Assessing the impact of maternal Omega-3 LCPUFA DHA on the body composition of children at 7 years of age using Air Displacement Plethysmography (ADP)

Transcription

1 Assessing the impact of maternal Omega-3 LCPUFA DHA on the body composition of children at 7 years of age using Air Displacement Plethysmography (ADP) Katie Wood MND BNutFoodSc School of Agriculture, Food and Wine The University of Adelaide South Australia A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy July 2016 P a g e

2 TABLE OF CONTENTS LIST OF FIGURES... vi ABSTRACT... viii DECLARATION... x ACKNOWLEDGEMENTS... xi LIST OF ABBREVIATIONS... xiii CHAPTER INTRODUCTION AND LITERATURE REVIEW... 2 PART CONTRIBUTORS TO CHILDHOOD OBESITY AND THE ROLE OF N-3 LCPUFA The Early Origins of Obesity N-3 LCPUFA in Perinatal Nutrition Perinatal N-3 LCPUFA and Body Composition Animal Models and in vitro studies Maternal n-3 LCPUFA supplementation and infant body composition: Evidence from human studies Other Contributors to Childhood Obesity Dietary intake Methods of measuring dietary intake in children Physical Activity and Screen Time PART ASSESSMENT OF BODY COMPOSITION IN CHILDREN Defining overweight and obesity in children: the use of Body Mass Index (BMI) Limitations of BMI for defining overweight and obesity Current approaches for measuring body Composition in children Challenges in measuring body composition Body composition compartment methods Skinfold Thickness (SFT) Dual energy X-ray Absorptiometry (DXA) i P a g e

3 1.7.5 Bioelectrical Impedance Analysis (BIA) and Bioelectrical Impedance Spectroscopy (BIS) Air Displacement Plethysmography using the BOD POD Validation of the BOD POD Comparison of BIA and ADP measurements in children Other/additional limitations to the use of the BOD POD in paediatric populations Other Considerations: The importance of fat distribution Conclusion Rationale and outline of this thesis CHAPTER METHODS AND MATERIALS Study Population Follow-up and Clinic Appointments Anthropometric outcomes Body composition outcomes Bioelectrical Impedance Spectroscopy (BIS) measurements Air Displacement Plethysmography (ADP) BOD POD reproducibility Dietary Intake Quantification of diet quality Physical Activity and Screen Time Other Measures CHAPTER RESULTS EFFECT OF MATERNAL DHA SUPPLEMENTATION ON BODY COMPOSITION OF THE CHILD Introduction Methods Flow of participants BOD POD and BIS testing Sociodemographic characteristics ii P a g e

4 3.2.4 Statistical Analysis Baseline characteristics of participants Sociodemographic Characteristics at the time of the 7 year follow-up Body composition measurements (fat mass and fat free mass) using BIS and the BOD POD Body composition outcomes by treatment group Male vs Female children Anthropometric measurements Anthropometric outcomes by treatment group Male vs Female children Summary of the results Discussion Limitations Overall conclusions and implications CHAPTER RESULTS DIETARY INTAKE, PHYSICAL ACTIVITY AND SCREEN TIME Introduction Methods Assessment of Dietary intake, Physical Activity and Screen Time Study population Diet and family food environment Determination of DGI-CA SCORE Physical Activity and Screen Time Statistical analysis RESULTS Dietary intake Dietary intake by treatment group Male vs Female children Family food environment iii P a g e

5 4.3.5 Relationship between DGI-CA, fat mass % and BMI z-score Physical Activity and Screen Time Physical Activity by treatment group Male vs Female children Screen Time (television and computer games) by treatment group Screen Time (television and computer games) in Male vs Female children Relationship between Physical Activity and Screen Time and BMI z- score and fat mass % Summary of the results Discussion Limitations Conclusion CHAPTER RESULTS COMPARISON OF BIOELECTRICAL IMPEDANCE SPECTROSCOPY (BIS) AND BOD POD MEASUREMENTS OF FAT/FAT FREE MASS Introduction METHODS Study population BOD POD and BIS testing Statistical analysis RESULTS BOD POD vs BIS fat mass percentage Relationship between fat mass measures using BOD POD or BIS and BMI z- score BOD POD vs BIS compliance Summary of the results Discussion Limitations Overall conclusions and implications iv P a g e

6 CHAPTER General discussion Concluding remarks and future direction APPENDIX DIET AND PHYSICAL ACTIVITY QUESTIONNAIRE REFERENCES v P a g e

7 LIST OF FIGURES Figure 1.1 Diagram representation of major system components BOD POD machine. 38 Figure 2.1 Clinic room Norwich building, North Adelaide where all appointments were conducted Table 2.2 Components of the Dietary Guideline Index-Children and Adolescents (DGI- CA) reflecting the 2013 Australian Dietary Guidelines Figure 3.1 Flow of study participants Table 3.1 Demographic characteristics of primary and secondary carers of the DOMInO study at 7 years of age Table 3.2 Body composition outcomes by treatment group Table 3.3 Body composition outcomes by sex Table 3.4 Anthropometric outcomes by treatment group Table 3.5 Anthropometric outcomes by sex Table 3.6 BMI for age cut off levels for overweight and obesity in 7 year old children 89 Table 3.7 Differences between number of children classified as overweight and obese using WHO BMI-for-age (using average age) vs WHO BMI z-score (using actual age) Table 4.1 DGI-CA total and component scores and percentage of children in the control and DHA groups achieving maximum component score Table 4.2 DGI-CA total and component scores and percentage of male and female children achieving maximum component score Figure 4.1 Relationship between BOD POD Fat Mass % and Total DGI-CA score Table 4.3 Treatment Group Comparison of family food environment, physical activity and screen time at 7 years of age vi P a g e

8 Table 4.4 Male vs Female Comparison of family food environment, physical activity and screen time at 7 years of age Table 5.1 Body composition outcomes Figure 5.1 Relationship between BOD POD FM% and BIS FM % at 7 years of age in male and female children Figure 5.2 Bland-Altman of mean differences between percentage body fat (BF%) in male and female 7 year olds measured with the BOD POD and BIS Figure 5.3 Bland-Altman of mean differences between percentage body fat (BF%) in female 7 year olds measured with the BOD POD and BIS Figure 5.4 Bland-Altman of mean differences between percentage body fat (BF%) in male 7 year olds measured with the BOD POD and BIS Figure 5.5 Relationship between BIS FM% and BMI at 7 years of age in male and female children Figure 5.6 Relationship between BOD POD FM% and BMI at 7 years of age in male and female children Figure 5.7 Relationship between BIS FM% and BMI z-score at 7 years of age in male and female children Figure 5.8 Relationship between BOD POD FM% and BMI z-score at 7 years of age in male and female children vii P a g e

9 ABSTRACT The first 1000 days of a child s life, from conception to their second birthday, is a critical window of development during which environmental exposures have a particularly important role in determining the future health outcomes of a child. Omega-3 long chain polyunsaturated fatty acids (LCPUFAs) are essential fatty acids that play an important role in the health of the mother and growth and development of the fetus. On the basis of the role of n-3 LCPUFAs in decreasing fat deposition in adult rodents and in vitro studies, it has been hypothesised that an increased supply of n-3 LCPUFA, in particular docosahexaenoic acid (DHA) before birth, could reduce body fat mass later in childhood. However, there is a lack of robust evidence from human studies to support this. The first aim of this thesis was to assess the impact of maternal n-3 fatty acid supplementation, chiefly as DHA, during pregnancy on the body fat mass of children at 7 years of age using Air Displacement Plethysmography (ADP). This study was a followup of the largest RCT to date to examine the effect of maternal n-3 LCPUFA supplementation on maternal/infant outcomes, the DOMInO (DHA to Optimise Mother Infant Outcome) trial in which the women were supplemented with 800mg DHA/day from 20 weeks gestation to delivery. The second aim was to undertake a comparison of the body composition results obtained using the BOD POD with Bioelectrical Impedance Spectroscopy (BIS). A total of 252 children participated in this study. During clinic appointments, body composition testing was conducted using both ADP and BIS. Anthropometric measurements including weight, height, waist and hip circumference were obtained. The child's dietary information was obtained using a Diet and Physical Activity Questionnaire viii P a g e

10 which included 116 food frequency questions. Nutritional intake was assessed using the Healthy Diet Index for children (HDI-CA). I found no effect of maternal DHA supplementation on body fat mass, assessed by either BIS or BOD POD, BMI z-score or any other anthropometric measures in the children at 7 years of age. While the results for body fat mass obtained using the BIS and BOD POD were significantly correlated, BIS consistently over estimated body fat mass in comparison with the BOD POD, especially in girls. Most children showed a poor compliance to the Australian Dietary Guidelines, and this was not affected by maternal DHA supplementation. The findings from this thesis provide new and important insights into our understanding of the relative importance of the nutritional environment before birth, in early infancy and later in childhood on body composition and the risk of obesity. The results from the comparison of body composition methods in this thesis highlights the variation that can occur between methods and therefore the importance of using the same method in studies where the aim is to directly compare results. In conclusion, high dose maternal DHA supplementation had no significant positive or negative effect on childhood growth/body composition, at least until 7 years of age, and is therefore not likely to be an effective strategy for addressing the current obesity epidemic. ix P a g e

11 DECLARATION I certify that this work contains no material which has been accepted for the award of any other degree or diploma in my name, in any university or other tertiary institution and, to the best of my knowledge and belief, contains no material previously published or written by another person, except where due reference has been made in the text. In addition, I certify that no part of this work will, in the future, be used in a submission in my name, for any other degree or diploma in any university or other tertiary institution without the prior approval of the University of Adelaide and where applicable, any partner institution responsible for the joint-award of this degree. I give consent to this copy of my thesis, when deposited in the University Library, being made available for loan and photocopying, subject to the provisions of the Copyright Act I also give permission for the digital version of my thesis to be made available on the web, via the University s digital research repository, the Library Search and also through web search engines, unless permission has been granted by the University to restrict access for a period of time... Katie Wood x P a g e

