DIET FOR THE TREATMENT OF GESTATIONAL
DIABETES MELLITUS
Jovana Mijatovic
BMedSc, MNutrDiet
A thesis is submitted in fulfilment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
Primary Supervisor: Professor Jennie Brand-Miller
Associate Supervisor: Associate Professor Glynis Ross
Faculty of Science
School of Life and Environmental Sciences
Charles Perkins Centre
The University of Sydney
2019
i
Statement of the Author
I, Jovana Mijatovic, hereby declare that this thesis is my own work and that it contains no material
previously published, in part or whole for the award of another degree.
In accordance with the Faculty of Science, this thesis does not exceed 80,000 words.
Name: Jovana MIJATOVIC
Date of submission: 9th November 2018
ii
Declaration of funding and support
This study had funding from the University of Sydney’s internal revenue.
iii
Main thesis abstract
Gestational diabetes mellitus (GDM) is a transient intolerance to carbohydrates affecting an
increasing number of pregnant women worldwide in parallel with obesity. The condition is
associated with adverse pregnancy outcomes, including long-term effects on the offspring through
metabolic programming and epigenetic changes in utero. GDM promotes a vicious cycle of metabolic
diseases for future generations. Medical Nutrition Therapy is currently the cornerstone of GDM
managements. The conflicting clinical evidence (and low quality overall for that evidence) has led to
the lack of expert consensus. Establishing an effective and safe diet for management of GDM is an
urgent priority.
Lower carbohydrate (LC) diets have been growing in popularity as a means of lowering blood glucose
levels (BGL) and have been endorsed by prestigious endocrine societies for GDM management. Aside
from reducing BGL, LC diets increase the formation of ketones through increased fat catabolism
(mainly as beta-hydroxybutyrate, BHB), particularly when the glucagon to insulin ratio is high. In
pregnancy, ketone formation is exaggerated due to a shift in maternal metabolism. Therefore, the
independent metabolic impact of LC diets is superimposed on gestational metabolism. High 3rd
trimester serum BHB levels have been inversely correlated with child’s intelligence at 2-5 years old,
but the quality of evidence is poor. In the present thesis, the aim was to investigate the current
literature on LC diets and generate evidence of their safety in GDM.
In Chapter 1, our literature review indicated conflicting evidence in relation to LC diets promoting
weight loss, lowering blood glucose and insulin levels, and improving cholesterol and triglyceride
concentrations in non-pregnant populations. This may be traced to heterogeneity in daily targets of
carbohydrate and differences in both study duration and study design. However, evidence on safety
of LC diets in GDM populations was lacking, highlighting the need for further research.
iv
To bring clarity, we conducted a systematic review and meta-analysis of prospective observational
studies in which information on dietary intake and physical activity (PA) levels were collected during
preconception or early pregnancy (Chapter 2). We found that frequent intake of potato and high
protein intake (% energy) derived from animal sources suggested a higher risk of GDM, whereas the
Mediterranean diet (MedDiet), Dietary Approaches to Stop Hypertension (DASH) diet and a higher
Alternate Healthy Eating Index (AHEI) score resulted in 15-38% reduced risk. In addition,
engagement in >90 min/week in leisure time PA reduced the odds of GDM by 46%. Therefore,
modifiable lifestyle factors such as diet and PA could play a critical role in disease prevention and
provide direction for lifestyle management of GDM.
To address the gap in knowledge on safety of LC diets, we conducted the MAMI 1 study
(Macronutrient Adjustment in Mothers to Improve GDM). This was a 6-week pilot, 2-arm randomised
controlled trial (RCT) comparing the effects of a Modestly Lower Carbohydrate diet (MLC, 135 g/day
carbohydrate) and Routine Care diet (RC, 180-200 g/day carbohydrate) on blood BHB levels, and
pregnancy outcomes. In total, 45 women were recruited and 33 completed the full protocol. The
results suggested no differences in BHB levels (MLC 0.1 ± 0.0 mmol/L vs RC 0.1 ± 0.0 mmol/L; P =
0.308), glycaemia (MLC 6.1 ± 0.1 mmol/L vs RC 6.0 ± 0.1 mmol/L, P = 0.307) or insulin dose (MLC 14.6
± 1.8 units vs RC 21.2 ± 3.9 units, P = 0.126) between the study groups. Analysis of 3-day food records
confirmed that carbohydrate intake was lower in the intervention arm, (mean ± SEM, carbohydrate
MLC 165 ± 7 g vs RC 190 ± 9 g, P = 0.042). However, we also observed significantly lower energy (MLC
7040 ± 240 kJ vs RC 8230 ± 320 kJ, P = 0.006), lower protein (85 vs 103 g/day, P = 0.006), and lower
micronutrient intake (including iron and iodine) in the MLC group.
The most surprising finding was a statistically smaller infant head circumference in the MLC group
(MLC 33.9 ± 0.1 cm vs RC 34.9 ± 0.3 cm; P = 0.046), which remained significant after adjustment for
gestational weight gain (GWG), gestational age at delivery and infant sex (P = 0.043). Head
circumference ranged from the 10th to 25th percentile in the MLC group and between 25th to 50th
v
percentile for the RC diet group. Because head circumference is a proxy for brain volume and
development, this finding suggests the need for caution on LC dietary advice in GDM.
Due to the slow rate of recruitment in MAMI 1, we conducted a cross-sectional observational study
(MAMI 2) to assess whether dietary carbohydrate consumption in the previous 12 hours (including
dinner, supper and breakfast) was correlated with morning urine and blood ketones levels, or with
pregnancy outcomes in women with GDM. Of the total number of women recruited (n = 160), only
14% were positive for ketonuria. Blood BHB levels and urinary ketones were highly correlated (rS =
0.717, P <0.001), but there was no correlation between ketonuria and carbohydrate intake.
Compared to the highest tertile of carbohydrate intake (% energy), the lowest tertile had a 2-fold
increased odds of higher blood ketone levels(OR = 2.14, 95% CI: 0.98 – 4.64, P = 0.055, adjusted for
pre-pregnancy BMI, energy intake and GWG), although not statistically significant.
Collectively, the studies in this thesis suggest the need for larger, appropriately-powered studies to
determine the safety and risks associated with recommending even modestly lower carbohydrate
intake in the management of GDM. Although ketonaemia may not be of concern, total energy and
micronutrient intake may be compromised, with unintended and potentially adverse effects on
offspring. Dietary recommendations in GDM should not be based on intuition and anecdotal
evidence, but rather robust scientific evidence that guarantees improved outcomes for the mother
and her offspring.
vi
Table of contents
Statement of the Author ………………………………………………………………………………………………….. i
Declaration of funding and support ………………………………………………………………………………….. ii
Main thesis abstract ………………………………..……………………………………………………………….………. iii
Table of contents ……………………………………………………………………………………………………..………. vi
List of tables ……………………………………………………………………………………………………………..……… x
List of figures …………………………………………………………………………………………………………….……… xii
List of abbreviations ………………………………………………………………………………………………….……… xiv
Presentations arising from present thesis ………………………………………………………………………… xvi
Acknowledgements …………………………………………………………………………………………………..……… xvii
Chapter 1 - Literature review: Low carbohydrate diets and their safety in pregnancy 1
Part 1: Low Carbohydrate Diets 2
1.1 Introduction …………………………………………………………………………………………………. 2
1.2 LC diet definitions ………………………………………………………………………………………… 3
1.3 Current popularity of LC diets ………………………………………………………………….…. 4
1.4 Benefits and mechanisms of LC diets …………………………………………………………. 5
1.4.1 Weight loss …………………………………………………………………………………………..…. 6
1.4.2 Blood insulin levels and glycaemia ……………………………………………………….…. 7
1.4.3 LC diets and blood lipids ……………………………………………………………………….…. 8
1.4.4 Mood ………………………………………………………………………………………………………. 10
1.4.5 Treatment of epilepsy and cancer ………………………………………………………..…. 10
1.4.6 Sports performance ……………………………………………………………………………..…. 11
1.5 Side effects of LC diets ………………………………………………………………………………. 11
1.6 Carbohydrates ………………………………………………………………………………………..…. 13
1.7 Ketone bodies ………………………………………………………………………………………..…. 15
1.8 What influences ketone bodies? …………………………………………………………….…. 17
1.9 Measurement of ketones ……………………………………………………………………….…. 19
vii
Part 2: Understanding pregnancies and gestational diabetes 21
1.10 Normal pregnancy vs GDM ………………………………………………………………………. 21
1.11 Diagnosis of GDM ………………………………………………………………………………….…. 22
1.12 Concerns and consequences of a GDM pregnancy ……………………………………. 24
1.13 GDM monitoring and management ……………………………………………………….…. 25
1.14 LC diets and ketonaemia in pregnancy …………………………………………………..…. 26
1.15 Conclusion …………………………………………………………………………………………….…. 27
Chapter 2 - Associations of diet and physical activity with risk for gestational diabetes
mellitus: a systematic review and meta-analysis
29
2.1 Introduction …………………………………………………………………………………………..…. 31
2.2 Materials and methods ………………………………………………………………………….…. 33
2.2.1 Eligibility criteria ……………………………………………………………………………………. 33
2.2.2 Information sources and search ………………………………………………………….…. 33
2.2.3 Quality assessment and data extraction ……………………………………………..…. 34
2.2.4 Statistical analysis ……………………………………………………………………………….…. 34
2.3 Results ………………………………………………………………………………………………….…. 35
2.3.1 Studies identified .……………………………………………………………………………….…. 35
2.3.2 General characteristics of studies ……………………………………………………….…. 37
2.3.3 Diet related studies …………………………………………………………………………….…. 59
2.3.3.1 Carbohydrates (fruit, fibre, beverages, potato) ……………………………………. 60
2.3.3.2 Fat intake (i.e. total, monounsaturated fatty acids, dietary cholesterol,
egg intake) ………………………………………………………………………………………….. 61
2.3.3.3 Protein intake (i.e. meat, iron, heme) ……………………………………………..…. 62
2.3.3.4 Caffeine ………………………………………………………………………………………………. 62
2.3.3.5 Fast food intake ……………………………………………………………………………….…. 63
2.3.3.6 Calcium/dairy intake ……………………………………………………………………….…. 63
2.3.3.7 Recognised dietary patterns ………………………………………………………………. 64
2.3.4 Physical activity ………………………………………………………………………………….…. 65
2.3.5 Meta-analysis and assessment of bias ……………………………………………….…. 66
2.4 Discussion ……………………………………………………………………………………………..…. 72
viii
2.4.1 Diet and GDM risk …………………………………………………………………………………. 70
2.4.2 Physical Activity and GDM ………………………………………………………….…………. 74
2.4.3 Strengths and limitations ………………………………………………………………………. 75
2.5 Conclusions …………………………………………………………………………………………..…. 76
Chapter 3 - A modestly lower carbohydrate diet for the management of gestational
diabetes
77
3.1 Introduction …………………………………………………………………………………………..…. 79
3.2 Methods ………………………………………………………………………….……………………….. 80
3.2.1 Participant recruitment …………………………………………………………………………. 81
3.2.2 Baseline data collection …………………………………………………………………………. 82
3.2.3 Randomisation and stratification ………………………………………………………..…. 83
3.2.4 Dietary intervention, safety and compliance ……………………………………….…. 83
3.2.5 Additional outcome measures …………………………………………………………….…. 87
3.2.6 Statistics ……………………………………………………………………………………………..…. 88
3.2.7 Power calculation ……………………………………………………………………………….…. 89
3.3 Results …………………………………………………………………………………………………..…. 89
3.3.1 Blood ketone levels (BHB) …………………………………………………………………….. 97
3.3.2 Pregnancy outcomes ………………………………….……………………………………….…. 97
3.4 Discussion ………………………………….………………………………………………………….…. 104
3.5 Conclusion ………………………………….……………………………………………………………. 108
Chapter 4 - Ketone levels in women with gestational diabetes mellitus: a pilot cross
sectional study
109
4.1 Introduction …………………………….…………………………………………………………….…. 111
4.2 Methods …………………………….………………………………………………………………….…. 112
4.2.1 Statistical analysis …………………………….……………………………………………………. 116
4.3 Results …………………………….………………………………………………………………….……. 117
4.3.1 Neonatal characteristics and anthropometry …………………………….……….…. 123
4.3.2 Blood ketone and carbohydrate intake …………………………….……………………. 125
4.3.3 Blood ketone and urine ketone …………………………….…………………………….…. 127
4.4 Discussion …………………………….……………………………………………………………….…. 129
ix
4.5 Conclusion …………………………….…………………………………………………………………. 132
Chapter 5 – Discussion of main findings and future directions 133
Reference List 141
Appendices 192
x
List of tables
Chapter 1
Table 1.1 Diet stratification based on carbohydrate content in grams and percent
(%) energy of total daily intake.
3
Table 1.2 Guidelines used in diagnosis of gestational diabetes mellitus (sourced
from WHO 2013, (1)).
24
Chapter 2
Table 2.1 Modified quality assessment & risk of bias form obtained from the
Evidence Analysis Manual: Steps in the academy evidence analysis
process (2).
39
Table 2.2 Characteristics of observational studies. 41
Chapter 3
Table 3.1 Participant selection criteria. 79
Table 3.2a Sample meal plan for the Modestly Lower Carbohydrate (MLC) diet
group.
83
Table 3.2b Sample meal plan for the Routine Care (RC) diet group. 84
Table 3.3 MAMI 1 (Macronutrient Adjustments in Mothers to Improve GDM)
study collection plan.
85
Table 3.4a Baseline characteristics of participants that received education. 90
Table 3.4b Baseline characteristics of participants that withdrew from the study
prior to randomisation compared to women who completed the study.
91
Table 3.5 Maternal baseline diet in the two intervention groups. 92
Table 3.6 Dietary intakes of study participants at the end of the intervention. 93
Table 3.7 Sub-analysis of women that met their assigned carbohydrate target
intake.
94
Table 3.8 Biochemistry at baseline and end of the study for both intention-to-
treat (ITT) and compliant participants.
97
Table 3.9 Pregnancy outcomes in the two intervention groups. 98
Table 3.10 Infant characteristics at delivery. 98
xi
Chapter 4
Table 4.1 MAMI 2 study collection plan 112
Table 4.2 Maternal characteristics combined or stratified according to their
research sites.
117
Table 4.3 Maternal dietary characteristics based on the 12-hour recall,
combined or stratified according to their research sites.
118
Table 4.4 Infant anthropometry outcomes combined or stratified according to
their research sites, where possible.
121
Table 4.5 Tertiles of carbohydrate intake (%E). 123
Table 4.6 Carbohydrate content and odds of developing elevated ketones. 123
xii
List of figures
Chapter 1
Figure 1.1 Chapter 1 overview. 2
Figure 1.2 Incremental area under the curve of low and high glycaemic index foods
(Image sourced from GlycemicIndex.com)
14
Figure 1.3 Dietary manipulation and effects on insulin sensitivity and resistance
(Modified from Weickert et al. 2012) (3).
15
Figure 1.4 Hepatic production of ketones through ketogenesis and ketolysis in
extrahepatic tissues. (Figure amended from Cotter et al. 2013, (4)).
16
Figure 1.5 Multi-parameter urine dipstick test and dual blood and ketone monitor. 20
Chapter 2
Figure 2.1 PRISMA flow diagram of screening, selection process and inclusion of
studies.
36
Figure 2.2 Confounding variables that were adjusted for in studies collecting
information on dietary intake and physical activity levels.
57
Figure 2.3 Meta-analysis of participation in any physical activity (PA) versus none
and odds of gestational diabetes (GDM).
66
Figure 2.4 Meta-analysis of participation in high versus low level of leisure time
physical activity (LTPA) and odds of gestational diabetes (GDM).
67
Figure 2.5 Meta-analysis of participation in high versus low level of leisure time
physical activity (LTPA) before pregnancy in metabolic equivalents
(MET.hr/week) and odds of gestational diabetes (GDM).
68
Figure 2.6 Meta-analysis of high versus low level of leisure time physical activity
(LTPA) before pregnancy reported in hr/week and odds of gestational
diabetes (GDM).
68
Figure 2.7 Assessing the risk of publication bias using funnel plots for different meta-
analyses.
69
Chapter 3
Figure 3.1 Target carbohydrate distribution for the control and intervention arms of
the MAMI 1 study.
82
Figure 3.2 Flow diagram depicting progress of a 2-group parallel randomised trial. 88
Figure 3.3 Cumulative frequency of participants consenting to take part in the study
at Royal Prince Alfred and Campbelltown Hospitals.
88
xiii
Figure 3.4 Birthweight stratified by weeks’ gestation at delivery and infant gender
using the Australian National Birthweight percentiles (1998-2007) as the
comparator.
99
Figure 3.5 Infant outcomes based on maternal dietary intervention group. 100
Chapter 4
Figure 4.1 Air displacement plethysmography (Pea Pod) device. 111
Figure 4.2 FreeStyle Optium Neo meter and corresponding ketone strips. 113
Figure 4.3 Cumulative recruitment of participants at Royal Prince Alfred (RPA) and
Campbelltown Hospitals.
115
Figure 4.4 Association between maternal pre-pregnancy BMI and age. 119
Figure 4.5 Correlation of maternal weight gain at enrolment and maternal pre-
pregnancy BMI.
119
Figure 4.6 Odds ratios (OR) of higher birthweight (BW) or percent Fat Free Mass
(%FFM) when Institute of Medicine’s (IOM) weight gain guidelines are
exceeded.
122
Figure 4.7 Urine ketone, glucose, protein and leukocytes in pregnant women
diagnosed with gestational diabetes mellitus (GDM).
124
Figure 4.8 Correlation between urine samples testing positive for ketone and their
corresponding blood ketone levels.
125
Chapter 5
Figure 5.1 Framework for further investigation on the possible relationship between
modestly lower carbohydrate (MLC) diet and infant head circumference.
134
xiv
List of Abbreviations
AcAc Acetoacetate
Acetyl CoA Acetyl Coenzyme A
ADA American Diabetes Association
ADIPS Australasian Diabetes in Pregnancy Society
AHEI Alternative Healthy Eating Index
ANOVA Analysis of variance
ANZCTR Australian New Zealand Clinical Trials Registry
BGL Blood glucose level
BHB Beta hydroxybutyrate
BMI Body mass index
CDC Centres for Disease Control and Prevention
CI Confidence interval
CVD Cardiovascular diseases
DASH Dietary approaches to stop hypertension (diet)
EI Energy intake
FAO Fatty acid oxidation
FFA Free fatty acid
GDM Gestational diabetes mellitus
GI Glycaemic index
GL Glycaemic load
HAPO Hyperglycaemia and Adverse Pregnancy Outcome
HbA1c Glycated haemoglobin
HDL-C High-density-lipoprotein cholesterol
KE Ketone ester
IADPSG International Association of the Diabetes and Pregnancy
Study Groups
IOM Institute of Medicine
LDL-C Low-density-lipoprotein cholesterol
LGA Large-for-gestational age
lnOR Log (natural) Odds Ratio
xv
MAMI Macronutrient adjustments in mothers to improve
gestational diabetes mellitus
MedDiet Mediterranean diet
MLC Modestly lower carbohydrate (diet)
MNT Medical nutrition therapy
OGTT Oral glucose tolerance test
OR Odds ratio
PA Physical activity
RC Routine care (diet)
REMA Random-effects meta-analysis
RCT Randomised controlled trial
RPA Royal Prince Alfred (Hospital)
RR Relative risk
SEM Standard error of the mean
SGA Small-for-gestational age
SSB Sugar-sweetened beverages
SWSLHD South-Western Sydney Local Health District
TAG Triglyceride
T1DM Type 1 diabetes mellitus
T2DM Type 2 diabetes mellitus
WHO World Health Organization
xvi
Presentations arising from this project
Oral presentation
1. Lifespan Research Network, Charles Perkins Centre 2015 – MAMI 1 protocol.
2. Royal Prince Alfred Hospital, Endocrinology and Obstetrics meeting, 2015 – MAMI 1
protocol.
3. Westmead Hospital, Endocrinology meeting, 2017 – Systematic review and meta-analysis.
4. Early and Mid-Career Symposium, Charles Perkins Centre 2017 – Systematic review and
meta-analysis.
5. Dietitians Association Australia, Diabetes Interest Group 2018 – MAMI 1 and MAMI 2 results,
systematic review and meta-analysis results.
Poster presentations
1. Higher Degree by Research student poster 2015 – MAMI 1 Protocol
2. 77th Scientific Session, American Diabetes Association (San Diego) 2017 – Systematic review
only.
3. Westmead Hospital 2017 - Systematic review only.
xvii
Acknowledgements
Realisation of this thesis could not have been possible without guidance and assistance from the
wonderful people I have met during the last few years.
To my primary supervisor, Professor Jennie Brand-Miller, the words “Thank you” cannot describe
how grateful I am. Your selfless time, care and believing in me were all that kept me going at times.
I feel honoured to have been under your supervision and greatly value the knowledge you have
shared with me during my thesis journey.
To my secondary supervisor, Associate Professor Glynis Ross, thank you for always finding time to
help me in the clinic and sharing your clinical knowledge despite the fast-paced hospital
environment.
A special gratitude goes out to Dr Fiona Atkinson, Dr Ros Muirhead and Mrs Shannon Brodie for our
many discussions, the many encouragements as well as assistance with the much-dreaded dietary
data entry. Thank you for making my thesis journey more enjoyable!
To my dear friend Ms Marion Buso, it was wonderful collaborating with you to tackle statistical
problems I often faced, but also for the many conversations about life and travel which made the
last few weeks pass so quickly.
Thank you to Professor Vicki Flood for conceptualising the systematic review and meta-analysis
found in Chapter 2 of this thesis and for allowing me to become a part of it. To all the authors who
contributed, your expertise and assistance ensured that we published the review.
In running both the MAMI 1 and MAMI 2 studies, this could not have been possible without the help
from RPA Hospital team members including Ms Anna Jane Harding, Ms Kim Nicholls as well as
Campbelltown Hospital team including Professor David Simmons, Ms Elizabeth Fletcher.
Thank you to all the mothers who selflessly set their time aside to participate in our MAMI 1 study.
Without your efforts and help, this thesis would not exist.
Finally, I wish to thank my family for the moral support and unconditional love you have showered
me with, especially in difficult times when I could not see the finish line in sight. I hope I did you
proud.
1
Chapter 1
___________________________________________________
Literature Review: Low carbohydrate diets and
their safety in pregnancy
2
Chapter 1, Part 1
Low carbohydrate diets
1.1 Introduction
The concept of a lower carbohydrate (LC) diet is not a new one. The first reported recommendation
was by William Banting in 1869 for weight loss (5). Since then, like the swing of a pendulum, the LC
diet has gone in and out of fashion as exemplified by the emergence of Atkins (6), ketogenic (7) and
Palaeolithic diets (8), with weight loss as the primary goal. Aside from weight loss, particularly in
obesity, LC diets have shown favourable effects on diabetes management (9, 10). In this chapter, we
examine 1) the scientific evidence behind LC diets and 2) its applicability in pregnancy, particularly in
gestational diabetes mellitus (GDM).
Figure 1.1. Chapter 1 overview.
PART 1 - LC diets
• Defining LC diets, growing popularity and suggested health benefits
• Side effects of LC diets
• Carbohydrates and ketone metabolism
• Ketones monitoring and factors influencing their levels
PART 2 - GDM and LC diets
• Defining GDM
• GDM diagnosis, prevalence and risk factors
• Metabolic differences between a healthy and GDM pregnancy
• Maternal lipid metabolism and accelerated starvation
• GDM monitoring & management
• Ketonaemia in pregnancy, are there any concerns?
3
1.2 LC diet definitions
Defining a LC is a challenge, primarily due to inconsistencies in daily carbohydrate targets. Among
the popular diets, the Atkins’ diet recommends <20 g/day or <100 g/day, depending on the diet
phase, while the ketogenic diet specifies <50 g/day (11). Dr Bernstein’s diabetes solution diet aims
for 30 g/day (11). More recently, Hashimoto and colleagues stratified LC diets into either very-low
(<50 g/day, carbohydrates comprising ~10% of total energy intake) or mild (~200 g/day,
carbohydrates comprising ~40% of total energy intake) carbohydrate content (12). On the other
hand, Wheeler et al. categorised carbohydrate diets based on 4 levels of carbohydrate restriction
(13). A summary of diet stratification based on carbohydrate content is shown in Table 1.1.
Low levels of carbohydrate intake contrast with national dietary guidelines in Australia (14) and
United States (15), which recommend that carbohydrates comprise 45-65% of total energy intake
(16). The acceptable macronutrient distribution ranges were established by Institute of Medicine’s
(IOM) Food and Nutrition Board on the basis that carbohydrate intake >65% total energy with a
reciprocal fall in fat intake (<20%) decreased the risk of coronary heart disease (16), whereas low
carbohydrate intake followed by compensatory increases in fat (>45% total energy intake) were
associated with higher risk of obesity (17).
Table 1.1 Diet stratification based on carbohydrate content in grams and percent (%) energy of total daily intake.
Carbohydrate diets Grams/day % Energy Very Low 21 – 70 11
• Atkins - -
- Phase 1 <20 11 -
- Phase 2 80 – 100 11 -
• Dr Bernstein’s 30 -
• Ketogenic <50 11 10 12
• Joslin Diet - 25 18
Modestly Low - 30 – 40 13
Moderate 200 13 40 – 65 11
High - >65 13
1 Wheeler et al. 2010 (13); 2 Fields et al. 2016 (11); 3 Hashimoto et al. 2016 (12); Osler et al. 1923 (18)
4
1.3 Current popularity of LC diets
The re-emergence of LC diets is driven in-part by the obesity epidemic (19). Currently, it is estimated
that 375 million women and 266 million men across the globe are obese (BMI ≥30), a 5- to 8-fold
increase since 1975, respectively (20). When geographic location is considered, it becomes clearer
that certain populations are more predisposed to overweight and obesity than others. The
prevalence of female obesity in some developing Caribbean and Middle Eastern countries has
reached 40-50% and exceeded 50% on several Polynesian and Micronesian islands (20).
The rates of overweight and obesity in developed countries are also high, with almost two-thirds of
adults classed as either overweight or obese in Australia (21) and United States (22). One-third of
women of a childbearing age are above the normal weight threshold (23, 24) and the numbers are
expected to rise further (25). In fact, by 2030, it is anticipated that 3 billion or 40% of the total world
population will be either overweight or obese (26, 27). Obesity is also the strongest risk factor for
chronic diseases such as type 2 diabetes (T2DM) (20), GDM (28, 29), cardiovascular disease (CVD)
(30) and cancer (31), which show the same upward trend over time.
The potential causes of obesity and concurrent metabolic diseases include diet, physical activity and
the environment in which we live. Mechanisation has considerably reduced our physical activity
levels and consequently energy expenditure in the last five decades (32). Even incidental activity such
as walking to our local supermarkets have become a rarity. In an urban setting, the size of our
supermarkets was reported to have a positive relationship with obesity (33). Certainly, our busy
lifestyles may have led to less opportunity to shop and cook food from basic ingredients. The
purchase of processed and pre-prepared foods with longer shelf-life are therefore favoured over
perishable options (33).
Many individuals pursuing weight loss diets have regained the weight within months or years (34).
This suggests that obesity is under complex biological control (34) and not merely an imbalance of
energy intake and expenditure (35). Nonetheless, there is an understandable desire to achieve rapid
5
weight loss (19), leading to a booming weight loss industry. Currently, it is estimated that $64 billion
was spent on weight loss products in United States alone (11), almost doubling since 1999 (36). The
number and availability of specific low carbohydrate products in supermarkets has increased over
the years (37, 38), with estimated sales up 500% in 2004 when compared to sales in 2001 (39). It
comes as no surprise that 12 million copies of Atkins’ diet books have been sold across the globe,
making it an all-time best-selling diet book (40).
The popularity of LC diets and their products can also be attributed to the marketing tactics employed
by manufacturers. The presence of nutrition claims such as “sugar free”, “no added sugar”, "carb
smart” or “carb conscious” could potentially mislead consumers to believe that the product is
healthful (38, 41), while the fat, added sugar and overall energy content could be relatively high (41,
42). In recent times, the low carbohydrate products are targeted at a wider weight-conscious
population where women are greater consumers (42, 43), not merely people pursuing LC diets (41).
People with diabetes are more likely to be conscious of low carbohydrate nutrition claims (43)
because they are taught that carbohydrate is the main macronutrient that affects postprandial
glycaemia (44). Educating individuals to actively check claims and read nutrition information panels
before deciding to purchase a product is one approach to overcoming marketing tactics (42).
1.4 Benefits and mechanisms of LC diets
Since the 1970s, LC diets such as the Atkins diet have claimed to be effective in mobilising lipids
within adipose cells via fatty acid oxidation to provide accelerated weight loss (19, 45). Rapid initial
weight loss may be attributed to depletion of glycogen stores, loss of water associated with the
glycogen structure (46, 47) or a reduction in the overall energy intake (19, 36, 48-51). The
mechanisms behind LC diets have now been the subject of intense investigation, with increased
interest in insulin dynamics and glucose homeostasis as well as in blood lipid changes. LC diets have
6
also been studied in the context of mood, epilepsy, cancer and sports performance where other
mechanisms may be relevant (51-54).
1.4.1 Weight loss
Several randomised controlled trials (RCTs) have demonstrated that LC diets promote greater weight
loss when compared to diets of higher carbohydrate content (36, 48, 49, 55-58). A systematic review
and meta-analysis by Bueno and colleagues, consisting of 13 studies and 1400 overweight or obese
patients, suggested that LC diets resulted in ~1 kg (weighted mean difference kg = -0.91, 95% CI: -
1.65, -0.17, P = 0.02) greater weight loss than low fat/high carbohydrate diets (59). A meta-regression
analysis of intervention studies lasting >4 weeks indicated a mean difference of 1.6–1.7 kg. The best
weight loss results occurred when the diet contained <41% energy from carbohydrates (60).
However, the weight loss effects of LC diets may be short term (3-6 months) (10, 48, 51, 61), and are
often accompanied by weight regain at 12 months (50, 56). Greater dietary adherence to LC diets
may help explain why the results initially appear more favourable (50, 59), although long term
adherence may be difficult (45). However, majority of the studies did not account for quality of
carbohydrates consumed.
LC diets usually induce a state of ketosis, i.e. the presence of elevated ketone levels (See 1.7 Ketones)
which are commonly believed to inhibit appetite. A systematic review by Gibson and colleagues of
26 studies provided convincing evidence of appetite reduction when following LC diets, despite a
period of severe energy restriction (9). These authors also suggested that the threshold level of
ketones (specifically beta-hydroxybutyrate, BHB) required to reduce appetite could be 0.5 mmol/L,
as higher levels of ketosis did not provide any additional benefit (9). In contrast, Foster and
colleagues reported no relationship between ketosis and weight loss (48), but appetite was not
assessed.
7
Other mechanisms may contribute, including decreases in the hormone ghrelin and increases in
leptin concentrations which together reduce appetite, and thereby leading to lower energy intake
(9, 46, 51). Greater satiety on LC diets may also be related to increases in dietary protein intake and
greater thermogenesis, leading to higher energy expenditure (62).
Unfortunately, a comprehensive systematic review consisting of 107 articles on LC diets suggested
that there were no significant differences in weight loss between different levels of carbohydrate
intake (19). When protein-sparing, modified fast diets (low energy, high protein and low
carbohydrate diets) were compared to low and very low-calorie diets, there were no observed
differences in weight loss (63). Similarly, in the context of T2DM, a systematic review by Snorgaard
et al. reported that LC diets did not provide any added benefit to weight loss when compared to
other higher carbohydrate diets (61), although, once again, carbohydrate quality was not considered.
Certainly, heterogeneity with respect to daily carbohydrate targets and levels of energy restriction
(vs ad libitum), as well as study duration and design were evident in majority of the systematic
reviews. The relationship between LC diets and weight loss is therefore inconclusive at present.
1.4.2 Blood insulin levels and glycaemia
Since carbohydrates have a direct effect on glycaemia (44), limiting carbohydrate intake should
theoretically lead to reductions in serum glucose levels. According to a 2-year RCT by Shai et al., LC
diets reduced glycated haemoglobin (HbA1c, a measure of average glucose levels, mean ± SD = 0.9 ±
0.8%, P = 0.05) (55), but did not contribute to improvements in fasting blood glucose levels. Of 5
systematic reviews and a meta-analysis on LC diets in a population of either higher BMI or T2DM (10,
19, 59, 61, 64, 65), 1 reported short term improvements in fasting glucose concentrations (-0.06
mmol/L [95% CI-1.67 to -0.44]) (64) and 2 indicated a small reduction in HbA1c levels (-0.21 [95% CI
8
- 0.24 to -0.18] or -0.34% [95% CI -0.63 to -0.06]) (61, 64). Protein-sparing, modified fast diets
produced a greater reduction in both fasting plasma glucose as well HbA1c levels when compared to
low and very-low calorie diets which contain carbohydrate (63).
The conflicting effects of LC diets on glycaemia were observed in a recent systematic review of diet
in type 1 diabetes mellitus (T1DM). The review of 9 studies noted differences in study design and
sample size (66). The significant heterogeneity among studies and lack of HbA1c reporting led to the
conclusion that the benefits of LC diets on improving glycaemia cannot be established at present. As
HbA1c is a superior indicator of glycaemic control than fasting glucose, future studies should
specifically report HbA1c as well as other cardiovascular risk markers to ascertain the benefits of LC
diets, particularly in populations with diabetes.
Fasting insulin levels may also be relevant when they reflect improvements in insulin resistance
rather than impairments in beta-cell function. Santos and colleagues reported a significant decrease
in fasting insulin following a period of LC diet consumption (-2.24 µIU/L) (64). While findings from an
observational study suggested that a lower fasting insulin in middle-aged and elderly non-diabetic
people may be an indicator of reduced insulin resistance and risk of metabolic syndrome (67), in the
context of LC diets, the lowering effect of fasting insulin could potentially reflect beta-cell
dysfunction (68). Moreover, a systematic review Bravata et al. (19) and a meta-analysis by Bueno et
al. (59) reported no differences in insulin concentrations, even when the degree of weight loss was
taken into consideration (19). For these reasons, the controversy surrounding LC diets remains.
1.4.3 LC diets and blood lipids
The majority of the systematic reviews on LC diets versus low-fat diets report an improvement in
certain risk factors for CVD, i.e. an increase in serum high-density-lipoprotein cholesterol (HDL-C)
(0.04 to 0.12 mmol/L) and a decrease in triglyceride (TAG) concentrations (-0.17 to -0.8 mmol/L) (10,
9
59, 64, 65). The rise in HDL-C could partially be attributed to the replacement of carbohydrates with
dietary fats (69), although careful consideration should be employed when selecting fat type due to
their differing effects on metabolism. For instance, higher intake of saturated fatty acid acts
unfavourably towards HbA1c and insulin sensitivity when compared to higher intake of
polyunsaturated fatty acids (70).
The effect of LC diets on HDL-C and TAG could be beneficial for overall health, since higher HDL-C
cholesterol was reported to be protective against onset of coronary artery disease (71), while higher
serum TAG was associated with an increased risk of CVD (72). Interestingly, 1 mmol/L reduction in
TAG was reported to decrease CVD death risk by 23% (73). Other diets, such as the Mediterranean
and some low-fat/high-carbohydrate diets, have also promoted a positive change in HDL-C
concentrations. While LC diets had a superior effect in the short term (55), there were no long term
benefits (17-24 months) (59, 65).
Unfortunately, RCTs have demonstrated that LC diets have been associated with an increase in low-
density-lipoprotein cholesterol (LDL-C) concentrations at 6 months (0.14-0.17 mmol/L) and 12
months (0.20-0.37 mmol/L) (10, 65). Elevated LDL-C concentrations in LC/high-fat diets may be
related to lower insulin levels (at least in the short term) and abundance of free fatty acids (74).
Considering that high LDL-C concentration is associated with risk of CVD-related (73) and all-cause
mortality (75), the rise in LDL-C following LC diets is a concern. In particular, higher saturated fat
intake is associated with a rise in LDL-C (59, 69), prompting the need for closer monitoring of type of
fat following periods of carbohydrate restriction. Interestingly, many LC diet studies distinguished
between different dietary fats (e.g. saturated, polyunsaturated and monosaturated fats) in their
analyses (64), but failed to compare the effects of different carbohydrates, that are well known to
have differing effects on the metabolism (see 1.6 Carbohydrates).
