; n (%) 48 (39.0%) 12 (30.0%) 14 (38.9%) 21 (50.0%) 0.809

Current smoker; n (%) 12 (9.8%) 5 (12.5%) 3 (8.3%) 4 (9.5%) 0.481

Alcohol intake; n (%) - Never to monthly 80 (65%) 22 (55.0%) 24 (66.7%) 25 (80.6%) 0.032

- Weekly to daily 41 (33.3%) 18 (45.0%) 12 (33.3%) 9 (21.4%)

Binge drinking; n (%) 31 (25.2%) 8 (20.0%) 9 (25.0%) 13 (31.0%) 0.940

Alcohol intake (gram/day) 5.1 (1.7-11.0-) 6.3 (1.2-12.5) 4.4 (1.2-9.5) 2.3 (0.6-7.2) 0.204

Sun exposure (SED) 0.10 (0.04-0.21) 0.10 (0.03-0.19) 0.10 (0.05-0.24) 0.11 (0.05-0.22) 0.352 Physical activity 1020 (525-1875) 1020 (330-1950) 960 (262.5-1552.5) 990 (570-2145) 0.645 (Met-min/day) 1Data are median (interquartile range) or number (percent) 2p-value obtained from ANOVA or Kruskal Wallis for continues variables and Chi square for categorical variables, p < 0.05 considered as statistically significant 3Defined as having BMI of 30 kg/m2 or greater 4Defined as having waist circumferences of greater than 80 cm

169

Table 5.1.2: Summary of baseline 25 OHD, metabolic profiles, blood pressure and anthropometric measurements 1 Behavioural group Supplement group Control group p-value2 (n=40) (n=41) (n=42) 25 OHD (nmol/L) 55±15 52±14 55±16 0.491 Metabolic profiles Glucose (mmol/L) 4.6 (4.4-4.8) 4.8 (4.5-5.1) 4.7 (4.5-5.1) 0.224 Insulin (mU/L) 8 (6.6-10.7) 9.4 (6.0-11.7) 8.2 (6.9-11.7) 0.787 HbA1c (mmol/mol Hb) 31 (30-33) 32.5 (31-35) 32 (30-35) 0.068 HOMA-IR 1.7 (1.2-2.2) 2.03 (1.28-2.43) 1.8 (1.4-2.6) 0.584 HOMA-B 1439 (112-212) 135 (108-198) 142 (92-181) 0.346 QUICKI 1.04±0.12 1.03±0.16 1.03±0.17 0.932 TC (mmol/L) 4.4 (3.7-5.3) 4.6 (3.9-5.0) 4.5 (4.0-5.2) 0.511 TG (mmol/L) 0.8 (0.5-1.1) 0.8 (0.6-1.0) 0.8 (0.7-1.1) 0.882 LDL (mmol/L) 2.5±0.7 2.6±0.6 2.9±1.2 0.466 HDL (mmol/L) 1.4 (1.2-1.6) 1.4 (1.3-1.8) 1.4 (1.3-1.7) 0.927 Non-HDL cholesterol 2.9 (2.2-3.5) 2.9 (2.5-3.5) 3.3 (2.5-3.8) 0.549 (mmol/L) HDL-C/total-C 0.34±0.08 0.34±0.07 0.32±0.09 0.603 Blood pressure SBP (mmHg) 112±11 108±9 112±10 0.203 DBP (mmHg) 72 (68-76) 69 (62-76) 73 (67-77) 0.124 Anthropometric measurements Height (cm) 166.8±6.7 164.9±6.7 166.4±5.8 0.112 Weight (kg) 64.5 (58.4-73.8) 60.4 (53.6-80.5) 67.7 (58.1-79.0) 0.272 Waist (cm) 77.5 (71.2-82.6) 76.5 (70.6-89.1) 80.5 (72.0-91.5) 0.711 Hip (cm) 98.5 (95.1-104.0) 95 (89.8-106.2) 104.0 (92.7-112.0) 0.162 WHR 0.7±0.1 0.8±0.1 0.7±0.1 0.356 BMI (kg/m2) 23.0 (21.2-26.3) 22.0 (20.5-29.4) 25.2 (21.4-27.7) 0.364 BSA (m2) 1.73 (1.63-1.88) 1.55 (1.66-1.91) 1.76 (1.65-1.95) 0.207 hs-CRP (mg/L) 1.3 (0.7-3.6) 1.2 (0.4-7.2) 1.3 (0.5-2.7) 0.880

170

1Data are mean ± standard deviation or median (interquartile range). 2p-value by ANOVA or Kruskal Wallis, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D; HbA1c: haemoglobin A1c; HOMA-IR: homeostatic model assessment of insulin resistance; HOMA-B: homeostatic model assessment of beta cell function; QUICKI: Quantitative insulin sensitivity check index; TC: total cholesterol TG: triglyceride; LDL: low density lipoprotein; HDL: high density lipoprotein; SBP: systolic blood pressure; DBP: Diastolic blood pressure; WHR: waist to hip ratio BMI: body mass index; BSA: body surface area; hs-CRP: high sensitivity C-reactive protein

171

Table 5.1.3: Mean (± standard deviation) changes in 25 OHD, metabolic profiles and anthropometric measurements after 4 months of intervention 1

Behavioural group Supplement group Control group p-value2 (n=40) (n=41) (n=42) 25 OHD (nmol/L) 11±22* 30±28* 9±21 <0.001

BMI (kg/m2) 0.2±1.0 0.3±0.9 -0.1±1.1 0.102 HbA1c (mmol/mol Hb) 1.3±2.2 0.3±2.6 0.9±2.3 0.173 Non-HDL cholesterol 0.08±0.37 0.02±0.54 0.06±0.47 0.811 (mmol/L) HDL-C/total-C 0±0.38 0±0.04 0±0.03 0.557 1 Data are mean change ±SD based on intention to treat 2p-value by ANOVA or Kruskal Wallis for differences among three groups, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D; BMI: Body mass index; HbA1c: haemoglobin A1c; Non-HDL cholesterol: Non-high density lipoprotein cholesterol; HDL-C/total-C: high density lipoprotein cholesterol to total cholesterol ratio *Statistically-significant compared to control group, using post hoc Tukey test

172

Table 5.1.4: Paired t-test or Wilcoxon comparing 25 OHD levels and CVD risk factors prior to and following 4 months of intervention

Behavioural group Supplement group Control group (n=40) (n=41) (n=42) t or z statistic p-value2 t or z statistic p-value2 t or z statistic p-value2

25 OHD 3.20 0.031 6.95 <0.001 2.71 0.270 BMI 1.59 0.119 -2.33 0.031 -0.75 0.459 HbA1c -3.28 0.001 0.88 0.372 -2.65 0.008 Non-HDL cholesterol 1.19 0.237 0.24 0.816 0.85 0.391 HDL-C/total-C 1.58 0.115 0.16 0.879 0.44 0.651 1 Data are t-statistic or z-statistic based on intention to treat 2p-value by paired t-test or Wilcoxon, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D; BMI: Body mass index; HbA1c: haemoglobin A1c; Non-HDL cholesterol: Non-high density lipoprotein cholesterol; HDL-C/total-C: high density lipoprotein cholesterol to total cholesterol ratio

173

Supplementary Table 5.1.5: Mean (± standard deviation) changes in 25 OHD, metabolic profiles and anthropometric measurements over 4 month intervention - per protocol analysis 1

Behavioural group Supplement group Control group p-value2 (n=29) (n=36) (n=31) 25 OHD 11±25* 38±31* 7±25 <0.001 BMI 0.3±1.2 0.3±1.0 -0.2±1.3 0.098 HbA1c 1.3±2.0 0.2±2.7 1.1±2.5 0.160 Non-HDL cholesterol 0.11±0.42 0.01±0.04 0.06±0.51 0.727 HDL-C/total-C 0±0.04 0±0.55 0±0.03 0.487 1 Data are mean change±SD based on per protocol analysis 2p-value by ANOVA or Kruskal Wallis for differences among three groups, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D; BMI: Body mass index; HbA1c: haemoglobin A1c; Non HDL cholesterol: Non high density lipoprotein cholesterol; HDL-C/total-C: high density lipoprotein cholesterol to total cholesterol ratio *Statistically significant compared to control group, using post hoc Tukey test

174

Supplementary Table 5.1.6: Paired t-test or Wilcoxon comparing 25 OHD levels and CVD risk factors prior to and following 4 month intervention - per protocol analysis

Behavioural group Supplement group Control group (n=29) (n=36) (n=31) t or z statistic p-value2 t or z statistic p-value2 t or z statistic p-value2 25 OHD 2.56 0.016 7.30 <0.001 1.51 0.142 BMI 1.29 0.207 -2.06 0.046 -1.01 0.319 HbA1c -3.60 0.001 0.59 0.553 -2.64 0.013 Non-HDL cholesterol 1.50 0.144 0.05 0.954 0.69 0.490 HDL-C/total-C 1.39 0.174 0.23 0.813 0.20 0.840

1 Data are t-statistic or z-statistic based on per protocol analysis 2p-value by paired t-test or Wilcoxon, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D; BMI: Body mass index; HbA1c: haemoglobin A1c; Non-HDL cholesterol: Non-high density lipoprotein cholesterol; HDL-C/total-C: high density lipoprotein cholesterol to total cholesterol ratio

175

Table 5.1.7 Summary of adverse events occurred in each group

Adverse events Behavioural intervention Grade Supplement Grade Control group Grade n (%) group (n=40) group (n=41) (n=42) Abdominal pain Not reported - Not reported - 1 (2%) 1 Syncope Not reported - Not reported - 1 (2%) 3 Sunburn 3 (7%) 1 3 (7%) 1 5 (12%) 1 Menorrhagia Not reported - 1 (2%) 1 Not reported - Anaphylaxis Not reported - 1 (2%) 3 Not reported - Increased 1 (2%) 1 Not reported - Not reported - anxiety/depression Grades are according to the National Institute of Health (NIH) grading system, version 5

176

5.2 The effects of vitamin D supplementation and behavioural intervention on metabolic profiles over one year follow-up

The primary analysis was done based on the intention to treat (ITT) with multiple imputation used to deal with missing data. A per protocol analysis was used to ensure robustness of findings.

In total 102 participants were included in the intention to treat analysis (30 in the behavioural group, 34 in the supplement and 38 in the control group). The mean and SD of serum 25

OHD levels, metabolic profiles, blood pressure, anthropometric measurements and hs-CRP levels in each arm are shown in Table 5.2.1. Mean±SD of 25 OHD levels were 76.35±20.03 nmol/L in the behavioural intervention group, 61.24±17.26 nmol/L in the control and

54.7±17.7 nmol/L in the supplement group. Mean and SD changes in metabolic profiles from baseline to one year follow-up are shown in Table 5.2.2. There were no significant differences in glucose, insulin, HOMA-IR, HOMA-B, QUICKI, total cholesterol, triglyceride, LDL, HDL and HDL to total cholesterol levels, systolic and diastolic blood pressure, height, weight, BMI, waist and hip circumferences, waist to hip ratio and hs-CRP levels among the three groups after 12 months of follow-up. HbA1c levels ware increased in all three groups after 12 months of intervention. However, increase in HbA1c was significantly higher in the behavioural intervention group compared to the other two groups

(2.40±2.21 in behavioural group versus 0.89±2.64 in supplement group and 1.09±2.39 in control, p-value for differences among three groups = 0.03). As it was expected 25 OHD levels were significantly increased more in the supplement group compare to the behavioural and control group. 25 OHD levels increased 24.3±24.2 nmol/L in supplement group and

6.7±17.7 nmol/L in behavioural group and 1.7±19.0 nmol/L in control group (p-value for differences among three groups <0.001) (Table 5.2.3).

177

Paired t-test or Wilcoxon were used to compare changes in 25 OHD levels and metabolic profiles, blood pressure, anthropometric measurements and hs-CRP levels from baseline to

12 months follow-up in each group. After 12 months of intervention there were a significant reduction in insulin (p = 0.03), QUICKI (p = 0.02), diastolic blood pressure (p = 0.02), waist

(p < 0.001) and hip circumferences (p = 0.01) in the behavioural intervention group.

However, in this group significant increases were seen in HbA1c levels (p < 0.001), HDL to total cholesterol ratio (p = 0.03) and BMI (p = 0.03). There was a marginally significant increase in non HDL cholesterol (p = 0.07) and marginally significant decrease in HOMA-IR

(p = 0.07) and LDL levels (p = 0.07). In the supplement group after 12 months of follow-up, there was significant reduction in glucose levels (p = 0.012), HOMA-B (p = 0.014), hip circumference (p = 0.004) and there was a significant increase in HbA1c levels (p = 0.03) and

BMI (p = 0.003). Marginally significant decrease was also observed in triglyceride levels (p

= 0.064) in this group. In the control group HbA1c significantly increased (p = 0.017) and waist circumference (p = 0.004) and WHR (p = 0.011) were significantly decreased after 12 months.

Sensitivity analyses were conducted on per protocol dataset, including all participants with no protocol deviation and completed all measurements for the primary variables. Per protocol analysis has been done on 84 participants (26 in behavioural group, 27 in supplement and 31 in control group). Per protocol analysis revealed similar results to the intention to treat analysis (Table 5.2.4 and 5.2.5). However, in the per protocol analysis no significant association were seen in HbA1c levels among the behavioural, supplement and control groups. Wilcoxon test revealed that HbA1c significantly increased (p = 0.001); however,

QUICKI and diastolic blood pressure decreased (p < 0.001 and p = 0.04, respectively) in the behavioural intervention group. In the supplement group glucose and QUICKI significantly decreased (p = 0.02 and p < 0.001, respectively) according to the per protocol analysis. In the

178 control group QUICKI (p < 0.001), waist circumference (p = 0.01) and WHR (p = 0.03) decreased significantly. Similar to two other groups HbA1c significantly increased in the control group (p = 0.01). Sun exposure changes in the behavioural group are presented with more details in chapter 7.

179

Table 5.2.1: The 12 months 25 OHD, metabolic profiles, blood pressure and anthropometric measurements of study participants (ITT analysis) 1 Behavioural group Supplement group Control group (n=30) (n=38) (n=34)

25 OHD (nmol/L) 76±20 54±17 61±17

Metabolic profiles Glucose (mmol/L) 4.64±0.14 4.64±0.13 4.69±0.13 Insulin (mU/L) 9.7±1.7 9.7±1.3 9.5±2.7 HbA1c (mmol/mol Hb) 33.5±0.7 33.8±0.8 33.4±1.2 HOMA-IR 2.0±0.3 2.0±0.2 2.0±0.5 HOMA-B 172.4±39.9 174.2±38.4 161.5±45.8 QUICKI 0.60±0.02 0.60±0.02 0.61±0.04 TC (mmol/L) 4.53±0.58 4.61±0.55 4.90±1.41 TG (mmol/L) 0.90±0.34 1.10±0.41 0.98±0.39 LDL (mmol/L) 2.64±0.51 2.57±0.46 2.99±1.27 HDL (mmol/L) 1.47±0.21 1.53±0.28 1.46±0.29 Non-HDL cholesterol 3.05±0.55 3.08±0.51 3.44±1.29 (mmol/L)

180

HDL-C/total-C 0.32±0.05 0.33±0.05 0.31±0.07

Blood pressure SBP (mmHg) 111.4±9.2 108.6±8.7 109.5±8.5

DBP (mmHg) 71.3±5.8 68.5±5.5 70.4±5.5 Anthropometric measurements

Height (cm) 168.2±5.6 165.4±6.5 166.7±5.5 Weight (kg) 73.1±20.2 68.2±16.6 72.1±19.0

Waist (cm) 84.5±13.6 80.8±11.6 84.4±14.1 Hip (cm) 104.8±14.1 101.0±10.9 103.6±13.4

WHR 0.80±0.04 0.79±0.03 0.81±0.04 BMI (kg/m2) 25.7±6.4 24.7±5.3 25.8±6.3

hs-CRP (mg/L) 2.99±2.70 2.77±2.38 3.24±2.65

1Data are mean ± standard deviation or median (interquartile range). 25 OHD: 25 hydroxy vitamin D; HbA1c: haemoglobin A1c; HOMA-IR: homeostatic model assessment of insulin resistance; HOMA-B: homeostatic model assessment of beta cell function; QUICKI: Quantitative insulin sensitivity check index; TC: total cholesterol TG: triglyceride; LDL: low density lipoprotein; HDL: high density lipoprotein; SBP: systolic blood pressure; DBP: Diastolic blood pressure; WHR: waist to hip ratio; BMI: body mass index; hs-CRP: high-sensitivity C-reactive protein

181

Table 5.2.2: Mean ±SD changes in 25 OHD, metabolic profiles and anthropometric measurements over the 12 months intervention (ITT analysis) 1

Changes Behavioural group Supplement group Control group p-value2

(n=30) (n=38) (n=34)

25 OHD (nmol/L) 6±17 24±24 1±19 <0.001

Metabolic profiles

Glucose (mmol/L) -0.01±0.31 -0.14±0.37 -0.05±0.31 0.26

Insulin (mU/L) 0.70±2.31 0.21±3.38 -0.64±8.52 0.60

HbA1c (mmol/mol Hb) 2.40±2.21 0.89±2.64 1.09±2.39 0.03

HOMA-IR 0.12±0.54 -0.04±0.82 -0.19±1.93 0.59

HOMA-B 8.37±50.77 10.85±77.65 -0.71±109.8 0.83

QUICKI -0.020±0.043 -0.021±0.087 -0.021±0.082 0.99

TC (mmol/L) 0.10±0.41 0.01±0.57 0.05±0.57 0.79

TG (mmol/L) 0.01±0.45 0.17±0.48 0.09±0.41 0.39

LDL (mmol/L) 0.13±0.37 -0.04±0.51 -0.01±0.50 0.26

HDL (mmol/L) -0.030±0.18 0.002±0.127 0.017±0.180 0.49

Non-HDL cholesterol 0.13±0.41 0.01±0.53 0.03±0.50 0.56 (mmol/L)

182

HDL-C/total-C -0.017±0.043 -0.003±0.04 -0.004±0.035 0.29

Blood pressure

SBP (mmHg) -1.51±6.12 0.23±5.08 -2.12±7.21 0.24

DBP (mmHg) -1.19±5.9 -0.26±4.52 -1.89±5.66 0.43

Anthropometric measurements

Height (cm) 0.16±0.31 0.32±0.32 0.30±0.45 0.19

Weight (kg) 1.77±4.40 1.88±3.92 0.18±4.37 0.18

Waist (cm) 3.20±3.95 0.95±5.70 2.47±5.59 0.20

Hip (cm) 2.18±4.0 2.85±5.9 1.08±6.4 0.38

WHR 0.01±0.03 0.01±0.04 0.01±0.04 0.20

BMI (kg/m2) 0.60±1.52 0.59±1.38 0.04±1.5 0.12

hs-CRP (mg/L) -0.37±4.20 -2.96±10.98 0.48±3.16 0.24

1 Data are Mean change±SD based on intention to treat 2P-value by ANOVA or Kruskal Wallis, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxy vitamin D; HbA1c: haemoglobin A1c; HOMA-IR: homeostatic model assessment of insulin resistance; HOMA-B: homeostatic model assessment of beta cell function; QUICKI: Quantitative insulin sensitivity check index; TC: total cholesterol TG: triglyceride; LDL: low density lipoprotein; HDL: high density lipoprotein; SBP: systolic blood pressure; DBP: Diastolic blood pressure, WHR: waist to hip ratio; BMI: body mass index; hs-CRP: high-sensitivity C-reactive protein

183

Table 5.2.3: Paired t-test or Wilcoxon comparing 25 OHD levels and CVD risk factors prior to and following 12 months intervention (ITT analysis) 1

Behavioural group Supplement group Control group (n=30) (n=38) (n=34)

t or z statistic p-value2 t or z statistic p-value2 t or z statistic p-value2

25 OHD 2.197 0.028 0.134 0.894 4.387 <0.001

Metabolic profiles Glucose -0.154 0.877 -0.932 0.351 -2.509 0.012 Insulin -2.067 0.039 -1.257 0.209 -0.964 0.335 HbA1c 4.103 <0.001 2.393 0.017 2.110 0.035 HOMA-IR -1.800 0.072 -1.154 0.248 -0.370 0.712 HOMA-B -1.162 0.245 -1.205 0.228 -2.458 0.014 QUICKI -2.335 0.020 -1.701 0.089 -1.095 0.274 TC -1.368 0.171 -0.957 0.338 -0.138 0.890 TG -0.113 0.910 -1.428 0.153 -1.849 0.064 LDL -1.762 0.078 -0.116 0.908 -0.786 0.627

184

HDL 0.627 0.530 -0.420 0.675 -0.399 0.690 Non-HDL cholesterol 1.800 0.072 0.915 0.360 0.189 0.850 HDL-C/total-C 2.067 0.039 0.436 0.663 0.413 0.679 Blood pressure SBP -1.121 0.262 -1.821 0.069 -0.196 0.845 DBP -2.229 0.026 -0.229 0.819 -0.059 0.953 Anthropometric measurements Waist -3.552 <0.001 -2.886 0.004 -1.457 0.145 Hip -2.414 0.016 -0.742 0.458 -2.879 0.004 WHR -1.731 0.084 -2.546 0.011 -1.196 0.232 BMI 2.149 0.032 0.556 0.578 2.937 0.003 hs-CRP -1.656 0.098 -1.667 0.096 -0.703 0.482

1 Data are t-statistic or z-statistic based on intention to treat 2P-value by paired t-test or Wilcoxon, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxy vitamin D; HbA1c: haemoglobin A1c; HOMA-IR: homeostatic model assessment of insulin resistance; HOMA-B: homeostatic model assessment of beta cell function; QUICKI: Quantitative insulin sensitivity check index; TC: total cholesterol TG: triglyceride; LDL: low density lipoprotein; HDL: high density lipoprotein; SBP: systolic blood pressure; DBP: Diastolic blood pressure; WHR: waist to hip ratio; BMI: body mass index; hs-CRP: high-sensitivity C-reactive protein

185

Table 5.2.4: The effects of vitamin D supplementation and behavioural intervention on metabolic profiles/anthropometric measurements (per protocol analysis)1

Changes Behavioural group Supplement group Control group p-value2

(n=26) (n=31) (n=27)

Metabolic profiles

Glucose (mmol/L) 0.012±0.417 -0.135±0.307 -0.017±0.367 0.22

Insulin (mU/L) -0.03±3.66 -0.32±3.51 1.61±7.85 0.32

HbA1c (mmol/mol Hb) 2.19±2.56 0.80±2.53 1.22±2.25 0.10

HOMA-IR 0.008±0.836 -0.116±7.193 0.300±1.595 0.31

HOMA-B -13.41±76.41 -1.36±71.40 8.80±59.10 0.39

QUICKI 0.444±0.144 0.449±0.204 0.430±0.240 0.47

TC (mmol/L) 0.05±0.39 0.09±0.65 0.08±0.58 0.99

TG (mmol/L) -0.070±0.339 0.260±1.601 0.096±0.390 0.20

LDL (mmol/L) 0.096±0.339 0.006±0.740 -0.0143±0.479 0.82

HDL (mmol/L) 0.008±0.193 0.001±0.272 0.057±0.223 0.67

Non-HDL cholesterol 0.06±0.37 0.09±0.50 0.01±0.47 0.88 (mmol/L)

HDL-C/total-C -0.002±0.044 -0.014±0.047 0.003±0.037 0.35

186

Blood pressure

SBP (mmHg) -3.36±8.30 1.16±8.06 -1.71±11.10 0.27

DBP (mmHg) -3.04±7.99 0.41±6.97 -0.92±8.16 0.16

Anthropometric measurements

Height (cm) 0.30±0.63 0.26±0.51 0.31±0.76 0.88

Weight (kg) 0.05±4.55 1.71±4.86 0.81±6.64 0.73

Waist (cm) 2.20±7.25 0.52±7.30 3.52±9.12 0.21

Hip (cm) 2.23±7.03 2.80±6.28 1.57±8.05 0.81 WHR 0.002±0.055 -0.01±0.068 0.020±0.055 0.40

2 BMI (kg/m ) -0.03±1.55 0.58±1.78 0.19±2.36 0.47

hs-CRP (mg/L) -0.36±3.48 -3.71±12.10 0.41±3.17 0.23

1 Data are Mean change±SD 2P-value by ANOVA or Kruskal Wallis, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D; HbA1c: haemoglobin A1c; HOMA-IR: homeostatic model assessment of insulin resistance; HOMA-B: homeostatic model assessment of beta cell function; QUICKI: Quantitative insulin sensitivity check index; TC: total cholesterol TG: triglyceride; LDL: low density lipoprotein; HDL: high density lipoprotein; SBP: systolic blood pressure; DBP: Diastolic blood pressure, WHR: waist to hip ratio; BMI: body mass index; hs-CRP: high-sensitivity C-reactive protein

187

Table 5.2.5: Paired t-test or Wilcoxon comparing 25 OHD levels and CVD risk factors prior to and following 12 months intervention (per protocol analysis) 1

Behavioural group Supplement group Control group (n=26) (n=31) (n=27)

t or z statistic p-value2 t or z statistic p-value2 t or z statistic p-value2

Metabolic profiles Glucose -0.42 0.66 -0.40 0.68 -2.22 0.02 Insulin -0.56 0.57 -0.72 0.46 0.66 0.50 HbA1c 3.39 0.001 2.53 0.01 1.70 0.08 HOMA-IR -0.36 0.71 -0.56 0.56 0.92 0.35 HOMA-B -1.14 0.25 -0.31 0.75 0.46 0.64 QUICKI -4.37 <0.001 -4.60 <0.001 -4.86 <0.001 TC -0.78 0.43 0.70 0.48 0.52 0.59 TG -1.25 0.20 -1.34 0.17 -0.07 0.93 LDL -1.23 0.21 -0.15 0.87 0.05 0.58 HDL -0.18 0.85 -1.27 0.20 -0.73 0.46 Non-HDL cholesterol -0.94 0.34 -0.12 0.90 -0.71 0.47 HDL-C/total-C -0.05 0.95 -0.38 0.70 -1.32 0.18

188

Blood pressure SBP -1.74 0.08 -0.88 0.37 0.37 0.71 DBP -2.02 0.04 -1.68 0.09 -0.60 0.54 Anthropometric measurements Height -2.07 0.038 -2.31 0.02 -2.48 0.01 Weight -0.40 0.68 -0.75 0.44 -1.68 0.09 Waist -1.54 0.12 -2.51 0.01 -0.75 0.45 Hip -1.49 0.13 -0.98 0.32 -1.79 0.07 WHR -0.27 0.78 -2.16 0.03 0.92 0.35 BMI -0.24 0.80 -0.57 0.56 -1.54 0.12 hs-CRP -0.53 0.59 -1.03 0.30 -1.12 0.19

1 Data are t-statistic or z-statistic 2P-value by paired t-test or Wilcoxon, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D; HbA1c: haemoglobin A1c; HOMA-IR: homeostatic model assessment of insulin resistance; HOMA-B: homeostatic model assessment of beta cell function; QUICKI: Quantitative insulin sensitivity check index; TC: total cholesterol TG: triglyceride; LDL: low density lipoprotein; HDL: high density lipoprotein; SBP: systolic blood pressure; DBP: Diastolic blood pressure; WHR: waist to hip ratio; BMI: Body mass index; hs-CRP: high-sensitivity C-reactive protein

189

Chapter 6

THE EFFECTS OF VITAMIN D SUPPLEMENTATION AND BEHAVIOURAL INTERVENTION ON BODY COMPOSITION OVER ONE YEAR FOLLOW-UP

190

6.1 Introduction and literature review

In this chapter the associations of 25 OHD levels with body composition and fat distribution among young women are presented. Moreover, the effects of vitamin D supplementation and behavioural intervention on body composition and fat distributions are assessed.

Vitamin D deficiency affects almost half of the worldwide population, regardless of age and ethnicity [Mithal et al., 2009]. It is suggested that serum 25 OHD levels are lower in obese people due to the sequestration of 25 OHD in fat tissue and many previous studies showed an association of lower 25 OHD levels with higher BMI [Vanlint et al., 2013].

Adipose tissue is no longer considered an organ just to store energy, as it secretes more than

260 different hormones and proteins in the body [Lehr et al., 2012]. Recently, it has been suggested that body composition, including the amount and distribution of fat mass, and amount and distribution of lean mass, can be associated with many health outcomes, including cardiovascular diseases and diabetes [Liberato et al., 2013 and Vitezova et al.,

2017]. Although BMI is widely used to estimate obesity, it cannot reflect body composition and cannot distinguish lean mass from fat mass, or distribution of fat in different parts of the body. Recent studies have reported that adipose tissue stored in different locations in the body has different impacts on health outcomes [Despres et al., 2012].

A numbers of studies have shown a relationship between lean mass and vitamin D status

[Strazdienelow et al., 2012; Bentes et al., 2017], but results from previous studies on the association of vitamin D status and lean mass are inconsistent. Studies focused on the region of fat stored in young women are limited.

Some observational studies showed a negative association between 25 OHD and obesity and body fat mass [Vanlint et al., 2013]. However, it is still not clear whether 25 OHD levels are associated with fat mass stored in specific regions in the body. Also, evidence to date is not

191 adequate to establish a causative link between 25 OHD levels and obesity/body fat mass.

Extending knowledge in this area is important because it might help to elucidate the mechanisms of the association between 25 OHD levels and obesity.

Most studies in this area have been cross-sectional in design and have shown that lower 25

OHD levels are associated with higher body fat percent. A cross-sectional study by Vitezova et al. in 2016 included 2158 participants aged over 55 years and showed that participants with vitamin D deficiency (25 OHD <50 nmol/L) had higher body fat percentage compared to those with adequate 25 OHD levels (25 OHD >75 nmol/L) (Beta =1.29, 95% CI: 0.55, 2.04).

No association was found with lean mass (beta=0.01, 95% CI: 0.55, 2.02) [Vitezova et al.,

2017]. Another study, by Kremer et al., showed a strong association between low 25 OHD levels and both higher visceral fat mass (24.7 cm2 among vitamin D replete participants vs

44.8 cm2 among vitamin D deficient participants, p = 0.009) and subcutaneous fat (203.3 cm2 among vitamin D replete participants vs 288.1 cm2 among vitamin D deficient participants, p

= 0.029), measured by computed tomography [Kremer et al., 2009]. Moschonis et al. in 2008 showed similar results among 112 postmenopausal women and indicated a negative association between 25 OHD levels and total body fat mass and all measures of regional body fat mass [Moschonis et al., 2009]. A recent meta-analysis showed a weak but significant association between higher BMI and lower serum 25 OHD levels. However, this study did not analyse the relationship between body fat mass (fat percentage) and serum 25 OHD levels

[Saneei et al., 2013].

Clinical trials in this area are scarce and the results have been inconsistent. A randomised trial by Wamberg et al. in 2013 evaluated the effects of 7000 IU per day vitamin D supplementation for 26 weeks in 52 obese subjects aged 18 to 50 years. They found that increasing 25 OHD levels had no effect on total fat percent (p = 0.54) or visceral fat (p =

0.16) [Wamberg et al., 2013]. Another intervention study by Salehpor et al. in 2012 showed

192 that supplementation with vitamin D (25 mcg per day) for 12 weeks among 18 to 50 year-old subjects with BMI more than 25 kg/m2 caused a significant decrease in total body fat mass (p

< 0.001) [Salehpour et al., 2012].

Findings between studies may have differed due to use of different measures of body fat, differences in ethnicity, age, sex, regional adiposity and status of central obesity, different dosage of vitamin D supplementation and different duration of intervention.

Therefore, the existence of a very small number of clinical trials in this area, inconsistent results from previous studies and the potential importance of understanding the relationship between vitamin D status and body composition led us to pursue this project.

6.2 Aims

In this study we aimed to evaluate the effects of behavioural intervention (using a custom- designed, mobile-based application) or vitamin D supplementation (1000 IU per day) on body composition and fat distribution, measured by DXA scan over 12 months of follow-up.

6.3 Methods and Materials

Study design and participant information are presented in detail in chapter 2 and published elsewhere [Tabesh et al., 2016]. In summary, 123 young women, living in Victoria, Australia with insufficient 25 OHD levels (25 OHD between 25 and 75 nmol/L) were recruited and randomised to one of three groups: to receive a vitamin D supplement, or behavioural intervention using a mobile-based application, or control group. Follow-up was for 12 months. Serum 25 OHD levels, body composition, UV exposure, vitamin D intake and anthropometric measurements were obtained at baseline and after 12 months of intervention.

All data except the body composition data were also obtained after 4 months of intervention.

These data are presented in chapter 5.

193

Dual-energy X-ray absorptiometry (DXA; QDR 4500A densitometer, Hologic Inc., Bedford,

USA) was used to measure body composition. The DXA machine sends a low-dose x-rays beam with two different energy peaks. The amount of energy absorbed by the tissue shows the density and content of the body in that specific region [Berger et al., 2002]. We used whole-body scan to determine lean mass and fat mass in different regions including arms, lower limbs, head and trunk regions (Figure 6.1). Total fat mass was calculated by adding fat in arm, leg and trunk areas. Total fat percentage was calculated as total fat as a percentage of total body weight.

To determine the visceral fat area, manual delineation of the whole-body DXA scan was performed (Figure 6.2). The region for determination of visceral fat was identified on the whole body DXA scan before re-analysing. Visceral fat area was determined as the abdominal region from just above the iliac crest to above the second vertebra (L2) and from each side the lateral margins were aligned with muscle of the abdomen (Figure 6.3). Visceral fat percentage was calculated as total visceral fat as a percentage of total body fat mass

[Prichard et al., 1988].

To determine fat content in the gynoid region, manual delineation of the whole-body scan was performed and gynoid area defined as the area between the head of femur and mid-thigh

(Figure 6.3) [Stults-Kolehmainen et al., 2013]. Twenty scans were reanalysed for the gynoid region and a significant positive association was observed between gynoid fat and leg fat data

(p < 0.001, r2=0.78). Therefore, due to the time limitation we used the leg fat data to evaluate the association of lower body fat with CVD risk factors [Van Pelt et al., 2005].

Total fat to total lean ratio, visceral fat to total fat ratio and trunk fat to total fat ratio were also calculated.

194

Serum 25 OHD levels were measured at the Melbourne Health Pathology laboratories using

CMIA method, using the Architect 25-OH vitamin D reagent kit with imprecision of 10.9% at 20 nmol/L and 6.9% at 70 nmol/L [Hutchinson et al., 2017].

Statistical Analysis: The Statistical Package for Social Science (IBM SPSS Statistics for

Windows, version 22.0. Armonk, NY: IBM Corp) was used for all statistical analyses [IBM

Corp, 2013]. The primary analysis was done based on the ITT with multiple imputation used to deal with missing data. To examine the effects of either vitamin D supplementation or behavioural intervention on body composition the Kruskal-Wallis test was used. To compare changes in each intervention group with the control group post hoc the Tukey test was used.

Paired t-test or Wilcoxon were used to compare changes in 25 OHD levels and body composition from baseline to 12 month follow-up in each group. The t or z statistics reported from the paired t-test or Wilcoxon show the mean differences from baseline to 12 month follow-up. Linear regression analysis was used to evaluate the association of 25 OHD change and change in body composition over 12 months. A p-value less than 0.05 was considered statistically significant.

6.4 Results

General characteristics of participants were presented in chapter 3. Baseline 25 OHD levels and body composition are presented in Table 6.1. A Kolmogorov Smirnov test was used to check the distribution of variables; none of the variables were normally-distributed so non- parametric tests were used. No significant differences were observed in 25 OHD levels, total fat mass, total fat percent, visceral fat, visceral fat percent, trunk fat, leg fat, trunk to total fat ratio or visceral to total fat ratio at baseline among the three groups. Mean ± SD changes in

25 OHD and body composition over the 12-month intervention are shown in Table 6.2. Mean changes in body composition over 12 months of intervention were not statistically-significant

195 among the three groups. However, change in 25 OHD levels over 12 months of intervention was statistically significant between the three groups, as it was increased more in the supplement group compared to the other two groups (38.16 ± 31.34 nmol/L in supplement group vs 11.61±25.23 nmol/L in the control group and 7.20 ± 25.70 nmol/L in the behavioural intervention group). Serum 25 OHD levels also increased significantly in the behavioural intervention group compared to the control group (p = 0.028). Before and after intervention comparisons in each group are shown in Table 6.3. Body composition was not significantly changed in the control or supplement group after 12 months of intervention.

However, in the behavioural intervention group trunk fat to total fat ratio decreased significantly after 12 months (z statistic: -3.000, p-value = 0.003). Other body composition components did not change significantly in the behavioural intervention group.

To evaluate the association of changes in 25 OHD levels and body composition independent of group allocation, all the three groups were combined together. The association between change in 25 OHD levels and change in body composition over 12 months, independent from the group allocation, are shown in Table 6.4. Regression analysis revealed no significant associations between the increase in 25 OHD levels and total fat mass change, total fat percent change, visceral fat mass change, visceral fat percent change, leg fat mass change and visceral to total fat mass ratio change over 12 months. However, increase in 25 OHD levels were negatively associated with trunk fat mass change (p = 0.029) and trunk fat to total fat ratio change (p = 0.040) over 12 months. Sun exposure data are presented in more detail in chapter 7.