12 ACKNOWLEDGEMENTS It seems in life your direction can change at any given moment and for this reason I am grateful. In the beginning of this PhD chapter in my life I was lucky enough to have two mentors, who not only encouraged this direction in my research career but passed on their passion for what research is all about, that we can make a difference. For this, my heartfelt thanks firstly goes to Bev Muhlhausler and Evangeline Mantzioris for your never ending support throughout this process. I also wish to sincerely thank my co-supervisors Jennifer Couper and Barbara Lingwood for your involvement and feedback throughout my PhD. Bev, thank you for being the most amazing primary supervisor anyone could ask for. In times of doubt you reassured me that everything was fine. I always felt comfort in the fact you were there if I needed you. Your support throughout my PhD has been constant, your never ending feedback has helped me grow and your love of science has inspired me in so many ways to what is possible. The excitement of receiving corgi stickers has never faded! Evangeline, I thank you for your ongoing support and guidance throughout these last few years with your experience and honesty. You were always there to put me on the right track when things weren t going to plan. I loved being able to chat over a coffee, enjoy a laugh and have the opportunity to work together. You have not only been an incredible supervisor, but I value you as a mentor and a friend. To my wonderful colleagues in the FOODplus Research group at Waite. Your support and friendship throughout this time has been incredible. I cannot name you all but you know who you are. In particular, a big thanks to Yichao Huang who has been there from the beginning, we have enjoyed many conversations about life and our PhD studies and you have always had the time to help me when I needed it. Liu Ge, for teaching and xi P a g e

13 explaining to me many statistical tests along the way and for your cheeky personality. Jing Zhou, who sat beside me for many years, thank you for being a beautiful friend. John Carragher, for your honesty, advice and sense of humour. I couldn t have asked for a more fantastic group of people to work with! To the staff at WCHRI who helped me set up my clinic room, showed me how to measure children, assisted me when I needed it, your help along the way is greatly appreciated. To the DOMInO families who participated in this study, this research would not have been possible without you. To the most important of all, my friends and family. To my parents for always being there unconditionally, always believing in me and moulding me into the person I am today. To my beautiful children, Lucy and Mitchell, you are my world, and I m hoping that I have shown you that anything in life is possible. Reach for the stars! To my husband Anthony, my world and my rock. I could not have done this without you. Juggling family and study and work we did this together. You have never questioned my ability in anything, you told me I could do things when I said I couldn t possibly do them. You helped me get to the finish line in so many ways and I cannot express what that means to me. This PhD would not be a thesis without all of the support I have received over the last few years and it represents something that I never dreamt possible. This experience has been life changing, challenging, and in the end, worth every moment. xii P a g e

14 LIST OF ABBREVIATIONS 2C 3C 4C AA ADP AGHE AHS ALA BF BIA BIS BMC BMI CDC CHD CNRC CT CV Db DGI-CA DHA DOMInO DVD DXA EAPM EPA ESPEN FFM FFQ FM FMC FSANZ Two compartment model Three compartment model Four compartment model Arachidonic acid Air Displacement Plethysmography Australian Guide to Healthy Eating Australian Health Survey Alpha Linolenic Acid Body Fat Bioelectrical Impedance Analysis Bioelectric Impedance Spectroscopy Bone Mineral Content Body Mass Index Centers for Disease Control and Prevention Coronary Heart Disease Child Nutrition Research Centre Computed Tomography Coefficient of Variation Body Density Dietary Guideline Index for Children and Adolescents Docosahexaenoic acid DHA for Optimising Mother and Infant Outcomes Digital Video Disc Dual energy x-ray absorptiometry European Association of Perinatal Medicine Eicosapentaenoic acid European Society for Clinical Nutrition Fat Free Mass Food Frequency Questionnaire Fat Mass Flinders Medical Centre Food Standards Australia New Zealand xiii P a g e

15 HDI Healthy Diet Index HDL High-density lipoprotein IOTF The International Obesity Taskforce IQR Interquartile Range ISSFAL International Society for the study of Fatty Acids and Lipids LA Linoleic acid LCPUFA Long Chain Polyunsaturated Fatty Acids LDL Low-density lipoprotein MRI Magnetic Resonance Imaging N-3 Omega 3 N-6 Omega 6 NEAF National Ethics Application Form NHMRC National Medical and Research Council NNPAS National Nutrition and Physical Activity Survey NNS National Nutrition Survey PUFA Polyunsaturated Fatty Acids RCT Randomised Controlled Trial SAA Surface Area Artefact SD Standard Deviation SFT Skinfold Thickness SSA Site Specific Application TBW Total Body Water TGV/VTG Thoracic gas volume Vbraw WCH WCHN WHO WtHR Raw Body Volume Women s and Children s Hospital Women s and Children s Health Network World Health Organisation Waist:Height ratio xiv P a g e

16 CHAPTER 1 Introduction and Literature Review 1 P a g e

17 CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW The worldwide incidence of childhood obesity has doubled in the last 30 years, with 42 million children under the age of five being classified as overweight or obese in 2013 [1]. In 2015, the World Health Organisation (WHO) identified childhood obesity as one of the most serious public health challenges of the 21st century [2]. In Australia, 1 in 4 children aged between 5 and 17 years are classified as overweight or obese. While there are some suggestions that rates have plateaued since , the proportion remains high, and is still increasing in lower SES groups. Thus, the proportion of Australian children who are classified as overweight or obese remains a significant health issue [3]. Overweight and obese children typically exhibit a number of established risk factors for coronary heart disease (CHD), including hyperlipidaemia, impaired glucose tolerance, hypertension and vascular abnormalities [4, 5], and the vast majority (~85%) of these children remain overweight or obese as adolescents and adults [6]. It is widely recognised that childhood obesity is a multi-faceted issue. Two factors at the core of its development and implicated as major targets for its prevention are a child s diet and their level of physical activity. Consumption of diets containing excess energy or low nutritional quality combined with low levels of physical activity results in weight gain and, ultimately, overweight and obesity. However, health strategies to avoid excessive energy intake and to engage in regular physical activity, in order to avoid excessive weight gain, have thus far been ineffective in curbing the obesity epidemic [1, 7]. This may be due in part to the finding that the susceptibility to obesity is programmed very early in life. 2 P a g e

18 A world-wide series of epidemiological and clinical studies [8-11] have established that nutrition during the first 1000 days of life (from conception to a child s second birthday) plays a critical role in determining the health of an individual through the life course. This includes the individual s body weight and body composition, and thus their susceptibility to obesity and poor metabolic health. This implies that it is likely to be important to apply nutritional interventions early, rather than later in life, in order to be most effective at overcoming the increasing rates of childhood obesity. The omega 3 (n-3) long chain polyunsaturated fatty acids (LCPUFA) are widely utilised as a nutritional supplement with reported benefits for a number of health conditions including cardiovascular disease, rheumatoid arthritis, asthma and inflammatory bowel disease [12]. N-3 LCPUFA supplements are consumed by over 1/3 of women during pregnancy due to the established importance of an adequate supply of these fatty acids for the optimum growth, physical and neurological development of the fetus [13, 14]. In addition to the established benefits of n-3 LCPUFA for infant/child development, there is growing interest in the potential for exposure to an increased supply of n-3 LCPUFA before birth to influence the long-term metabolic health of the child, in particular their body fat mass [15]. However, the number of studies that have evaluated the relationship between maternal n-3 LCPUFA intake and body composition in the child is limited, and results have been inconsistent. Furthermore, many existing studies have been of low quality, including small sample sizes and high rates of attrition [16]. The existing studies therefore do not provide robust evidence as to the potential effect of maternal n-3 LCPUFA supplementation on the body fat mass of the child. 3 P a g e

19 Accurately measuring the body composition of children is central to testing this hypothesis. A further limitation of existing studies in the area is their use of only indirect methods for estimating child body composition. Assessing the body composition of children to determine percentage fat mass requires techniques that are non-invasive, but also quick and accurate. While there are a number of established methods used for assessing body composition in paediatric populations, these vary in their accuracy and appropriateness. Disagreement exists regarding the best approach for assessing body composition in children. Air displacement plethysmography (ADP) assessed using the BOD POD system is one of the newest of these methods and is rapidly becoming the method of choice for assessing body composition in both children and adults [17]. However, no previous studies of prenatal n-3 LCPUFA supplementation have utilised this approach for assessing body fat of children. This literature review will first outline and evaluate the current evidence relating to the possible link between maternal n-3 LCPUFA supplementation during pregnancy and body composition, in particular the percentage of body fat mass, in the children. The role of diet and physical activity as contributing factors to childhood obesity is also discussed. The second half of this chapter describes the approaches currently being used in clinical and research settings to assess body composition in children, with particular emphasis on the initial studies using air displacement plethysmography (ADP) and the BOD POD. Finally, the aims of this thesis, and each of its constituent Chapters, will be introduced. 4 P a g e

20 PART 1 CONTRIBUTORS TO CHILDHOOD OBESITY AND THE ROLE OF N-3 LCPUFA 1.1 The Early Origins of Obesity The nutritional environment experienced by the fetus and neonate plays an important role in the developmental and later health outcomes of the offspring, including the risk of obesity [8, 10]. The developmental origins of health and disease (DOHaD) hypothesis states that an increased risk of overweight/obesity later in life originates from environmental exposures in fetal life and early infancy, and highlights the importance of the early nutritional environment for the programming of obesity and metabolic disease [8]. The importance of an increased nutrient supply for developmental programming was first demonstrated by studies in infants of diabetic mothers. Pregnant women who have been unable to control their gestational diabetes have excess glucose concentrations in their blood and typically give birth to heavier babies [18, 19]. Importantly, these infants have a greater neonatal fat mass than infants of the same birthweight born to mothers without gestational diabetes [20]. Increased placental and somatic growth, particularly fetal adipose deposition, is also seen in infants born to overweight or obese women, and this is suggested to be a result of exposure to an increased supply of nutrients/substrates to the fetus in utero [19, 20]. More recent studies have focussed on determining the mechanisms underlying the developmental origins of obesity and have identified the developing fat cell, or adipocyte, as an important target. These studies have demonstrated that exposure to an excessive or deficient nutrient supply before birth alters the development of the fat cell, or adipocyte, 5 P a g e

21 and results in a permanent increase in the capacity to form new cells in adipose depots (adipogenesis) or to store lipid in existing adipocytes (lipogenesis) [9, 18]. The process of adipocyte development in humans occurs primarily in the late stages of the last trimester of pregnancy and in early post-natal life, and it has been demonstrated that adipogenesis is highly sensitive to the nutritional environment experienced during this time [9]. For example, it has been shown in both human and experimental animal studies that adipocytes exposed to high levels of glucose during their major period of development had an increased capacity to store lipids in postnatal life [18, 21]. Consequently, exposure of the adipocyte (fat cell) to excess glucose/nutrient availability during critical periods of development can permanently increase the capacity of adipose cells for lipid storage. This phenomenon has been suggested to explain, at least in part, why infants born to overweight/obese and diabetic mothers are at an increased risk of obesity later in life [8, 11, 22]. 1.2 N-3 LCPUFA in Perinatal Nutrition The omega 3 (n-3) and omega 6 (n-6) polyunsaturated fatty acids (PUFA), represent two of the most biologically active classes of fatty acids in the human diet. Both n-3 and n-6 PUFA are essential components of human diets, meaning that humans are not able to synthesise them de novo and must obtain them solely from the diet [23]. The key members of the n-3 fatty acid family are the plant-derived α-linolenic (ALA 18:3-n-3) and n-3 LCPUFAs eicosapentaenoic acid (EPA 20:5n-3) and docosahexaenoic acid (DHA 22:6n- 3) [24]. EPA and DHA are mainly consumed pre-formed as either fish/seafood or fish-oil supplements. However, they can also be synthesized de novo through conversion of the 6 P a g e