In their systematic review consisting of 94 LC dietary interventions reporting data for 3268
participants, Bravata and colleagues reported no changes in serum LDL-C levels but noted that
10
reductions in LDL-C were associated with weight loss, energy restrictions and longer study
participation (19). Again, evidence of heterogeneity among included studies prevented conclusive
statements in favour of LC diets, especially in patients with lifelong metabolic syndrome who are
already at an increased risk of hypertension, elevated LDL-C and obesity (76).
1.4.4 Mood
The relationship between dieting and negative effects on mood is well established (77, 78). In fact,
Quehl et al. suggested that there was an association between depressive symptoms and a lower
healthy diet score (β = -0.016, 95% CI: -0.029 to -0.003, P = 0.017) in young female students (79).
According to a recent systematic review of the psychological and social outcomes of weight loss diets,
there are no added benefits in following a LC diet when compared to other weight loss diets. In
addition, they reported that any observed psychosocial benefit from weight loss diets is independent
of their macronutrient composition (52).
1.4.5 Treatment of epilepsy and cancer
Epilepsy is a medical condition affecting the brain, characterised by unexpected seizures (80). Almost
a century ago, LC diets were recommended for the treatment of seizures, but gradually became less
popular following the emergence of anticonvulsant medications (54). In recent years, drug-resistance
as well as avoidance of undesirable medication side effects prompted the resurfacing of LC diets for
the treatment of epilepsy (54). The hypothesised mechanism of action comes from the formation of
ketone bodies following carbohydrate restriction, which are believed to have anticonvulsant
properties, reducing neuronal excitability and subsequent seizure incidence (51). According to a
Cochrane review, seizure incidence was reduced up to 85% after 3 months following a LC diet
composed of 4:1 fat to protein ratio (81). However, as anticipated, the diet was associated with
11
adverse effects including gastro-intestinal disturbances, resulting in participant withdrawals from the
study (81).
Following their success in reduction of tumour size in mice and tumour progression in humans, LC
diets have also been prescribed for the treatment of certain cancers (51). LC diets are thought to
‘deprive’ cancer cells of glucose, subsequently reducing their rapid replication (82). Although LC diets
were reported to improve the response to chemotherapy, a systematic review concluded that the
effects cannot be generalised due to heterogeneity in study design, type and location of cancers (82).
1.4.6 Sports performance
The conventional dietary message for athletes is to consume a diet consisting of ~40-60% energy
derived from carbohydrates (83). However, LC diets have been prescribed to older athletes in the
hope of overcoming progressive insulin resistance with increasing weight and oxidative stress.
Following LC diets, endurance athletes often exhibit metabolic adaptations, such as ketosis where
they utilise an abundance of fat stores to fuel their performance without necessitating consumption
of carbohydrates. Emerging evidence suggests that ketones may fuel most of the brain’s energy
demands when carbohydrate intake is limited and promote better muscle recovery following
exercise (53). This subject remains controversial and is likely a long way from resolution.
1.5 Side effects of LC diets
Multiple side effects have been associated with LC diets. Elevated blood and urine ketone
concentrations as well as compensatory increases in protein content of the diet following
carbohydrate restriction, were reported to drive an impairment in insulin sensitivity, liver and renal
12
function, and hyperdyslipidaemia in children and adults (19). Long-term exposure to ketogenic/LC
diets may also cause kidney damage and development of kidney stones in children (84). The
propensity for kidney stone formation as well as bone loss has been observed within 6-weeks of
following a LC diet containing high protein content in a healthy population (85). However, modestly
higher protein diets (~25% energy intake) have not had adverse effects on estimated glomerular
filtration rate in adults with pre-diabetes (86).
LC diets invariably increase the intake of protein and fat, saturated fatand cholesterol, all factors long
associated with the risk of CVD (87). A 12-month RCT by Wycherley et al. demonstrated that a LC
diet vs low fat/high carbohydrate diet decreased flow mediated dilatation in overweight and obese
subjects (3.7 ± 0.5% vs 5.5 ± 0.7%, respectively). This suggested that the long-term LC diet pattern
may be unsafe due to unfavourable changes in the endothelial function (88). Recent studies have
shown a U-shape relationship between carbohydrate intake and risk of mortality, with low (<40%
energy) and high carbohydrate intakes (>70% energy) both associated with increased risk of death
(pooled hazard ratio 1.20, 95% CI: 1.09–1.32 vs 1.23, 95% CI: 1.11–1.36, respectively) (89).
Interestingly, when carbohydrates were substituted with protein and fat of animal origin, the risk
increased, but decreased when these macronutrients were derived from plant-based sources
(pooled hazard ratio 1.18, 95% CI: 1.08–1.29 vs 0.82, 95% CI: 0.78–0.87, respectively) (89).
Among the more symptomatic adverse effects, LC diets have been associated with constipation,
fatigue (19, 36), headaches, halitosis and diarrhoea (36). Salt and water depletion following a rapid
weight loss, may also drive feelings of general weakness, hypotension and constipation (19). Since
the LC diet usually involves lower consumption of fruits, starchy vegetables and wholegrain cereals
(all are sources of fibre for the large bowel microbial flora), the occurrence of constipation is not
surprising (87). This constellation of symptoms is likely to lead to low long-term compliance (49, 51).
13
1.6 Carbohydrates
Carbohydrates are a heterogenous class of nutrients ranging from simple sugars to starches and
dietary fibre. Carbohydrate foods include fruit, starchy vegetables, refined grains, wholegrains, dairy
products, table sugar and sugar-sweetened beverages. Due to their differing properties, they also
have varying effects in the human body and overall metabolism. Fermentable carbohydrates
increase the odds of dental caries through increases in oral acidity (90). However, no such effect was
observed with higher quartiles of dietary fibre intake (90, 91). Based on comprehensive meta-
analyses and systematic reviews, higher intake of sugar-sweetened beverages and refined cereals
has been associated with weight increase and higher risk of T2DM, while higher consumption of
foods naturally high in dietary fibre, such as wholegrain cereals, fruits and vegetables, is protective
(92).
Jenkins and colleagues pioneered the glycaemic index (GI) concept - a method of categorising
carbohydrates with respect to their propensity to raise blood glucose levels when compared to the
same amount of carbohydrate in the reference food (93). As depicted in Figure 1.2, low GI foods
produce lower glycaemic fluctuations, a desirable trait for individuals with impairments in
carbohydrate metabolism. Other favourable effects associated with low GI diets include
improvements in inflammatory markers (C-reactive protein), fasting insulin and TAGs, glucose
tolerance and a reduced risk of developing T2DM, breast cancer and gallbladder disease (94-96).
Figure 1.2 Incremental area under the curve of low and high glycaemic index foods (Image sourced from GlycemicIndex.com).
14
Dietary fibre was reported to improve insulin sensitivity and slow down absorption of glucose
independently of a food’s GI value (97). A study examining the effects of 3 cereal-containing meals
consumed at dinner-time, such as white bread, pasta (low GI, low fibre) and barley (low GI, high
fibre), demonstrated that a high fibre meal resulted in greater glucose tolerance at breakfast the
following morning (95), possibly via ‘second-meal’ phenomenon – i.e. presence of a late glycemic
response. Aside from positive effects on glycaemia, dietary fibre positively interacted with certain
gut microbiota (Lactobacillus and Bifidobacterium) to reduce inflammation, lower LDL-C as well as
the risk of certain cancers, onset of T2DM and CVD (98).
The link between dietary fibre and gut health was further emphasised in a study examining the
effects of a high-protein, animal-based diet vs a high-fibre, plant-based diet. Within a day, as the
contents reached the colon, Bilophila wadsworthia and Prevotella microorganisms increased sharply
in the stools of individuals following animal vs plant-based diets, respectively (99). Bilophila has been
previously associated with inflammation, dysfunction of the intestinal barrier and dysmetabolism of
bile acid (100), while Prevotella improved glucose metabolism (101). Diets with low fibre content
(including many LC diets) were reported to reduce the stool concentration of short chain fatty acids,
which play an important role in maintaining the immune system, appetite and glucose homeostasis
(102).
Given that high fibre diets are generally high in carbohydrates, this raises questions about
carbohydrate types when drawing conclusions with respect to health benefits and detriments of LC
diets. One meta-analysis compared the effects of LC diets to high carbohydrate diets and exclusively
removed studies with high fibre to remove its confounding effects from the analysis (103). Although
the majority of metabolic parameters were comparable, fasting insulin and triglycerides were higher
and HDL-C lower, in the high carbohydrate diet groups (103). However, it is likely that a higher fibre
version of the high carbohydrate diets could have ameliorated these effects. While it cannot be
denied that refined carbohydrates increase the risk of certain chronic diseases such as diabetes (104),
15
GI and dietary fibre content of a diet should never be neglected as carbohydrate quality plays a
greater role in population health than quantity (105). Figure 1.3 summarises the effects of multiple
dietary components on insulin sensitivity.
Figure 1.3 Dietary manipulation and effects on insulin sensitivity and resistance (Modified from Weickert et al. 2012) (3). BCAA – branched chain fatty acid; GI – glycaemic index; MUFA – monounsaturated fatty acid; SFA – saturated fatty acid; TFA – trans fatty acid.
1.7 Ketone bodies
Ketone bodies are a by-product of the breakdown of fat as a fuel source. They are produced in the
liver as acetoacetate (AcAc), BHB (most abundant, 75% ketones produced) and acetone (least
abundant) (106-108). When the glucagon to insulin ratio is high, ketogenesis is initiated in
mitochondria of hepatic cells. Free fatty acid oxidation (FAO) and subsequent formation of AcAc,
shifts the equilibrium towards BHB formation via a NAD+/NADH- coupled reaction (Figure 1.4) (4, 107,
109). BHB is a 4-carbon compound that moves out of the cell via solute carrier 16A (SLC16A), en route
16
to extrahepatic tissues (4, 110). Upon entry into the extrahepatic cells using the same carrier, BHB
migrates to mitochondria where ketolysis ensues. In a multi-step process, BHB is broken down to
Acetyl Coenzyme A (Acetyl CoA), which enters the tricarboxylic acid cycle to undergo terminal
oxidation (4).
Figure 1.4 Hepatic production of ketones through ketogenesis and ketolysis in extrahepatic tissues. (Figure amended from Cotter et al. 2013, (4); AcAc – Acetoacetate: CoA –Coenzyme A; BHB – Beta hydroxybutyrate; FAO – Fatty acid oxidation; TCA – Tricarboxylic acid).
Only 1-4% of ketone bodies are synthesised from non-fatty acid sources, such as glucose metabolism
and amino acid (particularly leucine) catabolism (4). Others have suggested that between 15-20% of
the carbon structure could be obtained from amino acids, with estimated overall ketone body
production ~115-180 g/day (110). Following a meal, the ratio between BHB:AcAc is normally 1, but
can rise to 6 in instances of a prolonged fast (106). Serum concentrations of BHB can vary across
populations due to age, fasting duration and differences in basal metabolic rates (106). Normal levels
of BHB are usually <0.5 mmol/L, while hyperketonaemia is defined as 0.5-3.0 mmol/L (some suggest
>1 mmol/L) and ketoacidosis as >3 mmol/L. Ketoacidosis is seen in uncontrolled type 1 diabetes and
requires medical intervention (106, 107). Collectively, ketone bodies including BHB and AcAc are 20%
more energy efficient than glucose and can fuel approximately half of the basal energy usage (106,
17
110, 111). However, the only organ not capable of metabolising ketone bodies for energy is the liver
(112).
Compared to smaller animals, humans have a slower basal metabolic rate enabling them to produce
and utilise greater amounts of ketone bodies. Ketotic sheep and dogs can oxidise only 50% of ketone
bodies to carbon dioxide, while humans can oxidise a greater proportion, possibly due to a larger
brain or larger adipose tissue stores in the case of obese subjects (113). The human brain accounts
for only ~2% of total body weight, but consumes up to 20% of total energy intake (114). Although
preference lies in utilisation of glucose, the brain can also oxidise ketone bodies for energy (114).
Ketogenesis is often halted or dampened by the action of insulin which acts on Foxa2 to supress
transcription of Hmgcs2 (4, 115).
1.8 What influences ketone levels?
Under normal circumstances, low levels of ketone bodies are produced (106, 116) but the process is
accelerated during periods of starvation, or consumption of LC diets, high-fat diets or calorie
restricted diets. Higher levels of ketones are produced during prolonged exercise, pregnancy and
uncontrolled metabolic disease states, such as T1DM, T2DM, cortisol deficiency and growth hormone
deficiency (106, 110, 117).
Obese humans can live ~2 months during starvation, provided they have access to water (118). Given
that hepatic stores of glycogen amount to ~70 g in a non-fasted state, they would be consumed
within 18-24 hours of fasting if gluconeogenesis and mobilisation of free fatty acids were not initiated
(109). After several days of fasting, blood glucose levels often stabilise and the brain obtains most
of its energy requirements from BHB and AcAc (which equates to 100-145 g/day of glucose under
normal circumstances) as they pass through the blood brain barrier (118). Amazingly, during
prolonged starvation, the body spares the use of muscle protein for energy to ensure survival of the
18
host, accompanied by a decline in insulin concentration so that the production of ketone bodies is
not supressed (118). Much like starvation, LC diets reduce the availability of glucose, through
intentional restriction of carbohydrate foods. Although nutritional ketosis ensues, BHB
concentrations are much lower than those observed in individuals experiencing diabetic ketoacidosis
(119). BHB and AcAc gradually become the main source of energy in LC diets (120).
Diabetic ketoacidosis is a life-threatening condition characterised by a decline in blood’s pH value
due to a rise in metabolic acidosis. This is more common in individuals with recent diagnosis of T1DM
and therefore complete insulin deficiency or uncontrolled T2DM (110, 121). In uncontrolled T2DM,
hyperinsulinaemia contributes to cytoplasmic localisation and inactivation of a mitochondrial Foxa2
transcription factor (responsible for inhibiting the production of BHB), resulting in hepatic
accumulation of lipids that leads to higher BHB production, thereby furthering insulin resistance
(115).
A healthy pregnancy is associated with a progressive rise in fasting free fatty acids (122), followed by
significant increases in BHB concentrations in the 3rd trimester due to greater insulin resistance (109,
123). In addition, there is an exaggerated 3-fold increase in maternal ketone levels when breakfast
is omitted (124). This manifestation is termed “accelerated starvation” (124), which preserves
glucose for fetal needs, while providing BHB for maternal use (109). During a therapeutic abortion,
84 hours of fasting resulted in a 30-fold rise (~4.2 mmol/L) in BHB levels, far more pronounced than
in fasted non-pregnant women (109). While glucose is the preferred source of energy for maternal
brain function and fetal development, ketones can cross the placenta without affecting fetal
secretion of insulin (107, 110, 125, 126). Newborns have a slightly higher rate of ketone production
initiated by the high fat content of maternal milk, although this is deemed normal in the absence of
ketonuria (106). The presence of ketonuria, however, is indicative of a disturbed metabolism (106).
Children also have a greater propensity for hyperketonaemia following 24-hours of fasting, due to
their low stores of hepatic glycogen (106, 107).
19
Ketones can also be consumed from exogenous sources. In the absence of carbohydrate and energy
restriction, synthetic (R)-3-hydroxybutyl (R)-3-hydroxybutyrate ketone ester (KE) dampened muscle
glycolysis despite periods of intense physical activity (116). Another KE (R-3-hydroxybutyrate-R-1,3-
butanediol monoester) was recently reported to increase resting energy expenditure and markers of
adipose tissue thermogenesis in obese mice fed a high-fat diet (127). This led to less weight gain and
lower fat mass compared with pair-fed controls.
1.9 Measurement of ketones
At present, body ketone levels can be established by a means of urinalysis or blood (capillary or
venous) and breath samples (Figure 1.5). Urinalysis involves immersing a reagent strip (dipstick) into
a “mid-stream” urine sample for a microscopic assessment (128) of ketones. Blood ketones in the
form of BHB can be measured using hand-held monitors and accompanying ketone test strips
requiring a small sample of blood (~0.6 µL). These strips were previously validated and show a high
degree of reliability and reproducibility (129).
Figure 1.5 Multi-parameter urine dipstick test (left) and dual blood and ketone monitor (right).
20
Blood ketones assessed by the hand-held monitors have been reported as superior to ketone
urinalysis, particularly in diagnosing ketoacidosis (130). In fact, ketonuria may be high before blood
ketones begin to rise (131). Ketone urinalysis may produce false positive results, particularly when
dealing with highly pigmented urine samples (132), whereas some medications (sulphydril (108) or
levodopa medications (132)), as well as strip air exposure may lead to false-negative results (108).
Nitroprusside-type test strips cannot detect urinary BHB (108), as they have been designed to
establish the presence of the other major ketone body, AcAc (132). Certainly, urinalysis is qualitative,
meaning that a change in strip colour may be subject to misinterpretation by the observer (133). On
the other hand, capillary assessment of ketones is expensive.
Breath ketone tests have recently been gaining popularity, as it is a non-invasive alternative in
assessing body ketone concentrations (134). Unlike a hand-held monitor, which determines
concentrations of BHB, the breath test assesses acetone as it is a volatile compound that can easily
be exhaled in air (134). Under normal living conditions, acetone is present in breath and its
concentrations have been correlated with fasting blood glucose, serum and urine ketones, LDL-C,
creatinine, and blood urea nitrogen (134). Therefore, breath acetone concentrations are deemed as
a reliable method for diagnosing and monitoring diabetic ketosis. The common procedure is to
exhale alveolar air into a collection bag, which will be subsequently analysed by gas chromatography
(135). However, should a participant fail to provide their alveolar air, this could lead to false negative
results. Other reported limitations include high cost, difficulty in storing bags of gas samples and
variability in acetone due to ethnicity, gender, age, dietary intake, weight and medication use (136).
In recent years, portable acetone breathalysers were developed and have suggested precision up to
100 parts per billion by volume (137).
21
Chapter 1, Part 2
Understanding pregnancies and gestational diabetes
1.10 Normal pregnancy vs GDM
Pregnancy is a transient period in a woman’s life accompanied by a plethora of hormonal changes
that correspond with crucial developmental stages in the fetus. In the 1st trimester, fasting blood
glucose levels show a temporary decrease (~10%), even though the embryo is still too small to
require glucose as a fuel (122, 138, 139). However, obese pregnant women often experience either
no decline or a rise that is attributed to higher insulin resistance (140). In early pregnancy, maternal
metabolism is anabolic, promoting accumulation of adipose tissue (141), with little or no change in
insulin secretion (142). As pregnancy progresses, particularly in the 2nd and 3rd trimester, the
metabolism milieu shifts towards catabolism, increasing lipolysis and ultimately contributing to
insulin resistance (141). This mechanism appears to be further amplified in women with GDM (143).
In the 3rd trimester of a healthy pregnancy, studies have reported a further, though modest decline
in fasting blood glucose levels (123, 139), which often increases in GDM (144).
Glucose is an important source of energy for the developing fetus, and readily crosses the placenta
(145, 146). The fetus is entirely dependent on maternal circulatory glucose concentrations as fetal
gluconeogenesis is low (147). Fetal glucose levels are ~15-20% lower than maternal (138) or
equivalent to 0.5-1.1 mmol/L (109), providing a distinct concentration gradient for glucose to cross
the placenta via facilitated diffusion. Since maternal insulin cannot cross the placenta (145), elevated
maternal blood glucose concentrations force the fetus to produce sufficient insulin (145) to
normalise glucose levels. Indeed, infants of women with uncontrolled diabetes are at higher risk of
hyperinsulinism (138), fetal overgrowth and subsequently hypoglycaemia at birth (148).
As pregnancy progresses and the placenta grows, the levels of human placental lactogen,
cortisol(138, 149), glucagon (149, 150), prolactin and maternal progesterone rise, antagonising the
22
normal actions of insulin (138). By the 3rd trimester, insulin sensitivity can be reduced by 50-70%,
when compared to non-pregnant women (151). To maintain blood glucose in the normal range,
maternal pancreatic beta-cells undergo hypertrophy to achieve a greater beta-to-alpha cell ratio
(152) and greater insulin than glucagon secretion (122, 138, 152). Higher insulin-to-glucagon ratio
directs more nutrients towards the growing fetus (150), but some women experience marked
hyperglycaemia possibly due to insulin resistance, often resulting in a diagnosis of GDM (138).
In humans, FFA also serve as a fuel source for fetal brain development and fat deposition (147),
although this is largely dictated by maternal glucose concentrations (153). Compared to glucose, the
transfer of TAG is more difficult and often requires TAG lipases to break it down to simple FFA, which
are then delivered to the fetus via transporter proteins (147). Serum cholesterol and TAG can rise
25-50% and 200-400%, respectively (154), with oestrogen and insulin resistance driving this increase
(155). High glucose concentrations observed in women with GDM (all on insulin therapy) lead to a
~30% decline in FAO, accompanied by a 3-fold higher TAG concentration when compared to healthy
controls (156). This is thought to contribute to fetal macrosomia (147).
1.11 Diagnosis of GDM
GDM is a transient form of diabetes characterised by glucose intolerance observed for the first time
in pregnancy [145]. In developed nations, pregnant women are routinely screened for GDM, usually
at 24-28 weeks gestation (157). In 2008, the HAPO (Hyperglycaemia in Pregnancy Outcomes) study
provided firm evidence that there were adverse maternal and fetal outcomes associated with
glucose levels lower than previously appreciated, even outside of diabetes (158). As a result, the
International Association of Diabetes and Pregnancy Study Groups (IADPSG) proposed new
diagnostic criteria for GDM which was based on a fetal growth outcome. Following a fasting 75 g oral
glucose tolerance test (OGTT), any one of the following are deemed sufficient for GDM diagnosis:
fasting BGL ≥5.1 mmol/L, 1-hour BGL hour ≥10.0 mmol/L and 2-hour BGL ≥8.5 mmol/L (159).
23
Although the Australasian Diabetes in Pregnancy Society (ADIPS) adopted the diagnostic guidelines,
they may not be consistently used in clinical practice. Some centres still prefer to use older criteria
(ADIPS: fasting BGL ≥5.5 mmol/L or the 2-hour ≥ 8.0 mmol/L (160)) or other criteria for GDM
diagnosis (Table 1.2). Although minor variations in diagnosis are justified to reflect Australian local
conditions, the progression towards an international diagnostic method is recommended (160).
A recent survey conducted among health practitioners and members of the Society for Maternal–
Fetal Medicine in the United States suggested that 90% use a 2-step diagnostic test (i.e. 50 g glucose
challenge and 100 g OGTT) and ~80% used the Carpenter-Coustan criteria for the 3-hour OGTT (161).
With respect to the cut-off point for the 50 g glucose challenge test, it varied between 130-140 mg/dL
(equivalent to ~7.2-7.8 mmol/L) (161). Therefore, both Australia and United States demonstrate
inconsistency with regards to diagnostic criteria used to ascertain GDM status.
Table 1.2 Guidelines used in diagnosis of gestational diabetes mellitus (sourced from World Health Organization, WHO 2013, (1)). Glucose concentration is given in mmol/L.
24
1.12 Concerns and consequences of a GDM pregnancy
Exposure to a hyperglycaemic environment during fetal development has long been hypothesised to
be associated with fetal macrosomia (162). According to the Barker hypothesis of “fetal origins of
adult disease”, not only does the uterine environment influence fetal growth and development, but
also plays a role in “programming” long-term health, largely through epigenetic changes (163-166).
The Dutch famine birth cohort demonstrated that food shortages during pregnancy increased the
risk of T2DM, obesity and coronary heart disease in the offspring later in life, particularly if
undernutrition occurred during early gestation (164).
Evidence from the HAPO study demonstrated that healthy pregnant women with moderately
elevated glycaemia not only had a higher risk of developing pre-eclampsia, higher induction of labour
rates and caesarean section delivery, but also a higher risk of neonatal complications including
shoulder dystocia, neonatal macrosomia and hypoglycaemia (158). The maternal risk of progressing
to T2DM in later life was as high as 70% (167, 168), with infants also likely to develop the disease in
early adulthood (169), along with obesity in childhood (170), especially if they were born large-for-
gestational age (170, 171). Maternal obesity and excessive gestational weight gain (GWG) were
reported to be independent factors that predicted undesirable pregnancy outcomes in T1DM, T2DM
and GDM (172, 173). The concept of intrauterine environment influencing offspring health long-term
was further corroborated by others. McLean et al. demonstrated that female offspring of GDM
women were twice as likely to develop GDM themselves, compared with female offspring of fathers
with diabetes (174). According to a recent study, GDM altered the expression of 26 microRNA (micro
Ribonucleic Acid) signatures in the feto-placental endothelial cells in a sex specific manner (22 in
females and 4 in males), suggesting greater metabolic derangements in female than male neonates
(175).
The prevalence of GDM continues to rise due to advanced maternal age and obesity, as well as
adoption of the new diagnostic criteria as suggested by the IADPSG (160, 176, 177). While the
25
proportions have reached ~30% in some populations (160, 178), there is some evidence of
improvement. In fact, several studies to date have demonstrated that lifestyle changes have the
potential to improve GDM pregnancy outcomes (179). A recent meta-demonstratedfor the first time
that nutrition modification (across different types) does reduce offspring birthweight, as well as
maternal fasting and postprandial glucose concentrations (180) (See section 1.13).
1.13 GDM monitoring and management
Medical nutrition therapy (MNT) is the first line of treatment for women recently diagnosed with
GDM, with specific focus on monitoring quality and quantity of carbohydrate intake, in conjunction
with prescribing more physical activity. The ADIPS guidelines suggested the following target BGLs for
GDM: fasting ≤5.0 mmol/L, 1-hour postprandial ≤7.4 mmol/L and 2-hour postprandial ≤6.7 mmol/L
(181). When lifestyle approaches are insufficient, women are often started on insulin or other
hyperglycaemic lowering agents (182).
At present, it was acknowledged that women should consume a minimum 175g carbohydrate/day
(183), but there is no established consensus on the most effective dietary strategy for the
management of GDM (184, 185). One probable reason is that lower carbohydrate intake is often
accompanied by an increase in fat, which is worrisome as exemplified by animal models (186) and
some human studies regarding a positive association between maternal lipids and offspring growth
(187). The lack of agreement provides fuel for apprehension and confusion among pregnant women.
The American Endocrine Society and the American College of Obstetricians and Gynaecologists
recommend a lower carbohydrate diet with ~30-45% energy derived from carbohydrates, with
glucose management as the primary goal (188, 189). However, only a handful of studies have
investigated the effects of LC diets in a GDM population, often with mixed findings regarding
maternal glucose levels and pregnancy outcomes (44, 190, 191).
26
According to a Cochrane review and meta-analysis, the combined effect of various lifestyle
interventions compared to a control/standard care group resulted in reductions in risk of postnatal
depression, large-for-gestational age infants and neonatal fat mass (180, 192). More specifically, one
review by Viana and colleagues suggested that a low GI diet was associated with favourable
outcomes including less frequent insulin use and lower infant birthweight (193). This demonstrates
that lifestyle interventions may be able to dampen the effect of intrauterine stressors commonly
associated with GDM pregnancies (e.g. hyperglycaemia) and therefore improve pregnancy
outcomes.
1.14 LC diets and ketonaemia in pregnancy
In the 1980s, Churchill and colleagues were among the first to suggest a correlation between
ketonuria in mothers with diabetes and lower intelligence quotient of their offspring in a
retrospective observational study (194). Unfortunately, the association disappeared when the data
were later adjusted for multiple factors including psychomotor development (e.g. Down Syndrome)
(195).
It was not until a study published in the prestigious New England Journal of Medicine in 1991 (196),
that ketone bodies gained another wave of attention. Pregnant women with pre-gestational diabetes
(n = 89), GDM (n = 99) and a healthy pregnant control group (n = 25) were included with the aim of
collecting several metabolic parameters, such as serum BHB and glucose concentration between 2nd-
3rd trimester. After adjustment for multiple confounding variables (e.g. socioeconomic status,
diabetes status and ethnicity), Rizzo and colleagues reported a strong inverse correlation between
3rd trimester maternal BHB concentrations within all 3 groups (196) and child’s intelligence at 2-5
years old. What was more concerning was that BHB serum concentrations (range = 0.14-0.18
mmol/L) were well below the threshold for ketonaemia (<0.5 mmol/L) and the effect of BHB on
27
children’s intelligence was observed even in the healthy pregnant women. Similarly, a meta-analysis
consisting of 12 observational cohort studies suggested that infants born to mothers with diabetes
(GDM and pregestational) had lower mental and psychomotor development scores than infants born
to healthy mothers. However, dietary intake and BHB concentrations were not assessed and the
studies were reasonably heterogenous (197). In the limited number of studies available on
carbohydrate reduction in GDM, only urinary ketone levels (44, 190, 191) have been assessed. The
popularity of LC diets for the management of GDM makes it more important than ever to resolve the
controversy surrounding serum BHB and pregnancy outcomes.
To date, several animal models have investigated the effects of LC, high-fat diets on pregnancy
outcomes. One study suggested that a ketogenic diet during gestation in mice resulted in changes
in embryonic organ growth including the brain, which was smaller at 13.5 days gestation but larger
at 17.5 days gestation when compared to a standard high carbohydrate diet (198). Another study in
a different mouse model reported that LC, high-fat diets resulted in greater production and activity
of arginase enzyme in the lung, contributing to obesity in the offspring, without causing an
impairment of the glucose metabolism (199). Because both animal models indicated negative effects
of LC diets on the offspring, more studies in human pregnancy are required, particularly in
metabolically-challenged populations such as GDM.
1.15 Conclusion
At present, there is a lack of consensus surrounding the use of LC diets in pregnancy, despite an
urgent need to establish the most effective treatment strategy for women with GDM. The increasing
incidence of GDM together with the popularity of LC diets has resulted in the need for a greater
understanding of the safety and potential health consequences for the mother and her offspring.
The following chapters of this thesis address in turn the lifestyle risk factors associated with GDM
(Chapter 2), a pilot RCT called MAMI 1 (Macronutrient Adjustments in Mothers to Improve GDM)
28
examining the effects of a LC diet in GDM management and pregnancy outcomes (Chapter 3), and a
cross-sectional observational study (MAMI 2) investigating the relationship between recent
carbohydrate intake and random blood ketone concentrations in women with GDM (Chapter 4).
Given the new findings, the final chapter suggests future directions for research on diet in GDM
(Chapter 5).
Chapter 2
___________________________________________________
29
Associations of diet and
physical activity with risk for
gestational diabetes mellitus: A
systematic review and meta-
analysis
Note: Chapter 2 was published in Nutrients Journal, May 2018. While the content remained the
same, the chapter was reformatted to match the rest of this thesis.
Citation:
Mijatovic-Vukas, J.; Capling, L.; Cheng, S.; Stamatakis, E.; Louie, J.; Cheung, N.W.; Markovic, T.; Ross, G.; Senior, A.; Brand-Miller, J.C.; Flood, V.M. Associations of diet and physical activity with risk for gestational diabetes mellitus: A systematic review and meta-analysis. Nutrients 2018, 10, 698.
Abstract
Rising rates of gestational diabetes mellitus (GDM) and related complications have prompted calls
to identify potentially modifiable risk factors that are associated with gestational diabetes mellitus
(GDM). We systematically reviewed the scientific literature for observational studies examining
specific dietary and/or physical activity (PA) factors and risk of GDM. Our search included PubMed,
Medline, CINAHL/EBSCO, Science Direct and EMBASE, and identified 1167 articles, of which 40 met
our inclusion criteria (e.g., singleton pregnancy, reported diet or PA data during pre-
30
pregnancy/early pregnancy and GDM as an outcome measure). Studies were assessed for quality
using a modified Quality Criteria Checklist from American Dietetic Association. Of the final 40
studies, 72% obtained a positive quality rating and 28% were rated neutral. The final analysis
incorporated data on 30,871 pregnant women. Dietary studies were categorised into either
caffeine, carbohydrate, fat, protein, calcium, fast food and recognized dietary patterns. Diets such
as Mediterranean Diet (MedDiet), Dietary Approaches to Stop Hypertension (DASH) diet and
Alternate Healthy Eating Index diet (AHEI) were associated with 15–38% reduced relative risk of
GDM. In contrast, frequent consumption of potato, meat/processed meats, and protein (% energy)
derived from animal sources was associated with an increased risk of GDM. Compared to no PA,
any pre-pregnancy or early pregnancy PA was associated with 30% and 21% reduced odds of GDM,
respectively. Engaging in >90 min/week of leisure time PA before pregnancy was associated with
46% decreased odds of GDM. We conclude that diets resembling MedDiet/DASH diet as well as
higher PA levels before or in early pregnancy were associated with lower risks or odds of GDM
respectively. The systematic review was registered at PROSPERO (www.crd.york.ac.uk/PROSPERO)
as CRD42016027795.
2.1 Introduction
Gestational diabetes mellitus (GDM) is defined as diabetes or glucose intolerance occurring for the
first time in pregnancy (200). Although diagnostic criteria vary, GDM affects approximately 15-20%
(201) of pregnancies, reaching 30% (178, 202) in some parts of the world. The World Health
Organization’s (WHO) new diagnostic criteria explain some, but not all, of the recent rise in
prevalence (178). Delayed age of motherhood, obesity and migration of higher risk population
groups to regions of lower risk also contribute to increasing prevalence of GDM (203). Women with
GDM have up to 70% chance of progressing to type 2 diabetes mellitus (T2DM) within 28 years post-
31
partum (167). GDM also increases the risk of macrosomia, hypoglycaemia (204) and epigenetic
changes in the infant, resulting in a new generation susceptible to obesity and type 2 diabetes later
in life (205).
Current treatment options include changes in lifestyle, e.g. more physical activity (PA) and improved
diet quality through MNT (192). MNT aims to achieve and maintain euglycaemia through
consumption of appropriate meal portions, distribution of carbohydrates (206) and consumption of
foods with a lower GI (178, 192). A recently published Cochrane review and meta-analysis of RCTs
demonstrated that provision of more intensive health care such as additional dietary counseling and
exercise, lead to an improved glycated hemoglobin (HbA1c), lower incidence of large-for-gestational-
age (LGA) infants, decreased weight gain in pregnancy and lower rate of depression in mothers three
months post-partum, compared to standard care (192). However, when lifestyle approaches alone
are not sufficient, insulin or other anti-hyperglycaemic pharmacological therapies are often
prescribed (178, 192).
The increasing prevalence of GDM has prompted calls to identify key lifestyle factors that either
prevent or promote onset of the disease. The Finnish Gestational Diabetes Prevention Study (RADIEL)
study suggested that the risk of GDM can be reduced by approximately 40% by following a physically
active lifestyle and a diet enriched with fruits, vegetables and wholegrain cereals as per Nordic
Nutrition Recommendations (207). While higher fruit and vegetable intake has also been reported
to reduce all-cause mortality, particularly cardiovascular mortality in a general population (208),
other dietary patterns, habits and components may be relevant risk factors for GDM.
PA has long been prescribed to patients with diabetes due to improvements in glycaemia and insulin
sensitivity (209). It has been proposed that PA achieves these benefits by promoting an increase in
skeletal muscle glucose uptake in resistance training or increase mitochondrial density and
expression of glucose transporter proteins that are evident in aerobic exercises (209, 210). Women
are currently advised to take part in 150-300 min/week in moderate-intensity aerobic PA both during
32
pregnancy and after delivery (211) as there is a strong inverse association between PA and excessive
weight gain in pregnancy, postpartum depression as well as GDM (212). However, despite the
benefits, only 23-29% of pregnant women meet the recommendations (212). Women generally
decrease their incidental PA as pregnancy progresses (213) and spend the majority of their day being
sedentary (up to 60%), as shown by motion sensor data from the US (214).
The aim of the present study was to undertake a systematic literature search of observational studies
investigating the associations between diet and PA aspects before and in early pregnancy that are
associated with risk of GDM. In addition, we conducted a meta-analysis on PA studies to examine the
associations of specific types or duration of PA with GDM risk.
2.2 Materials and Methods
2.2.1 Eligibility Criteria
Longitudinal and cohort studies containing information on diet and PA either prior to or at the
commencement of a singleton human pregnancy were included. Although GDM was the main
outcome measure, studies that reported other outcomes (e.g. pre-eclampsia, macrosomia and
large/small for gestational age offspring) were also included. Studies were excluded if they were
published in a language other than English, subjects had diabetes mellitus in pregnancy diagnosed at
33
the start of the pregnancy, or an underlying medical condition that affected digestion and absorption
of nutrients (e.g. sleeve gastrectomy or gastric bypass surgery).