196

Table 6.1: The baseline 25 OHD and body composition of study participants Behavioural group Supplement group Control group p-value2

(n=40) (n=41) (n=42)

25 OHD (nmol/L) 52.1±12.0 53.0±13.6 54.1±14.3 0.84

Total fat mass (grams) 20459.6 15890.4 19253.3 0.38 (17086.10, 26324.05) (13597.68, 27353.64) (12284.97, 28914.64)

Total fat % 31.84 28.63 29.05 0.79 (26.33, 36.55) (24.84, 34.38) (21.85, 38.92)

Trunk fat mass (grams) 8178.41 5946.12 7137.46 0.38 (5893.94, 10957.14) (4887.72, 13255.53) (4707.07, 13027.55)

Visceral fat mass 1933.00 1458.4 1767.80 0.39 (grams) (1389.20, 2645.70) (1047.70, 2696.50) (1122.77, 2958.72)

Visceral fat % 2.86 (2.30, 3.61) 2.65 (1.92, 3.62) 2.70 (2.01, 3.89) 0.68

Leg fat mass (grams) 10221.08 8235.78 10491.56 0.50 (7806.09, 12075.25) (6769.69, 12171.66) (6873.36, 13905.54)

Visceral fat to total fat 0.09 (0.08, 0.10) 0.08 (0.07, 0.10) 0.09 (0.08, 0.10) 0.75 ratio Trunk fat to total fat 0.39 (0.37, 0.43) 0.38 (0.35, 0.41) 0.38 (0.35, 0.42) 0.55 ratio

197

1Data are Mean±SD or median (interquartile range). 2P-value of the differences between the three groups by Kruskal Wallis, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D

198

Table 6.2: Mean ±SD changes in 25 OHD and body composition over 12 months intervention 1

Behavioural group Supplement group Control group p-value2

(n=40) (n=41) (n=42)

25 OHD (nmol/L) 11.6±25.2* 38.1±31.3* 7.2±25.7 <0.001 Total fat mass (grams) -864±3729 -382±3124 -765±5007 0.918

Total fat % -1.1±3.1 -0.7±3.2 -1.5±3.9 0.796

Trunk fat mass (grams) -968±2205 -857±2970 -835±3042 0.873

Visceral fat mass (grams) 50±552 84±495 -43±694 0.626

Visceral fat % 0.10±0.60 0.08±0.61 -0.04±0.61 0.562

Leg fat mass (grams) -82±1437 87±1325 -97±1834 0.673

Visceral fat to total fat ratio 0.007±0.020 0.006±0.019 0.006±0.019 0.926

Trunk fat to total fat ratio -0.038±0.076 -0.038±0.096 -0.036±0.086 0.582

1 Data are mean change from baseline to 12 months intervention ±SD based on intention to treat 2P-value of the differences between the three groups by Kruskal Wallis, p < 0.05 considered as statistically significant *Statistically significant compared to control group, using post hoc Tukey test 25 OHD: 25 hydroxyvitamin D

199

Table 6.3: Paired t-test or Wilcoxon comparing 25 OHD levels and body composition prior to and following 12 months intervention 1

Behavioural group Supplement group Control group

(n=40) (n=41) (n=42)

t or z statistic p-value2 t or z statistic p-value2 t or z statistic p-value2

25 OHD 2.5 0.010 1.5 0.140 7.3 <0.001

Total fat mass -0.82 0.412 -0.72 0.466 -0.70 0.481

Total fat % -1.41 0.158 -0.16 0.092 -1.21 0.224

Trunk fat mass -1.65 0.098 -1.07 0.285 -1.07 0.281

Visceral fat mass -0.79 0.427 -0.43 0.665 -1.03 0.299

Visceral fat % -0.87 0.382 0.56 0.569 -0.49 0.624

Leg fat mass 0.04 0.968 -0.22 0.820 -0.60 0.544

Visceral fat to total fat ratio -1.57 0.115 -1.11 0.265 -1.05 0.290

Trunk fat to total fat ratio -3.00 0.003 -1.93 0.053 -1.92 0.055

1 Data are t-statistic or z-statistic based on intention to treat 2P-value by paired t-test or Wilcoxon, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D

200

Table 6.4: Association of change in 25 OHD levels and change in body composition over 12 months, independent of group allocation

1 Change in 25 OHD B 95 % CI p-value

Total fat mass 16.88 -25.77 to 59.54 0.433

Total fat % 0.025 0.175 to 0.063 0.175

Trunk fat mass -32.34 3.42 to 61.26 0.029

Visceral fat mass 1.08 -5.18 to 7.34 0.733

Visceral fat % 0.002 -0.005 to 0.008 0.566

Leg fat mass -2.55 -19.04 to 13.92 0.758

Visceral fat to total fat -0.000003 -0.000214 to 0.000207 0.976 ratio

Trunk fat to total fat -0.001 0.00004 to 0.00186 0.040 ratio

1P-value by linear regression analysis adjusted for group allocation, p < 0.05 considered as statistically significant 25 OHD: 25 hydroxyvitamin D

201

Figure 6.1 Usual whole body DXA scan

Figure 6.2 Visceral fat analyses on whole body DXA scan

202

Figure 6.3 Gynoid fat analyses on whole body DXA scan

203

6.5 Discussion

In this single-blinded, open-labelled randomised controlled trial after one-year intervention, serum 25 OHD levels increased significantly in both the supplement group and the behavioural intervention group, but no significant changes in total fat mass, total fat percent, visceral fat mass, visceral fat percent, trunk fat mass, trunk fat percent, visceral to total fat ratio were observed. The trunk to total fat ratio was significantly decreased (10%) only in the behavioural intervention group, though the decrease approached significance in the other two groups.

Observational studies have shown a relationship between 25 OHD levels and excess fat mass or visceral fat mass [Vogt et al., 2016 and Hao et al., 2014]. However, whether 25 OHD level improvements could affect body composition including total fat mass, visceral fat mass or trunk fat mass has been unclear. In 2012, Salehpour et al. showed in a clinical trial on 77 overweight and obese women that vitamin D supplementation for 12 weeks led to body fat mass reduction (-2.7±2.1 kg in the supplement group vs. -0.47±2.1 kg in the control group; p

< 0.001) [Salehpour et al., 2012]. Another clinical trial (Dong et al. in 2010) on 49 African-

American boys and girls showed that total body fat mass was inversely associated with 25

OHD levels in response to 2000 IU per day vitamin D supplementation for 16 weeks [Dong et al., 2010]. They concluded that plasma 25 OHD levels in response to the supplementation were negatively modulated by adiposity. Major et al. in a clinical trial assessed the effects of a weight loss program combined with vitamin D and calcium supplementation compared to the weight loss program alone. They showed that adding supplementation of 200 IU per day vitamin D plus 1200 mg elemental calcium to the weight loss program did not increase fat mass loss [Major et al., 2007]. Another clinical trial in line with our findings was conducted by Wamberg et al. in 2013. They assessed the effects of 7000 IU per day vitamin D supplementation on body composition among 52 people aged 18 to 50 years and with BMI

204 more than 30 kg/m2. They reported that 25 OHD levels significantly increased after 26 weeks of taking the supplement but it had no effects on total body fat mass or visceral fat mass

[Strazdienelow et al., 2012]. Very few studies in this area showed significant effects of vitamin D plus calcium supplementation in reduction on fat mass or visceral fat mass. For example a clinical trial by Rosenblum et al. evaluated the effects of consuming vitamin D plus calcium-supplemented orange juice on weight reduction and visceral fat reduction over

16 weeks. They showed that the average weight loss did not differ significantly between groups, but visceral fat was reduced more in the vitamin D plus calcium group compared to the control [Rosenblum et al., 2012].

These inconsistent results could be due to the heterogeneity in doses and types of vitamin D supplements, prevalence of vitamin D deficiency, duration of follow-up, and the fact that most previous studies assessed effects of vitamin D combined with calcium on body composition. Our clinical trial is, to the best of our knowledge, the first study which has evaluated the effects of vitamin D supplementation and a smart-phone-based behavioural intervention on improving vitamin D levels and their effects on body composition among young healthy women.

A number of mechanisms have been proposed to explain possible effects of vitamin D on body composition, and vice versa. First, the bioavailability of vitamin D potentially is decreased in people with excess fat mass because of the sequestration of vitamin D in adipose tissue [Wortsman et al., 2000]. Second, excess fat and obesity increase systemic inflammation which can decrease the 25 OHD levels by increasing the clearance of vitamin D

[Compher et al., 2008]. In the other direction, low vitamin D levels can increase PTH levels which may affect lipid storage [McCarty et al., 2003]. Moreover, 1,25(OH)2D was found to regulate expression of some genes involved in synthesis of adipose tissue [Kong et al, 2006].

Finally, 1,25(OH)2D bound to the vitamin D receptor in muscle tissue activates muscle

205 growth and improves muscle function and may thereby influence overall body composition

[Boland et al., 1986].

Strengths of this project include the collection of numerous population characteristics, as well as the single-blind randomised clinical trial design of the study which allows for causative conclusions. Moreover, measurements of body composition using DXA scan provides an accurate method of body composition measurement [Pritchard et al., 1993]. Another strength is the very high compliance in the supplement group.

The present study has some potential limitations. First, our results may not be generalizable to men or older adults as all our subjects were young women. Moreover, most of the participants were white so our results cannot be extrapolated to other ethnic groups. Finally, we were not able to include the gynoid region data in our analysis as the DXA Hologic software is not providing gynoid region data in the standard analysis to get the information on gynoid region customise delination of the original whole body scan is required.

In conclusion, increasing circulating 25 OHD levels did not lead to any significant changes in total fat mass, total fat percent, visceral fat mass, visceral fat percent, trunk fat mass, trunk fat percent, or visceral to total fat ratio after a one-year intervention. Only the trunk to total fat ratio was significantly decreased in the app group. Improvement in 25 OHD levels does not appear to affect body composition in young healthy women.

206

Chapter 7

SUN-RELATED BEHAVIOURS AMONG YOUNG AUSTRALIAN WOMEN AND EFFECTS OF USING M-HEALTH ON CHANGE IN SUN PROTECTION BEHAVIOUR

207

7.1 Introduction

Production of vitamin D in the body is highly dependent on sun exposure. It is estimated that

80 to 90 percent of 25 OHD levels are from UV-induced production [Holick et al., 2007].

However, excessive exposure to sun light and UV radiation is the major risk factor for all types of skin cancer. Sun-related behaviours include the amount of sun exposure during a day, protection while under direct sunlight, use of sunbeds or suntanning and attitude towards getting suntanned. Despite a good knowledge among young adults about the association of high levels of sun exposure and the risk of skin cancer, an improvement in knowledge and attitudes about sun tanning is still needed, especially among young girls [Day et al., 2014].

Having safe sun-related behaviours are important as they can affect both vitamin D status and risk of cancer, particularly among young adults. In the cross-sectional section of this chapter we evaluated aspects of sun-related behaviours among our study population, and further we evaluated if sun-related behaviours are associated with vitamin D status. In the randomised controlled trial part, we examined that whether behavioural intervention, using the Safe-D mobile application, can influence sun-related behaviours and affect vitamin D status and metabolic profiles.

7.1.1 Skin cancer

Skin cancer is one of the most common forms of cancer around the world [Apalla et al.,

2017]. Currently, between 2 and 3 million non-melanoma skin cancers and 132,000 melanoma skin cancers occur globally each year [World Health Organisation. 2017]. In

Western countries, skin cancer causes over 16,000 deaths annually. Australia has one of the highest rates of skin cancer in the world [Australian Institute of Health and Welfare. 2017], causing more than 2000 deaths in 2014 [AIHW. 2017]. Approximately two in three

Australians are diagnosed with skin cancer by the time they are reach 70 years, with more

208 than 750,000 people treated for one or more non-melanoma skin cancers in Australia each year [www.cancer.org.au/about-cancer/what-is-cancer/facts-and-figures.html]. Skin cancer occurs with abnormal growth of skin cells, most often due to exposure to the sun. There are three main types of skin cancer: squamous cell carcinoma, basal cell carcinoma, and the most dangerous form - melanoma [www.sunsmart.com.au/skin-cancer/skin-cancer-types]. Non- melanoma skin cancer is more common in men, with almost double the incidence compared to women. However, women have a higher incidence from the ages of 20 to 44 years compared to men [AIHW. 2017]. The economic burden of skin cancer in Australia is critically high [Cancer in Australia. 2017]. In 2005, approximately 260 million dollars were spent on non-melanoma skin cancer and 30 million dollars on melanoma [Cancer in

Australia. 2017], including hospital admission and care, medication and out-of-hospital medical expenses. Therefore, in Australia significant effort has gone into trying to prevent skin cancer and reduce the expense related to treatment.

The majority of skin cancers in Australia are caused by exposure to ultraviolet radiation (UV) in sunlight. UV appears to play a critical role in the development of skin cancer particularly during childhood and the adolescent years [Green et al., 2011]. Around 80% of melanoma and 90% of non-melanoma skin cancers are associated with UV exposure [Armstrong et al.,

1993]. Sunburn, tanning and the use of sunbeds are the main sources of overexposure to sunlight or UV radiation. During the last decades, efforts have been made to teach

Australians to ‘slip (slip on sun protective clothing that covers as much of the body as possible), slop (slop on SPF (sun protection factor) 30 or higher broad-spectrum, water- resistant sunscreen, at least 20 minutes before sun exposure), slap (slap on a broad-brimmed hat that shades face, neck and ears), seek (seek shade) and slide (slide on sunglasses)’ to protect themselves from sun exposure and prevent skin cancer [www.cancer.org.au].

209

7.1.2 Sun-related behaviour

The term ’sun-related behaviour’ refers to various activities and behaviours including time spent exposed to the sun and the amount of sun exposure, sun-protection, sun tanning and attitudes to sun tanning which we will discuss further in this chapter.

7.1.3 Sun exposure

There are different definitions for sun exposure. In 2005, Glanz et al. [Glanz et al., 2013] defined sun exposure as the number of hours spent out-of-doors between 10 am and 4 pm.

For this study, we examined sun exposure by asking the following four questions. How much time was spent in the sun on a typical day over a summer weekend? How much time was spent in the sun on a typical summer weekday? How much time was spent in the sun on a typical day in a winter weekend? How much time was spent in the sun on a typical winter weekday. People’s activities and patterns of sun exposure are different on weekends compared to weekdays, for example people may work indoors and experience most of their sun exposure during leisure time on weekends, while others work outside most weekdays and rest indoors on the weekends. Therefore, we asked separate questions pertaining to weekday and weekend sun exposure. Moreover, as UV index is very different in summer compared to winter, particularly in Victoria, Australia, we also divided the questions into sun exposure during summer and winter.

7.1.4 Sun protection behaviour

Sun protection behaviour was recently defined as avoiding direct sun exposure or using protective clothes, hat, sunglasses and applying sunscreen when exposed to direct sunlight for more than 15 minutes [www.cancer.org.au/preventing-cancer/sun-protection/preventing-skin- cancer]. Recommendations for sun protection behaviour have changed during the last decade

210 due to the availability of new products such as sunscreens with high SPF [Lazovich et al.,

2011]. Volkov et al. [Volkov et al., 2013] evaluated the trends in sun protection behaviour and sunburn among Australian adolescents and adults for seven years and indicated that sun protection behaviours during summer weekends were relatively low and adolescents used less sun protection than adults. It was also shown that one in eight adolescents were sunburnt on summer weekends between the years 2010 to 2011.

7.1.5 Suntanning behaviours

Tanning behaviour refers to purposefully exposing the skin to direct sunlight, using solariums or applying fake tan. We evaluated suntanning behaviour by asking the following questions.

During holidays, how often did you go out to get a suntan? Outside of holidays, how often did you go out to get a suntan? Do you like to get a suntan? Have you used a sunbed in the last 2 months?

Solaria, also known as sunbeds, are used to artificially-induce tanning. Solarium machines deliver UV radiation through sun lamps. However, it is not a safe way and increases the risk of skin cancer. It has been shown that 281 incidences of melanoma and 43 deaths occurred as a result of using solaria in Australia in 2008 [www.betterhealth.vic.gov.au].Commercial sunbeds have been banned in most of Australia since 2015. However, some illegal use is said to continue.

7.1.6 Fake suntan

A fake tan is when a chemical substance is applied on the skin surface to simulate a tan, and does not involve direct exposure to sunlight. Fake tan chemicals usually contain dihydroxyacetone which is approved as safe by the United States Food and Drug

Administration. However, using the chemical near the eyes or mouth is not approved and not

211 recommended. Fake tanning has recently become popular, especially among young people, as tanned skin is perceived to indicate good health and is fashionable. The prevalence of fake tanning has been reported to be around 8.7% of adults (age >18 years) in South Australia. It is a more common practice among younger adults, especially women, with 28% of young women aged 18-24 years old trying it at least once [Beckmann et al., 2001]. While there is no evidence that use fake tanning increase the risk of skin cancer, a survey conducted in

Victoria, Australia, in 1993 showed a higher prevalence of sunburn among those using fake tan (66% vs 46%) [Borland et al., 1992]. According to a study conducted in the US, using fake tanning was more frequent among older girls, those with a greater desire for a suntan and those with a positive attitude towards using tanning products. Moreover, the use of fake tanning was associated with a higher frequency of sunburn and higher sun exposure [Buller et al., 2012].

7.1.7 Suntan attitudes

Knowledge about the need for protection from sun exposure by itself is not sufficient to influence the sun-related behaviour of young people. Favourable attitudes, rather than awareness, are what determine changes in behaviour. According to the Cancer Council

Australia in 2014 around 38 % of Australian adolescents aged 12 to 17 years old liked to get a suntan [www.cancer.org.au/preventing-cancer/sun-protection/preventing-skin-cancer]. In

Europe in 2014, it was shown that the motivation and attitudes towards sun tanning were the barriers to achieving healthy sun protection behaviour and were mainly motivated by aesthetic reasons [Fernández-Morano et al., 2015]. Increasing public knowledge and changing attitudes towards sun tanning are the key goals of public health campaigns to improve sun protection behaviour, particularly among adolescents.

212

7.1.8 Balance between sun protection and receiving enough vitamin D

Vitamin D deficiency is another current health problem and affects at least one third of

Australians [Daly et al., 2012]. Finding a balance between producing enough vitamin D from sunlight exposure and protecting the skin from cancer caused by UV exposure has been causing a great deal of public confusion and has attracted research attention during the last decades. [Cancer Council Victoria, 2017]. It is important to have safe and sufficient sun exposure and practice proper ‘SunSmart’ behaviour to get adequate vitamin D and prevent skin cancer. SunSmart is a health promotion campaign in Australia that promotes a balance between the benefits and harms of sun light exposure [SunSmart, 2017].

7.1.9 Effects of using mobile-based application to improve sun-related behaviour

In the literature, there have been a few observational studies [Brock et al., 2013; Kim et al.,

2012] evaluating sun exposure and sun protection behaviours and their association with vitamin D levels. Also, a number of studies have tried to determine the optimal amount of sun exposure for each population to balance vitamin D status and the risk of skin cancer [Holick et al., 2008]. However, very few intervention studies exist that have evaluated improving knowledge of sun exposure or provided personalized information on how much sun exposure is enough. According to previous studies [Daly et al., 2012] and our findings in the cross- sectional part of this project [Tabesh et al., 2017], the level of sun exposure remains high among young Australian women. Moreover, despite a good knowledge among young adults about the association of high levels of sun exposure and the risks of skin cancer, knowledge and attitudes about sun tanning still need to be improved, especially among young girls [Day et al., 2014]. Gender and age are the best predictors of sun protective behaviour with females of increasing age demonstrating more protective behaviour [Starfelt et al., 2016]. It has been suggested that interventions that attempt to increase sun protection behaviour in young people

213 must include educational components. It is very important for individuals to know how much sun exposure is safe and if it is sufficient to produce adequate vitamin D levels in the body.

Providing personalized information on how much sun is likely to be enough and is safe is very important, and is dependent on skin type, skin colour, type of clothing, the use of sunscreen and the UV index. UV index is dependents on time of day, weather condition, season, latitude and altitude. In our study, we tried to improve sun-related behaviour by providing information about how much sun exposure was enough for each individual based on their skin type, skin colour, use of sunscreen, clothes, location and UV index and also by improving knowledge about sun exposure and vitamin D by sending information messages to participants through a Safe-D mobile application.

7.2 Literature review

7.2.1 Sun-related behaviour

Previous studies in this area have shown that younger people report far more recent sunburn than older people, highlighting the importance of sun protection education for school children, adolescents and young adults [Green et al., 2011]. It has also been shown that risky sun-protection behaviour was more common among girls than boys of the same age [Volkov et al., 2013]. Females were more likely than males to report that their peers believed a suntan was desirable and made them feel healthier and more attractive [Kyle et al., 2014]. Almost all the studies (e.g., a recent small Australian study [Cargill et al., 2013] and a larger Danish population-based study [Koster et al., 2017]) used validated questionnaires to evaluate sun- related behaviours. In 2017, Vuadens et al. [Vuadens et al., 2017] indicated that sun-related knowledge was high among Swiss school children, but it was not associated with sun protection behaviour and attitudes. A Scottish study which evaluated the sun protection behaviour of adolescents revealed poor habits, with 51% reporting being sunburnt the

214 previous summer. Sunscreen use was also infrequent with only one fifth of participants using it. The study suggested urgent action to reduce levels of sunburn among Scottish adolescents.

They also reported that females were more likely to sunbed, use fake tanning and get sunburnt than males [Kyle et al., 2014]. Some other studies also showed similar results; that risky sun behaviour was more common among female adolescents despite their greater awareness regarding skin cancer risk factors [Wright et al., 2017; Kyle et al., 2013].

Cockburn et al. investigated the association between knowledge, attitudes, beliefs and sun protection behaviour among school children in 1986 and 1987 in Australia. They found that just 30% of the sample students used sun-protection measures and that knowledge was not associated with better sun protection behaviour [Cockburn et al., 1989]. Fernandez-Morano in

2014 [Fernández-Morano et al., 2014] indicated the importance of positive attitudes about sun protection in order to achieve a change in the behaviour of adolescents. They showed that young people liked sunbathing and considered a tan synonymous with beauty and health.

Improving attitudes is therefore an important method of reducing the desire for a tan. In their study the boys thought it more worthwhile to use sunscreen even if they did not tan. This contrasted with to the fact that adolescent girls had more positive attitudes towards tanning than boys [Fernández-Morano et al., 2014]. In a 2009 study of 315 college students of

Midwestern University, USA, Dennis et al. showed that individuals had a strong desire to tan and participate in tanning regularly, even though they had good knowledge on the dangers of

UV radiation [Dennis et al., 2009]. Participants spent a substantial amount of time outdoors during peak sunlight (10am–4pm), yet often did not use sunscreen. Even though about 25 percent of the participants indicated that they often reapplied sunscreen while at the beach, this may not be the case when partaking in other outdoor activities where they are unaware they are being exposed to harmful UV radiation. The authors also reported that participants who had skin cancer or who had a friend or family member with skin cancer, were more

215 likely to have been sunburnt in the previous 12 months (87.3%) than participants without skin cancer or whose friends or family did not have skin cancer (57.8%, p < 0.001). Also, in the previous 12 months, participants who had skin cancer or a friend or family member with skin cancer were more likely to have sunbathed outside (66.7%) than those participants who did not have a friend or family member with skin cancer (48.9%, p = 0.008). The respondents in the above study did not report high levels of using protective clothing as defined by the

American Academy of Dermatology [Christoph et al., 2016]. Koster in a cross-sectional study found that 66% of the participants used sunscreen to prolong their time in the sun and sunburn became less frequent with increasing age (OR) 4.44; 15-19 years old vs. 50-59 years old) and skin type (OR 2.57; skin type I vs. skin type III) [Køster et al., 2010].

7.2.2 Sun-related behaviour and circulating 25 hydroxyvitamin D (25 OHD) levels

Sun protection is recommended to prevent skin cancer. However, very little evidence is available on sun-related behaviour among young Australians and its association with 25 OHD levels. Youl et al. in an observational study in Australia showed that from 2005 to 2009 the percentage of people believing that sun protection could lead to inadequate vitamin D levels had increased by 15%. People who agreed that protecting themselves from the sun’s rays led to inadequate vitamin D levels were more likely to use a sunbed or have positive attitudes about using sunbeds [Youl et al., 2009]. There is some evidence that sun protection behaviour, such as using sunscreen, causes vitamin D deficiency in healthy individuals.

Matsuoka et al. documented that in a laboratory situation, applying sunscreen to patches of normal human skin prevented UV-induced conversion of 7-dehydrocholesterol to pre-vitamin

D3 [Matsuoka et al., 1987]. However, in a randomized controlled clinical trial, Marks et al. demonstrated that sunscreen use with SPF-17 was sufficient to prevent actinic keratosis, but did not induce vitamin D insufficiency or deficiency [Marks et al., 1995]. A population study

216 in the US from 2003 to 2006 evaluated the association of sun protective behaviours and 25

OHD levels among 6000 adults aged 18 to 60 years old. This study showed that seeking shade and wearing long sleeve shirts were significantly associated with lower 25 OHD levels

(p < 0.001 and p = 0.001, respectively). These associations were stronger for Caucasians, and were not statistically significant among Hispanics or African Americans. Conversely, they demonstrated that frequent use of sunscreen and wearing hats were not associated with vitamin D deficiency [Wehner et al., 2012]. There are few studies evaluating the relationship between specific sun protective behaviours such as seeking shade, using sunscreen, wearing hats or sunglasses, and vitamin D levels in the general population, especially in Australia.

7.2.3 Effects of using mobile-based applications to improve sun-related behaviour

Most clinical trials in this field have focussed on young adults or adolescents. Some studies used face to face sessions to attempt to improve sun protection behaviour, while others used social media and technology such as sending text messages or emails about improving sun- related behaviour. It has been shown that Australians have good knowledge of the dangers of sun exposure; however, young people in particular engage in relatively few sun protection practices [Dobbinson et al., 2008]. Previous public health campaigns have relied largely on media such as television and print-based advertisements, and have been designed based on social-cognitive theories of health behaviour change [Myers et al., 2006]. More recently, research has been focussed on better ways to bridge the gap between knowledge, attitudes and actual behaviour.

New modes of communication, such as mobile phones or email, allow health promotion messages to be timed and individualised towards the user, and can be delivered flexibly and on demand. Australia has one of the highest rates of mobile phone ownership in the world: therefore access to a mobile phone is almost ubiquitous [Mackay et al., 2011]. Mobile

217 messaging and self-management phone applications have already been successful in delivering health interventions and knowledge improvement about healthy lifestyle [Daniel et al., 2015]. However, very few studies have evaluated the effects of using mobile-based applications to change attitudes or actual behaviours.

Sachse et al. in 2016 evaluated the effects of face to face sun protection training and sending text messages to improve sun protection behaviour in adolescents who received organ transplants. The study showed that text messaging recommendations for sun protection was technically feasible and well accepted by adolescents. Further, 81% of participants demonstrated good knowledge of the dangers of sun exposure and skin cancers. However, only 50% used sunscreen and only 60% used a hat or cap as sun protection [Sachse et al.,

2016]. Mair et al. studied the acceptability and feasibility of delivering sun protection text messages to people aged 18 to 40 years old and found that 80% of participants agreed to receive some form of sun protection advice. Greater use of sun protection was associated with higher willingness to receive electronic message about UV. Careful attention to message framing and timing of message delivery, as well as a focus on the short-term effects of sun exposure, such as sunburn, and long-term effects such as skin ageing, would likely increase the effectiveness of such messages to young people [Mair et al., 2012]. In a randomised clinical trial, Armstrong et al. evaluated the effectiveness of using text messages as reminders for sunscreen application on 70 participants aged over 18 years old. The results were that adherence to daily sunscreen application amongst participants who received text message reminders was almost twice that in the control group, and the results remained significant after adjusting for daily weather patterns [Armstrong et al., 2009]. In a trial by Buller et al. in

2012 on 604 adults the effectiveness of a mobile application on sun protection was assessed.

The application provided personalised, real time advice on sun protection. They showed that

218 those who used the mobile app reported spending less time in the sun and also seeking shade more than the control group [Buller et al., 2015].

Another clinical trial with 1,589 adolescents aged between 11 and 18 years evaluated the effects of receiving a physician consultation on sun protection behaviour, and resulted in participants being significantly more likely to report regular sunscreen use (43%) and intermittent wide-brimmed hat use (15%) compared to those who did not receive counselling

(30% and 9%, respectively). The counselling did not advise avoiding peak sun exposure, seeking the shade, or wearing protective clothing. The consultations had significantly positive associations with regular sunscreen specific practices, including using SPF 15+ sunscreen on the face and body when at the beach or pool, and sunscreen reapplication when in the sun all day. Parents who received counselling were also more likely to regularly insist on summer sunscreen use for their children (35%) compared to those who did not receive counselling

(26%), but this relationship was not evident for other rules, including insistence on wearing shirts, hats or staying in the shade or under an umbrella [Cokkinides et al., 2010]. A systematic review in 2017 of 19 studies evaluated the effects of texting and mobile-based applications to improve disease-prevention behaviour, including contraceptive use, physical activity, smoking cessation and sun protection behaviours among adolescents. Despite the promising feasibility and acceptability of text messaging and mobile phone apps in improving preventive health behaviours among adolescents, the overall findings were modest in terms of efficacy [Badawy et al., 2017].

The Safe-D app was produced based on the SunSmart app. SunSmart app was produced by

Cancer Council Victoria to prevent skin cancer by providing daily alert on UV radiation and how much and when sun protection is needed and when it's safe to get some sun for vitamin

219

7.3 Significance and Aims

Given the high prevalence of melanoma and non-melanoma skin cancer in Australia and the fact that a history of sunburn and cumulative sun exposure during childhood and adolescence are the most important risk factors for the development of skin cancer, the need to find a safe balance between sufficient sun exposure for vitamin D synthesis versus avoiding an increased risk of skin cancer led us to undertake this project. Moreover, sun protection behaviour still needs to be improved among adolescents, especially girls, as they are particularly at risk for risky sun-related behaviours. There is lack of studies in this area among young Australian women. There are very few studies that address the relationship between specific sun- protective behaviours and 25 OHD levels in the general population [Glanz et al., 2008]. It is still unclear whether typical usage of sunscreen may result in vitamin D deficiency [Samanek et al., 2006 and Callister et al., 2011]. Even less is known about the impact that other recommended sun protective behaviours, such as seeking shade, wearing long sleeves, and wearing hats or sunglasses, have on 25 OHD levels. Furthermore, data on the prevalence of different sun protection behaviours in Australia are limited. Therefore, in the cross-sectional part of this project we aimed to evaluate the prevalence of various sun-related behaviours, including time spent in the sun, sun protection behaviour, sunbed behaviour and suntan attitudes, of young women living in Victoria, Australia. We also aimed to investigate the associations of each sun-related behaviour with 25 OHD levels.

In the intervention part of the project, we aimed to evaluate the effects of using a mobile- based application on changes in sun exposure, attitudes to suntans and sunbathing behaviours over a 12 months intervention.

220

7.4 Methods and Materials

All information on study design and methods and material information were presented in detail in Chapter 2. Serum 25 OHD levels were measured in fasted blood by using the chemiluminescent microparticle immunoassay in the Melbourne Health Pathology laboratory.

Sun-related behaviours were assessed using an online survey questionnaire at baseline, after 4 months and after 12 months of intervention. We used validated, sun behaviour-related questionnaires [Sneyd et al. 2006]. Sun-related questions were categorised into sun exposure, sun protection behaviour, sunbathing behaviour and suntan attitude sections.

1- Sun exposure information was collected by asking the following questions:

a) How much time would you spend in the sun on a typical day on a summer weekend?

b) How much time would you spend in the sun on a typical summer weekday?

c) How much time would you spend in the sun on a typical day on a winter weekend?

d) How much time would you spend in the sun on a typical winter weekday?

Participants could select from six responses (less than 30 minutes, 30 to 60 minutes, 1-2 hours, 2-3 hours, 3-4 hours or more than 4 hours). For the analysis the responses were combined into three categories (less than 60 minutes, 1-3 hours and more than 3 hours).

2- Sun protection behaviour was assessed by asking the following questions:

a) In summer, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you seek shade or avoid direct sun?

b) In summer, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you cover your head?

c) In summer, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you wear clothes, such as a long sleeve shirt, to protect your skin from

the sun?

221

d) In summer, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you wear sunglasses?

e) In summer, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you use sunscreen on any part of your body that is not covered up?

f) In winter, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you seek shade or avoid direct sun?

g) In winter, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you cover your head?

h) In winter, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you wear clothes, such as a long sleeve shirt, to protect your skin from

the sun?

i) In winter, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you wear sunglasses?

j) In winter, when you are in the sun for 15 minutes or more between 11 am and 4 pm,

how often do you use sunscreen on any part of your body that is not covered up?

There were seven response categories for each question (always, often, sometimes, rarely, never, I don’t know or not applicable). Each response was given a score (I don’t know or not applicable=0, always=1, often 2, sometimes=3, rarely=4, and never=5) and scores for all ten questions were summed to calculate the sun protection scores with a minimum of zero and a maximum of 50 points. For the statistical analysis, we combined the responses into four categories (always/often, sometimes, rarely/never, I don’t know/not applicable).

3- Sun tanning behaviour was assessed by asking the following questions:

a) During holidays how often did you go out in the sun to get suntan? (daily, 2-3 times a

week, once a week, less often, not during my holiday)

222

b) Outside of your holiday how often did you go out in the sun to get a suntan during this

summer? (daily, 2-3 times a week, once a week, every 2 weeks, monthly, less often,

not during this summer)

c) Do you like to go out in the sun to get suntan? (yes or no)

d) Have you used a sunbed in the last two months? (yes or no),

e) How often have you used a sunbed during last two months? (daily, weekly, every 2

weeks, monthly, less often).

The sun tanning behaviour score was calculated by adding up the five above-mentioned question scores with a minimum of zero and maximum of 18 points.

4- Suntan attitudes were evaluated by asking seven questions with seven responses categories for each question (strongly disagree, disagree, mildly disagree, neither agree nor disagree, mildly agree, strongly agree, can’t say/don’t know). Questions were as follows:

a) I feel healthier with a suntan.

b) A suntan makes me feel more attractive to others.

c) This coming summer I intend to use a sunbed regularly to get a suntan.

d) Most of my friends think a suntan is a good thing.

e) A suntan makes me feel better about myself.

f) Most of my close family think that a suntan is a good thing.

g) A suntan protects you against melanoma and other skin cancers.

For the analysis we combined the responses into four categories (strongly disagree/disagree/mildly disagree, neither agree nor disagree, mildly agree/strongly agree, can’t say/don’t know). To calculate the suntan attitude score we summed the scores from the above mentioned questions (strongly disagree=1, disagree=2, mildly disagree=3, neither agree nor disagree=4, mildly agree=5 and strongly agree=6), with a maximum of 42 points.

223

Data on skin type, physical activity, ethnicity and sensitivity to sunlight were also collected from an online questionnaire. Participants were asked to select the closest match to their skin type from following choices: reddish, pale, olive, brown and dark brown. Skin sensitivity to sun exposure was evaluated by asking participants to choose one of the following options: your face is very sensitive, sensitive, sometimes sensitive, resistant or never had a problem with sun. Ethnicity was assessed by asking participants “what is your ancestry? Consider the origins of your parents and grandparents”.

All the questionnaires were completed at baseline (cross-sectional part), after 4 months’ follow-up and at the end of the study (at 12 months follow-up).

Objective UV exposure was collected by asking participants to wear a sun monitoring watch

(dosimeter) for 14 consecutive days before each site visit. Detailed methodology provided in chapter 2.

For the intervention section, participants were randomly assigned to receive vitamin D supplements, a mobile-based application as a behavioural intervention group, or a control group. Those in the app group received instructions to download the Safe-D application on their smart phone. Details on the Safe-D application are presented in chapter 2. The app delivered advice each morning about how to obtain safe (as per the SunSmart guidelines) and adequate sun exposure during that day, including an estimate of the required sun exposure necessary for adequate vitamin D production; it would also send push notifications each morning to provide information about sun protection behaviours and vitamin D. Changes in each sun-related behaviour were measured in the app group and compared with the other two groups after 12 months.

224

7.5 Statistical analysis

Analysis methods involved descriptive statistics, including frequencies and percentages. A

Chi-square test was used to compare changes in sun-related behaviour among the three groups at 12 months follow-up. McNemar's test was used to compare sun-related data at 12 months follow-up with baseline in each group. McNemar's test was used to compare the number of participants whose sun exposure decreased after 12 months follow-up in each group. Kruskal–Wallis test was used to see any differences in risky sun behaviour scores or suntan protection behaviour scores, among the three groups at 12 months follow-up. The

Wilcoxon signed rank test was used to evaluate the mean differences in risky sun behaviour scores or suntan protection behaviour scores over the 12 months in each group.

As the suntan attitude score was normally distributed, ANOVA was used to evaluate differences among the three intervention groups at the 12 months follow-up. To compare changes in each intervention groups with the control group post hoc Tukey test was used. To see any changes in suntan attitude scores from baseline in each group, a paired t-test was used. Any association of vitamin D with changes in sun-related behaviour over the 12 months was evaluated by using Spearman correlation and regression analysis. Confounders including skin type, baseline physical activity and BMI were adjusted. The association of sun exposure with ethnicity, skin type, physical activity and skin sensitivity to UV, the association of 25

OHD levels and UV exposure, and the association of 25 OHD and sun protection behaviour score were assessed using the Spearman correlation or Chi square test. The responses of

“can’t say” or “I don’t know” were excluded from the analysis. Bonferroni correction was used to counteract the problem of multiple comparisons.

225

7.6 Results

In the cross-sectional part of the study, sun-related data and 25 OHD levels were generated for 449 participants. Median (IQR) of serum 25 OHD levels was 64 (46-86) nmol/L. Median

(IQR) of BMI was 22.9 (20.9-25.6) kg/cm2. In total, 85% of participants were born in

Australia and 47.8% of participants reported having a sensitive or very sensitive skin to the sun. Most of participants had pale skin type (62.6%), 31.0 % had beige or olive skin type, 3.8

% had brown, 2.0 % had reddish and 0.7 % had dark brown skin type. The amounts of sun exposure and the prevalence of sun protection behaviours are shown in Tables 7.1 and 7.2.

Participants spent more time in the sun on weekends than on weekdays, and also spent more time in the sun in winter than in summer. During summer, around half of the participants never, or rarely, wore hats or long sleeve shirts to protect themselves from the sun; 58.6% would seek shade and 68.4% used sunglasses always or often. The results indicated that people used sunscreen or sought shade more than they wore hats or covered dress to protect themselves from the sun in summer. In winter, sun protection behaviour was different from summer, with only 6.2% of participants seeking shade and only 11.4% always or often using sunglasses. The percentages of participants who always or often covered their heads or wore covered clothing in winter were also low (5.8% and 38.5%, respectively). The sun tanning behaviour data are shown in Table 7.3. In total, 37.4% of participants liked to get a suntan.

However, only two participants (0.4%) had used sunbeds during the previous two months, and one of them was a weekly user. The details of attitudes toward tanning are presented in

Table 7.4. More than half of the participants (55.0%) thought that a suntan made them feel more attractive and 51% considered that their friends thought a suntan was a good thing.

Around 47% agreed that a suntan made them feel better about themselves. Ninety-seven participants (21.6%) intended to use a sunbed regularly during the upcoming summer. Only

31.8% stated that they felt healthier with a suntan and 20.5% stated that most of their family

226 thought a suntan was a good thing. Almost all of the participants (94.4%) disagreed that suntans protected them from melanoma or other skin cancers, which shows that participants had some basic knowledge about skin cancer.

In total, 123 participants were recruited into the clinical trial part of the study. Median (IQR) of age for participants recruited to the trial part was 23 (21-24) years. Almost 17.1% were obese (BMI>30 kg/m2) and 39 % had central obesity (waist circumferences above 80 cm). In total, 87.3 % of participants were born in Australia and 68% reported their ethnicity as

European. Moreover, 48.4% of participants reported their skin was sensitive or very sensitive skin to the sun. Most of participants had pale skin type (64.8%), 29.5% had beige or olive skin type, 2.5 % had brown, 2.5 % had reddish and 0.8 had dark brown skin type. General characteristics of participants were not significantly different among participants recruited to the cross-sectional part and those recruited to the intervention study. Sun related behaviour comparison between participants in the cross-sectional group and those recruited to the trial are presented in Table 7.5. No significant differences were observed between those recruited into the cross-sectional part and those recruited to the trial part in terms of sun-related risky behaviours, suntan protection behaviour score and suntan attitude score. At the 4 months follow-up, data were obtained from 95 participants and at the 12 months follow-up visit, data from 80 participants were available. Table 7.6 shows the amount of time participants spent in the sun in summer and winter, i.e. <60 minutes, 1 - 3 hours or more than 3 hours per day, for each intervention group at baseline, 4 months follow-up and 12 months follow-up. The results showed that on a summer’s weekend, the proportion of participants who spent less than 60 minutes in the sun was 32.8% at baseline, 40% at the 4 months follow-up and 36.2% at the 12 months follow-up. How much sun exposure was changed among the participants in each group over 12 months is shown in Table 7.7. The amount of sun exposure increased in

16.5% of participants during the 12 months follow-up, decreased in 31.6% of participants and

227 showed no significant changes in 51.9 % of participants. However, there were no significant differences among the three arms at 12 months’ follow-up in terms of sun exposure. When comparing the sun exposure changes from baseline to 12 months in each group, no significant differences were observed in the control group after 12 months. In the app intervention group, time spent in the sun on winter weekends decreased significantly (p = 0.033); however, there were no significant differences in time spent in the sun on summer or winter weekdays or in summer on weekends in this group. In the supplement group, over 12 months there was a decrease in time spent in the sun on summer weekdays and on winter weekdays (p = 0.032 and p = 0.032, respectively).