22 plant-derived short-chain n-3 PUFA precursor, ALA [25]. However, the efficiency of this process in adult humans is generally low, with in vivo studies in humans reporting that only ~5% of ALA is converted to EPA and <0.05% is further elongated to DHA [26, 27]. The main dietary n-6 PUFA are linoleic acid (LA 18:2n-6) and arachidonic acid (AA 20:4n-6). LA is found at high levels in the majority of vegetable oils and spreads and in most processed foods, whilst AA is found predominantly in animal products [24]. EPA and DHA have been been shown to have a number of important health benefits in humans, particularly for protection against cardiovascular disease, cancer and inflammatory and autoimmune diseases [23, 28]. It is also important to note, however, that recent studies and meta analyses have presented less clear conclusions regarding the relationship between increased n-3 LCPUFA intakes and protection against cardiovascular disease [29, 30]. A recent systematic review conducted on behalf of the National Heart Foundation of Australia found a positive role for n-3 LCPUFA in the treatment of hypertriglyceridaemia but no significant effect (beneficial or otherwise) of n-3 LCPUFA supplementation in primary or secondary prevention of coronary heart disease in adults [31]. Thus, while n-3 LCPUFA undoubtedly have physiological properties which are cardioprotective, their role in cardiovascular health in adult humans may be less important than previously thought. Nutrition during pregnancy is an important determinant of health of the mother and growth and development of the fetus, and an adequate supply of n-3 LCPUFAs during this time is important for maternal and infant outcomes [32]. During the prenatal period, the only source of LCPUFAs for the fetus is from the mother via transfer across the placenta [33]. Maintaining an adequate supply of n-3 LCPUFA (especially DHA) to the 7 P a g e

23 fetus is particularly critical during the last trimester, which represents the period of most rapid development of the central nervous system, including the brain and retina, and deposition of bone and adipose tissue [33, 34]. The demand for DHA by the fetus is highest in the last five weeks of pregnancy [35], with an intrauterine accretion rate of up to 75 mg DHA per day during this time [36]. Maternal dietary intake and adipose tissue stores of DHA are the sole source of DHA for the fetus and the breastfed infant [32]. An intake of at least 200mg/day DHA during pregnancy and lactation has been recommended by the Perinatal Lipid Intake Working Group which included representation from several key international health bodies including the Child Health Foundation, International Society for the study of Fatty Acids and Lipids (ISSFAL), European Society for Clinical Nutrition (ESPEN) and the European Association of Perinatal Medicine (EAPM) [13, 32]. The majority of women aged years (which is likely to include pregnant and lactating women) in Australia are not meeting the recommended dietary intake for n-3 LCPUFA DHA with an average intake of 51mg/day DHA, equating to only 10% of this population meeting requirements [37]. The guidelines set by Food Standards Australia New Zealand (FSANZ) surrounding fish intake during pregnancy and avoidance or limitation of certain types of fish due to the risk of mercury toxicity also contributes to lower fish intakes in pregnant women, with many preferring to avoid fish/seafood altogether [38]. 8 P a g e

24 1.3 Perinatal N-3 LCPUFA and Body Composition Animal Models and in vitro studies Research into the impact of maternal dietary n-3 LCPUFA supplementation on maternal/infant outcomes has previously concentrated primarily on outcomes associated with neurodevelopment, visual acuity, cognitive function and gestation length [14]. Consequently, there is currently limited information on the effects of dietary n-3 LCPUFA supplementation during pregnancy on body composition in children. Evidence for a role of n-3 LCPUFA in decreasing adipogenesis and lipogenesis was first provided by in vitro studies and studies in adult rodents. In vitro studies showed that the n-3 LCPUFA, in particular DHA, acted to suppress adipocyte differentiation [39] and showed hormone-like effects in regulating expression of genes in preadipocytes which play a role in subsequent adipocyte proliferation and differentiation [40]. Animal studies also showed that incorporating n-3 LCPUFA into obesogenic high fat diets resulted in a reduced accumulation of body fat, particularly visceral fat, in comparison with animals fed the same obesogenic diet without added n-3 LCPUFA [41]. Modulation of gene expression in adipocytes, improved insulin sensitivity and reduced accumulation of body fat was also reported in 3-4 month old obesity-prone mice fed a high fat diet (20-35% w/w) containing varying levels of EPA and DHA (1-12% EPA + DHA w/w) compared with a low-fat control diet. Importantly, the authors attributed this effect predominantly to DHA, due to the low EPA to DHA ratio of the diet fed to the rats in this study [42]. Similar findings have been reported in a study of adult rats fed isocaloric diets containing 20% triglycerides with different fatty acid compositions. This study reported that the group allocated to the fish oil (n-3 LCPUFA) diet had a smaller adipose cell size and 9 P a g e

25 adipose mass after 3 weeks compared to rats fed a saturated fat-based diet, and that this was due largely to decreased triglyceride storage in adipose depots [43]. In contrast to the relatively strong and consistent evidence provided from adult rat studies showing the anti-adipogenic and anti-lipogenic effect of n-3 LCPUFA, the effect of increased n-3 LCPUFA exposure during the period of adipocyte development on subsequent fat deposition in the offspring is less clear. There are currently only 2 animal studies in which exposure to an increased supply of n-3 LCPUFA has been confined to the fetal and suckling period, and these studies have produced conflicting results [44, 45]. In one study, Korotkova et al. [44] found that pups born to dams fed a high ALA diet for the last 10 days of gestation and during lactation had a lower body weight and percentage body fat mass compared to pups of dams fed a standard rodent diet. In contrast, Muhlhausler et al. [45] reported that both male and female offspring of dams fed a diet supplemented with n-3 LCPUFA during pregnancy and lactation had a higher percentage of body fat at 6 weeks of age, although there were no differences in the percentage body fat in the offspring at 3 months of age in this same study. In summary, the results from research in vitro and in adult rodents into the effect of n-3 LCPUFA on modulating lipid synthesis/storage in adipose cells is encouraging in supporting their potential role in suppressing fat deposition and thus reducing fat mass. However, the effect of increased perinatal n-3 LCPUFA exposure on subsequent fat mass in the offspring from the limited studies conducted is inconclusive and further studies are required to obtain more robust data. 10 P a g e

26 1.4 Maternal n-3 LCPUFA supplementation and infant body composition: Evidence from human studies To date, 8 human studies, including 2 epidemiological studies and 6 randomised control trials [34, 46-51] have studied the association between maternal n-3 LCPUFA intake/status and infant/child fat mass. The epidemiological studies in this area have generally supported the existence of an inverse relationship between maternal n-3 LCPUFA intake/status and body composition in childhood. Donahue et al. reported that higher mid and late pregnancy n-3 LCPUFA intakes and higher n-3 LCPUFA concentrations in umbilical cord plasma phospholipids were both associated with a lower percentage body fat mass, as assessed by skinfold thicknesses and BMI percentile, in children at 3 years of age [52]. A prospective cohort study conducted by Moon et al. found no relationship between maternal n-3 LCPUFA and fat mass in their children at either 4 or 6 years, but did identify a positive correlation with children s lean mass [53]. More recently, a large prospective cohort study including 4830 mothers and their children reported that higher maternal n-3 PUFA concentrations (EPA, DPA and DHA) and lower n-6 PUFA concentrations during pregnancy (taken at median gestation 20.5 weeks) were associated with a lower fat mass % as assessed by DXA at 6 years of age [54]. While epidemiological studies can provide an indication of possible associations, these are unable to establish causality. Therefore, data from randomised controlled trials is essential for determining if there is in fact a cause and effect relationship between increased perinatal n-3 LCPUFA exposure and reduced body fat mass later in childhood [55]. Five out of the six randomised controlled trials conducted to date that have assessed the relationship between perinatal n-3 LCPUFA exposure and childhood body composition in term infants showed no differences in body composition measurements 11 P a g e

27 between the intervention and control groups. In addition, studies that reported body composition at 2 time points have all shown that effects observed at younger ages were no longer present later in childhood [34, 56, 57]. A study conducted by Bergman et al. found children whose mothers were supplemented with 200mg/day DHA from weeks gestation and until 3 months after delivery had a lower BMI at 21 months of age. However the differences between groups was no longer present at 6 years of age in the same study population [34]. A Danish study in which women were supplemented with n- 3 LCPUFA during the first 4 months of lactation reported that BMI, weight and head circumference were increased in the fish oil supplemented group compared to the placebo group at 2.5 years, but these differences were no longer present at 7 years [56, 57]. Most of the studies conducted in this area to date have used only basic anthropometric measurements, mostly BMI, as a proxy measure of body fat mass/obesity, which may not accurately reflect the balance of fat/fat free mass. A study by Hauner and colleagues [47] is the only one to date to use a direct measure of body fat mass, and used ultrasonography to determine abdominal subcutaneous and pre-peritoneal fat depth; an indicator of visceral fat deposition [58]. The study reported that thickness of abdominal subcutaneous and preperitoneal fat depots at 3 5 d, 6 wk, 4 and 12 months postpartum (n=188) were comparable between treatment groups and concluded that increased maternal n-3 LCPUFA, even when coupled with a reduction in dietary AA, does not influence offspring adipose tissue deposition across the first year of life. It is also important to note that this trial had a longer duration of n-3 LCPUFA supplementation (15 weeks gestation 4 months post-partum) in comparison with other trials and participants received a very high dose of n-3 LCPUFA per day (1020 mg DHA and 180 mg EPA). This is also the only study that also aimed to reduce the dietary intake of n-6 AA in conjunction with n- 12 P a g e