2.2.2 Information Sources and Search
This systematic review was registered at PROSPERO (www.crd.york.ac.uk/PROSPERO) as
CRD42016027795. A systematic literature search was undertaken initially 22nd October 2015 and was
later updated 2nd February 2017 by three independent researchers (J.M-V., L.C. and S.C.). The search
was conducted using Medline, PubMed, Science Direct, EMBASE and Cumulative Index to Nursing
and Allied Health Literature (CINAHL) databases with interest in studies that reported on the
relationship between diet and/or PA during pre-pregnancy or early pregnancy and risk of GDM.
Additionally, we hand searched reference lists to obtain more studies. For a complete list of search
terms, please refer to Table 2.1. A time frame 1985-2017 was selected as it reflected the current
context of high prevalence of GDM as well as to capture good quality cohort and longitudinal studies.
2.2.3 Quality Assessment and Data Extraction
To determine study quality, studies were assessed independently (J.M-V., L.C., S.C.) using a modified
version of the Quality Criteria Checklist found in the American Dietetic Association (ADA) Evidence
Manual (2). Information about confounding variables and adjusted data was also collected from
included studies to highlight study strengths or weaknesses. We assigned either a positive, neutral
or a negative value for studies that were of excellent, good and poor quality respectively. Study
characteristics were extracted into a pre-determined table that collected information including
34
author, year of publication, participant number, study selection criteria, data collection methods,
GDM diagnostic criteria as well as results and how they were deduced. Data were cross-checked by
co-author (V.F.) for any errors and discrepancies.
2.2.4 Statistical Analysis
Our primary outcome measure was onset of GDM. We quantified study results using odds ratios
(OR), which along with their sample variances, were calculated using the escalc function in metafor
R package (Maastricht University, Maastricht, Netherlands) (215). A positive effect size suggests that
the odds of GDM are higher in group A (active) than B (inactive). Where results were reported as
stratified across groups a combined effect-size was calculated following Borenstein and colleagues
(216). Where results were reported in text as relative risk (RR), we used the following equation from
Deeks and Altman (217) to convert values reported to OR, where pc is the typical event rate without
treatment:
For analysis, odds ratios were log transformed (i.e. lnOR), and in places we back-transform overall
lnOR for interpretation by raising e (exponential) to the power of the lnOR.
We analyzed effect sizes using a random-effects meta-analysis (REMA) model, implemented in the R
package metafor. Statistical heterogeneity was quantified via the heterogeneity statistic, I² - a type
of intra-class correlation (218). I² corresponds to the percentage of among effect sizes variance, that
cannot be attributed to sampling. An I² value determined variability of results between different
studies as either low (25%), moderate (50%) or high (75%). We were unable to perform a meta-
OR = RR (1 – pc)
1 – pcRR
35
analysis on dietary studies as they were too diverse in the aspects of the diet they report on.
Therefore, we report on meta-analyses conducted on PA studies only.
We assessed the risk of publication bias across studies by developing a funnel plot with lnOR scale
and inverse standard error as x and y-axis respectively. We also used a regtest function in metafor
to determine funnel plot asymmetry. It should be noted that tests of funnel plot asymmetry can be
unreliable indicators of publication bias where the number of effect sizes is small (<10) and/or there
is substantial heterogeneity. Thus, in the main text, we only present and interpret results from
publication bias tests where the number of effect sizes was greater than 10. Statistical significance
of the overall effect of PA was inferred when the p-value was <0.05 and 95% Confidence Intervals
(CI) did not contain 0.
2.3 Results
2.3.1 Studies Identified
We extracted 1166 articles from the database searches, and one article was identified through hand-
search. After screening and assessing for eligibility, we identified 40 journal articles which met
inclusion criteria (Figure 2.1). Of these 40 articles, 23 reported data only on dietary intake, 15 only
on PA and two articles reported on both dietary intake and PA. Twenty-nine studies (72%) obtained
a positive quality rating and 11 (28%) were rated neutral (Table 2.1). All the articles reported findings
from prospective cohort studies of which there were multiple publications from four major studies
including Nurses’ Health Study II (n = 14) (219-232), Omega (n = 7) (233-239), Australian Longitudinal
Study on Women's Health (n = 4) (240-243) and Project Viva (n = 2) (244, 245). The most common
reasons for exclusion of publications were unavailability of full texts, no relevant data collected
36
necessary for the present review and late recruitment of study participants, consequently capturing
dietary and PA information that were not reflective of the pre-pregnancy or early pregnancy period.
Figure 2.1. PRISMA flow diagram of screening, selection process and inclusion of
studies.
2.3.2 General Characteristics of Studies
The review captured data on 30,871 pregnancies of which 1980 (7%) developed GDM. The studies
provided information on women from multiple populations including 26 American (219-239, 244-
248), five Australian (240-243, 249), two Hispanic American (250, 251) and one each for the
following: Iranian (252), Danish (253), Canadian (254), Pakistani (255), Norwegian (256), Spanish
(257), and multi-centre Mediterranean Study (Algeria, France, Greece, Italy, Lebanon, Malta,
37
Morocco, Serbia, Syria and Tunisia) (258). The number of participants per study ranged from 97 to
71,239 and were published between 1997-2016, with an age range of 16-48 years as reported in 27
studies. Of 22 studies that reported retention rate, 14 had ≥80% (233, 235, 239, 242, 244, 245, 247,
249, 251, 253, 255-258), 7 had 50-79% (236-238, 243, 246, 250, 254), and only one <50% (242).
The reported GDM diagnostic methods included a 100g (n = 12) (233, 235, 236, 238, 244, 245, 248,
250-252, 255, 259), 75g (n = 5) (242, 243, 249, 256, 258), 50g (n = 1) (246), a combination of these (n
= 2) (254, 257) OGTT or were extrapolated from medical records (n = 20) (219-232, 234, 239-241,
247, 253, 260). Multiple diagnostic criteria were used to ascertain GDM status and included 1.) 1997
American Diabetes Association (ADA) criteria (n = 2; fasting ≥105 mg/dL, 1-hr ≥190 mg/dL, 2-hr ≥165
mg/dL, 3-hr ≥145 mg/dL) (234, 235), 2.) 2004 ADA criteria (n = 10; fasting ≥95 mg/dL, 1-hr ≥180
mg/dL, 2-hr ≥155 mg/dL, 3-hr ≥140 mg/dL) (236-238, 244, 245, 250-252, 255, 257), 3.) 2010
International Association of the Diabetes and Pregnancy Study Group (IADPSG) criteria (n = 3; fasting
≥5.1 mmol/L, 1-hr ≥10.0 mmol/L, 2-hr ≥8.5 mmol/L) (249, 256, 258), 4.) 1998 Australasian Diabetes
in Pregnancy Society (ADIPS) criteria (n = 3; fasting ≥5.6 mmol/L and/or 2-hr ≥8.0 mmol/L) (241-243)
or were not reported (n = 22) (219-233, 239, 240, 246-248, 253, 254). Thirty-six studies reported on
pre-pregnancy body mass index (BMI), of which 20 were within the normal range (219-225, 228-231,
234, 236-238, 244, 252, 256, 257), one overweight (246), one obese (140) and 14 were categorised
into multiple groups (227, 232, 235, 242, 243, 245, 247, 248, 250, 251, 253-255, 258) rather than
providing an overall average. Twenty-two publications reported data from the pre-pregnancy period
(219, 221-233, 236, 237, 240-242, 246, 257), ten focused on early pregnancy (234, 243, 245, 247-
249, 252, 253, 255, 258), seven on both (235, 238, 244, 250, 251, 254, 256) and one was unclear
(239). Refer to Table 2.2 for more information on study characteristics.
38
Table 2.1 Modified quality assessment & risk of bias form obtained from the Evidence Analysis Manual: Steps in the academy evidence analysis process (2).
Au
tho
r, Y
ear
Res
earc
h q
ues
tio
n
clea
rly
stat
ed
Par
tici
pan
ts
rep
rese
nta
tive
of
a
GD
M p
op
ula
tio
n
Res
po
nse
Rat
e
Att
riti
on
Rat
e
Exp
osu
re le
vel
des
crib
ed
Die
t o
r P
A
asse
ssm
ent
too
ls
valid
ated
Met
ho
d o
f G
DM
dia
gno
sis
stat
ed
Ap
pro
pri
ate
stat
isti
cal a
nal
ysis
Co
nfo
un
din
g fa
cto
rs
adju
sted
Dis
cuss
ion
of
fin
din
gs, b
ias(
es)
&
stu
dy
limit
atio
ns
iden
tifi
ed &
dis
cuss
ed
Fun
din
g o
r
spo
nso
rsh
ip b
ias
un
likel
y
Qu
alit
y R
atin
g
Adeney et al. 2007 (233) Y Y * Y Y X Y Y Y Y Y Neutral Badon et al. 2016 (234) Y Y NA NA Y N Y Y Not E Y Y Neutral Bao et al. 2013 (219) Y Y NA NA Y Y Y Y Y Y Y Positive
Bao et al. 2014a (220) Y Y NA NA Y Y Y Y Y Y Y Positive Bao et al. 2014b (221) Y Y NA NA Y Y Y Y Y Y Y Positive Bao et al. 2016 (222) Y Y NA NA Y Y Y Y Y Y Y Positive Baptiste-Robert et al. 2011 (246)
Y Y Y NA Y Y Y Y Not E Y Y Positive
Behboudi-Gandevani et al. 2013 (252)
Y * X X Y Y Y Y Not E Y * Neutral
Bowers et al. 2011 (223) Y Y NA NA Y Y Y Y Y Y Y Positive Bowers et al. 2012 (224) Y Y NA NA Y Y Y Y Y Y Y Positive Chasan-Taber et al. 2008 (251)
Y Y Y Y Y Y Y Y Not E Y Y Positive
Chasan-Taber et al. 2014 (250)
Y Y Y Y Y Y Y Y Not E Y Y Positive
Chen et al. 2009 (225) Y Y NA NA Y Y Y Y Not E Y Y Neutral Chen et al. 2012 (226) Y Y NA NA Y Y Y Y Not E Y Y Neutral Currie et al. 2014 (254) Y Y Y Y Y Y Y Y Not E Y Y Positive Dempsey et al. 2004 (235) Y Y Y Y Y X Y Y Not E Y Y Positive Dominguez et al. 2014 (257) Y Y Y X Y Y Y Y Y Y * Neutral Dye et al. 1997 (247) Y Y Y Y Y X Y Y Not E Y Y Positive Gresham et al. 2016 (240) Y Y Y * Y Y Y Y Y Y Y Positive Harrison et al. 2012 (249) Y X Y Y Y Y Y Y Not E Y Y Positive Hinkle et al. 2015 (253) Y Y Y Y Y * * Y Not E Y * Neutral Iqbal et al. 2007 (255) Y Y Y Y Y Y Y Y Not E Y Y Positive Karamanos et al. 2014 (258) Y Y Y X Y Y Y Y Y Y Y Positive Morkrid et al. 2007 (256) Y Y Y Y Y Y Y Y Not E Y Y Positive
39
Au
tho
r, Y
ear
Res
earc
h q
ues
tio
n
clea
rly
stat
ed
Par
tici
pan
ts
rep
rese
nta
tive
of
a
GD
M p
op
ula
tio
n
Res
po
nse
Rat
e
Att
riti
on
Rat
e
Exp
osu
re le
vel
des
crib
ed
Die
t o
r P
A
asse
ssm
ent
too
ls
valid
ated
Met
ho
d o
f G
DM
dia
gno
sis
stat
ed
Ap
pro
pri
ate
stat
isti
cal a
nal
ysis
Co
nfo
un
din
g fa
cto
rs
adju
sted
Dis
cuss
ion
of
fin
din
gs, b
ias(
es)
&
stu
dy
limit
atio
ns
iden
tifi
ed &
dis
cuss
ed
Fun
din
g o
r
spo
nso
rsh
ip b
ias
un
likel
y
Qu
alit
y R
atin
g
Oken et al. 2006 (244) Y Y Y Y Y Y Y Y Not E Y Y Positive Osorio-Yanez et al. 2016 (236)
Y Y Y Y Y Y Y Y Y Y Y Positive
Putnam et al. 2013 (248) Y Y Y X Y Y Y Y Not E Y Y Positive Qiu et al. 2011a (237) Y Y Y Y Y Y Y Y Y Y Y Positive Qiu et al. 2011b (238) Y Y X Y Y Y Y Y Y Y Y Positive Radesky et al. 2008 (245) Y Y Y Y Y Y Y Y Not E Y * Neutral Rudra et al. 2006 (239) Y Y Y X Y Y Y Y Not E Y Y Neutral Schoenacker et al. 2015 (242)
Y Y Y Y Y Y Y Y Y Y Y Positive
Schoenacker et al. 2016 (241)
Y Y Y Y Y Y Y Y Y Y Y Positive
Solomon et al. 1997 (227) Y Y Y X Y X Y Y Not E Y Y Neutral Tobias et al. 2012 (228) Y Y NA NA Y Y Y Y Y Y Y Positive Van der Ploeg et al. 2011 (243)
Y Y Y X Y Y Y Y Not E Y Y Neutral
Zhang et al. 2006a (229) Y Y NA NA Y Y Y Y Y Y Y Positive Zhang et al. 2006b (230) Y Y NA NA Y Y Y Y Y Y Y Positive Zhang et al. 2006c (231) Y Y NA NA Y Y Y Y Y Y Y Positive Zhang et al. 2014 (232) Y Y NA NA Y Y Y Y Y Y Y Positive Key: Y = Yes; N = No; NA = Not Available; * = Unclear Abbreviations: E – Energy; GDM - Gestational Diabetes Mellitus; PA - Physical Activity
40
Table 2.2 – Characteristics of observational studies.
DIET & PHYSICAL ACTIVITY (PA)
Source Aim & Study population Selection Criteria
Diet Assessment
Method
Physical Activity
Assessment Method
Diagnostic Method for
GDM
Statistical Analysis & Adjusted factors
Selected Main Findings (RR, OR etc.)
Quality Rating, Retention
Baptiste-Robert
et al. 2011 (246)
To determine pre-
pregnancy PA & dietary
intake in early pregnancy
& its effect on glucose
tolerance test.
n = 152
Age: 30.1 (SD = 5.2)
Country: United States
Study: Parity,
Inflammation & DM
Inclusion: <14
weeks
gestation,
no history of
DM, consent to
participate.
Validated
Rapid Food
Screener
Interview
questionnai
re (not
validated)
50g, 1-hr
GCT,
Medical
records
Multiple logistic
regressions
Adjustments: race,
age, parity,
gestational weight
gain & BMI.
68% less likely to have a 1-hr
GCT response >140 mg/dL
with a leisure score of ≥2.75
when compared to <2.75 [RR
= 0.32, 95% CI: 0.12-0.86, P
<0.05]. No association
between dietary intake &
response to 1-hr GCT
response.
Positive,
64.4%
Zhang et al. 2014
(232)
To examine the effect of
lifestyle characteristics on
risk of GDM.
n women = 14 437
n pregnancies = 20 136
Age: 24-44
Country: United States
Study: NHS II
Inclusion: No
history of GDM,
T2DM, CVD &
cancer.
Exclusion:
Pregnancies
after GDM.
Validated
FFQ.
Validated
physical
activity
questionnai
re
(not in a
pregnant
population)
.
Medical
records
Multivariable log
binomial models
with generalized
estimating
equations
Adjustments: age,
parity, family history
of DM, history of
infertility, race/
ethnicity, alcohol
intake,
questionnaire period
& total EI.
Adhering to any 4 low risk
lifestyle factors (AHEI-2010,
PA, BMI, Smoking) before
pregnancy, risk of GDM was
lower by 83% when compared
to those that did not adhere
to any [RR = 0.17, 95% CI:
0.12-0.25]. Highest quintile of
PA (≥210min/week) vs lowest
(<30min/week) reduced the
risk of GDM by 22% [RR =
0.78, 95% CI: 0.64-0.94].
Positive,
NA%
41
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Adeney et al.
2007 (233)
To examine the
relationship between
coffee consumption &
the risk of GDM.
n = 1744
Age: 32.1 (0.1) yrs
Country: United
States
Study: Omega
Inclusion: <16 weeks
gestation, knowledge of
English language.
Exclusion: <18 yrs, non-
term pregnancy, did
not plan to deliver at
the research hospitals.
121-item
semi-
quantitative
FFQ (not
validated).
100g, 3-hr
OGTT,
Medical
records
Generalized linear model using a
log-link function
Adjusted factors: age, race, BMI,
parity, smoking, alcohol use before
pregnancy, smoking during
pregnancy & chronic hypertension.
Moderate pre-pregnancy
caffeinated coffee intake
significantly reduced the risk of
GDM by 52% when compared
with non-consumers [RR = 0.48,
95% CI: 0.28-0.82].
Neutral,
87.2%
Bao et al. 2013
(219)
To examine the
association between
dietary protein &
GDM.
n women = 15 294
n pregnancies = 21
457
Age: 25-44 years
Country: United
States
Study: NHS II
Inclusion: singleton
pregnancy, >6 months
long, years 1991-2001.
Exclusion: Previous
GDM, T2DM, cancer,
CVD prior to pregnancy,
FFQ not delivered or
incomplete with
unrealistic values.
Semi-
quantitative
FFQ
(validated)
Medical
records
Multivariate logistic regression using
generalized estimating equations
Adjustments: age, parity,
race/ethnicity, family history DM,
smoking, alcohol intake, PA, total EI,
intakes of
saturated/monounsaturated/
trans/polyunsaturated fatty acids,
dietary cholesterol, glycemic load,
dietary fiber, mutual adjustment for
animal protein & vegetable protein &
BMI.
Animal protein intake
significantly increased GDM risk
by 49% [RR = 1.49, 95% CI: 1.03-
2.17], whereas vegetable protein
intake significantly reduced the
risk of GDM by 31% [RR = 0.69,
95% CI: 0.50-0.97].
Positive,
NA%
Bao et al. 2014a
(221)
To examine the
association between
pre-pregnancy fried
food consumption &
risk of incident GDM.
n women = 15 027
n pregnancies = 21
079
Age: 25-44
Country: United
States
Study: NHS II
Inclusion: No history of
GDM, T2DM,
cardiovascular disease
& cancer.
Exclusion: no pre-
pregnancy
FFQ, an incomplete
form or unrealistic EI
(<600 or
>3500kcal/day).
Semi-
quantitative
FFQ
(validated)
Medical
records
Generalized estimating equations
with log-binomials models
Adjustments: age, parity, race/
ethnicity, family history of DM,
smoking, PA, total EI, diet quality
(AHEI-2010 score) & BMI.
Frequent fried food intake
especially away from home, was
associated with a greater risk of
GDM when comparing frequency
of ≥7/week vs <1/week [RR =
2.18, 95% CI: 1.53-3.09]. BMI
adjustment resulted in
attenuated but significant risk of
GDM.
Positive,
NA%
42
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Bao et al. 2014b
(220)
To examine the
association of 3 pre-
pregnancy low
carbohydrate (CHO)
diet patterns with risk
of GDM.
n women = 15 265
n pregnancies = 21
411
Age: 25-44
Country: United
States
Study: NHS II
Inclusion: No history of
GDM, T2DM, CVD or
cancer.
Exclusion: no pre-
pregnancy FFQ or an
incomplete form with
unrealistic EI (<600 or
>3500kcal/day).
Semi-
quantitative
FFQ
(validated)
Medical
records
Log-binomials models with
generalized estimating equation
Adjustments: age, parity, race/
ethnicity, family history of DM,
smoking, alcohol intake, PA, BMI &
total EI.
Low CHO diet high in animal
protein increases the risk of GDM
by 36% [RR = 1.36, 95% CI: 1.13-
1.64, P-trend = 0.003], however
opposite is true for high
vegetable protein & fat, reducing
GDM by 16% [RR = 0.84, 95% CI:
0.69-1.03, P-trend = 0.08].
Overall low CHO diet is
associated with an increased risk
of GDM [RR = 1.27, 95% CI: 1.06-
1.51, P-trend = 0.03].
Positive,
NA%
Bao et al. 2016
(222)
To examine the
association between
pre-pregnancy potato
consumption & risk of
GDM.
n = 21 693
Age: 24-44
Country: United
States
Study: NHS II
Inclusion: No history of
GDM, T2DM, CVD or
cancer.
Exclusion: no pre-
pregnancy FFQ or an
incomplete form with
unrealistic EI (<600 or
>3500kcal/day).
FFQ
(validated)
Medical
records
Log-binomials models with
generalized estimating equation.
Adjustments: age, parity, race, family
history of DM, smoking, PA, EI &
AHEI-2010 score.
Consuming ≥5 servings/week of
potatoes compared to <1
serving/week significantly
increases the risk of GDM by 62%
[RR = 1.62, 95% CI: 1.24-2.13, P
<0.001].
Positive,
NA
43
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Behboudi-
Gandevani et al.
2013 (252)
To investigate the
association between
maternal iron/zinc
serum levels &
women’s nutritional
intake in early
pregnancy with GDM.
n = 1 033
Age: 27.57 (SD = 4.84)
Country: Iran
Inclusion: singleton
pregnancy, 20-35 yrs,
14–20 weeks gestation,
attending prenatal
clinics in specified
hospitals.
Exclusion: disease of
glucose metabolism
(T1DM/T2DM),
abortions, infections,
chronic illness, or
medical treatments.
Semi-
quantitative
FFQ
(validated)
100g, 3-hr
OGTT (2004
American
Diabetes
Association
criteria)
Mann–Whitney, chi-square &
multiple logistic regression tests
Adjustments: age, BMI, education,
parity, passive smoking, history of
GDM & family DM, serum zinc/iron &
hemoglobin levels, & deficient
zinc/iron intakes in early pregnancy.
Higher early pregnancy maternal
serum iron levels increased risk
of GDM [mean (SD) = 143.8 (48.7)
versus 112.5 (83.5) μg/dL in GDM
and non-GDM women
respectively, P <0.0001]. No
significant difference in zinc
levels & iron/zinc nutritional
intake between these groups [OR
= 1.006, 95% CI: 1.002-1.009, P =
0.001].
Neutral,
NA%
Bowers et al.
2011 (223)
To determine if pre-
pregnancy dietary &
supplemental iron
intakes are associated
with risk of GDM.
n = 13 475
Age: 22-44
Country: United
States
Study: NHS II
Inclusion: 22-44 yrs,
singleton pregnancy, no
history of GDM/T1DM/
T2DM, CVD or cancer.
Exclusion: no pre-
pregnancy FFQ,
incomplete form,
unrealistic EI (<600 or
>3500kcal/day), peri-
menopausal at
baseline, missing
information on age/iron
intake.
133-item
semi-
quantitative
FFQ
(validated)
Medical
records
Pooled logistic regression, restricted
cubic spline regressions
Adjustments: Age, parity, BMI, PA,
glycemic index, cereal fiber,
polyunsaturated fatty acids, smoking
status, alcohol, total calories, &
family history of DM.
Dietary heme iron is positively
associated with GDM risk when
comparing highest vs lowest
quintile [RR = 1.58, 95% CI 1.21-
2.08]. Every 0.5mg/day increase
in heme iron intake increases risk
of GDM by 22% [RR = 1.22, 95%
CI 1.10-1.36].
Positive,
NA%
44
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Bowers et al.
2012 (224)
To determine
whether the total
amount, type
& source of pre-
pregnancy dietary
fats is related to risk
of GDM.
n = 13 475
Age: 22-44
Country: United
States
Study: NHS II
Inclusion: age 22-44
yrs, singleton
pregnancy >6 months
(1991-2001).
Exclusion: unrealistic
total EI (<500 or
3500kcal/ day), DM,
GDM, CVD, cancer, or
missing information on
age/iron intake or peri-
menopausal at
baseline.
133-item
semi-
quantitative
FFQ
(validated)
Medical
records
Pooled logistic regression
Adjustments: age, parity, current
smoking, BMI, PA, family history of
DM, smoking, alcohol, race, & total
EI, cereal fiber, dietary cholesterol,
glycemic load & mutual adjustment
for the specific fatty acids or source
of fats.
Higher animal fat & cholesterol
intakes increased GDM risk by
88% [RR = 1.88, 95% CI: 1.36-
2.60, P=0.05] and 45% [RR = 1.45,
95% CI: 1.11-1.89, P = 0.04]
respectively, when comparing
highest vs lowest quintile.
Positive,
NA%
Chen et al. 2009
(225)
To examine the
association between
regular pre-gravid
sugar sweetened
beverage (SSB)
consumption & the
risk of GDM.
n = 13 475
Age: 24-44
Country: United
States
Study: NHS II
Exclusion: Incomplete
FFQ in 1991, >70 items
left blank (FFQ),
unrealistic total EI,
multiple gestation, no
PA data in 1991, history
of DM, GDM, cancer or
CVD.
133-item
semi-
quantitative
FFQ
(validated)
Medical
records
Cox proportional hazards models &
multivariate adjustments
Adjustments: age & parity.
Higher SSB significantly increased
the risk of GDM by 23% when
comparing ≥5 servings/week vs
<1/month [RR = 1.23, 95% CI
1.05-1.45, P-value = 0.005].
When SSB intake was treated as a
continuous variable, each
serving/day increment was
associated with a 23% increase in
GDM risk [RR = 1.23, 95% CI:
1.05-1.43, P-value = 0.01].
Neutral,
NA%
45
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Chen et al. 2012
(226)
To examine the
association of pre-
pregnancy habitual
consumption of fruits
& fruit juices & GDM
risk.
n = 13 475
Age: 22-44
Country: United
States
Study: NHS II
Inclusion: women that
did not have DM &
major chronic diseases
at baseline.
133-item
semi-
quantitative
FFQ
(validated)
Medical
records
Cox proportional hazards models &
restricted cubic spline regressions
Adjustments: age, parity, race,
smoking, alcohol intake, PA, family
history of DM, BMI, & dietary factors
(cereal fiber, processed meat/red
meat, SSB & fruit juice or apple).
Higher consumption of whole
fruits
is not associated with an
increased GDM risk, when
comparing highest vs lowest
quintile [RR = 0.93, 95% CI: 0.76-
1.16]. The association of fruit
juices with GDM risk appears to
be nonlinear, with lowest risk
reported in women with
moderate fruit juice
consumption.
Neutral
NA%
Dominguez et
al. 2014 (257)
To investigate the
incidence of GDM
according to the
consumption of fast
food in a cohort of
university graduates.
n = 3 048
Country: Spain
Study: Seguimiento
Universidad de
Navarra (SUN)
Inclusion: Graduates
from the University of
Navarra & other
Spanish universities,
registered nurses &
other health
professionals from
different Spanish
provinces.
Exclusion: Extremely
low/ high total EI, had
previous GDM or DM.
Semi-
quantitative
FFQ
(validated)
50g or 100g
OGTT (2004
American
Diabetes
Association
criteria)
Non-conditional regression models
Adjustments: age, total EI, smoking,
PA, family history of DM,
cardiovascular disease/
hypertension, parity, adherence to
MedDiet pattern score, alcohol
intake, fiber intake, and SSB intake
and BMI.
Fast food consumption was
significantly associated with an
86% higher risk of incident GDM
when compared to the lowest
category of fast food
consumption [OR = 1.86, 95% CI:
1.13-3.06].
Neutral
97.2%
46
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Gresham et al.
2016 (240)
To assess whether
diet quality before or
during pregnancy
predicts adverse
pregnancy & birth
outcomes in
Australian women.
n = 1 907
Age: 20.8 (SD 1.4)
Country: Australia
Study: Australian
Longitudinal Study on
Women’s Health
Exclusion: not classified
as pre-conception or
pregnant when
completing the FFQ,
multiple birth,
incomplete FFQ.
74-item FFQ
(validated)
Self-report Multiple logistic regressions
Adjustments: level of education, age,
weight, area of residence, smoking
status, parity, and level of exercise.
When comparing highest to
lowest quintile, diet quality was
not associated with GDM [OR =
1.7, 95% CI: 0.7-4.0].
Positive,
NA%
Hinkle et al.
2014 (253)
To examine the
relation between first
trimester coffee & tea
intake & the risk of
GDM.
n = 71 239
Age: 16-48 yrs
Country: Denmark
Study: Danish
National Birth Cohort
Inclusion: first singleton
pregnancy.
Exclusion: pre-existing
DM, data of relevant
covariates missing.
Interview Self-report &
medical
records
Chi-square statistics for bivariate
analyses & modified Poisson
regression
Adjustments: age, parity, smoking
status, cola intake, BMI, SES.
Suggested a protective, but non-
significant association with
increasing coffee [≥8 vs 0
cups/day RR = 0.89, 95% CI: 0.64-
1.25] and tea intake [≥8 vs 0
cups/day RR = 0.77, 95% CI: 0.55-
1.08].
Neutral,
82.4%
47
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Karamanos et
al. 2014 (258)
To investigate the
association of
MedDiet with the
incidence of GDM in
Mediterranean
regions.
n = 1 003
Country: Algeria,
France, Greece, Italy,
Lebanon, Malta,
Morocco, Serbia,
Syria & Tunisia).
Inclusion: women with
oral glucose tolerance
test results, women
with/without a history
of GDM.
Exclusion: history of
T1DM or T2DM.
Questionnai
re
(validated)
& MedDiet
Index.
75g, 1 & 2-hr
OGTT (2010
International
Association in
Diabetes and
Pregnancy
Study Group
criteria)
Binary logistic regression
Adjustments: age, BMI, family
history of DM, gestational weight
gain, EI.
GDM incidence was lower in
subjects with better MedDiet
adherence, 8.0% vs 12.3% [OR =
0.62, 95% CI 0.40-0.95, P = 0.030]
by American Diabetic Association
2010 and 24.3% vs 32.8% [OR =
0.66, 95% CI: 0.50-0.87, P =
0.004] according to International
Association of Diabetes &
Pregnancy Study Group 2012
criteria.
Positive
93.2%
Osorio-Yáñez et
al. 2016 (236)
To examine the
association between
dietary Calcium
intake and risk of
GDM.
n = 3 414
Age: 32.8
Country: United
States
Study: Omega
Inclusion: >18 yrs, <20
weeks gestation, spoke
& read English,
delivered at specified
hospitals.
Exclusion: history of
DM/GDM, multi-
gestation, pregnancy
<20 weeks, iron
deficiency anaemia,
incomplete FFQ,
unrealistic levels of
total EI (<500 kcal/day
or >3500 kcal/ day).
121-item
FFQ
(validated)
100g, 3-hr
OGTT (2004
American
Diabetic
Association
criteria)
Generalized linear models with log-
link function, log Poisson regression
model and robust standard errors.
Adjustments: total energy, age,
race/ethnicity, education, smoking
status, BMI, prenatal vitamin use, PA,
family history of DM, alcohol, coffee,
SSB, red & processed meats, fatty
fish, total fiber intake & dietary
covariates (vitamin D & Mg).
Higher dietary Calcium intake
compared to lower was
inversely (though not statistically)
associated with GDM risk [RR =
0.57, 95% CI: 0.27-1.21). Calcium
intake ≥795 mg/day resulted in a
42% reduction in GDM risk when
(<795 mg/day) [R = 0.58, 95% CI:
0.38-0.90, P-value = 0.02).
Positive,
74.2%
48
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Qiu et al. 2011a
(237)
To investigate the
association of egg
intake and dietary
cholesterol & GDM
risk in a cohort study.
n = 3 158
Age (mean): 32.7 yrs
Country: United
States
Study: Omega
Inclusion: pre-natal
care <20 weeks, >18
yrs, spoke/read English,
to deliver at either of 2
study hospitals.
Exclusion: DM, multi-
gestation, incomplete
or unrealistic dietary
intake (<500 or
>3500kcal/day).
121-item
semi-
quantitative
FFQ
(validated)
100g, 3-hr
OGTT (2004
American
Diabetic
Association
criteria)
Multivariable models, generalized
linear models using a log-link
function
Adjustments: EI, age, race/ethnicity,
parity, PA, pre-pregnancy BMI,
dietary fiber, vitamin C, intake red &
processed meats, saturated fat
intake.
Higher eggs and cholesterol
intake during the pre-pregnancy
and early pregnancy period were
associated with a greater GDM
risk [RR (≥10 eggs/week) = 2.52,
95% CI: 1.11-5.72; RR (294 vs
<151 mg/day cholesterol) = 2.35,
95% CI: 1.35-4.09 respectively].
Positive,
79%
Qiu et al. 2011b
(238)
To examine the
associations of
dietary heme & non-
heme iron with the
risk of GDM.
n = 3 158
Age: 32.7 yrs
Country: United
States
Study: Omega
Inclusion: pre-natal
care <20 weeks, >18
yrs, spoke/read English,
to deliver at either of 2
selected hospitals.
Exclusion: DM, multi-
gestation, incomplete
or excessive dietary
intake (<500 or
>3500kcal/day).
121-item
semi-
quantitative
FFQ
(validated)
100g, 3-hr
OGTT (2004
American
Diabetic
Association
criteria)
Generalized linear models using a
log-link function
Adjustments: EI, age, race/ethnicity,
parity, PA, pre-pregnancy BMI,
dietary fiber, vitamin C.
Higher heme iron intake is
associated with an increased
GDM risk [RR = 1.57, 95% CI
0.95–2.61] when comparing
highest to quartile. Women who
reported very high heme iron
intake (≥1.52 mg/ day) had a
2.26-fold increased risk (95% CI
1.09–4.69) of GDM compared
with women reporting lower
levels.
Positive,
79%
Radesky et al.
2008 (245)
To report results from
an analysis of diet
quality & risk of
abnormal glucose
tolerance among a
cohort of women.
n = 1 733
Age: 32.2 (4.9) yrs
Country: United
States
Study: Project Viva
Inclusion: <20 weeks,
singleton pregnancy,
complete study forms
in English.
Exclusion: missing or
incomplete oral glucose
tolerance test & diet,
history of T2DM or
T2DM, or polycystic
ovarian syndrome.
Self-
administere
d Semi-
quantitative
FFQ
(validated)
100g, 3-hr
OGTT (2004
American
Diabetes
Association
criteria)
Multinomial regression
Adjustments: age, pre-pregnancy
BMI, race/ ethnicity, family history of
DM, history of GDM.
Alpha-linolenic acid was
associated with increased risk for
GDM [OR = 1.29, 95% CI: 1.04-
1.60) for each 300 mg/day after
adjustment for confounders &
other fats. Overall women with
GDM had higher average n-3
fatty acid intake, lower n-6/n-3
ratio, and slightly higher
polyunsaturated fat intake than
normo-glycaemic women.
Neutral,
81.4%
49
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Schoenacker et
al. 2015 (242)
To examine the
associations between
pre-pregnancy dietary
patterns & risk of
GDM.
n = 3 853
n pregnancies = 6626
Age: 28 (1.4) yrs
Country: Australia
Study: Australian
Longitudinal Study on
Women’s Health
Inclusion: Australian
women without pre-
existing DM.
Exclusion: T2DM or
T2DM, pregnant with
their first child in 2003,
did not report a live
birth at consecutive
surveys in
2006/2009/2012,
missing data, had GDM,
unrealistic EI (<2093 or
>14654kJ/d).
Questionnai
re
(validated)
75g, 1-hr
OGTT;
Self-report
(1998
Australasian
Diabetes in
Pregnancy
Society
criteria)
Generalized estimating equation,
Log-binomial models or Log-Poisson
Adjustments: age, EI, parity,
hypertensive disorders of pregnancy,
highest education, smoking status,
PA, BMI, polycystic ovarian
syndrome.
No association between fruit &
low-fat dairy or cooked
vegetables with GDM risk.
Mediterranean-style diet
associated with 15% lower GDM
risk [RR = 0.85, 95% CI: 0.76-
0.98]. Each SD increase in score
of the meats, snacks & sweets
pattern was associated with 41%
higher GDM risk [RR = 0.59, 95%
CI:1.03-1.91]. This association
was no longer statistically
significant after additional
adjustment including BMI [RR =
1.35, 95% CI: 0.98-1.81].
Positive,
42.4%
Schoenacker et
al. 2016 (241)
To determine how
much pre-pregnancy
BMI mediates the
association between
a pre-pregnancy
MedDiet &
development of
GDM.
n = 3 378
Country: Australia
Study: Australian
Longitudinal Study on
Women’s Health
Inclusion: not pregnant
at baseline and who
reported ≥1 live birth
during the 9-y follow-
up.
Exclusion: women in
rural or remote areas.
FFQ
(validated)
Self-report
(1998
Australasian
Diabetes in
Pregnancy
Society
criteria)
Linear or logistic regression
Adjustments: education, parity,
polycystic ovarian syndrome, EI and
PA.