Sun protection behaviour data are shown in Table 7.8 and include the following variables: seeking shade, covering head, wearing clothes to protect from sun, wearing sunglasses and using sunscreen, for both summer and winter seasons. Sun risky behaviour scores in each group are shown in Table 7.9. The risky sun behaviour score was calculated by adding the scores of each of the previously mentioned variables (with a score of 1 for ‘always protect’ and 5 for ‘never’). The median and interquartile range of risky sun behaviour is presented in

Table 7.9. At baseline, significant differences were observed for the three arms in the risky sun behaviour score (p = 0.045), with a higher score (e.g more risky behaviour) in the app group compared to the control or supplement groups. At 4 months and 12 months follow-up visits there were no significant differences among the three groups in terms of risky sun behaviour scores (p = 0.690 in summer at 4 months, and p = 0.947 in winter at 4 months, p =

0.924 in summer at 12 months and p = 0.793 in winter in 12 months).

Mean changes in the risky sun behaviour score at the 12 months follow-up in each group are presented in Table 7.10. In the app group, risky sun behaviour decreased significantly in summer after 12 months (-1.41±3.03, p = 0.033); however, the decrease was not statistically significant in winter for this group (-1.36±3.26, p = 0.078). For the control group, the mean of

228 risky sun behaviour tended to increase, but it was not statistically significant (p = 0.163 for summer and p = 0.178 for winter). In the supplement group, risky sun behaviour also tended to increase after 12 months but again it was not statistically significant, either in summer or in winter (p = 0.737 in summer and p = 0.225 in winter).

The results of the suntan behaviour analysis are shown in Table 7.11. At baseline, 73 participants had holidayed the previous summer and 45 (68.5%) of them answered never, or less often to the question on how many times they use a sunbed during their holiday.

Approximately 31% of participants said that they liked to get a suntan. At baseline, 4 months or 12 months follow-up, none of the participants had used sunbeds during the previous 2 months. Table 7.12 shows suntan behaviour scores, divided by groups. The mean and interquartile range for suntan behaviour scores at baseline, 4 months and 12 months follow- ups are as follows: 11.0 (7.5-13.0); 11.5 (8.0-13.0), 11.0 (9.0-13.0), respectively. No significant differences were observed among the three groups at baseline, 4 months or 12 months follow-ups in suntan behaviour (Table 7.13).

The suntan attitudes data are shown in Tables 7.14 and 7.15. No significant differences were observed in the mean ± standard deviation of suntan attitude scores among the three groups at each time point (p=0.66). The changes in the mean of the suntan attitudes over 12 months in the app group was 1.12±8.84, in the control group 0.10±10.75 and in the supplement group -1.85±12.22. No significant changes were observed in any of the groups over 12 months (p = 0.61 for app group, p = 0.96 for control group and p = 0.49 for supplement group).

The association of 25 OHD and sun-related behaviour data is shown in Table 7.16. In the crude model, changes in 25 OHD levels over 12 months were negatively associated with changes in risky sun behaviour in winter (p = 0.022). It means higher 25 OHD levels were

229 associated with less risky sun behaviour. Moreover, higher 25 OHD levels were associated with a lower suntan attitude score in the control group (p = 0.033) and less sun risky behaviour score in the supplement group (p = 0.045). No significant association was found between 25 OHD levels and sun risky behaviour score, suntan protection score or suntan attitude score in the behavioural intervention group. After adjustment for skin type, baseline physical activity and BMI, the negative association of 25 OHD and risky sun behaviour score change in winter and suntan attitude score changes in the control group, remained statistically significant (p = 0.038 and p = 0.035, respectively). However, the association of 25 OHD with risky sun behaviour score changes in winter in the supplement group was not significant after adjustment for confounders (p = 0.050). In the control group, after adjustment for confounders the positive association of 25 OHD with suntan protection behaviour scores became statistically significant (p = 0.006). No significant associations were observed between 25 OHD levels and changes in risky sun behaviour in summer or suntan protection behaviour scores in any of the intervention groups.

The association of objective sun exposure (obtained from UV dosimeter), with skin type, ethnicity, physical activity and skin sensitivity to sunlight are presented in Table 7.17. In our study, no significant association was observed between UV exposure and different skin types

(p = 0.332), ethnicity (p = 0.642) or skin sensitivity to sunlight (p = 0.331). However, a significant positive association was observed between sun exposure and physical activity (p =

0.026).

The association of 25 OHD levels and objective UV exposure is shown in Table 7.18.

Significant positive association was observed between 25 OHD and UV exposure only in the app group (p = 0.013). This association remained statistically significant after adjustment for skin type, baseline physical activity and BMI. Such associations were not observed in the supplement or control group.

230

The associations between 25 OHD levels and each parameter of sun protection behaviour are shown in Table 7.19. No significant associations were observed between 25 OHD levels and seeking shade, wearing hats or protective clothes, using sunglasses or sunscreen, in summer or in the winter time.

After 12 months of intervention, no significant differences in sun exposure were observed among the three groups. However, sun exposure in winter significantly decreased in the app group after 12 months, and decreased in the supplement group in both the summer and winter time when compared to baseline.

Sensitivity analyses were also conducted on the per protocol dataset, including all participants with no protocol deviations and who completed all measurements for the primary variables to ensure robustness of findings. Per protocol analysis showed a similar result so we only reported the ITT analysis in detail.

7.7 Discussion

7.7.1 Sun exposure

It is recommended that in Australia during summer, 2 to 14 minutes of sun exposure, three to four times per week at midday is safe and will provide fair-skinned people with the recommended amount of vitamin D [Cancer Council Victoria, 2017]. Kimlin et al. in an observational study in Melbourne, Australia, evaluated the impact of the current Australian sun exposure guideline to provide adequate 25 OHD levels. They have shown that in summer; almost all of the participants could achieve the recommended amount of sun exposure which was likely to result in adequate 25 OHD levels. However, in winter, participants were much less likely to achieve the recommended amount of sun exposure.

231

Even if this occurred, it was unlikely to result in adequate 25 OHD levels [Kimlin et al.,

2016].

So, in peak summer between 10am and 3pm, it is recommended that people should use protection against the sun by seeking shade, wearing hats, suitable protective cloths and sunglasses, and applying sunscreen, if they are in the sun for more than 30 minutes

[www.cancer.org.au/preventing-cancer/sun-protection/preventing-skin-cancer/].

In Southern Australia, where our study was performed, the UV index varies considerably in different seasons and it is usually lower than northern Australia. Sun protection may not be necessary in on most days in this region in late autumn and winter. However, during summer, the UV index reached above 3 and sun protection was recommended [www.cancer.org.au].

Our findings show that in summer, only 10.8% (7.3% on weekends and 14.3% on weekdays) of participants spent less than 30 minutes in the sun, which indicates the necessity of sun protection and SunSmart behaviour. However, in winter, 35.6% of participants spent less than

30 minutes in the sun which suggests that people may try to avoid sun exposure in summer more than in winter. This could be because during late autumn and winter there are fewer sunny days in Victoria, Australia, and people cover themselves more in winter due to unpleasant weather.

No associations were observed between sun exposure and ethnicity or skin types in our study.

This null association could be due to the fact that our study sample was quite homogeneous ethnically. Similar to our findings, Samanek et al. in 2006, documented the amount of beneficial and harmful sun exposure time over the year in selected Australian locations and found no association between sun exposure and ethnicity [Wright et al., 2005]. It has also been demonstrated that although skin colour and ethnicity could be variables strongly

232 associated with sun exposure or 25 OHD levels, this relationship is becoming weaker [Lawler et al., 2007].

Recently, it has been suggested that the effects of skin colour or ethnicity on sun exposure or health outcomes such as skin cancer and vitamin D status depend on a range of environmental and individual factors, in particular UV levels and personal attitudes and behaviours related to sun exposure, diet or exercise [Bränström et al., 2010].

We showed a significant association between sun exposure and physical activity. Many similar studies showed a close link between sun exposure and physical activity. Physical activity generally takes place outdoors which increases time spent in the sun. A population study showed that those who were inactive were more concerned about sun exposure and skin damage, which may have actually been acting as a barrier to being physically active.

However, active and inactive people tend to use different sun protection measures, with active people using higher levels of protection [Filiz et al., 2006].

Sensitivity and reactivity to UV radiation differs greatly between individuals and is mainly dependent on several phenotypic factors. In general, individuals with fair skin, red hair and freckles burn more easily in the sun than those with browner pigmented skin and dark hair

[Livingston et al., 2007]. However, individuals with high self-estimated skin UV sensitivity and those with blonde or red hair tend to be more cautious in the sun than those with low self- estimated skin UV sensitivity and dark hair [American Academy of Dermatology. 2017]. In our study, no significant associations were observed between sun exposure and self-reported

UV skin sensitivity. This null association could have been due to the fact that almost all of our participants had light skin colour, 62.4% had pale skin, and almost half reported to have a sensitive or very sensitive skin type.

233

7.7.2 Sun protection behaviour

According to the Australian Bureau of Statistics, in a total sample of more than 350,000,

86.5% of people aged 15 years and over wore hats as a sun protection measure during summer, while 85.1% wore sunglasses and 77.4% used sunscreen. An estimated 59,700 people sometimes used an umbrella during summer while an estimated 83,800 (23.8%) never deliberately avoided direct sunlight in summer [www.abs.gov.au/AUSSTATS/abs]. In our study, using sunglasses was the most common sun protection behaviour during summer

(68.4%). Wearing long sleeves or other protective clothing was the least common protective behaviour, with only 27% of participants practising this behaviour. This could have been because girls tend to wear short sleeve tops and skirts during summer in Australia which provides only moderate protection.

Those who wear hats have been reported to wear sunglasses more often and generally use other sun protection measures [Kyle et al., 2014]. These findings are in line with our study.

A study in Turkey showed that older students (≥17 years old) tended to get sunburnt more often, even if they used more sunscreen, which could have been because they spent more time at the beach [Matsuoka et al., 1987]. Livingston et al. evaluated the association between skin type, suntan preference and SunSmart behaviour among Australian adolescents over a 10 year period. They showed less compliance with sun protective measures in Australian children with a desire for darker skin [Norval et al., 2009]. Decreases in sun protection behaviour were demonstrated over time, regardless of tanning preference and skin type.

Although there was a reduction in the proportion of adolescents wanting a tan, it did not result in better sun protection outcomes. The desire for a darker tan was related to decreased use of SunSmart behaviour. The majority of adolescent students in that study rarely or never used all the three mentioned sun protection behaviours [Prichard et al., 2015].

234

Solarium use in our participants was low, with only 0.4% having used a sunbed in the previous 2 months. The same results were observed in a recent Australian study by Prichard et al. in 2015 which reported 7.1% solarium use among female university students in South

Australia [Prichard et al., 2015]. Another study in Queensland in 2009 reported 2.5% of participants used sunbeds in the previous 12 months [Gordon et al., 2012]. Studies showed a decrease in sunbed use from 2003 to 2007 [Francis et al., 2010]. This decline could be due to the effects of publicity surrounding the skin cancer risk associated with sunbed use and media attention on this matter [Norval et al., 2009]. It is worth mentioning that commercial solaria have been banned since 2015 which could have resulted in the very low reported solarium use.

7.7.3 Suntan attitudes

Young people like using sunbeds and consider a tan as synonymous with beauty and health.

Importance must therefore be given to improving attitudes, mainly by reducing the desire for a tan. In our study, 55% of participants thought that a suntan would make them more attractive. Other studies have shown similar results with a high percentage of participants believing a suntan to be a good thing, making them look healthier and more attractive. It has also been reported that adolescent girls have more positive attitudes towards tanning than boys [Wichstrøm et al., 1994]. However, in our study we could not evaluate suntan attitudes in boys, as all our participants were girls. It would seem though that attitudes towards sun tanning are a barrier to improving SunSmart behaviour. A recent study of the tanning practices of college students revealed that 83% of students reported that having a tan was very, or somewhat important [Dennis et al., 2009]. Our findings confirm that we still need to improve knowledge about tanning in young women.

235

7.7.4 Associations of 25 OHD levels and sun-related data

Sun exposure is necessary for vitamin D synthesis in the body and it has been proven that sun exposure is positively associated with 25 OHD levels [Nair et al., 2012]. However, it is still unclear how much sun exposure is required to achieve optimal balance between the risk of skin damage/skin cancer and vitamin D deficiency. We showed no significant associations between 25 OHD levels and seeking shade, wearing hats, sunglasses and protective clothing, or using sunscreen, in summer or winter. This finding suggests that sun protection behaviour and having SunSmart behaviour does not result in vitamin D reduction. However, our participants spent much longer in the sun than recommended by the Cancer Council of

Victoria.

There are unfortunately limited studies addressing the association between specific sun protective behaviours and 25 OHD levels in the general population [Marks et al., 1995;

Thompson et al., 1993]. Sunscreen use can reduce vitamin D production [Matsuoka et al.,

1987], but it is unclear how much usage of sunscreen results in deficiencies. Even less is known about the impact that other recommended sun protective behaviours (e.g., staying in the shade, wearing long sleeves and hats) have on vitamin D status.

Although in our study we could not see any association between 25 OHD levels and sun protection behaviours, in a study by Eleni Linos et al., staying in the shade and wearing long sleeves were significantly associated with lower 25 OHD levels, especially in Caucasian people, but wearing a hat and using sunscreen were not associated with lower 25 OHD levels.

These results remained significant even after adjustment for milk intake and supplement use, season of blood collection, race and sensitivity to sun [Linos et al., 2012].

Our findings are consistent with other studies in Australia which report that using sunscreen does not cause lower 25 OHD levels [www.cdc.gov].

236

Previous research has shown that darker skin colour is a significant predictor of 25 OHD

[Rockell et al., 2008]. Similar to our study, Dix et al. indicated that there was no relationship between sun exposure and serum 25 OHD concentrations in 50 participants aged over 18

[Dix et al., 2017].

7.7.5 Effects of using mobile based applications to improve sun-related behaviour

In the intervention part of the project, we found that 12 months of behavioural intervention could be effective in improving the sun protection behaviour in young women. However, the behavioural intervention had no significant effects on sunbathing behaviour or suntan attitudes compared to baseline or the two other intervention groups. Most similar intervention studies focused on improving people’s knowledge of sun exposure and sun protection behaviours. Girgis et al. in 1994 evaluated the effects of education sessions on sun protection behaviour, knowledge and attitudes of 142 outdoor workers. They randomised participants into two groups - one group attended 30-minute educational sessions; the other was the control group. They revealed a significant improvement in both knowledge and sun protection behaviour in the intervention group [Girgis et al., 1994]. The strength of the study was using diaries to record sun exposure and sun protection behaviours, while in our study we used a self-reported questionnaire which could cause recall bias. However, using diaries itself might influence sun-related behaviours.

Because it has been shown that the risk of skin cancer is associated particularly with sun exposure during childhood or adolescence, most studies in this area have focused on younger age groups. One of the strongest studies in this area was conducted on 648 children aged between 9 and 11 years old, from 11 different schools. They evaluated the effects of intensive intervention and standard intervention on sun protective behaviours. Diary-based reports of sun protective behaviours increased five weeks and eight months after implementation of the

237 intensive intervention program, with participants in the intensive intervention group more than three times more likely to use sun protective behaviours than participants in the control group. There were no differences in protective behaviours between the standard care and control groups [Girgis et al., 1993]. Another school-based study randomly assigned one school to receive a five-unit sunshine and skin health curriculum compared with another school as a control. The intervention was associated with markedly increased knowledge and negative attitudes about tanning immediately and eight weeks after intervention. In addition, they reported increased sunscreen use and increased use of protective clothing [Buller et al.,

2005]. A large randomized study of 543 adolescents (12–16 year old) conducted at seven different schools, compared four intervention groups with a control group. The interventions differed in their increasing intensities, but they all resulted in increased knowledge and appropriate attitudes about tanning compared to the control group [Rodrigue et al., 1996].

Some of the inconsistencies between study results can be attributed to differences in behaviour assessment. For example, Girgis et al. [Girgis et al., 1994] measured protective behaviours using daily diaries (over a five-day period), while other studies assessed protective behaviours at one time with a general self-report questionnaire. Moreover, studies were conducted for different age groups, genders and countries.

There have been a few intervention studies that targeted a whole population and showed an improvement in skin protection behaviour. These studies used television, newspaper or social media to improve knowledge on sun exposure and sun protective behaviours. The most successful interventions for changing sun protection behaviour were the community-based studies [Dietrich et al., 1998]. One of these studies was conducted in Victoria, Australia, as part of the “SunSmart” campaign. This program included behavioural objectives such as activity scheduling and shade-seeking. Pre-campaign (n=560) and post-campaign (n=605) cross-sectional data were collected from households sampled in clusters from census

238 collectors’ districts across Victoria. Those exposed to the campaign were more likely to use protective behaviours than were residents of a control community. Sunscreen use was the most common precaution used despite efforts by the “SunSmart” program to emphasize increasing other behaviours over chemical prevention. Hat and shirt use were the next most commonly cited behaviours. Nearly half of the respondents reported increased sun protection behaviours during the summer of the campaign. Two-thirds of those surveyed indicated that they had encouraged others to adopt precautionary behaviours for sun protection [Borland et al., 1990].

Another community-based study in Texas evaluated the effects of special UV light meters that measure the intensity of the sun’s dangerous UVB rays mounted on unshaded rooftops on sun protection behaviours. They showed that self-reported sun protection behaviour increased in households that received the UVB monitoring device [Boutwell et al., 1995].

7.7.6 Limitations

While the cross-sectional component had a substantial sample size, the sample size for the randomised controlled trial may well have been limiting. Moreover, most of our subjects were Australian, had pale skin and skin sensitive to sun exposure. In this study, we were not able to see the association of gender or age with sun-related behaviour, as our participants were all women aged 16 to 25 years old. We relied on self-reporting of sun-related behaviour information and did not directly assess these behaviours. However, the validity of the questionnaires has already been shown to be adequate. Recall bias is another limitation of our study, as it may have been difficult for participants to remember the sun-related behaviours from the previous year. Another limitation of this project is low compliance to the behavioural intervention (22%), which makes it difficult to draw conclusions on the effects of

239 behavioural intervention on sun related behaviours. Finally, samples could be biased as people with UV sensitive skin could be more interested to participate in this project.

7.7.7 Strengths

An important strength of our study was that we objectively measured UV exposure with a

UV dosimeter rather than solely using a questionnaire to evaluate sun exposure. Using validated questionnaires was another strength of our project. The extensive collection of questionnaires and biodata also allowed adjustment for many potential confounding factors and relevant covariates. Moreover, our sample was broadly representative of young women in

Victoria [Fenner et al., 1995].

7.8 Conclusion and future direction

In summary, in this chapter we found that at baseline sun-related risky behaviour in summer was significantly lower in the control group compared to the other two intervention groups.

Over a 12 months follow-up period, reported risky sun behaviour decreased significantly in the app group. This change was not significant for the control or supplement group. Over the

12 months, sun exposure on winter weekends but not summer weekends decreased significantly in the app group. It also decreased significantly in the supplement group on summer and winter weekdays. The other sun exposure variables, suntan behaviour or suntan attitudes, did not statistically change over the 12 months of intervention in any of the groups.

No significant association was found between 25 OHD levels and sun risky behaviour score, suntan protection score or suntan attitude score in the behavioural intervention group.

However, higher 25 OHD levels were associated with less risky sun behaviour and less suntan attitude score in the control group and less sun risky behaviour score in the supplement group. Serum 25 OHD levels were positively associated with dosimeter-

240 measured UV exposure only in the behavioural group. This association was not statistically significant in the control or supplement group.

More research is needed to evaluate sun-related behaviour in men. Although rates of tanning are higher in women [Buller et al., 2011], the incidence of melanoma are 50% higher in men than in women in Australia [www.aihw.gov.au/acim-books]. Most of our participants were

Australian with fair skin colour. We recommend future studies to include a range of ethnicities and skin colours. We also recommend future research to evaluate other factors relevant to sun-related behaviours, such as motivation for tanning and social norms or culture. Moreover, as it has been shown that sun protection behaviour, attitudes to tanning and other sun-related behaviours vary across different age ranges, more study on different age ranges, including younger and older people, is suggested.

241

Table 7.1 Self-reported time spent in the sun in the cross-sectional part

N=449 <30min 30-60min 1-2hrs 2-3hrs 3-4hrs >4hrs

How much time spend in 33 (7.3%) 79 (17.6%) 103 (22.9%) 96 (21.4%) 70 (15.6%) 68 (15.1%) sun in summer on weekend

How much time spend in 64 (14.3%) 140 (31.2%) 115 (25.6%) 59 (13.1%) 40 (8.9%) 31 (6.9%) sun in summer on weekdays How much time spend in 127 (28.3%) 134 (29.8%) 115 (25.6%) 41 (9.1%) 22 (4.9%) 10 (2.2%) sun in winter on weekend

How much time spend in 193 (43.0%) 137 (30.5%) 72 (16.0%) 34 (7.6%) 7 (1.6%) 6 (1.3%) sun in winter on weekdays

Data presented as number (percentage)

242

Table 7.2 Self-reported sun protection behaviour in the cross-sectional part

N=449 Always\Often Sometimes Rarely\Never Don’t know

In summer Seek shade 263 (58.6%) 120 (26.7%) 59 (13.1%) 7 (1.6%) Cover head 141 (31.4%) 113 (25.2%) 192 (42.8%) 3 (0.7%) Wear cloths to 124 (27.6%) 123 (27.4%) 201 (44.8%) 1 (0.2%) protect from sun Wear sunglasses 307 (68.4%) 70 (25.6%) 71 (15.8%) 1 (0.2%) Apply sunscreen 276 (61.5%) 111 (24.7%) 61 (13.6%) 1 (0.2%)

In winter Seek shade 28 (6.2%) 51 (11.4%) 362 (80.6%) 8 (1.8%) Cover head 26 (5.8%) 74 (16.5%) 346 (77.1%) 3 (0.7%) Wear cloths to 173 (38.5%) 54 (12.0%) 214 (47.7%) 8 (1.8%) protect from sun Wear sunglasses 133 (29.6%) 112 (24.9%) 202 (45.0%) 2 (0.4%) Apply sunscreen 51 (11.4%) 74 (16.5%) 319 (71.0%) 5 (1.1%)

Data presented as number (percentage)

243

Table 7.3 Self-reported suntan behaviour in the cross-sectional part

N=298 Suntan behaviour responses During holiday-how Daily 41 (13.8%) often suntan 2-3 times a week 53 (17.8%) Once a week 38 (12.8%) Less often 55 (18.5%) Not during my 111 (37.2%) holiday

Not during holiday- Daily 9 (3.0%) how often suntan 2-3 times a week 28 (9.4%) Once a week 36 (12.1%) Every 2 weeks 25 (8.4%)

Monthly 14 (4.7%) Less often 66 (22.1%) Not during this 120 (40.3%) summer

Do you like to get Yes 168 (37.4%) suntan Used sunbed in last 2 Yes 2 (0.4%) months How often use sunbed Daily 0 Weekly 1 (50%)

Every 2 weeks 0 Monthly 0 Less often 1 (50%) Data presented as number (percentage)

244

Table 7.4 Self-reported suntan attitude in the cross-sectional part

Strongly disagree/ Neither agree nor Mildly Can’t say/don’t Disagree/Mildly disagree disagree agree/Strongly know agree

I feel more healthy with suntan 221 (49.2%) 76 (16.9%) 143 (31.8%) 9 (2.0%) A suntan make me feel more 134 (29.8%) 58 (12.9%) 247 (55.0%) 11 (2.4%) attractive This coming summer I tend to 292 (65.0%) 53 (11.8%) 97 (21.6%) 7 (1.6%) sunbathe regularly Most of my friend thinks suntan 142 (31.6%) 60 (13.4%) 230 (51.2%) 17 (3.8% is a good thing Suntan make me feel better 173 (38.5%) 57 (12.7%) 211 (47.0%) 8 (1.8%) about myself Most of my family think suntan 267 (59.5%) 72 (16.0%) 92 (20.5%) 18 (4.0%) is a good thing

Suntan protects you from 424 (94.4%) 7 (1.6%) 2 (0.4%) 16 (3.6%) Melanoma and other skin cancer

Data presented as number (percentage)

245

Table 7.5 Sun related behaviour of Safe-D Part A (n=407) and Part B baseline (n=123) 1

Variable Part A Part B baseline P-value 2 (n= 407) (n= 123) Sun risky behaviour score (summer) 13.0 (11.0-15.0) 13.0 (11.0 - 15.7) 0.821 Sun risky behaviour score (winter) 18.1 (17.0-22.0) 19.0 (17.0 – 22.0) 0.924 Suntan protection behaviour score 11.0 (1.0-17.0) 11.0 (7.5-13.0) 0.984 Suntan attitude score 22.4±8.8 20.45±8.03 0.788 1Data are median (IQR) or mean±SD 2P-value obtained from the independent t-test (2-tailed, equal variance assumed) for continues variables

246

Table 7.6 Time spent in sun in each group at baseline, 4 month and 12 month of intervention

Baseline Baseline Baseline All (n=123) Behavioural Supplement Control All (n=95) Behavioura Supplement Control All (n=80) Behavioural Supplement Control (n=40) (n=41) (n=42) l (n=29) (n=36) (n=31) (n=26) (n=31) (n=27) How much spend in sun <60min 40 (32.8%) 14 (35.0%) 13 (31.8%) 13 (31.8%) 38 (40.0%) 14 (48.4%) 12 (35.2%) 12 (37.6%) 29 (36.2%) 13 (52.0%) 10 (34.6%) 6 (23.0%) in summer on weekend 1-3hrs 54 (44.3%) 19 (47.5%) 17 (41.4%) 18 (43.8%) 43 (45.3%) 12 (41.4%) 19 (55.9%) 12 (37.6%) 37 (46.3%) 8 (32.0%) 13 (44.8%) 16 (61.5%) >3hrs 28 (22.9%) 7 (17.5%) 11 (26.8%) 10 (24.4%) 14 (14.7%) 3 (10.2%) 3 (8.9%) 8 (4.8%) 14 (17.5%) 4 (16.0%) 6 (20.7%) 4 (15.5%) How much spend in sun <60min 68 (55.7%) 26 (65.0%) 20 (48.8%) 22 (53.6%) 66 (69.4%) 19 (75.8%) 24 (70.6%) 20 (62.5%) 56 (70.1%) 19 (76%) 19 (65.5%) 18 (96.3%) in summer on weekdays 1-3hrs 40 (32.9%) 10 (25.0%) 16 (39.0%) 14 (34.1%) 24 (25.2%) 5 (17.2%) 9 (26.4%) 10 (30.3%) 22 (27.5%) 5 (20.0%) 10 (34.5%) 7 (26.9%) >3hrs 14 (11.4%) 4 (10.0%) 5 (12.2%) 5 (12.2%) 5 (5.2%) 2 (7.0%) 1 (3.0%) 2 (6.2%) 2 (2.4%) 1 (4.0%) 0 1 (3.8%) How much spend in sun <60min 78 (63.9%) 26 (65%) 25 (61.0%) 27 (65.8%) 65 (68.4%) 22 (75.9%) 22 (64.7%) 21 (65.6%) 61 (76.3%) 21 (96%) 21 (72.4%) 19 (73.1%) in winter on weekend 1-3hrs 38 (31.2%) 10 (27.5%) 14 (33.2%) 13 (31.8%) 29 (30.5%) 6 (20.6%) 12 (35.3%) 11 (34.4%) 16 (20.0%) 4 (4.0%) 7 (24.1%) 5 (19.2%) >3hrs 6 (4.9%) 3 (7.5%) 2 (4.8%) 1 (2.4%) 1 (1.1%) 1 (3.5%) 0 0 3 (3.7%) 0 1 (3.4%) 2 (7.7%) How much spend in sun <60min 94 (77.0%) 32 (80.0%) 28 (68.3%) 34 (83.0%) 85 (89.5%) 25 (86.2%) 32 (94.1%) 28 (87.6%) 72 (90.0%) 24 (96%) 25 (86.2%) 23 (88.5%) in winter on weekdays 1-3hrs 26 (21.3%) 7 (17.5%) 12 (29.3%) 7 (17.0%) 9 (9.5%) 31 (10.3%) 2 (5.9%) 4 (12.4%) 8 (10.0%) 1 (4.0%) 4 (23.8%) 3 (11.5%) >3hrs 2 (1.6%) 1 (2.5%) 1 (2.4%) 0 1 (1.0%) 1 (3.4%) 0 0 0 0 0 0

Data are simply proportional data which are presented as number (percentage)

247

Table 7.7 Time spent in sun changes over 12 months of intervention

Over 12 All (n=79) p-value* Behaviour p- Supplement p-value** Control p-value** months; n al (n=25) value** (n=29) (n=26) (%) How much spend Increased 13 (16.5%) 0.677 3 (12%) 0.105 7 (24.2%) 0.801 8 (30.8%) 0.549 in sun in summer No change 25 (31.6%) 6 (31.6%) 13 (44.8%) 8 (30.8%) on weekend Decreased 41 (51.9%) 16 (64%) 9 (31%) 10 (38.4%) How much spend Increased 19 (24%) 0.387 4 (16%) 0.247 5 (17.3%) 0.032 5 (19.2%) 0.368 in sun in summer No change 26 (32.9%) 8 (32%) 8 (27.5%) 9 (34.6%) on weekdays Decreased 34 (43.1%) 13 (52%) 16 (55.2%) 12 (46.2%) How much spend Increased 16 (20.3%) 0.460 2 (8%) 0.033 8 (27.6%) 0.319 6 (23.1%) 0.435 in sun in winter No change 37 (46.9%) 11 (44%) 12 (41.4%) 14 on weekend Decreased 26 (32.8%) 12 (48%) 9 (31%) (53.8%) 6 (23.1%) How much spend Increased 9 (11.4%) 0.393 0 0.259 6 (20.7%) 0.032 3 0.564 in sun in winter No change 45 (57%) 17 (68%) 10 (34.5%) (11.55%) on weekdays Decreased 25 (31.6%) 8 (32%) 13 (44.8%) 19 (73%) 4 (15.5%)

Data presented as number (percentage)

*P-value obtained from Chi square_comparing change between groups at 12 months follow-up (change in category over 12 month/group)

**P-value obtained from McNemar's test_ comparing 12 months follow-up with baseline in each group

248

Table 7.8 Sun protection behaviour changes over 12 months intervention

Baseline 4 months 12 months All (n=123) Behavioural Supplement Control All (n=95) Behavioura Supplement Control All (n=80) Behavioura Supplement Control (n=40) (n=41) (n=42) l (n=29) (n=36) (n=31) l (n=26) (n=31) (n=27) S-seek shade Always\Often 77 (63.2%) 28 (70%) 22 (53.7%) 27 (65.9%) 55 (57.9%) 19 (65.6%) 18 (53%) 18 (56.3%) 50 (62.5%) 16 (%64) 18 (62.1%) 16 (61.6%) Sometimes 32 (26.2%) 7 (17.5%) 14 (34.1%) 11 (26.8%) 27 (28.4%) 5 (17.2%) 12 (35.3%) 10 (31.3%) 19 (23.8%) 6 (24%) 6 (20.7%) 7 (26.9%) Rarely\Never 12 (9.8%) 5 (12.5%) 4 (9.8%) 3 (7.3%) 12 (12.7%) 5 (17.2%) 4 (11.7%) 3 (9.4%) 91 (1.2%) 2 (8%) 4 (13.8%) 3 (11.5%) Don’t now 1 (0.8%) 0 1 (2.4%) 0 1 (1.1%) 0 0 1 (3.1%) 2 (2.5%) 1 (4%) 1 (3.4%) 0 S-cover head Always\Often 42 (34.4%) 8 (20%) 16 (39%) 18 (43.9%) 35 (36.9%) 10 (34.5%) 11 (22.4%) 14 (43.7%) 30 (37.5%) 10 (40%) 9 (31.0%) 11 (42.3%) Sometimes 27 (22.2%) 13 (32.5%) 5 (12.2%) 9 (22%) 27 (28.4%) 10 (34.5%) 82 (3.5%) 9 (28.1%) 21 (26.3%) 5 (20%) 10 (34.5%) 6 (23.1%) Rarely\Never 52 (42.7%) 19 (47.5%) 19 (46.4%) 14 (34.1%) 32 (33.7%) 9 (31.0%) 15 (44.1%) 8 (25%) 29 (36.2%) 10 (40%) 10 (34.5%) 9 (34.6%) Don’t now 1 (0.8%) 0 1 (2.4%) 0 1 (1.1%) 0 0 1 (3.1%) 0 0 0 0 S-wear cloths to Always\Often 31 (25.4%) 7 (17.5%) 9 (21.9%) 15 (36.6%) 31 (32.7%) 11 (37.9%) 9 (26.5%) 11 (34.4%) 26 (32.4%) 9 (36%) 8 (27.5%) 9 (34.6%) protect Sometimes 37 (30.3%) 15 (37.5%) 11 (26.8%) 11 (26.8%) 32 (33.7%) 8 (27.6%) 13 (38.2%) 11 (34.4%) 27 (33.8%) 9 (36%) 9 (31%) 9 (34.6%) Rarely\Never 54 (44.3%) 19 (45%) 21 (51.3%) 15 (36.6%) 31 (32.6%) 10 (34.4%) 12 (35.5%) 9 (28.2%) 27 (33.8%) 7 (28%) 12 (41.3%) 8 (30.7%) Don’t now 0 0 0 0 1 (1.1) 0 0 1 (3.1%) 0 0 0 0 S-sunglasses Always\Often 85 (69.7%) 28 (70.0%) 28 (68.3%) 29 (70.7%) 33 (66.3%) 20 (67.9%) 21 (64.8%) 21 (65.6%) 58 (72.5%) 18 (72%) 23 (79.3%) 17 (65.4%) Sometimes 16 (13.1%) 5 (12.5%) 6 (14.6%) 5 (12.2%) 18 (18.9%) 5 (17.2%) 6 (17.6%) 7 (21.9%) 15 (18.8%) 4 (16%) 4 (13.8%) 7 (26.9%) Rarely\Never 21 (17.1%) 7 (17.5%) 7 (17.1%) 7 (17.1%) 14 (14.8%) 4 (13.7%) 6 (17.6%) 4 (21.5%) 7 (8.8%) 3 (12%) 2 (6.8%) 2 (7.7%) Don’t now 0 0 0 0 0 0 0 0 0 0 0 0 S-sunscreen Always\Often 73 (59.9%) 26 (25%) 24 (58.6%) 23 (56%) 60 (63.2%) 23 (78.4%) 17 (49.0%) 20 (62.4%) 52 (65%) 17 (68%) 17 (58.6%) 18 (69.3%) Sometimes 25 (20.5%) 6 (15%) 10 (24.4%) 9 (22%) 23 (24.2%) 4 (13.8%) 13 (38.2%) 6 (18.8%) 18 (22.5%) 7 (28%) 5 (17.2%) 6 (23.1%) Rarely\Never 24 (19.7%) 8 (20%) 7 (17.0%) 9 (22%) 12 (12.6%) 2 (6.8%) 4 (11.8%) 6 (18.8%) 10 (13.5%) 1 (4%) 7 (24%) 2 (7.7%) Don’t now 0 0 0 0 0 0 0 0 0 0 0 0 W-seek shade Always\Often 8 (6.5%) 3 (7.5%) 2 (4.8%) 3 (7.4%) 4 (4.2%) 2 (6.8%) 1 (2.9%) 1 (3.1%) 3 (3.7%) 1 (4%) 1 (3.4%) 1 (3.8%) Sometimes 10 (8.2%) 1 (2.5%) 5 (12.2%) 4 (9.8%) 12 (12.6%) 3 (10.5%) 6 (17.7%) 3 (9.3%) 6 (7.6%) 2 (8%) 3 (10.3%) 1 (3.8%) Rarely\Never 101 36 (90%) 32 (78.2%) 33 (80.4%) 78 (82.1%) 24 (82.7%) 27 (79.4%) 27 (84.5%) 71 (88.7%) 2 (88%) 25 (86.3%) 24 (92.4%) Don’t now (82.8%) 0 2 (4.8%) 1 (2.4%) 1 (1.0%) 0 0 1 (3.1%) 0 0 0 0 3 (2.5%) W-cover head Always\Often 7 (5.7%) 3 (7.5%) 1 (2.4%) 3 (7.3%) 7 (7.4%) 2 (6.9%) 3 (8.8%) 2 (6.3%) 5 (6.3%) 1 (4.0%) 1 (3.4%) 3 (11.5%) Sometimes 22 (18.3%) 4 (10%) 9 (22.0%) 9 (22.0%) 12 (12.6%) 4 (13.8%) 5 (14.7%) 3 (9.4%) 16 (20.0%) 5 (20.0%) 6 (20.7%) 5 (19.2%) Rarely\Never 83 (68%) 33 (82.5%) 31 (75.6%) 29 (70.7%) 75 (78.9%) 23 (79.3%) 26 (76.5%) 26 (81.3%) 59 (73.8%) 19 (76.0%) 22 (75.9%) 18 (69.2%) Don’t now 0 0 0 0 1 (1.1%) 0 0 1 (3.1%) 0 0 0 0 W-wear cloths Always\Often 54 (44.4%) 13 (32.5%) 18 (43.9%) 23 (56.1%) 39 (41.1%) 11 (37.9%) 14 (41.2%) 13 (40.6%) 37 (46.3%) 10 (40.0%) 14 (48.3%) 12 (46.2%) to protect Sometimes 12 (9.8%) 4 (10%) 5 (12.2%) 3 (7.3%) 13 (13.7%) 4 (13.8%) 4 (11.8%) 5 (15.6%) 13 (16.3%) 8 (32.0%) 2 (6.9%) 3 (11.5%) Rarely\Never 53 (43.4%) 23 (57.5%) 17 (42.5%) 13 (31.7%) 41 (43.2%) 12 (41.4%) 16 (47.1%) 13 (40.6%) 30 (37.5%) 6 (24.0%) 13 (44.8%) 10 (38.5%) Don’t now 3 (2.4%) 0 1 (2.4%) 2 (4.9%) 2 (2.1%) 1 (3.4%) 0 1 (3.1%) 0 0 0 0 W-sunglasses Always\Often 33 (27%) 10 (25%) 10 (24.4%) 13 (31.7%) 22 (23.2%) 8 (27.6%) 8 (23.5%) 6 (18.8%) 18 (22.5%) 6 (24.0%) 6 (20.7%) 6 (23.1%) Sometimes 27 (22.2%) 7 (17.5%) 13 31.7%) 7 (17.1%) 20 (21.1%) 2 (6.9%) 10 (29.4%) 8 (25.0%) 23 (28.8%) 7 (28.0%) 11 (37.9%) 5 (19.2%) Rarely\Never 61 (50%) 23 (57.5%) 18 (43.9%) 20 (48.8%) 53 (55.8%) 19 (65.5%) 16 (47.1%) 18 (56.3%) 39 (48.8%) 12 (48.0%) 12 (41.4%) 15 (57.7%) 249

Don’t now 1 (0.8%) 0 0 1 (2.4%) 0 0 0 0 0 0 0 0 W-sunscreen Always\Often 11 (9.0%) 2 (5%) 4 (9.8%) 5 (12.2%) 12 (12.6%) 3 (10.3%) 4 (11.8%) 5 (15.6%) 13 (16.3%) 4 (16.0%) 3 (10.3%) 6 (23.1%) Sometimes 18 (14.7%) 6 (15%) 8 (19.5%) 4 (9.8%) 8 (8.4%) 3 (10.3%) 4 (11.8%) 1 (3.1%) 10 (12.5%) 4 (16.0%) 3 (10.3%) 3 (11.5%) Rarely\Never 92 (75.5%) 32 (80%) 29 (70.7%) 31 (75.6%) 75 (78.9%) 23 (79.3%) 26 (76.5%) 26 (81.3%) 57 (71.3%) 17 (68.0%) 23 (79.3%) 17 (65.4%) Don’t now 1 (0.8%) 0 0 1 (2.4%) 0 0 0 0 0 0 0 0 Data presented as number (percentage)

250

Table 7.9 Sun risky behaviour score in each group at baseline, 4 months and 12 months

Baseline 4 months 12 months All (n=123) Behaviour Supplemen Control p- All (n=95) Behaviour Supplement Control p-value* All (n=80) Behaviour Supplemen Control p-value* al (n=40) t (n=41) (n=42) value* al (n=29) (n=36) (n=31) al (n=26) t (n=31) (n=27) S-sun risky 13.0 12.5 13.0 12.0 0.632 13.0 13.0 13.0 13.0 0.690 13.0 12.0 13.0 13.0 0.924 behaviour (11.0 - 15.7) (11.2-15.0) (11.0-16.0) (11.0-15.0) (10.0-15.0) (9.5-15.5) (10.0-16.2) (10.0-15.0) (13.0-15.2) (9.2-16.7) (10.2-14.0) (10.0-15.2)

W-sun risky 19.0 21.0 ** 19.0 18.5 0.045 19.0 19.0 19.5 19.0 0.947 19 20.0 18.0 19.0 0.793 behaviour (17.0 – 22.0) (19.0-22.0) (17.0-22.0) (16.0-21.0) (17.0-21.0) (17.0-21.0) (17.0-21.0) (17.0-20.0) (16-22) (16.0-23.0) (16.0-22.0) (16.0-21.0)

Data are median (IQR)

S: summer; W: winter

* Bonferroni corrected P-value obtained from Kruskal–Wallis (not normally distributed). Differences between three intervention groups at each time point

**Statistically-significant compared to control group, using post hoc Tukey test

251

Table 7.10 Sun risky behaviour score changes over 12 months of intervention

All (n=84) Behavioural (n=26) supplement 3 (n=31) Control (n=27) Mean p-values * Mean changes p-values * Mean p-values * Mean changes p-value changes changes S- change in sun risky -0.25±2.83 0.560 -1.41±3.03 0.033 -0.07±2.69 0.737 0.65±2.48 0.163 behaviour score over 12 months W- change in sun risky -0.43±3.26 0.317 -1.36±3.26 0.078 -0.69±3.14 0.225 0.76±3.13 0.178 behaviour score over 12 months

Data are mean change ±SD

S: summer; W: winter

*p-value obtained from Wilcoxon signed rank test (score changes over 12 months were not normally distributed). Mean differences from baseline.