28 3 LCPUFA supplementation, due to AA s association with increased rates of adipogenesis [47]. Intervention studies in pre-term infants also give us valuable information regarding the potential effect of increased n-3 LCPUFA exposure in early life on growth rate/body composition. A randomised controlled trial by Groh-Wargo et al. [59] found that preterm infants (born <33 weeks) fed a formula fortified with DHA + AA until 1 year gestationcorrected age did not differ to the control group in terms of overall weight, but had a significantly higher percent lean body mass and reduced percent fat mass as determined by DXA measurements at 35 weeks, 40 weeks, 4 and 12 months [59]. Fat mass is usually gained faster in formula fed pre-term infants in the early post-natal period in comparison with formula-fed babies born at term, and these results therefore raise the possibility that increasing the proportion of energy/fat from n-3 LCPUFA DHA during periods of rapid fat deposition could potentially limit the expansion of fat depots in the fetus/infant. The randomised controlled trials that have provided maternal DHA supplementation during pregnancy have varied in dosage, duration and form in which the supplementation was provided. In the 6 studies conducted to date, DHA dosage ranged from 200mg mg/day. This wide variation in dosage may have contributed to variability in study findings and makes it challenging to combine results from different studies to reach an overall conclusion. Participants were typically supplemented in the form of capsules with the exception of two trials conducted by Lauritzen et al. [57] and Helland et al [48]. The study by Lauritzen et al. provided the DHA/placebo in the form of fortified muesli bars or cookies (containing olive oil or fish oil) [57], while cod liver oil or corn oil (placebo) was provided in the study by Helland et al. [48]. The lowest supplement dose of n-3 13 P a g e

29 LCPUFA DHA (200mg/day) given from 21 weeks gestation to 3 months post-partum, was used in the study by Bergmann and colleagues [34]. These researchers found a lower BMI and weight at the age of 21 months in the DHA supplemented group but no significant differences between the DHA and placebo groups in length, weight, head circumference and skin-fold thickness at the age of 6 years. The trial with the highest dose of DHA supplementation (1183mg/day), conducted by Helland and colleagues, showed no relationship between DHA supplementation and BMI at 7 years of age [48]. The wide variation in n-3 LCPUFA supplementation dosages used in different trials, provides no indication of a dose-response relationship between n-3 LCPUFA exposure in the perinatal period and fat mass in childhood. Despite some positive findings, there is currently limited evidence that n-3 LCPUFA exposure before birth and/or in early infancy has any influence on body composition later in childhood. In addition, no studies to date have demonstrated any effect of n-3 LCPUFA supplementation during pregnancy on body composition in children that persist beyond 2 years of age. However, there are potential issues with the quality of existing studies in this area including modest sample sizes, timing of exposure and high attrition rates (32.5% %) [60]. The lack of direct measures of body composition, and reliance on a single indirect surrogate measure (eg. BMI z-score, weight, waist circumference) is also a limitation of existing studies in this area. Taken together, these factors make it difficult to draw robust conclusions as to the effect of increased perinatal n-3 LCPUFA exposure on body composition later in childhood. The number of studies looking specifically at body composition of children following maternal n-3 LCPUFA supplementation is clearly limited and there is inconsistency in 14 P a g e

30 the findings. Thus, larger studies using more accurate testing methods are needed to better understand the potential role of n-3 LCPUFA in growth and growth quality in children and to definitively answer the question of whether n-3 LCPUFA supplementation has the potential to decrease fat accumulation in early childhood. 15 P a g e

31 1.5 Other Contributors to Childhood Obesity Dietary intake There are multiple factors that contribute to childhood obesity [7]. The obesogenic environment, defined as the sum of influences that the surroundings, opportunities or conditions of life have on promoting obesity in individuals or populations [61] includes physical, economic, policy and socio-cultural environments. Included in the physical environment are diet/food choices and physical activity [62]. Within the food environment, social interactions between family and friends plays an important role in shaping the dietary choices made for and by the child. In young children parents/primary carers will be a major influence on the child s eating habits and patterns. Previous reviews have found consistent associations between parental influences (parental food intake and education) and the risk of obesity in children aged 3-18 years [63]. As they grow into adolescents, peer pressure becomes more important, and parental influence is reduced [64]. Therefore, early childhood is seen as the most appropriate period to implement obesity prevention programs centered on improved diet/lifestyle choices and when such interventions are likely to have maximum efficacy [65]. There are many potential environments/settings in which individuals interact, including schools, homes and neighbourhoods. These micro environments are influenced by the macro environments in which they belong eg. health and education systems, food industry and government [61]. Studies investigating the influence of the food environment on overweight and obesity in young children have generally concluded that a reduction in food promotion to young people, providing alternatives to sugar-sweetened soft drinks 16 P a g e

32 and increasing the availability of smaller portions may assist in future obesity prevention [66]. Poor dietary quality has been known for many years to play a key role as a risk factor for chronic diseases [67]. The fundamental cause of childhood overweight and obesity is an imbalance between calories consumed and calories expended which has a direct influence on body weight [1]. The importance of dietary intake for the nutritional status of a child is also well recognised [68]. Both globally, and in Australia, there has been a shift in the composition of the typical western diet since the second half of the twentieth century from a mainly plant based diet to diets high in energy-dense nutrient-poor foods, and corresponding changes in the structure of children s typical dietary patterns has been seen worldwide [69]. In addition, changes in dietary patterns over this time have seen an increased dietary intake of n-6 PUFA with a subsequent decrease in the intakes of n-3 LCPUFA. This change in the ratio of n-6 to n-3 PUFA, which is now heavily dominated by n-6 PUFA, is a growing concern for future health outcomes, and has been suggested to be one potentially important contributor to the increased incidence of obesity [70, 71]. The composition of the current western diet is characterised by higher amounts of processed foods, reduced intakes of complex carbohydrates and dietary fibre, and reduced fruit and vegetable intake in comparison to traditional diets. Overall, this has resulted in a reduction in the nutritional quality of the diet, and in the dietary content of vitamins, minerals and phytonutrients [4, 67]. Scientific opinion as to which macronutrient is the main driver of obesity is heavily debated. While fat was heavily implicated as a major driver in early studies, studies looking at the relationship between dietary fat and adiposity have not been consistent in 17 P a g e

33 their findings/conclusions [72-74]. There is also limited published data on children s dietary intake in the context of obesity interventions [75]. The studies that have been undertaken have predominantly investigated the potential influences of fat and carbohydrates, energy density, portion size, fast food and the consumption of sugarsweetened soft drinks on obesity risk [74]. The majority of fast foods are high in energy and low nutrient dense, contain high levels of fat, particularly saturated fats, have a high glycaemic index and are usually served in a large portion [74]. Results from studies have shown an association between consumption of fast food and excessive energy intake and excess weight gain in adolescents and adults [76, 77]. Similarly, a prospective observational study conducted in 548 ethnically diverse American middle school aged children from 1995 to 1997 (mean age 11.7 years) reported that the risk of obesity increased by 60% for each additional serving of sugar-sweetened beverage consumed during the follow-up period [78]. A 5 year follow-up of a longitudinal prospective cohort study of 281 Australian children (aged 13 years) also found that the intake of soft drink/cordial, but not fruit juice/fruit drinks or milk at 8 years was associated with increased weight gain 5 years later [79]. There is no doubt that the diet of a child plays a critical role in the development and prevention of obesity. However, it is only one factor amongst many in this multi-faceted obesity crisis. 18 P a g e

34 1.5.2 Methods of measuring dietary intake in children Measuring dietary intake accurately is important in both the clinical and research settings. Measuring dietary intake comes with its challenges and all methods have their own strengths and limitations. Under and over reporting of food consumed is a common contributor to error in assessing both adult and children s diet [80]. Assessing the dietary intake of children is understandably more difficult than in adults due to factors such as the need to rely on parental recall for younger children, day to day variation in appetite/food habits, and limited ability of younger children to recall details of foods eaten throughout the day and the quantity/portion size of the food consumed in comparison to adolescents and adults[81]. The reliance on parental reporting of a child s diet is not always accurate, especially if the child is cared for outside of the home. The most frequently used methods are 24 hour recall (administered to either the child or the parent), food frequency questionnaires (FFQ) and food records (weighed or estimated) usually conducted over 3, 5 or 7 days [82]. Pictures of food and portion sizes are commonly incorporated in methods used for children to assist in increasing the accuracy of quantifying amounts of food and drink consumed [82]. The age of the child is a major consideration when deciding on the method to use, with children from the age of 8 generally considered to be able to recall details sufficiently to be able to self-report food intake, while younger children require parental/carer input [80]. Dietary surveys that rely on recall such as the 24 hour recall and FFQ rely on memory, which can inevitably lead to an increase in error. However, these methods are less time consuming and therefore less burden is placed upon the child/parent in comparison to weighed food diaries. Validation studies suggest that 24 hour recalls and weighed or estimated food records provide more accurate dietary intake estimates of children in 19 P a g e

35 younger age groups than FFQs [80]. However, FFQs have been used much more widely to assess food intake in children, largely due to logistical issues. A number of FFQs have been validated in children [82], but there are an even greater number of FFQs in use that have not been validated, or for which details of validation are not provided. Validation of a dietary assessment method such as the FFQ is completed comparing it with another method eg. weighed food record over the same period of time as the FFQ is quantifying eg.7 days. Other methods of validation include doubly labelled water and micronutrient concentrations [82] Currently, the Australian Child and Adolescent Eating Survey is the first specific FFQ validated and available for ranking the dietary intakes of children aged 9-16 years [83]. Validated questionnaires for use in younger children in Australia and worldwide are limited. A 56 item FFQ based on the 1995 Australian National Nutrition Survey was designed and validated for use in 5-8 year old children and was later modified for use by our research group at 3, 5 and 7 years [84]. A recent study in 2-5 year old Australian children validated a short 17 question FFQ which provided reliable and moderately valid information regarding the diets of this age group [85]. The use of technology in the form of mobile apps and websites is rapidly replacing the use of the traditional paper based management of dietary surveys. The latest technology includes interactive computer-based technologies, Personal Digital Assistants (PDAs), web-based technologies, mobile devices, specialised cameras and tape recorders, and scan and sensor technologies [86]. Self-administered computerised assessments, which can include audio support, may assist in overcoming literacy problems, be translated and are useful for younger age groups, but less so for those unfamiliar with computers. Selfadministered 24-hour questionnaires utilising computers have been shown to provide comparable data to paper-based approaches, but it is important to note that all of these 20 P a g e