BMI contributes 32% to the total
effects and relationship between
pre-pregnancy MedDiet and odds
of GDM [OR = 1.35, 95% CI: 1.02-
1.60].
Positive,
84.5%
50
DIET ONLY
Source Aim & Study
Population Selection Criteria
Diet
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Tobias et al.
2012 (228)
To assess usual pre-
pregnancy adherence
to well-known dietary
patterns & GDM risk.
n = 15 254
Age: 24-44
Country: United
States
Study: NHS II
Inclusion: singleton
pregnancy, no GDM
history, no history of
DM/cancer/ CVD event.
Exclusion: pregnancies
after GDM, pre-
pregnancy FFQ, left >70
FFQ items blank, or
reported unrealistic
total EI (<500 or
3500kcal/ day).
Semi-
quantitative
FFQ
(validated)
Medical
records
Multi-variable marginal logistic
using Generalized estimating
equation
Adjustments: age, EI, race/ethnicity,
PA, BMI, family history of DM,
gravidity, smoking status.
Comparing high to low dietary
adherence, the risk of GDM was
24% lower with the alternate
MedDiet score [RR = 0.76, 95%
CI: 0.60, 0.95, P-value = 0.004],
34% lower with the Dietary
Approaches to Stop Hypertension
(DASH) score [RR = 0.66, 95% CI
0.53, 0.82, P-trend = 0.0005], &
46% lower with the AHEI score
[RR = 0.54, 95% CI: 0.43- 0.68, P-
trend <0.0001].
Positive,
NA%
Zhang et al.
2006a (229)
To examine whether
pre-gravid dietary
fiber consumption
from cereal, fruit, &
vegetable sources &
dietary glycemic load
was related to GDM.
n = 13 110
Age: 24-44
Country: United
States
Study: NHS II
Inclusion: pregnant
women.
Exclusion: did not
complete FFQ in 1991,
incomplete FFQ, dietary
intake was unrealistic
total EI (500 kcal/day or
3,500 kcal/day),
multiple gestation or
history DM/cancer/CVD
or GDM.
133-item
Semi-
quantitative
FFQ
(validated)
Medical
records
Cox proportional hazards analysis
Adjustments: parity, age, BMI,
smoking status, race/ ethnicity, PA,
family history of DM & dietary
variables (total fat expressed as %
energy), cereal fiber, fruit &
vegetable fiber, alcohol
consumption, EI & glycaemic load.
Dietary total fiber & cereal & fruit
fiber were strongly inversely
associated with GDM risk. Each
10g/day increment in total fiber
intake was associated with 26%
(RR = 0.74, 95% CI: 0.51-0.91)
reduction in risk. Each 5g/day
increment in cereal or fruit fiber
was associated with a 23% (9 –
36) or 26% (5– 42) reduction
respectively.
Positive,
NA%
Zhang et al.
2006b (230)
To examine whether
dietary patterns are
related to risk of
GDM.
n = 13 110
Age: 24-44
Country: United
States
Study: NHS II
Inclusion: pregnant
women
Exclusion: did not
complete FFQ in 1991,
> 9 items blank in FFQ,
unrealistic total EI (500
kcal/day or 3,500
kcal/day), multiple
gestation.
133-item
semi-
quantitative
FFQ
(validated)
Medical
records
Cox proportional hazards analysis
Adjustments: parity, age, BMI,
smoking status, race/ ethnicity, PA,
family history of diabetes & dietary
variables including total fat (%
energy), cereal fiber, alcohol intake,
total EI & glycaemic load.
Comparing the highest with the
lowest quintile of the Western
pattern scores, RR = 1.63 (95% CI:
1.20–2.21, P = 0.001) &
conversely comparing the lowest
with the highest quintile of the
prudent pattern scores, RR = 1.39
(95% CI: 1.08–1.80, P = 0.018).
Positive,
NA%
51
PHYSICAL ACTIVITY ONLY
Source Aim & Study Population Selection Criteria
Physical
Activity
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Badon et al.
2016 (234)
To investigate the
associations of Leisure
Time Physical Activity
(PA) before and during
pregnancy with GDM risk.
n = 3 449
Age: 32.6 (SD 4.4)
Country: United States
Study: Omega
Inclusion: >18 yrs,
speak & read in English
language, prenatal care
<20 weeks gestation,
deliver at allocated
hospitals.
Exclusion: Pre-
pregnancy or early
pregnancy PA of ≥35
metabolic equivalents
(MET-hrs/week),
missing data on PA, had
prior T1/T2DM.
Questionnaire
(Invalidated)
100-g, 3-hr
OGTT (1997
American
Diabetic
Association
criteria)
Multivariable Poisson regression
Adjustments: age, race, education,
marital status, nulliparity, pre-
pregnancy BMI category, gestational
weight gain, smoking during
pregnancy, alcohol use during
pregnancy & year of study
enrollment.
Leisure time PA during both
pre-pregnancy and early
pregnancy was associated with
a 46% reduced risk of GDM [RR
= 0.54, 95% CI: 0.32-0.89] when
compared with inactivity.
Neutral,
NA
Chasan-Taber
et al. 2008
(251)
To determine whether PA
during pregnancy reduces
the risk of GDM in
Hispanic women.
n = 1006, (710 for mid-
pregnancy data)
Age: 16-40 yrs
Country: United States
Inclusion: age 16-40
yrs, <24 weeks
gestation.
Exclusion: Non-
Hispanic, T2DM,
hypertension, heart
disease, chronic renal
disease, medications
that influence glucose
tolerance, multi-
gestation & previous
participation in the
study.
Kaiser PA
Survey &
Pregnancy PA
Questionnaire
(validated in a
pregnant
population)
100g, 3-hr
OGTT (2004
American
Diabetic
Association
criteria),
medical
records
Logistic regression
Adjustments: age & BMI.
Higher levels of household/
caregiving activity in early (OR =
0.2, 95% CI: 0.1-0.8, P-trend =
0.03) & mid-pregnancy (OR =
0.2, 95% CI: 0.1-0.8, P-trend =
0.004) were associated with a
reduced risk of GDM. Higher
level of total PA was also
associated with reduced odds
of GDM (OR = 0.4, 95% CI: 0.1-
1.2, P-trend = 0.06).
Positive,
81.7%
52
PHYSICAL ACTIVITY ONLY
Source Aim & Study Population Selection Criteria
Physical
Activity
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Chasan-Taber
et al. 2014
(250)
To examine the
relationship between PA
during pre, early & mid
pregnancy & risk of
abnormal glucose
tolerance & GDM.
n = 1241
Age: 16-40 yrs
Country: United States
Study: Proyecto Buena
Salud
Inclusion: born in the
Caribbean Islands or
had a parent or ≥2
grand-parents born in
the Caribbean Islands.
Exclusion: history of
DM/hypertension/heart
or renal disease, <16 or
>40 yrs old, multi-
gestation or
medications that
influence glucose
tolerance.
Pregnancy PA
Questionnaire
(validated in
pregnant
women)
100g, 3-hr
OGTT (2004
American
Diabetic
Association
criteria);
medical
records
Logistic regression
Adjustments: age, BMI, gestational
weight gain, education level,
generation in the United States.
Women in the top quartile of
moderate intensity PA in early
pregnancy had a 52%
decreased risk of abnormal
glucose result when compared
to the lowest quartile [OR =
0.48, 95% CI: 0.27-0.88, P-trend
= 0.03]
Positive,
76.3%
Currie et al.
2014 (254)
To examine if physical
activity in the year pre-
pregnancy & in the first
half of pregnancy is
associated with maternal
& neonatal outcomes.
n = 1 749
Age: 31 (mean)
Country: Canada
Exclusion: >20 weeks
gestation, pre-existing
DM, early pregnancy
loss or pregnancy
termination, any
missing information,
contraindications to PA
present before 20
weeks gestation.
Kaiser PA
Survey
(validated in
pregnant
women)
50g, 1-hr
GCT or 100g
1 & 2-hr
OGTT,
Medical
records
Logistic regression
Adjustments: age, pre-pregnancy
BMI, education, parity, & history of
GDM.
Relative to the lowest tertile of
pre-pregnancy household PA,
women in the middle & the
highest tertiles were at
decreased risk of GDM [OR =
0.29, 95% CI: 0.12 – 0.74 & OR
= 0.33, 95% CI: 0.12 - 0.88]
respectively, albeit statistically
insignificant.
Positive,
79.5%
Dempsey et
al. 2004 (235)
To examine the
relationship between
recreational PA before &
during pregnancy & risk
of GDM.
n = 909
Country: United States
Study: Omega
Inclusion: <16 weeks
gestation
Exclusion: <18 yrs, did
not speak/read English,
did not carry to term, if
they did not plan to
deliver at the selected
hospitals.
Questionnaire
(Invalidated)
100-g, 3-hr
OGTT (1997
American
Diabetes
Association
criteria),
Medical
records
Generalized linear models using a
log-link function
Adjustments: maternal age, race,
parity, & pre-pregnancy BMI.
Compared with those who
were inactive, women who
participated in any recreational
PA in the pre-pregnancy period,
had a 56 % GDM risk reduction
(RR = 0.44, 95% CI: 0.21 - 0.91).
Women who engaged in PA
before & during pregnancy had
a 69% GDM reduced risk (RR =
0.31, 95% CI: 0.12, 0.79).
Positive,
90.9%
53
PHYSICAL ACTIVITY ONLY
Source Aim & Study Population Selection Criteria
Physical
Activity
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Dye et al.
1997 (247)
To determine whether
exercise has a preventive
role in the development
of GDM in women living
in central New York State
on a population-based
birth registry.
n = 12 799
Country: United States
Inclusion: women that
delivered a livebirth
within the New York
State between
1/10/1995-31/07/1996.
Exclusion: conditions
that affect exercise (e.g.
heart disease, multi-
gestation, incompetent
cervix, previous
preterm delivery & low
birth weight infant &
chronic hypertension).
Personal
interview
Medical
records
Chi-square statistics, Logistic
regression
Adjustments: age, race, parity, pre-
pregnancy BMI, gestational weight
gain & insurance coverage.
When stratified by pre-
pregnancy BMI category,
exercise was associated with
reduced rates of GDM only
among women with a BMI >33
[OR = 1.9, 95% CI: 1.2-3.1].
Positive,
89.1%
Iqbal et al.
2007 (255)
To identify lifestyle
predictors of GDM in
South Asian women.
n = 611
Age: 29.4 (4.7)
Country: Canada
Inclusion: women of
South Asian origin, ≤18
weeks of gestation &
did not have known
diabetes.
Exclusion: missing data,
terminating a
pregnancy, refusing
oral glucose tolerance
test.
Interviewer
administered
Monitoring
Trends &
Determinants
of
Cardiovascula
r Disease
(Monica)
Optional
Study of PA,
(Validated)
100g, 3-hr
OGTT (2004
American
Diabetic
Association
criteria)
Logistic regression
Adjustments: age, family history of
DM, education, height, parity BMI,
PA level (kcal/day) & rate of weight
gain/week.
Increase in PA (100 kcal),
decreased the risk of GDM by
11% [OR = 0.89, 95% CI: 0.79-
0.99].
Positive,
81.6%
54
PHYSICAL ACTIVITY ONLY
Source Aim & Study Population Selection Criteria
Physical
Activity
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Morkrid et al.
2007 (256)
To assess the association
between objectively
recorded PA in early
gestation & GDM
identified at multiethnic
cohort.
n = 759
Age: 29.9 (4.4)
Country: Norway
Study: Stork Groruddalen
Study
Inclusion: lived in one
of the selected districts,
to give birth in one of
the 2 selected
hospitals, <20 weeks
gestation, could speak
one of the 9 listed
languages & to provide
written consent.
Exclusion: known
diabetes or other
diseases requiring
frequent hospital visits.
Questionnaire
(validated)
75g, 2-hr
OGTT
(amended
2010
Internationa
l Association
of Diabetes
&
Pregnancy
Study Group
criteria)
Logistic regression
Adjustments: ethnic origin, weeks
gestation, age, parity, & pre-
pregnancy BMI.
Significant associations
between the following 3
components GDM risk:
objectively recorded steps/day
in early gestation [OR = 0.79,
95% CI: 0.65 –0.97], self-
reported regular PA before
pregnancy [OR = 0.66, 95% CI
0.46-0.94] & self-reported
aerobic PA ≥ 150 min/week 3
months before pregnancy [OR =
0.69, 95% CI: 0.49-0.97].
Positive,
92.2%
Oken et al.
2006 (244)
To examine the
associations of PA &
television viewing before
& during pregnancy, with
risk for GDM & abnormal
glucose tolerance.
n = 1 805
Age: 32.1 (5.0)
Country: United States
Study: Project Viva
Exclusion: history of
T1DM or type 2
diabetes no
measurement of
blood glucose levels
during pregnancy, no
data on PA or TV
viewing, no records of
pre-pregnancy BMI.
Questionnaire
; modified
from the
leisure time
activity
section of the
PA Scale for
the Elderly
(validated on
an elderly
population).
100g, 3-hr
OGTT (2004
American
Diabetic
Association
criteria)
Logistic regression Adjustments: age,
race/ethnicity, pre-pregnancy BMI,
history of GDM in a previous
pregnancy, & mother’s history of
DM.
Vigorous activity during the
year before pregnancy reduced
the risk of GDM by 44% [OR =
0.56, 95% CI: 0.33-0.95].
Vigorous activity before
pregnancy & light-to-moderate
or vigorous activity during
pregnancy appeared to reduce
the risk of GDM [OR = 0.49,
95% CI: 0.24-1.01].
Positive,
84.8%
55
PHYSICAL ACTIVITY ONLY
Source Aim & Study Population Selection Criteria
Physical
Activity
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Putnam et al.
2013 (248)
To determine association
between daily physical
activity & pregnancy &
neonatal outcomes in
stay at home military
wives.
n = 190
Age: 28.3 (5.5)
Country: United States
Inclusion: unemployed,
married to an active-
duty or reserve service
member, aims to
complete prenatal care
& delivery within the
specified medical
facility.
Exclusion: preexisting
hypertension/ DM or
thrombophilia, multiple
gestation, or history of
preterm delivery.
Validated
questionnaire
describing
their
domestic PA
on a typical
day during
the previous 4
weeks (1st
trimester).
100g, 3-hr
OGTT, no
further
information
Logistic regression Adjustments:
maternal BMI at first visit & delivery,
number of children at home,
gravidity, & parity.
Highest incidence rate of GDM
occurred in the group with the
least average daily energy
expenditure (P = 0.025).
Positive,
NA%
Rudra et al.
2006 (239)
To examine the relation
between perceived
exertion & GDM within
sub-groups of women
categorize by energy
expenditure.
n = 897
Country: United States
Study: Omega
Inclusion: women who
initiated prenatal care
before 16 weeks
gestation.
Exclusion: <18 yrs, did
not speak/read English,
did not plan to carry
the pregnancy to term,
or did not plan to
deliver at either of the
specified hospitals.
Stanford 7-
Day PA Recall
& the
Minnesot
Leisure-Time
PA
Questionnaire
, (validated
among men &
non-pregnant
women).
Medical
records
Logistic regression models
Adjustments: age, race/ethnicity pre-
pregnancy hypertension, nulliparity,
& pre-pregnancy BMI.
Women reporting strenuous &
very strenuous maximal
exertion had 37% [OR = 0.63,
95% CI: 0.31-1.29] & 43% [OR =
0.57, 95% CI: 0.24-1.37] lower
risk of GDM respectively, when
compared with negligible-
moderate exertion. Women
reporting ≥15.0 MET-
hours/week experienced 86%
GDM risk reduction when
compared to inactive women
[OR = 0.14, 95% CI: 0.05-0.38].
Neutral,
89.7%
56
PHYSICAL ACTIVITY ONLY
Source Aim & Study Population Selection Criteria
Physical
Activity
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Solomon et al.
1997 (227)
To assess whether
recognized determinants
of NIDDM may also be
markers for increased risk
of GDM.
n = 14 613
Age: 25-42 yrs
Country: United States
Study: NHS II
Inclusion: no history of
GDM or diabetes,
singleton pregnancy
between 1990 & 1994,
pregnancy lasting >6
months.
Exclusion: multiple
pregnancy.
PA (1989) -
assessed as
average MET
expenditures.
In 1991 -
women were
questioned
about the
number of
times /week
they engaged
in PA to
perspire
heavily.
Medical
records
Logistic regression
Adjustments: age, BMI & parity.
No association between total
MET score in 1989 &
subsequent GDM risk. GDM risk
appeared slightly lower with
frequent participation in
vigorous PA, albeit statistically
insignificant [RR(≥4/week) =
0.78, 95% CI: 0.47-1.29].
Neutral,
NA%
Van der Ploeg
et al. 2011
(243)
To examine the
relationships between
PA, sedentary behavior &
the development of GDM
n = 3 529
Age: 24-34 yrs
Country: Australia
Study: Australian
Longitudinal Study on
Women’s Health
Inclusion: Women in
Australia.
Exclusion: T1DM, type
2 diabetes were
pregnant at the second
survey, were pregnant
with their first child at
the third survey or did
not have a live-birth
between survey 2 & 3.
Australian
Longitudinal
Study on
Women’s
Health
modification
of the 7-day
recall Active
Australia
questionnaire,
non-validated
75 g, 2-hr
OGTT, self-
reported
(1998
Australasian
Diabetes in
Pregnancy
Society
criteria).
Generalized estimating equations
Adjustments: EI, overweight &
obesity, age, BMI, parity, age at birth
of first child, country of birth, &
education.
Neither total PA nor sedentary
behavior were associated with
the risk of GDM. Analyses for
self-reported vigorous PA
showed no significant
relationships with the
development of GDM, with
OR=1.23 [95% CI 0.83-1.81] &
OR=0.95 [95% CI 0.62-1.46] for
1–90 min/week & >90
min/week, respectively.
Neutral,
S2: 82.0%
S3: 75.3%
57
PHYSICAL ACTIVITY ONLY
Source Aim & Study Population Selection Criteria
Physical
Activity
Assessment
Method
Diagnostic
method for
GDM
Statistical Analysis & Adjusted
factors
Selected Main Findings (RR, OR
etc.)
Quality
Rating,
Retention
Zhang et al.
2006c (231)
To assess whether the
amount, type, & intensity
of pre-gravid PA &
sedentary behaviors are
associated with GDM risk.
n = 21 765
Age: 24-44 yrs
Country: United States
Study: NHS II
Inclusion: singleton
pregnancy lasting 6
months or longer.
Exclusion: history of
GDM/diabetes/cancer
or cardiovascular
disease, were pregnant
in 1989 questionnaire,
no PA data, multiple
gestation.
Questionnaire
(validated)
Medical
records
Cox proportional hazards analysis
Adjustments: parity, nulliparous
women, age, smoking status, race or
ethnicity, family history of diabetes &
dietary variables (total fat, % energy,
cereal fiber, alcohol, GI, total EI) &
BMI.
Highest quintile of vigorous PA
significantly reduced the risk of
GDM by 23%, when compared
to the lowest quintile [RR=0.77,
95% CI 0.69-0.94, P-
trend=0.002].
Positive
69.8%
Abbreviations: AHEI-2010 – Alternative Healthy Eating Index – 2010, BMI – Body mass index, CHO – Carbohydrates, CIs – confidence intervals, CVD – Cardiovascular Disease, DM – Diabetes Mellitus, EI – Energy Intake, FFQ – food frequency
questionnaire, GCT – Glucose Challenge Test, GDM – gestational diabetes mellitus, GI – Glycemic index MedDiet – Mediterranean Diet, MET - metabolic equivalent, NA – Not Available, NHS I/II – Nurse’s Health Study I or II, NIDDM – non-insulin
dependent diabetes mellitus, OGTT – Oral Glucose Tolerance Test, OR – odds ratio, PA – physical activity, RCT – randomised controlled trial, RR – relative risk, SD – Standard Deviation, SE – Standard Error, SES – socioeconomic status, SSB – Sugar
sweetened beverage, T1/T2DM – Type 1 or Type 2 Diabetes Mellitus,
58
2.3.3 Diet Related Studies
To account for the diversity of 25 dietary studies identified after inclusion, we categorised them into
one of seven themes: carbohydrates, fat intake, protein, fast food intake, caffeine, calcium intake
and commonly recognised dietary patterns. The predominating dietary collection method was a
validated Food Frequency Questionnaire (FFQ) (n = 23) (219-226, 228-230, 236-238, 240-242, 252,
257, 258). The remaining two studies used a rapid food screener (246) and an interview (253) to
collect dietary data. Studies focusing on early pregnancy collected dietary data <22 weeks into
pregnancy. When studies reporting on diet and PA were compared with respect to adjusted
confounding variables (Figure 2.2), we observed that age, BMI and parity were most common in
both. Only 70% of diet related and 10% PA studies adjusted for energy intake.
Figure 2.2. Confounding variables that were adjusted for in studies collecting information on dietary intake (white bars) and physical activity levels (blue bars). Age, BMI and parity were most commonly adjusted confounding variables in observational studies reporting on either diet or physical activity.
0
10
20
30
40
50
60
70
80
90
100
Pro
po
rtio
n o
f st
ud
ies
wit
h a
dju
sted
co
nfo
un
din
g va
riab
les
(%)
Confounding variables
59
2.3.3.1 Carbohydrates (Fruit, Fiber, Beverages, Potato)
Five studies reported on high carbohydrate foods, including fruit, fiber, potato and beverage intake
and their respective associations to GDM risk (221, 222, 225, 226, 229). High pre-pregnancy fruit
intake was not associated with an increase in GDM risk (RR high vs low intake = 0.93, 95% CI: 0.76 -
1.16) (226), however fruit fiber (229) was reported to be protective (RR fruit fiber = 0.66, 95% CI:
0.51 - 0.86). Although high compared to low apple intake suggested non-significant protection from
GDM risk (RR apple = 0.81, 95% CI: 0.65 - 1.01), the overall trend across quintiles of apple
consumption reached statistical significance (P-trend <0.05) (226). Protective effects were also
evident when consumption of total dietary fiber and cereal fiber were examined (RR total fiber =
0.67, 95% CI: 0.51 - 0.90; RR cereal fiber = 0.76, 95% CI: 0.59 -0.99) (229).
Higher frequency of potato intake increased the risk of GDM (RR high vs low frequency intake = 1.62,
95% CI: 1.24 - 2.13) (222). However, frequent consumers of potato tended to be current smokers,
had higher BMI and lower diet quality as assessed by the Alternate Healthy Eating Index (AHEI) 2010
score (222). In contrast, a study by Karamanos and colleagues found that women who went on to
develop GDM consumed less potatoes and cereals than those that did not develop it (258). Replacing
two servings of potatoes per week for other vegetables types, legumes or wholegrain foods resulted
in a 9%, 10% and 17% GDM risk reduction, respectively (222). No significant association was observed
between potato crisps or corn chips and GDM risk after adjustment of confounding variables
including age, parity, race, family history of diabetes, smoking, PA, energy intake, diet quality and
BMI (222).
The relationship between 100% fruit juice consumption and GDM onset was nonlinear, with the
lowest risk observed in women with moderate fruit juice intake (226). In contrast, higher sugar
sweetened beverage (SSB) intake was associated with GDM risk (RR ≥5 week = 1.23, 95% CI: 1.05 -
1.45, P-value = 0.005) (225). When different sub-types of SSB were taken into account, the strongest
60
association was observed for sugar sweetened cola (RR high vs low intake = 1.29, 95% CI: 1.07 - 1.55)
but not for non-cola SSB (RR high vs low = 0.99, 95% CI: 0.78 - 1.25) (225).
2.3.3.2 Fat Intake (i.e. Total, Monounsaturated Fatty Acids,
Dietary Cholesterol, Egg Intake)
Higher intake of animal, cholesterol and monounsaturated fatty acids (MUFA) were significantly
associated with increased risk of GDM (224). When comparing highest to lowest quintile of animal
fat intake (%EI), the risk increased by ~90% (RR = 1.88, 95% CI: 1.36 - 2.60) (224). Similarly, a
comparison between the highest and lowest quintile of cholesterol intake elucidated a positive
relationship with GDM risk (RR = 2.35, 95% CI: 1.35, 4.09). On the contrary, Baptise-Roberts and
colleagues reported no association between either cholesterol or total fat intake with a high glucose
response following a glucose challenge test (246).
While no associations were observed between total omega-3 or total omega-6 fatty acids and risk of
GDM in one study (224), another noted that women who developed GDM had a lower n-6/n-3 ratio,
a higher intake of n-3 fatty acid and polyunsaturated fats than their non-GDM counterparts (245).
Each 300 mg/day intake of alpha-linolenic acid, was associated with an increased risk for GDM, with
an OR = 1.29 (95% CI: 1.04 - 1.60) (245). Karamanos and colleagues reported that olive oil was
consumed in higher quantities in women who went on to develop GDM compared to those that did
not (258), however no association analysis was presented (258). With regards to egg consumption,
one study suggested that high intakes increased the risk of GDM by a 1.77 fold (237), while another
found no such association (219).
61
2.3.4.3 Protein Intake (i.e. Meat, Iron, Heme)
Bao and colleagues reported that an intake of protein from animal origin increased the risk of GDM
by ~50%, whereas an intake of protein sourced from vegetables was protective by 30% (219).
Similarly, a low carbohydrate dietary pattern with high animal protein and animal content was
associated with a 36% increase risk of GDM, whereas a low carbohydrate diet containing high intake
of plant-sourced protein and fat was not associated with any increased risk (220). Replacing 5%
energy of animal protein for protein of plant origin reduced GDM risk by 51% (219).
Neither a high meat score before pregnancy calculated using The Rapid Food Screener (246) or a high
red meat intake in early pregnancy (245) were able to predict GDM risk in the two studies. On the
contrary, three studies concluded that women with a high pre-pregnancy red meat intake had
between 1.4-2.0 times the risk of developing GDM (219, 230, 242). There are similar inconsistent
findings for processed meat intake and GDM risk. Two studies reported a statistically significant
increased risk for GDM, ranging between 48-68% during the pre-pregnancy period (219, 230),
whereas the remaining study found no such association (245). A positive relationship was observed
between higher pre-pregnancy iron or heme intake and GDM (223, 238). While Behboudi-Gandevani
and collegues (252) observed no statistically significant differences in iron or zinc intake in women
with or without GDM, women with GDM did have a statistically significant higher serum iron level in
early pregnancy.
2.3.3.4 Caffeine
Two studies (233, 253) reported on caffeine intake and risk of GDM. Whilst both captured caffeine
intake during the pre-pregnancy period, Hinckle and colleagues additionally looked at tea intake
(253). Coffee consumption was reported to have a protective affect against GDM in one study [RR =
62
0.48 (95% CI: 0.28 - 0.82)] (233), but failed to reach a statistical significance in the other (RR ≥8 vs 0
cups/day = 0.89, 95% CI: 0.64 - 1.25) (253). Consumption of decaffinated coffee was not associated
with risk reduction (233). Increasing frequency of tea intake indicated a potential protective effect
against GDM risk, albeit statistically insignificant (RR ≥8 vs 0 cups/day = 0.77, 95% CI: 0.55 - 1.0) (253).
2.3.3.5 Fast Food Intake
Increasing frequency of fast food intake prior to pregnancy was associated with a statistically
significant increased risk (221) or incidence (257) of GDM. The reported RR for ≥7/week vs <1/week
= 2.18, 95% CI: 1.53 - 3.09) (221) and Odds Ratio (OR) for highest vs lowest frequency intake = 1.86,
95% CI: 1.13 - 3.06) (257). Women with greater fast food consumption were typically younger,
current smokers, multiparous, less physically active and followed diets that were either less adherent
to the Mediterranean Diet (MedDiet) pattern (257) or had an overall lower AHEI-2010 diet quality
score (221).
2.3.4.6 Calcium/Dairy Intake
Total pre-pregnancy dairy intake was not associated with risk of GDM (219, 236). Habitual maternal
intake of low-fat dairy suggested a non-significant inverse association with GDM risk (RR highest vs
lowest quintile = 0.57, 95% CI: 0.32 - 1.02) (236), however the overall trend across quartiles of low-
fat dairy intake reached statistical significance (P-trend <0.05). Interestingly, Schoenacker and
colleagues (2015) observed fruit and low-fat dairy as a dietary pattern but found no association with
GDM risk (242). With respect to calcium intake, an inverse association with GDM risk was observed,
albeit statistically insignificant (236). When quintiles were grouped into higher vs lower level of
63
intake, women who consumed ≥795 mg Calcium/day had a 42% GDM risk reduction when compared
to those who had <795mg/day (236).
2.3.4.7 Recognised Dietary Patterns
MedDiet was the most consistently reported protective dietary pattern against GDM risk, reaching
statistical significant in all four studies (228, 241, 242, 258). In a comprehensive review by Radd-
Vagenas, MedDiet is defined as a diet containing higher bread, cereal, legume, vegetable, fruit, fish
and olive oil intake and smaller or limited intake of animal fat, meat and eggs (261). The extent of
MedDiet protectiveness ranged from 15-38%. Although women with better MedDiet compliance had
lower incidence of GDM than their non-compliant counterparts, Karamanos and colleagues reported
that GDM incidence greatly differed when comparing ADA and IADPSG diagnostic criteria between
the compliant groups (8% vs 24%, respectively) (258).
Adherence to a diet with a high AHEI 2010 score was associated with a reduced risk of GDM by 19%
(232) or 46% (228). When additional lifestyle factors were taken into account such as regular PA,
normal BMI, non-smoker, the association with risk reduction was 83% (232). Similarly to the AHEI
scoring system, some studies used an Australian Recommended Food Score (ARFS) (240) or Dietary
Approaches to Stop Hypertension (DASH) score (228). A high ARFS was not associated with GDM risk
(240), whereas a greater DASH diet compliance was associated with a 34% GDM risk reduction (228).
With respect to Prudent and Western diets, there were some conflicting results. Whilst compliance
to a Prudent or Western diet resulted in a negative and positive association with GDM risk
respectively in one study (230), a second study observed no such relationships (245).
64
2.3.5 Physical Activity
The relationship between PA and risk of GDM was examined by 17 publications, including the Nurse’s
Health Study II (NHS II, n = 3), OMEGA Study (n = 3), Projecta VIVA (n = 1), the Australian Longitudinal
Study on Women's Health (n = 1). Data collection methods included interviews (n = 2), questionnaires
(n = 14, of which 12 were validated) and a self-report (n = 1, also validated). PA levels were captured
during the pre-pregnancy (n = 10) and early pregnancy (n = 9) stages.
Overall, PA was reported to be protective against GDM in 13 of 17 studies and the degree of
protection generally increased with greater levels of PA. Eleven studies reported that PA before
pregnancy was beneficially associated with reduced risk by 22-86% (231, 232, 234, 235, 239, 244,
246, 251, 256), with only two studies not reaching statistical significance (227, 254). The degree of
potential protection depended on type and duration of PA. Similarly, ten studies that assessed early
pregnancy PA levels also reported a reduction in GDM risk with higher PA, ranging between 11-52%,
however two studies not reaching statistical significance (235, 243). When both pre-pregnancy and
early pregnancy PA levels were taken into account, there was an even lower risk of GDM observed
(RR = 0.31, 95% CI: 0.12 - 0.79) (235).
The most apparent associations between PA and GDM risk were observed in 13 studies reporting on
Leisure Time PA (LTPA). Of these, ten reported a significant reduction in GDM risk (231, 232, 234,
235, 239, 244, 246, 250, 255, 256). Two studies that examined LTPA volumes, suggested ≥150
min/week (256) or ≥ 210 min/week (232) as being sufficient to reduce the risk of GDM. Higher leisure
activity score (246) before pregnancy was associated with a 68% reduced risk of a high 1-hr glucose
challenge test. In terms of intensity-weighted PA volume (Metabolic Equivalent (MET) hours/week),
the beneficial association of PA in the year before pregnancy were observed at ≥ 15 (239) or ≥ 21
MET hours/week (235). Solomon et al. (227) and Van der Ploeg et al. (243) suggested a lower GDM
risk with higher frequency and volume of vigorous PA, but this association did not reach statistical
significance. In contrast, one study (247) reported that LTPA was associated with reduced rates of
65
GDM only among women in the obese pre-pregnancy BMI category. Engaging in LTPA (234) before
and during pregnanacy was associated with a 46% GDM risk reduction.
The three studies that reported on total PA levels suggested that higher total PA volume (251, 256)
or score (254) was associated with lower risk of GDM (251, 254, 256), although one did not reach
significance (254) and another was borderline statistically significant (251). Two studies suggested
that higher domestic (e.g. child and elderly caregiving, meal preparation, cleaning, shopping,
gardening) PA levels were associated with lower risk of GDM (251, 254), although one did not reach
statistical significance (254). Putman and colleagues observed that GDM risk was highest among
women with the least average daily energy expenditure (≤2200 kcal) (248).
2.3.5.1 Meta-analysis and Assessment of Bias
Of 17 PA studies, 16 were suitable for meta-analyses (227, 231, 232, 234, 235, 239, 243, 244, 246-
248, 250, 251, 254-256) as they reported associations with sufficient statistical evidence. We were
able to test a priori different indicators of PA and results were consistent. Engaging in any type of PA
compared to none during the pre-pregnancy period was associated with approximately 30% reduced
odds of GDM (OR = 0.70, 95% CI = 0.57 - 0.85; I² = 52% (medium), P-value = 0.0006), whereas engaging
in any PA early in pregnancy suggested reduced odds of GDM by 21% (OR = 0.79, 95% CI = 0.64 -
0.97, I² = 26% (low), P-value = 0.03) as evident in Figure 2.3. Taking part in any LTPA compared to
none either before (OR = 0.65, 95% CI = 0.43 - 1.00; I² = 90% (high), P-value = 0.05) or during early
pregnancy (OR = 0.69, 95% CI = 0.50 - 0.96; I² = 15% (low), P-value = 0.03) suggested a beneficial
association with GDM (Figure 2.4), with the prior not achieving statistical level of significance.
When comparing the studies reporting pre-pregnancy LTPA in MET.hr/week, our analysis suggested
that >15 MET.hr/week was associated with 48% reduced odds of GDM (OR = 0.52, 95% CI = 0.27 -
1.00; I² = 95%, P-value = 0.05) (Figure 2.5). Taking part in approximately >90 min/week in LTPA before
66
pregnancy was associated with 46% reduced odds of GDM (OR = 0.54, 95% CI = 0.34 - 0.87; I² = 70%
(medium), P-value = 0.01) (Figure 2.6). It was not possible to perform a meta-analysis for the early
pregnancy period for any LTPA indicator due to insufficient number of studies.
Tests revealed a variable degree of heterogeneity ranging from 15-95%. Based on an almost
symmetrical distribution of data points in funnel plots Figure 7A (n studies = 10, z = -1.52, p = 0.13)
and 7B (n studies = 10, z = -0.65, p = 0.52), there was no evidence of publication bias. Remaining
funnel plots are presented in supplementary materials section as they contained <10 studies in their
analyses. This includes the following figures found in the Appendix section of this thesis Figure A1a
(n studies = 9, z = -1.90, p = 0.06) with some symmetry present and Figure A1b (n studies = 6, z = -
2.96, p = 0.003) and Figure A1c (n studies = 4, z = -2.34, p = 0.02) suggesting presence of asymmetry
and potentially publication bias. The raw values of back-transformed natural log ORs are shown in
the Appendix section of this thesis in Table A1.
67
3A.)
3B.)
Figure 2.3. Meta-analysis of participation in any physical activity (PA) versus none and odds of gestational diabetes (GDM). Estimates are expressed as odds ratios (OR) with their corresponding 95% confidence intervals, however x-axis uses lnOR scale. A.) Engaging in any PA before pregnancy suggested 30% reduced odds of GDM (OR = 0.70, 95% CI = 0.57 - 0.85; I² = 52% (Medium), P-value = 0.0006). B.) Engaging in any PA during early pregnancy suggested 21% lower odds of GDM (OR = 0.79, 95% CI = 0.64 - 0.97, I² = 26% (low), P-value = 0.03).
68
4A.)
4B.)
Figure 2.4. Meta-analysis of participation in high versus low level of leisure time physical activity (LTPA) and odds of gestational diabetes (GDM). Estimates are expressed as odds ratios (OR) with their corresponding 95% confidence intervals, however x-axis uses lnOR scale. A.) Engaging in any LTPA before pregnancy suggested possible reduced odds of GDM (OR = 0.65, 95% CI = 0.43 - 1.00; I² = 90% (high), P-value = 0.05). B.) Engaging in any LTPA during early pregnancy suggests reduced odds of GDM (OR = 0.69, 95% CI = 0.50 - 0.96; I² = 15% (low), P-value = 0.03).