252

Table 7.11 Suntan behaviour in each group at baseline, 4 months and 12 months

Baseline 4 months 12 months All Behavioural Supplement Control All Behavioural Supplement Control All Behavioural Supplement Control Had holiday Yes 73 (61.9%) 24 (61.5%) 22 (56.4%) 27 (67.5%) 56 (58.6%) 17 (58.6%) 20 (58.8%) 19 (59.4%) 45 (57.0%) 17 (70.8%) 14 (48.3%) 14 (53.8%) How often have Daily/2-3 times a 28 (31.5%) 10 (41.7%) 8 (36.4%) 10 (37.0%) 21 (37.5%) 5 (29.4%) 7 (35.0%) 9 (47.4%) 11 (24.4%) 6 (35.3%) 2 (6.9%) 3 (11.5%) you suntan week/Once a week during your holiday Less often/Not during 45 (68.5%) 14 (58.3%) 14 (63.6%) 17 (63.0%) 35 (62.5%) 12 (70.6%) 13 (65.0%) 10 (52.6%) 34 (75.6%) 11 (64.7%) 12 (93.1%) 11 (88.5%) my holiday How often have Daily/2-3 times a 11 (15.1%) 5 (20.8%) 3 (13.6%) 3 (11.0%) 9 (16.1%) 3 (17.7%) 3 (15.0%) 3 (15.9%) 6 (13.3%) 3 (11.8%) 3 (6.8%) 2 (7.7%) you suntaned week/Once a week outside of holiday Every 2 62 (84.9%) 19 (79.2%) 19 (86.4%) 24 (89.0%) 47 (83.9%) 14 (82.3%) 17 (85.0%) 16 (84.1%) 39 (86.7%) 15 (88.2%) 12 (93.2%) 12 (92.3%) weeks/Monthly /Less often/Not during this summer

Do you like to Yes 37 (30.6%) 11 (27.5%) 13 (31.5%) 13 (32.5%) 29 (30.5%) 11 (37.9%) 10 (29.4%) 8 (25%) 19 (24.1%) 7 (29.2%) 7 (24.1%) 5 (19.2%) get suntan

Data are presented as number (percentage)

253

Table 7.12 Suntan protection behaviour score over 12 months of intervention

Baseline 4 months 12 months All Behavioural Supplement Control p- All Behavioural Supplement Control p- All Behavioural Supplement Control p- value* value* value* Suntan protection 11.0 11.0 11.0 11.0 0.802 11.5 13 11.5 10.0 0.476 11.0 10.0 12.5 13 0.168 behaviour score (7.5-13.0) (6.0-13.0) (7.5-13.0) (8.0-13.0) (8.0-13.0) (9.0-13.0) (7.2-13.0) (8.0-13.0) (9.0-13.0) (10.0-11.5) (7.9-13) (8.0-13.0) Data are median (IQR)

*P-value obtained from Kruskal–Wallis (not normally distributed). Differences between three intervention groups

254

Table 7.13 Suntan protection behaviour score changes over 12 months of intervention

All Behavioural Supplement Control Mean p-value* Mean changes p-value* Mean changes p-value* Mean changes p-value* changes Change in Suntan 0.32±3.59 0.562 0.00±3.9 1.000 0.81±3.99 0.400 0.18±3.09 0.888 protection behaviour score over 12 months Data are presented as mean±SD *p-value obtained from Wilcoxon signed rank test (score changes over 12 months were not normally distributed). Mean differences from baseline

255

Table 7.14 Suntan attitude in each group at baseline, 4 months and 12 months

Baseline 4 months 12 months All (n=123) Behavioura Supplement Control All (n=95) Behavioura Supplement Control All (n=80) Behavioura Supplement Control l (n=40) (n=41) (n=42) l (n=29) (n=36) (n=31) l (n=26) (n=31) (n=27) I feel more Strongly disagree/ Disagree/Mildly disagree 62 (50.8%) 19 (47.5%) 20 (48.8%) 23 (56.1%) 51 (53.7%) 16 (55.2%) 16 (47.1%) 19 (59.4%) 38 (47.5%) 10 (40.0%) 15 (51.7%) 13 (50.0%) healthy with Neither agree nor disagree 23 (18.9%) 10 (25.0%) 8 (19.5%) 5 (12.2%) 16 (16.8%) 7 (24.1%) 5 (14.7%) 4 (12.5%) 15 (18.8%) 4 (16.0%) 6 (20.7%) 5 (19.2%) suntan Mildly agree/Strongly agree 35 (28.7%) 11 (27.5%) 12 (29.3%) 12 (29.3%) 28 (29.5%) 6 (20.7%) 13 (38.2%) 9 (28.1%) 23 (28.8%) 9 (36.0%) 7 (24.1%) 7 (26.9%) Can’t say/don’t know 2 (1.6%) 0 1 (2.4%) 1 (2.4%) 0 0 0 0 3 (3.8%) 1 (4.0%) 1 (3.4%) 1 (3.8%)

A suntan Strongly disagree/ Disagree/Mildly disagree 42 (34.4%) 14 (35.0%) 13 (31.7%) 15 (36.6%) 31 (32.6%) 10 (34.5%) 10 (29.4%) 11 (34.4%) 26 (32.5%) 6 (24.0%) 11 (37.9%) 9 (34.6%) make me feel Neither agree nor disagree 12 (9.8%) 4 (10.0%) 4 (9.8%) 4 (9.8%) 7 (7.4%) 2 (6.9%) 3 (8.8%) 2 (6.3%) 6 (7.5%) 2 (8.0%) 3 (10.3%) 1 (3.8%) more Mildly agree/Strongly agree 64 (52.5%) 20 (50.0%) 23 (56.1%) 21 (51.2%) 56 (58.9%) 17 (58.6%) 20 (58.8%) 19 (59.4%) 43 (53.8%) 15 (60.0%) 14 (48.3%) 14 (53.8%) attractive Can’t say/don’t know 4 (3.3%) 2 (5.0%) 1 (2.4%) 1 (2.4%) 1 (1.1%) 0 1 (2.9%) 0 4 (5.0%) 1 (4.0%) 1 (3.4%) 2 (7.7%)

This coming Strongly disagree/ Disagree/Mildly disagree 85 (69.7%) 28 (70.0%) 28 (68.3%) 29 (70.7%) 78 (82.1%) 25 (86.2%) 26 (76.5%) 27 (84.4%) 62 (77.5%) 17 (68.0%) 25 (86.2%) 20 (76.9%) summer I Neither agree nor disagree 16 (13.1%) 7 (17.5%) 6 (14.6%) 3 (7.3%) 6 (6.3%) 2 (6.9%) 2 (5.9%) 2 (6.3%) 8 (10.0%) 3 (12.0%) 1 (3.4%) 4 (15.4%) tend to Mildly agree/Strongly agree 18 (14.8%) 5 (12.5%) 6 (14.6%) 7 (17.1%) 11 (11.6%) 2 (6.9%) 6 (17.6%) 3 (9.4%) 7 (8.8%) 3 (12.0%) 2 (6.9%) 2 (7.7%) sunbathe Can’t say/don’t know 3 (2.5%) 0 1 (2.4%) 2 (4.9%) 0 0 0 0 2 (2.5%) 1 (4.0%) 1 (3.4%) 0 regularly Most of my Strongly disagree/ Disagree/Mildly disagree 40 (32.8%) 10 (25.0%) 12 (29.3%) 18 (43.9%) 37 (38.9%) 9 (31.0%) 15 (44.1%) 13 (40.6%) 32 (40.0%) 10 (40.0%) 10 (34.5%) 12 (46.2%) friend thinks Neither agree nor disagree 15 (12.3%) 6 (15.0%) 8 (19.5%) 1 (2.4%) 11 (11.6%) 4 (13.8%) 4 (11.8%) 3 (9.4%) 13 (16.3%) 3 (12.0%) 4 (13.8%) 6 (23.1%) suntan is a Mildly agree/Strongly agree 60 (49.2%) 23 (57.5%) 19 (46.3%) 18 (43.9%) 40 (42.1%) 15 (51.7%) 14 (41.2%) 11 (34.4%) 29 (36.3%) 10 (40.0%) 12 (41.1%) 7 (26.9%) good thing Can’t say/don’t know 7 (5.7%) 1 (2.5%) 2 (4.9%) 4 (9.8%) 7 (7.4%) 1 (3.4%) 1 (2.9%) 5 (15.6%) 5 (6.3%) 1 (4.0%) 3 (10.3%) 1 (3.8%)

Suntan make Strongly disagree/ Disagree/Mildly disagree 52 (42.6%) 18 (45.0%) 18 (43.9%) 16 (39.0%) 41 (43.2%) 13 (44.8%) 14 (41.2%) 14 (43.8%) 34 (42.5%) 8 (32.0%) 14 (48.3%) 12 (46.2%) me feel better Neither agree nor disagree 13 (10.7%) 4 (10.0%) 5 (12.2%) 4 (9.8%) 10 (10.5%) 4 (13.8%) 4 (11.8%) 2 (6.3%) 6 (7.5%) 1 (4.0%) 3 (10.3%) 2 (7.7%) about myself Mildly agree/Strongly agree 56 (45.9%) 18 (45.0%) 18 (43.9%) 20 (48.8%) 44 (46.3%) 12 (41.4%) 16 (47.1%) 16 (50.0%) 36 (45.0%) 14 (56.0%) 11 (37.9%) 1 (3.8%) Can’t say/don’t know 1 (0.8%) 0 0 1 (2.4%) 0 0 0 0 3 (3.8%) 1 (4.0%) 1 (3.4%) 1 (3.8%)

Most of my Strongly disagree/ Disagree/Mildly disagree 73 (59.8%) 22 (55.0%) 24 (58.5%) 27 (65.9%) 60 (63.2%) 19 (65.5%) 19 (55.9%) 22 (68.8%) 50 (62.5%) 15 (60.0%) 21 (72.4%) 14 (53.8%) family think Neither agree nor disagree 18 (14.8%) 6 (15.0%) 10 (24.4%) 2 (4.9%) 13 (13.7%) 3 (10.3%) 6 (17.6%) 4 (12.5%) 11 (13.8%) 4 (16.0%) 1 (3.4%) 6 (23.1%) suntan is a Mildly agree/Strongly agree 22 (18.0%) 10 (25.0%) 5 (12.2%) 6 (14.6%) 19 (20.0%) 7 (24.1%) 8 (23.5%) 6 (18.8%) 15 (18.8%) 4 (16.0%) 5 (17.2%) 6 (23.1) good thing Can’t say/don’t know 9 (7.4%) 1 (2.5%) 2 (4.9%) 6 (14.6%) 3 (3.2%) 0 1 (2.9%) 2 (6.3%) 3 (3.8%) 1 (4.0%) 2 (6.9%) 0

Suntan Strongly disagree/ Disagree/Mildly disagree 117 37 (92.5%) 39 (95.1%) 41 88 (92.6%) 28 (96.6%) 30 (88.2%) 30 (93.8%) 71 (88.8%) 21 (84.0%) 25 (86.2%) 25 (96.2%) protects you Neither agree nor disagree (95.9%) 1 (2.5%) 0 (100.0%) 4 (4.2%) 0 3 (8.8%) 1 (3.1%) 2 (2.5%) 1 (4.0%) 1 (3.4%) 0 from Mildly agree/Strongly agree 1 (0.8%) 0 0 0 1 (1.1%) 0 1 (2.9%) 0 2 (2.5%) 1 (4.0%) 0 1 (3.8%) Melanoma Can’t say/don’t know 0 2 (5.0%) 2 (4.9%) 0 2 (2.1%) 7 (24.1%) 0 1 (3.1%) 4 (5.0%) 1 (4.0%) 3 (10.3%) 0 and other 4 (3.3%) 0 skin cancer

256

Data are presented as number (percentage)

257

Table 7.15 Suntan attitude score changes over 12 months of intervention

Baseline 4 months 12 months

All (n=123) Behavioura Supplement Control p-value* All (n=95) Behavioural Supplement Control p-value* All (n=80) Behavioura Supplement Control p- l (n=40) (n=41) (n=42) (n=29) (n=36) (n=31) l (n=26) (n=31) (n=27) value* 20.45±8.03 21.00±8.03 21.17±8.38 19.21±7.74 0.53 20.17±7.93 19.50±7.73 22.00±8.47 18.65±7.27 0.24 19.91±8.09 21.10±8.75 18.83±7.65 20.00±8.15 0.65 Suntan attitude score

Data presented as mean±SD

*P-value obtained from ANOVA. Differences between three intervention groups.

258

Table 7.16 Association of 25 OHD levels with sun related behaviour change over 12 months of intervention

25 OHD (nmol/L) All Behavioural Supplement Control Spearman p-value Spearman p-value Spearman p-value Spearman p-value correlation or correlation or correlation or correlation or Beta Beta Beta Beta S-sun risky Crude a -0.103 0.380 b -0.027 0.904 -0.304 0.140 -0.054 0.793 behaviour score Model 1 -0.093 0.441 c -0.059 0.752 -0.263 0.248 -0.049 0.770 change W-sun risky Crude -0.266 0.022 0.198 0.353 -0.404 0.045 -0.187 0.371 behaviour score Model 1 -0.248 0.038 0.022 0.909 -0.409 0.050 -0.187 0.285 change Suntan protection Crude 0.068 0.700 -0.415 0.179 -0.243 0.471 0.337 0.310 behaviour score Model 1 0.094 0.643 -0.288 0.428 -0.199 0.559 0.571 0.006 change Suntan attitude Crude -0.071 0.603 -0.367 0.162 0.262 0.252 -0.489 0.033 score change Model 1 -0.086 0.536 -0.342 0.106 0.268 0.268 -0.404 0.035

a Crude: Unadjusted

b p-value obtained from Spearman correlation

c p-value obtained from regression analysis adjusted for skin type, baseline physical activity and BMI

259

Table 7.17 Association of objective sun exposure with ethnicity, skin type, physical activity and skin sensitivity to sun exposure

UV exposure (SED) p-value Skin type 0.332 Ethnicity 0.642 Physical activity 0.026 Skin sensitivity 0.331

*p-value in crude model obtained from Spearman correlation or Chi square

260

Table 7.18 Association of 25 OHD levels and UV exposure (obtained from dosimeter)

25 OHD All Behavioural Supplement Control (nmol/L) Beta p-value Beta p-value Beta p-value Beta p-value UV exposure Crude 0.179 0.078 0.456 0.013 0.021 0.902 0.054 0.765 Model 1 0.141 0.193 0.458 0.018 0.014 0.942 0.029 0.883 *p-value in crude model obtained from Spearman correlation or Chi square *p-value in model 1 obtained from regression analysis adjusted for skin type, baseline physical activity and BMI

261

Table 7.19 Association of 25 OHD levels and sun protection behaviour

25 OHD (nmol/L) All p-value Seeking shade Summer 0.586 Winter 0.465 Wearing hat Summer 0.468 Winter 0.266 Wearing cloths Summer 0.450 Winter 0.578 Using sunglasses Summer 0.427 Winter 0.430 Using sunscreen Summer 0.464 Winter 0.449 *p-value obtained from Chi square

262

Chapter 8

CONCLUSIONS AND FUTURE PERSPECTIVES

263

The prevalence of vitamin D deficiency among women is high and a small proportion of women across Australia achieve optimal circulating 25 OHD levels (>75 nmol/L) [Gill et al., 2014, Daly et al., 2012, Holick et al., 2011]. According to Daly et al. nearly one-third (31%) of the

Australian population are vitamin D deficient (25 OHD <50 nmol/L). This prevalence is higher among women (39%) than men (22%). In the cross-sectional part of our project, we observed that 340 out of 557 of participants (61%) had vitamin D insufficiency, defined as 25 OHD levels less than 75 nmol/L. Vitamin D deficiency is potentially an important health risk for young women as it is suggested to be associated with many chronic diseases including cardiovascular diseases and diabetes [Gouni-Berthold et al., 2009 and Dalgård et al., 2011]. Results from previous studies are inconsistent and it remains unclear whether low serum 25 OHD levels are associated with an increased risk of cardiovascular disease [Skaaby et al., 2017 and Pilz et al.,

2016]. Moreover, it is still uncertain whether low 25 OHD levels are a causal risk factor or are simply related to the CVD risk factors due to reverse causation or confounding.

According to previous evidence, there are some gaps and limitations in this area. No one doubts that UV radiation increases vitamin D production, but the magnitude of the effect in individuals with different characteristics is less clear, due to trials being small and heterogeneous, and few addressing the issue of finding the balance between sufficient sunlight exposures to maintain healthy vitamin D status, while controlling the risk of skin cancer. Moreover, there is a lack of studies in young people, and there are lower quality randomized clinical trials in younger populations. Furthermore, most previous trials have been limited by the imprecision of the assays used to measure serum 25 OHD, with a critical need to employ state-of-the-art liquid chromatography-tandem mass spectrometry to measure serum 25 OHD concentrations with

264 higher precision and sensitivity and measure 25 OHD metabolites [Van den Ouweland et al.,

2013].

In this project, we tried to address these limitations and help to fill these knowledge gaps. This study was conducted with 16 to 25 year old women because 1) little evidence is available in this age group, 2) the high prevalence of vitamin D deficiency in this population, 3) the importance of this life stage, when people become more independent, and 4) the important role individual environmental/behavioural factors play in shaping health patterns that have long-term consequences, and 5) the popularity of communication using mobile and social media technologies in this demographic [www.sensis.com.au]. In the cross-sectional analysis we demonstrated that the serum concentration of 25 OHD was inversely associated with BMI, total fat mass, trunk fat, and visceral fat and was positively associated with triglyceride levels in young women. However, no significant associations with other lipids, glycaemic profiles or anthropometric measurements were observed.

In our project, the behavioural intervention was designed to evaluate an alternative approach to a pharmacological solution to vitamin D deficiency, which may be more affordable and appealing to many young women than taking supplements, has potential to achieve better long-term adherence than supplementation and enables guidance regarding SunSmart behaviour to be provided. We also attempted to address the important question whether it is feasible for most young Australian women to achieve sufficient vitamin D levels while still being SunSmart. We used a purpose-built, innovative, interactive, mobile application for lifestyle intervention to improve engagement, retention and data quality, and to ease participant burden. Moreover, we

265 used Facebook advertising for recruitment to increase the sample size. Any volunteer study has potential volunteer bias, but the high usage of Facebook in this demographic (>90%) and the general representativeness of our recruits should have minimised such bias. We used both the

LC-MS/MS and CMIA methods, to measure serum 25 OHD levels for all cross-sectional and 4 month follow-up analysis. LC-MS/MS method has the highest sensitivity and selectivity and is currently the most accurate method of serum 25 OHD measurements. However, for the 12 months follow-up analysis, 25 OHD levels only measured by CMIA method were available. The reason why we used the CMIA method for the 12 month data was that the assay provider had technical difficulties with their assay to measure the 25 OHD levels using LC-MS/MS method, but fortunately we had back-up results with an alternative laboratory and method. We also used

HOMA-IR and HOMA-B, two well-validated methods, to measure insulin resistance and beta- cell function. In addition, this study comprised a sample size of women which provided adequate power to perform multivariable analysis. This sample is broadly representative of this age group in Victoria [Fenner et al., 2012]. However, participants in this study were more educated and had higher BMI than the general population in Victoria [Fenner et al., 2012]. The extensive collection of questionnaires and biodata allowed adjustment for many potential confounding factors and relevant covariates in data analysis. Direct measurement of body composition and adiposity by using DXA, in addition to BMI, to estimate adiposity is another strength of this project. DXA is a widely-accepted standard method to measure body composition. Finally, in this study we measured sun exposure directly by using UV dosimeters in addition to employing well-established questionnaires; most previous studies used only a questionnaire to collect sun exposure data and did not directly measure sun exposure.

266

We conducted the clinical trial component of this project primarily to measure the effectiveness of a (1) behavioural and (2) pharmacological intervention to increase circulating 25 OHD levels over 12 months, compared with a control group in a cohort of young women.

Our secondary aims were to compare the effectiveness of the pharmacological and behavioural interventions to increase circulating 25 OHD levels over 12 months, to measure compliance in relation to the primary outcomes for the two intervention groups, and to measure the effects of pharmacological and behavioural intervention on obesity and cardiovascular risk factors including lipid profiles, glucose metabolism, anthropometric measurements, blood pressure, body composition and fat distribution.

According to our findings, using a mobile-based application to increase safe sun exposure and vitamin D supplementation both resulted in significant increases in 25 OHD levels after 4 months of intervention. However, this increase was greater in the supplement group than in the behavioural intervention group. Moreover, the behavioural intervention significantly reduced reported risky sun exposure in summer. This is important as the behavioural intervention improved vitamin D status while also reducing reported risky sun exposure.

During the 12 months of intervention, 32 adverse events occurred including sun burn. In the control group one abdominal pain, one syncope, one worsening of allergy reaction, one wound infection and six sunburn, in the behavioural intervention 12 adverse events (one worsening of depression/anxiety, one hip pain, one broken hand, two gastrointestinal pain, one wisdom teeth removal and six sunburn) and in the supplement group ten adverse events ( one menorrhagia, two

267 anaphylaxis episode, one tonsillitis, one worsening of allergic reaction, one flu, one peri acetabular osteotomy and one 3 sunburn) happened.

The compliance rate in the supplement group was 91.4% at the 4 months and 76.8% at 12 months follow-up. In the behavioural intervention group compliance was 22.3% at the 4 months follow-up and 12.8% at the 12 months follow-up. However, sensitivity analysis showed that there were no significant differences in outcomes between the lowest quartile and highest quartile of app compliance. Our original concept was that the app would be a learning tool and that users eventually would change their behaviour with respect to sun exposure and they would not be dependent on app use in the longer term. So the percentage of participants who changed their sun-related behaviour to be more protective was 52% in 4 months follow-up and 44% in the

12 months follow-up.

In the clinical trial, vitamin D supplementation resulted in a slight reduction in BMI and behavioural intervention resulted in a significant reduction in HbA1c compared to baseline after

4 months follow-up. However, changes in HbA1c and BMI were small and of uncertain clinical significance. Improvement in 25 OHD levels did not affect cardiovascular risk factors in young women. After 12 months of intervention, 25 OHD levels were significantly increased, more in the supplement group compared to the behavioural and control group. Compared to baseline, 25

OHD levels improved in both intervention groups. Serum 25 OHD levels increased 24.3±24.2 nmol/L in supplement group and 6.7±17.7 nmol/L in behavioural group and 1.7±19.0 nmol/L in the control group (p < 0.001). There were no significant between-group differences at 12 months in glucose, insulin, HOMA-IR, HOMA-B, QUICKI, total cholesterol, triglyceride, LDL, HDL

268 levels and HDL to total cholesterol ratio, systolic and diastolic blood pressure, height, weight,

BMI, waist and hip circumferences, waist to hip ratio and hs-CRP levels. HbA1c levels were increased from baseline in all three groups after 12 months of intervention. However, the increase in HbA1c was significantly greater in the behavioural intervention group compared to the other two groups (2.40±2.21 in the behavioural group versus 0.89±2.64 in the supplement group and 1.09±2.39 in control, p = 0.03). However, these significant changes were quantitatively small and of uncertain clinical significance. Moreover, the increase in 25 OHD levels did not lead to any significant changes in total fat mass, total fat percent, visceral fat mass, visceral fat percent, trunk fat mass, trunk fat percent, visceral to total fat ratio after the one year intervention. Only the trunk to total fat ratio was significantly decreased in the app group.

Improvement in 25 OHD levels did not affect body composition in young healthy women.

Evaluating the association of change in 25 OHD levels and change in CVD risk factors and body composition over 12 months of intervention, independent from the group allocation, revealed no significant associations between 25 OHD levels change and change in any of the metabolic profiles, blood pressure, total fat mass, total fat percent, visceral fat mass, visceral fat percent, leg fat mass and visceral to total fat mass ratio, over 12 months. However, changes in 25 OHD levels were negatively associated with change in waist circumference, WHR, trunk fat mass and trunk fat to total fat ratio.

In chapter 7, we evaluated sun-related behaviour changes in each group over 12 months of intervention. We also evaluated weather changes in sun-related behaviours were associated with

25 OHD levels. We found that 12 months of behavioural intervention could improve the sun

269 protection behaviour in young women. However, the behavioural intervention had no significant effects on suntan behaviour or suntan attitudes compared to baseline or compared to the supplement or control groups. Behavioural intervention also had no significant effects on CVD risk factors independent of 25 OHD levels.

An exploratory aim was to determine the proportion of young women in each arm of the trial in whom recommended vitamin D status was achieved at 4 and 12 months.

At the 4 months follow-up, in the supplement group 22 out of 41 participants (54 %) and in the behavioural intervention group 16 out of 42 participants (38%) reached the optimal 25 OHD levels (25 OHD >75 nmol/L). At 12 months follow-up in the supplement group 20 out of 38 participants (53 %) and in the behavioural intervention group 4 out of 34 participants (12%) reached the optimal levels of vitamin D (25 OHD >75 nmol/L). However, in the control group only 9 out of 42 (21.4%) participants at 4 months follow-up and 4 out of 34 participants (11.7%) at 12 months reached the optimal 25 OHD levels.

Strengths of the present study include 1) the design of the project which included two parts, a cross-sectional part with more than 400 participants which provided adequate power to perform multivariable analysis and a controlled randomised clinical trial part which allowed us to look at potential causal effects 2) the sample size in this study was broadly representative of this age group in Victoria 3) the evaluation of an under-studied demographic, young adult women 4) the collection of extensive data which allowed adjustment for many potential confounding factors and relevant covariates 5) measurement of 25 OHD levels using the LC-MS/MS method, as well as measurements of both 25 OHD metabolites in the cross-sectional part and in 4 months follow-

270 up. However, in the 12 months follow-up only 25 OHD levels measured by chemiluminescent microparticle immunoassay method in the Melbourne Health laboratory were available 6) using

HOMA-IR and HOMA-B, two well-validated methods, to measure insulin resistance and beta- cell function 7) direct measurement of adiposity by using DXA in addition to BMI or skin fold measurements to estimate adiposity 8) measurement of sun exposure directly by using UV dosimeters, an objective measure and 9) high compliance rate in the supplement group.

Our study has some potential limitations. Firstly, we were not able to directly measure insulin resistance and beta cell function, because of their invasive nature and costs in studies with a large sample size. We used HOMA-IR and HOMA-B to estimate the insulin resistance and beta cell function. While these methods are non-invasive, convenient ways of measuring insulin resistance and beta cell function, their validity in young non-diabetic people is still unknown. Second, blood pressure was measured only twice and not three times as per the World Health

Organization guidelines and we have not used 24-hour blood pressure monitoring. Another limitation was the low number of participants in the 25 OHD <25 nmol/L group (n=13), which reduced the study’s potential to evaluate associations with severe vitamin D deficiency.

Limitations of the trial part of the project includes low compliance in the behavioural intervention group (22.3%). Although app compliance was poor according to our definition for compliance (which was number of days app opened by participant), our original concept was that the app would be a learning tool and that users eventually would change their behaviour about sun exposure and then have little need to open the app on a regular basis. Another limitation is that in this trial all study staff and laboratory staffs were blinded, but we were not able to blind

271 participants to their group allocation and thus the study used an open label design. However, it has been shown that open label design studies are lower in cost and have greater similarity to standard clinical practice, which make results more applicable [Hansson et al., 1992]. Difficulty in booking the follow-up visits at exactly 4 months and 12 months from starting intervention was another limitation of our project. We aimed to book the 4 months and 12 months follow-up visits within a window period of one week before and one week after the exact site visit due dates.

Another limitation is that we did not reach our targeted sample size in the time available.

However, we ran an interim analysis and power calculation which showed us that our collected sample size was large enough to provide adequate power. The Safe-D app also faced some issues and glitches during the study. These problems related particularly to changes in phone operating systems. However, an unblinded study team member advised the app developer as quickly as possible to fix the problems and they generally responded promptly. The app faced some problems before starting the trial that necessitated suspending the intervention for about a month to address multiple glitches. In this study we considered 25 OHD levels less than 75 nmol/L as borderline vitamin D status; however there is still an ongoing controversy about what constitutes a sufficient vitamin D level. Also, we were not able to include the gynoid region data in our body composition analysis as the DXA Hologic software did not provide gynoid region data.

However, custom-design analysis has been done on 20 whole body scans to get the gynoid region data. A strong correlation was seen between the gynoid fat data and leg fat data.

Therefore, only leg region data were used as being representative of the lower body region. We relied on self-reported sun-related behaviour information and did not directly assess these behaviours. However, we did objectively measure UV exposure for 2-week sampling periods

272 using UV dosimeters worn by participants. Moreover, the validity of the questionnaires we used had been established previously. Finally, the results of our study may not be generalizable to men, other ethnic groups or other age groups given that all participants were women, aged 16 to

25 years and most of them white.

Future directions could include a clinical trial with a larger sample size, longer follow-up duration, using different doses of vitamin D supplementation, studies with a male population, and triple measurement of blood pressure as a standard method. We also recommend future studies to be done with population samples including a range of ethnicity and skin colour. Future improvements to the Safe-D app include: showing the users if they will get enough vitamin D based on UV index, especially in days with low UV index; automatic app updates following iOS or android software updates; quality improvement of the app to cause fewer glitches during the course of the study; app compliance can be improved by considering rewards after positive behaviour or making the app more user friendly.

273

APPENDICES

274

A) Protocol paper

Improving Vitamin D Status and Related Health in Young Women: The Safe-D study –

Part B

Monitoring Editor: Gunther Eysenbach Reviewed by Stephanie Alley and Erik Nelson Marjan Tabesh, MSc,1 Suzanne Marie Garland, MD,2,3,4 Alexandra Gorelik, MSc,1,5 Alison Nankervis, MD,1,6 Skye Maclean, MN(Prac),1 Emma Teresa Callegari, Biomed(Hons),1 Shanton Chang, PhD,7 Kayla Heffernan, MIS,7 and John Dennis Wark, MD, PhDcorresponding author18 1Department of Medicine, Royal Melbourne Hospital, University of Melbourne, VIC, Australia 2Murdoch Childrens Research Institute, Royal Children’s Hospital, VIC, Australia 3Women’s Centre for Infectious Diseases, Royal Women’s Hospital, VIC, Australia 4Department of Obstetrics and Gynaecology, University of Melbourne, VIC, Australia 5Melbourne EpiCentre, Royal Melbourne Hospital, VIC, Australia 6Diabetes Service, Royal Women's Hospital, VIC, Australia 7Department of Computing and Information Systems, University of Melbourne, VIC, Australia 8Bone and Mineral Medicine, Royal Melbourne Hospital, VIC, Australia John Dennis Wark, Department of Medicine, Royal Melbourne Hospital, University of Melbourne, Parkville, VIC, 3052, Australia, Phone: 61 8344 3258, Fax: 61 8344 3258, Email: ua.ude.bleminu@krawdj.

275

Abstract

Background

Vitamin D deficiency is highly prevalent and associated with increased risk of a number of chronic health conditions including cardiovascular disease, poor bone and muscle health, poor mental health, infection, and diabetes. Vitamin D deficiency affects millions of Australians, potentially causing considerable suffering, economic loss, and mortality.

Objective

To measure the effectiveness of a (1) mobile-based app (behavioural) and (2) pharmacological intervention to increase circulating 25-hydroxyvitamin D (serum 25 OHD) levels and health outcomes over 4 months of intervention compared with a control group in a cohort of young women with suboptimal serum 25 OHD levels (25-75 nmol/L).

Methods

Participants with 25 OHD levels 25 to 75 nmol/L are invited to participate in this study.

Participants are randomized to one of three groups in 1:1:1 ratio: a mobile phone–based application, vitamin D supplementation (1000 IU/day), and a control group. Data collection points are at baseline, 4, and 12 months post baseline with the major endpoints being at 4 months. A wide-range of information is collected from participants throughout the course of this study. General health, behavioural and demographic information, medications, smoking, alcohol and other substance use, health risk factors, nutrition, eating patterns and disorders, and mental health data are sourced from self-administered, Web-based surveys. Clinical data include

276 anthropometric measurements, a silicone skin cast of the hand, cutaneous melanin density, bone mineral density, and body composition scans obtained through site visits. Main analyses will be conducted in two ways on an intention-to-treat (ITT) basis using the last observation carried forward approach as an imputation for missing data, and on a per protocol basis to compare the intervention arms against the control group at 4 and 12 months.

Results

Publication of trial results is anticipated in 2017.

Conclusions

The study will allow assessment of the effects of a mobile-based app behavioural intervention and vitamin D supplementation on vitamin D status and will evaluate the effects of improving vitamin D levels on several health outcomes.

Keywords: vitamin D, young women, health outcomes, intervention, Safe-D study, behavioural intervention, app, vitamin D supplementation, m-Health

277

Introduction

Low vitamin D status is an issue of concern in today’s society and its prevalence has been reported to be high, with approximately 50% of the world population thought to be affected [1].

Low vitamin D levels have been shown to be associated with an increased risk of numerous chronic health conditions, including poor musculoskeletal health and cardiovascular disease [2].

Vitamin D deficiency also impacts on young women’s ability to achieve optimal peak bone mass increasing the risks of osteoporosis and osteoporotic fractures, which are major public health problems in the aging population [3,4]. Twenty-seven percent of women across Australia achieve optimal vitamin D levels (defined as serum 25-hydroxyvitamin D (25 OHD) greater than 75 nmol/L) [5]. Individual factors such as habitual sun exposure and skin colour are likely to play major roles in determining vitamin D status [5,6]. Vitamin D deficiency is an important health risk factor for young women, particularly during their childbearing years, when deficiency can harm both the mother and the unborn child [7]. Despite the potentially serious effects of vitamin

D insufficiency (serum 25 OHD between 26 and 49 nmol/L) and deficiency (serum 25 OHD <25 nmol/L), very few vitamin D studies have focused on young women; the majority of studies in the literature have studied general population cohorts or have exclusively focussed on elderly participants [8]. This study is of particular importance as many risks factor for certain cancers, autoimmune diseases, infections, neurological diseases, diabetes, and poor mental health are established during youth [1]. Vitamin D status is one of the factors that may impact on these disease risks.

The Safe-D study aims to (1) examine the links between vitamin D and various health indicators

278 in a young female cohort (Part A) [9], and (2) evaluate the effectiveness of a smartphone app compared with vitamin D supplementation in improving both serum 25 OHD levels and several health measures that have been associated with vitamin D deficiency (Part B). By focusing on a younger cohort and using a successful recruitment strategy via social media [10], the study aims to achieve a much larger sample size of this population than previous research [11,12]. In addition, the relationship between vitamin D status and health will be comprehensively studied using state-of-the-art information technology-based data collection methods that have not been used in any similar studies, which have based results largely on self-reported data [13].

While vitamin D deficiency is associated with many poor health outcomes, its potential impact on young Australian women’s health has yet to be established. Moreover, there are evidence gaps about the safest and most effective interventions to improve vitamin D status. Of importance, much of the previous research has been limited by imprecise 25 OHD measurements

[2].

Addressing factors that can lead to vitamin D deficiency earlier in life might be beneficial for long-term health, productivity, and quality-of-life of young women. Part A of the Safe-D study is described elsewhere [9]. Part B, a randomized controlled trial involving the use of a digital Web- based smartphone app, is described here.

Study Rationale

This study focuses on 16- to 25-year-old women, because of: (1) the high prevalence of vitamin

D deficiency in young people [1,2,14,15], (2) the importance of this life stage, as individuals become more autonomous and independent, and individual environmental as well as behavioural

279 factors play an increasing role in shaping health patterns that have long-term consequences [16],

(3) the popularity in this demographic of communication using mobile and social media technologies with which we have previous experience, and harness in this study [10], (4) vitamin

D deficiency impacting young women’s bone mass increasing the risks of osteoporosis and osteoporotic fractures later in life [17], and (5) women’s smaller skeletons make them more likely to develop osteoporosis due to biological sex differences in the skeleton with ageing [18].

While there have been many studies examining the associations between vitamin D and health, only small, restricted studies have been conducted with young women. Therefore, this study aims to ensure that comprehensive health data are collated for this population subgroup.

Methods

Study Design

The Safe-D study comprises two distinct, but overlapping and interrelated components. Part A is a cross-sectional study of healthy women aged 16 to 25 years, aimed at investigating associations between 25 OHD levels and musculoskeletal health (bone density, bone turnover markers, muscle function), mood/mental health, body composition and weight, and atopic/allergic symptoms [9]. Part B is an open-label, blinded-endpoint, randomized controlled trial with three arms; a behavioural intervention, a pharmacological intervention, and control group [19,20].