36 methods require supervision if children are involved. Nevertheless, these methods show promise in improving accuracy of estimating portion sizes which children generally find difficult. The increased use of technology also decreases the workload for researchers with less one-on-one assistance required [87]. Selecting the most appropriate diet recording method in studies with children requires consideration of the objectives of the study, size of the cohort and age of the children with the choice aimed to obtain the highest quality diet data possible Physical Activity and Screen Time Energy expenditure through physical activity is an important part of the energy balance equation that determines body weight. Sedentary behavior ie. very low levels of physical activity, is a potential risk factor for the development of obesity and cardiometabolic diseases in children [88]. Children who are overweight or obese are significantly more sedentary overall and report higher amounts of screen time than children of normal weight [89]. Conversely, in a study of 709 children aged 7 to 12 years in the United States of America children not meeting physical activity and screen time recommendations were found to be three to four times more likely to be overweight compared with children who met the recommendations [90]. Screen time includes television viewing, watching DVDs and the use of computer and video games. All forms of screen time are seen to have an impact on a child s activity level. Television, especially commercial TV, is often viewed as the most detrimental. This is because, in addition to decreased physical activity it also increases the exposure of a child to food marketing and promotes consumption of unhealthy snack foods [91]. The clustering of obesity-related behaviours, including television viewing and food 21 P a g e

37 consumption, has also been confirmed in Australian studies in children aged 5 to 18 years old [92]. The increased food consumption associated with these activities is one of the potential mechanisms linking lower physical activity/sedentary behavior to obesity [93, 94]. The majority of studies in this area, but not all, have found a positive association between time engaged in television viewing and obesity risk/bmi [95]. Results from a 3 year longitudinal Australian study of children aged 5-6 (n=87) and (n=123) identified the combination of greater amounts of television viewing and increased consumption of energy dense food across this period was the strongest predictor of the likelihood of overweight/obesity in the child at the end of the 3 year follow up (odds ratio=2.8; 95% confidence interval: 1.1, 6.9; P<0.05) [92]. Results from a large study (n=7792) of boys and girls from the USA aged 9 to 16 years reported a 0.09 increase in BMI (P<0.001) for each hour per day increase in reported television viewing [96]. The combination of low amounts of moderate to vigorous physical activity, high amounts of television viewing and short sleep duration were also found to be significant predictors of obesity in a multinational cohort including 9 to 11 year old children from 12 countries [93]. Another component of screen time is the use of computer/video games. Computer games can either be classified as sedentary or active. In the case of active games, there is the potential for positive influences on activity levels and therefore a reduction in obesity risk. Food consumption is also lower for children engaged in active video games [91, 97]. A recent systematic review and meta-analysis including 109 studies with children found ccomputer use was not related to BMI. This is potentially explained by the fact that less is known about the association of computer games with eating behaviours and that, depending on the age of the child, television viewing takes up more time than playing 22 P a g e

38 computer games during the day. In addition, the majority of studies in the review did not separate active and passive gaming, which clearly have different impacts on overall activity levels/obesity risk [98]. Overall, there has been a small but steady increase in participation of children in organised sport in Australia since 2000 [99]. However, recent data reports only 60% of children on average are meeting the Australian Children s Physical Activity and Sedentary Behaviour Guideline recommendation of engaging in at least one hour of moderate to vigorous physical activity per day [100]. Coupled with this, 70% of children in Australia have been reported to exceed the recommended level of screen time of no greater than 2 hours per day [101]. Findings from a recent systematic review and meta-analysis reported greater improvements in cholesterol, fasting glucose fasting insulin measures when dietary interventions in children and adolescents aged up to 18 years were combined with an increase in exercise/activity levels [102]. Thus, the available data would clearly suggest that being physically active is a protective factor against excess weight gain/fat deposition, and reduced levels of physical activity are likely to be a key factor contributing to the current epidemic of childhood obesity. In addition, it is important to acknowledge that additional factors, including genetics, poor quality/lack of sleep and imbalances in the gut microbiome, that were not measured in this study have also been implicated as contributing factors to the risk of obesity in children [103]. Many studies have found an association of shorter sleep duration and chronic sleep deprivation with an increased risk of obesity in children [104, 105]. In addition, overweight and obesity in children has been correlated with dysregulation of the immune system, stress axis, and gastrointestinal barrier function, all of which are 23 P a g e

39 associated with alterations in the composition of the gut microbiota [106]. Importantly, several recent studies have revealed a possible causative relationship between dysregulation of the gastrointestinal microbiota and the development of obesity in both children and adults [107]. 24 P a g e

40 PART 2 ASSESSMENT OF BODY COMPOSITION IN CHILDREN 1.6 Defining overweight and obesity in children: the use of Body Mass Index (BMI) Overweight and obesity is currently defined predominately on the basis of Body Mass Index (BMI), calculated as weight (in kg) divided by height (in metres) squared (kg/m 2 ), in both children and adults. Overweight and obesity are determined by expressing a child s BMI as a z-score or converting it to an adult equivalent BMI value. In both cases, this is achieved by comparing their BMI to a reference BMI of children the same age and sex. These reference BMI values are derived from standardised BMI charts for children [108]. The first of these BMI classification charts for children was published in the early 1980s [109]. Since this time, various recommendations have been put forward regarding the most appropriate way to use BMI to categorise children as overweight or obese. Typically children whose BMI lies above the 85 th and 95 th percentiles (z-scores of >1 and >2 SDs above the mean respectively) are classified as overweight and obese [110]. The three most commonly used classification systems are those produced by the WHO [111], the US Centers for Disease Control and Prevention (CDC) and The International Obesity Taskforce (IOTF) [5]. The use of the WHO or CDC guidelines are recommended by the National Medical and Research Council, Australia. The Australian Department of Health recommends CDC for children and adolescents aged between 2 and 18 years who have a BMI greater than the 85 th percentile and The World Health Organization (WHO) percentile charts are used for infants and children under 2 years of age [112]. 25 P a g e

41 The CDC growth charts represent BMI as sex specific BMI-for-age and z scores (SD scores) allowing any child from the age of 2 to 20 years to be compared to others of the same age and sex [113, 114]. The 85 th and 95 th percentiles of BMI-for-age are used in the CDC growth charts with a BMI greater than the 95 th percentile being classified as obese and children between the 85 th and 94 th percentile being overweight [110]. The IOTF classification system for overweight (IOTF-25) and obese (IOTF-30) is based on the analysis of a large data set including data from children from six countries (Brazil, Great Britain, Hong Kong, the Netherlands, Singapore and the United States) [115]. Agespecific BMI cut offs were identified that were equivalent to adult BMI cut off points [115], in this case centiles corresponding to a BMI of 25 kg/m 2 (overweight) and 30kg/m 2 (obese) at the age of 18 years [116]. The new WHO guidelines, released in 2006, were developed using longitudinal (birth to 24 months) and cross-sectional (ages 18 to 71 months) data collected from 6669 healthy breast fed children from six sites in Brazil, Ghana, India, Norway, Oman and the USA [111]. Following this, in 2007, the 1977 National Centre for Health Statistics (NCHS)/WHO growth reference charts for children aged 5 to 19 years were reconstructed using the original NCHS sample, which consisted of three pooled data sets from the USA [117]. 26 P a g e

42 1.6.1 Limitations of BMI for defining overweight and obesity BMI is a simple, convenient and non-invasive measurement, and for this reason has been widely used for defining overweight and obesity in children, both clinically and at a population level. However, while BMI provides an indication of the ratio of weight to height, and therefore body proportions, it provides very limited information on body composition, i.e. the balance of fat and fat free mass. The limitations of using BMI to predict fat mass have been well-documented in adults [118, 119], and a number of studies which have measured fat mass in children have reported that BMI z-scores are even more poorly correlated with body fatness in children [ ]. This is thought to be due to the rapid changes in the normal balance of fat and fat free mass, density of muscle and bone and hydration status which occur during childhood, and which make it difficult to ascribe increases in BMI to increases in fat mass as distinct from lean tissue and bone mass [117, 120]. In addition, there is evidence supporting that the interpretation of BMI values may differ between different racial and ethnic groups and that health risks associated with certain levels of BMI eg. overweight or obese may vary depending on ethnicity [123]. Differences between individual children in the timing of the adiposity rebound, which marks the point at which body fat mass (and BMI) begins to increase steadily following an initial decrease across infancy and early childhood [ ], also complicates the interpretation of BMI. While the adiposity rebound typically occurs at around 5 to 7 years of age, the timing varies significantly between individuals, and epidemiological studies have suggested that an earlier onset is a risk factor for the later development of obesity [124, 125]. 27 P a g e

43 The relationship between BMI z-scores and fat mass is also not consistent across the fat mass range, with the association becoming stronger at the higher end of the fat mass scale [122]. In addition to the importance of fat mass, it is becoming increasingly clear that the distribution of body fat is an equally, if not more important, indicator of metabolic health in children as well as adults [127, 128]. Thus, the excess accumulation of fat in the visceral (abdominal) compartment is more detrimental to cardiometabolic health than accumulation of the equivalent amount of adipose tissue in peripheral or subcutaneous depots [129, 130]. 28 P a g e

44 1.7 Current approaches for measuring body Composition in children Challenges in measuring body composition The limitations of using BMI to determine body composition/fat mass in children has led to the search for appropriate methods for measuring these parameters in the paediatric population. As with BMI, the measurement of body composition in children presents additional challenges in comparison to adults. A number of approaches for assessing body composition in children are currently used in both clinical and research settings, including skinfold thickness (SFT), dual energy x-ray absorptiometry (DXA), bioelectric impedance spectroscopy (BIA/BIS) and air-displacement plethysmography (ADP). However, the reliability, accuracy and suitability of these methods for assessing fat mass and fat free mass in children varies considerably [131, 132]. Firstly, these methods measure body composition indirectly with the dissection of cadavers being the only direct method available. Secondly, there are practical difficulties associated with using gold standard techniques such as hydrostatic weighing and DXA in children. However, the more significant challenges relate to the changes in bone density, hydration status and body proportions which occur across childhood and which violate many of the key assumptions in equations used to convert the raw measurement data obtained into fat mass and fat free mass [133, 134]. While attempts have been made to overcome this, by applying age and sex-specific correction factors to adult equations, there is nevertheless a lack of appropriately validated paediatric equations, and this represents a major barrier to the accurate assessment of body composition in paediatric populations. 29 P a g e

45 1.7.2 Body composition compartment methods Body composition testing methods currently in use are broadly categorised into two, three and four compartment models based on the number of body compartments which they measure. Two compartment (2C) methods measure only fat mass and fat free mass, three compartment (3C) methods measure fat mass, fat free mass and bone mass and four compartment (4C) systems individually measure the water, protein, bone and mineral components of the fat free mass [135, 136]. These different models have been used and validated to varying degrees of accuracy in studies with children. Four compartment (4C) models are considered the gold standard in body composition testing as they directly measure the individuals body density, body water and bone mass, and eliminate the need to make assumptions about the density of fat free mass and hydration level [137]. As a result it is generally considered optimal to use 4C models for the validation of new body composition assessment methods wherever possible [137, 138]. The validation process essentially involves a comparison between different body composition methods, with each method based on its own assumptions and approximations using different software, equations and formulas. For this reason it must be noted that there is no true gold standard in measuring body composition, but the goal should be to find the most accurate method for the population to be tested. The disadvantage of 4C methods is that relatively complex procedures are required to undertake the measurements, in particular the measure of body volume. Traditionally, this is measured by hydrodensitometry (underwater weighing) [139]. This method of determining body volume has been used for decades as a reference model for adults [140], however it is a challenging procedure to perform in children, particularly young children. 30 P a g e