69
Figure 2.5. Meta-analysis of participation in high versus low level of leisure time physical activity (LTPA) before pregnancy in metabolic equivalents (MET.hr/week) and odds of gestational diabetes (GDM). Estimates are expressed as odds ratios (OR) with their corresponding 95% confidence intervals, however x-axis uses lnOR scale. Taking part in ~ >15 MET.hr/week suggested 52% reduced odds of GDM (OR = 0.52, 95% CI = 0.27 - 1.00; I² = 95%, P-value = 0.05). Due to insufficient number of studies reporting on LTPA in MET.hr/week in early pregnancy, a meta-analysis could not have been performed.
Figure 2.6. Meta-analysis of high versus low level of leisure time physical activity (LTPA) before pregnancy reported in hr/week and odds of gestational diabetes (GDM). Estimates are expressed as odds ratios (OR) with their corresponding 95% confidence intervals, however x-axis uses lnOR scale. Longer hours (>90 min/week) of LTPA/week reduced the odds of GDM by 46% (OR = 0.54, 95% CI = 0.34 - 0.87; I² = 70% (medium), P-value = 0.01). Due to insufficient number of studies reporting on LTPA in hr/week in early pregnancy, a meta-analysis could not have been performed.
70
A.)
B.)
Figure 2.7. Assessing the risk of publication bias using funnel plots for different meta-analyses. A.) Any type pre-pregnancy physical activity (PA) versus none (n studies = 10, z = -1.52, p = 0.13). B.) Pre-pregnancy leisure time PA (LTPA), comparing high versus none regardless of units reported (n studies = 10, z = -0.65, p = 0.52) Due to insufficient number of studies reporting on early pregnancy period, a funnel plot could not have been performed for some meta-analyses.
71
2.4 Discussion
The present systematic review identified 40 studies reporting on the relationship between diet or PA
and subsequent risk of GDM. It is the first review to examine a range of specific dietary factors and
indicators of PA, with a view to shedding light on the relative importance of these crucial lifestyle
health behavior factors. We identified more observational studies than previous reviews (262, 263),
including those that additionally reported on PA levels, thereby providing a more comprehensive
overview of lifestyle factors involved in the development of GDM. In addition, we visually depicted
different types of confounding variables that were adjusted for in each individual study (Figure 2).
While age, BMI and parity topped the list, we discovered that only one-in-three dietary studies
adjusted for energy intake. This raises concerns about interpretation of data as many nutrients are
associated with energy intake (264).
2.4.1 Diet and GDM Risk
Our present study investigated consumption of different types of beverages including fruit juice, SSB,
coffee and tea intake in relation to GDM risk. While coffee consumption appeared to be protective
against GDM in one study, SSB consumption resulted in a statistically significant positive association.
There are now concerns over excessive SSB intake, particularly due to the reported association with
obesity and risk of chronic diseases (265). At the population level, there has been a decline in the
overall SSB intake in several countries (265-267) over the same time frame, but there may be
segments of the population such as young adults who continue to consume SSB in high amounts
(268). Added sugars from SSB are likely to be part of a poorer quality diet and lifestyle (269). In one
study (225), women with higher intake of SSB tended to have a diet lower in total dietary fiber, fruits
and vegetables prompting the need to focus on a dietary pattern and quality in understanding any
associations with GDM risk.
72
In particular, several studies from our literature search indicate that a MedDiet may be protective
and possibly reduce the risk of GDM by 15-38%. The protective association may extend to reducing
future risk of type 2 diabetes (by 19-23%), increasing chance of remission (by 49%) (270) and
protection against cardiovascular diseases with extra virgin olive oil and plant-based dietary
components in the diet (261). There are however, drawbacks in defining what specifically constitutes
a ‘traditional’ MedDiet due to variability among different regions of the Mediterranean. At present,
a ‘traditional’ MedDiet is characterized by larger quantities of fruits, vegetables, legumes, nuts,
unprocessed cereals and grains, extra virgin olive oil, moderate fish and wine and small amount of
meat intake with low amounts of discretionary foods (261). High red or processed meat consumption
before pregnancy was associated with an increased risk of GDM. In fact, two meta-analyses
conducted in a healthy adult population found that processed meat intake was associated with a
greater risk of coronary heart disease (42%) and type 2 diabetes (19-32%) (271, 272). The proposed
mechanism of coronary heart disease and type 2 diabetes onset include excess sodium and oxidative
stress due to high levels of iron and advanced glycation end products (272) but warrants further
discussion. Given the traditional MedDiet is low in meat consumption, this could be one of the
reasons why the diet persistently provides health benefits across different age groups and stages of
life.
The risk of GDM in an Australian population following a MedDiet was partly (32%) mediated by pre-
pregnancy BMI (241). While it cannot be denied that MedDiet provides multiple health benefits, the
extent to which BMI explains the association with GDM comes as no surprise as obesity promotes
insulin resistance (273). In fact, a study by Janssen et al. (274) reported significant changes in insulin
and leptin levels in the first trimester of women with a high BMI that was comparable to that of
women with a normal BMI in the third trimester. This has enormous implications not only for
maternal metabolism but also for fetal growth trajectory (274).
73
2.4.2 Physical Activity and GDM
The present meta-analysis suggested a protective association of PA (21-46%) from GDM when
comparing any type of PA to none in either the pre-pregnancy or early pregnancy period. In a pooled
data set from 6 studies, consisting of 661 137 men and women, Arem and colleagues (275) reported
a similar protective association for all-cause mortality, which was steepest for comparisons between
none (referent) and the equivalent of 150 min/week moderate-intensity LTPA (or 7.5 MET.hr/week).
While current PA guidelines for pregnancy recommend 150 min/week (212, 276) of moderate or 75
min/week vigorous intensity PA (277), the majority of adults still fail to meet the PA guidelines (278)
Our meta-analysis, however, suggests a potentially lower odds of GDM (46%) at >90 min/week.
Similarly, O’Donovan and colleagues reported that taking part in 1 or 2 sessions/week in moderate-
vigorous intensity PA resulted in CVD and all-cause mortality risk reduction regardless of an
individual’s adherence to the current guidelines (279). Whilst the potential benefits of structured
LTPA are undisputed, volumes of PA below the recommended levels and even light intensity PA may
have measurable health benefits (280).
We also observed that women who engaged in any type of PA compared to none in the year before
pregnancy had potentially 10% lower odds of developing GDM than women who engaged in PA also
compared to none during early pregnancy. The findings are further strengthened by presence of low-
medium level of heterogeneity, no evidence of publication bias and data collected by predominatly
validated questionnaires. On the other hand, Cordero and colleagues (281) suggested that engaging
in structured PA 150-180 min/week during pregnancy could reduce the risk of GDM up to 90% when
compared to standard care in their RCT. While it can be argued that study design may have an effect
on the strength of the results, it cannot be denied that increasing PA appears to have an inverse
association with risk of GDM. Since pregnancy is a temporary phase in a woman’s life, accompanied
by many physiological and physical changes, preconception period should be perceived as a window
74
of opportunity to adopt a healthier lifestyle. This could include incorporating more PA, achieving a
healthy BMI and following a diet rich in plant-based food groups such as MedDiet to prevent
undesirable pregnancy outcomes.
2.4.3 Strengths and Limitations
This study has strengths and limitations. Due to strict the selection criteria, a few larger population
studies were excluded which may have limited the findings of this review. For example, the Coronary
Artery Risk Development in Young Adults (CARDIA) study (282) (n = 1488) did not report on the
relationship between diet or PA and risk of GDM, but rather differences in health behaviours
between the pre-pregnancy period to several years after pregnancy. On the other hand, a study by
Deierlein and colleagues (283) (n = 1437) did not report GDM as an outcome measure but used the
term ‘hyperglycaemia’ instead. Studies that were included in other systematic reviews (263), such as
Saldana et al. (284) and He et al. (285), were excluded in our review due to late study recruitment
and subsequently collection of dietary data that was not reflective of the early pregnancy period.
Since all included studies are observational, they are susceptible to the effects of bias, confounding,
potential measurement error, and under/over reporting of dietary intake. In the meta-analysis of PA,
we were able to minimize these effects by using a random-effects model in statistical analysis
(assuming heterogeneity) and in all the reviewed studies we conducted independent assessment of
study quality, including whether validation of data collection methods had occurred (Table 2.2).
What adds strength to the present study is that different indicators of LTPA were tested a priori with
consistent findings. However, caution should be applied when interpreting sub-analysis of LTPA
studies, particularly as there are <10 data sets available to be able to determine with confidence
whether asymmetry is real or a coincidental occurrence (286).
75
2.5 Conclusions
Assuming that the associations we identified reflect causal relationships, our review suggests that a
MedDiet and PA are promising interventions for the prevention of GDM. However, a greater degree
of protection may occur when both lifestyle factors are incorporated before pregnancy and followed
throughout pregnancy. Engaging in any PA even below the guidelines suggested a protective
association with GDM risk. The finding in part raises the importance of individualized-patient care as
the level of PA set in the current guidelines may be unachievable by some, however they could still
potentially achieve similar health benefits at a lower PA threshold. There is an opportunity for future
RCTs to further explore interventions with both PA and MedDiet pattern, especially in the pre-
conception period to ensure best outcomes throughout pregnancy.
76
Chapter 3
___________________________________________________
A modestly lower carbohydrate diet for the management of
gestational diabetes
Abstract
77
Background: Individuals with diabetes are commonly instructed to monitor their carbohydrate
intake for better blood glucose control. Diets with reduced carbohydrates have the potential to
increase levels of ketones in blood, which are negatively associated with infant brain development.
Currently, the safety and benefit of lower carbohydrate diets in managing gestational diabetes
(GDM) remain unclear.
Objective: To investigate blood ketone levels, risk of ketonaemia and pregnancy outcomes in women
with GDM prescribed a Modestly Lower Carbohydrate (MLC) diet vs Routine Care (RC) diet.
Method: Women diagnosed with GDM were invited to participate in the MAMI 1 (Macronutrient
Adjustments in Mothers to Improve GDM) study, a pilot, 2-arm randomised controlled trial
conducted at Campbelltown and Royal Prince Alfred (RPA) Hospitals, Sydney, Australia. Using block
randomisation, eligible participants were randomised to MLC diet (carbohydrate target 135 g/day)
or RC (180-200 g/day). Blood ketones and 3-day food diaries were collected at baseline and after a
6-week intervention. Dietary compliance and random blood ketone levels were assessed at clinic
visits. Pregnancy outcomes were obtained from medical records. The trial was registered in the
Australia and New Zealand Clinical Trials Register (ANZCTR): 12616000018415.
Results: Forty-six women completed the study, 24 in MLC and 22 in RC. Carbohydrate and total
energy intake were significantly lower in MLC vs RC (mean ± SEM, carbohydrate 165 ± 7 g vs 190 ± 9
g; P = 0.042; energy, 7040 ± 240 kJ vs 8230 ± 320 kJ; P = 0.006, respectively). There were no detectable
differences in blood ketones (MLC 0.1 ± 0.0 mmol/L vs RC 0.1 ± 0.0 mmol/L; P = 0.308) with mean
levels well below the threshold of ketonaemia. Infant head circumference was significantly lower in
the MLC group (MLC 33.9 ± 0.11 cm vs RC 34.9 ± 0.3 cm; P = 0.046), before and after adjustment for
GWG, weeks gestation at delivery and infant sex (P = 0.043).
Conclusion: A MLC diet provided sufficient carbohydrates to prevent ketonaemia but may also
reduce overall energy and nutrient intake with a potentially negative impact on brain development.
78
3.1 Introduction
Carbohydrates have played an important role in human evolution with marked changes in gene
frequency related to starch and lactose intake (287). In early pregnancy, maternal BGLs typically fall
despite increases in hepatic glucose production (139). In famine and prolonged fasting, maternal
glucose is diverted to the fetus while the mother temporarily switches to fat metabolism to generate
energy (150). Similarly, in a non-pregnant state, when macronutrients are consumed simultaneously,
the body has greater selectivity towards glucose absorption (288) and carbohydrate oxidation (289)
over other macronutrients. Aside from energy, carbohydrate foods are sources of vitamins, minerals
and dietary fibre (290). National guidelines advise dietary intake of carbohydrate to be 45-65% of
total energy intake (EI) (16), although estimates of actual consumption fall within a wider range of
40-80% (287).
Since 1961, global cereal production has increased 5-fold (291), concurrently with an increase in
world population (292) and a 3-fold increase in the prevalence of obesity and type 2 diabetes mellitus
(T2DM) (293). Carbohydrate is the main component of cereals and the primary determinant of
postprandial glucose (44). In recent studies, limiting carbohydrate intake in people with T2DM has
produced significant improvements in glycated haemoglobin A1c (HbA1c) and body mass index (BMI)
(294). However, few studies have investigated the effects of a lower carbohydrate diet in treatment
of GDM (44, 190, 191, 295-297). In some instances, there were improvements in postprandial
glucose levels (190, 191, 295, 296), a reduction in insulin requirements (191) and lower risk for LGA
infants (191). One study reported no differences in any outcome (44).
Currently there is no consensus on the most effective diet for management of GDM (185). This is
concerning as GDM is associated with a plethora of negative pregnancy outcomes including
macrosomia (298), caesarean section and pelvic trauma (299), and the incidence of GDM continues
to rise (178, 299). MNT is the first line of treatment for GDM and focuses on carbohydrate quality,
quantity and distribution throughout the day to achieve euglycaemia (300). While MNT does not
79
specify a daily carbohydrate target, although it is generally accepted that the minimum daily
carbohydrate intake should be 175g (183), the American Endocrine Society and American College of
Obstetrics and Gynaecologists advise women to follow a lower carbohydrate diet, where
carbohydrates comprise ~35-45% (301) and 33-40% (189) of their total EI, respectively. Typically,
carbohydrate restriction reduces the ratio of glucagon to insulin, thereby promoting oxidation of free
fatty acids to BHB and other ketones (302). The suggested %EI from carbohydrates for GDM
management should not be associated with elevated ketone levels in the blood, although there are
no trials to confirm the assumption. In studies of carbohydrate reduction in GDM, only urinary ketone
levels have been assessed (44, 190, 191).
While there is a strong correlation between urine and blood ketone levels, blood ketone
measurements are considered superior (130) and reflective of current ketone status in the body.
Given the potentially serious negative effects of ketonaemia on fetal brain development (196), we
aimed to investigate actual blood levels and risk of ketonaemia in women with GDM following a
modestly lower carbohydrate (MLC) diet (carbohydrate target 135 g/day) versus routine care (RC)
(180-200 g/day). Pregnancy outcomes, including birth weight and head circumference were also
measured. We hypothesised that a modest reduction in dietary carbohydrate intake would not
increase blood ketone levels or risk of adverse pregnancy outcomes.
3.2 Methods
The present study was a pilot 6-week, 2-arm, parallel randomised controlled trial conducted at the
antenatal clinics of RPA Hospital (Camperdown) and Campbelltown Hospital, Australia. Recruitment
took place between April 2016 to May 2018. All personnel and participants were blinded to the
dietary group allocation, except for the study dietitian (J.M.). We selected the intervention period of
6-weeks, as this timeframe was sufficient to promote changes in maternal and neonatal outcomes
80
in previous studies (303, 304). The trial was registered in the Australia and New Zealand Clinical Trials
Register (ANZCTR): 12616000018415. Ethics approval was obtained from the South-Western Sydney
Local Health District (HE16/367) and Human Research Ethics Committee of the Sydney South West
Area Health Service (RPA Hospital Zone HREC/15/RPAH/397).
3.2.1 Participant recruitment
Pregnant women aged 18-45 years with a singleton pregnancy, between 24-32 weeks gestation and
a GDM diagnosis were eligible to take part in the study (Table 3.1).
Table 3.1 Participant selection criteria.
Criteria
Inclusion
✓ 18-45 years old
✓ 24-32 weeks gestation
✓ Women diagnosed with gestational diabetes
✓ Singleton pregnancy
✓ Understanding the English language
Exclusion
X Women with special dietary requirements (e.g. vegan/vegetarian)
X Existence of co-morbidities other than obesity, hypertension or dyslipidaemia
X Pre-existing diabetes
X Undesirable lifestyle habits such as smoking and alcohol consumption
X Pregnancy achieved by assisted reproduction (in-vitro fertilisation)
GDM diagnosis was based on a fasting 75 g OGTT, using the 2010 IADPSG diagnostic criteria: Fasting
BGL ≥5.1 mmol/L, 1-hr BGL hour ≥10.0 mmol/L and 2-hr BGL ≥8.5 mmol/L (305). The hospitals
differed in their blood glucose monitoring targets, with a fasting BGL target of ≤5.3 mmol/L and 2-hr
post meal ≤6.8 mmol/L used by the Campbelltown Hospital, and a fasting BGL ≤5.2 mmol/L and 1-hr
post meal ≤7.4 mmol/L used by RPA Hospital. Women were excluded if they had current alcohol or
81
smoking status, followed a gluten free or vegan/vegetarian diet, could not understand the English
language or had major surgery (e.g. Roux-En-Y) in the past 5 years that affected nutrient absorption.
Once deemed eligible, study procedures were explained, and willing participants completed the
consent and enrolment forms. The enrolment form captured demographic data including age, parity,
past and present medical conditions, self-reported pre-pregnancy weight and PA levels, and degree
of exposure to nutrition information. Gestational age was based on last menstrual period and
corrected if indicated otherwise by an ultrasound scan. Since weight was measured as part of usual
care at both hospitals, we obtained participant weight from medical records at each visit.
3.2.2 Baseline data collection
Once enrolled, participants were required to complete a 3-day food diary and a 2-day blood ketone
diary. The food diary consisted of any 2 weekdays and 1 day of the weekend to account for day-to-
day variability in dietary intake (306). Ongoing blood glucose monitoring was required as per usual
care. Blood finger prick glucose levels were assessed 4 times/day, with first measurement collected
in the morning following an overnight fast and subsequent measurements taken 1-hour (RPA
Hospital) or 2-hours (Campbelltown Hospital) after each of the 3 main meals. To ascertain baseline
glucose management, HbA1c levels were extracted from medical records.
Blood ketone levels in the form of BHB were determined using Optium™ meter and Optium™ β-
ketone test strips (Abbott, Macquarie Park, Australia). The ketone test strips contained 3 electrodes,
including a fill trigger, working and a reference electrode (307). To operate, the strips required ~0.6
μL of whole blood, which was obtained using a disposable lancet (Accu-Chek®, Roche Diagnostics
GmbH, Mannheim, Germany). The research dietitian demonstrated how to measure blood samples,
operate the ketone monitor and instructed participants to collect 3 blood samples per day during
the 2-day blood ketone collection period. The first measurement took place in the morning after an
82
overnight fast. Two other measurements were collected at noon and in the evening, both prior to a
main meal to determine the likely highest rise in ketones. The following criteria were used to define
safe ketone levels: normal: <0.5 mmol/L, hyperketonaemia: >1.0 mmol/L and ketoacidosis: >3.0
mmol/L (106). All baseline measurements were cross-checked by the research dietitian.
3.2.3 Randomisation and stratification
Participants were randomly allocated to either a MLC or RC diet using block randomisation technique
- a method which ensured a balance in participant numbers and characteristics at baseline (308, 309).
In practice, this generated 4 boxes stratified according to age (18≤age≤30 or 30<age≤45) and BMI
(≤27 or >27) category. Each box contained an equal number of concealed “intervention” or “control”
cards (i.e. 8 intervention and 8 control cards). If a participant met the criteria for any of the 4 boxes,
a card was drawn from that box by the research dietitian to achieve random assignment. Once
assigned, the card was disposed. Aside from the research dietitian, participants and other health
professionals were blinded to the diet allocation.
3.2.4 Dietary intervention, safety and compliance
The MLC diet aimed for an absolute target of 135 g carbohydrate/day (estimated average
requirement (EAR)) in line with Institute of Medicine recommendation for carbohydrate that would
meet the needs of 50% of the pregnant population (310), without restricting overall EI. The RC diet
aimed for an absolute target of 180-200 g carbohydrate/day. Additionally, we asked participants to
strive for an even daily distribution of carbohydrates for both diets, as depicted in Figure 3.1. A
sample meal plan is shown in Table 3.2a and Table 3.2b.
83
Routine care (RC) diet - control
Breakfast Snack (AM) Lunch Snack (PM) Dinner Supper
2-3 1-2 3-4 1-2
3-4 1-2
Modestly lower carbohydrate (MLC) diet - intervention
Breakfast Snack (AM) Lunch Snack (PM) Dinner Supper
2 1 2 1 2 1
Figure 3.1 Target carbohydrate distribution (as exchanges) for the control and intervention arms of the MAMI 1 study.
To assist in achieving the set carbohydrate target, we provided a pictorial booklet to both groups.
While the booklets contained the same content such as images of different food categories, detailing
their carbohydrate content, number of carbohydrate exchanges as well as their GI, they emphasised
different carbohydrate targets. Study visits were made to coincide with visits to the antenatal clinic,
approximating every 2 weeks.
As safety was critical, the research dietitian closely monitored patient medical symptoms and
biochemistry, with specific focus on blood ketone levels. A strategy to avoid potential adverse effects
was developed before the study participants commenced MLC diet. If participants were generally
feeling unwell or dizzy, they were instructed to assess both blood ketone and glucose levels. If the
ketone level was >0.5 mmol/L, they were advised to consume a carbohydrate containing snack and
repeat the blood ketone measurement. If blood ketone levels remained elevated (i.e. >0.5 mmol/L)
for more than 2 days, the patient was advised to contact the hospital to arrange an immediate clinical
review to establish the cause of elevated blood ketones. They were also instructed to increase their
dietary carbohydrate to 180 g/day and monitor blood ketones daily until the levels stabilised (for at
least 48 hours). The obstetrician or materno-fetal medicine fellows reviewed both maternal and fetal
wellbeing. If and when these events occurred, participants were excluded from the study.
Versus
84
Table 3.2a Sample meal plan for the modestly lower carbohydrate (MLC) diet group.
Meal Meal Description Amount
BF
Oats (Traditional) 30 g
Skim milk 150 mL
Blueberries ½ cup
AM Wholegrain bread 1 slice
Avocado ¼ avocado
Lunch
Sandwich
- Wholegrain bread
2 slices
- Chicken breast 40 g
- Cheese (gouda) 20 g
- Avocado ¼ avocado
- Vegetables of choice (1/2 medium tomato,
1/3 capsicum, 1/2 medium carrot, few
leaves spinach, 2 olives)
1 cup
PM Medium apple
Mixed unsalted nuts
150 g
30 g
Dinner
Basmati rice steamed ½ cup = 100 g
Garden salad (with olive oil and balsamic dressing) 1 cup
- Lettuce ½ cup
- Cucumber ¼ cucumber
- Tomato ¼ tomato
Roasted vegetables
- Sweet potato
- Pumpkin
- Capsicum
- Broccoli
30 g
30 g
30 g
30 g
Roasted (lean beef/ chicken) >85 g
Supper Yoghurt (Yoplait Forme, strawberry sensation)
Mixed unsalted nuts
150 g
30 g
85
Table 3.2b Sample meal plan for the routine care (RC) diet group.
Meal Meal Description Amount
BF
Oats (Traditional) 45 g Skim milk 200-250 mL
Blueberries ½ cup
AM
Wholegrain bread 1 slice
Cheese (cheddar) 1 slice
Apple 1 medium
Lunch
Sandwich - Wholegrain bread
2 slices
- Chicken breast 40 g - Cheese (gouda) 20 g - Avocado 5 g
- Vegetables of choice (1/2 medium tomato, 1/3 capsicum, 1/2 medium carrot, few leaves spinach, 2 olives)
1 cup
PM Medium apple 150 g
Wholegrain cracker (Vita-Weat™) 4 crackers
Dinner
Basmati rice steamed ½ cup = 100 g Garden salad (with olive oil and balsamic dressing) 1 cup
- Lettuce ½ cup
- Cucumber ¼ cucumber
- Tomato ¼ tomato
Roasted vegetables - Sweet potato - Pumpkin - Capsicum - Broccoli
30 g 30 g 30 g 30 g
Roasted (lean beef/ chicken) > 85 g
Supper Yoghurt (Yoplait Forme, strawberry sensation) 150 g
A 24-hour recall, approximately every 2 weeks or at Visits 1-3, was used to assess treatment fidelity
with compliance to prescribed diets. On Visit 3, participants were asked to complete a 2nd 3-day food
diary and 2-day blood ketone diary. Overall glucose control during the intervention period was
ascertained using HbA1c levels, which were extracted from medical records. The return of the
booklets on their 4th Visit to the clinic marked the end of their participation in the study. An overview
of study data collection points is shown in Table 3.3.
86
Dietary information from food diaries and 24-hour recalls were entered by the research dietitian into
the Australian nutrition analysis computer software (FoodWorks Professional Version 8, 2015, Xyris
Software, Brisbane, Australia), based on AUSNUT 2011-2013 database. GI values were cross checked
and compared to the international table of GI and glycaemic load (GL) values (311).
Table 3.3 MAMI 1 (Macronutrient Adjustments in Mothers to Improve GDM) study collection
plan.
Consent Maternal anthropometry
Blood ketone
Blood glucose
24-hour diet Recall
3-day food diary
Infant anthropometry
Enrolment (24-32 weeks)
Baseline
Follow up 1
Follow up 2
Follow up 3
Final
3.2.5 Additional outcome measures
To determine the potential effects of MLC diet on the neonate, we obtained information on
birthweight, length, head circumference, percent fat mass (%FM) and percent fat free mass (%FFM)
(RPA Hospital only) from electronic medical records. We used the WHO gender specific growth charts
(weight for gestational age) to assess birthweight. Neonates below the 10th percentile were
categorised as small-for-gestational age (SGA) and those exceeding the 90th percentile were classed
87
as large-for-gestational age (LGA). Macrosomia was defined as birthweight >4000 g. In addition,
birthweights were compared to the Australian national birthweight percentiles (312). Both the
Centre for Disease Control and Prevention (CDC) growth charts and Australian birthweight database
were used to determine head circumference percentiles (313-315),
GWG was calculated as the difference between the last measured weight before delivery and pre-
pregnancy weight and compared to the 2009 IOM weight gain guidelines (316), specific for each BMI
category. Full term pregnancy was defined as ≥37 weeks gestation.
3.2.6 Statistics
Statistical analyses were conducted using SPSS (version 24, IBM Australia, St Leonards, Australia) and
SAS Statistical Software, Version 9.4 (SAS Institute Inc., Cary, NC). Descriptive data are presented as
mean ± SEM for continuous variables and percentages for frequency variables. For continuous
outcomes or to assess differences between groups, we used an independent samples t-test, one-
way ANOVA or Mann-Whitney U test where appropriate. Pearson’s chi-square test of independence
was used to compare method of delivery, induction rate, infant birthweight.
Bivariate analyses (Pearson (rρ) or Spearman (rS) correlation coefficients where appropriate) were
used to assess the association between selected maternal variables and infant outcomes.
We used a crude and adjusted logistic linear regression to assess the association between dietary
intervention and odds of having elevated blood ketone levels. Blood ketones were stratified into 3
categories according to change from baseline, namely if there was a decrease (n = 17), an increase
(n = 4) or no change (n = 19) in measurements. The model adjusted for maternal age and categories
of BMI (BMI <25 and BMI ≥25). In addition, median level of daily carbohydrate intake was used
stratify participants into higher (≥median) or lower (<median) consumers regardless of their group
88
assignment. This allowed us to ascertain whether levels of actual carbohydrate intake affected
ketone status (adjusting for age and categories of BMI).
Although MLC diet and RC groups were prescribed absolute carbohydrate targets, we allowed a 20%
deviance (MLC diet ≤162 g/day; RC >162 g/day) to account for the potential difficulty in reaching the
target. Gestational age at delivery was stratified as <39 or ≥39 weeks gestation.
3.2.7 Power calculation
The study was designed to provide 80% statistical power to detect approximately 0.04 mmol
difference in blood ketone levels with 25 participants required for each of the 2 study arms.
Considering a potential dropout rate of 25%, our study required a total of 65 participants, or
approximately 32 per study group. The calculations were based on a study by Gin et al. 2006 (317).
3.3 Results
A total of 297 pregnant women diagnosed with GDM were approached, of which 76 gave consent.
Twenty-nine of these withdrew before randomisation (Figure 3.2), leaving 46 who were randomised
to either MLC or RC group. One participant reported extremely low energy intake (~4500 kJ) at
baseline and was removed from further analysis. Our intention-to-treat (ITT) analysis consisted of 45
participants. A secondary analysis consisted of 33 participants who completed all requirements of
the study. The most common reason for non-completion included early delivery (n = 10), followed
by loss of interest (n = 2) and medical reasons (n = 1). The 5 women who were recruited at
Campbelltown Hospital subsequently withdrew before randomisation and baseline data collection.
Figure 3.3 shows the cumulative recruitment rate on a week-by-week basis.
89
Figure 3.2 Flow diagram depicting progress of a 2-group parallel randomised trial.
Figure 3.3 Cumulative frequency of participants consenting to take part in the study at Royal Prince Alfred and Campbelltown Hospitals.
Analysed:
n = 17
Analysis (ITT)
RC Diet
(n = 22)
MLC Diet
(n = 24)
Assessed for eligibility
(n = 297)
Analysed; n = 24
Randomised
Enrolled
(n = 75)
Withdrew (n = 29) Lost interest (n = 17)
Lost contact (n = 6)
No timely baseline data (n = 4)
Busy schedule (n = 2)
Analysis
(Completers)
Analysed:
n = 16
Withdrew (n = 8) Delivered before final
data collection (n = 7)
Lost interest (n = 1)
Withdrew (n = 4) Delivered before final
data collection (n = 3)
Medical reasons (n = 1)
0
10
20
30
40
50
60
70
80
Cu
mu
litiv
e fr
equ
ency
of
con
sen
tin
g p
arti
cip
ants
Under-reporter; n = 1
Analysed; n = 21
90
Participant characteristics at baseline are shown in Table 3.4a and Table 3.4b. The two dietary
groups were comparable with respect to age, mean pre-pregnancy BMI, weeks gestation at
diagnosis, OGTT results and HbA1c at baseline. Both groups had a strong family history of T2DM and
hypertension, and 100% pregnancy multivitamin use. MLC had a higher proportion of women in the
obese pre-pregnancy BMI category but a lower proportion using thyroid medication. Women who
withdrew from the study were similar to the two treatment groups. Both groups were also
comparable in absolute intake of nutrients and macronutrient energy distribution at baseline (Table
3.5).
Among the women who completed the full protocol, including 3-day food diaries (n = 33), the MLC
group consumed less carbohydrate (165 ± 7 vs 190 ± 9 g/day, P = 0.042) as predicted, but also less
total energy (7040 ± 240 vs 8230 ± 320 kJ/day, P = 0.006) (Table 3.6). Consequently, the MLC group
consumed less protein (85 ± 4 vs 103 ± 4 g/day, P = 0.006), and fewer micronutrients (including
significantly less iron and iodine). As a proportion of energy, carbohydrate was similar in both groups.
A sub-group analysis of women (n = 23) who complied with their assigned carbohydrate target (Table
3.7) indicated a significantly lower intake of carbohydrate in the MLC vs RC group (143 ± 4 vs 196 ±
6 g/day, P <0.0001), as well as sugars (55 ± 4 vs 77 ± 3 g/day, P <0.0001), starch (88 ± 5 vs 117 ± 5
g/day, P <0.0001), and glycaemic load (79 ± 10 vs 104 ± 13, P <0.0001), respectively. However, intake
of total energy was lower still in the MLC group (6640 ± 240 vs 8040 ± 210 kJ/day, P <0.0001). On
average, women in the MLC group met the predicted proportion of energy as carbohydrate intake
(36% EI), but those in the RC group consumed less than instructed (41% EI). Both dietary groups
exceeded the 10% EI target for saturated fat. There were no differences in intake of polyunsaturated
fatty acids including linoleic, alpha-linolenic, eicosapentaenoic, docosapentaenoic and
docosahexaenoic fatty acid between groups at baseline, among completers or in the compliant sub-
analysis (results not shown).
91
Table 3.4a Baseline characteristics of participants that received education.
n MLC n RC P
Age, y 24 32.5 ± 0.9 21 33.9 ± 0.9 0.281
BMI (kg.m2) 24 25.8 ± 1.0 21 28.0 ± 1.6 0.229
BMI category
- Underweight (%) 0 0 2 8.7
- Normal (%) 12 50.0 8 34.8
- Overweight (%) 8 33.3 5 21.7
- Obese I (%) 2 8.3 4 17.4
- ≥ Obese II (%) 2 8.3 2 8.7
Ethnicity (Asian vs Caucasian) 0.278†
Asian (%) 13 54.2 16 69.7
- East Asian (%) 1 4.2 5 21.7
- South Asian (%) 9 37.5 4 17.4
- Southeast Asian (%) 3 12.5 7 30.4
Caucasian (%) 11 45.8 7 30.4
Other (%) 0 0 0 0
Nulliparous n (%) 14 58.3 10 43.5 0.472†
Weeks at GDM diagnosis 24 20.2 ± 1.1 21 20.7 ± 1.2 0.431
75-g OGTT results (mmol/L)
Fasting 23 4.8 ± 0.1 20 4.7 ± 0.1 0.261
1-hour 23 9.4 ± 0.3 19 9.9 ± 0.3 0.383
2-hour 23 8.0 ± 0.4 19 8.3 ± 0.3 0.569
HbA1c % 20 5.1 ± 0.1 20 5.0 ± 0.1 0.522
Education 0.727†
- Secondary (%) 4 16.7 3 13.0
- Tertiary (%) 20 83.3 18 87.0
Marital status
- Single (%) 1 4.2 1 4.3
- De-facto (%) 1 4.2 6 26.1
- Married (%) 22 91.7 16 69.6
Smoking Hx (%) 5 20.8 5 21.7 0.811†
GDM Hx 2 8.3 1 4.3
Family History
- T2DM 18 75.0 16 69.6 0.926†
- HT 18 75.0 14 60.9 0.538†
- Ow/Ob 7 29.2 6 26.1 0.965†
Insulin use (%) 6 25.0 6 28.6 0.787†
Thyroid Medication (%) 6 25.0 2 9.5
Metformin (%) 1 4.2 1 4.8
Aspirin (%) 1 4.2 1 4.8
Supplement use 24 100 21 100
- Pregnancy multivitamin 24 100 21 100
Abbreviations: GDM – gestational diabetes mellitus; HT – hypertension; Hx – history; MLC – modestly lower carbohydrate (diet) OGTT – oral glucose tolerance test; Ow/Ob – Overweight/Obesity; RC – routine care (diet); SEM – standard error mean; T2DM – type 2 diabetes mellitus;
Independent samples t-test, Mann Whitney tests or Pearson’s chi-square test of independence (†) were performed. P <0.05 deemed significant.