Study participants are monitored for a period of 12 months. A comprehensive, Web-based questionnaire is completed by all participants at 0 and 12 months; an abbreviated version of this survey is completed at 4 months. The survey links are sent through LimeSurvey (an open-source, password-protected, secure software survey tool) to each participants [21]. The questionnaires

280 comprise five modules, which the participants are able to complete either altogether or on separate occasions over a 2-week period before their site visit. The modules cover health areas including demographics, medical history, use of health care professionals, use of medications and allergy data to establish current and/or past health conditions [14]. Nutrition, dietary behaviours, and weight management data are collected to identify change in body composition and weight and investigate the relationship between obesity and vitamin D. Body image, alcohol use, tobacco use, and illicit drug use information is collected because it has been suggested that vitamin D has a neuroprotective effect on dopaminergic pathways in the adult brain and may have a role in the management of drug dependence, and also smoking and alcohol use can affect vitamin D status [22]. Diet, physical activity, pain and injuries, sun exposure, and mental health data are collected to assess dietary intake of vitamin D, to control for physical activity as a contributor to weight and to investigate a relationship between vitamin D and musculoskeletal health. Sun exposure is collected to investigate a relationship between change in vitamin D and ultraviolet (UV) exposure.

Dietary intake is a confounder for the relationship between vitamin D and obesity [23]. The

Cancer Council Victoria Questionnaire (comprehensive dietary questionnaire) is used to evaluate the diet type and portion sizes [23].

In addition, site visits take place at our study centre at each time point to collect a range of clinical health information, including blood tests, bone density (baseline and 12 months), and tests of muscle health (Figure 1).

Subject Selection

281

Participants in this study are female, aged between 16 to 25 years upon inclusion in the study, and residing in Victoria, Australia, for the duration of the study. Inclusion criteria are completion of all components of Part A of the Safe-D study, serum 25 OHD levels between 25 and 75 nmol/L, plus ownership and regular use of a smartphone with Apple or Android operating systems.

Exclusion criteria include a history of skin melanoma (or having a first-degree relative (parent or sibling) who has had a melanoma), current pregnancy, breastfeeding, an intention to conceive in the next 12 months, current supplementation with ≥800 IU vitamin D daily, an intention to move out of Australia during the course of the study, any chronic health condition or medication that may disturb vitamin D metabolism or action or cause safety concerns, any medical condition, or using any medication that increases sensitivity to sun light or UV radiation.

The inclusion and exclusion criteria were designed so that any harm to study participants is minimized as far as practically possible, while selecting as broad a sample as possible to maximize the generalizability of the findings. The exclusion and inclusion criteria also assist to reduce confounding of results. If participants meet any of the exclusion criteria, at any time in the study they are withdrawn from intervention as appropriate for good clinical care. Where possible, these participants complete all data collection including study visits for the purpose of an intention-to-treat (ITT) analysis.

Proposed Sample Size

A meta-analysis of 16 studies found an increased serum 25 OHD concentration of approximately

1 to 2 nmol/L for each 100 IU per day of supplemental vitamin D [24]. Assuming average

282 supplementation of 1000 IU vitamin D/day, we expect changes in serum 25 OHD concentration of approximately 10 to 20 nmol/L in the pharmacological intervention group. By using ITT analysis, a sample size of 62 per arm at 4 months (corresponding to 78 per arm at baseline) will

85% power to detect a difference of 15 nmol/L in 25 OHD levels between groups (assuming a standard deviation of 25 nmol/L and total level of significance of 0.05). This gives an 80% power to detect a 15 nmol/L difference in per protocol analysis, assuming 85% adherence to the protocol, and making the assumptions of at least 50% of participants fall in the range of 25 to 75 nmol/L at entry, 20% attrition at 4 months, and 30% total attrition at 12 months [25].

We expect to be able to include 48 participants per arm (n=56 for ITT analysis) in the 12-month per protocol analyses (accounting for attrition and noncompliance), giving us 80% power (85% in ITT) at a 5% significance level to detect a 9 to 10 nmol/L difference in change from baseline between the two intervention arms (assuming a lower standard deviation of 20 nmol/L due to seasonal matching).

Based of the above, the study team aim to recruit 468 young women into Part A of the study, which should lead to approximately 234 participants in Part B assuming at least 50% of participants fall in the serum 25 OHD range of 25 to 75 nmol/L, meet other eligibility criteria for

Part B and agree to participate.

Recruitment

Participants who complete Part A of the Safe-D study (including completion of a comprehensive questionnaire, wearing of an UV dosimeter for a period of 14 consecutive days and participation in the study site visit) and meet the eligibility criteria for Part B are invited to participate in the

283 study. Upon receipt of the serum 25 OHD and other pathology results from the laboratory, eligible participants are contacted by a member of the study team to inform them that they are eligible to participate. A verbal consent process takes place during the telephone call. Each volunteer is given sufficient time to review the detailed participant information and consent form

(PICF), and is given the opportunity to ask questions about the study. Subjects are then asked to provide written informed consent in order to participate. All participants under 18 years are assessed as a mature minor and those deemed unable to provide informed consent require the consent of their parent or guardian to participate in the study [26].

All participants are offered compensation for their time in the form of an AU$30 gift voucher at each assessment. Participants identified as having high depressive and/or anxiety symptoms in study survey responses at any time point are sent information booklets (beyondblue booklet;

“What works for anxiety disorders?” [27] and “What works for depression in young people”)

[28]. Participants who answer positively to suicidal ideation are contacted by a study team member with Applied Suicide Intervention Skills Training, to ensure the participant’s safety and well-being [29].

Participants whose serum 25 OHD levels show moderate to severe deficiency (<25 nmol/L) are contacted by the study team and strongly advised to review the results with their primary care physician. This group is expected to be <8% of the target population and are not randomized into

Part B. Subjects whose levels drop below 25 nmol/L at 4 months are also immediately referred for appropriate treatment; however, they are encouraged to remain in the trial to reduce attrition.

Verbal, Written, and Electronic Consent

284

Verbal consent is obtained from all participants in this study by telephone communication.

Participants then are sent email links, so that they can complete the Web-based surveys. Each of the surveys includes the PICF and this is being used as a means of obtaining electronic consent.

Prior to the site visit, all participants are sent a hard copy of the PICF with a welcome letter. The

PICF is reviewed at the commencement of the study visit prior to the collection of any biological data.

An electronic consent (e-consent) form is offered to participants, which is obtained through a secure link sent by the study team via LimeSurvey. Information contained in the PICF is displayed on screen and requires participants to click “I agree” to a statement that they have read and understood the information and freely agree to participate in the research project.

Participants giving electronic consent are asked to provide written consent at the first subsequent opportunity to do so. The e-consent form allows recruitment and allocates the participants to the intervention groups to allow for a more streamlined process. Eligible participants were randomized into one of three groups.

Trial Interventions

Behavioural Intervention Group

All participants randomized to the behavioural intervention group receive instructions on how to download the Safe-D study mobile-based app (Safe-D app) to use for 12 months following randomization [20]. The Safe-D app is designed for both Apple and Android operating systems.

The Safe-D app delivers advice about how to obtain safe and effective sun exposure daily, as per

285 guidelines developed by the Safe-D team in conjunction with the SunSmart guidelines [20]. An algorithm was incorporated into the app to estimate the time required with direct exposure to sunlight to achieve adequate vitamin D levels [30]. The mechanics of algorithm are not apparent to users. It is the messages themselves that convey the complex information simply, and the use of game elements to further simplify understanding. In Safe-D app UV-exposure is shown as different shape of sunflower to users to convey message simply and effectively. Advice is tailored according to the individual’s characteristics and reported behaviours, including

Fitzpatrick skin type [31], clothing, sunscreen use, and local UV forecast (location determined using the smartphone’s global positioning system or by manually entering location details) sourced from the Australia Bureau of Meteorology and the Australian Radiation Protection and

Nuclear Safety Agency. As there is a need for improved education in the community about

SunSmart behaviours and safe methods to achieve the best possible vitamin D production, general advice is also delivered.

The Safe-D app enables tracking of UV-exposure, records missed exposure, and monitors participant progress regarding time spent in the sun. Participants start the app timer when they are in direct sunlight and stop when they are no longer exposed. The app sends tailored messages to participants depending on their sun exposure records. The messages are sent as push notification, rather than requiring participants to check the app daily [32-34]. Educational messages are also sent to maintain participant motivation and interest in the study. Messages encourage appropriate and safe levels of UV-exposure and explain the importance of vitamin D, especially in women, health consequences of vitamin D deficiency, and tips to improve vitamin

D status. Three types of messages are sent to participants through the app: automatic push

286 notifications, automatic in-app messages, and tailored and personalized mail messages [35].

Participants are able to turn off the automatic push notification, though they are encouraged by the study team during their randomization call not to do so. Safety is monitored within the app and any participant who exceeds the recommended time in the sun receives an auto generated safety warning and the application reports any consequences of overexposure to study staff. Use of the application is also measured by the number of times participants open it [20].

Pharmacological Intervention Group

All participants in this group receive a 1-year supply of 1000 IU vitamin D supplements.

Participants are informed of the prescribed dose and route of administration of the supplements

(oral), and the recommended storage conditions. Participants are sent weekly regular SMS text messaging with (short message service, SMS) reminders to take the equivalent dose of 1 capsule per day, and reminded that if they miss any doses throughout the week, that they may take the missed tablets all at once that day, without any risk of toxicity. Messages are sent weekly for the first month of participation, taper to fortnight, then monthly. Standard protocols for the receipt, dispensing, return, and disposal of these supplements have been developed with the assistance of a research pharmacist.

Control group

All participants in this group receive general advice in the form of the “How much sun is enough?” pamphlet produced by Cancer Council Victoria [36]. This pamphlet is sent to participants via email. It provides information about achieving adequate vitamin D status using safe levels of UV exposure and diet, and a contact number to obtain further advice, with links to

287

SunSmart Victoria’s Web-based vitamin D resources.

Randomization

Participants are randomized into one of the three intervention groups using stratified block randomization with computer-generated varying block sizes (3, 6, and 9), based on baseline serum 25 OHD levels 25 to 49 nmol/L and 50 to 74 nmol/L. The study statistician (AG) is responsible for the generation of the randomization schedule and preparing the codes. An electronic process is used to generate the codes. As each eligible participant is identified from

Part A of the Safe-D study and consented, the statistician is emailed by an unblinded researcher, and a randomly generated allocation group is assigned, details of which are kept in the participant’s notes and entered into the unblinded database. All other study team members who collect outcome data are blinded to intervention group allocation.

Blinding

As far as practically possible, procedures are in place to maintain blinding of team members who collect outcome data. They are blinded to the participant’s treatment allocated Two databases are used to avoid inadvertent unblinding, one for blinded members and one for unblinded members.

Participants are blinded to their vitamin D stratification, and only receive the details of their vitamin D results during the trial if clinically necessary. All vitamin D results will be provided to participants upon conclusion of the study.

Data Collection

A wide-range of information is collected from participants throughout the course of this study

288

[9]. These data are sourced from self-administered Web-based surveys and clinical data are obtained through site visits. Data are collected from all participants at baseline (0 month) and at the end of the study (after 12 months of study participation). All data except the bone density scans are collected at 4 months, which is the approximate duration predicted for vitamin D levels to reach a steady state with intervention [37]. A modified version of the questionnaires is used at

4 months. Bone density testing is not performed at 4 months because significant changes are unlikely to be detected at that time-point.

Questionnaires Content

The Cancer Council Victoria Questionnaire was comprised of dietary intake and portion sizes data. The other questionnaires comprise four modules, which the participants are able to complete either altogether or on separate occasions over a 2-week period before the site visit.

The modules cover groups of health areas as described as follows:

Module A: demography, medical history, use of health care professionals, use of medications and allergy data.

Module B: nutrition, dietary behaviours, and weight management data.

Module C: body image, alcohol use, tobacco use, and illicit drug use information.

Module D: diet, physical activity, pain and injuries, sun exposure, and mental health data.

Site Visit Assessment and Rationale

Participants are asked to attend Royal Melbourne Hospital for a 2-hour study site visit baseline

289 and 12 months for a health check including a physical examination, blood collection, silicone skin cast of the hand, skin reflectance, bone density and body composition scans, and Leonardo mechanography testing [38]. They are also asked to attend for a 1-hour visit at the 4-month follow-up for the above tests except that bone density tests are not performed at this time point.

Physical Examination

Blood pressure, resting heart rate, waist circumference, hip circumference, height, and weight are measured for general health assessment. Blood pressure is measured twice with two different machines and recorded as systolic and diastolic blood pressure. Height is measured to the closest of 0.1 cm by using a wall-mounted stadiometer. Weight is measured to the closest of 0.01 kg.

Blood Collection

A fasting, morning blood sample is collected to test for analytes by standard methods including

25 OHD, thyroid stimulating hormone, prolactin, insulin, HbA1c, glucose, lipids (total cholesterol, high-density lipoprotein, low-density lipoprotein, and triglyceride), calcium, parathyroid hormone, albumin, creatinine and C-reactive protein [9]. VivoPharm Laboratories employ a highly-sensitive, accurate, and precise liquid chromatography-tandem mass spectrometry (LC-MS/MS) method using Applied Biosystems 4000 Q Trap and Agilent LC-

MS/MS instruments, which is used to measure serum 25 OHD3 and serum 25 OHD2 concentrations for this study. Serum 25 OHD concentration is the best indicator of vitamin D status. Ligand-binding assays are the standard method for measuring serum 25 OHD but have some limitations including poor agreement between assays and laboratories [39], inability to distinguish between serum 25 OHD2 and 25 OHD3, and systematic under- or overestimation of

290

25 OHD levels [40]. The current “gold standard” method for determining vitamin D levels is LC-

MS/MS, which is more accurate and precise, uses standards of defined concentrations, needs smaller sample volume and shorter turnaround time than alternative methods, and distinguishes between D2 and D3 metabolites. For these reasons, we chose the LC-MS/MS method to measure vitamin D levels.

Skin Cast of Hand

Actinic skin damage is measured at baseline, after 4 and 12 months of intervention by taking a silicone rubber cast of the dorsum of the hand to assess skin damage [41]. The Beagley and

Gibson grading system is used to score the skin casts [42]. This visual system of grading is not time-consuming and it is well suited for use in studies with large sample size. Skin casts are graded on a scale of 1 to 6, with 1 indicating undamaged skin that is evenly spaced with fine lines of equal depth. Grade 6 is indicative of maximum photo-damage, specifically with a more flattened appearance to the skin surface [43].

Skin Reflectance

Cutaneous melanin density is measured at each visit using a spectrophotometer as skin colour is a covariate to be controlled for in assessing vitamin D response. Melanin density is measured at both a UV-unexposed region (inner side of upper arm) and exposed regions (back of hand and facial cheek) .

Bone Density and Body Composition Scans

291

Participants attend the Bone Densitometry Unit at the Royal Melbourne Hospital for dual energy x-ray absorptiometry (DXA) and pQCT scanning. DXA is used to measure areal bone mineral density and bone mineral content as well as soft tissue composition. The parts of the body to be scanned are lumbar spine, total hip, femoral neck, and total body [44]. Peripheral QCT of the tibia is used to assess volumetric bone density, bone geometry, and muscle cross-sectional area to investigate the relationship between 25 OHD and these measures [45].

Leonardo Mechanography

Muscle measurements are taken using a Leonardo jumping mechanography ground reaction force platform for muscle strength and muscle performance, to examine the relationship between these measurements and 25 OHD [38]. Single two leg jump, multiple one leg hop, and balance testing are performed to estimate muscle strength, efficiency of movement, maximum voluntary force, maximum acceleration, stiffness, energy storage capacity, and Esslinger Fitness Index.

Sun Exposure/SunSmart Behaviour

Real-time UVB exposure is measured objectively in all participants at baseline, at 4 and after 12 months of intervention using a small, discreet, wearable UV dosimeter, with a sampling interval of 30 seconds to provide real-time profiles of UV-exposure during 14 consecutive days before the visit. The dosimeter is worn on the wrist like a watch. Participants also complete a log and standard questionnaire about sunlight exposure and SunSmart behaviour [46], for comparison with the objective data. Logs are to report clothing worn, sunscreen use, sunburn, and when they took the watch off/put on.

292

Statistical Analysis

The primary outcome in this study is the change in serum 25 OHD concentration at 4 months.

Secondary outcomes include: objectively measured sunlight exposure, SunSmart behaviour, compliance rates for the two interventions, and musculoskeletal health measures at 12 months.

Exploratory outcomes include metabolic profiles, body composition and weight, atopic/allergic symptoms, mood and mental health, knowledge about sun-safe behaviour, as well as defining the determinants of vitamin D status in young women using baseline data and investigating the effects of vitamin D improvement on metabolic profiles, body composition and weight, atopic/allergic symptoms, mood, and mental health after 4 months. Main analyses will be conducted on an ITT basis to compare the intervention arms against the control group at 4 months, using various imputation strategies to account for missing data arising from sample attrition.

Due to possible protocol violations (noncompliance with the prescribed treatment and unblinding occurrences), a secondary per protocol analysis designed to adjust for noncompliance will be undertaken at 4 and 12 months and results compared with the ITT analysis. Subjects in the control and behavioural intervention arms who start taking vitamin D supplements during the trial and noncompliers with the interventions were excluded from the per protocol analysis.

Kolmogrov-Smirnov test and histogram chart will be used to assess the normality of continuous variables. Baseline general characteristics will be examined using two-way analysis of variance

(ANOVA) for continuous variables and chi-square for categorical variables. Two-way ANOVA will be used to determine the effects of supplementation and behavioural intervention. Tukey's

293 post-hoc comparisons will be used to identify pair wise differences when we reach a significant finding in multivariate regression. P-values <0.05 will be considered as significant. All statistical analyses will be performed using the Statistical Package for Social Science version 22.

Compliance and Withdrawal

The measurement of study compliance may vary between the three groups. The app contains built-in timers and data that allow the study team to determine how often the participants in the behavioural intervention group access the app.

Unblinded study staff undertake a manual count of the supplements at the 4 and 12 months visits, on return of the container of supplements. These data are recorded and a compliance percentage is determined. Unblinded study staff call participants in the control group to ask if they have read and understood the information brochure 2 weeks after randomization and then each month.

Any participants who have commenced vitamin D supplementation throughout the course of the study, and are not in the pharmacological intervention arm, remain in the study, and their data are analysed using the ITT analysis. Subsequently, per protocol analyses are performed, excluding the results of these participants.

Any participant who fails to meet the eligibility criteria for the duration of the study is invited to complete their remaining study visits for ITT analyses. In addition, any participant who commences participation in the study but, at any stage of the study, withdraws their consent, is withdrawn from the study. Any data that they have contributed to the study continues to be used, unless the participant states specifically that they would like their data to be deleted. This process

294 is formally stated to the participant before they commence the study.

The study team make reasonable attempts to contact a participant before withdrawing them from the study. This includes sending emails, text messages, making telephone calls, and sending written letters to the participant’s home address. If no contact can be made with the participant using all mentioned methods three times, the participant is withdrawn from the study and a letter is sent to them to explain this decision.

Ethical and Legal Considerations

This study has received approval from the Melbourne Health Human Research and Ethics

Committee (HREC) and is conducted according to the principles and rules laid down in the

Declaration of Helsinki and its subsequent amendments. It is carried out according to the revised

National Statement on Ethical Conduct in Research Involving Humans (2007) produced by the

National Health and Medical Research Council of Australia [38]. This national statement was developed to protect the interests of people who participate in research studies.

Mandatory reporting requirements pertaining to physical or sexual abuse of minors are adhered to by the study team and incorporated into the PICF so that participants are aware of the obligations of the study team. Data are kept confidential except if required by law. Information regarding illegal drug use may be disclosed to relevant authorities if required by law.

Clinically-Significant Results

All participants who have an abnormal pathology result, which is considered to be clinically significant after review by the principal investigator are contacted and/or receive the results by

295 mail or telephone, depending on the urgency of the matter. Specifically, any participant who records a serum 25 OHD result lower than 25 nmol/L is withdrawn and is referred to their treating general practitioner (GP) for advice and follow-up.

Adverse Events

An adverse event is defined as any occurrence that has unfavourable and/or unintended effects on research subjects, regardless of severity or study-relatedness. Adverse events may manifest as new findings (signs, symptoms, diagnoses, laboratory results) or alterations in pre-existing conditions. All adverse events occurring during the study are recorded whether or not they are considered to be serious and/or related to the study. Any major adverse events are reported in writing to the Melbourne Heath HREC. Based on the self-reported and UV dosimeter data, any subjects receiving UV-B exposure at levels deemed to place them at risk are advised that they are at high risk and provided with further information about SunSmart behaviour and safe sun exposure.

Results

Recruitment is currently underway. Publication of trial results is anticipated in 2017.

Discussion

Trial Implications

Causes of vitamin D deficiency include decreased sun exposure, use of UV-B blocking sunscreens, low dietary intake of vitamin D, obesity, and possibly smoking. Sunlight exposure is the major source of vitamin D (through the internal synthesis of vitamin D3), accounting for 80%

296 to 90% of circulating vitamin D metabolites, few foods other than fatty fish contain vitamin D

[47].

To the investigators’ knowledge, this study is the first randomized trial to assess the effectiveness of an m-health lifestyle intervention to safely improve vitamin D status in young women, and addresses the limitations of previous studies that have been small, limited by the imprecision of traditional assays for measuring 25 OHD levels, and/or lacking data on adolescents and younger women. In addition, recruitment via Facebook and subsequent enrolment into an intervention trial is novel in this study. Recruitment via Facebook can be effective at reaching a large number of potential participants, as well as improving affiliated costs. Also it is affordable and reach a wide diverse population, thus increasing generalizability

[48]. If successful, the smartphone m-health intervention developed for this study would be readily deployable throughout Australia and internationally, translatable to different demographics, and has the potential to be incorporated into health promotion initiatives and clinical care. The extensive data collection being undertaken by this study allows for the identification of possible relationships between vitamin D status and a range of other health conditions. As there are seasonal differences in 25 OHD levels, all measurements are repeated after 1 year of intervention.

Limitations

This study has a number of limitations. First, blinding is a difficult issue to address. As far as practically possible, study staff are blinded to the group to which participants are allocated throughout the study. However, we are not able to blind participants to their intervention group

297 allocation. Another limitation of our study is using Facebook advertising for recruitment, which can cause selection bias. However, in the pilot study on feasibility for the same age group reasonable representativeness with the general population was achieved apart from selecting those with a higher level of education [10]. This is common to most methods used for recruitment of subjects from general populations.

Acknowledgments

Authors would like to thank the Boosted Human members for developing the Safe-D application,

Amanda Hawker, Anna-Louise Ponsonby and Robyn Lucas, Ashwin Swami Nathan, Rachel

Slatyer, Jessica Cargill for skin casting and photographing, Cancer Council Victoria, Melbourne

Health Pathology Service and Stefanie Hartley, Adele Rivers, Anna Leigh Scobie, Jemma

Christie, Ashwini Kale, Asvini Subasinghe, Yasmin Jayasinghe, Kerryn King, Catherine Segan,

Adrian Bickerstaffe, Elisa Young, Jen Makin, Alison Brodie, Johannes Willnecker, Maria

Bisignano, George Varigos, Kim Bennell, Nicola Reavley, Marie Pirotta, Steve Howard (dec.),

Tony Jorm Peter Gies and all other members of the Safe-D study and YFHI study. Authors also would like to thank all participants for their time and effort. The Safe-D study has received in- kind support from Swisse Wellness to provide the supplements. This project is funded by a

NHMRC project grant, APP1049065.

298

References

1. Holick M, Chen T. Vitamin D deficiency: a worldwide problem with health consequences. Am J Clin Nutr. 2008;87:1080S–1086S. http://www.ajcn.org/cgi/pmidlookup?view=long&pmid=18400738. [PubMed] 2. Theodoratou E, Tzoulaki I, Zgaga L, Ioannidis JPA. Vitamin D and multiple health outcomes: umbrella review of systematic reviews and meta-analyses of observational studies and randomised trials. BMJ. 2014;348:g2035. http://www.bmj.com/cgi/pmidlookup?view=long&pmid=24690624. [PMC free article] [PubMed] 3. Colón-Emeric C, Saag K. Osteoporotic fractures in older adults. Best Pract Res Clin Rheumatol. 2006;20:695–706. [PMC free article] [PubMed] 4. Marjo KL, Timo TM, Ilpo ON, Kerttu MI, Aila EL, Jorma SV. Vitamin D and attainment of peak bone mass among peripubertal Finnish girls: a 3-y prospective study1-3. Am J Clin Nutr. 2002;76:1446–1453. [PubMed] 5. Faden VB, Day NL, Windle M, Windle R, Grube JW, Molina BSG, Pelham WE, Gnagy EM, Wilson TK, Jackson KM, Sher KJ. Collecting longitudinal data through childhood, adolescence, and young adulthood: methodological challenges. Alcohol Clin Exp Res. 2004;28:330–340. http://europepmc.org/abstract/MED/15112941. [PMC free article] [PubMed] 6. Australian Bureau of Statistics (ABS) National Health Survey. Canberra (ACT): Catalogue 21. 2001;4364 7. World Health Organisation (WHO) Women and Health: Today's Evidence Tomorrow's Agenda, November. 2009. [2016-04-07]. http://www.refworld.org/docid/4af82d3d2.html webcite. 8. Bruyère O, Cavalier E, Souberbielle J, Bischoff-Ferrari H, Beaudart C, Buckinx F, Reginster J, Rizzoli R. Effects of vitamin D in the elderly population: current status and perspectives. Arch Public Health. 2014;72:32. [PMC free article] [PubMed] 9. Callegari ET, Reavley N, Garland SM, Gorelik A, Wark JD. Vitamin D status, bone mineral density and mental health in young Australian women: the Safe-D Study. J Public Health Res. 2015;4:594. doi: 10.4081/jphr.2015.594. http://dx.doi.org/10.4081/jphr.2015.594. [PMC free article] [PubMed] [Cross Ref] 10. Fenner Y, Garland S, Moore E, Jayasinghe Y, Fletcher A, Tabrizi S, Gunasekaran B, Wark J. Web-based recruiting for health research using a social networking site: an exploratory study. J Med Internet Res. 2012;14:e20. doi: 10.2196/jmir.1978. http://www.jmir.org/2012/1/e20/ [PMC free article] [PubMed] [Cross Ref] 11. Institute of Medicine . Dietary Reference Intakes for Calcium and vitamin D. Washington, DC: The National Academies Press; 2011. 12. Putman MS, Pitts SAB, Milliren CE, Feldman HA, Reinold K, Gordon CM. A randomized clinical trial of vitamin D supplementation in healthy adolescents. J Adolesc Health. 2013;52:592–598. doi: 10.1016/j.jadohealth.2012.10.270. http://europepmc.org/abstract/MED/23608721. [PMC free article] [PubMed] [Cross Ref]

299

13. Suliivan SS, Cobb JL, Rosen CJ, Holick MF, Chen TC, Kimlin M, Michael G, Parisi A. Assessment of sun exposure in adolescent girls using activity diaries. Nutr Res. 2003;23:631– 644. 14. Daly RM, Gagnon C, Lu ZX, Magliano DJ, Dunstan DW, Sikaris KA, Zimmet PZ, Ebeling PR, Shaw JE. Prevalence of vitamin D deficiency and its determinants in Australian adults aged 25 years and older: a national, population-based study. Clin Endocrinol (Oxf) 2012;77(1):26–35. doi: 10.1111/j.1365-2265.2011.04320.x. [PubMed] [Cross Ref] 15. Gordon CM, DePeter KC, Feldman HA, Grace E, Emans SJ. Prevalence of vitamin D deficiency among healthy adolescents. Arch Pediatr Adolesc Med. 2004;158:531–537. doi: 10.1001/archpedi.158.6.531. [PubMed] [Cross Ref] 16. WHO Diet, nutrition, the prevention of chronic diseases. Report of a WHO Study Group. 1990:6–13. 17. Barnes M, Robson P, Bonham M, Strain J, Wallace JMW. Effect of vitamin D supplementation on vitamin D status and bone turnover markers in young adults. Eur J Clin Nutr. 2006;60:727–733. doi: 10.1038/sj.ejcn.1602374. [PubMed] [Cross Ref] 18. Orwoll E, Bliziotes M. Heterogeneity in osteoporosis. Men versus women. Rheumatic diseases clinics of North America. 1994;20 [PubMed] 19. Cancer Council Victoria, Be SunSmart. 2015. [2016-04-07]. http://www.cancervic.org.au/preventing-cancer/be-sunsmart webcite. 20. Heffernan K, Chang S, Maclean S, Callegari E, Garland S, Reavley N, Varigos G, Wark J. The potential of eHealth apps to support targeted complex health messages. J Gen Practice. 2014:182. 21. LimeSurvey project team. [2016-04-07]. LimeSurvey: An open Source Survey Tool https://www.limesurvey.org/ webcite. 22. Eserian J. Vitamin D as an effective treatment approach for drug abuse and addiction. Journal of Medical Hypotheses and Ideas. 2013;7(2):35–39. 23. Ireland P, Jolley D, Giles G, O'Dea K, Powles J, Rutishauser I, Wahlqvist ML, Williams J. Development of the Melbourne FFQ: a food frequency questionnaire for use in an Australian prospective study involving an ethnically diverse cohort. Asia Pac J Clin Nutr. 1994;3:19–31. http://apjcn.nhri.org.tw/server/APJCN/3/1/19.pdf. [PubMed] 24. Cranney A, Horsley T, O'Donnell S, Weiler H, Puil L, Ooi D, Atkinson S, Ward L, Moher D, Hanley D, Fang M, Yazdi F, Garritty C, Sampson M, Barrowman N, Tsertsvadze A, Mamaladze V. Effectiveness and safety of vitamin D in relation to bone health. Evid Rep Technol Assess. 2007:231–235. [PMC free article] [PubMed] 25. Leonard A, Hutchesson M, Patterson A, Chalmers K, Collins C. Trials. Recruitment and retention of young women into nutrition research studies: practical considerations. 2014;16:15– 23. [PMC free article] [PubMed] 26. Sanci LA, Sawyer SM, Weller PJ, Bond LM, Patton GC. Youth health research ethics: time for a mature-minor clause? Med J Aust. 2004;180:336–338. [PubMed] 27. Reavley N, Allen N, Jorm A, Morgan A, Purcell R. beyondblue: Melbourne. 2010. [2016-04- 07]. A Guide to What Works for Anxiety Disorders https://www.beyondblue.org.au/docs/default-source/resources/bl-0762-guide-to-what-works-for- anxiety-02-14.pdf?sfvrsn=2 webcite.

300

28. Jorm AF, Allen N, Morgan A, Purcell R. beyondblue: Melbourne, August. 2009. [2016-04- 07]. A Guide to What Works for Depression https://www.beyondblue.org.au/docs/default- source/resources/bl-0556-guide-to-what-works-for-depression-12-14.pdf?sfvrsn=2 webcite. 29. LivingWorks Australia . http://www.livingworks.com.au/ Australia: Living works,; 2015. [2016-04-04]. ASIST http://www.livingworks.com.au/programs/asist/ webcite. 30. McKenzie R, Liley J, Björn L. UV radiation: balancing risks and benefits. Photochem Photobiol. 2009:88–98. [PubMed] 31. Sachdeva S. Fitzpatrick skin typing: applications in dermatology. Indian J Dermatol Venereol Leprol. 2009;75:93–96. http://www.ijdvl.com/article.asp?issn=0378- 6323;year=2009;volume=75;issue=1;spage=93;epage=96;aulast=Sachdeva. [PubMed] 32. Fitzpatrick TB. The validity and practicality of sun-reactive skin types I through VI. Arch Dermatol. 1988;124:869–871. [PubMed] 33. Fogg B. A Behavior Model for Persuasive Design. ACM Proceedings of the 4th International Conference on Persuasive Technology ; 40; 2009; Claremont, CA. 2009. 34. Fine MR. Beta Testing for Better Software. New York: John Wiley & Sons; 2002. 35. Dolan R, Matthews J. Maximizing the utility of customer product testing: beta test design and management. Journal of Product Innovation Management. 1993;10:318–330. 36. How much sun is enough? [2016-04-07]. http://www.sunsmart.com.au/downloads/vitamin- d/how-much-sun-is-enough-vitamin-d.pdf webcite. 37. Wootton AM. Improving the measurement of 25-hydroxyvitamin D. Clin Biochem Rev. 2005;26:33–36. http://europepmc.org/abstract/MED/16278775. [PMC free article] [PubMed] 38. Matheson L, Duffy S, Maroof A, Gibbons R, Duffy C, Roth J. Intra- and inter-rater reliability of jumping mechanography muscle function assessments. J Musculoskelet Neuronal Interact. 2013;13:480–486. [PubMed] 39. Singh RJ. Are clinical laboratories prepared for accurate testing of 25-hydroxy vitamin D? Clin Chem. 2008;54:221–223. doi: 10.1373/clinchem.2007.096156. http://www.clinchem.org/cgi/pmidlookup?view=long&pmid=18160734. [PubMed] [Cross Ref] 40. Carter GD, Carter R, Jones J, Berry J. How accurate are assays for 25-hydroxyvitamin D? Data from the international vitamin D external quality assessment scheme. Clin Chem. 2004;50:2195–2197. doi: 10.1373/clinchem.2004.040683. http://www.clinchem.org/cgi/pmidlookup?view=long&pmid=15375018. [PubMed] [Cross Ref] 41. Lucas RM, Ponsonby A, Dear K, Taylor BV, Dwyer T, McMichael AJ, Valery P, van der Mei I, Williams D, Pender MP, Chapman C, Coulthard A, Kilpatrick T. Associations between silicone skin cast score, cumulative sun exposure, and other factors in the ausimmune study: a multicenter Australian study. Cancer Epidemiol Biomarkers Prev. 2009;18:2887–2894. doi: 10.1158/1055-9965.EPI-09-0191. http://cebp.aacrjournals.org/cgi/pmidlookup?view=long&pmid=19843682. [PubMed] [Cross Ref] 42. Battistutta D, Pandeya N, Strutton GM, Fourtanier A, Tison S, Green AC. Skin surface topography grading is a valid measure of skin photoaging. Photodermatol Photoimmunol Photomed. 2006;22:39–45. doi: 10.1111/j.1600-0781.2006.00194.x. [PubMed] [Cross Ref] 43. Holman CD, Armstrong BK, Evans PR, Lumsden GJ, Dallimore KJ, Meehan CJ, Beagley J, Gibson IM. Relationship of solar keratosis and history of skin cancer to objective measures of actinic skin damage. Br J Dermatol. 1984;110:129–138. [PubMed]

301

44. Crabtree NJ, Arabi A, Bachrach LK, Fewtrell M, El-Hajj FG, Kecskemethy HH, Jaworski M, Gordon CM, International Society for Clinical Densitometry Dual-energy X-ray absorptiometry interpretation and reporting in children and adolescents: the revised 2013 ISCD Pediatric Official Positions. J Clin Densitom. 2014;17:225–242. doi: 10.1016/j.jocd.2014.01.003. [PubMed] [Cross Ref] 45. Neu CM, Manz F, Rauch F, Merkel A, Schoenau E. Bone densities and bone size at the distal radius in healthy children and adolescents: a study using peripheral quantitative computed tomography. Bone. 2001;28:227–232. [PubMed] 46. Glanz K, Yaroch A, Dancel M, Saraiya M, Crane L, Buller D, Manne S, O'Riordan D, Heckman C, Hay J, Robinson J. Measures of sun exposure and sun protection practices for behavioral and epidemiologic research. Archives of dermatology. 2008;144:217. [PubMed] 47. National HealthMedical Research Council (NHMRC) National Statement on Ethical Conduct in Research Involving Humans. 2007. [2016-04-07]. http://www.nhmrc.gov.au/book/national- statement-ethical-conduct-human-research webcite. 48. Lane TS, Armin J, Gordon JS. Online Recruitment Methods for Web-Based and Mobile Health Studies: A Review of the Literature. J Med Internet Res. 2015;17:e183. doi: 10.2196/jmir.4359. http://www.jmir.org/2015/7/e183/ [PMC free article] [PubMed] [Cross Ref]

302

B) Facebook advertisement

Heading Link Text Target Location Target Age

Are you aged 16-25? Contribute to Women’s Health http://safedstudy.org/ women's health research & receive a $30 Victoria, Australia 16-25 Matters gift voucher.

303

C) Safe-D Part B brochure

304

305

D) Participant information sheet & consent form (PICF)

Safe-D Study PART B Participant Information Sheet & Consent Form

RMH HREC No: 2013.215

Title: Improving Vitamin D Status and Related Health in Young Women (the Safe-D Study) – Part B

Short Title: Improving Vitamin D Status in Young Women - Part B

Principal Investigator: Professor John D Wark

Associate Investigators: Dr Nicola Reavley, Associate Professor Marie Pirotta, Prof George Varigos, Ms Alexandra Gorelik, Professor Suzanne M. Garland, Professor Tony Jorm, Professor Kim Bennell, Dr Tharshan Vaithianathan, Associate Professor Shanton Chang, Ms Marjan Tabesh, Ms Adele Rivers, Ms Skye Maclean, Ms Emma Callegari

Location: Royal Melbourne Hospital

1. Introduction

You are invited to take part in this research project. You are being invited because you participated in Part A of this project, where it was found your vitamin D levels were low or below optimal.

This Participant Information and Consent Form (PICF) tells you about the research project and explains what is involved, to help you decide whether you want to take part. Please read this information carefully and ask questions about anything that you don’t understand or would like more information about.

306

Before deciding whether or not to take part, you might want to talk about it with a relative, friend, local health worker, or with a member of the study team (whose contact details are on the last page).

Participation in this research is voluntary. If you do not wish to take part, you do not have to.

If you decide you want to take part in the research project, you will be asked to sign the consent section.

By signing it you are telling us that you:

 understand what you have read;  consent to take part in the research project;  consent to participate in the research processes that are described;  consent to the use of your personal and health information as described

You will be given a copy of this Participant Information and Consent Form to keep.

2. What is the purpose of this research?

Vitamin D deficiency is an important health risk for young women, particularly during child-bearing years, and as many as 50% of Australian women are below optimal vitamin D levels. Vitamin D deficiency has been linked with many chronic health conditions including poor bone and muscle health, heart disease, autoimmune diseases, infections, neurological diseases, cancer, diabetes and mental health problems. Major causes of vitamin D deficiency are decreased exposure to sunlight and low intake of dietary forms of vitamin D.

The Safe-D study aims to investigate the relationships between vitamin D levels and other aspects of women’s health, including bone and muscle health, ultraviolet light (sun) exposure, mental health, exercise, nutrition and allergies. The findings from this study will lead to better education and policies to improve vitamin D status in young Australian women.

Part B of this study aims to investigate the safest and most effective way to increase vitamin D levels in vitamin D deficient young women. Specifically, the research aims to see if safe levels of sun exposure can increase vitamin D levels as much as taking a vitamin D supplement. We aim to recruit 234 women aged 16-25 years from across Victoria, to participate in part B of this study.

307

This research has been initiated by the principal investigator, Professor Wark, and has been funded by the National Health and Medical Research Council (NHMRC).

3. What does participation in this research involve?

We invite you to participate in all components of the study described below. It may be possible to participate in a limited way, if that is your preference, or you have a condition or restriction that prevents full participation. We anticipate that your participation in this study will be no longer than 12 months.

(a) Randomisation to different methods to increase vitamin D levels

Participants in this study will be randomised to one of three groups. This means a computer will randomly select a group for you to be in. There is a one in three chance of you being randomised to any one of the groups. You cannot choose which group you are in. Each group will use a different way of increasing vitamin D levels, so the researchers can study which way is safest and most effective.

Participants in group 1 will be given general information on improving vitamin D levels. Participants in group 2 will be asked to use a smart phone app to receive information and advice on safely improving vitamin D levels for 12 months. Participants will be able to enter information into the app about their sun exposure, and receive advice on safe and healthy sun exposure. An estimate of time necessary to spend in the sun to achieve healthy vitamin D levels will be provided to app users, along with daily and weekly achievements of sun exposure targets and comparisons with the rest of the group. Participants in group 3 will be asked to take a vitamin D capsule once a day for 12 months.