46 In addition to the complications associated with underwater weighing itself, accurately performing this measurement requires lung volume to be assessed at the same time as underwater weighing, and accurate measures of lung volume are difficult to obtain in children [132, 139]. Lung volume is therefore usually measured out of the water and these measurements then applied to underwater weighing measurement calculations, which is clearly not ideal [139]. Thus, while 4C models are highly accurate, the need for multiple measurements to be conducted in combination, as well as the sophisticated set-up required, makes these methods expensive, time consuming and, depending on the approach, largely impractical for measuring the body composition in children. Three compartment models (3C) measure whole body composition and can separate body compartments into fat, lean mass and bone [137]. This approach is based on the fact that these compartments all have different absorption properties, and can therefore be separated based on differences in the absorption of radiation that is passed through them [141]. The most commonly used 3C method in both adults and children is DXA. DXA has been established as a reliable and highly reproducible approach for quantifying fat and fat free mass in adults and children, and as a result is commonly used as a criterion method for validation of other techniques [131]. However, DXA has several limitations when it comes to assessing body composition in children and/or in large populations. These include exposure of the subject to ionising radiation, the need for expensive equipment and the need for the subject (child) to remain still for relatively extended periods of time whilst being tested [141]. Two compartment (2C) models are able to estimate body composition based on measurements of fat mass and fat free mass on the basis of the assumption that fat free 31 P a g e

47 mass (water, mineral and protein) has a constant density [136, 142]. While these assumptions appear to be largely valid in adults, their validity in children is less clear. There is evidence that a consistent composition of fat mass and fat free mass in compartments is not achieved until adulthood and that there is a gradual change in the water content of fat free mass, and therefore its density, during childhood and adolescence [135, 142]. In an attempt to overcome this issue, specific paediatric equations have been developed, based on tests of fat mass and fat free mass in children of different ages, which theoretically correct for these age and sex-related differences in fat free mass composition [134, 135]. However, few of these equations have been appropriately validated in separate cohorts of children to those in which they were developed, and their ability to fully overcome these issues is unclear. Densitometry, including ADP and measurement of total body water (TBW) using deuterium oxide, are 2C methods that are commonly used as reference models for assessing body composition in children [135]. ADP has been the most recent addition to this list of methods, facilitated by the release of a commercial ADP system, the BOD POD (Life Measurement Instruments, Concord, CA) [143]. This system stands out from other 2C methods by having a number of features that have been specifically designed for children, and initial studies suggest that this method produces reliable and repeatable results in children aged as young as 2 years of age [144]. For this reason the BOD POD is now utilised as a criterion method in some studies [145] and the primary method used to measure body composition in others [146, 147]. The availability of a commercial ADP system thus represents an important advance in reliably testing paediatric body composition and a step towards determining normative ranges for fat mass and fat free mass in children. 32 P a g e

48 1.7.3 Skinfold Thickness (SFT) Traditional anthropometric measurements are widely used due to their practicality and non-invasive nature, and are commonly used for assessing fatness in children. One of the methods commonly used in clinical and population studies to measure body composition is the measurement of skinfold thickness. Skinfold thickness is used to provide an estimate of body fat using a number of separate measurements completed at different standardised sites on the body using specialised calipers [148]. Skinfold thickness specifically measures subcutaneous fat, the layer of fat just below the skin, and provides no information on central or visceral fat accumulation. The measurements also require considerable expertise, and the examiners conducting these measurements must be extensively trained to avoid variability in results [149]. Existing studies have reported that total body fatness predicted from triceps and subscapular skinfold thicknesses were comparable to the same measures obtained by DXA [149]. A limited number of studies in paediatric populations have suggested that skinfold thicknesses provide a better prediction of body fatness than BMI, although the difference in the predictive ability between the two methods in these studies was relatively small [114, 148]. The accuracy of skinfold thickness measures for predicting body fatness also appears to vary according to the degree of adiposity [150], number and location of sites measured and the prediction equation used for the calculation [148]. Both BMI and subscapular and triceps skinfold thicknesses are strongly related to biochemical abnormalities including changes in serum lipid profile (HDL/LDL cholesterol, triglycerides), high fasting insulin concentration and high systolic and diastolic blood pressure among children [148, 151]. However, studies conducted to date suggest that 33 P a g e

49 skinfold thickness measurements (triceps and subscapular) provide limited additional information about metabolic health compared to that derived from BMI-for-age [149] Dual energy X-ray Absorptiometry (DXA) DXA is a widely accepted method used for the assessment of body composition in children, adolescents and adults, and is often used as a reference or criterion method against which new techniques are validated [136, 152, 153]. DXA was initially designed principally for the measurement of bone mass and bone mineral content (BMC), and remains the standard clinical approach for these assessments [154]. More recently, this method has been increasingly applied to the determination of fat mass and fat free mass [153]. DXA is classified as a 3C model, but, due to its capability to measure BMC, is quite commonly combined with other assessment methods to create a 4C model. The ability of DXA to accurately measure bone, fat and fat free mass is based on the fact that the X-ray beams which are generated by the machine pass with different efficiencies through these different body compartments. Since fat mass is less dense than muscle, for example, X-rays pass more readily though fat in the body than through lean tissue [155]. While DXA is widely recognised as being an accurate and reproducible method for assessing body composition [152, 156], applying this method to paediatric populations is not entirely straightforward. In most cases the complete measurement takes up to 15 minutes, and requires the child to stay still for the duration, which is clearly not practical for younger age groups [157]. However, the time needed to complete the DXA test is dependent on the machine used with some studies now using machines requiring the children to stay still for 5 minutes [158]. The fact that the measurement exposes the child to ionising radiation, albeit a low dose, also raises ethical issues about the use of this 34 P a g e

50 method in children, particularly for repeated measurements [132]. Furthermore, the correct conversion of the raw data obtained to a measurement of fat, fat free and bone mass in the paediatric population needs to take into account the normal changes in the density and relative size of the fat, fat free and bone mass which occur during childhood. Nevertheless, a recent study conducted by Jackson et al. comparing the results obtained using skinfold thicknesses, DXA and ADP to those obtained from the deuterium dilution method in 4 year old children concluded that DXA was the most accurate [157] Bioelectrical Impedance Analysis (BIA) and Bioelectrical Impedance Spectroscopy (BIS) BIA and BIS have been widely applied in paediatric populations due to their portability, ease of use and speed of measurement. Both BIA and BIS are based on the underlying principle that the resistance (R) of a length of a conducting homogenous material of uniform cross sectional area is proportional to its length (L) and inversely proportional to its cross sectional area (A) [159]. In both methods, the human body is assumed to consist of five interconnecting cylinders (two arms, two legs and the trunk) [160]. The separate body segments are recognised as if they are linked together (and not separate compartments) with shorter and thicker segments contributing less to the overall resistance [159, 161]. The components of total body water (fluid and electrolytes) act as natural conductors and therefore the higher the body water content the lower the resistance to the flow of current through the cylinder. Since fat free mass contains an average of 73% of water in the healthy adult human, there is a strong inverse relationship between resistance measures obtained using BIA/BIS and fat free mass [159, 162, 163]. 35 P a g e

51 BIA estimates total body water (TBW) by measuring the impedance or resistance of a low-energy single frequency current, generally 50 khz, as it passes through the body [159]. BIS uses the same basic principle as BIA but measures impedance across multiple frequencies; ~5 to 1000 khz, which makes the measurement more robust. In addition, due to the fact that low frequency currents do not pass across cell membranes while higher frequencies will travel through both the intracellular and extracellular space, measuring across a range of frequencies enables separate estimates of extracellular water and intracellular water to be made [164], which is not possible using single frequency BIA [159]. BIA and BIS use mixture theory or empirically derived prediction equations to provide an estimate of total body water from the resistance measurements. This is converted to fat free mass and fat mass based on the hydration of fat free mass, taking into account differences in sex, age and ethnicity [141, 161]. While BIA and BIS are convenient and practical methods of analysing body composition that can be used easily with children, the accuracy of these methods in children, particularly young children, has been questioned. This is largely because these methods are highly dependent on the hydration status of the subject, which is much more variable in children than in adults, and changes with age [133, 165]. In addition, different types/brands of BIA/BIS devices produce different results and this is another factor to consider when comparing results between studies. The equations that are applied to convert the raw impedance data to a measure of fat mass and fat free mass generally assume that the hydration status of fat free mass is constant for all test children of a given sex and age which may not be the case [159]. There are numerous BIA prediction equations used to calculate fat mass or TBW, and the results 36 P a g e

52 obtained from the same raw impedance data can vary significantly depending on which equation is used. This makes it difficult to compare results between studies most of which use different equations. In a recent systematic review of the use of BIA in children and adolescents, 40 different prediction equations were used amongst the 50 studies included in the review [166]. This review also highlighted the limited number of studies that made use of appropriate equations, including age and sex specific hydration factors [166]. It is also important to note that the equations currently used in paediatric settings have been developed almost exclusively in children whose weight is within a healthy weight range (between the 5 th and 85 th percentile) [167], and their applicability to children outside this range, in particular those who are overweight and obese, remains unclear. It is known that children carrying excess weight have a higher fat free mass hydration status than children with lower amounts of body fat, making the use of these standard equations questionable [168]. 37 P a g e

53 1.7.6 Air Displacement Plethysmography using the BOD POD The BOD POD (Life Measurement Instruments, Concord, CA), is the first commercially available ADP system. The BOD POD represents a convenient, simple, non-invasive system that can be used in a wide range of populations including children as young as 2 years of age, adults, pregnant women, disabled persons and the elderly [144, 169, 170]. The BOD POD is classified as a 2C model and consists of two chambers, an inside chamber with an internal volume of ~450 L, in which the test subject sits, and a reference chamber, with a volume of 300 L, behind (Figure 1). A diaphragm oscillates between the chambers, producing sinusoidal volume perturbations which result in pressure changes within the chambers (±1 cm water). Figure 1.1 Diagram representation of major system components BOD POD machine Picture adapted from [171] ADP measures the volume of an object indirectly by measuring the volume of air displaced inside the enclosed chamber (plethysmograph). This is done by measuring the volume changes that occur when the compartment is empty vs. occupied. Calculating volume from pressure measurements involves the application of two gas laws, Boyle s Law and Poisson s Law. Since the air within the BOD POD chamber behaves 38 P a g e