92
Table 3.4b Baseline characteristics of participants who withdrew from the study prior to randomisation compared to women that completed the study.
n Withdrawn n MLC n RC PA P
Age, y 29 31.7 ± 1.0 16 32.7 ± 1.0 17 33.9 ± 1.1 0.400 0.371
BMI (kg.m2) 29 26.9 ± 1.0 16 27.0 ± 1.2 17 29.2 ± 1.8 0.337 0.659
BMI category
- Underweight (%) 0 0 0 0 2 11.8
- Normal (%) 13 46.4 6 37.5 5 29.4
- Overweight (%) 7 25.0 6 37.5 2 11.8
- Obese I (%) 5 17.9 2 12.5 4 23.5
- ≥ Obese II (%) 3 10.7 2 12.5 4 23.5
Ethnicity (Asian vs Caucasian)
0.881† 0.836†
Asian (%) 13 48.1 9 56.3 10 58.8
- East Asian (%) 3 23.1 0 0 4 23.5
- South Asian (%) 5 38.5 7 43.8 3 17.6
- Southeast Asian (%) 5 38.5 2 12.5 3 17.6
Caucasian (%) 13 48.1 7 43.8 7 41.2
Other (%) 1 3.7 0 0 0 0
Nulliparous n (%) 16 57.1 8 50.0 6 35.3 0.392† 0.424†
Weeks at GDM diagnosis 24 20.9 ± 1.2 16 21.5 ± 1.5 17 20.4 ± 1.9 0.656 0.964
75-g OGTT results
(mmol/L)
Fasting 24 4.9 ± 0.1 15 4.7 ± 0.1 16 4.7 ± 0.1 0.599 0.587
1-hour 23 10.1 ± 0.3 15 9.3 ± 0.4 15 9.7 ± 0.4 0.473 0. 333
2-hour 23 7.2 ± 0.5 15 7.9 ± 0.5 15 8.2 ± 0.3 0.713 0.156
HbA1c % 28 5.1 ± 0.1 13 5.1 ± 0.1 16 5.0 ± 0.1 0.720 0.976
Education (Secondary vs Tertiary) 0.605† 0.711†
- Secondary (%) 8 28.6 4 25.0 3 17.6
- Tertiary (%) 20 71.4 12 75.0 14 82.4
Marital status
- Single (%) 2 7.4 0 0 0 0
- De-facto (%) 5 18.5 1 6.3 6 35.3
- Married (%) 20 74.1 15 93.7 11 64.7
Smoking Hx (%) 8 29.6 3 18.8 4 23.5 0.737† 0.801†
GDM Hx 5 25.0 2 12.5 1 5.9
Family History
- T2DM 22 81.5 12 75.0 12 70.6 0.776† 0.922†
- HT 15 55.6 13 81.3 10 58.8 0.161† 0.146†
- Ow/Ob 7 25.9 3 18.8 6 35.3 0.286† 0.533†
Insulin use (%) 8 28.6 4 25.0 6 35.3 0.520† 0.787†
Thyroid Medication (%) 4 14.3 6 37.5 2 11.8
Metformin (%) 0 0 1 6.3 1 5.9
Aspirin (%) 4 14.3 1 6.3 1 5.9
Supplement use 29 96.6 16 100 17 100
- Pregnancy
multivitamin
28 92.3 16 100 17 100
Abbreviations: GDM – gestational diabetes mellitus; HT – hypertension; Hx – history; MLC – modestly lower carbohydrate (diet) OGTT – oral glucose tolerance test; Ow/Ob – Overweight/Obesity; RC – routine care (diet); SEM – standard error mean; T2DM – type 2 diabetes mellitus; Values presented as mean ± SEM. PA – Independent samples t-test (between MLC and RC) or Pearson’s chi-square test of independence
(†), P – One-way ANOVA were performed or Pearson’s chi-square test of independence (†).
P <0.05 deemed significant.
93
Table 3.5 Maternal baseline diet in the two intervention groups.
Baseline
MLC RC PA
n 24 21 -
Energy (kJ) 7480 ± 320 7510 ± 370 0.949
Carbohydrate (g) 167 ± 6 164 ± 12 0.857
Sugars (g) 62 ± 4 61 ± 5 0.802
Starch (g) 104 ± 4 102 ± 9 0.844
Dietary fibre (g) 25 ± 1 24 ± 1 0.806
Protein (g) 100 ± 6 99 ± 5 0.882
Total fat (g) 74 ± 5 77 ± 6 0.732
- Saturated (g) 24 ± 2 27 ± 2 0.733
- Long chain FA-3 (g) 0.6 ± 0.2 0.5 ± 0.1 0.682
Carbohydrate (%EI) 38 ± 1 36 ± 2 0.529
Protein (%EI) 23 ± 0.8 23 ± 0.9 0.964
Total fat (%EI) 36 ± 1 37 ± 2 0.505
- MFA (% total fat) 45 ± 1 45 ± 2 0.985
- PFA (% total fat) 19 ± 1 18 ± 1 0.773
- SFA (% total fat) 36 ± 1 36 ± 1 0.853
SFA (%EI) 12 ± 1 12 ± 1 0.617
GI 54 ± 1 52 ± 1 0.339
GL 92 ± 3 88 ± 7 0.619
Iron (mg) 10 ± 1 11 ± 1 0.581
Iodine (µg) 161 ± 11 144 ± 10 0.261
Total folate (µg) 490 ± 40 440 ± 30 0.509
Abbreviation: GI – Glycaemic index; GL – Glycaemic load; FA – Fatty acid; MFA – monounsaturated fatty acid; MLC – modestly lower carbohydrate (diet); EI – energy intake; RC – routine care (diet) Values presented as mean ± SEM; Independent samples t-test or Mann Whitney tests were performed. P <0.05 deemed significant.
94
Table 3.6 Dietary intakes of study participants at the end of the intervention.
End of intervention
ITT Completers only
MLC RC Pꓮ
MLC RC PA
n 24 21 -
16 17 -
Energy (kJ) 7070 ± 201 7750 ± 222 0.028 7040 ± 240 8230 ± 320 0.006
Carbohydrate (g) 164 ± 4 176 ± 8 0.199 165 ± 7 190 ± 9 0.042
Sugars (g) 65 ± 3 70 ± 4 0.255 65 ± 4 78 ± 5 0.080
Starch (g) 98 ± 4 105 ± 6 0.373 99 ± 7 110 ± 7 0.252
Dietary fibre (g) 23 ± 1 24 ± 1 0.370 24 ± 1 26 ± 2 0.262
Protein (g) 90 ± 4 100 ± 5 0.102 85 ± 4 103 ± 4 0.006
Total fat (g) 70 ± 4 77 ± 4 0.199 71 ± 5 82 ± 5 0.136
- Saturated (g) 24 ± 1 27 ± 4 0.101 24 ± 2 29 ± 2 0.105
- Long chain FA-3 (g) 0.6 ± 0.2 0.3 ± 0.1 0.820 0.4 ± 0.1 0.4 ± 0.1 0.264
Carbohydrate (%EI) 39 ± 1 38 ± 2 0.516 39 ± 2 38 ± 1 0.613
Protein (%EI) 22 ± 1 22 ± 1 0.927 21 ± 1 21 ± 1 0.330
Total fat (%EI) 36 ± 1 36 ± 1 0.735 37 ± 2 37 ± 1 0.996
- MFA (% total fat) 45 ± 1 44 ± 1 0.500 46 ± 1 44 ± 1 0.240
- PFA (% total fat) 17 ± 1 17 ± 1 0.983 17 ± 1 17 ± 1 0.836
- SFA (% total fat) 38 ± 1 39 ± 1 0.584 37 ± 1 39 ± 1 0.294
SFA (%EI) 12.1 ± 0.4 12.6 ± 0.6 0.454 12.3 ± 0.7 13.0 ± 0.6 0.492
GI 53 ± 1 52 ± 1 0.982 53 ± 1 51 ± 1 0.313
GL 87 ± 3 93 ± 5 0.241 87 ± 4 98 ± 6 0.144
Iron (mg) 9.1 ± 0.4 10.0 ± 0.4 0.075 8.7 ± 0.4 10.6 ± 0.4 0.003
Iodine (µg) 160 ± 9 164 ± 11 0.778 147 ± 11 196 ± 14 0.009
Total folate (µg) 443 ± 21 467 ± 21 0.413 451 ± 25 488 ± 31 0.365
Abbreviations: EI – energy intake; GI – Glycaemic index; GL – Glycaemic load; ITT – intention-to-treat; MFA – monounsaturated fatty acid; MLC – modestly lower carbohydrate (diet); PFA – polyunsaturated fatty acid; RC – routine care (diet); SEM – standard error of the mean; SFA – saturated fatty acid. Values presented as mean ± SEM; Independent samples t-test or Mann Whitney tests were performed. P <0.05 deemed significant.
95
Table 3.7 Sub-analysis of women that met their assigned carbohydrate target intake.
End of intervention
MLC RC Pꓮ
n 10 15 -
Energy (kJ) 6640 ± 240 8040 ± 210 <0.0001
Carbohydrate (g) 143 ± 4 196 ± 6 <0.0001
Sugars (g) 55 ± 4 77 ± 3 <0.0001
Starch (g) 88 ± 5 117 ± 5 <0.0001
Dietary fibre (g) 21 ± 1 24 ± 1 0.135
Protein (g) 94 ± 6 97 ± 3 0.610
Total fat (g) 66 ± 4 78 ± 5 0.082
- Saturated (g) 22 ± 1 28 ± 2 0.063
- Long chain FA-3 (g) 0.4 ± 0.1 0.3 ± 0.1 0.912
Carbohydrate (%EI) 36 ± 1 41 ± 1 0.056
Protein (%EI) 24 ± 1 21 ± 1 0.023
Total fat (%EI) 36 ± 1 35 ± 1 0.528
- MFA (% total fat) 46 ± 1 44 ± 1 0.408
- PFA (% total fat) 17 ± 1 17 ± 1 0.982
- SFA (% total fat) 37 ± 2 39 ± 2 0.542
SFA (%EI) 12 ± 1 13 ± 1 0.737
GI 54 ± 2 52 ± 1 0.489
GL 79 ± 10 104 ± 13 <0.0001
Iron (mg) 8.8 ± 0.6 9.7 ± 0.4 0.207
Iodine (µg) 143 ± 10 175 ± 13 0.091
Total folate (µg) 428 ± 22 480 ± 25 0.156
Abbreviations: EI – energy intake; GI – Glycaemic index; GL – Glycaemic load; ITT – intention-to-treat; MFA – monounsaturated fatty acid; MLC – modestly lower carbohydrate (diet); PFA – polyunsaturated fatty acid; RC – routine care (diet); SEM – standard error of the mean; SFA – saturated fatty acid.
P <0.05 deemed significant; Pꓮ - Independent samples t-test or Mann Whitney tests were performed where appropriate.
96
3.3.1 Blood ketone levels (BHB)
There was no difference in average daily blood ketone levels between the groups at baseline (Table
3.8). However, closer inspection indicated higher fasting values in the MLC group (P = 0.013). At the
end of the intervention, there was no difference in our primary outcome of blood ketone levels
between groups and no instances of high levels (>1.5 mmol/L). One participant in the RC group
displayed a single high blood ketone value of 0.7 mmol/L, which was normalised (<0.5 mmol/L)
following a meal. On the whole, the MLC diet group had reduced odds (OR = 0.72, 95% CI: 0.19-2.58)
of showing high blood ketone concentration, but the difference remained non-significant after
adjusting for maternal age and pre-pregnancy BMI category. The model appeared to be influenced
by maternal BMI (OR = 0.38, 95% CI:0.10 -1.30), with higher BMI associated with lower blood ketone
levels. Even when participants were stratified into groups of higher and lower intakes of
carbohydrates (based on the median intake), lower carbohydrate intake was still associated with
reduced odds of having higher ketones (OR = 0.42, 95% CI: 0.11 - 1.54).
3.3.2 Pregnancy outcomes
While there were no differences in total GWG between groups, MLC group had a higher proportion
of women meeting the IOM weight gain guidelines than RC (P = 0.039). Among different modes of
delivery, natural birth predominated and was higher in the carbohydrate restricted group (MLC 71%
vs RC 52%), although weighted comparison between modes of delivery revealed no statistical
difference between the two groups (P = 0.203). We observed a trend towards higher induction rates
in the MLC diet group (P = 0.055), which could not have been explained by gestational age at delivery
(MLC 38.7 weeks ± 0.2 vs RC 38.6 ± 0.2 weeks, P = 0.973). Three of the 4 women induced in the MLC
group had a BMI >30. Almost 60% of participants in both groups were using insulin. There were no
97
significant differences between groups with respect to insulin dosing and mean change in units from
baseline (Table 3.9).
Of 45 births obtained from medical records, 4 were born prematurely (i.e. 36-37 weeks gestation).
There were no differences between infants with respect to length, %FM and %FFM. We observed a
moderate correlation between maternal pre-pregnancy BMI and total GWG (rρ = -0.450, n = 45, P =
0.002), although no correlation between total GWG or GWG category and infant birthweight (results
not shown). Among the women randomised to the MLC diet during pregnancy, there was reduced
odds of having a higher birthweight infant (OR = 0.49, 95% CI: 0.10 – 2.05), although this did not
reach statistical significance. Figure 3.4 shows infant birthweight according to gender using the
Australian National Birthweight percentile as a comparator.
Only 15 infants (33%) had Pea Pod data collection for determination of body composition. One-
quarter of MLC diet infants (n = 6) were classed as SGA when compared to RC (n = 3; 14%). Infants
whose mothers followed the MLC diet had significantly smaller head circumference RC (33.9 ± 0.3
cm versus 34.9 ± 0.3 cm; P = 0.046) (Table 3.10). The difference remained significant when the model
was adjusted for maternal GWG, weeks gestation at delivery and infant sex (P = 0.043) (Figure 3.5b).
98
Table 3.8 Biochemistry at baseline and end of the study for ITT and participants who
complied to the respective carbohydrate targets.
Baseline End of Intervention
ITT n MLC n RC P n MLC n RC P
HbA1c (%) 20 5.1 ± 0.1 20 5.0 ± 0.1 0.518 16 5.1 ± 0.1 15 5.3 ± 0.1 0.209
Glucose mmol/L
(average) 22 6.0 ± 0.1 17 6.2 ± 0.1 0.261 13 6.1 ± 0.1 15 6.0 ± 0.1 0.307
Ketone mmol/L
(average) 24 0.1 ± 0.0 21 0.2 ± 0.0 0.189 21 0.1 ± 0.0 19 0.1 ± 0.0 0.308
Fasting 24 0.1 ± 0.0 21 0.2 ± 0.0 0.013 14 0.1 ± 0.0 18 0.1 ± 0.0 0.178
Noon 24 0.2 ± 0.0 21 0.2 ± 0.0 0.240 15 0.1 ± 0.0 18 0.1 ± 0.0 0.770
Evening 24 0.1 ± 0.0 21 0.2 ± 0.0 0.239 13 0.1 ± 0.0 18 0.1 ± 0.0 0.754
Sub-group analysis of compliant participants
HbA1c (%) 9 5.0 ± 0.1 14 5.1 ± 0.1 0.859 8 5.2 ± 0.1 12 5.3 ± 0.1 0.402
Glucose mmol/L
(average) 9 5.7 ± 0.1 11 6.3 ± 0.2 0.093 6 6.2 ± 0.2 11 6.0 ± 0.1 0.318
Ketone mmol/L
(average) 10 0.2 ± 0.0 15 0.2 ± 0.0 0.293 6 0.1 ± 0.0 14 0.1 ± 0.0 1.000
Fasting 10 0.2 ± 0.0 15 0.2 ± 0.0 0.753 6 0.1 ± 0.0 14 0.1 ± 0.0 0.452
Noon 10 0.2 ± 0.0 15 0.2 ± 0.0 0.521 6 0.2 ± 0.0 14 0.1 ± 0.0 0.141
Evening 10 0.2 ± 0.0 15 0.2 ± 0.0 0.389 6 0.1 ± 0.0 14 0.1 ± 0.0 0.780
Abbreviation: MLC – modestly lower carbohydrate (diet); PFA – polyunsaturated fatty acid; RC – routine care (diet) Values presented as mean ± SEM; P <0.05 deemed significant; P value obtained from Independent samples t-test.
99
Table 3.9 Pregnancy outcomes in the two intervention groups.
n MLC n RC P Total weight gain (kg) 24 10.9 ± 0.9 21 8.2 ± 1.5 0.209
Meeting target vs not meeting weight gain target 24 - 21 - 0.039†
¥Below target (%) 41.7 12 57.1
¥Within target (%) 41.7 3 14.3
¥Above target (%) 16.7 6 28.6
Gestational age (weeks) 24 38.7 ± 0.2 21 38.6 ± 0.2 0.973
Mode of delivery (Vaginal vs Caesarean) 24 - 21 - 0.203†
Vaginal delivery (%) 24 70.8 52.4
Normal (% Vaginal) 88.2 72.7
Vacuum Extraction (% Vaginal) 5.9 18.2
Forceps-Liftout (% Vaginal) 5.9 9.1
Elected Caesarean (%) 12.5 38.1
Emergency Caesarean (%) 16.7 9.5
Induction (Yes, %) 16 66.7 8 38.1 0.055†
Insulin treatment (Yes at term, %) 14 58.3 12 57.1 0.936†
Final daily insulin dose (units) 14 14.6 ± 1.8 12 21.2 ± 3.9 0.126
∆ Insulin from enrolment (units) 14 7.1 ± 1.8 12 8.7 ± 4.6 0.748
Abbreviations: MLC – modestly lower carbohydrate (diet); RC – routine care (diet); LGA – large-for-gestational age; PI – Ponderal Index; SGA – small-for-gestational age; SEM – standard error of the mean; Values presented as mean ± SEM; P <0.05 deemed significant; P value obtained from Independent samples t-test; † Pearson’s chi-square test of independence
100
Table 3.10 Infant characteristics at delivery.
n MLC n RC P
Sex 23 19 0.853† Male (%) 52.2 55.0
Female (%) 47.8 45.0
Birthweight (g) 24 3125 ± 101 20 3278 ± 79 0.253
Birthweight within vs outside normal range
24 - 20 - 0.408†
SGA (%) 24 25.0 20 14.3 <0.0001
LGA (%) 24 0 20 4.8 <0.0001
Macrosomia (%) 24 4.2 20 4.8 <0.0001
Length (cm) 16 47.9 ± 0.7 10 49.2 ± 0.4 0.195
Head circumference (HC, cm) 22 33.9 ± 0.3 17 34.9 ± 0.3 0.046
HC/length 16 0.70 ± 0.01 10 0.71 ± 0.01 0.301
HC/birthweight 22 0.11 ± 0.00 17 0.11 ± 0.00 0.270
PI (kg/m3) 16 2.7 ± 0.1 10 2.7 ± 0.1 0.832
Fat Mass (%) 7 7.2 ± 2.2 8 10.1 ± 1.0 0.233
Fat Free Mass (%) 7 92.8 ± 2.2 8 89.9 ± 1.0 0.233
Abbreviations: MLC – modestly lower carbohydrate (diet); RC – routine care (diet); LGA – large-for-gestational age; PI – Ponderal Index; SGA – small-for-gestational age; SEM – standard error of the mean; Values presented as mean ± SEM; Differences between groups
determined using Independent t-test; P <0.05 deemed significant. † Pearson’s chi-square test of independence.
101
Figure 3.4 Birthweight stratified by weeks gestation at delivery and infant gender using the Australian National Birthweight percentiles (1998-2007) as the comparator. MLC – modestly lower carbohydrate (diet); RC – routine care (diet).
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
34 36 38 40
95th90th75th50th25th10th
5th
RCMLC
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
34 36 38 40
95th90th75th50th25th10th
5th
Gestational Age (weeks)
MALE
FEMALE Bir
thw
eig
ht
(g)
102
Figure 3.5 Infant outcomes based on maternal dietary intervention group. A) There was no
association between diet and birthweight (n = 43, P = 0.149). B) A modestly lower
carbohydrate (MLC) diet was associated with a statistically lower head circumference in
the infant (n = 39, P = 0.043) compared to routine care (RC) diet. C) There was no
association between diet and fat mass (FM%), (n = 15, P = 0.264). All 3 models adjusted
for maternal gestational weight gain and weeks gestation at delivery. Unlike A and B,
model C did not adjust for infant sex due to small sample size.
103
3.4 Discussion
In this pilot RCT, women with GDM who were instructed to follow a MLC diet had average blood
ketone levels and glucose control similar to those of women assigned to routine management of diet
in GDM. Even after a typical overnight fast when carbohydrate stores in liver are likely to be
depleted, there was no sign of ketonaemia in either treatment group. Our most surprising finding,
however, was a significantly smaller head circumference in infants born to mothers assigned to the
MLC diet (MLC 33.9 vs RC 34.9 cm). All infants in the MLC group fell within the range of the 10-25th
percentile, while those receiving RC fell within 25-50th percentile range. This difference was seen in
the absence of differences in birthweight and birth length, and remained significant after adjustment
for infant sex, weeks gestation and maternal weight gain. In addition to lower carbohydrate intake,
mothers assigned to the MLC diet group ingested less energy and fewer micronutrients, including
iron and iodine. These additional factors raise concerns and may explain the difference in head
circumference. Given the potential of head circumference to reflect brain development (318), larger
studies of restricted carbohydrate intake in GDM are warranted.
There are relatively few studies in GDM with which to compare the present study, although
carbohydrate-restricting diets often led to a reduction in energy intake (319, 320) in animal models.
Interestingly, even with energy restriction of the order of 30-33%, there was no evidence of
ketogenesis (321, 322). However, when energy restriction reached 50%, ketonuria increased 2 to 3-
fold (321). While energy restriction has been reported to reduce maternal weight gain (323, 324) and
birthweight in early studies (325) in animal models, there are concerns over safety (150). In
pregnancy, energy-restricted eating patterns are likely to negatively impact daily nutrient targets
(326). In the present study, the MLC diet group reported significantly lower dietary absolute intakes
of iron and iodine. Any deficiency in dietary intake may have been overcome by additional
micronutrients derived from the pregnancy multivitamin (100% usage in our study). Even carefully
designed low carbohydrate diet models failed to provide sufficient iron to non-pregnant women
104
(326). While others have reported no differences in micronutrients between diets containing <40%
vs ≥40% carbohydrate (327), a recent study suggested that low carbohydrate diets, particularly
during early pregnancy, are associated with a greater risk of neural tube defects (328), likely due to
reduced folic acid intake. Even in the context of frequent dietetic counselling, the recommended
daily micronutrient targets for pregnancy were not achieved from dietary intake alone. These
findings underline the importance of vitamin and mineral supplements to complement dietary intake
to help achieve the daily nutrient targets during pregnancy.
The MLC diet did not appear to provide any additional benefit for glucose management. Average
glucose levels and HbA1c were comparable in the two treatment groups. However, a systematic
review and meta-analysis exploring varying degrees of carbohydrate restriction, suggested that
moderate and high carbohydrate diets were both capable of reducing HbA1c levels in non-pregnant
populations with type 1 and type 2 diabetes (61). Indeed, Hernandez and colleagues demonstrated
that when compared to a lower carbohydrate diet, a high carbohydrate intake in GDM pregnancy
resulted in improvements in glycaemia (296), insulin sensitivity and a trend towards lower infant
adiposity (329).
The controversy surrounding the most effective therapeutic diet for GDM management continues
unabated. On one side, a high carbohydrate diet was reported to reduce the risk of macrosomia in
an observational study (330) or suggested greater infant thinness (331). However, other studies
reported either no association between birthweight and carbohydrate intake (332) or suggested a
reduced risk of LGA infants with a low carbohydrate diet (191). The mixed findings may relate to the
heterogeneity of carbohydrates (333) with varying effects on glycaemia, insulin resistance and health
outcomes. At the present time, it would be unacceptable to prescribe a diet of specific carbohydrate
quantity without consideration of its quality, including GI, dietary fibre and free sugar content.
Carbohydrate restriction has the potential to increase dietary fat intake, including saturated fat,
which may promote accelerated fetal growth (334). Our study, on the other hand demonstrated the
105
opposite, i.e. the relative difficulty in increasing protein and fat intake despite dietary counselling,
resulting in an energy difference of ~1000-1500 kJ. The higher cost of meat and food safety concerns
in pregnancy may partially explain our results (335, 336). Interestingly, with respect to fat, our study
indicated that both groups exceeded the recommendation for saturated fat intake (<10% EI).
Saturated fat intake has been associated with increased infant birthweight (337) and unfavourable
effects on glycaemia (HbA1c), C-peptide and insulin sensitivity (70). However, there was no
difference in birthweight or glycaemia in the two groups.
Carbohydrate intake in the MLC group was approximately 3 times greater than the minimum
suggested to induce ketosis in a non-pregnant population (338, 339). The blunting of ketosis cannot
be attributed solely to having adequate amount of carbohydrates, as protein, although less efficient,
was also shown to reduce ketosis (338). A 2-week 150 g carbohydrate/day intervention in pregnant
obese women with GDM suggested a modest rise in blood ketone levels (~0.26 mmol/L) (297), but
lower than the threshold for ketonaemia (>0.5 mmol/L). In other LC intervention studies in GDM,
only a small number of participants tested positive to ketonuria in either the intervention (n = 2)
(191) or control groups (n = 1) (44). These findings suggest that ketogenesis is low when carbohydrate
is only modestly reduced.
Interestingly, when the statistical model in the present study was adjusted for age and pre-pregnancy
BMI category, higher BMI appeared to offer some protection from high blood ketone levels. In non-
pregnant women, obesity is associated with significantly lower blood ketone concentration, despite
elevated free fatty acids (FFA) (340). Since ketone production is usually strongly correlated to FFA
concentration (341), these findings imply that obesity may impair the liver’s ability to convert FFA to
ketones (340), potentially due to fatty liver or inflammation. In GDM pregnancies, elevated levels of
FFA are associated with increased insulin resistance (342) and excess fetal growth (343).
The finding of lower infant head circumference in the MLC group raises some concern because of its
direct relationship to brain volume (344) and association with intelligence (345). According to one
106
study, high birthweight and head circumference predicted better cognitive performance at around
10 years of age (346). Aside from infant sex (347), covariates such as socio-economic status (348)
and maternal dietary exposure to acrylamide (349) have been linked to infant head circumference.
However, no association was observed with respect to maternal pre-pregnancy BMI (350) and
maternal stress during pregnancy (351). Our finding contrasts with that of Rhodes and colleagues
who found that obese women who followed a low GL diet (~49% energy from carbohydrates) had
infants of significantly larger head circumference at birth when compared to a low-fat diet (~52%
energy from carbohydrates) (352). Moses and colleagues reported no differences in infant head
circumference in women prescribed a higher GI diet during pregnancy, although the higher GI diet
suggested higher birthweight in infants (353). Our intervention had a much lower proportion of
energy derived from carbohydrates and our population of pregnant women had greater metabolic
disturbance. In the present study, there was no difference in dietary GI between the intervention
groups.
Our study has strengths and limitations. An important constraint was the precision of the Optium™
meter, which reported measurements to 1 decimal point. We were therefore unable to distinguish
values such as 0.11 from 0.14 mmol/L. While this may have meant that we missed small differences
in ketone concentration, the clinical significance would be uncertain. The candidate (a graduate
dietitian) was responsible for all aspects of the study, including screening, recruitment,
randomisation, education, and blood ketone assays in the clinic. The lack of blinding to treatment
may have introduced bias in interpretation of results. All neonatal outcomes including head
circumference were obtained from the medical records and may be prone to error including but not
limited to inexperience of data collector and infant’s hair volume (354). Our study was underpowered
to due to slow recruitment rate at the clinic. In the last few weeks of the study, a temporary
researcher assisted with recruitment. Dietary compliance and treatment fidelity proved to be an
important issue as some GDM participants were already restricting their intake of carbohydrate at
baseline. Previous studies have suggested greater underreporting in pregnant women with higher
107
BMI (355, 356) and lower education (356). Finally, we excluded an extreme outlier with a very low
energy intake (>2 SD below the mean of the group).
3.5 Conclusion
Modest reductions in carbohydrate intake do not result in greater ketone concentration or improved
glucose control in GDM. However, the lower carbohydrate dietary strategy resulted in significantly
lower intake of energy and of important micronutrients. Although there were no adverse pregnancy
outcomes, the lower infant head circumference in the lower carbohydrate treatment group suggests
that additional studies with appropriate power are warranted to further examine the suitability of
carbohydrate restriction during pregnancy.
108
Chapter 4
___________________________________________________
Ketone levels in women with gestational diabetes mellitus: a
pilot cross-sectional study
109
Abstract
Background: There are growing concerns over elevated ketone levels in pregnancy due to potential
negative effects on fetal brain development. The increasing popularity of low carbohydrate diets
among pregnant women with gestational diabetes mellitus (GDM) makes further research more
urgent.
Objective: To investigate random blood and urine samples for presence of ketones in women
diagnosed with GDM and assess whether there is an association with the maternal diet. Additionally,
maternal diet was compared to infant’s body composition at birth.
Method: A total of 161 women were recruited into a pilot cross-sectional study (MAMI 2) conducted
at Campbelltown (n = 40) and Royal Prince Alfred (RPA) Hospitals (n = 121) in Sydney, Australia.
Maternal anthropometry, 12-hour dietary recall and blood and urine samples were collected at
enrolment. Neonatal anthropometry and body composition (RPA only) data were taken from medical
records.
Results: Participants were comparable between study sites, with the exception of age and
carbohydrate quality, with Campbelltown women being younger and more likely to consume foods
of a higher GI and GL. Average random blood ketone levels were similar between sites (0.1 mmol/L).
There was a trend toward higher blood ketone in the lowest vs highest tertiles of carbohydrate (%EI)
(OR = 2.14, 95% CI: 0.98 – 4.64, P = 0.055) after adjustment for pre-pregnancy BMI, energy intake
and GWG (2nd – 3rd trimester). Urinalysis indicated the presence of ketones in 14% of the overall
population with a positive correlation between urinary and blood ketone levels (n = 19, rS = 0.717, P
<0.001).
Conclusion: Women consuming lower levels of carbohydrate intake (%E) may have higher levels of
blood ketones, suggesting that carbohydrate intake was not sufficient to prevent ketogenesis from
fat stores. This difference may not be clinically significant (<0.5 mmol/L).
110
4.1 Introduction
Pregnancy is a transient state in a woman’s life characterised by metabolic changes to support the
normal growth and development of her fetus (150). However, women with GDM experience greater
metabolic and oxidative stress than their non-GDM counterparts (357), and often have greater
complications of pregnancy including macrosomia and shoulder dystocia (298). With prevalence
reaching 30% in some regions (178), developing cost-effective GDM management strategies is an
urgent priority.
Diet advice and PA are often prescribed as the first line of management for women with GDM, with
insulin and other hyperglycaemic lowering agents as secondary interventions. Self-regulated dietary
manipulation, such as a reduction in carbohydrate intake, has been reported in an observational
study comparing intake of pregnant women with GDM and their non-pregnant counterparts (317),
presumably in the hope of achieving better glycaemic control. Current dietary approaches of
mothers-to-be may be influenced by the growing popularity of LC diets, as exemplified by the
emergence of a plethora of diet books and dietary products (37). With reductions in carbohydrate
intake, the body shifts towards greater fat than carbohydrate oxidation (358), with ketone bodies
produced as a by-product.
Ketones such as BHB are part of normal fat metabolism (106) and can provide energy to the brain
with up to 20% greater efficiency than glucose (111). However, despite this, the fetus has a greater
preference for glucose (153). In fact, an accelerated starvation response can be observed within 16
hours of fasting in the late stage of pregnancy (124), whereby fatty acids are mobilised for maternal
use and glucose is diverted for fetal needs (109). While some studies are in favour of lowering
carbohydrate intake for management of GDM pregnancies (191), others have cautioned about this
approach due a potential link between higher ketone concentrations (Chapter 1.12) and negative
effects on fetal brain development (194, 196).
111
In most clinical settings, pregnant women have urinalysis to test for ketones and other metabolic
markers, including protein, glucose and leukocytes. However, blood ketone levels are deemed more
reflective of the current ketone status in the body than urinalysis, as exemplified by a faster clearance
of ketones in blood that in urine (359). Since ketone levels rise following a prolonged fast in
pregnancy and are considered undesirable for fetal growth and development, our aim was to
determine the range of ketone levels in a random blood and urine sample of women diagnosed with
GDM. We hypothesised that blood and urine ketone levels would be higher in women who consumed
less dietary carbohydrate in the previous 12 hours and may be associated with infant birth weight
and body composition.
4.2 Methods and materials
MAMI 2 was a pilot cross-sectional study that emerged because of slow recruitment rate
encountered during the main RCT, MAMI 1. MAMI 2 provided us with an opportunity to answer a
simple question on the relationship between dietary carbohydrates and body ketones as well as
explore pregnancy outcomes in a GDM population. The study was conducted between December
2016 to March 2018 at Campbelltown and RPA Hospitals in Sydney, Australia. Campbelltown is a
socio-economically disadvantaged suburb in Greater Sydney area with a population close to 160 000
(360). RPA Hospital is situated in the Inner West Sydney area and is a teaching hospital of the
University of Sydney. While the number of residents in the Inner West Sydney area are similar to that
of Campbelltown area, they generally have a higher income (361).
Participants were approached by the research dietitian (J.M.) during GDM education sessions at the
antenatal GDM clinic, either in the morning (RPA) or afternoon (Campbelltown). Recruitment criteria
included pregnant women diagnosed with GDM, ≥18 years and 24-35 weeks gestation. Women were
not excluded if they had multiple-gestation pregnancy, medication and insulin use, or current alcohol
112
or smoking status, but needed to understand the English language. The women previously
underwent a 2-hour, 75 g OGTT using the ADIPS with either an older (1998 ADIPS) (362) or more
recent (2013 ADIPS) (363) diagnostic criteria. Any one of the following were deemed sufficient for
GDM diagnosis:
1998 ADIPS
1.) Fasting BGL ≥5.5 mmol/L;
2.) 2-hour post 75 g oral glucose load ≥8.0 mmol/L.
2013 ADIPS
1.) Fasting BGL 5.1–6.9 mmol/L;
2.) 1-hour post 75 g oral glucose load ≥10.0 mmol/L;
3.) 2-hour post 75 g oral glucose load 8.5–11.0 mmol/L.
The study was approved by the South-Western Sydney Local Health District (HE16/367) and Human
Research Ethics Committee of the Sydney South West Area Health Service (RPA Hospital Zone
HREC/15/RPAH/397) Ethics Committee.
Last recorded maternal weight, mode of delivery, gestation length and neonatal anthropometry,
including birthweight, length, head circumference and Pea Pod® data (RPA hospital only) were all
obtained from electronic medical records. Pea Pod® employs air displacement plethysmography to
assess infant’s body composition, including fat mass (%FM) and fat-free mass (%FFM) (364). While
Pea Pod® data collection is part of routine care at RPA, only 59 neonates (49.1%) had their body
composition assessed. This could be related to several reasons but not limited to mother being
placed in a ward further away from Pea Pod® data collection site. The WHO growth charts (weight
for gestational age, gender specific) (365, 366) and the Australian national birthweight percentiles
(312) were used to determine whether neonates were born <10th percentile (small-for-gestational-
age, SGA) or >90th percentile (large-for-gestational-age, LGA).
113
Figure 4.1 Air displacement plethysmography (Pea Pod®) device.
Consenting participants completed a 1-page form detailing their anthropometry (including self-
reported pre-pregnancy weight), self reported sleep duration, blood glucose levels, as well as a 12-
hour dietary recall, capturing food intake from dinner the previous day, to breakfast the present day.
Twelve-hour recall was employed to ascertain the effects of acute dietary carbohydrate consumed
in the 12 hours prior and their effects on ketone concentrations. Women were also asked to provide
urine and blood samples for ketone analysis (Table 4.2). At RPA, urine analysis is routinely collected
and analysed using 77 Elektronika Kft.® LabStrip U11 Plus. The urine strip tests for 11 parameters
including presence of blood, glucose, specific gravity, bilirubin, protein, nitrite, urobilinogen,
leucocytes, ketones, pH and ascorbic acid. At the initial visit, a nurse demonstrates to participants
how to assess their urine sample against a standard colour chart. Upon subsequent visits, pregnant
women were encouraged to conduct the urine analysis themselves. Since urine testing was not part
of Campbelltown Hospital’s standard practice, urine samples were analysed by the study research
114
dietitian using SIEMENS Multistix® 10 SG. Much like LabStrip U11 Plus, Multistix also assays multiple
parameters including ketones, glucose, bilirubin, ketone, specific gravity, presence of blood, pH,
protein, urobilinogen, nitrite and leukocyte esterase. A ketone conversion factor (0.17212) was used
to convert urine results from mg/dL to a standard metric unit, mmol/L. Aside from ketone, the
additional parameters observed in the urine included glucose, leukocytes and protein.
Table 4.1 MAMI 2 study collection plan
Consent Maternal Anthropometry
Blood and Urine Ketone
Blood Glucose
12-Hour Diet Recall
Infant Anthropometry
Enrolment (24-35 weeks)
Birth of baby
Non-fasting blood ketone tests were conducted by the research dietitian using an Optium™ meter
and Optium™ β-ketone test strips (Abbott, Macquarie Park, Australia). The monitor was designed to
measure either blood glucose or ketone levels (in the form of BHB), provided that corresponding
strips are inserted. Optium™ β-ketone test strips contain 3 electrodes, including a fill trigger, working
and reference electrodes (307). Participants were required to have clean and dry hands prior to
every finger prick using the disposable lancets provided (Accu-Chek, Roche Diagnostics GmbH,
Mannheim, Germany). The ketone monitor requires ~0.6 μL of blood sample on the ketone test strip
to be able to produce a reading, often within 10 seconds. A previous report suggested that hand-
held monitors and their electrodes were accurate and comparable to the laboratory analysis of blood
ketone (129) levels, with strip sensitivity ranging between 0.1-2.0 mmol/L (317).
115
Figure 4.2 FreeStyle Optium Neo meter and corresponding ketone strips (Abbott).