(b) Site visits at start of study; 4 months; and 12 months

If you consent to participate in Part B of this study, and it is within 2 weeks of your Part A Site Visit, all data collected at this visit will be used as a baseline for the Part B study. If it is between 2 and 8 weeks, you will be asked to have a repeat blood test and answer a brief questionnaire. If the visit is more than 8 weeks but less than 4 months, abbreviated online questionnaires and a brief site visit will need to be undertaken (no repeat of scans is necessary). If your Part A site visit was more than 4 months ago we will need to repeat all assessments, including modified questionnaires.

Participants in all groups will be asked to attend our study centre for a health check at 4 months and again at 12 months for a final comprehensive visit. This is so the study doctors can see whether your vitamin D levels are increasing over time, and which group has the greatest increase in vitamin D. We will also do tests to check whether sun exposure is having negative effects, such as skin damage. The site visits will be conducted at the Royal Melbourne Hospital, Parkville, Victoria. Study centre visits will take up to 1.5 – 2 hours. Visits will be scheduled in the morning and you will be asked to fast overnight and not eat or drink anything (apart from water) in the morning before your visit. At the study centre you will

308 be met by members of the study team who will guide you through your visit and answer any questions you may have. At the study centre visit you will have:

1. A physical examination including measures such as height, weight, blood pressure, resting heart rate, as well as a test of your balance and jumping power.

2. Collection of a sample of your blood which will be tested for markers of health and fitness (bone health, calcium, vitamin D, cholesterol, blood count, iron levels, human papillomavirus antibodies and hormone tests to check the function of your ovaries). Some of these blood markers may be stored to be tested at completion of the study subject to the availability of funds to cover the cost of the tests. We would also like to keep some blood in storage for future testing related to this study, for research purposes only. An appropriately trained member of the research team will be collecting the blood sample. They will draw approximately 50-60ml (about four tablespoons) of blood from a vein in your arm using a syringe and needle.

3. A silicon rubber cast of the back of your hand will be taken to measure effects on your skin from sun exposure. We will apply a liquid to the back of your hand and leave it to set for 7 min. The rubber cast is then removed, allowed to dry, and stored in a bag for a member of the research team to examine.

4. Skin pigmentation will be measured using a skin reflectance instrument called a spectrophotometer; this will take skin photographs of your hand, upper arm and cheek, to further evaluate your skin pigmentation.

5. Urine sample will be collected to test for pregnancy, where the possibility exists.

6. Bone density scans will be performed to measure bone mineral density and bone mineral content (to estimate bone strength), as well as measuring soft tissue (non-bone) body composition (lean or muscle mass and fat mass). The parts of your body to be scanned are regional (your lower leg, lower back, and hip) plus total body. For these tests, you’ll need to remove all metal jewellery and clothing with zips and/or buttons. Gowns will be provided if necessary.

You will be informed of the results of any of these tests that are related to your health.

In Part A of this study, you will have already completed a comprehensive online survey. If there is more than a 2 week and less than an 8 week gap we will ask you to answer a brief questionnaire. If the visit is more than 8 weeks and less than 4 months, an abbreviated questionnaire is to be undertaken. If the visit is over 4 months from your Part A visit we will need you to redo the baseline questionnaire component of

309 study (not including questions related to sexual health and sexual experiences). We will invite you to complete a modified version (much shorter) of this survey after participating in the study for 4 months, and then complete the full survey again after 12 months. Each time you complete the survey, it will take around 2 hours to complete (45 to 60 mins for the modified version). You may complete these questionnaires from your home, or wherever you have internet access. You do not need to complete the entire survey in one go; instead, you are welcome to complete the survey over a period of 1 week, and we will send reminder emails and/or text messages to help you to remember to finish the survey. The questionnaires will cover a range of issues including general health and wellbeing, mental health, sunlight exposure and SunSmart behaviour, muscle and bone health, medical history, physical activity, diet, allergies, complementary medicine and supplement use, smoking, medication use, illicit drug use, relationships, periods, contraception (such as use of the pill), pregnancies, how you feel about your quality of life, and how you feel about your body. You will be asked questions of a sensitive and personal nature, including your mental health status. You will also be asked about illegal drug use, including questions such as “in your lifetime have you ever used cocaine or opioids?” If your answers indicate that you are clinically depressed or anxious, or at risk of self-harm, a member of the Safe-D team may contact you via email or telephone to follow up on your welfare.

(d) Wearing of a UV monitor

As with Part A of this study, you will be asked to wear a small device that measures ultraviolet (UV) radiation from the sun for a period of 14 consecutive days. This will occur after participating in the study for 4-months and again at 12-months. The study team will send the UV monitor to you in an express post parcel.

Reimbursement

You will not be paid for your participation in this research, but you will receive some compensation for participating, as a token for your time. Upon completion of the initial online questionnaire and return of the UV dosimeter (in Part A of the study), you will receive a $30 voucher. On completion of each site visit in Part B, a further $30 voucher will be provided to help compensate you for your time and travel costs for your participation in this research.

If you are randomised to group 3 in the study, you will be given vitamin D capsules free of charge for 12 months.

4. What will happen to my test samples?

Following your attendance at the Royal Melbourne Hospital for a site visit, your samples will be managed as follows:

310

Blood samples: The blood will be stored in a container to be sent to the pathology laboratory, a private provider in Melbourne. This laboratory will perform blood tests to measure chemical markers of bone health and other health indicators. After preparation by the laboratory, another container of blood will be sent to a different laboratory in Melbourne, Victoria to test for vitamin D levels.

Rubber cast: The rubber cast of the back of your hand will be assessed by a trained researcher to assess the level of sun damage to your skin. The casts will be stored in a folder in a secure location at the Royal Melbourne Hospital.

Skin Photographs: The photographs of your skin will be assessed by a trained researcher in order to evaluate skin pigmentation. The photographs will be stored in a folder in a secure location at the Royal Melbourne Hospital. You will not be able to be identified from the photographs and your name will not be used to label these.

All blood samples sent to laboratories for testing will be labelled with your participant number, initials and date of birth; your name will not be used when labelling the samples. The rubber cast and skin photographs will be labelled with your participant number, initials and date of birth; your name will not be used.

Optional storage of samples for future research

After the samples have been analysed, we would like your permission to store your samples indefinitely in freezers at the Royal Melbourne Hospital to be used again in related future research. This part of the study is optional. You can choose not to consent to this and still participate in the study. Future testing is unlikely to have a direct use for your future health; however you will be contacted and informed of your results if the results are deemed of direct use. It is not foreseen that future testing will be of a genetic nature. You will be informed if there are any changes to the nature of the testing and can revoke your consent for the study team to use the samples at any time.

5. What are the possible benefits of taking part?

We cannot guarantee or promise that you will receive any benefits from this research. However, possible benefits include:

 You may increase your vitamin D levels  You may gain a better understanding of some aspects of health.

311

 You may benefit from tests/assessments that identify risk factors towards future health problems, such as low vitamin D levels, low bone mineral density, high UV (sun) exposure, which may be avoided with healthy lifestyle and behaviour modification or medical treatment.  The potential contribution to assist the future medical care of young women.  The potential contribution to health resources for young women and health professionals.

6. What are the possible risks?

Possible risks, side effects and discomforts include:

 This project may lead to you increasing your exposure to sunlight. Too much sun exposure can contribute to the development of skin changes and skin cancers. The aim of this project is to see whether safe levels of sun exposure can increase vitamin D levels and reduce vitamin D deficiency. The periods of time in the sun recommended in this project are considered safe for most people. However, if you or a member of your immediate family have ever had a melanoma (type of skin cancer), you will not be able to participate in this research project as you may be at greater risk from sun exposure.  The collection of blood may sometimes cause side effects. These include momentary discomfort when the needle is inserted, soreness (pain), redness, fainting or light-headedness (dizziness). These reactions usually last only a short time. Rarely, there could be a minor infection, bleeding or the formation of a local blood clot in the vein that was punctured. If this happens, it can be easily treated.  Taking part in the study may uncover risk factors for your health in the future. For example having low bone density may put someone at risk of osteoporosis later in life. The good thing is that many of the health risks we are testing for are reversible or at least modifiable with changes in health behaviour and with medical treatment, and the study team may assist with organising follow-up with your local doctor as required. However, participation in this project should not be considered an alternative to your normal medical care with your usual doctor.  The survey asks sensitive health questions about mental health. These questions may be embarrassing or disturbing to some. We will use standard ways of collecting data about these important issues. While the questions are raised as sensitively as possible, they are also quite direct. If you expect to feel unduly embarrassed, confronted or disturbed by these questions, you need not answer those questions. Your name will not be recorded on the survey; however your study participant number will be recorded on your survey. As such, your answers will be re- identifiable to study researchers with access to the password secure study database.  During this research study you may undergo radiographic examinations that you would not normally receive. This research study involves quantitative dual energy x-ray analysis (DEXA) and peripheral quantitative computed tomography (pQCT) scans of your skeleton, with exposure to a very small amount of radiation that you would not normally receive. As part of everyday living, everyone is exposed to naturally occurring background radiation and receives a dose of about 2 millisievert (mSv) each year. The effective dose from this study is about 0.011 mSv. At this dose level, no harmful effects of radiation have been demonstrated as any effect is too small to

312

measure. The risk is believed to be minimal. If you are under the age of 18 years, you should inform us of any other studies that you have participated in that involve the use of radiation.  Exposure to ionising radiation can adversely affect an unborn child and newborn baby. Because of this, it is important that participants receiving ionising radiation exposure through bone scans are not pregnant. You must not participate in the bone scans if you are pregnant. If you are sexually active, we will try to schedule the bone scans during first 12 days of your menstrual cycle, to help minimise risk, and you may be required to have a pregnancy test prior to the bone scans.  There is minimal risk of injury from falling while undergoing tests of your balance and jumping power.

Our research staff can arrange for counselling or other appropriate support for you, if you become distressed. Any counselling or support will be provided by professionals who are not members of the research team. You may suspend or end your participation in the project if distress occurs.

There may be additional risks that the researchers do not expect or do not know about. Tell a member of the research team immediately about any new or unusual symptoms that you get.

7. What if new information arises during this research project?

During the research project, new information about the risks and benefits of the project may become known to the researchers. If this occurs, you will be told about this new information and the researcher will discuss whether this new information affects you.

8. Are there alternatives to participation?

Participation in this research is not your only option. You may discuss alternative options to improve your vitamin D status with your regular physician. Discuss these options with your healthcare worker before deciding whether or not to take part in this research project.

9. Do I have to take part in this research project?

Participation in any research project is voluntary. If you do not wish to take part, you do not have to. If you decide to take part and later change your mind, you are free to withdraw from the project at a later stage. If you do consent to participate, you may withdraw at any point.

Your decision whether to take part or not, or to take part and then withdraw, will not affect your relationship with the researchers, the University of Melbourne or The Royal Melbourne Hospital.

313

10. Am I eligible to take part in this research project?

All participants must meet the following eligibility criteria to be able to participate in this study:

- Female - Aged between 16 and 25 years on the date you expressed an interest in participating in this study - Currently living in Victoria, with no plans to relocate interstate in the next 12 months

In addition, if you are in any of the following categories, you will not be able to participate in this study:

- Currently pregnant or breastfeeding, or trying to conceive - Have ever had a melanoma, or have a first-degree relative who has ever had a melanoma (mother, father, sister, brother) - Suffer from a medical condition that makes you sensitive to UV and/or sunlight - Do not currently own and use a smartphone

11. What if I withdraw from this research project?

If you decide to withdraw from the project, please notify a member of the research team. This notice will allow that person or the research supervisor to discuss any health risks or special requirements linked to withdrawing. If you decide to leave the project, the researchers will keep your survey data, and results of samples that have already been tested. This is to help them make sure that the results of the research can be evaluated properly. If you do not want them to do this, you should choose not to enter the research project.

If you withdraw we will also ask you if we can continue to store and use the samples that have been collected.

12. How will I be informed of the results of this research project?

When the study is over, we will write up the results in a short report and distribute them to you via email and on our study’s group page on Facebook. We will also publish study results in various scientific journals. You will not be identifiable in the results.

You can also elect to request results of any specific tests that are relevant to your health and are performed as part of the study. Any results that are show significant abnormalities and/or are relevant to clinical care will be communicated to you and/or your usual doctor with your permission. Please note, abnormal results do not necessarily indicate that you have a condition or medical problem. Therefore we advise that you seek further advice regarding the results of any test results.

314

13. What else do I need to know? a) What will happen to information about me?

By signing the consent form you consent to the relevant research staff collecting and using personal information about you for the research project. Any information obtained in connection with this research project that can identify you will remain confidential. The information collected for this study will be re-identifiable. This means that the researchers will assign a code (study participation number) to your name. This code will be used, along with your initials and date of birth, to record all information collected about you; your name will not appear on these records. Study researchers will be able to connect the code to your identifiable information (e.g. full name and address) only if they have access to the password protected study database. All of the information will be stored on a password-protected computer, on a secure network, in a highly-secure research building with 24-hour security personnel. Your information will only be used for the purpose of this research project and possibly for future related research as described in this document. It will only be disclosed with your permission, except as required by law. This means that information you provide regarding illegal drug use may be disclosed to relevant authorities if required by law. The information collected in this research project will be kept for at least 15 years.

We would also like permission to contact you in the future for future research studies related to this project. For example, as a result of this study we may develop some health and education tools that we would like to test. All future studies will require ethical approval from a committee designed to protect the rights of people in research studies.

It is anticipated that the results of this research project will be published and/or presented in a variety of forums. In any publication and/or presentation, information will be provided in such a way that you cannot be identified. b) How can I access my information?

In accordance with relevant Australian and/or Victorian privacy and other relevant laws, you have the right to access the information collected and stored by the research team. You also have the right to request that any information with which you disagree be corrected. Please contact one of the study team members named at the end of this document if you would like to access your information. c) What happens if I am injured as a result of participating in this research project?

If you suffer an injury as a result of participating in this research project, you should contact the study team as soon as possible and you will be assisted with arranging appropriate medical treatment. You can

315 receive any medical treatment required to treat the injury or complication, free of charge, as a public patient in any Australian public hospital. d) Is this research project approved?

The ethical aspects of this research project have been approved by the Melbourne Health Human Research Ethics Committee. This project will be carried out according to the National Statement on Ethical Conduct in Human Research (2007). This statement has been developed to protect the interests of people who agree to participate in human research studies.

14. Consent a.) Electronic Consent

Those participants who are at least 18 years old can opt to provide initial electronic consent rather than written consent. Electronic consent would be obtained over a secure link sent by the study team. Information contained in the Participant Information Sheet will be displayed onscreen and requires participants to click “I agree” to a statement that they have read and understood the information and freely agree to participant in the research project. If you are not randomised (that is, allocated to a group in part-B of the Safe-D study) within 2 weeks from the time of your part A site visit, another baseline visit is needed to check your vitamin D status.

You would be asked to provide written consent at a later date as confirmation of your agreement to participate. b.) Written Consent

 I have read the Participant Information Sheet or someone has read it to me in a language that I understand.  I understand the purposes, procedures and risks of the research described in the project.  I have had an opportunity to ask questions and I am satisfied with the answers I have received.  I freely agree to participate in this research project as described and understand that I am free to withdraw at any time during the study without affecting my future health care.  I understand that I will be given a signed copy of this document to keep.

Optional Consent: (Initial to indicate your choice)

I consent to the storage and use of my samples for use in future research that is related to this research as described in Section 4 of this document. Yes: No:

316

I consent to be contacted in the future to consider participating in future related research

Name of Researcher (please print) ______

Signature ______Date ______

Yes: No:

Name of Participant (please print) ______

Signature ______Date ______

Declaration by Researcher (OFFICE USE ONLY)

I have given a verbal explanation of the research project, its procedures and risks and I believe that the participant has understood that explanation.

Further information and who to contact

The person you may need to contact will depend on the nature of your query. Therefore, please note the following:

For further information:

317

If you want any further information concerning this project or if you have any problems which may be related to your involvement in the project (for example, feelings of distress), you can contact the following:

Name Professor John Wark

Position Head, Bone and Mineral Medicine, Royal Melbourne Hospital

Telephone 03 8344 3258

Email [email protected]

Name Skye Maclean

Position Research Assistant

Telephone 0466 520 387

Email [email protected]

For complaints: If you have any complaints about any aspect of the project, the way it is being conducted or any questions about being a research participant in general, then you may contact:

Name Ms. Jessica Turner

Position Manager Melbourne Health Human Research Ethics Committee

Telephone (03) 9342 8530

318

E) Photos

Spectrometer

Skin cast

Dosimeter device

319

Safe-D app

320

F) LimeSurvey questionnaire

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

337

338

G) Clothing chart

339

340

H) Sun monitoring device log

Safe-D Study (Part-B) Improving vitamin D status and related health in young women Royal Melbourne Hospital

ID No: ______Name: ______

When did you START wearing the watch (day 1) ? Date: ____/____/____ Time: ____:____ AM / PM

When did you STOP wearing the watch (day 14) ? Date: ____/____/____ Time: ____:____ AM / PM

Sometimes you will remove your watch: You may wish to remove the watch for periods of time over the week. Please write down when you take the watch off, when you put it back on and where you were during the time.

341

NOTE: The sun monitoring device is not waterproof; therefore you’ll need to remove the monitor when undertaking activities where the monitor is at risk of getting wet, e.g. showering or swimming. Please re-attach the monitor as soon as you can and record these periods using the activity log provided below

342

Time monitor put Time monitor Where were you during this Why did you remove the sun Did you go on taken off time? Date outside monitoring watch? today? Outside Outside e.g. swimming, showering, sleeping Inside (Shade) (Sun)

Example 01/01/2014 Yes/no 1:30 AM / PM 12:45 AM / PM Swimming 

: AM/PM : AM/PM

Day 1 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 2 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 3 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 4 : AM/PM : AM/PM

: AM/PM : AM/PM

343

Day 5 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 6 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 7 : AM/PM : AM/PM

: AM/PM : AM/PM

344

Time monitor put Time monitor Where were you during this Why did you remove the activity Did you go on taken off time? Date outside monitor? today? Outside Outside e.g. sleeping, dancing, resting Inside (Shade) (Sun)

Example 01/01/2014 Yes/no 12:45 AM / PM 1:30 AM / PM Resting inside 

: AM/PM : AM/PM

Day 8 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 9 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 10 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 11 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 12 : AM/PM : AM/PM

345

: AM/PM : AM/PM

Day 13 : AM/PM : AM/PM

: AM/PM : AM/PM

Day 14 : AM/PM : AM/PM

: AM/PM : AM/PM

346

BIBLIOGRAPHY

347

Abboud, M., Gordon-Thomson, C., Hoy, A.J., Balaban, S., Rybchyn, M.S., Cole, L., Su, Y., Brennan-Speranza, T.C., Fraser, D.R., Mason, R.S. (2014). Uptake of 25- hydroxyvitamin D by muscle and fat cells. J Steroid Biochem Mol Biol, 144:232-236.

Agbalalah, T., Hughes, S.F., Freeborn, E.J., Mushtaq, S. (2017). Impact of vitamin D supplementation on endothelial and inflammatory markers in adults: A systematic review. J Steroid Biochem Mol Biol, 173:292-300.

Australian Institute of Health and Welfare. (2017). Cancer in Australia 2017. Cancer series no. 101. Cat. No. CAN 100. Canberra: AIHW.

Ainsworth, B.E., Haskell, W.L., Whitt, M.C., Irwin, M.L., Swartz, A.M., Strath, S.J., O'Brien, W.L., Bassett D.R., Schmitz K.H., Emplaincourt P.O., Jacobs D.R. Jr., Leon A.S. (2000). Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc, 32(9):498-504.

Ajabshir, S., Asif, A., Nayer, A. (2014). The effects of vitamin D on the renin-angiotensin system. J Nephropathol 3(2): 41-43.

Al-Dujaili, E.A.S., Munir, N., Revuelta Iniesta, R. (2016). Effect of vitamin D supplementation on cardiovascular disease risk factors and exercise performance in healthy participants: a randomized placebo-controlled preliminary study. Ther Adv Endocrinol Metab, 7(4): 153-165.

Aloia, J.F., Patel, M., Dimaano, R., Li-Ng, M., Talwar, S.A., Mikhail, M., Pollack, S., Yeh, J.K. (2008). Vitamin D intake to attain a desired serum 25-hydroxyvitamin D concentration. Am J Clin Nutr, 87(6):1952–1958.

Alshishtawy, M.M. (2012). Vitamin D Deficiency. Sultan Qaboos University Medical Journal, 12(2): 140-152.

Al Shoumer, K.A., Al-Essa, T.M. (2015). Is there a relationship between vitamin D with insulin resistance and diabetes mellitus? World J Diabetes, 25; 6(8): 1057-1064.

Aly, Y.E., Abdou, A.S., Rashad, M.M., Nassefb, M.M. (2016). Effect of exercise on serum vitamin D and tissue vitamin D receptors in experimentally induced type 2 Diabetes Mellitus. J Adv Res, 7(5): 671-679.

American Academy of Dermatology. (n.d.). Skin cancer. Retrieved from

348 https://www.aad.org/media/stats/conditions/skin-cancer

Amini, Z., Bryant, S., Smith, C., Singh, R., Kumar, S. (2013). Is the serum vitamin D- parathyroid hormone relationship influenced by obesity in children? Horm Res Paediatr, 80(4):252-256.

Andersen, R., Brot, C., Mejborn, H., Mølgaard, C., Skovgaard, L.T., Trolle, E., Ovesen, L. (2009). Vitamin D supplementation does not affect serum lipids and lipoproteins in Pakistani immigrants. European Journal of Clinical Nutrition, 63(9),1150-1153.

Anta, R.M.O., González-Rodríguez, L.G., Jiménez Ortega, A.I., Estaire Gómez, P., Rodríguez-Rodríguez, E. , Perea Sánchez, J.M , Vizuete, A.A. (2012). Insufficient Intake of Vitamin D in Spanish Schoolchildren: Determinants of the Problem and Basis for Its Improvement. Nutr Hosp, 27 (5), 1437-1443.

Apalla, Z., Nashan, D., Weller, R.B., Castellsagué, X. (2017). Skin Cancer: Epidemiology, Disease Burden, Pathophysiology, Diagnosis, and Therapeutic Approaches. Dermatol Ther (Heidelb), 7:5-19.

Appelman, Y., van Rijn, B.B., Ten Haaf, M.E., Boersma, E., Peters, S.A. (2015). Sex differences in cardiovascular risk factors and disease prevention. Atherosclerosis, 241(1):211-218.

Armstrong, A.W., Watson, A.J., Makredes, M., Frangos, J.E., Kimball, A.B., Kvedar, J.C. (2009). Text-message reminders to improve sunscreen use: a randomized, controlled trial using electronic monitoring. Arch Dermatol, 145(11):1230-1236.

Armstrong, B.K., Kricker, A. (1993). How much melanoma is caused by sun exposure? Melanoma Res, 3(6):395-401.

Asemi, Z., Hashemi, T., Karamali, M., Samimi, M., Esmaillzadeh, A. (2013). Effects of vitamin D supplementation on glucose metabolism, lipid concentrations, inflammation, and oxidative stress in gestational diabetes: a double-blind randomized controlled clinical trial. Am J Clin Nutr, 98(6):1425-1432.

Atkinson, M.A., Eisenbarth, G.S., Michels, A.W. (2014). Type 1 diabetes. Lancet, 4; 383(9911): 69-82.

Au, L.E., Rogers, G.T., Harris, S.S., Dwyer, J.T., Jacques, P.F., Sacheck, J.M. (2013).

349

Associations of vitamin D intake with 25-hydroxyvitamin D in overweight and racially/ethnically diverse US children. J Acad Nutr Diet, 113(11):1511-1516.

Australian Bureau of Statistics (2015). National Health Survey: First Results. Retrieved from www.abs.gov.au

Australian Bureau of Statistics. (2016). Household Income and Wealth, Australia, 2015- 16. Retrieved from http://www.abs.gov.au/ausstats/[email protected]/mf/6523.0.

Australian Bureau of Statistics. (2017). Australian Social Trends. Retrieved from http://www.abs.gov.au/AUSSTATS/[email protected]/2f762f95845417aeca25706c00834efa/47f1 51c90ade4c73ca256e9e001f8973

Australian Bureau of Statistics. (2017). Census of Population and Housing: Socio- Economic Indexes for Areas (SEIFA), Australia. Retrieved from http://www.abs.gov.au/ausstats/[email protected]/mf/2033.0.55.001.

Australian Health Survey: Biomedical Results for Nutrients, 2011-12 (2013, December). Retrieved from http://www.abs.gov.au/ausstats/abs

Australian Institute of Health and Welfare (2003). The Active Australia survey. Canberra. AIHW.

Australian Institute of Health and Welfare (2011). Cardiovascular disease: Australian facts 2011. Cardiovascular disease series. Cat. no. CVD 53. Canberra: AIHW.

Australian Institute of Health and Welfare. (2017). Authoritative information and statistics to promote better health and wellbeing. Cardiovascular disease, diabetes and chronic kidney disease-Australian facts: risk factors. Retrieved from www.aihw.gov.au

Azadbakht, L., Mirmiran, P., Esmaillzadeh, A., Azizi, F. (2005). Dairy consumption is inversely associated with the prevalence of the metabolic syndrome in Tehranian adults. Am J Clin Nutr,82(3):523-530.

Badawy, S.M., Kuhns, L.M. (2017). Texting and Mobile Phone App Interventions for Improving Adherence to Preventive Behavior in Adolescents: A Systematic Review. JMIR Mhealth Uhealth, 5(4):e50.

Baker, C.P., Kulkarni, B., Radhakrishna, K.V., Charyulu, M.S., Gregson, J., Matsuzaki,

350

M., Taylor, A.E., Prabhakaran, D., Mamidi, R.S., Wells, J., Wilkinson, I., McEniery, C., Yasmin Davey Smith, G., Ben-Shlomo, Y., Kuper, H., Kinra, S. (2015). Is the Association between Vitamin D and Cardiovascular Disease Risk Confounded by Obesity? Evidence from the Andhra Pradesh Children and Parents Study (APS). PLoS One, 10(6):e0129468.

Bansilal, S., Castellano, J.M., Fuster, V. (2015). Global burden of CVD: focus on secondary prevention of cardiovascular disease. Int J Cardiol, 201 Suppl 1:S1-7.

Barnes, M.S., Horigan, G., Cashman, K.D., Hill, T.R., Forsythe, L.K., Lucey, A.J., McSorley, E.M., Kiely, M., Bonham, M.P., Magee, P.J., Strain, J.J., Wallace, J.M. (2011). Maintenance of wintertime vitamin D status with cholecalciferol supplementation is not associated with alterations in serum cytokine concentrations among apparently healthy younger or older adults. J Nutr, 141(3):476-481.

Beckmann, K.R., Kirke, B.A., McCaul, K.A., Roder, D.M. (2001). Use of fake tanning lotions in the South Australian population. Med J Aust, 174(2):75-78.

Belenchia, A.M., Tosh, A.K., Hillman, L.S., Peterson, C.A. (2013). Correcting vitamin D insufficiency improves insulin sensitivity in obese adolescents: a randomized controlled trial. Am J Clin Nutr, 97(4): 774-781.

Bell, N.H., Epstein, S., Greene, A., Shary, J., Oexmann, M.J., Shaw, S. (1985). Evidence for alteration of the vitamin D-endocrine system in obese subjects. J Clin Invest, 76(1): 370-373.

Bell, N.H., Shaw, S., Turner, R.T. (1984). Evidence that 1,25-dihydroxyvitamin D3 inhibits the hepatic production of 25-hydroxyvitamin D in man. J Clin Invest, 74: pp. 1540-1544.

Bellien, J., Thuillez, C., Joannides, R. (2008). Contribution of endothelium-derived hyperpolarizing factors to the regulation of vascular tone in humans. Fundam Clin Pharmacol, 22(4):363-377.

Bellone, S., Esposito, S., Giglione, E., Genoni, G., Fiorito, C., Petri, A., Bona, G., Prodam, F. (2014). Vitamin D levels in a paediatric population of normal weight and obese subjects. J Endocrinol Invest, 37(9):805-809.

351

Bentes, C.M., Costa, P.B., Resende, M., Miranda, H.L., Silva, C.M.V., Netto, C.C., Marinheiro, L.P.F. (2017). Association between muscle function and body composition, vitamin D status, and blood glucose in postmenopausal women with type 2 diabetes. Diabetes Metab Syndr, 11 Suppl 2:S679-S684.

Bergmann, K. (2010). Non-HDL Cholesterol and Evaluation of Cardiovascular Disease Risk, 21(3): 64-67.

Better Health. (2017). Solarium (sunbeds and tanning beds). Retrieved from https://www.betterhealth.vic.gov.au/health/conditionsandtreatments/solariums-sunbeds- and-tanning-beds.

Birken, C.S., Lebovic, G., Anderson, L.N., McCrindle, B.W., Mamdani, M., Kandasamy, S., Khovratovich, M., Parkin, P.C., Maguire, J.L. (2015). Association between Vitamin D and Circulating Lipids in Early Childhood. PLoS One, 10(7):e0131938.

Bischoff-Ferrari, H.A., Dawson-Hughes, B., Staehelin, H.B., Orav, J.E., Stuck, A.E., Theiler, R., Wong, J.B., Egli, A., Kiel, D.P., Henschkowski, J. (2009). Fall prevention with supplemental and active forms of vitamin D: a meta-analysis of randomised controlled trials. BNJ, 339:b3692.

Bischoff-Ferrari, H.A., Willett, W.C., Orav, E.J., Lips, P., Meunier, P.J., Lyons, R.A., Flicker, L., Wark, J., Jackson, R.D., Cauley, J.A., Meyer, H.E., Pfeifer, M., Sanders, K.M., Stahelin, H.B., Theiler, R., Dawson-Hughes, B. (2012). A pooled analysis of vitamin D dose requirements for fracture prevention. N Engl J Med, 367(1):40-49.

Blum, M., Dolnikowski, G., Seyoum, E., Harris, S.S., Booth, S.L., Peterson, J., Saltzman, E., Dawson-Hughes, B. (2008). Vitamin D3 in fat tissue. Endocrine, 33(1): 90–94.

Boland, R. (1986). Role of vitamin D in skeletal muscle function. Endocr Rev, 7: pp. 434- 448.

Borland, R., Hill, D., Noy, S. (1990). Being sun smart: Changes in community awareness and reported behavior following a primary prevention program for skin cancer control. Behav. Change, 7:126-135.

Borland, R., Marks, R., Noy, S. (1992). Public knowledge about characteristics of moles and melanomas. Aust J Public Health, 16(4):370-375.

352

Boulos, M.N.K., Wheeler, S., Tavares, C., Jones, R. (2011). How smartphones are changing the face of mobile and participatory healthcare: an overview, with example from eCAALYX. Biomed Eng Online, 10: 24.

Boutwell, W.B. (1995). The under cover skin cancer prevention project: a community- based program in four Texas cities. Cancer, 75:657-660.

Bränström, R., Kasparian, N.A., Chang, Y.M., Affleck, P., Tibben, A., Aspinwall, L.G., Azizi, E., Baron-Epel, O., Battistuzzi, L., Bergman, W., Bruno, W., Chan, M., Cuellar, F., Debniak, T., Pjanova, D., Ertmanski, S., Figl, A., Gonzalez, M., Hayward, N.K., Hocevar, M., Kanetsky, P.A., Leachman, S.A., Heisele, O., Palmer, J., Peric, B., Puig, S., Schadendorf, D., Gruis, N.A., Newton-Bishop, J., Brandberg, Y. (2010). Predictors of sun protection behaviors and severe sunburn in an international online study. Cancer Epidemiol Biomarkers Prev, 19(9):2199-2210.

Brock, K.E., Ke, L., Tseng, M., Clemson, L., Koo, F.K., Jang, H., Seibel, M.J., Mpofu, E., Fraser, D.R., Mason, R.S. (2013). Vitamin D status is associated with sun exposure, vitamin D and calcium intake, acculturation and attitudes in immigrant East Asian women living in Sydney. J Steroid Biochem Mol Biol, 136:214-217.

Brot, C., Jorgensen, N.R., Sorensen, O.H. (1999). The influence of smoking on vitamin D status and calcium metabolism. Eur J Clin Nutr, 53(12):920-926.

Buller, D.B., Berwick, M., Lantz, K., Buller, M.K., Shane, J., Kane, I., Liu, X. (2015). Evaluation of immediate and 12-week effects of a smartphone sun-safety mobile application: a randomized clinical trial. JAMA Dermatol, 151(5):505-512.

Buller, D.B., Berwick, M., Lantz, K., Buller, M.K., Shane, J., Kane, I., Liu, X. (2015). Smartphone Mobile Application Delivering Personalized, Real-Time Sun Protection Advice: A Randomized Clinical Trial. JAMA Dermatol, 151(5): 497-504.

Buller, D.B., Buller, M.K., Kane, I. (2005). Web-based strategies to disseminate a sun safety curriculum to public elementary schools and state-licensed child-care facilities. Health Psychol, 24(5):470-476.

Buller, D.B., Cokkinides, V., Hall, H.I., Hartman, A.M., Saraiya, M., Miller, E., Paddock, L., Glanz, K. (2011). Prevalence of sunburn, sun protection, and indoor tanning behaviors among Americans: review from national surveys and case studies of 3 states. J Am Acad

353

Dermatol, 65(5 Suppl 1):S114–S123.

Caan, B., Neuhouser, M., Aragaki, A., Lewis, C.B., Jackson, R., LeBoff, M.S., Margolis, K.L., Powell, L., Uwaifo, G., Whitlock, E., Wylie-Rosett, J., LaCroix, A. (2007). Calcium plus vitamin D supplementation and the risk of postmenopausal weight gain, Arch Intern Med,167(9):893-902.

Cafazzo, J., Casselman, M., Hamming, N., Katzman, D.K., Palmert, M.R. (2012). Design of an mHealth app for the self-management of adolescent type 1 diabetes: A pilot study. J Med Internet Res, 14(3):e70.

Callegari, E.T., Reavley, N., Garland, S.M., Gorelik, A., Wark, J.D. (2015).Vitamin D status, bone mineral density and mental health in young Australian women: the Safe-D Study. J Public Health Res, 4:594.

Callister, P., Galtry, J., Didham, R. (2011). The risks and benefits of sun exposure: should skin colour or ethnicity be the main variable for communicating health promotion messages in New Zealand? Ethn Health, 16(1):57-71.

Calton, E.K., Keane, K.N., Newsholme, P., Soares, M.J. (2015). The Impact of Vitamin D Levels on Inflammatory Status: A Systematic Review of Immune Cell Studies. PLoS One, 10(11): e0141770.

Cancer Council Asutralia. (2017). Protect your skin. Retrieved from http://www.cancer.org.au/preventing-cancer/sun-protection/preventing-skin-cancer/.

Cancer Council Australia. (2017). Cancer in Australia. Retrieved from http://www.cancer.org.au/about-cancer/what-is-cancer/facts-and-figures.html.

Cancer Council Victoria. (2017). https://www.cancer.org.au/news/blog/prevention/ vitamin-d-and-sun-protection

Cancer in Australia: Actual incidence data from 1982 to 2013 and mortality data from 1982 to 2014 with projections to 2017. (2017). Australian Institute of Health and Welfare. Asia Pac J Clin Oncol, 20.

Cargill, J., Lucas, R.M., Gies, P., King, K., Swaminathan, A., Allen, M.W., Banks, E. (2013). Validation of brief questionnaire measures of sun exposure and skin pigmentation against detailed and objective measures including vitamin D status. Photochem Photobiol,

354

89(1):219-226.

Caro, Y., Negrón, V., Palacios, C. (2012). Association between Vitamin D Levels and Blood Pressure in a Group of Puerto Ricans. Health Sci J, 31(3):123-129.

Carruth, B.R., Skinner, J.D. (2001). The role of dietary calcium and other nutrients in moderating body fat in preschool children. International Journal of Obesity, 25: 559-566.

Cashman, KD. (2007). Vitamin D in childhood and adolescence. Postgrad Med J, 83(978): 230-235.

Castellon, X., Bogdanova, V. (2016). Chronic Inflammatory Diseases and Endothelial Dysfunction. Aging Dis, 7(1): 81-89.

CDC Centers for Disease Control and Prevention, National Center for Health Statistics. National health and nutrition examination survey data sets and related documentation. (2009). Retrieved from http:// www.cdc.gov/nchs/nhanes/nhanes_questionnaires.htm

Ceglia, L. (2009). Vitamin D and Its Role in Skeletal Muscle. Curr Opin Clin Nutr Metab Care, 12(6): 628-633.

Chan, A., Kow, R., Cheng, J.K. (2017). Adolescents’ Perceptions on Smartphone Applications (Apps) for Health Management. Journal MTM 6:2:47-55.

Chandler, P.D., Wang, L., Zhang, X., Sesso, H.D., Moorthy, M.V., Obi, O., Lewis, J., Prince, R.L., Danik, J.S., Manson, J.E., LeBoff, M.S., Song, Y. (2015). Effect of vitamin D supplementation alone or with calcium on adiposity measures: a systematic review and meta-analysis of randomized controlled trials. Nutr Rev,73(9):577-593.

Chang, E., Kim, Y. (2017). Vitamin D Insufficiency Exacerbates Adipose Tissue Macrophage Infiltration and Decreases AMPK/SIRT1 Activity in Obese Rats. Nutrients, 9(4): 338.

Chaudhuri, J.R., Mridula, K.R., Anamika, A., Boddu, D.B., Misra, P.K., Lingaiah, A., Balaraju, B., Bandaru, V.S. (2013). Deficiency of 25-hydroxyvitamin d and dyslipidemia in Indian subjects. J Lipids, 2013:623420.

Chen, S., Law, C.S., Grigsby, C.L., Olsen, K., Gardner, D.G. (2010). A role for the cell cycle phosphatase Cdc25a in vitamin D-dependent inhibition of adult rat vascular smooth

355 muscle cell proliferation. J Steroid Biochem Mol Biol, 122(5):326-332.

Cheng, S., Massaro, J.M., Fox, C.S. (2010). Adiposity, cardiometabolic risk, and vitamin D status: the framingham heart study. Diabetes, 59(1): 242-248.

Christesen, H.T., Falkenberg, T., Lamont, R.F., Jørgensen, J.S. (2012). The impact of vitamin D on pregnancy: a systematic review. Acta Obstet Gynecol Scand, 91(12):1357- 1367.

Christie, J.J., Nowson, C.A., Garland, S.M., Wark, J.D. (2013). Emerging nutritional and lifestyle risk factors for bone health in young women: a mixed longitudinal twin study. Nutritional influences on bone health, Springer, London, England, 61-69.

Christoph, S., Cazzaniga, S., Hunger, R.E., Naldi, L., Borradori, L., Oberholzer, P.A. (2016). eCollection. Ultraviolet radiation protection and skin cancer awareness in recreational athletes: a survey among participants in a running event. Swiss Med Wkly, 21;146:14297.

Cipriani, C., Pepe, J., Piemonte, S., Colangelo, L., Cilli, M., Minisola, S. (2014). Vitamin D and Its Relationship with Obesity and Muscle. Int J Endocrinol, 2014: 841248.

Cockburn, J., Hennrikus, D., Scott, R., Sanson-Fisher, R. (1989). Adolescent use of sun- protection measures. Med J Aust, 7;151(3):136-140.

Cokkinides, V.E., Bandi, P., Weinstock, M.A., Ward, E. (2010). Use of sunless tanning products among US adolescents aged 11 to 18 years. Arch Dermatol,146(9):987-992.