54 adiabatically, i.e. is allowed to freely gain and lose heat during compression and expansion [138], the air temperature fluctuates as volume changes (due to the subject s presence) [172]. By measuring the change in temperature and volume of the empty chamber, the gas laws can be used to calculate the change in air volume in the chamber with the subject inside, and hence their body volume [138, 172]. Earlier versions of this technology were relatively inefficient and inconvenient as they required isothermal conditions for correct operation. However, modifications since have allowed this system to operate without temperature controlled surroundings [138]. The raw measurement of body volume (Vbraw) is converted into a final volume by applying adjustments which account for surface area artefact (SAA) and thoracic gas volume (VTG) [173]. In children under the age of 6 lung volume is predicted by equations based on reference data in healthy children [ ]. Air on the skin and on clothing and hair (SAA) are areas of isothermal air which are minimised by the subject wearing tight fitting swimwear and a swim cap during the measurement [138]. Each subject s surface area is automatically calculated and the formula used is dependent on height, with subjects greater than or less than 110cm allocated different formulas [177, 178]. Once the volume of the subject is determined, their body density (Db) is calculated using the formula; Db= Body Mass/ (Vbraw VTG SAA) [171]. The subject s density is then inserted into a density model equation to determine fat mass and fat free mass. As with BIA and BIS, the equations used for converting density measures to measures of body composition have varied between studies and constant density values for fat and fat free mass are dependent on ethnicity, gender and age. Few of these equations have been appropriately validated to date. The SIRI equation is the most widely used in general adult 39 P a g e

55 population studies [179], but it has been shown in most studies to overestimate body fat percentage in children, and has therefore been deemed inappropriate for use in paediatric populations [139, 180, 181]. Due to differences in race of the study population, values for African American males and females are found in equations by Schutte et al. and Ortiz et al. [182, 183]. The Lohman density model [134], a modified version of the Siri model [179], and the Fomon [184] model both take into account predicted linear changes in the water and mineral content of fat free mass with increasing age of the child [179]. The Lohman model is designed to be used for children 17 or under with the Fomon model an alternative option that can be used instead of the Lohman model in children under the age of 6 [134, 184]. Siri and Lohman density equations. Density (Db) is calculated from body mass divided by volume X and Y are constants derived from the equation d2 represents the constant density of fat (0.9g cm 3 ) and d1 represents the assumed density of fat free mass as stated by either Siri [179] or Lohman [134]. While the Siri and Lohman equations are the most commonly used in paediatric studies, they are far from universally applied. Instead, a range of equations have been utilised, including the Siri (adult) equation and constants developed in-house by researchers or their colleagues [ ]. A study by Butte et al. [188] provided updated reference values for body composition in children up to 2 years of age based on the density values published by Fomon, but these have not been widely utilised thus far. Consistency is required in this area, and there is a critical need for a universal equation to be established 40 P a g e

56 in order for standard normative ranges for percentage fat mass and fat free mass in children to be developed. The more recent introduction of paediatric additions to the BOD POD in [144] has increased the scope and accuracy of testing children using this instrument. These additions include a paediatric seat, which fits securely within the BOD POD chamber that assists in safely testing small children between the ages of 2 and 6 years and increases precision by minimising movement [132, 144]. Enhanced calibration with a smaller volume measure specifically designed for use with children 6 years and under complements the updated software which includes specific equations designed to assess the fat mass % and fat free mass % in children of different ages [144]. In total, 29 studies have tested the utility of the BOD POD system for measuring fat mass/fat free mass in children across a range of ages from 2 to 18 years, by comparing the results obtained with other established assessment methods. 41 P a g e

57 1.8 Validation of the BOD POD The first study validating the BOD POD in adults was published in 1995 with studies in children following shortly thereafter [189]. The use of piglets and bovine tissue phantoms have been reported in the validation of the smaller version of the BOD POD called the PEA POD which is able to accommodate infants up to the age of 6 months. The validation test using sedated live piglets ( kg) in comparison with biochemical analysis reported reasonable precision and accuracy using the PEA POD, the results showed an underestimation of mean fat mass by the PEA POD of 0.66% (SD 1.73%) (P=0.031). The accuracy also decreased in piglets with a lower fat mass % [190]. Differences between the physiology of piglets and babies were adjusted for via modified equations and calculations for surface area in this study and this needs to be taken into consideration when interpreting these results. The study using tissue phantoms which varied in mass between 1 and 10 kgs found no significant difference between fat mass % between ADP and chemical analysis [172]. While there are no reports of studies using phantoms in the BOD POD, a number of studies have validated the accuracy and reproducibility of this system in measuring the volume and density of an inanimate object (50 L aluminium cylinder). When this object was measured 20 consecutive times on two separate days the results obtained were highly consistent, with a coefficient of variation (CV) of 0.025% on day 1 and 0.027% on day 2 [171]. Further to this, a study which tested a range of volumes between 25 and 150 L using cubes of various sizes reported a high precision across multiple sequential measurements [171]. Initial validation studies conducted by Fields and colleagues [144], provided promising results in relation to the reproducibility of the BOD POD system for 42 P a g e

58 measures of fat mass in 2 to 6 year old children. This study reported a CV of 3.5% between repeated measures of the same child on the same day. The accuracy of the measurements for fat mass and fat free mass obtained using the BOD POD system in children have been largely evaluated by comparing these results with those obtained in the same child using existing body composition methods, usually DXA (14 studies) and/or hydrodensitometry (6 studies). The majority of validation studies comparing the BOD POD with hydrodensitometry show an agreement of within 1% body fat for children aged between 6 and 19 [138]. While one study found significant differences in measurements for percentage fat between ADP and hydrodensitometry, the authors of this manuscript specifically stated that their study was inadequately powered [191]. Similarly, studies comparing the BOD POD and DXA in children typically report a high degree of agreement between the values obtained for percentage fat mass using these two approaches; mostly within 2% body fat [138]. Results obtained for fat mass % using DXA and the BOD POD also closely correlated in the majority of studies, with r 2 values of >0.80 in studies by Lockner et al. (r=0.94), Nunez et al. (r=0.90) and Radley et al. (r=0.94) [140, 192, 193]. Correlations assist in determining a strength and direction of a linear relationship between two variables but they cannot be used to infer equivalence of the results, with the use of additional analyses such as Bland Altman required to assess this. In the study by Nunez et al. Bland Altman analyses revealed no significant % fat bias by either BOD POD or hydrodensitometry vs DXA in either children or adults [192]. Similarly, the study conducted by Radley et al. reported limits of agreement to be relatively similar for all % fat estimates, ranging from +/-6.57 to +/-7.58%, with comparisons of the individual differences between the BOD 43 P a g e

59 POD and DXA revealing a significant bias associated with increased % fat (DXA), only in girls (P<0.01) [193]. In comparison with a 4C model, DXA, hydrodensitometry (underwater weighing) and TBW all showed bias in fat mass estimation with the BOD POD found to be the most accurate and precise in a group of twenty-five 9 to 14 year olds. Similarly, results from a study conducted by Gately et al. in a group of thirty 12 to 16 year olds found the BOD POD and TBW to be the most accurate in estimating fat mass [136, 137]. However, in a study by Crook et al., ADP was found be less accurate in predicting fat mass % when compared to DXA in 3 to 5 year olds [194]. Therefore, further studies are required to confirm the accuracy of the BOD POD in predicting body composition in paediatric populations. 1.9 Comparison of BIA and ADP measurements in children Comparisons of the body composition results obtained using BIA and ADP has been reported in 7 studies with children and adolescents [131, 142, ]. The first of these studies included both obese and non-obese children and adolescents aged 5 to 22 years, and suggested that BIA significantly underestimated percent fat mass in comparison to ADP, particularly in children with a percentage fat mass greater than 25% [195]. Similar findings have since been reported by others [142]. A study by Elberg et al. [131, 200] that compared a number of different body composition assessment methods with DXA in a group of 86 overweight and non-overweight boys and girls, reported that BIA, but not ADP, significantly overestimated changes in percentage body fat over a 15 month period when compared to DXA [131]. Comparison of BIA and the BOD POD (ie. 2C models) with a reference 3C model (measurement of body volume using the BOD POD and TBW using deuterium dilution), 44 P a g e

60 showed wide intervals of agreement, with both BIA and ADP overestimating body fat mass when compared with the 3C model [200]. Overestimation of body fat mass by BIA in comparison to HW, ADP and SFT has also been reported in another study conducted in adolescent males [197, 198]. Overall, it is apparent that differences between methods exists and that it is not possible to use different methods interchangeably or to directly compare body composition measures in children assessed using different methods/equations Other/additional limitations to the use of the BOD POD in paediatric populations The main limitation of the BOD POD, apart from its limited portability and issues with consistent equations described earlier, is the need for lung volume of children to be measured as part of the testing process, since obtaining accurate measures requires the subject to maintain a consistent tidal breathing pattern for several minutes [192]. The BOD POD system does, however, offer the option of using predicted lung volume (based on subjects weight, age, total volume and sex) in both adults and children and initial studies suggest that using predicted values does not alter the final result compared to using measured lung volumes [173, 192]. However, few studies have been undertaken in very young children, and it is possible that using predicted equations in these populations may be less accurate than in older children and adults [174]. The small size of a child s body volume in comparison to the BOD POD chamber volume also increases the measurement error, particularly for children with a volume below 40 L, which has the potential to reduce accuracy of body composition assessments [201]. 45 P a g e

61 1.11 Other Considerations: The importance of fat distribution While the review of current body composition assessment measures has focussed on their accuracy in measuring total body fat mass, it is important to note that all of these methods are limited in their ability to assess body fat distribution. This is significant, since it is becoming increasingly recognised that excess accumulation of fat in the visceral (abdominal) compartment is associated with greater cardio-metabolic risk in comparison with a more peripheral or subcutaneous fat distribution in children and adolescents, as well as in adults [129, 130]. The gold standard approaches for quantifying body fat distribution involve techniques which can produce detailed cross-sectional images of the body, such as computed tomography (CT) and magnetic resonance imaging (MRI) [202]. However, these methods are expensive and technically challenging, and therefore impractical for larger scale studies. This has led to a search for alternate approaches for assessing fat distribution in both adults and children. A number of software packages have now been developed for use in conjunction with DXA which allow total body scans to be divided into areas corresponding to different body regions (ie. head, trunk, arms and legs) so that the amount of fat at these different anatomical sites can be assessed separately [155, 203]. These software packages also allow users to manually trace the outlines of specific regions of interest, thereby enabling estimates of abdominal fat mass to be made based on defined anatomical landmarks [204, 205]. The most recent software packages, including the latest version of CoreScan produced by GE Healthcare [206], also incorporate automated features for quantification of visceral and subcutaneous fat from DXA scans [204]. 46 P a g e