Dietary information was entered by the research dietitian into the Australian nutrition analysis
computer software (FoodWorks Professional Version 8, 2015, Xyris Software, Brisbane, Australia),
based on AUSNUT 2011-2013 database.
4.1.1 Statistical analysis
Statistical analyses were conducted using SPSS (version 24, IBM Australia, St Leonards, Australia) and
SAS Statistical Software, Version 9.4 (SAS Institute Inc., Cary, NC). Descriptive data are presented as
mean ± SEM for continuous variables and numbers (n) and percentages (%) for frequency variables.
Variables were checked for normality. For categorical outcomes, comparisons between groups were
conducted using either the chi square or Fisher’s exact test. For continuous outcomes or differences
between 2 study sites in terms of dietary intake, we used an independent samples t-test or Mann-
Whitney U test. Associations between maternal variables (including age, dietary intake and
anthropometry) and infant body composition were calculated using bivariate analysis, with either
Pearson (rρ) or Spearman (rS) correlation coefficients, where it was deemed appropriate. A
116
participant with twins was excluded from further analysis, thereby bringing the final number of
participants to 160.
Data were transformed if non-normal to meet the assumptions for linear regression. We used crude
and adjusted logistic linear regression to assess the association between varying amounts of
carbohydrate intake on odds of developing elevated blood ketone levels. Ketone levels were
stratified into 3 categories, namely 1) <0.1 mmol/L (n = 28), 2) 0.1 mmol/L (n = 92) and 3) ≥0.2 mmol/L
(n = 41). The latter logistic linear regression model adjusted for maternal pre-pregnancy body mass
index (BMI), age and GWG at enrolment. Carbohydrates, GI and GL were split into their respective
tertiles based on 12-hour maternal distribution of food intake.
Firth corrected logistic regression was used to assess the association between GWG and infant
outcomes (birthweight and body composition), while adjusting for maternal pre-pregnancy BMI
(continuous), age (continuous), weeks gestation at delivery, infant gender and maternal percent (%)
glycosylated haemoglobin (HbA1c) levels (high vs low). The Firth method assists in eliminating bias
associated with small sample sizes (<50 people) (367). Infants’ birth weights were stratified into
either a lower (≤3230g, n = 77) or a higher range (≥3235, n = 76) based on the median value. Total
and 3rd trimester GWG were compared against IOM weight gain guidelines for each BMI category.
Since the study participants varied in their gestational age at enrolment (25-35 weeks), we defined
3rd trimester as 10-13 weeks duration, i.e. from 27 weeks onwards (n = 37). Odds Ratios (OR) and
95% confidence intervals (CIs) were calculated and a P-value <0.05 was used to define statistical
significance.
4.3 Results
A total of 204 women were approached at RPA and 110 at Campbelltown Hospital, of which 121
(59%, with 1 later excluded) and 40 (36%) consented to take part in the study, respectively (Total n
117
= 160). Cumulative recruitment rate of study participants is shown in Figure 4.3. Key screening
outcomes were obtained for most participants and their offspring, including maternal blood ketone
(100%) and glucose (99%) levels, urine analyses (91%), Pea Pod® data (47%, RPA only), birthweight
(96%), baby length (71%), head circumference (89%) and dietary intake (100%).
Figure 4.3 Cumulative recruitment of participants at Royal Prince Alfred (RPA)
and Campbelltown Hospitals.
On average, participants from Campbelltown had comparable BMI, total and GWG at enrolment to
women at RPA, however they were slightly younger and had higher HbA1c (Table 4.2). Blood ketone
levels were similar at both study sites (mean ± SEM mmol/L, RPA 0.1 ± 0.0 versus Campbelltown 0.1
± 0.0) and well below the threshold for ketonaemia (>0.5 mmol/L). Based on the 12-hour recall,
maternal diet was generally comparable between the study sites. The exception was GI and GL, which
were statistically lower in the RPA group (Table 4.3).
0
20
40
60
80
100
120
140
26-Nov-16 06-Mar-17 14-Jun-17 22-Sep-17 31-Dec-17 10-Apr-18
Cu
mu
lati
ve R
ecru
itm
ent
of
par
tici
pan
ts
RPA
Campbelltown
118
We observed a time-gap ~1.4 weeks between delivery and date of last recorded maternal weight.
When stratified according study site, the gap was larger at Campbelltown Hospital (P = 0.002). When
the rate of weight gain for the 3rd trimester (10-13 weeks until delivery) was stratified according to
BMI category and compared to IOM recommended rate of GWG, we found that 59% of women were
below the target, ~24% were above and only 15% met the target rate. Third trimester rate of GWG
was not associated with increased odds of having a larger infant birthweight (results not shown).
When total GWG was compared to IOM’s guidelines, 38% of women did not gain sufficient weight,
31% exceeded and 31% met the weight criteria for their respective BMI category. Using alternate
weight gain recommendations suggested by Faucher et al. (368), 36% of women did not gain
sufficient weight, 34% exceeded and 30% met the weight criteria (results not shown). We also
observed weight loss in some obese women, reaching up to -15 kg by end of pregnancy.
Maternal age was strongly and positively correlated to pre-pregnancy BMI (rS = 0.254, P = 0.001),
however BMI was negatively associated with 2nd-3rd trimester weight gain (rS = -0.256, P = 0.001)
(Figure 4.4). The relationship was also significant when total GWG was compared to maternal pre-
pregnancy BMI (results not shown, n =152, rS = -0.252, P = 0.002). We found no association either
between sleep and blood ketone levels, or between sleep and blood glucose levels.
119
Table 4.2 Maternal characteristics combined or stratified according to their research sites.
Overall
(n = 160)
RPA
(n = 120)
Campbelltown
(n = 40)
P
Age (years) 32.9 ± 0.4 33.3 ± 0.5 31.5 ± 0.8 0.019
Pre-pregnancy BMI 26.3 ± 0.5 25.5 ± 0.5 28.8 ± 1.4 0.060
Underweight n (%) 8 (5.0) 5 (4.1) 3 (7.5)
Normal n (%) 79 (49.1) 66 (54.5) 13 (32.5)
Overweight n (%) 37 (22.3) 27 (22.3) 10 (25.0)
Obese n (%) 37 (22.3) 23 (19.0) 14 (35.0)
Weeks gestation (enrolment) 30.9 ± 0.2 31.0 ± 0.3 30.7 ± 0.5 0.617
Weeks gestation (delivery) 38.7 ± 0.2 38.7 ± 0.1 38.6 ± 0.2 0.340
Weight gain (enrolment) (kg) 8.6 ± 0.4 8.3 ± 0.5 9.6 ± 0.9 0.364
Total weight gain (kg) 10.8 ± 0.5 10.7 ± 0.6 11.5 ± 1.0 0.435 ¥Below target n (%) 58 (38.2) 47 (40.5) 10 (27.8)
¥Within target n (%) 47 (30.9) 34 (29.3) 14 (38.9) ¥Above target n (%) 47 (30.9) 35 (30.2) 12 (33.3)
Rate of weight gain (3rd trimester, n = 34) ¥Below target n (%) 20 (58.8) 15 (55.5) 5 (71.4)
¥Within target n (%) 6 (14.7) 5 (18.5) 1 (14.3) ¥Above target n (%) 8 (23.5) 7 (25.9) 1 (14.3)
Ketone (mmol/L) 0.1 ± 0.0 0.1 ± 0.0 0.1 ± 0.0 0.885
BGL (mmol/L) 5.9 ± 0.1 6.0 ± 0.1 5.4 ± 0.1 0.007
HbA1c (%) 5.1 ± 0.0 5.1 ± 0.1 5.4 ± 0.0 <0.0001
Mode of delivery
Vaginal delivery n (%) 105 (67.3) 76 (65.5) 29 (72.5)
Normal n (%) 75 (71.4) 53 (69.7) 22 (75.9)
Vacuum Extraction n (%) 15 (14.3) 8 (10.5) 7 (24.1)
Forceps-Liftout n (%) 15 (14.3) 15 (19.7) 0 (0)
Elected caesarean n (%) 30 (19.2) 19 (16.4) 11 (27.5)
Emergency caesarean n (%) 21 (13.5) 21 (18.1) 0 (0)
Abbreviations: BGL – blood glucose levels; RPA – Royal Prince Alfred (Hospital) ¥ Institute of Medicine gestational weight gain criteria; Values presented as mean ± SEM; Differences between groups determined using Independent t-test, P <0.05 deemed significant.
120
Table 4.3 Maternal dietary characteristics based on the 12-hour recall, combined or stratified according to the research sites.
Overall
(n = 160)
RPA
(n = 120)
Campbelltown
(n = 40) P
Energy (kJ) 3890 ± 90 3870 ± 100 3940 ± 190 0.724
Protein (g) 52.6 ± 1.6 52.9 ± 2.0 51.8 ± 2.9 0.918
Total fat (g) 38.1 ± 1.4 37.9 ± 1.7 38.7 ± 1.1 0.820
Saturated fat (g) 13.4 ± 0.6 13.1 ± 0.6 14.5 ± 1.1 0.295
Carbohydrate (g) 87.2 ± 2.2 86.0 ± 2.5 90.6 ± 5.2 0.375
Sugars (g) 30.8 ± 1.3 30.5 ± 1.4 31.6 ± 2.8 0.720
Starch (g) 56.0 ± 1.7 55.1 ± 1.9 58.7 ± 3.7 0.356
Dietary fibre (g) 12.3 ± 0.5 12.4 ± 0.5 12.0 ± 1.1 0.466
GI 52 ± 1 51 ± 1 55 ± 1 0.008
GL 46 ± 1 44 ± 2 50 ± 3 0.050
Protein (%EI) 23.1 ± 0.5 23.3 ± 0.6 22.7 ± 0.9 0.720
Carbohydrates (%EI) 38.4 ± 0.9 38.3 ± 1.0 38.7 ± 1.7 0.816
Total fat (%EI) 35.0 ± 0.7 34.8 ± 0.9 35.4 ± 1.4 0.730
Saturated fat (%EI) 12.4 ± 0.3 12.1 ± 0.4 13.4 ± 0.7 0.163
Abbreviations: EI – Energy intake; GI – Glycaemic index; GL – Glycaemic load; RPAH – Royal Prince Alfred (Hospital). Values presented as mean ± SEM; Differences between groups determined using Independent t-test, P <0.05 deemed significant.
121
Figure 4.4 Association between maternal pre-pregnancy BMI and age. Bivariate analysis conducted using a Spearman correlation (rS = 0.254, P = 0.001).
Figure 4.5 Correlation of maternal weight gain at enrolment and maternal pre-pregnancy BMI. Grey areas indicate Institute of Medicine (IOM) gestational weight gain recommendations by BMI category. Bivariate analysis conducted using Spearman correlation (n = 157, rS = -0.255, P = 0.001).
-10
-5
0
5
10
15
20
25
30
15 25 35 45 55
Ges
tati
on
al w
eigh
t ga
in a
t en
rolm
ent
(kg)
Pre-pregnancy BMI
15
20
25
30
35
40
45
50
55
10 20 30 40 50 60
Age
(ye
ars)
Pre-pregnancy BMI
rS = 0.254 P = 0.001
n = 157 rS = -0.255 P = 0.001
122
4.3.1 Neonatal characteristics and anthropometry
Of 159 births with information from medical records, 5 were born prematurely (<37 weeks
gestation). Overall, there were more male than female neonates at both sites and no differences in
their birthweights (Table 4.4). Based on the WHO birthweight-for-age growth chart, approximately
11% of all birthweights met the SGA criteria, 4% were LGA. Using the Australian National birthweight
percentiles, proportions of neonates deemed SGA were higher (14%) and comparable for LGA (5%).
When compared to RPAH, Campbelltown site had higher rates of LGA regardless of which criteria
was used to ascertain birthweight percentile.
Infants from Campbelltown Hospital had greater birth length (P = 0.020) and subsequently lower
ponderal index (PI) (P = 0.006) than infants from RPA Hospital. Higher maternal pre-pregnancy BMI
was a predictor of greater birthweight (rS = 0.304, P <0.0001), infant head circumference (rS = 0.301,
P <0.0001) and birth length (rS = 0.235, P <0.011). Maternal GWG at enrolment (24-35 weeks
gestation) was positively correlated with birthweight (rS = 0.170, P = 0.036). When compared to
women meeting the total gestational weight gain guidelines, those exceeding them had 3.4-fold
increased odds of having an infant of higher birthweight (OR = 3.40, 95% CI:1.26 - 9.74, P = 0.007)
(Figure 4.6). This model was strongly and significantly influenced by maternal age and gestational
age at delivery.
Body composition data were available for 59 neonates at the RPA Hospital. Female neonates had a
significantly higher %FM than their male counterparts (P <0.0001). Infant %FM was positively
correlated with birthweight (rρ = 0.424, P = 0.003) and infant length (rρ = 0.399, P = 0.001). In
addition, we found no increased odds of having an infant with higher %FM at birth in women who
exceeded total GWG guidelines (OR = 1.04, 95% CI: 0.20 - 5.25, P = 0.869) (Figure 4.6). In addition,
we found no association with maternal dietary intake and any infant outcome (data not shown).
123
Table 4.4 Infant anthropometry outcomes combined or stratified according to their research sites, where possible.
Overall (n = 160) RPA (n = 120) Campbelltown (n = 40) P
Male n (%) 84 (56.0) 65 (57.5) 19 (51.4)
Female n (%) 66 (44.0) 48 (42.5) 18 (48.6)
Birthweight (kg) 3250 ± 40 3250 ± 40 3250 ± 90 0.952
Male mean (kg) 3260 ± 50 0.922
Female mean (kg) 3270 ± 50
WHO criteria SGA n (%) 17 (11.3) 12 (10.0) 5 (12.5)
WHO criteria LGA n (%) 6 (4.0) 2 (1.7) 4 (10.0)
ANBP SGA n (%) 21 (13.9) 14 (12.3) 7 (18.9)
ANBP LGA n (%) 7 (4.6) 2 (1.8) 5 (13.5)
Length (cm) 50 ± 0.2 49 ± 0.2 51 ± 0.5 0.020
HC (cm) 34.4 ± 0.1 34.4 ± 0.1 34.5 ± 0.3 0.842
PI (kg/m3) 2.7 ± 0.0 2.7 ± 0.0 2.5 ± 0.1 0.006
Fat Mass (%) 10.1 ± 0.6
Male (%) 9.3 ± 0.9 0.044
Female (%) 11.8 ± 0.7
¥Male (%) 8.6 ± 0.5 <0.0001 ¥Female (%) 11.8 ± 0.7
Fat Free Mass (%) 89.9 ± 0.6
Abbreviations: ANBP - Australian National birthweight percentiles (312); %FFM/FM – Percent fat free mass/fat mass; HC - Head circumference; LGA – large-for-gestational age; SD – standard deviation; SGA – small-for-gestational age; Ponderal index – PI; WHO criteria – World Health Organization criteria (365, 366). ¥ Comparing mean of %FM between male and female neonates, after removal of an extreme outlier. Values presented as mean ± SEM; Differences between groups determined using Independent t-test, P <0.05 deemed significant.
124
0
12O
dds R
atio
(95%
Confidence Inte
rval)
Birthweight (kg)
1
3.4
1.0
% Fat Free Mass
Figure 4.6 Odds ratios (OR) of higher birthweight or percent Fat Free Mass when Institute of Medicine’s (IOM) weight gain guidelines are exceeded. Multivariable models adjusted for BMI, maternal age, weeks gestation at delivery, baby gender and Hba1c category.
4.3.2 Blood ketone and carbohydrate intake
Tertiles of carbohydrate intake (%EI) are shown in Table 4.5. Compared to the highest tertile, the
lowest tertile had a 2-fold increased odds of having higher blood ketone levels, although the increase
was not statistically significant at the 5% level (OR = 2.05, 95% CI: 0.97 – 4.34, P = 0.060) (Table 4.6).
When the model was adjusted for pre-pregnancy BMI, energy intake and GWG at the time of
enrolment, we observed similar findings (OR = 2.14, 95% CI: 0.98 – 4.64, P = 0.055).
However, we found no relationship between highest vs lowest tertile of GI or GL and blood ketone
levels (crude GI OR = 0.67, 95% CI: 0.32 – 1.41, P = 0.300; crude GL OR = 1.08, 95% CI: 0.52 - 2.26, P
= 0.832). Further adjustment for pre-pregnancy BMI, energy intake and GWG at the time of
enrolment yielded similar results (GI OR = 0.68, 95% CI: 0.32 - 1.44, P = 0.315; GL OR = 1.11, 95% CI:
0.53 - 2.34, P = 0.776).
125
Table 4.5 Tertiles of carbohydrate intake (%EI) from the 12-hour recall.
Percent (%EI) kJ Carbohydrate Mean Min Max
Tertile 1 26.4 4.3 33.4
Tertile 2 38.5 33.5 43.6
Tertile 3 50.0 43.7 63.2
Table 4.6 Carbohydrate content and odds of developing elevated ketones.
OR 95% CI Upper 95% CI Lower
P
Model 1† Tertile 3 versus 1 2.05 0.97 4.34 0.060
Tertile 3 versus 2 0.75 0.36 1.58 0.448
Model 2‡ Tertile 3 versus 1 2.14 0.98 4.64 0.055
Tertile 3 versus 2 0.77 0.37 1.63 0.500
Abbreviations: CI - confidence intervals; OR = Odds Ratio;
P <0.05 deemed statistically significant
† Model 1 - Crude logistic regression model
‡ Model 2 - adjusted for pre-pregnancy body mass index (BMI), age and gestational weight gain at enrolment. Carbohydrates
tertiles based on the distribution of intake from the 12-hour diet recall.
126
4.3.3 Blood ketone and urine ketone
Random urinalysis suggested presence of ketone (14%), glucose (7%), protein (29%) and leukocytes
(40%) in a proportion of women at both study sites (Figure 4.7). Compared with RPA, Campbelltown
Hospital had a greater proportion of women testing positive for protein and ketone in their urine
(protein P = 0.026; ketone P <0.0001). In women that tested positive for ketones in urine, their
values were strongly and positively correlated with blood ketone levels (n = 19, R² = 0.37, rS = 0.717,
P <0.001) (Figure 4.8).
Figure 4.7 Urinalysis of pregnant women with gestational diabetes mellitus (GDM). Blue represents the Royal Prince Alfred (RPA) and green represents the Campbelltown hospital. Chi square test and Fisher’s test were used to ascertain statistical difference, with P <0.05 (*) and P <0.001 (***) deemed as statistically significant.
127
Figure 4.8 Correlation between urine samples testing positive for ketone and their corresponding blood ketone levels. Bivariate analysis conducted using Spearman correlation (n = 19, rS = 0.717, P = 0.001).
-5
0
5
10
15
20
25
30
0.0 0.1 0.2 0.3 0.4 0.5
Uri
ne
Ke
ton
e (m
mo
l/L)
Blood Ketone (mmol/L)
n = 19 rS = 0.717 P = 0.001
128
4.4 Discussion
Our pilot cross-sectional study suggested a trend of increased (non-fasting) blood ketone levels when
lowest tertile of carbohydrate intake in the previous 12 hours (~40% EI) was compared to highest in
women with GDM, although non-significant. In another study of similar design and population,
consumption of carbohydrates at bedtime resulted in lower fasting ketone concentrations (369).
Approximately 40% of participants at Campbelltown Hospital had ketonuria (vs 3% at RPAH). Given
that urinalysis was self-reported at RPAH and assessed by the researcher at Campbelltown Hospital,
the difference could be attributed to observer bias (133) as well as sample collection at different
time of day (morning vs afternoon). In a study by Robinson et al. 2018 ~22% of women between 16
to 28 weeks gestation demonstrated detectable ketones, which subsequently dropped to 8% at 36
weeks gestation (370).
Our findings contrast with an observational study by Gin and colleagues (317), which found no
correlation between dietary variables and ketonaemia. This could be explained in part by the
relatively high intake (>200 g) of carbohydrate per day in their cohort of women, a level which may
have been insufficiently low to induce higher level of ketones in blood. We also demonstrated a
strong correlation between urine and blood ketone levels, which is in agreement with previous
studies (130, 369).
We found a negative correlation between maternal pre-pregnancy BMI and GWG. However, ~70%
of pregnancies still did not meet total GWG guidelines. While the current IOM’s GWG
recommendation is fixed for all classes of obesity (5-9 kg), evidence from a comprehensive
systematic review suggested ideal GWG should differ according to classes of obesity (Class I: 5-9 kg,
Class II: 1 to <5 kg, Class III: no weight gain) as well as ethnicity (368). In fact, when different classes
of obesity were taken into consideration and then compared to IOM guidelines, 3% (IOM 31% vs
Faucher et al. 34%) more women exceeded weight gain recommendations.
129
Our study did not find any association between rate of GWG in the 3rd trimester and infant
birthweight. Sridhar and colleagues (371) reported similar findings. However, their study suggested
that GWG below IOM’s guidelines in the 2nd trimester generally increased the risk of having a SGA
infant, but not among overweight or obese mothers (371). Excess weight gain in women of normal
BMI in all trimesters increased the risk of LGA infant, but only 3rd trimester weight gain for overweight
or obese women increased the risk (371). This suggests that there is still an opportunity to intervene
and control the rate of GWG in GDM pregnancy (diagnosis 24-28 weeks gestation) and influence
infant birthweight.
While our study observed a weight loss of up to 15 kg among obese pregnant women, a retrospective
study consisting of ~26,000 participants suggested that 3rd trimester weight loss in overweight and
obese women with GDM improved maternal and neonatal outcomes (372). However, this was at the
expense of increased odds of preterm delivery and having a SGA infant (372). With overweight and
obesity on the rise (25) and women entering pregnancy with the excessive weight (373), more
research on weight management and weight loss in overweight and obese women prior to
conception is warranted.
Despite being part of RPA Hospital’s routine care to collect Pea Pod® values at birth, only 49% of
infants had their body composition measured. Female infants had higher %FM when compared to
males, a finding supported by other studies (374, 375). When Pea Pod® measurement was compared
to maternal dietary intake on previous 12 hours, we found no observable association with either
%FM or FFM. In contrast, Kizirian and colleagues reported that higher maternal GI and carbohydrate
intake during the 3rd trimester of pregnancy were associated with lower FFM and FM index,
respectively (303). The lack of correlation in the present study could partially be attributed to our
dietary collection tool, which did not capture a full day’s worth of food intake. While the relative
validity of an electronic 12-hour dietary recall was in good agreement with a food frequency
130
questionnaire and four dietary records (376), a 3-day food diary capturing maternal weekday and
weekend dietary intake would have provided a better insight into a participant’s typical diet (377).
Although the present study found no association between sleep duration and blood glucose values,
others have reported that lack of sleep could impair glucose metabolism (378, 379). Since sleep was
self-reported, it may not be reflective of participant’s actual sleep duration, nor could it take into
account periods of sleep disturbances, a common occurrence (~46%) among expectant mothers
(380). While Coronary Artery Risk Development in Young Adults (CARDIA) Sleep Study suggested a
moderate correlation between self-reported and objectively measured sleep, the mean of self-
reported sleep duration was 1-hr longer than mean of objectively measured sleep (381). Since sleep
has clinical implications for blood glucose levels, this raises the importance of establishing sleep
patterns, particularly in individuals with elevated blood glucose levels.
Our study had strengths and limitations. We were limited in the reporting of blood ketone levels due
to the narrow working range of Optium™ ketone monitors. This meant that values such as 0.11 vs
0.14 mmol/L were not different (i.e. 0.1 mmol/L) and a 0.0 mmol/L reading could mislead the
observer because under normal living conditions ketones are always present in the blood (106).
Interestingly, Gin and colleagues measured blood ketone levels using the same device, but reported
their measurements with up to 2 decimal places (317). Considering the device’s working range
limitations, we resorted to using a categorical logistic regression analysis when determining the
relationship to maternal carbohydrate intake. While this may have been the better option in terms
of analysis, we still had an imbalance within the 3 categories of ketones, which could have influenced
our results. Since we investigated non-fasting blood ketones, their levels could have declined due to
a surplus of carbohydrates following breakfast, thereby preventing us from determining the true
extent of their rise in the morning. We believe that future studies, which specifically assess fasting
blood ketone levels using a device with greater precision will demonstrate greater odds of having a
rise in blood ketones levels due to lower consumption of carbohydrate (%EI). In addition, due to
131
time-restraint, the study did not capture data on ethnicity, which could have potentially explained
the differences in blood glucose concentrations between the sites.
4.5 Conclusion
Under standard living conditions in GDM population, we did not find extreme high blood ketone
levels. Our study, however, suggested a trend between lower compared to higher tertiles of
carbohydrate intake and rise in blood ketone levels, although non-significant and below the
threshold for ketonaemia (<0.5 mmol/L).
132
Chapter 5
___________________________________________________
Conclusions and directions for the future
133
Discussion of main findings and future directions
Overweight and obesity, delayed age of motherhood and adoption of the new GDM diagnostic
criteria are just some of the factors driving an increase in the rates of GDM (160, 176, 177). GDM is
associated with numerous complications in pregnancy as well as epigenetic changes in the offspring,
instigating a vicious cycle of metabolic diseases for many generations to come (165). In Chapter 1,
we summarised current evidence supporting the use of LC diets to normalise blood glucose
concentrations within the accepted ranges for GDM. Unfortunately, there were only a handful of
studies available (44, 190, 191, 295). On closer inspection of the literature, studies on LC diets in non-
pregnant populations indicated conflicting evidence with respect to their effects on weight loss,
serum glucose, insulin, cholesterol (LDL-C and HDL-C) and triglyceride concentrations. Inconclusive
findings were largely attributed to heterogeneity in study design, duration and differences in target
populations (e.g. obesity, T2DM).
In Chapter 2, a systematic review and meta-analysis was conducted specifically targeting prospective
observational studies that reported on dietary intake and PA levels before and in early pregnancy on
the risk of developing GDM. With respect to dietary studies, frequent consumption of potato and
protein (% energy) derived from animal sources, particularly processed meats, suggested a higher
risk of GDM. On the other hand, the Mediterranean diet (MedDiet), dietary approaches to stop
hypertension (DASH) diet and higher alternate healthy eating index (AHEI) reduced the risk by ~15-
38% (182). While engagement in PA has been previously shown to improve insulin sensitivity (382),
our meta-analysis suggested for the first time that engagement in leisure time PA before and in early
pregnancy above >90 min/week, may reduce the odds of GDM by 46% (182). These findings raise the
importance of modifiable lifestyle factors in disease prevention and provide hope and possible
direction for future management of GDM.
While there is no consensus at present on the most effective diet for GDM treatment (185), certain
endocrine societies recommend a LC diet to achieve normoglycaemia (188, 383), despite the lack of
134
evidence for their safety in pregnancy. In Chapter 1, we established that LC diets can augment
enhanced FAO to BHB and that this metabolic parameter has been inversely correlated to child’s
intelligence (196). The implication is that LC diets in pregnancy might reduce child IQ by increasing
the concentration of maternal blood ketones. Therefore, MAMI 1 RCT was a pilot study undertaken
to investigate the safety of LC diets (Chapter 3) and MAMI 2 explored the association of carbohydrate
intake in the previous 12 hours with BHB levels (Chapter 4). To our knowledge, the RCT is the longest
(6 weeks) intervention study which compared the effects of modestly lower carbohydrate diets and
conventional healthy diets (135 vs 180-200 g/day carbohydrate) on blood BHB levels in GDM. While
there were no observed differences in BHB levels following a reduction in absolute carbohydrate
intake, there were also no apparent benefits for glycaemia and insulin dosing. Certainly, the most
surprising finding was a statistically smaller (~1cm) head circumference in the treatment group when
compared to routine dietary advice. According to the Australian birthweight database and Centre for
Disease Control and Prevention (CDC) growth charts (313-315), this meant that the head
measurements were either within the 10th to 25th percentile, or between 25th to 50th percentile for
MLC and RC diets, respectively. Because there is a strong correlation between head circumference
and brain volume (344), as well as brain volume and intelligence (345), this raises questions of safety
or negative long-term effects on offspring brain development. However, in the context of GDM, it
could also be suggested that a smaller head circumference may be beneficial in promoting more
vaginal deliveries, however we found no statistical difference between modes of delivery between 2
study arms of MAMI 1 study (P = 0.203).
A follow up study on MAMI 1 offspring could be useful in determining whether differences in head
circumference persist into childhood and whether there are any implications for their psychomotor
performance. However, factoring in the limited number of participants who completed MAMI 1
study (n = 33), the sample size may be too low. To help address this issue and to further investigate
the relationship between carbohydrate restriction and infant head circumference, there are 3
possible paths to undertake (Figure 5.1). Firstly, we could explore the published literature on dietary
135
intake in pregnancy and neonatal measurements, in the hope of obtaining not just carbohydrate
intake and head circumference, but also other metabolic parameters of interest (e.g. glycaemia,
BHB). According to a recent study exploring the effects of western, healthy and traditional dietary
patterns in early pregnancy, there were no reported differences in infant head circumference
between the study groups. However, in all these groups, carbohydrate intake (210-341 g/day)
exceeded the amount prescribed in the MAMI 1 study (384). On the other hand, a cross sectional
study suggested that higher %EI from carbohydrate was associated with lower infant head
circumference (385). The authors, however, noted that the study effect size estimate was reasonably
small (β ≤−0.01), thereby limiting generalisability of their findings (385). Since neither study
investigated serum BHB concentrations in a GDM population, it may be difficult to collate evidence
from studies which have been designed to test a different hypothesis.
136
Figure 5.1 Framework for further investigation of the possible relationship between
modestly lower carbohydrate (MLC) diet or energy restriction and infant head
circumference.
Secondly, we could use animal models comparing low and high carbohydrate diets as this has been
a relatively popular focus of study. In one study, a ketogenic diet in mice resulted in embryonic brain
growth deviating from the established norm, suggesting possible behavioural changes after birth
(198). According to another rodent model, carbohydrate quantity rather than type and quality
influenced fetal brain mass, with lower intake associated with a reduction in brain mass (386).
Restriction of maternal carbohydrates was also reported to lower concentrations of an important
Reported differences
in embryonic organ
development
Overcome energy and
nutrient imbalance
Likely to be a healthy
pregnant population
Differences in study design
and data collection
timepoints
Lower brain mass
Energy restriction
Lower head
circumference ?
Lower concentration of 5-
hydroxytryptamine
neurotransmitter
Systematic
review/Meta-analysis
is warranted
Not all metabolic
parameters
investigated
Studies on long-term
effects of LC diets on
behaviour are needed
Follow up studies on
offspring behaviour
are needed
Reduce pain and discomfort
of GDM pregnancy (e.g.
finger pricks)
Assessment of neonatal
biochemistry at birth
Human studies on
pregnancy
Animal studies
during gestation Larger MAMI 1
sample size
Modestly lower
carbohydrate intake
137
neurotransmitter (5-hydroxytryptamine) for brain development (386). These findings could explain
the mechanism behind lower head circumference observed in the MLC group of our MAMI 1 study.
Clearly, animal studies provide sufficient reason to undertake human studies of larger sample size.
Lastly, we propose that the MAMI 1 study is conducted on a larger scale to confirm findings reported
in our pilot study. Apart from simply increasing participant numbers, incorporating a control group
of healthy pregnant women without GDM could assist in establishing whether the head
circumference phenomenon was mediated by differences in glycaemia, insulinaemia, or ketonaemia.
The pilot study also demonstrated the relative difficulty in increasing protein and fat intake without
producing an energy deficit (~1000-1500 kJ), despite dietary counselling. We speculate that the
relatively high cost of meat (336) and advice to restrict intake of fish, processed or deli-style cold
meats due to food safety concerns (335), may be partly to blame. The protein leverage hypothesis is
also relevant. It proposes that individuals have a protein target that must be reached each day
irrespective of carbohydrate and fat intake. Thus higher intake of high-protein foods brings about a
reduction in intake of carbohydrate and fat, and therefore energy (387). On the other hand, advice
to consume a high carbohydrate (or a high fat diet) may inadvertently increase total energy intake in
order to reach the protein target. In the scientific literature, high protein intake is positively
associated with neonatal head circumference (388), but this is largely driven by the adverse effects
of very low protein diets.
High fat diets have also been recommended to bring about reductions in carbohydrate intake
without increasing protein. However, in GDM, this may be a cause for concern. High-fat diets during
pregnancy have been associated with an increase in insulin resistance in rodents (389), fetal
overgrowth, particularly if maternal FFAs were high (343, 390), and effects on child neurobehavioral
development (391). In future studies, it may therefore be helpful to compare 3 diets in one RCT,
wherein the third diet achieves a low carbohydrate intake by increasing the intake of beneficial fats
(Mediterranean-style diet). It would also be informative to assess neonatal biochemical and
138
behavioural parameters. Altered maternal glucose metabolism in diabetes can cause neonatal
hypoglycaemia, hyperbilirubinaemia and hypocalcaemia, which in turn have been associated with
fetal congenital malformations (392). At present, we have little understanding of whether these
biochemical markers improve (or not) when the mother consumes a MLC diet.
Aside from an energy deficit, the intervention resulted in a decreased intake of carbohydrate-rich
foods, such as breads. In Australia, most breads are fortified with iodised salt (by law) and are a
major source of iodine. Recent studies suggest that many Australian women do not ingest enough
iodine to satisfy the increased requirements during pregnancy, and this is associated with deficits in
scores for numeracy and literacy (393). While iodine undoubtedly plays an important role in fetal
neurological development and fetal thyroid function (394), women ingesting LC and MLC diets should
be closely monitored for iodine, particularly in participants that do not take pregnancy supplements.
Improved methods of tracking ketones are also needed in future studies. The typical pregnant
woman is diagnosed with GDM at ~24 weeks gestation and delivers at ~40 weeks. During that time,
she will have performed ~450 finger prick tests to ascertain fasting and postprandial glucose
concentrations. In the MAMI 1 study, the mother was expected to perform additional finger prick
tests, involving more time, pain and discomfort. To overcome the barrier and promote more
frequent measurements, non-invasive devices will be appealing (395). ‘Flash’ glucose testing and
other new devices should be considered in the planning of future studies. GlucoTrack is one such
non-invasive device, which can determine glucose fluctuations using a clip which attaches to the
earlobe and has a cable extending to a reader. According to a recent study in a T2DM population,
GlucoTrack was reasonably accurate, regardless of individual’s duration of diabetes, %HbA1c and
smoking history (396).
Unfortunately, with regards to ketone assessment, there are no real-time devices on the market to
assess blood BHB concentration. Current ketone detectors are more suitable for detecting
ketoacidosis in individuals with T1DM or poorly management of T2DM (397). There may be a market
139
for continuous ketone readings in people following LC diets as well as GDM pregnancy. This would
potentially map out important periods of BHB spikes. This information could be useful for dietary
counselling and meal timing to prevent further BHB peaks and potential undesirable effects on fetal
brain development.
Given the limited and often conflicting evidence in the scientific literature, it is not surprising that
there is lack of consensus regarding dietary management of GDM. The studies described in this thesis
indicate that it is critical that the potential unintended consequences of well-meaning dietary advice
be considered. There is room for improvement in the usual nutrition advice given to women with
GDM. It should not be acceptable for advice in GDM to be based on ‘intuition’ and anecdontal
evidence. Rather, dietary advice backed by strong scientific evidence is likely to promote better
outcomes for the mother and her offspring. Taken together, our findings provide a rational basis for
appropriately powered studies investigating the safety of LC diets in pregnancy.
140
Reference List
1. WHO. Diagnostic criteria and classification of hyperglycaemia first detected in pregnancy
Switzerland; 2013.
2. Academy of Nutrition and Dietetics. Evidence analysis manual: steps in the academy: Evidence
analysis process. Research and Strategic Business Development. 2012.
3. Weickert MO. Nutritional modulation of insulin resistance. Scientifica. 2012;2012:1-15.
4. Cotter DG, Schugar RC, Crawford PA. Ketone body metabolism and cardiovascular disease.
American Journal of Physiology Heart and Circulatory Physiology. 2013;304(8):H1060-H76.
5. Banting W. Letter on corpulence, addressed to the public: Third edition. Obesity Research.
1993;1(2):153-63.
6. Atkins RC. Dr. Atkins’ Diet revolution: The high calorie way to stay thin forever. New York, NY:
D. McKay Co; 1972.
7. Freeman JM, Kossoff EH, Hartman AL. The ketogenic diet: One decade later. Pediatrics.
2007;119(3):535-43.
8. Johnson AR. The Paleo diet and the American weight loss utopia, 1975–2014. Utopian Studies.
2015;26(1):101-24.