Cole T.J., Bellizzi M.C., Flegal K.M., Dietz W.H. (2000). Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ, 320:1240-1243.

Compher, C.W, Badellino, K.O., Boullata, J.I. (2008). Vitamin D and the bariatric surgical patient: a review. Obes Surg, 18:220-224.

Coppack, S.W. (2001). Pro-inflammatory cytokines and adipose tissue. Proc Nutr Soc, 60(3):349-356.

Cranney, A., Horsley, T., O'Donnell, S., Weiler, H., Puil, L., Ooi, D., Atkinson, S., Ward, L., Moher, D., Hanley, D., Fang, M., Yazdi, F., Garritty, C., Sampson, M., Barrowman,

356

N., Tsertsvadze, A., Mamaladze, V. (2007). Effectiveness and safety of vitamin D in relation to bone health. Evid Rep Technol Assess (Full Rep), 158:1-235.

Dalgård, C., Petersen, M.S., Weihe, P., Grandjean, P. (2011). Vitamin D status in relation to glucose metabolism and type 2 diabetes in septuagenarians. Diabetes Care, 34(6):1284- 1288.

Daly, R.M., Gagnon, C., Lu, Z.X., Magliano, D.J., Dunstan, D.W., Sikaris, K.A., Zimmet, P.Z., Ebeling, P.R., Shaw, J.E. (2012). Prevalence of vitamin D deficiency and its determinants in Australian adults aged 25 years and older: a national, population-based study. Clin Endocrinol (Oxf), 77(1):26-35.

Damilakis, J., Adams, J.E., Guglielmi, G., Link, T.M. (2010). Radiation exposure in X- ray-based imaging techniques used in osteoporosis. Eur Radiol, 20(11): 2707-2714.

Daniel, C.L., Armstrong, G.T., Keske, R.R., Davine, J.A., McDonald, A.J., Sprunck- Harrild, K.M., Coleman, C., Haneuse, S.J., Mertens, A.C., Emmons, K.M., Marghoob, A.A., Elkin, E.B., Dusza, S.W., Robison, L.L., Geller, A.C. (2015). Advancing Survivors' Knowledge (ASK) about skin cancer study: study protocol for a randomized controlled trial. Trials, 24;16:109.

Davis, W.A., Peters, K.E., Makepeace, A., Griffiths, S., Bundell, C., Grant, S.F.A., Ellard S., Hattersley, A.T., Paul Chubb, S.A., Bruce, D.G., Davis ,T.M.E. (2018). Prevalence of diabetes in Australia: insights from the Fremantle Diabetes Study Phase II. Intern Med J, 48(7):803-809.

Davis, J.E., McDonald, J.M., Jarett, L. (1978). A high-performance liquid chromatography method for hemoglobin A1c. Diabetes, 27(2):102-107.

Day, A.K., Wilson, C., Roberts, R.M., Hutchinson, A.D. (2014). The Skin Cancer and Sun Knowledge (SCSK) Scale: Validity, Reliability, and Relationship to Sun-Related Behaviors Among Young Western Adults. Health Educ Behav, 41(4):440-448.

De Boer, I.H., Tinker, L.F., Connelly, S., Curb, J.D., Howard, B.V., Kestenbaum, B., Larson, J.C., Manson, J.E., Margolis, K.L., Siscovick, D.S., Weiss, N.S. (2008). Women's Health Initiative Investigators. Calcium plus vitamin D supplementation and the risk of incident diabetes in the Women's Health Initiative. Diabetes Care, 31(4):701-707.

357

Definition and diagnosis of diabetes mellitus and intermediate hyperglycaemia (2006). Retrieved from http://www.who.int/diabetes/publications/diagnosis_diabetes2006/en/

Dennis, L.K., Lowe, J.B., Snetselaar, L.G. (2009). Tanning behavior among young frequent tanners is related to attitudes and not lack of knowledge about the dangers. Health Educ J, 68 (3):232-243.

Despre´s, J. (2012). Body Fat Distribution and Risk of Cardiovascular Disease An Update. Circulation, 126:1301-1313.

Dietrich, A.J., Olson, A.L., Sox, C.H., Stevens, M., Tosteson, T.D., Ahles, T., Winchell, C.W., Grant-Petersson, J., Collison, D.W., Sanson-Fisher, R. (1998). A community-based randomized trial encouraging sun protection for children. Pediatrics,102(6):E64.

Dix, C.F., Bauer, J.D., Martin, I., Rochester, S., Duarte Romero, B., Prins, J.B., Wright, O.R.L. (2017). Association of Sun Exposure, Skin Colour and Body Mass Index with Vitamin D Status in Individuals Who Are Morbidly Obese. Nutrients, 4;9(10).

Dobbinson, S.J., Wakefield, M.A., Jamsen, K.M., Herd, N.L., Spittal, M.J., Lipscomb , J.E., Hill, D.J. (2008). Weekend sun protection and sunburn in Australia trends (1987 – 2002) and association with SunSmart television advertising. Am J Prev Med, 34:94-101.

Dong, Y., Stallmann-Jorgensen, I.S., Pollock, N.K., Harris, R.A., Keeton, D., Huang, Y., Li, K., Bassali, R., Guo, D.H., Thomas, J., Pierce, G.L., White, J., Holick, M.F., Zhu, H. (2010). A 16-week randomized clinical trial of 2000 international units daily vitamin D3 supplementation in black youth: 25-hydroxyvitamin D, adiposity, and arterial stiffness. J Clin Endocrinol Metab, 95(10):4584-4591.

Dougkas, A., Reynolds, C.K., Givens, I.D., Elwood, P.C., Minihane, A.M. (2011). Associations between dairy consumption and body weight: a review of the evidence and underlying mechanisms. Nutr Res Rev, 24(1):72-95.

Duclos, V., Yé, M., Moubassira, K., Sanou, H., Sawadogo, N.H., Bibeau, G., Sié, A. (2017). Situating mobile health: a qualitative study of mHealth expectations in the rural health district of Nouna, Burkina Faso. Health Res Policy Syst, 12;15(Suppl 1):47.

Dyląg, H., Rowicka, G., Strucińska, M., Riahi, A. (2014). Assessment of vitamin D status in children aged 1-5 with simple obesity. Rocz Panstw Zakl Hig, 65(4):325-330.

358

Elkassaby, S., Harrison, L.C., Mazzitelli, N., Wentworth, J.M., Colman, P.G., Spelman, T., Fourlanos, S. (2014). A randomised controlled trial of high dose vitamin D in recent- onset type 2 diabetes. Diabetes Res Clin Pract, 106(3):576-582.

Fenner, Y., Garland, S.M., Moore, E.E., Jayasinghe, Y., Fletcher, A., Tabrizi, S.N., Gunasekaran, B., Wark, J.D. (2012). Web-based recruiting for health research using a social networking site: an exploratory study. J Med Internet Res, 14:e20.

Fernandes, M.R., Barreto, W.D.R. (2017). Association between physical activity and vitamin D: A narrative literature review. Rev Assoc Med Bras, 63(6):550-556.

Fernández-Morano, T., De Troya-Martín, M., Rivas-Ruiz, F., Blázquez-Sánchez, N., Del Boz-González, J., Fernández-Peñas, P., Buendía-Eisman, A. (2014). Behaviour, attitudes and awareness concerning sun exposure in adolescents on the Costa del Sol. Eur J Dermatol, 24(1):85-93.

Fernández-Morano, T., Troya-Martín, M., Rivas-Ruiz, F., Blázquez-Sánchez, N., Buendía-Eisman, A. (2015). Sensitivity to change of the Beach Questionnaire to behaviour, attitudes and knowledge related to sun exposure: quasi-experimental before- after study. BMC Public Health, 15: 60.

Filiz, T.M., Cinar, N., Topsever, P., Ucar, F. (2006). Tanning youth: knowledge, behaviors and attitudes toward sun protection of high school students in Sakarya, Turkey. J Adolesc Health, 38(4):469-471.

Fitzpatrick, T.B. (1988). The validity and practicality of sun reactive skin types I through VI. Arch Dermatol, 124: 869-871.

Flood-Nichols, S.K, Tinnemore, D., Huang, R.R., Napolitano, P.G., Ippolito, D.L. (2015). Vitamin D Deficiency in Early Pregnancy. PLOS One, 10(4): e0123763.

Food and drug administration. What is a Serious Adverse Event? Retrived from https://www.fda.gov/Safety/MedWatch/HowToReport/ucm053087.htm

Ford, J.A., MacLennan, G.S., Avenell, A., Bolland, M., Grey, A., Witham, M., for the RECORD Trial Group. (2014). Cardiovascular disease and vitamin D supplementation: trial analysis, systematic review, and meta-analysis. Am J Clin Nutr, 100:746-755.

Forman, J.P., Curhan, G.C., Taylor, E.N. (2008). Plasma 25-hydroxyvitamin D levels and

359 risk of incident hypertension among young women. Hypertension, 52(5):828-832.

Forman, J.P., Scott, J.B, Ng, K., Drake, B.F., Suarez, E.G., Hayden, D.L., Bennett, G.G., Chandler, P.D., Hollis, B.W., Emmons, K.M., Giovannucci, E.L., Fuchs, C.S., Chan, A.T. (2013). Effect of vitamin D supplementation on blood pressure in blacks. Hypertension, 61(4):779-785.

Forsythe, L.K., Livingstone, M.B., Barnes, M.S., Horigan, G., McSorley, E.M., Bonham, M.P., Magee, P.J., Hill, T.R., Lucey, A.J., Cashman, K.D., Kiely, M., Strain, J.J., Wallace, J.M. (2012). Effect of adiposity on vitamin D status and the 25- hydroxycholecalciferol response to supplementation in healthy young and older Irish adults. Br J Nutr, 107(1):126-134.

Francis, K., Dobbinson, S., Wakefield, M., Girgis, A. (2010). Solarium use in Australia, recent trends and context. Aust N Z J Public, 34(4):427-430.

Free, C., Phillips, G., Felix, L., Galli, L., Patel, V., Edwards, P. (2010). The effectiveness of M-health technologies for improving health and health services: a systematic review protocol. BMC Research Notes, 3:250.

Freeman, J., Wilson, K., Spears, R., Shalhoub, V., Sibley, P. (2015). Performance evaluation of four 25-hydroxyvitamin D assays to measure 25-hydroxyvitamin D 2. Clinical Biochemistry, 48 (16-17): 1097-1104.

Gagnon, C., Daly, R.M., Carpentier, A., Lu, Z.X., Shore-Lorenti, C., Sikaris, K., Jean, S., Ebeling, P.R. (2014). Effects of combined calcium and vitamin D supplementation on insulin secretion, insulin sensitivity and β-cell function in multi-ethnic vitamin D- deficient adults at risk for type 2 diabetes: a pilot randomized, placebo-controlled trial. PLoS One, 9(10):e109607.

Gallagher, E.J., Le Roith, D., Bloomgarden, Z. (2009). Review of haemoglobin A1c and the management of diabetes. J Diabetes, 1:9-17.

Gill, T.K., Hill, C.L., Shanahan, E.M., Taylor, A.W., Appleton, S.L., Grant, J.F., Shi, Z., Grande, E.D., Price, K., Adams, R.J. (2014). Vitamin D levels in an Australian population. BMC Public Health,14:1001

Ginde, A.A., Liu, M.C., Camargo, C.A. (2009). Demographic Differences and Trends of

360

Vitamin D Insufficiency in the US Population. ArchInternMed, 169(6):626-632.

Ginde, A.A., Sullivan, A.F., Mansbach, J.M., Camargo, C.A. (2010). Vitamin D insufficiency in pregnant and nonpregnant women of childbearing age in the United States. Am J Obstet Gynecol, 202(5):436.e1-8.

Ginsberg, H.N. (2000). Insulin resistance and cardiovascular disease. J Clin Invest, 106(4): 453-458.

Giovannucci, E., Liu, Y., Hollis, B.W., Rimm, E.B. (2008). 25-hydroxyvitamin D and risk of myocardial infarction in men: a prospective study. Arch Intern Med, 9;168(11):1174-1180.

Girgis, A., Sanson-Fisher, R.W., Tripodi, D.A., Golding, T. (1993). Evaluation of intentions to improve solar protection in primary schools. Health Educ, 20:275-287.

Girgis, A., Sanson-Fisher, R.W., Watson, A. (1994). A workplace intervention for increasing outdoor workers' use of solar protection. Am J Public Health, 84(1):77-81.

Glanz, K., Steffen, A.D., Schoenfeld, E., Tappe, K.A. (2013). Randomized trial of tailored skin cancer prevention for children: the Project SCAPE family study. J Health Commun, 18(11):1368-1383.

Glanz, K., Yaroch, A.L., Dancel, M., Saraiya, M., Crane, L.A., Buller, D.B., Manne, S., O'Riordan, D.L., Heckman, C.J., Hay, J., Robinson, J.K. (2008). Measures of sun exposure and sun protection practices for behavioral and epidemiologic research. Archives of Dermatology, 144:217.

Goldberg, T.B., da Silva, C.C., Peres, L.N., Berbel, M.N., Heigasi, M.B., Ribeiro, J.M., Suzuki, K., Josué, L.M., Dalmas, J.C. (2009) Calcium intake and its relationship with risk of overweight and obesity in adolescents. Arch Latinoam Nutr 59, 14-21.

Gonzalez, F.J., Moschetta, A. (2014). Potential role of the vitamin D receptor in control of cholesterol levels. Gastroenterology, 2(14):233.

Goodman, S., Morrongiello, B., Meckling, K. (2016). A randomized, controlled trial evaluating the efficacy of an online intervention targeting vitamin D intake, knowledge and status among young adults. Int J Behav Nutr Phys Act, 13(1):116.

361

Gordon, L.G., Hirst, N.G., Green, A.C., Neale, R.E. (2012). Tanning behaviors and determinants of solarium use among indoor office workers in Queensland, Australia. J Health Psychol, 17(6):856-865.

Gouni-Berthold, I., Krone, W. Berthold, H.K. (2009) Vitamin D and cardiovascular disease. Curr Vasc Pharmacol, 7(3):414-422.

Green, A.C., Wallingford, S.C., McBride, P. (2011). Childhood exposure to ultraviolet radiation and harmful skin effects: epidemiological evidence. Prog Biophys Mol Biol, 107(3):349-55.

Grimnes, G., Figenschau, Y., Almås, B., Jorde, R. (2011). Vitamin D, Insulin Secretion, Sensitivity, and Lipids. Diabetes, 60(11): 2748-2757.

Haarbo, J., Gotfredsen, A., Hassager, A., Christiansen, C. (1991). Validation of body composition by dual energy X-ray absorptiometry (DEXA). Clinical Physiology, 11:331- 341.

Hall, J.E. (1991). Control of blood pressure by the renin-angiotensin-aldosterone system. Clin Cardiol, 14(8 Suppl 4): IV6-21; IV51-5.

Han, S.S., Kim, M., Lee, S.M., Lee, J.P., Kim, S., Joo, K.W., Lim, C.S., Kim, Y.S., Kim, D.K. (2014). Association between body fat and vitamin D status in Korean adults. Asia Pac J Clin Nutr, 23(1):65-75.

Hansson, L., Hedner, T., Dahlöf, B. (1992). Prospective randomized open blinded endpoint (PROBE) study. A novel design for intervention trials. Prospective Randomized Open Blinded End-Point. Blood Press, 1(2):113-119.

Hao, Y., Ma, X., Shen, Y., Ni, J., Luo, Y., Xiao, Y., Bao, Y., Jia, W. (2014). Associations of Serum 25-Hydroxyvitamin D3 Levels with Visceral Adipose Tissue in Chinese Men with Normal Glucose Tolerance. PLoS One, 22;9(1):e86773.

Hardy, L.L., Jin, K., Mihrshahi, S., Ding, D. (2018). Trends in overweight, obesity, and waist-to-height ratio among Australian children from linguistically diverse backgrounds, 1997 to 2015. Int J Obes (Lond). doi: 10.1038/s41366-018-0139-5.

Harris, S.S., Dawson-Hughes, B. (1998). Seasonal changes in plasma 25-hydroxyvitamin D concentrations of young American black and white women. Am J Clin Nutr, 67:1232–

362

1236.

Harris, R.A., Pedersen-White, J., Guo, D.H., Stallmann-Jorgensen, I.S., Keeton, D., Huang, Y., Shah, Y., Zhu, H., Dong, Y. (2011). Vitamin D3 supplementation for 16 weeks improves flow-mediated dilation in overweight African-American adults. Am J Hypertens, 24(5):557-562.

Head, K.J., Noar, S.M., Iannarino, N.T., Grant Harrington, N. (2013). Efficacy of text messaging-based interventions for health promotion: a meta-analysis. Soc Sci Med, 97:41-48.

Heaney, R.P. (2008). Vitamin D and calcium interactions: functional outcomes. Am J Clin Nutr, 88(2):541S-544S.

Heffernan, K., Chang, S., Maclean, S., Callegari, E., Garland, S., Reavley, N., Varigos, G., Wark, J.D. (2014). The potential of eHealth apps to support targeted complex health messages. J Gen Practice, 2014:182.

Heffernan, K.J., Chang, S., Maclean, S.T., Callegari, E.T., Garland, S.M., Reavley, N.J., Varigos, G.A., Wark, J.D. (2016). Guidelines and Recommendations for Developing Interactive eHealth Apps for Complex Messaging in Health Promotion. JMIR Mhealth Uhealth, 4(1):e14.

Heikkinen, A.M., Tuppurainen, M.T., Niskanen, L., Komulainen, M., Penttilä, I., Saarikoski, S. (1997). Long-term vitamin D3 supplementation may have adverse effects on serum lipids during postmenopausal hormone replacement therapy. Eur J Endocrinol, 137(5):495-502.

Holick, M.F. (2008). Vitamin D and Sunlight: Strategies for Cancer Prevention and Other Health Benefits. Clin J Am Soc Nephrol, 3(5): 1548-1554.

Holick, M.F., Binkley, N.C., Bischoff-Ferrari, H.A., Gordon, C.M., Hanley, D.A., Heaney, R.P., Murad, M.H., Weaver, C.M. (2011). Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab, 96(7):1911-1930.

Holick, M.F. (2007). Vitamin D deficiency. N Engl J Med, 357:266-281

Holick, M.F., Binkley, N.C., Bischoff-Ferrari, H.A., Heike, A., Gordon, C.M., Hanley,

363

D.A., Heaney, R.P., Murad, M.H., Weaver, C.M. (2011). Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab, 96(7):1911-1930.

Hong, Y.M. (2010). Atherosclerotic Cardiovascular Disease Beginning in Childhood. Korean Circ J, 40(1): 1-9.

How much sun is enough?. (2017). Retrieved from http://www.sunsmart.com.au/downloads/vitamin-d/how-much-sun-is-enough-vitamin- d.pdf

Hruby, A., Hu, F.B. (2015). The Epidemiology of Obesity: A Big Picture. Pharmacoeconomics, 33(7): 673-689.

Hsia, J., Heiss, G., Ren, H., Allison, M., Dolan, N.C., Greenland, P., Heckbert, S.R., Johnson, K.C., Manson, J.E., Sidney, S., Trevisan, M., for the Women’s Health Initiative Investigators. (2007). Calcium/vitamin D supplementation and cardiovascular events. Circulation, 115:846-854.

Hu, G., Bouchard, C., Bray, G.A., Greenway, F.L., Johnson, W.D., Newton, R.L., Ravussin, E., Ryan, D.H., Katzmarzyk, P.T. (2011). Trunk Versus Extremity Adiposity and Cardiometabolic Risk Factors in White and African American Adults. Diabetes Care, 34(6): 1415-1418.

Huse, O., Hettiarachchi, J., Gearon, E., Nichols, M., Allender, S., Peeters, A. (2018). Obesity in Australia. Obes Res Clin Pract, 12(1):29-39.

Hutchinson, K., Healy, M., Crowley, V., Louw, M., Rochev, Y. (2017). Verification of Abbott 25-OH-vitamin D assay on the architect system. Pract Lab Med,17(7):27-35.

Hyppönen, E., Boucher, B.J., Berry, D.J., Power, C. (2008). 25-hydroxyvitamin D, IGF- 1, and metabolic syndrome at 45 years of age: a cross-sectional study in the 1958 British Birth Cohort. Diabetes, 57(2):298-305.

Hyppönen, E., Power, C. (2006). Vitamin D status and glucose homeostasis in the 1958 British birth cohort: the role of obesity. Diabetes Care, 29(10):2244-2246.

IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp

364

Ilinčić, B., Stokić, E., Stošić, Z., Kojić, NE., Katsiki, N., Mikhailidis, D.P., Isenovic, E.R. (2017). Vitamin D status and circulating biomarkers of endothelial dysfunction and inflammation in non-diabetic obese individuals: a pilot study. Arch Med Sci,13(1):53-60.

Ireland, P., Jolley, D., Giles, G., O'Dea, K., Powles, J., Rutishauser, I., Wahlqvist, M.L., Williams, J. (1994). Development of the Melbourne FFQ: a food frequency questionnaire for use in an Australian prospective study involving an ethnically diverse cohort. Asia Pac J Clin Nutr, 3:19-31.

Jones, G., Dwyer, T., Hynes, K.L., Parameswaran, V., Greenaway, T.M. (2005). Vitamin D insufficiency in adolescent males in Southern Tasmania: prevalence, determinants, and relationship to bone turnover markers. Osteoporos International, 16: 636-641.

Jorde, R., Figenschau, Y. (2009). Supplementation with cholecalciferol does not improve glycaemic control in diabetic subjects with normal serum 25-hydroxyvitamin D levels. Eur J Nutr, 48(6):349-354.

Jorde, R., Figenschau, Y., Emaus, N., Hutchinson, M., Grimnes, G. (2010). Serum 25- hydroxyvitamin D levels are strongly related to systolic blood pressure but do not predict future hypertension. Hypertension, 55(3):792-798.

Jorde, R., Figenschau, Y., Hutchinson, M., Emaus, N., Grimnes, G. (2010). High serum 25-hydroxyvitamin D concentrations are associated with a favorable serum lipid profile. Eur J Clin Nutr, 64(12):1457-1464.

Jorde, R., Sollid, S.T., Svartberg, J., Schirmer, H., Joakimsen, R.M., Njølstad, I., Fuskevåg, O.M., Figenschau, Y., Hutchinson, M.Y. (2016). Vitamin D 20,000 IU per Week for Five Years Does Not Prevent Progression From Prediabetes to Diabetes. J Clin Endocrinol Metab, 101(4):1647-1655.

Jorm, A.F., Allen, N., Morgan, A., Purcell, R. (2009). beyondblue: Melbourne, August. 2009. A Guide to What Works for Depression. Retrieved from https://www.beyondblue.org.au/docs/default-source/resources/bl-0556-guide-to-what- works-for-depression-12-14.pdf?sfvrsn=2 webcite.

Judd, S.E., Nanes, M.S., Ziegler, T.R., Wilson, P.W., Tangpricha, V. (2008). Optimal vitamin D status attenuates the age-associated increase in systolic blood pressure in white Americans: results from the third National Health and Nutrition Examination Survey. Am

365

J Clin Nutr, 87(1):136-141.

Kamel Boulos, M.N., Brewer, A.C., Karimkhani, C., Buller, D.B.,Dellavalle, R.P. (2014). Mobile medical and health apps: state of the art, concerns, regulatory control and certification. Online J Public Health Inform, 5(3): 229.

Kane, L., Moore, K., Lütjohann, D., Bikle, D., Schwartz, J.B. (2013). Vitamin D3 Effects on Lipids Differ in Statin and Non-Statin-Treated Humans: Superiority of Free 25-OH D Levels in Detecting Relationships. J Clin Endocrinol Metab, 98(11): 4400-4409.

Kannel, W.B., McGee, D.L. (1979). Diabetes and Cardiovascular Disease The Framingham Study. JAMA, 241(19):2035-2038.

Kao, K.T., Abidi, N., Ranasinha, S., Brown, J., Rodda, C., McCallum, Z., Zacharin, M., Simm, P.J., Magnussen, C.G., Sabin, M.A. (2015). Low vitamin D is associated with hypertension in paediatric obesity. J Paediatr Child Health, 51(12):1207-1213.

Karhapää, P., Pihlajamäki, J., Pörsti, I., Kastarinen, M., Mustonen, J., Niemelä, O., Kuusisto, J. (2010). Diverse associations of 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D with dyslipidaemias. J Intern Med, 268(6):604-610.

Katz, A., Nambi, S.S., Mather, K., Baron, A.D., Follmann, D.A., Sullivan, G., Quon, M.J. (2000). Quantitative insulin sensitivity check index: a simple, accurate method for assessing insulin sensitivity in humans. J Clin Endocrinol Metab, 85(7):2402-2410.

Kayaniyil, S., Vieth, R., Retnakaran, R., Knight, J.A., Qi, Y., Gerstein, H.C., Perkins, B.A., Harris, S.B., Zinman, B., Hanley, A.J. (2010). Association of vitamin D with insulin resistance and beta-cell dysfunction in subjects at risk for type 2 diabetes. Diabetes Care, 33(6):1379-1381.

Keast, D.R., Hill Gallant, K.M., Albertson, A.M, Gugger, C.K., Holschuh, N.M. (2015). Associations between Yogurt, Dairy, Calcium, and Vitamin D Intake and Obesity among U.S. Children Aged 8–18 Years: NHANES, 2005–2008. Nutrients, 7(3), 1577-1593.

Kelishadi, R., Farajzadegan, Z., Bahreynian, M. (2014). Association between vitamin D status and lipid profile in children and adolescents: a systematic review and meta- analysis. Int J Food Sci Nutr, 65(4):404-410.

Kelly, A., Brooks, L.J., Dougherty, S., Carlow, D.C., Zemel, B.S. (2011). A cross-

366 sectional study of vitamin D and insulin resistance in children. Arch Dis Child, 96(5):447-452.

Kennedy, C., Bajdik, C.D., Willemze, R., De Gruijl, F.R., Bouwes Bavinck, J.N., Leiden Skin Cancer Study. (2003). The influence of painful sunburns and lifetime sun exposure on the risk of actinic keratoses, seborrheic warts, melanocytic nevi, atypical nevi, and skin cancer. J Invest Dermatol, 120(6):1087-1093.

Kharlamov, A.N., Perrish, A.N., Gabiskiĭ, Ia.L., Ronne, K.h., Ivanova, E.I. (2012). Vitamin D in the treatment of cardiorenal syndrome in patients with chronic nephropathy. Kardiologiia, 52(3):33-44.

Khundmiri, S.J., Murray, R.D., Lederer, E. (2016). PTH and Vitamin D. Compr Physiol, 15;6(2):561-601.

Kim, B.H., Glanz, K., Nehl, E.J. (2012). Vitamin D beliefs and associations with sunburns, sun exposure, and sun protection. Int J Environ Res Public Health, 9(7):2386- 2395.

Kim, K.J., Kim, Y.J., Kim, S.H., An, J.H., Yoo, H.J., Kim, H.Y., Seo, J.A., Kim, S.G., Kim, N.H., Choi, K.M., Baik, S.H., Choi, D.S., Kim, N.H. (2015). Vitamin D status and associated metabolic risk factors among North Korean refugees in South Korea: a cross- sectional study. BMJ Open, 5(11): e009140.

Kimlin, M., Sun, J., Sinclair, C., Heward, S., Hill, J., Dunstone, K., Brodie, A. (2016). Are the current Australian sun exposure guidelines effective in maintaining adequate levels of 25-hydroxyvitamin D? J Steroid Biochem Mol Biol, 155(Pt B):264-270.

Klasnja, P., Consolvo, S., McDonald, D.W., Landay, J.A., Pratt, W. (2009). Using Mobile & Personal Sensing Technologies to Support Health Behavior Change in Everyday Life: Lessons Learned. AMIA Annu Symp Proc, 2009: 338-342.

Knekt, P., Laaksonen, M., Mattila, C., Härkänen, T., Marniemi, J., Heliövaara, M., Rissanen. H., Montonen. J., Reunanen, A. (2008). Serum vitamin D and subsequent occurrence of type 2 diabetes. Epidemiology,19(5):666-671.

Kong, J., Li, Y,C. (2006). Molecular mechanism of 1,25-dihydroxyvitamin D3 inhibition of adipogenesis in 3T3-L1 cells. Am J Physiol Endocrinol Metab, 290: E916-E924.

367

Kongsbak, M., Levring, T.B., Geisler, C., von Essen, M.R. (2013). The vitamin d receptor and T cell function. Front Immunol, 18;4:148.

Køster, B., Søndergaard, J., Nielsen, J.B., Allen, M., Olsen, A., Bentzen, J. (2017). The validated sun exposure questionnaire: association of objective and subjective measures of sun exposure in a Danish population-based sample. Br J Dermatol, 176(2):446-456.

Køster, B., Thorgaard, C., Philip, A., Clemmensen, I.H. (2010). Prevalence of sunburn and sun-related behaviour in the Danish population: a cross-sectional study. Scand J Public Health, 38 (5):548-552.

Kremer, R., Campbell, P.P., Reinhardt, T., Gilsanz, V. (2009). Vitamin D status and its relationship to body fat, final height, and peak bone mass in young women. J Clin Endocrinol Metab, 94(1):67-73.

Kyle, R.G., MacMillan, I., Forbat, L., Neal, R.D., O'Carroll, R.E., Haw, S., Hubbard, G. (2014). Scottish adolescents’ sun-related behaviours, tanning attitudes and associations with skin cancer awareness: a cross-sectional study. BMJ Open, 4(5): e005137.

Kyle, R.G., Macmillan, I., Rauchhaus, P., O'Carroll, R., Neal, R.D., Forbat, L., Haw, S., Hubbard, G. (2013). Adolescent Cancer Education (ACE) to increase adolescent and parent cancer awareness and communication: study protocol for a cluster randomised controlled trial. Trials, 14:286.

Lai, V., Cranwell, W., Sinclair, R. (2018). Epidemiology of skin cancer in the mature patient. Clin Dermatol, 36(2):167-176.

Lawler, S., Spathonis, K., Eakin E., Gallois, C., Leslie, E., Owen, N. (2007). Sun exposure and sun protection behaviours among young adult sport competitors. Aust N Z J Public Health, 31(3):230-234.

Lazovich, D., Vogel, R.I., Berwick, M., Weinstock, M.A., Warshaw, E.M., Anderson, K.E. (2011). Melanoma Risk in Relation to Use of Sunscreen or Other Sun Protection Methods. Cancer Epidemiol Biomarkers Prev, 20 (12): 2583-2593.

Lebovitz, H.E. (2001). Insulin resistance: definition and consequences. Exp Clin Endocrinol Diabetes, 109 Suppl 2:S135-148.

Lee, W.P., Lin, L.W., Yeh, S.H., Liu, R.H., Tseng, C.F. (2002). Correlations among

368 serum calcium, vitamin D and parathyroid hormone levels in the elderly in southern. Taiwan. J Nurs Res, 10(1):65-72.

Lehr, S., Hartwig, S., Lamers, D., Famulla, S., Müller, S., Hanisch, F.G., Cuvelier, C., Ruige, J., Eckardt, K., Ouwens, D.M., Sell, H., Eckel, J. (2012). Identification and validation of novel adipokines released from primary human adipocytes. Mol Cell Proteomics, 11(1):M111.010504.

Lenders, C.M., Feldman, H.A., Von Scheven, E., Merewood, A., Sweeney, C., Wilson, D.M., Lee, P.D., Abrams, S.H., Gitelman, S.E., Wertz, M.S., Klish, W.J., Taylor, G.A., Chen, T.C., Holick, M.F., Elizabeth Glaser Pediatric Research Network Obesity Study Group. (2009). Relation of body fat indexes to vitamin D status and deficiency among obese adolescents. Am J Clin Nutr, 90(3): 459-467.

Leon, B.M., Maddox, T.M. (2015). Diabetes and cardiovascular disease: Epidemiology, biological mechanisms, treatment recommendations and future research. World J Diabetes, 10; 6(13): 1246-1258.

Leonard, A., Hutchesson, M., Patterson, A., Chalmers, K., Collins, C. (2014). Trials. Recruitment and retention of young women into nutrition research studies: practical considerations,16:15-23.

Levy, R., Malech, H.L. (1991). Effect of 1,25-dihydroxyvitamin D3, lipopolysaccharide, or lipoteichoic acid on the expression of NADPH oxidase components in cultured human monocytes. J Immunol, 147: 3066-3071.

Liberato, S.C., Maple-Brown, L., Bressan, J., Hills, A.P. (2013). The relationships between body composition and cardiovascular risk factors in young Australian men. Nutr J, 12: 108.

LimeSurvey project team. LimeSurvey: An open Source Survey Tool. Retrieved from https://www.limesurvey.org/

Linos, E., Keiser, E., Kanzler, M., Sainani, K.L., Lee, W., Vittinghoff, E., Chren, M.M., Tang, J.Y. (2012). Sun protective behaviors and vitamin D levels in the US population: NHANES 2003–2006. Cancer Causes Control, 23(1): 133-140.

Lips, P. (2001). Vitamin D deficiency and secondary hyperparathyroidism in the elderly:

369 consequences for bone loss and fractures and therapeutic implications. Endocr Rev, 22(4):477-501.

Lips, P. (2012). Interaction between vitamin D and calcium. Scand J Clin Lab Invest, 243:60-64.

Living Works Australia. (2015). Retrieved from http://www.livingworks.com.au/programs/asist/ webcite.

Livingston, P.M., White, V.M., Hayman, J., Dobbinson, S. (2007). Australianadolescents’ sun protection behavior: Who are we kidding? Preventive Medicine, 44, 508-512.

Ljunghall, S., Lind, L., Lithell, H., Skarfors, E., Selinus, I., Sørensen, O.H., Wide L. (1987). Treatment with one-alpha-hydroxycholecalciferol in middle-aged men with impaired glucose tolerance--a prospective randomized double-blind study. Acta Med Scand, 222(4):361-367.

Lu, H.K., Zhang, Z., Ke, Y.H., He, J.W., Fu, W.Z., Zhang, C.Q, Zhang, Z.L. (2012). High prevalence of vitamin D insufficiency in China: relationship with the levels of parathyroid hormone and markers of bone turnover. PLoS One, 7:e47264.

Mackay, M.M., Weidlich, O. (2011). Australian Mobile Phone Lifestyle Index. Retrieved from http://www.aimia.com.au/enews/mobile/ 090929%20AIMIA_Report_FINAL.pdf

Madar, A.A., Knutsen, K.V., Stene, L.C., Brekke, M., Meyer, H.E., Lagerløv, P. (2014). Effect of vitamin D3 supplementation on glycated hemoglobin (HbA1c), fructosamine, serum lipids, and body mass index: a randomized, double-blinded, placebo-controlled trial among healthy immigrants living in Norway. BMJ Open Diabetes Res Care, 2(1):e000026.

Maghbooli, Z., Hossein-Nezhad, A., Karimi, F., Shafaei, A.R., Larijani, B. (2008). Correlation between vitamin D3 deficiency and insulin resistance in pregnancy. Diabetes Metab Res Rev, 24(1):27-32.

Mair, S., Soyer, H.P., Youl, P., Hurst, C., Marshall, A., Janda, M. (2012). Personalised electronic messages to improve sun protection in young adults. J Telemed Telecare, 18(5):247-252.

Major, G.C., Alarie, F., Doré, J., Phouttama, S., Tremblay, A. (2007). Supplementation

370 with calcium + vitamin D enhances the beneficial effect of weight loss on plasma lipid and lipoprotein concentrations. Am J Clin Nutr, 85(1):54-59.

Major, G.C., Alarie, F.P., Doré, J., Tremblay, A. (2009). Calcium plus vitamin D supplementation and fat mass loss in female very low-calcium consumers: potential link with a calcium-specific appetite control. Br J Nutr, 101(5):659-663.

Margolis, K.L., Martin, L.W., Ray, R.M., Kerby, T.J., Allison, M.A., Curb, J.D., Kotchen, T.A., Liu, S., Wassertheil-Smoller, S., Manson, J.E. (2012). Women’s Health Initiative Investigators. A prospective study of serum 25-hydroxyvitamin D levels, blood pressure, and incident hypertension in postmenopausal women. Am J Epidemiol, 175(1):22-32.

Margolis, K.L., Ray, R.M., Van Horn, L., Manson, J.E., Allison, M.A., Black, H.R., Beresford, S.A., Connelly, S.A., Curb, J.D., Grimm, R.H.Jr., Kotchen, T.A, Kuller, L.H., Wassertheil-Smoller, S., Thomson, C.A., Torner, J.C. (2008). Women's Health Initiative Investigators. Effect of calcium and vitamin D supplementation on blood pressure: the Women's Health Initiative Randomized Trial. Hypertension, 52(5):847-855.

Marks, R., Foley, P.A., Jolley, D., Knight, K.R., Harrison, J., Thompson, S.C. (1995). The effect of regular sunscreen use on vitamin D levels in an Australian population: results of a randomized controlled trial. Arch Dermatol, 131:415-421.

Marques-Vidal, P., Vollenweider, P., Guessous, I., Henry, H., Boulat, O., Waeber, G., Jornayvaz, F.R. (2015). Serum Vitamin D Concentrations Are Not Associated with Insulin Resistance in Swiss Adults. J Nutr, 145(9):2117-2122.

Martinez, H.F.R., Chacra, W., Strickman-Stein, N., Levis, S. (2007). “Outdoor exercise reduces the risk of hypovitaminosis D in the obese,” Journal of Steroid Biochemistry and Molecular Biology, 103(3–5): 679-681.

Martins, D., Wolf, M., Pan, D., Zadshir, A., Tareen, N., Thadhani, R., Felsenfeld, A., Levine, B., Mehrotra, R., Norris, K. (2007). Prevalence of cardiovascular risk factors and the serum levels of 25-hydroxyvitamin D in the United States: data from the Third National Health and Nutrition Examination Survey. Arch Intern Med, 11;167(11):1159- 1165.

Mason, C., Xiao, L., Imayama, I., Duggan, C.R., Bain, C., Foster-Schubert, K.E., Kong,

371

A., Campbell, K.L., Wang, C.Y., Neuhouser, M.L., Li, L.W., Jeffery, R., Robien, K., Alfano, C.M., Blackburn, G.L., McTiernan, A. (2011). Effects of weight loss on serum vitamin D in postmenopausal women. Am J Clin Nutr, 94(1):95-103.

Mathers, C.D., Loncar, D. (2006). Projections of global mortality and burden of disease from 2002 to 2030. PLoS Med, 3(11):e442.

Matheson, L., Duffy, S., Maroof, A., Gibbons, R., Duffy, C., Roth, J. (2013). Intra- and inter-rater reliability of jumping mechanography muscle function assessments. J Musculoskelet Neuronal Interact, 13:480-486.

Matsuoka, L.Y., Ide, L., Wortsman, J., MacLaughlin, J.A., Holick, M.F. (1987). Sunscreens suppress cutaneous vitamin D3 synthesis. J Clin Endocrinol Metab, 64(6):1165-1168.

Maynard, L.M., Wisemandle, W., Roche, A.F., Chumlea, W.C., Guo, S.S., Siervogel, R.M. (2001). Childhood Body Composition in Relation to Body Mass Index. Pediatrics, 107: 2.

Mayor, S. (2016). Oestrogen-containing contraceptive pills increase vitamin D levels, study finds. BMJ. 7;354:i4345.

Mazahery, H., Von Hurst, P.R. (2015). Factors Affecting 25-Hydroxyvitamin D Concentration in Response to Vitamin D Supplementation. Nutrients, 7(7): 5111-5142.

McCarty, M.F., Thomas, C.A. (2003). PTH excess may promote weight gain by impeding catecholamine-induced lipolysis-implications for the impact of calcium, vitamin D, and alcohol on body weight. Med Hypotheses, 61: 535-542.