62 The measurement of abdominal fat mass through waist circumference and Waist:Height (WtHR) are frequently used as indices for body fat distribution and as a diagnostic tool for coronary heart disease in adult populations. Waist circumference measurements are specific to the measurement of central adiposity and have been found to be highly sensitive in children, indicating that this is a valuable tool when identifying overweight and obese children at higher risk of developing metabolic and cardiovascular complications [207, 208] Conclusion The use of the BOD POD to measure body composition is now highly regarded with many studies reporting it to be both accurate and precise in adults and children [131, 138, 144]. This section of the literature review has described several different methods used to measure the body composition of children and proposed moving forward from the application of BMI to classify overweight and obesity in both adults and children. While a convenient approach for categorising overweight and obesity in large populations, BMI provides limited information on the percentage of fat mass and fat free mass. This is significant when it comes to predicting individual risk, since body fat mass is increasingly recognised as being a more important determinant of cardio-metabolic risk than body weight per se. While the majority of the studies investigating into this have been conducted in adults, there is emerging evidence that this association is also present in children and adolescents [209, 210]. While a variety of techniques are available for assessment of body composition, a lack of consensus on the most appropriate method and lack of universal equations for converting raw measures from the various methods to measures of body composition has hampered progress towards developing standard normative ranges for body fat mass in paediatric populations. Whilst Wells and 47 P a g e

63 colleagues attempted to establish normative fat mass ranges for 5 to 20 year olds in their 2012 paper there is far from a general consensus on this subject, and these figures have not been widely adopted [211]. As the incidence of childhood obesity, and its associated health consequences, continues to rise, establishing a standardised approach for monitoring body fat mass on both an individual and population level will be critical. ADP, using the BOD POD system, represents a recent addition to the suite of body composition testing methods. While it has only been used in a relatively limited number of paediatric studies to date, the results in relation to its reliability and accuracy in this population have been encouraging. Nevertheless, as indicated above, inconsistencies between studies in relation to the equations applied to calculate body composition prevent the data obtained to date to be used to establish normative ranges of body composition in children. Thus, future studies should endeavour to implement equations specific to age and sex, in order to allow for the results of individual studies to be directly compared, and data from different populations to be combined Rationale and outline of this thesis The rising prevalence of overweight and obesity is a major public health concern for Australia now and for the future. Childhood obesity has become a significant health issue in Australia, with 25.1% of children aged 2 to 17 years classified as overweight (18.2%) or obese (6.9%) [212]. Importantly, there is a clear tracking of obesity risk and associated diseases across the life course, and increased accumulation of body fat in infancy and early childhood is a major risk factor for obesity in adult life [130, 151]. It is therefore clear and widely recognised that strategies which reduce the deposition of adipose tissue in early life are essential in order to curb the rising incidence of obesity. 48 P a g e

64 There has been recent interest in the potential role of maternal n-3 LCPUFA supplementation in reducing the risk of obesity in children. There has been encouraging data from animal studies supporting the role of n-3 LCPUFA in decreasing fat cell formation and fat storage in adults. However, there has been relatively little research investigating the role of exposure to increased dietary n-3 LCPUFA, specifically DHA, in the fetal period/before birth on the body composition of children, and both animal and human studies in this area have shown mixed results. Thus, there is a lack of robust evidence as to whether maternal n-3 LCPUFA supplementation during pregnancy is likely to be an effective strategy for decreasing fat mass in children. The aim of Chapter 3 of this thesis was to determine if maternal n-3 LCPUFA supplementation (specifically DHA) can reduce body fat mass in children. This aim was addressed by conducting a follow-up of a sub-set of children of mothers involved in the DOMInO study (DHA for Optimising Mother and Infant Outcomes), in which mothers were supplemented with high dose DHA or placebo from 20 weeks gestation to delivery, at 7 years of age. This was an extension of a larger growth and body composition followup of the DOMInO children at 3 and 5 years of age. Extending this follow-up to 7 years in a sub-set of these children is important due to the major changes in the relative balance of muscle and fat which occurs during the 5 to 7 year age period [125]. This also coincides with the adiposity rebound; a period when BMI increases following a decline across the first few years of childhood [125]. Importantly, body composition of the children (fat mass and fat free mass) in this follow-up study was measured using two methods, BIS and the BOD POD, to compare fat mass % between groups which was only measured with BIS in the previous body composition follow-up of the DOMInO children at 3 and 49 P a g e

65 5 years. I hypothesised that children in the n-3 LCPUFA supplementation group would have a lower fat mass in comparison to the children in the placebo group. As discussed in the literature review, there are a number of other factors in addition to the nutritional environment experienced before birth that can influence the body composition of children. In particular, a child s diet and level of physical activity are important factors. Thus, the aim of Chapter 4 was to assess and describe the habitual diet, physical activity levels and screen time in the DOMInO children, to (a) determine whether there were any systematic differences between groups and (b) identify postnatal factors related to current fat mass % at 7 years. I hypothesised that any changes in fat mass % seen between the intervention and control groups would occur independently of differences in diet and physical activity/screen time between the groups. The Healthy Diet Index (HDI) was used to assess the quality of the children s diet as this method ascertains compliance of the child s dietary intake to the Australian National Dietary Guidelines. As indicated in the literature review (Part 2), there are a number of methods for assessing body composition in children, but no clear consensus as to the most appropriate method. The majority of studies in this area have used BMI and BMI z-score to estimate the degree of obesity, rather than more direct measures of body fat mass. BIS has been commonly used and has advantages in large populations, but its accuracy in children has been questioned. The BOD POD is a more recent addition to the range of methods for assessing the body composition of children and is gaining acceptance as a reliable and accurate method for measuring fat mass % in paediatric populations. Therefore, the aim of Chapter 5 was to compare the results obtained for fat mass % and fat free mass using the BOD POD with those obtained using BIS. I also aimed to compare the fat mass % results from 50 P a g e

66 both methods to BMI z-score, to determine to what extent BMI z-scores might be able to be used to estimate body fat % in 7 year old children. This represents the first direct comparison of BIS and the BOD POD in children, and will assist in understanding their compatibility and accuracy for future research. The findings from this thesis increase our understanding of the relative importance of the nutritional environment before birth, in early infancy and later in childhood on body composition and the future risk of obesity. This research will assist in determining optimum dietary guidelines for n-3 LCPUFA intakes during pregnancy and in understanding the long term effect of increased n-3 LCPUFA, specifically DHA supplementation during pregnancy, on the child. Furthermore, this study examines the influence of other factors including diet and physical activity, to body fat mass in early childhood. Finally, the assessment of current methods used to measure the body composition of children and their reliability/accuracy with children in this age group is important for future research in the area of childhood obesity. 51 P a g e

67 CHAPTER 2 Methods and Materials 52 P a g e

68 CHAPTER 2 METHODS AND MATERIALS 2.1 Study Population The DOMInO study (DHA to Optimise Mother and Infant Outcomes) is a multi-centre, double-blinded, randomised controlled trial which is the largest to date investigating the impact of high dose maternal n-3 LCPUFA supplementation in the second half of pregnancy on maternal and infant outcomes. Details of the DOMInO trial have been published previously [213]. Briefly, 2399 pregnant women who were less than 21 weeks gestation with singleton pregnancies were recruited from 5 Australian perinatal centres between October 2005 and January A total of 1,660 women were enrolled at Adelaide-based centres including the Women s and Children s Hospital (WCH) and Flinders Medical Centre (FMC). Women were excluded if they were already taking supplements containing DHA, had a known fetal abnormality, had a bleeding disorder in which tuna oil was contraindicated, were taking anticoagulant therapy, had a history of drug or alcohol abuse or if English was not the main language spoken in their home. The women were randomised, using a computer generated randomisation schedule, to either the DHA group or control (placebo) group. Women in the DHA group were asked to consume three 500mg capsules of DHA-rich fish oil concentrate per day which provided 800mg/d of DHA and 100 mg/d of eicosapentaenoic acid (EPA). Women in the control group were asked to consume three 500mg capsules containing a blend of vegetable oils without DHA each day [213]. 53 P a g e

69 The study described in this thesis was undertaken on a subset of children from the DOMInO study whose mothers were enrolled in one of the Adelaide-based centres (WCH or FMC). Children were eligible to participate if they had not withdrawn or died, were living in the Adelaide metropolitan area and were not participating in the neurodevelopmental follow-up of the DOMInO children at 7 years [214]. All procedures were conducted in accordance with the trial protocol. The National Ethics Application Form (NEAF) and Site Specific Application (SSA) were completed and approved by the Women s and Children s Health Network (WCHN) Human Research Ethics Committee and The University of Adelaide Human Research Ethics Committee. I completed all study appointments, measurements and analysis and was blinded to group allocation throughout the study. Written informed consent was obtained from the parents/carers of all participating children prior to the completion of the clinic appointment. 2.2 Follow-up and Clinic Appointments DOMInO participants whose children were eligible to participate in the 7 year BOD POD follow-up were sent an information sheet and consent form in the mail providing details of the study. Parents/carers were contacted by phone two weeks after this information was sent. If contact was not successful a message was left for them to return the call. If contact was not made by the participant after 7 days, subsequent contact was attempted. If they provided verbal agreement for their child to be involved, a clinic appointment was scheduled for the child to attend. Prior to the appointment, parents were sent a letter detailing the location of the clinic and a Diet and Physical Activity questionnaire, which they were asked to complete prior to their appointment. Written informed consent was obtained from each parent/carer prior to the commencement of the clinic appointment. 54 P a g e

70 Appointments were scheduled as close as possible to the child s 7 th birthday and were conducted between April, 2014 and June, Children whose parent/carer provided consent attended a 30 minute clinic appointment at the Norwich Building, North Adelaide (Figure 2.1). At this clinic appointment waist and hip circumference, weight and height were measured and percentage fat mass was assessed using bioelectrical impedance spectroscopy (BIS) which had previously been used in our research group with good accuracy and reproducibility [215] and by airdisplacement plethysmography (ADP) using the BOD POD. Basic demographic information was collected at the beginning of the appointment prior to commencement of measurements. Parents/carers were instructed that their child should fast for at least two hours before the scheduled appointment and children were instructed to void before the measurements were conducted. Figure 2.1 Clinic room Norwich building, North Adelaide where all appointments were conducted 55 P a g e