Mendis, S., Puska, P., Norrving, B. (2011). Global Atlas on Cardiovascular Disease Prevention and Control. World Health Organization (in collaboration with the World Heart Federation and World Stroke Organization), Geneva.

Michos, E.D., Blumenthal, R.S. (2007). Vitamin D Supplementation and Cardiovascular Disease Risk. Circulation, 115:827-828.

Millán, J., Pintó, X., Muñoz, A., Zúñiga, M., Rubiés-Prat, J., Pallardo, L.F., Masana, L., Mangas, A., Hernández-Mijares, A., González-Santos, P., Ascaso, J.F., Pedro-Botet, J. (2009). Lipoprotein ratios: Physiological significance and clinical usefulness in

372 cardiovascular prevention. Vasc Health Risk Manag, 5: 757-765.

Misra, M., Pacaud, D., Petryk, A. Collett-Solberg, P.F., Kappy, M., Drug and Therapeutics Committee of the Lawson Wilkins Pediatric Endocrine Society. (2008). Vitamin D deficiency in children and its management: review of current knowledge and recommendations. Pediatrics, 122(2):398-417.

Mitri, J., Muraru, M.D., Pittas, A.G. (2011). Vitamin D and type 2 diabetes: a systematic review. Eur J Clin Nutr, 65(9):1005-1015.

Molinari, C., Uberti, F., Grossini, E., Vacca, G., Carda, S., Invernizzi, M., Cisari, C. (2011). 1,25-Dihydroxycholecalciferol Induces Nitric Oxide Production in Cultured Endothelial Cells. Cell Physiol Biochem, 27:661-668.

Montague, M., Borland, R., Sinclair, C. (2001). Slip! Slop! Slap! and SunSmart, 1980 – 2000: skin cancer control and 20 years of population-based campaigning. Health Educ Behav, 28: 290-305.

Moschonis, G., Tanagra, S., Koutsikas, K., Nikolaidou, A., Androutsos, O., Manios, Y. (2009). Association between serum 25-hydroxyvitamin D levels and body composition in postmenopausal women: the Postmenopausal Health Study. Menopause, 16: 701-707.

Mosteller, R.D. (1987). Simplified calculation of body surface area. NEJM, 317:1098.

Munns, C., Zacharin, M.R., Rodda, C.P., Batch, J.A., Morley, R., Cranswick, N.E., Craig, M.E., Cutfield, W.S., Hofman, P.L., Taylor, B.J., Grover, S.R., Pasco, J.A., Burgner, D., Cowell, C.T. (2006). Prevention and treatment of infant and childhood vitamin D deficiency in Australia and New Zealand: a consensus statement. MJA, 185: 268-272.

Murni, I.K., Sulistyoningrum, D.C., Oktaria, V. (2016). Association of vitamin D deficiency with cardiovascular disease risk in children: implications for the Asia Pacific Region. Asia Pac J Clin Nutr, 25(Suppl 1):S8-S19.

Mutt, S.J., Hyppönen, E., Saarnio, J., Järvelin, M.R., Herzig, K.H. (2014). Vitamin D and adipose tissue-more than storage. Front Physiol, 24(5):228.

Myers, L.B., Horswill, M.S. (2006). Social cognitive predictors of sun protection intention and behavior. Behav Med, 32(2):57-63.

373

Nair, R., Maseeh, A. (2012). Vitamin D: The “sunshine” vitamin. J Pharmacol Pharmacother, 3(2): 118-126.

Nam, H., Kim, H.Y., Choi, J.S., Kweon, S.S., Lee, Y.H., Nam, H.S., Park, K.S., Ryu, S.Y., Choi, S.W., Oh, S.H., Kim, S.A., Shin, M.H. (2017). Association between Serum 25-hydroxyvitamin D Levels and Type 2 Diabetes in Korean Adults. Chonnam Med J, 53(1):73-77.

National Institutes of Health. (2018). Common Terminology Criteria for Adverse Events (CTCAE). Published: January 5, (Version 5).

National Statement on Ethical Conduct in Human Research (2007) - Updated May 2015 Retrieved from https://www.nhmrc.gov.au/guidelines-publications/e72.

Ni, W., Watts, S.W. Ng, M., Chen, S., Glenn D.J., Gardner, D.G. (2014). Elimination of Vitamin D Receptor in Vascular Endothelial Cells Alters Vascular Function. Hypertension, 64(6): 1290-1298.

Norval, M., Wulf, H.C. (2009). Does chronic sunscreen use reduce vitamin D production to insufficient levels? Br J Dermatol, 161(4):732-736.

Nutrient Reference Values (2018). www.nrv.gov.au

Ogurtsova, K., da Rocha Fernandes, J.D., Huang, Y., Linnenkamp, U., Guariguata, L., Cho, N.H., Cavan, D., Shaw, J.E., Makaroff, L.E. (2017). IDF Diabetes Atlas: Global estimates for the prevalence of diabetes for 2015 and 2040. Diabetes Res Clin Pract, 128:40-50.

Olokoba, A.B., Obateru, O.A., Olokoba, L.B. (2012). Type 2 Diabetes Mellitus: A Review of Current Trends. Oman Med J, 27(4): 269-273.

Ozdemir, A.A., Gundemir, Y.E., Kucuk, M., Saric, D.Y., Elgormuş, Y., Cag, Y., Bilek, G. (2018) Vitamin D Deficiency in Pregnant Women and Their Infants. J Clin Res Pediatr Endocrinol, 10(1):44-50.

Pacifico, L., Anania, C., Osborn, J.F., Ferraro, F., Bonci, E., Olivero, E., Chiesa, C. (2011). Low 25(OH)D3 levels are associated with total adiposity, metabolic syndrome, and hypertension in Caucasian children and adolescents. Eur J Endocrinol, 165(4):603- 611.

374

Palacios, C., Benedetti, P., Fonseca, S. (2007) Impact of calcium intake on body mass index in Venezuelan adolescents. P R Health Sci J 26, 199-204.

Palacios, C., Gonzales, L. (2014). Is vitamin D deficiency a major global public health problem? J Steroid Biochem Mol Biol, 144:138-145.

Pan, H., Fu, C., Huang, L., Jiang, Y., Deng, X., Guo, J., Su, Z. (2018). Anti-Obesity Effect of Chitosan Oligosaccharide Capsules (COSCs) in Obese Rats by Ameliorating Leptin Resistance and Adipogenesis. Mar Drugs, 16(6): 198.

Parikh, S., Guo, D.H., Pollock, N.K., Petty, K., Bhagatwala, J., Gutin, B., Houk, C., Zhu, H., Dong, Y. (2012). Circulating 25-hydroxyvitamin D concentrations are correlated with cardiometabolic risk among American black and white adolescents living in a year-round sunny climate. Diabetes Care, 35(5):1133-1138.

Parikh, S.J., Edelman, M., Uwaifo, G.I., Freedman, R.J., Semega-Janneh, M., Reynolds, J., Yanovski, J.A. (2004). The relationship between obesity and serum 1,25-dihydroxy vitamin D concentrations in healthy adults. J Clin Endocrinol Metab, 89(3):1196-1199.

Park, S., Leecorresponding, B.K. (2012). Vitamin D deficiency is an independent risk factor for cardiovascular disease in Koreans aged ≥ 50 years: results from the Korean National Health and Nutrition Examination Survey. Nutr Res Pract, 6(2): 162-168.

Parker, J., Hashmi, O., Dutton, D., Mavrodaris, A., Stranges, S., Kandala, N.B., Clarke, A., Franco, O.H. (2010). Levels of vitamin D and cardiometabolic disorders: systematic review and meta-analysis. Maturitas, 65(3):225-236.

Peng, W., Kanthawala, S., Yuan, S., Hussain, S.A. (2016). A qualitative study of user perceptions of mobile health apps. BMC Public Health, 16: 1158.

Pergola, GD, Silvestris, F. (2013). Obesity as a Major Risk Factor for Cancer. J Obesity; 2013: 291546.

Pfeifer, M., Begerow, B., Minne, H.W., Nachtigall, D., Hansen, C. (2001). Effects of a short-term vitamin D(3) and calcium supplementation on blood pressure and parathyroid hormone levels in elderly women. J Clin Endocrinol Metab, 86(4):1633-1637.

Piccolo, B.D., Dolnikowski, G., Seyoum, E., Thomas, A.P., Gertz, E.R., Souza, E.C., Woodhouse, L.R., Newman, J.W., Keim, N.L., Adams, S.H., Van Loan, M.D. (2013).

375

Association between subcutaneous white adipose tissue and serum 25-hydroxyvitamin D in overweight and obese adults. Nutrients, 5(9):3352-3366.

Pilz, S., Gaksch, M., Kienreich, K., Grübler, M., Verheyen, N., Fahrleitner-Pammer, A., Treiber, G., Drechsler, C., Hartaigh, B., Obermayer-Pietsch, B., Schwetz, V., Aberer, F., Mader, J., Scharnagl, H., Meinitzer, A., Lerchbaum, E., Dekker, J.M., Zittermann, A., März, M., Tomaschitz, A. (2015). Effects of Vitamin D on Blood Pressure and Cardiovascular Risk Factors A Randomized Controlled Trial. Hypertension, 65(6):1195- 201.

Pilz, S., Verheyen, N., Grübler, M.R., Tomaschitz, A., März, W. (2016). Vitamin D and cardiovascular disease prevention. Nat Rev Cardiol, 13(7):404-417.

Pittas, A.G., Chung, M., Trikalinos, T., Mitri, J., Brendel, M., Patel, K., Lichtenstein, A.H., Lau, J., Balk, E.M. (2010). Systematic review: Vitamin D and cardiometabolic outcomes. Ann Intern Med, 2;152(5):307-314.

Pittas, A.G., Dawson-Hughes, B. (2010). Vitamin D and diabetes. J Steroid Biochem Mol Biol, 121:425-429

Pittas, A.G., Dawson-Hughes, B., Li, T., Van Dam, R.M., Willett, W.C., Manson, J.E., Hu, F.B. (2006). Vitamin D and calcium intake in relation to type 2 diabetes in women. Diabetes Care, 29(3):650-656.

Pittas, A.G., Harris, S.S., Stark, P.C., Dawson-Hughes, B. (2007). The effects of calcium and vitamin D supplementation on blood glucose and markers of inflammation in nondiabetic adults. Diabetes Care, 30(4):980-986.

Poirier, P., Giles, T.D., Bray, G.A., Hong, Y., Stern, J.S., Pi-Sunyer, F.X., Eckel, R.H. (2006). Obesity and cardiovascular disease: pathophysiology, evaluation, and effect of weight loss: an update of the 1997 American Heart Association Scientific Statement on Obesity and Heart Disease from the Obesity Committee of the Council on Nutrition, Physical Activity, and Metabolism. Circulation,14;113(6):898-918.

Ponda, M.P., Huang, X., Odeh, M.A., Breslow, J.L., Kaufman, H.W. (2012). Vitamin D may not improve lipid levels: a serial clinical laboratory data study. Circulation, 126(3):270-277.

376

Prentice A. (2008). Vitamin D deficiency: a global perspective. Nutr Rev,66:153-164.

Prichard, I., Dobbinson, S., Wilson, C., H utchinson, A.D., Rayner, J., Makin, J. (2015). Perceptions of the solarium ban in Australia: 'Fake it, don't bake it'. Health Promot J Austr, 26(2):154-158.

Pritchard, J.E. (1988). The relationship of body composition to energy balance. PhD thesis. Melbourne, Australia.

Pritchard, J.E., Nowson, C.A., Strauss, B.J., Carlson, J.S., Kaymakci, B., Wark, J.D. (1993). Evaluation of dual energy X-ray absorptiometry as a method of measurement of body fat. Eur J Clin Nutr, 47(3):216-228.

Ramachandran, A., Snehalatha, C., Ram, J., Selvam, S., Simon, M., Nanditha, A., Shetty, A.S., Godsland, I.F., Chaturvedi, N., Majeed, A., Oliver, N., Toumazou, C., Alberti, K.G., Johnston, D.G. (2013). Effectiveness of mobile phone messaging in prevention of type 2 diabetes by lifestyle modification in men in India: a prospective, parallel-group, randomised controlled trial. Lancet Diabetes Endocrinol, 1(3):191-198.

Rathish, N., Arun, M. (2012). Vitamin D: The “sunshine” vitamin. J Pharmacol Pharmacother, 3(2): 118-126.

Rautiainen, S., Wang, L., Lee, I.M., Manson, J.E., Buring, J.E., Sesso, H.D. (2016). Dairy consumption in association with weight change and risk of becoming overweight or obese in middle-aged and older women: a prospective cohort study. Am J Clin Nutr, 103(4):979-988.

Reavley, N., Allen, N., Jorm, A., Morgan, A., Purcell, R. (2010). beyondblue: Melbourne. A Guide to What Works for Anxiety Disorders. Retrieved from https://www.beyondblue.org.au/docs/default-source/resources/bl-0762-guide-to-what- works-for-anxiety-02-14.pdf?sfvrsn=2 webcite

Reddy Vanga, S., Good, M., Howard, P.A., Vacek, J.L. (2010). Role of Vitamin D in Cardiovascular Health. Am J Cardiol, 106(6):798-805.

Reinehr, T. (2013). Type 2 diabetes mellitus in children and adolescents. World J Diabetes, 4(6): 270-281.

Rhee, S.Y., Hwang, Y.C., Chung, H.Y., Woo, J.T. (2012). Vitamin D and diabetes in

377

Koreans: analyses based on the Fourth Korea National Health and Nutrition Examination Survey (KNHANES), 2008-2009. Diabet Med, 29(8):1003-1010.

Rimm, E.B., Williams, P., Fosher, K., Criqui, M., Stampfer, M.J. (1999). Moderate alcohol intake and lower risk of coronary heart disease: meta-analysis of effects on lipids and haemostatic factors. BMJ, 11; 319 (7224):1523-1528.

Ripsin, C.M., Kang, H., Urban, R.J. (2009). "Management of blood glucose in type 2 diabetes mellitus". American Family Physician, 79 (1): 29-36.

Rock, C.L., Emond, J.A., Flatt, S.W., Heath, D.D., Karanja, N., Pakiz, B., Sherwood, N.E., Thomson, C.A. (2012). Weight loss is associated with increased serum 25- hydroxyvitamin D in overweight or obese women. Obesity (Silver Spring), 20(11):2296- 2301.

Rockell, J.E., Green, T.J., Skeaff, C.M., Whiting, S.J., Taylor, R.W., Williams, S.M., Parnell, W.R., Scragg, R., Wilson, N., Schaaf, D., Fitzgerald, E.D., Wohlers, M.W. (2005). Season and ethnicity are determinants of serum 25-hydroxyvitamin D concentrations in New Zealand children aged 5-14 y. J Nutr, 135(11):2602-2608.

Rockell, J.E.P., Skeaff, C.M., Williams, S.M., Green, T.J. (2008). Association between quantitative measures of skin color and plasma 25-hydroxyvitamin D. Osteoporos Int , 19:1639–1642.

Rodrigue, J.R. (1996). Promoting healthier behaviors, attitudes, and beliefs toward sun exposure in parents of young children. J Consult Clin Psychol, 64(6):1431-1436.

Rosenblum, J.L., Castro, V.M., Moore, C.E., Kaplan, L.M. (2012).Calcium and vitamin D supplementation is associated with decreased abdominal visceral adipose tissue in overweight and obese adults. Am J Clin Nutr, 95(1):101-108.

Ross, A.C., Taylor, C.L. (2011). Dietary Reference Intakes for Calcium and Vitamin D. Washington, DC: National Academies Press, Institute of Medicine.

Ross, A.C., Taylor, C.A., Yaktine, S.L., Del Valle, H.B. (2011). Dietary Reference Intakes for Calcium and Vitamin D. Washington (DC): National Academies Press (US).

Roth, G.A., Huffman, M.D., Moran, A.E., Feigin, V., Mensah, G.A., Naghavi, M., Murray, C.J. (2015). Global and regional patterns in cardiovascular mortality from 1990

378 to 2013. Circulation,132(17):1667-1678.

Ryu, O.H., Chung, W., Lee, S., Hong, K.S., Choi, M.G., Yoo, H.J. (2014). The effect of high-dose vitamin D supplementation on insulin resistance and arterial stiffness in patients with type 2 diabetes. Korean J Intern Med, 29(5):620-629.

Sabanayagam, C., Shankar, A., Somasundaram, S. (2012). Serum vitamin D level and prehypertension among subjects free of hypertension. Kidney Blood Press Res, 35(2):106-113.

Sachse, M.M., Böttcher, S., Pape, L., Wagner, G., Mehls, O., Klaus, G., Laschewski, G., Barz, M., Jahn, I., Zeeb, H. (2016). Face-to-face Sun Protection Training and Text Messages Improve Sun Protection Behaviour in Adolescent Organ Transplant Recipients: HIPPOlino Feasibility Study. Acta Derm Venereol, 96(3):341-345.

Salehpour, A., Hosseinpanah, F., Shidfar, F., Vafa, M., Razaghi, M., Dehghani, S., Hoshiarrad, A., Gohari, M. (2012). A 12-week double-blind randomized clinical trial of vitamin D3 supplementation on body fat mass in healthy overweight and obese women. Nutr J, 11: 78.

Samanek, A.J., Croager, E.J., Gies, P., Milne, E., Prince, R., McMichael, A.J., Lucas, R.M., Slevin, T. (2006). Estimates of beneficial and harmful sun exposure times during the year for major Australian population centres.; Skin Cancer Prevention. Med J Aust, 3;184(7):338-341.

Sanci, L.A., Sawyer, S.M., Weller, P.J., Bond, L.M., Patton, G.C. (2004). Youth health research ethics: time for a mature-minor clause? Med J Aust,180:336-338.

Saudek, C.D., Herman, W.H., Sacks, D.B., Bergenstal, R.M., Edelman, D., Davidson, M.B. (2008). A new look at screening and diagnosing diabetes mellitus. J Clin Endocrinol Metab, 93:2447-2453.

Scott-Sheldon, L.A., Lantini, R., Jennings, E.G., Thind, H., Rosen, R.K., Salmoirago- Blotcher, E., Bock, B.C. (2016). Text Messaging-Based Interventions for Smoking Cessation: A Systematic Review and Meta-Analysis. JMIR Mhealth Uhealth, 4(2):e49.

Scragg, R., Sowers, M., Bell, C. (2004). Third National Health and Nutrition Examination Survey.Serum 25-hydroxyvitamin D, diabetes, and ethnicity in the Third National Health

379 and Nutrition Examination Survey. Diabetes Care, 27(12):2813-2818.

Scragg, R., Sowers, M., Bell, C. (2007). erum 25-hydroxyvitamin D, ethnicity, and blood pressure in the Third National Health and Nutrition Examination Survey. Am J Hypertens, 20(7):713-719.

Seida, J.C., Mitri, J., Colmers, I.N., Majumdar, S.R., Davidson, M.B., Edwards, A.L., Hanley, D.A., Pittas, A.G., Tjosvold, L., Johnson, J.A. (2014). Clinical review: Effect of vitamin D3 supplementation on improving glucose homeostasis and preventing diabetes: a systematic review and meta-analysis. J Clin Endocrinol Metab, 99(10):3551-3560.

Sensis social media report. (2017). Retrieved from https://www.sensis.com.au/asset/PDFdirectory/Sensis_Social_Media_Report_2017- Chapter-1.pdf. Accessed 10/1/218

Shi, H., Dirienzo, D., Zemel, M.B. (2001). Effects of dietary calcium on adipocyte lipid metabolism and body weight regulation in energy-restricted aP2-agouti transgenic mice. FASEB J, 15:291-293.

Shrapnel, W., Truswell, S. (2006). Vitamin D deficiency in Australia and New Zealand: What are the dietary options? Nutrition & Dietetics, 63 (4).

Skaaby, T., Thuesen, B.H., Linneberg, A. (2017). Vitamin D, Cardiovascular Disease and Risk Factors. Adv Exp Med Biol, 996:221-230.

Skowrońska, B., Fichna, M., Fichna, P. (2005). The role of adipose tissue in endocrine system. Endokrynol Otyłość 1: 21-29

Sneve, M., Figenschau, Y., Jorde, R. (2008). Supplementation with cholecalciferol does not result in weight reduction in overweight and obese subjects. European Journal of Endocrinology,159:675-684

Sneyd, M., Cox, B. (2006). The control of melanoma in New Zealand. N Z Med J. 119(1242):U2169.

Snijder, M.B., van Dam, R.M., Visser, M., Deeg, D.J., Dekker, J.M., Bouter, L.M., Seidell, J.C., Lips, P. (2005). Adiposity in relation to vitamin D status and parathyroid hormone levels: a population-based study in older men and women. J Clin Endocrinol Metab, 90(7):4119-4123.

380

Spedding, S., Vanlint, S., Morris, H., Scragg, R. (2013). Does vitamin D sufficiency equate to a single serum 25-hydroxyvitamin D level or are different levels required for non-skeletal diseases? Nutrients, 5(12):5127-5139.

Starfelt Sutton, L.C., White, K.M. (2016). Predicting sun-protective intentions and behaviours using the theory of planned behaviour: a systematic review and meta-analysis. Psychol Health, 31(11):1272-1292.

Strazdienelow asterisk, V., Tamulaitiene, M., Mastaviciute, A., Alekna ,V. (2012). Association of vitamin D with lean mass and handgrip strength. Bone, 50 (1):198.

Stricker, H., Bianda, F.T., Guidicelli-Nicolosi, S., Limoni, S., Colucci, G. (2012). Effect of a Single, Oral, High-Dose Vitamin D Supplementation on Endothelial Function in Patients With Peripheral Arterial Disease: A Randomised Controlled Pilot Study. Eur J Vasc Endovasc Surg, 44 (3), 307-312.

Strobel, F., Reusch, J., Penna-Martinez, M., Ramos-Lopez, E., Klahold, E., Klepzig, C., Wehrle, J., Kahles, H., Badenhoop, K. (2014). Effect of a randomised controlled vitamin D trial on insulin resistance and glucose metabolism in patients with type 2 diabetes mellitus. Horm Metab Res, 46(1):54-58.

Stults-Kolehmainen, M.A., Stanforth, P.R., Bartholomew, J.B., Lu, T., Abolt, C.J., Sinha, R. (2013). DXA estimates of fat in abdominal, trunk and hip regions varies by ethnicity in men. Nutrition & Diabetes, 3:e64.

Sung, C., Liao, M., Lu, K., Wu, C. (2012). Role of Vitamin D in Insulin Resistance. J Biomed Biotechnol, 2012:634195.

SunSmart program (2017). Retrieved from http://www.sunsmart.com.au/about/sunsmart- program

SunSmart. (2017). Skin cancer types. Retrieved from http://www.sunsmart.com.au/skin- cancer/skin-cancer-types.

Swainson, M.G., Batterham, A.M., Tsakirides, C., Rutherford, Z.H., Hind, K. (2017). Prediction of whole-body fat percentage and visceral adipose tissue mass from five anthropometric variables. PLOS ONE, 12 (5).

Tabesh, M., Salehi-Abargouei, A., Tabesh, M., Esmaillzadeh, A. (2013). Maternal

381 vitamin D status and risk of pre-eclampsia: a systematic review and meta-analysis. J Clin Endocrinol Metab, 98(8):3165-3173.

Tabesh, M., Callegari, E.T., Gorelik, A., Garland, S.M., Nankervis, A., Subasinghe, A.K., Wark, J.D., on behalf of the YFHI and Safe-D study groups. (2017). Associations between vitamin D status, adiposity and lipid profiles in young women. 2018 Jan 24. doi: 10.1038/s41430-018-0086-1.

Tabesh, M., Garland, S.M., Gorelik, A., Nankervis, A., Maclean, S., Callegari, E.T., Chang, S., Heffernan, K., Wark, J.D. (2016). Improving Vitamin D Status and Related Health in Young Women: The Safe-D study - Part B. JMIR Res Protoc, 10; 5(2):e80.

Talmor, Y., Golan, E., Benchetrit, S., Bernheim, J., Klein, O. Green, J., Rashid, G. (2008). Calcitriol blunts the deleterious impact of advanced glycation end products on endothelial cells. Am J Physiol Renal Physiol, 294: F1059-F1064.

Tardelli, V.S., Lago, M.P.P.D., Silveira, D.X.D., Fidalgo, T.M. (2017). Vitamin D and alcohol: A review of the current literature. Psychiatry Res, 248:83-86.

Tay, I., Garland ,S., Gorelik, A., Wark, J.D. (2017). Development and Testing of a Mobile Phone App for Self-Monitoring of Calcium Intake in Young Women. JMIR Mhealth Uhealth,5(3): e27.

Thimbleby, H. (2013). Technology and the Future of Healthcare. J Public Health Res, 1; 2(3): e28.

Thompson, S.C., Jolley, D., Marks, R. (1993). Reduction of solar keratoses by regular sunscreen use. N Engl J Med. 14;329(16):1147-1151.

Truthmann, J., Busch, M.A., Scheidt-Nave, C., Mensink, G.B.M., Gößwald, A., Endres M., Neuhauser, H. (2015). Modifiable cardiovascular risk factors in adults aged 40–79 years in Germany with and without prior coronary heart disease or stroke. BMC Public Health, 15: 701.

Turner, R.C. (1985). Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia, 28(7):412-419.

Turner-McGrievy, G.M., Beets, M.W., Moore, J.B., Kaczynski, A.T., Barr-Anderson,

382

D.J., Tate, D.F. (2013). Comparison of traditional versus mobile app self-monitoring of physical activity and dietary intake among overweight adults participating in an mHealth weight loss program. J Am Med Inform Assoc. 20(3): 513-518.

Tyagi, S., Gupta, P., Singh Saini, A., Kaushal, C., Sharma, S. (2011). The peroxisome proliferator-activated receptor: A family of nuclear receptors role in various diseases. J Adv Pharm Technol Res, 2(4): 236-240.

Tzotzas, T., Vlahavas, G., Papadopoulou, S.K., Kapantais, E., Kaklamanou, D., Hassapidou, M. (2010). Marital status and educational level associated to obesity in Greek adults: data from the National Epidemiological Survey. BMC Public Health, 10: 732.

Uberti, F., Lattuada, D., V.Morsanuto, V., U.Nava, U., Bolis, G., Vacca, G., Squarzanti, D.F., Cisari, C., Molinari, C. (2013). Vitamin D protects Human Endothelial Cells from oxidative stress through autophagic and survival pathways. J Clin Endocrinol Metab, 99(4):1367-1374.

Valiña-Tóth, A.L.B. Lai, Z., Yoo, W., Abou-Samra, A., Gadegbeku, C.A., Flack, J.B. (2010). Relationship of Vitamin D and Parathyroid Hormone to Obesity and Body Composition in African Americans. Clin Endocrinol (Oxf), 72(5): 595-603.

Van den Ouweland, J.M., Vogeser, M., Bächer, S. (2013). Vitamin D and metabolites measurement by tandem mass spectrometry. Rev Endocr Metab Disord, 14(2):159-184.

Vanlint, S. (2013). Vitamin D and Obesity. Nutrients, 5(3): 949-956.

Ventola, CL. (2014). Mobile Devices and Apps for Health Care Professionals: Uses and Benefits. P T, 39(5): 356-364.

Vijan, S. (2010). "Type 2 diabetes". Annals of Internal Medicine. 152 (5): ITC31–ITC15; quiz ITC316.

Vilarrasa, N., Vendrell, J., Maravall, J., Elío, I., Solano, E., San José, P., García, I., Virgili, N., Soler, J., Gómez, J.M. (2010): Is plasma 25 (OH) D related to adipokines, inflammatory cytokines and insulin resistance in both healthy and morbidly obese population? Endocr 38: 235-242.

Vitezova, A., Muka, T., Zillikens, M.C., Voortman, T., Uitterlinden, A.G. (2017). Albert

383

Hofman, Fernando Rivadeneira, Jessica C. Kiefte-de Jong, Oscar H. Franco. Vitamin D and body composition in the elderly. Clinical Nutrition 36, 585-592

Vogt, S., Zierer, A., Laxy, M., Koenig, W., Linkohr, B., Linseisen, J., Peters, A., Thorand, B. (2016). Association of serum vitamin D with change in weight and total body fat in a German cohort of older adults. Eur J Clin Nutr, 70(1):136-139.

Volkov, A., Dobbinson, S., Wakefield, M., Slevin, T. (2013). Seven-year trends in sun protection and sunburn among Australian adolescents and adults. Aust N Z J Public Health, 37(1):63-69.

Von Websky, K., Hasan, A.A., Reichetzeder, C., Tsuprykov, O., Hocher, B. (2018). Impact of vitamin D on pregnancy-related disorders and on offspring outcome. J Steroid Biochem Mol Biol, 180:51-64.

Vu, D., Ong, J.M., Clemens, T.L., Kern, P.A. (1996). 1,25-Dihydroxyvitamin D induces lipoprotein lipase expression in 3T3-L1 cells in association with adipocyte differentiation. Endocrinology, 137(5):1540-1544.

Vuadens, A., Ackermann, S., Levi, F., Bulliard, J.L. (2017). Sun-related knowledge and attitudes of primary and secondary schoolchildren in western Switzerland. Eur J Cancer Prev, 26(5):411-417.

Wakasugi, M., Noguchi, T., Inoue, M., Kazama, Y., Tawata, M., Kanemaru, Y., Onaya, T. (1991). Vitamin D3 stimulates the production of prostacyclin by vascular smooth muscle cells. Prostaglandins, 42(2):127-136.

Wakayo, T., Whiting, S.J., Belachew, T. (2016). Vitamin D Deficiency is Associated with Overweight and/or Obesity among Schoolchildren in Central Ethiopia: A Cross-Sectional Study. Nutrients, 8(4): 190.

Walls, H.L., Magliano, D.J., Stevenson, C.E., Backholer, K., Mannan, H.R., Shaw, J.E., Peeters, A. (2012). Projected progression of the prevalence of obesity in Australia. Obesity (Silver Spring), 20(4):872-878.

Wamberg, L., Kampmann, U., Stødkilde-Jørgensen, H., Rejnmark, L., Pedersen, S,B., Richelsen, B. (2013). Effects of vitamin D supplementation on body fat accumulation, inflammation, and metabolic risk factors in obese adults with low vitamin D levels -

384 results from a randomized trial. Eur J Intern Med, 24(7):644-649.

Wang, H., Chen, W., Li, D., Yin, X., Zhang, X., Olsen, N., Zheng, S.G. (2017). Vitamin D and Chronic Diseases. Aging and Disease, 8(3): 346-353.

Wang, T.J., Pencina, M.J., Booth, S.L., Jacques, P.F., Ingelsson, E., Lanier, K., Benjamin, E.J., D'Agostino, R.B., Wolf, M., Vasan, R.S. (2008). Vitamin D deficiency and risk of cardiovascular disease. Circulation, 117(4):503-511.

Wang, Y., Rimm, E.B., Stampfer, M.J., Willett, W.C., Hu, F.B. (2005). Comparison of abdominal adiposity and overall obesity in predicting risk of type 2 diabetes among men. Am J Clin Nutr, 81(3):555-563.

Wehner, M.R., Shive, M.L., Chren, M.M., Han, J., Qureshi, A.A., Linos, E. (2012). Indoor tanning and non-melanoma skin cancer: systematic review and meta-analysis. BMJ, 2;345:e5909.

Whittaker, R., Borland, R., Bullen, C., Lin Ruey, B., McRobbie, H., Rodgers, A. (2009). Mobile phone-based interventions for smoking cessation. Cochrane Database of Systematic Reviews, 7;(4):CD006611.

Wichstrøm, L. (1994). Predictors of Norwegian adolescents sunbathing and use of sunscreen. Health Psychol, 13: 412-420

Williams, D.M., Fraser, A., Lawlor, D.A. (2011). Associations of vitamin D, parathyroid hormone and calcium with cardiovascular risk factors in US adolescents. Heart, 97(4):315-320.

Wilmot, E., Idris, I. (2014). Early onset type 2 diabetes: risk factors, clinical impact and management. Ther Adv Chronic Dis, 5(6): 234-244.

Wimalawansa, S.J. (2017). Vitamin D and cardiovascular diseases: Causality, J. Steroid Biochem. Mol. Biol, 175:29-43.

Witham, M.D., Dove, F.J., Dryburgh, M., Sugden, J.A., Morris, A.D., Struthers, A.D. (2010). The effect of different doses of vitamin D(3) on markers of vascular health in patients with type 2 diabetes: a randomised controlled trial. Diabetologia, 53(10):2112- 2119.

385

Witham, M.D., Dove, F.J., Sugden, J.A., Doney, A.S., Struthers, A.D. (2012). The effect of vitamin D replacement on markers of vascular health in stroke patients - a randomised controlled trial. Nutr Metab Cardiovasc Dis, 22(10):864-870.

Witham, M.D., Nadir, M.A., Struthers, A.D. (2009). Effect of vitamin D on blood pressure: a systematic review and meta-analysis. J Hypertens, 27(10):1948-1954.

Wood, A.D., Secombes, K.R., Thies, F., Aucott, L., Black, A.J., Mavroeidi, A., Simpson, W.G., Fraser, W.D., Reid, D.M., Macdonald, H.M. (2012). Vitamin D3 supplementation has no effect on conventional cardiovascular risk factors: a parallel-group, double-blind, placebo-controlled RCT. J Clin Endocrinol Metab, 97(10):3557-3568.

Wood, A.W., Karipidis, K. (2017). Non-ionizing radiation protection. John Wiley Nad Sons Inc.

Wootton, A.M. (2005). Improving the measurement of 25-hydroxyvitamin D. Clin Biochem Rev, 26:33-36.

World Health Organisation (WHO). Skin Cancer. (2017). Retrieved from http://www.who.int/uv/faq/skincancer/en/index1.html.

World Health Organization (WHO). (2014). Cardiovascular diseases (CVDs). Retrieved from http://www.who.int/mediacentre/factsheets/fs317/en/

World Health Organization (WHO). (2017, May). Cardiovascular disease. Retrieved from http://www.who.int/mediacentre/factsheets/fs317/en/

World Health Organization. (2003). Diet, nutrition and the prevention of chronic diseases. Retrieved from www.fao.org/DOCREP/005/AC911E/AC911E00.HTM. Accessed 26 July 2010.

World Health Organization. (2017). Waist circumference and waist-hip ratio. Retrieved from http://apps.who.int/iris/bitstream/10665/44583/1/, Accessed 5-10-2017

World Health Organization. (2017). Definition and diagnosis of diabetes mellitus and intermediate hyperglycemia. Retrieved from http://www.who.int/diabetes/publications/Definition%20and%20diagnosis%20of%20diab etes_new.pdf , Accessed 5-10-2017

386

World Medical Association. (2013). Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA, 27;310(20):2191-2194.

Wright, C.Y., Albers, P.N., Reeder, A.I., Mathee, A. (2017). Sunbeds and skin cancer risk: quantifying a baseline estimate of sunbed facilities in South Africa prior to implementation of sunbed regulations. Pan Afr Med J, 30;26:188.

Wright, C.Y., Reeder, A.I. (2005). Youth solar ultraviolet radiation exposure, concurrent activities and sun-protective practices: a review. Photochem Photobiol, 81(6):1331-1342.

Wu, C.H., Yao, W.J., Lu, F.H., Yang, Y.C., Wu, J.S., Chang, C.J. (2001). Sex differences of body fat distribution and cardiovascular dysmetabolic factors in old age. Age Ageing, 30(4):331-336.

Wulaningsih, W., Van Hemelrijck, M., Tsilidis, K.K., Tzoulaki, I., Patel, C., Rohrmann, S. (2017). Investigating nutrition and lifestyle factors as determinants of abdominal obesity: an environment-wide study. International Journal of Obesity, 41: 340-347.

Xiang, W., Liao, W., Yi, Z., He, X., Ding, Y. (2017). 25-Hydroxyvitamin D-1-α- hydroxylase in apoliporotein E knockout mice: The role of protecting vascular smooth muscle cell from calcification. Biomed Pharmacother, 88:971-977.

Yao, Y., Zhu, L., He, L., Duan, Y., Liang, W., Nie, Z., Jin, Y., Wu, X., Fang, Y. (2015). A meta-analysis of the relationship between vitamin D deficiency and obesity. Int J Clin Exp Med, 8(9): 14977-14984.

Yiu, Y.F., Yiu, K.H., Siu, C.W., Chan, Y.H., Li, S.W., Wong, L.Y., Lee, S.W., Tam, S., Wong, E.W., Lau, C.P., Cheung, B.M., Tse, H.F. (2013). Randomized controlled trial of vitamin D supplement on endothelial function in patients with type 2 diabetes. Atherosclerosis, 227(1):140-146.

Youl, P.H., Janda, M., Kimlin, M. (2009). Vitamin D and sun protection: the impact of mixed public health messages in Australia. Int J Cancer, 124(8):1963-1970.

Zhang, M., Li, P., Zhu, Y., Chang, H., Wang, X., Liu, W., Zhang, Y., Huang, G. (2015). Higher visceral fat area increases the risk of vitamin D insufficiency and deficiency in Chinese adults. Nutr Metab (Lond), 12: 50.

Zhang, H.Q., Teng, J.H., Li, Y., Li, X.X., He, Y.H., He, X., Sun, C.H. (2014). Vitamin D

387 status and its association with adiposity and oxidative stress in schoolchildren. Nutrition, 30(9):1040-1044.

Zhao, W., Katzmarzyk, P.T., Horswell, R., Wang, Y., Johnson, J., Hu, G. (2014). HbA1c and coronary heart disease risk among diabetic patients. Diabetes Care, 37(2):428-435.

Zhou Y., Rui L. (2013). Leptin signaling and leptin resistance. Front Med, 7(2): 207-222.

Zhou, J., Zhao, L.J., Watson, P., Zhang, Q., Lappe, J.M. (2010). The effect of calcium and vitamin D supplementation on obesity in postmenopausal women: secondary analysis for a large-scale, placebo controlled, double-blind, 4-year longitudinal clinical trial. Nutrition & Metabolism,7:62.

Zhou, J.C., Zhu, Y.M., Guo, P., Chen, Z., Xie, F.Z., Liu, X.L., He, S. (2013). Serum 25(OH)D and lipid levels in Chinese obese and normal weight males before and after oral vitamin D supplementation. Biomed Environ Sci, 26(10):801-807.

Zittermann A., Gummert J.F. (2010). Nonclassical Vitamin D Action. Nutrients,(4): 408- 425.

Zittermann, A., Frisch, S., Berthold, H.K., Götting, C., Kuhn, J., Kleesiek, K., Stehle, P., Koertke, H., Koerfer, R. (2009). Vitamin D supplementation enhances the beneficial effects of weight loss on cardiovascular disease risk markers. Am J Clin Nutr, 89(5):1321-1327.

388

Minerva Access is the Institutional Repository of The University of Melbourne

Author/s: Tabesh, Marjan

Title: Understanding the relationship between obesity/ fat distribution/metabolic profiles and vitamin D status in young women (the safe-D study)

Date: 2018

Persistent Link: http://hdl.handle.net/11343/216474

File Description: Understanding the Relationship between Obesity/ Fat distribution/Metabolic Profiles and Vitamin D Status in Young Women (The Safe-D study)

Terms and Conditions: Terms and Conditions: Copyright in works deposited in Minerva Access is retained by the copyright owner. The work may not be altered without permission from the copyright owner. Readers may only download, print and save electronic copies of whole works for their own personal non-commercial use. Any use that exceeds these limits requires permission from the copyright owner. Attribution is essential when quoting or paraphrasing from these works.