9. Gibson AA, Seimon RV, Lee CMY, Ayre J, Franklin J, Markovic TP, et al. Do ketogenic diets really
suppress appetite? A systematic review and meta‐analysis. Obesity Reviews. 2015;16(1):64-76.
141
10. Nordmann AJ, Nordmann A, Briel M, Keller U, Yancy WS, Jr., Brehm BJ, et al. Effects of low-
carbohydrate vs low-fat diets on weight loss and cardiovascular risk factors: a meta-analysis of
randomized controlled trials. Archives of Internal Medicine. 2006;166(3):285-93.
11. Fields H, Ruddy B, Wallace MR, Shah A, Millstine D. Are low-carbohydrate diets safe and
effective? The Journal of the American Osteopathic Association. 2016;116(12):788-93.
12. Hashimoto Y, Fukuda T, Oyabu C, Tanaka M, Asano M, Yamazaki M, et al. Impact of low-
carbohydrate diet on body composition: Meta-analysis of randomized controlled studies: Low-
carbohydrate diet and body composition. Obesity Reviews. 2016;17(6):499-509.
13. Wheeler ML, Dunbar SA, Jaacks LM, Karmally W, Mayer-Davis EJ, Wylie-Rosett J, et al.
Macronutrients, food groups, and eating patterns in the management of diabetes. Diabetes
Care. 2012;35(2):434-45.
14. NHMRC. Australian Dietary Guidelines. Canberra: NHMRC.; 2013.
15. U.S. Department of Health and Human Services and U.S. Department of Agriculture. 2015–2020
Dietary guidelines for Americans.; 2015.
16. Trumbo P, Schlicker S, Yates AA, Poos M, Food, Nutrition Board of the Institute of Medicine, The
National Academies. Dietary Reference intakes for energy, carbohydrate, fiber, fat, fatty acids,
cholesterol, protein and amino acids. Journal of the American Dietetic Association.
2002;102(11):1621-30.
17. NHMRC and Ministry of Health. Nutrient reference values for Australia and New Zealand -
Macronutrient balance 2014 [updated 2nd April 2014. Available from:
https://www.nrv.gov.au/chronic-disease/macronutrient-balance.
142
18. Osler W MT. The Principles and Practice of Medicine. New York: Appleton and Co.; 1923.
19. Bravata DM, Sanders L, Huang J, Krumholz HM, Olkin I, Gardner CD, et al. Efficacy and safety of
low-carbohydrate diets: A systematic review. The Journal of the American Medical Association.
2003;289(14):1837-50.
20. Di Cesare M, Bentham J, Stevens GA, Zhou B, Danaei G, Lu Y, et al. Trends in adult body-mass
index in 200 countries from 1975 to 2014: a pooled analysis of 1698 population-based
measurement studies with 192 million participants. The Lancet. 2016;387(10026):1377.
21. ABS. National Health Survey: First results, 2014–2015. Canberra: ABS; 2015.
22. Yang L, Colditz GA. Prevalence of overweight and obesity in the United States, 2007-2012. The
Journal of the American Medical Association Internal Medicine. 2015;175(8):1412-3.
23. Chu SY, Kim SY, Bish CL. Prepregnancy obesity prevalence in the United States, 2004-2005.
Maternal and Child Health Journal. 2009;13(5):614-20.
24. Davies GALMD, Maxwell CMD, McLeod LMD, Gagnon RMD, Basso MRN, Bos HMD, et al. Obesity
in pregnancy. Journal of Obstetrics and Gynaecology Canada. 2010;32(2):165-73.
25. Wise J. Obesity rates rise substantially worldwide. The British Medical Journal. 2014;348:g3582-
g.
26. Kelly T, Yang W, Chen CS, Reynolds K, He J. Global burden of obesity in 2005 and projections to
2030. International Journal of Obesity. 2008;32(9):1431-7.
27. UN News Centre. UN projects world population to reach 8.5 billion by 2030, driven by growth in
developing countries: UN; 2015
143
28. Hildén K, Hanson U, Persson M, Fadl H, Institutionen för h, Örebro u, et al. Overweight and
obesity: A remaining problem in women treated for severe gestational diabetes. Diabetic
Medicine. 2016;33(8):1045-51.
29. Ovesen P, Rasmussen S, Kesmodel U. Effect of prepregnancy maternal overweight and obesity
on pregnancy outcome. Obstetrics and Gynecology. 2011;118(2):305-12.
30. Ortega FB, Lavie CJ, Blair SN. Obesity and cardiovascular disease. Circulation Research.
2016;118(11):1752-70.
31. Berger NA. Obesity and cancer pathogenesis. Annals of the New York Academy of Sciences.
2014;1311(1):57-76.
32. Wiklund P. The role of physical activity and exercise in obesity and weight management: Time
for critical appraisal. Journal of Sport and Health Science. 2016;5(2):151-4.
33. Cameron AJ, Waterlander WE, Svastisalee CM. The correlation between supermarket size and
national obesity prevalence. BMC Obesity. 2014;1(1):27.
34. Greenway FL. Physiological adaptations to weight loss and factors favouring weight regain.
International Journal of Obesity. 2015;39(8):1188-96.
35. Romieu I, Dossus L, Barquera S, Blottière HM, Franks PW, Gunter M, et al. Energy balance and
obesity: What are the main drivers? Cancer Causes & Control. 2017;28(3):247-58.
36. Yancy JWS, Olsen MK, Guyton JR, Bakst RP, Westman EC. A low-carbohydrate, ketogenic diet
versus a low-fat diet to treat obesity and hyperlipidemia: A randomized, controlled trial. Annals
of Internal Medicine. 2004;140(10):769.
144
37. Samji S. Low carb diets. Student British Medical Journal. 2004;12:1-5.
38. Labiner-Wolfe J, Jordan Lin C-T, Verrill L. Effect of low-carbohydrate claims on consumer
perceptions about food products' healthfulness and helpfulness for weight management.
Journal of Nutrition Education and Behavior. 2010;42(5):315-20.
39. Pirovano T. Health & wellness trends - The speculation is over. Frozen Food Digest. 2006:29.
40. Lenzer J. Robert Coleman Atkins. The British Medical Journal. 2003;326(7398):1090.
41. Kemp E, Burton S, Creyer EH, Suter TA. When do nutrient content and nutrient content claims
matter? Assessing consumer trade-offs between carbohydrates and fat. The Journal of
Consumer Affairs. 2007;41(1):47-73.
42. Lopez M-T, Soliah L, Walter J. Low-carb food choices: The net effect. Journal of Family and
Consumer Sciences. 2005;97(2):70.
43. Seitz VA, Razzouk N, Triyangkulsri W. Factors influencing the purchasing decision process of low-
carbohydrate products. Journal of Food Products Marketing. 2007;13(4):23-38.
44. Moreno-Castilla C, Hernandez M, Bergua M, Alvarez MC, Arce MA, Rodriguez K, et al. Low-
carbohydrate diet for the treatment of gestational diabetes mellitus: A randomized controlled
trial. Diabetes Care. 2013;36(8):2233-8.
45. Erlanson-Albertsson C, Mei J, Aptitkontroll, Lund U, Appetite R, Lunds u. The effect of low
carbohydrate on energy metabolism. International Journal of Obesity. 2005;29(S2):S26-S30.
46. St Jeor ST, Howard BV, Prewitt TE, Bovee V, Bazzarre T, Eckel RH, et al. Dietary protein and
weight reduction: a statement for healthcare professionals from the Nutrition Committee of the
145
Council on Nutrition, Physical Activity, and Metabolism of the American Heart Association.
Circulation Research. 2001;104(15):1869-74.
47. Steer T. The Atkins Diet: Apparent endorsements by glamorous celebrities have boosted the
popularity of this high protein, low carbohydrate diet. Toni Steer looks at the evidence to
establish what advice practice nurses can give to patients. Practice Nurse. 2003:44.
48. Foster GD, Wyatt HR, Hill JO, McGuckin BG, Brill C, Mohammed BS, et al. A randomized trial of
a low-carbohydrate diet for obesity. The New England Journal of Medicine. 2003;348(21):2082-
90.
49. Samaha FF, Iqbal N, Seshadri P, Chicano KL, Daily DA, McGrory J, et al. Low-carbohydrate as
compared with a low-fat diet in severe obesity. The New England Journal of Medicine.
2003;348(21):2074-81.
50. Sacks FM, Bray GA, Carey VJ, Smith SR, Ryan DH, Anton SD, et al. Comparison of weight-loss
diets with different compositions of fat, protein, and carbohydrates. The New England Journal
of Medicine. 2009;360(9):859-73.
51. Paoli A, Rubini A, Volek JS, Grimaldi KA. Beyond weight loss: A review of the therapeutic uses of
very-low-carbohydrate (ketogenic) diets. European Journal of Clinical Nutrition. 2013;67(8):789-
96.
52. Ghoch ME, Calugi S, Grave RD. The effects of low-carbohydrate diets on psychosocial outcomes
in obesity/overweight: A systematic review of randomized, controlled studies. Nutrients.
2016;8(Generic):1-13.
146
53. Volek JS, Noakes T, Phinney SD. Rethinking fat as a fuel for endurance exercise. European Journal
of Sport Science. 2015;15(1):13-20.
54. Williams TJ, Cervenka MC. The role for ketogenic diets in epilepsy and status epilepticus in
adults. Clinical Neurophysiology Practice. 2017;2:154-60.
55. Shai I, Schwarzfuchs D, Henkin Y, Shahar DR, Witkow S, Greenberg I, et al. Weight loss with a
low-carbohydrate, Mediterranean, or low-fat diet. The New England Journal of Medicine.
2008;359(3):229-41.
56. Gardner CD, Kiazand A, Alhassan S, Kim S, Stafford RS, Balise RR, et al. Comparison of the Atkins,
Zone, Ornish, and LEARN Diets for Change in weight and related risk factors among overweight
premenopausal women: The A TO Z weight loss study: A randomized trial. The Journal of the
American Medical Association. 2007;297(9):969-77.
57. Brehm BJ, Seeley RJ, Daniels SR, D'Alessio DA. A randomized trial comparing a very low
carbohydrate diet and a calorie-restricted low fat diet on body weight and cardiovascular risk
factors in healthy women. The Journal of Clinical Endocrinology and Metabolism.
2003;88(4):1617-23.
58. Skov AR, Toubro S, Ronn B, Holm L, Astrup A. Randomized trial on protein vs carbohydrate in ad
libitum fat reduced diet for the treatment of obesity. International journal of obesity and related
metabolic disorders. 1999;23(5):528-36.
59. Bueno NB, de Melo ISV, de Oliveira SL, da Rocha Ataide T. Very-low-carbohydrate ketogenic diet
vs low-fat diet for long-term weight loss: A meta-analysis of randomised controlled trials. British
Journal of Nutrition. 2013;110(7):1178-87.
147
60. Krieger JW, Sitren HS, Daniels MJ, Langkamp-Henken B. Effects of variation in protein and
carbohydrate intake on body mass and composition during energy restriction: A meta-
regression. The American Journal of Clinical Nutrition. 2006;83(2):260-74.
61. Snorgaard O, Poulsen GM, Andersen HK, Astrup A. Systematic review and meta-analysis of
dietary carbohydrate restriction in patients with type 2 diabetes. The British Medical Journal.
2017;5(1):1-10.
62. Halton TL, Hu FB. The effects of high protein diets on thermogenesis, satiety and weight loss: A
critical review. Journal of the American College of Nutrition. 2004;23(5):373.
63. Avenell A, Brown TJ, McGee MA, Campbell MK, Grant AM, Broom J, et al. What are the long‐
term benefits of weight reducing diets in adults? A systematic review of randomized controlled
trials. Journal of Human Nutrition and Dietetics. 2004;17(4):317-35.
64. Santos FL, Esteves SS, da Costa Pereira A, Yancy Jr WS, Nunes JPL. Systematic review and meta‐
analysis of clinical trials of the effects of low carbohydrate diets on cardiovascular risk factors.
Obesity Reviews. 2012;13(11):1048-66.
65. Hession M, Rolland C, Kulkarni U, Wise A, Broom J. Systematic review of randomized controlled
trials of low-carbohydrate vs low-fat/low-calorie diets in the management of obesity and its
comorbidities. Obesity Reviews. 2009;10(1):36-50.
66. Turton JL, Raab R, Rooney KB. Low-carbohydrate diets for type 1 diabetes mellitus: A systematic
review. Public Library of Science One. 2018;13(3):e0194987.
148
67. Chen Y-H, Lee Y-C, Tsao Y-C, Lu M-C, Chuang H-H, Yeh W-C, et al. Association between high-
fasting insulin levels and metabolic syndrome in non-diabetic middle-aged and elderly
populations: A community-based study in Taiwan. The British Medical Journal. 2018;8(5):1-7.
68. Schmid H, Kettelhut IC, Migliorini RH. Reduced lipogenesis in rats fed a high-protein
carbohydrate-free diet. Metabolism: clinical and experimental. 1984;33(3):219-23.
69. Mensink RP, Katan MB. Effect of dietary fatty acids on serum lipids and lipoproteins: A meta-
analysis of 27 trials. Arteriosclerosis and Thrombosis: A Journal of Vascular Biology.
1992;12(8):911-9.
70. Imamura F, Micha R, Wu JHY, de Oliveira Otto MC, Otite FO, Abioye AI, et al. Effects of saturated
fat, polyunsaturated fat, monounsaturated fat, and carbohydrate on glucose-Insulin
homeostasis: A systematic review and meta-analysis of randomised controlled feeding trials.
Public Library of Science Medicine. 2016;13(7):e1002087.
71. Khera AV, Cuchel M, de la Llera-Moya M, Rodrigues A, Burke MF, Jafri K, et al. Cholesterol efflux
capacity, high-density lipoprotein function, and atherosclerosis. The New England Journal of
Medicine. 2011;364(2):127-35.
72. Nordestgaard BGP, Varbo AMD. Triglycerides and cardiovascular disease. The Lancet.
2014;384(9943):626-35.
73. Gould ALP, Davies GMD, Alemao ERMS, Yin DDP, Cook JRP. Cholesterol reduction yields clinical
benefits: Meta-analysis including recent trials. Clinical Therapeutics. 2007;29(5):778-94.
149
74. Hernandez TL, Sutherland JP, Wolfe P, Allian-Sauer M, Capell WH, Talley ND, et al. Lack of
suppression of circulating free fatty acids and hypercholesterolemia during weight loss on a
high-fat, low-carbohydrate diet. The American Journal of Clinical Nutrition. 2010;91(3):578-85.
75. Noto H, Goto A, Tsujimoto T, Noda M. Low-carbohydrate diets and all-cause mortality: A
systematic review and meta-analysis of observational studies. Public Library of Science One.
2013;8(1):e55030.
76. Aronson D, Edelman ER. Coronary artery disease and diabetes mellitus. Cardiology Clinics.
2014;32(3):439-55.
77. Stunkard AJ, Rush J. Dieting and depression reexamined. Annals of Internal Medicine.
1974;81(4):526.
78. Singh M. Mood, food, and obesity. Frontiers in psychology. 2014;5:925-.
79. Quehl R, Haines J, Lewis SP, Buchholz AC. Food and mood: Diet quality is inversely associated
with depressive symptoms in female university students. Canadian Journal of Dietetic Practice
and Research. 2017;78(3):124.
80. Gavvala JR, Schuele SU. Epilepsy. The Journal of the American Medical Association.
2016;316(24):2686-.
81. Martin K, Jackson CF, Levy RG, Cooper PN. Ketogenic diet and other dietary treatments for
epilepsy. Cochrane Database of Systematic Reviews. 2016;2:CD001903.
82. Chung H-Y, Park YK. Rationale, feasibility and acceptability of ketogenic diet for cancer
treatment. Journal of Cancer Prevention. 2017;22(3):127-34.
150
83. Noakes T, Volek JS, Phinney SD. Low-carbohydrate diets for athletes: What evidence? British
Journal of Sports Medicine. 2014;48(14):1077-8.
84. McNally MA, Pyzik PL, Rubenstein JE, Hamdy RF, Kossoff EH. Empiric use of potassium citrate
reduces kidney-stone incidence with the ketogenic diet. Pediatrics. 2009;124(2):e300.
85. Reddy ST, Wang C-Y, Sakhaee K, Brinkley L, Pak CYC. Effect of low-carbohydrate high-protein
diets on acid-base balance, stone-forming propensity, and calcium metabolism. American
Journal of Kidney Diseases. 2002;40(2):265-74.
86. Møller G, Sluik D, Ritz C, Mikkilä V, Raitakari OT, Hutri-Kähönen N, et al. A Protein Diet Score,
including plant and animal protein, investigating the association with HbA1c and eGFR - The
PREVIEW Project. Nutrients. 2017;9(7):763.
87. Frigolet M-E, Ramos Barragán V-E, Tamez González M. Low-carbohydrate diets: A matter of love
or hate. Annals of Nutrition and Metabolism. 2011;58(4):320-34.
88. Wycherley TP, Brinkworth GD, Keogh JB, Noakes M, Buckley JD, Clifton PM. Long-term effects
of weight loss with a very low carbohydrate and low fat diet on vascular function in overweight
and obese patients. Journal of Internal Medicine. 2010;267(5):452-61.
89. Seidelmann SB, Claggett B, Cheng S, Henglin M, Shah A, Steffen LM, et al. Dietary carbohydrate
intake and mortality: A prospective cohort study and meta-analysis. The Lancet Public Health.
2018;3(9):e419-e28.
90. Chapple ILC, Bouchard P, Cagetti MG, Campus G, Carra M-C, Cocco F, et al. Interaction of
lifestyle, behaviour or systemic diseases with dental caries and periodontal diseases: Consensus
151
report of group 2 of the joint EFP/ORCA workshop on the boundaries between caries and
periodontal diseases. Journal of Clinical Periodontology. 2017;44(S18):S39-S51.
91. Palacios C, Rivas-Tumanyan S, Morou-Bermúdez E, Colon AM, Torres RY, Elías-Boneta AR.
Association between type, amount, and pattern of carbohydrate consumption with dental
caries in 12-year-olds in Puerto Rico. Caries Research. 2016;50(6):560-70.
92. Fardet A, Boirie Y. Associations between food and beverage groups and major diet‐related
chronic diseases: An exhaustive review of pooled/meta‐analyses and systematic reviews.
Nutrition Reviews. 2014;72(12):741-62.
93. Jenkins DJ, Wolever TM, Taylor RH, Barker H, Fielden H, Baldwin JM, et al. Glycemic index of
foods: A physiological basis for carbohydrate exchange. The American Journal of Clinical
Nutrition. 1981;34(3):362-6.
94. Barclay AW, Petocz P, McMillan-Price J, Flood VM, Prvan T, Mitchell P, et al. Glycemic index,
glycemic load, and chronic disease risk: A meta-analysis of observational studies. The American
Journal of Clinical Nutrition. 2008;87(3):627-37.
95. Björck I, Elmståhl HL. The glycaemic index: importance of dietary fibre and other food
properties. The Proceedings of the Nutrition Society. 2003;62(1):201.
96. Schwingshackl L, Hoffmann G. Long-term effects of low glycemic index/load vs. high glycemic
index/load diets on parameters of obesity and obesity-associated risks: A systematic review and
meta-analysis. Nutrition, Metabolism and Cardiovascular Diseases. 2013;23(8):699-706.
97. Bessesen DH. The role of carbohydrates in insulin resistance. The Journal of Nutrition.
2001;131(10):2782S-6S.
152
98. Barazzoni R, Deutz NEP, Biolo G, Bischoff S, Boirie Y, Cederholm T, et al. Carbohydrates and
insulin resistance in clinical nutrition: Recommendations from the ESPEN expert group. Clinical
Nutrition. 2017;36(2):355-63.
99. David LA, Maurice CF, Carmody RN, Gootenberg DB, Button JE, Wolfe BE, et al. Diet rapidly and
reproducibly alters the human gut microbiome. Nature. 2014;505(7484):559-63.
100. Natividad JM, Lamas B, Pham HP, Michel M-L, Rainteau D, Bridonneau C, et al. Bilophila
wadsworthia aggravates high fat diet induced metabolic dysfunctions in mice. Nature
Communications. 2018;9(1):2802.
101. Kovatcheva-Datchary P, Nilsson A, Akrami R, Lee YS, De Vadder F, Arora T, et al. Dietary fiber-
induced improvement in glucose metabolism is associated with increased abundance of
Prevotella. Cell metabolism. 2015;22(6):971-82.
102. Morrison DJ, Preston T. Formation of short chain fatty acids by the gut microbiota and their
impact on human metabolism. Gut Microbes. 2016;7(3):189-200.
103. Kodama S, Saito K, Tanaka S, Maki M, Yachi Y, Sato M, et al. Influence of fat and carbohydrate
proportions on the metabolic profile in patients with type 2 diabetes: A meta-analysis. Diabetes
Care. 2009;32(5):959-65.
104. Ley SHP, Hamdy OMD, Mohan VMD, Hu FBP. Prevention and management of type 2 diabetes:
Dietary components and nutritional strategies. The Lancet. 2014;383(9933):1999-2007.
105. Ludwig DS, Hu FB, Tappy L, Brand-Miller J. Dietary carbohydrates: Role of quality and quantity
in chronic disease. The British Medical Journal. 2018;361:1-6.
153
106. Laffel L. Ketone bodies: A review of physiology, pathophysiology and application of monitoring
to diabetes. Diabetes/Metabolism Research and Reviews. 1999;15(6):412-26.
107. Meas T, Taboulet P, Sobngwi E, Gautier JF. Is capillary ketone determination useful in clinical
practice? In which circumstances? Diabetes & metabolism. 2005;31(3 Pt 1):299-303.
108. Brooke J, Stiell M, Ojo O. Evaluation of the accuracy of capillary hydroxybutyrate measurement
compared with other measurements in the diagnosis of diabetic ketoacidosis: A systematic
review. International Journal of Environmental Research and Public Health. 2016;13(9):837.
109. Rudolf MC, Sherwin RS. Maternal ketosis and its effects on the fetus. Clinics in Endocrinology
and Metabolism. 1983;12(2):413.
110. Mitchell GA, Kassovska-Bratinova S, Boukaftane Y, Robert MF. Medical aspects of ketone body
metabolism. Clinical and Investigative Medicine. 1995;18(3):193.
111. Manninen AH. Metabolic effects of the very-low-carbohydrate diets: Misunderstood "villains"
of human metabolism. Journal of the International Society of Sports Nutrition. 2004;1(2):7-11.
112. Cotter DG, d'Avignon DA, Wentz AE, Weber ML, Crawford PA. Obligate role for ketone body
oxidation in neonatal metabolic homeostasis. The Journal of Biological Chemistry.
2011;286(9):6902-10.
113. Reichard GA, Jr., Owen OE, Haff AC, Paul P, Bortz WM. Ketone-body production and oxidation
in fasting obese humans. The Journal of Clinical Investigation. 1974;53(2):508-15.
114. Sokoloff L. Metabolism of ketone bodies by the brain. Annual Review of Medicine. 1973;24:271-
80.
154
115. Wolfrum C, Asilmaz E, Luca E, Friedman JM, Stoffel M. Foxa2 regulates lipid metabolism and
ketogenesis in the liver during fasting and in diabetes. Nature. 2004;432(7020):1027-32.
116. Cox Pete J, Kirk T, Ashmore T, Willerton K, Evans R, Smith A, et al. Nutritional ketosis alters fuel
preference and thereby endurance performance in athletes. Cell Metabolism. 2016;24(2):256-
68.
117. Robinson AM, Williamson DH. Physiological roles of ketone bodies as substrates and signals in
mammalian tissues. Physiological Reviews. 1980;60(1):143-87.
118. Owen OE. Ketone bodies as a fuel for the brain during starvation. Biochemistry and Molecular
Biology Education. 2005;33(4):246-51.
119. ePaoli A, eBosco G, eCamporesi E, eMangar D. Ketosis, ketogenic diet and food intake control:
a complex relationship. Frontiers in Psychology. 2015;6.
120. Fukao T, Lopaschuk GD, Mitchell GA. Pathways and control of ketone body metabolism: On the
fringe of lipid biochemistry. Prostaglandins, Leukotrienes and Essential Fatty Acids.
2004;70(3):243-51.
121. Bonadio W. Pediatric diabetic ketoacidosis: an outpatient perspective on evaluation and
management. Pediatric Emergency Medicine Practice. 2013;10(3):1.
122. Persson B, Lunell NO. Metabolic control in diabetic pregnancy. American Journal of Obstetrics
and Gynecology. 1975;122(6):737-45.
123. Maresh M, Gillmer MD, Beard RW, Alderson CS, Bloxham BA, Elkeles RS. The effect of diet and
insulin on metabolic profiles of women with gestational diabetes mellitus. Diabetes. 1985;34
Suppl 2(Supplement_2):88-93.
155
124. Metzger B, Vileisis R, Ravnikar V, Freinkel N. "Accelerated starvation" and the skipped breakfast
in late normal pregnancy. The Lancet. 1982;319(8272):588-92.
125. Patel D, Kalhan S. Glycerol metabolism and triglyceride-fatty acid cycling in the human newborn:
Effect of maternal diabetes and intrauterine growth retardation. Pediatric Research.
1992;31(1):52-8.
126. Shambaugh GE, 3rd. Ketone body metabolism in the mother and fetus. Federation proceedings.
1985;44(7):2347-51.
127. Davis RAH, Deemer SE, Bergeron JM, Little JT, Warren JL, Fisher G, et al. Dietary R, S-1,3-
butanediol diacetoacetate reduces body weight and adiposity in obese mice fed a high-fat diet.
FASEB Journal. 2018:fj201800821RR.
128. Coad S, Friedman B, Geoffrion R. Understanding urinalysis: Clues for the obstetrician-
gynecologist. Expert Review of Obstetrics and Gynecology. 2012;7:269+.
129. Byrne HA, Tieszen KL, Hollis S, Dornan TL, New JP. Evaluation of an electrochemical sensor for
measuring blood ketones. Diabetes Care. 2000;23(4):500-3.
130. Taboulet P, Deconinck N, Thurel A, Haas L, Manamani J, Porcher R, et al. Correlation between
urine ketones (acetoacetate) and capillary blood ketones (3-beta-hydroxybutyrate)
in hyperglycaemic patients. Diabetes and Metabolism. 2007;33(2):135-9.
131. Fraser J, Fetter MC, Mast RL, Free AH. Studies with a simplified nitroprusside test for ketone
bodies in urine, serum, plasma, and milk. Clinica Chimica Acta. 1965;11(4):372-8.
132. Comstock JP GA. Ketonuria. 1990. In: Clinical Methods: The History, Physical, and Laboratory
Examinations [Internet]. Boston: Butterworths.
156
133. Laffel L. Improving outcomes with POCT for HbA1c and blood ketone testing. Journal of Diabetes
Science and Technology. 2007;1(1):133-6.
134. Qiao Y, Gao Z, Liu Y, Cheng Y, Yu M, Zhao L, et al. Breath ketone testing: A new biomarker for
diagnosis and therapeutic monitoring of diabetic ketosis. BioMed Research International.
2014;2014:869186-5.
135. Musa-Veloso K, Likhodii SS, Cunnane SC. Breath acetone is a reliable indicator of ketosis in adults
consuming ketogenic meals. The American Journal of Clinical Nutrition. 2002;76(1):65-70.
136. Ruzsanyi V, Peter Kalapos M. Breath acetone as a potential marker in clinical practice. Journal
of breath research. 2017;11(2):024002.
137. Blaikie TPJ, Couper J, Hancock G, Hurst PL, Peverall R, Richmond G, et al. Portable Device for
Measuring Breath Acetone Based on Sample Preconcentration and Cavity Enhanced
Spectroscopy. Analytical Chemistry. 2016;88(22):11016-21.
138. Reece EA, Homko C, Wiznitzer A. Metabolic changes in diabetic and nondiabetic subjects during
pregnancy. Obstetrical and Gynecological Survey. 1994;49(1):64-71.
139. Riskin-Mashiah S, Damti A, Younes G, Auslander R. Normal fasting plasma glucose levels during
pregnancy: A hospital-based study. Journal of Perinatal Medicine. 2011;39(2):209-11.
140. Mills JL, Jovanovic L, Knopp R, Aarons J, Conley M, Park E, et al. Physiological reduction in fasting
plasma glucose concentration in the first trimester of normal pregnancy: The diabetes in early
pregnancy study. Metabolism - Clinical and Experimental. 1998;47(9):1140-4.
141. Bronisz A, Ozorowski M, Hagner-Derengowska M. Pregnancy ketonemia and development of
the fetal central nervous system. International Journal of Endocrinology. 2018;2018:1242901.
157
142. Sonagra AD, Biradar SM, K D, Murthy D S J. Normal pregnancy- a state of insulin resistance.
Journal of Clinical and Diagnostic Research. 2014;8(11):CC01-CC3.
143. Barbour LA, McCurdy CE, Hernandez TL, Kirwan JP, Catalano PM, Friedman JE. Cellular
mechanisms for insulin resistance in normal pregnancy and gestational diabetes. Diabetes Care.
2007;30 (Supplement 2):S112.
144. Seabra G, Saunders C, de Carvalho Padilha P, Zajdenverg L, da Silva LBG, de Souza Santos MMA.
Association between maternal glucose levels during pregnancy and gestational diabetes
mellitus: An analytical cross-sectional study. Diabetology and Metabolic Syndrome.
2015;7(1):17.
145. Dabelea D. The predisposition to obesity and diabetes in offspring of diabetic mothers. Diabetes
Care. 2007;30 Suppl 2(Supplement 2):S169-S74.
146. Catalano PM, Tyzbir ED, Wolfe RR, Roman NM, B. Amini S, Sims EAH. Longitudinal changes in
basal hepatic glucose production and suppression during insulin infusion in normal pregnant
women. American Journal of Obstetrics and Gynecology. 1992;167(4, Part 1):913-9.
147. Brett KE, Ferraro ZM, Yockell-Lelievre J, Gruslin A, Adamo KB. Maternal–fetal nutrient transport
in pregnancy pathologies: The role of the placenta. International Journal of Molecular Sciences.
2014;15(9):16153-85.
148. Negrato CA, Mattar R, Gomes MB. Adverse pregnancy outcomes in women with diabetes.
Diabetology and Metabolic Syndrome. 2012;4:41.
158
149. Karpate SJ, Morsi H, Shehmar M, Dale J, Patel C. Euglycemic ketoacidosis in pregnancy and its
management: Case report and review of literature. European Journal of Obstetrics and
Gynecology and Reproductive Biology. 2013;171(2):386-7.
150. Butte NF. Carbohydrate and lipid metabolism in pregnancy: Normal compared with gestational
diabetes mellitus. The American Journal of Clinical Nutrition. 2000;71(5 Suppl):1256S-61S.
151. Nelson SM, Matthews P, Poston L. Maternal metabolism and obesity: Modifiable determinants
of pregnancy outcome. Human Reproduction Update. 2010;16(3):255-75.
152. Kalkhoff RK, Kim HJ. The influence of hormonal changes of pregnancy on maternal metabolism.
Ciba Foundation Symposium. 1978(63):29.
153. Desoye G, Nolan CJ. The fetal glucose steal: An underappreciated phenomenon in diabetic
pregnancy. Diabetologia. 2016;59(6):1089-94.
154. Huda SS, Brodie LE, Sattar N. Obesity in pregnancy: Prevalence and metabolic consequences.
Seminars in Fetal and Neonatal Medicine. 2009;15(2):70-6.
155. Feldt–Rasmussen UMDD, Mathiesen ERMDD. Endocrine disorders in pregnancy: Physiological
and hormonal aspects of pregnancy. Best Practice and Research: Clinical Endocrinology and
Metabolism. 2011;25(6):875-84.
156. Cozar-Castellano I, Bartha JL, Perdomo G, Visiedo F, Bugatto F, Sanchez V. High glucose levels
reduce fatty acid oxidation and increase triglyceride accumulation in human placenta. American
Journal of Physiology. 2013;305(1):E205.
157. Utz B, Kolsteren P, De Brouwere V. Screening for gestational diabetes mellitus: Are guidelines
from high-income settings applicable to poorer countries? Clinical Diabetes. 2015;33(3):152-8.
159
158. The HAPO Study Cooperative Research Group. Hyperglycemia and adverse pregnancy
outcomes. The New England Journal of Medicine. 2008;358(19):1991-2002.
159. International Association of Diabetes and Pregnancy Study Groups Consensus Panel.
International Association of Diabetes and Pregnancy Study Groups recommendations on the
diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care. 2010;33(3):676.
160. Moses RG, Morris GJ, Petocz P, San Gil F, Garg D. The impact of potential new diagnostic criteria
on the prevalence of gestational diabetes mellitus in Australia. The Medical Journal of Australia.
2011;194(7):338-40.
161. Bimson BE, Rosenn BM, Morris SA, Sasso EB, Schwartz RA, Brustman LE. Current trends in the
diagnosis and management of gestational diabetes mellitus in the United States. The Journal of
Maternal-fetal & Neonatal Medicine. 2017;30(21):2607-12.
162. Pedersen J. Weight and length at birth of infants of diabetic mothers. Acta endocrinologica.
1954;16(4):330-42.
163. Gluckman PD, Hanson MA, Cooper C, Thornburg KL. Effect of in utero and early-life conditions
on adult health and disease. The New England Journal of Medicine. 2008;359(1):61-73.
164. Calkins K, Devaskar SU. Fetal origins of adult disease. Current Problems in Pediatric and
Adolescent Health Care. 2011;41(6):158-76.
165. Ma RC, Tutino GE, Lillycrop KA, Hanson MA, Tam WH. Maternal diabetes, gestational diabetes
and the role of epigenetics in their long term effects on offspring. Progress in Biophysics and
Molecular Biology. 2015;118(1-2):55-68.
160
166. Hales CN, Barker DJ. Type 2 (non-insulin-dependent) diabetes mellitus: The thrifty phenotype
hypothesis. Diabetologia. 1992;35(7):595-601.
167. Kim C, Newton KM, Knopp RH. Gestational diabetes and the incidence of type 2 diabetes: A
systematic review. Diabetes Care. 2002;25(10):1862-8.
168. Song C, Lyu Y, Li C, Liu P, Li J, Ma RC, et al. Long-term risk of diabetes in women at varying
durations after gestational diabetes: A systematic review and meta-analysis with more than 2
million women. Obesity Reviews. 2018;19(3):421-9.
169. Burguet A. Long-term outcome in children of mothers with gestational diabetes. Diabetes and
Metabolism. 2010;36(6, Part 2):682-94.
170. Tam WH, Ma RCW, Ozaki R, Li AM, Chan MHM, Yuen LY, et al. In utero exposure to maternal
hyperglycemia increases childhood cardiometabolic risk in offspring. Diabetes Care.
2017;40(5):679-86.
171. Hammoud NM, Visser GHA, van Rossem L, Biesma DH, Wit JM, de Valk HW. Long-term BMI and
growth profiles in offspring of women with gestational diabetes. Diabetologia. 2018;61(5):1037-
45.
172. Ray JG, Vermeulen MJ, Shapiro JL, Kenshole AB. Maternal and neonatal outcomes in
pregestational and gestational diabetes mellitus, and the influence of maternal obesity and
weight gain: The DEPOSIT study. Quarterly Journal of Medicine: An International Journal of
Medicine. 2001;94(7):347-56.
173. Longmore DK, Barr ELM, Lee IL, Barzi F, Kirkwood M, Whitbread C, et al. Maternal body mass
index, excess gestational weight gain, and diabetes are positively associated with neonatal
161
adiposity in the Pregnancy and Neonatal Diabetes Outcomes in Remote Australia (PANDORA)
study. Pediatric obesity. 2019;14(4):e12490.
174. McLean M, Chipps D, Cheung NW. Mother to child transmission of diabetes mellitus: Does
gestational diabetes program Type 2 diabetes in the next generation? Diabetic Medicine.
2006;23(11):1213-5.
175. Strutz J, Cvitic S, Hackl H, Kashofer K, Appel HM, Thüringer A, et al. Gestational diabetes alters
microRNA signatures in human feto-placental endothelial cells depending on fetal sex. Clinical
Science. 2018:1-24.
176. Ovesen PG, Fuglsang J, Andersen MB, Wolff C, Petersen OB, #x00F8, et al. Temporal trends in
gestational diabetes prevalence, treatment, and outcomes at Aarhus University hospital, Skejby,
between 2004 and 2016. Journal of Diabetes Research. 2018;2018:6.
177. Carolan M, Davey M-A, Biro MA, Kealy M. Maternal age, ethnicity and gestational diabetes
mellitus. Midwifery. 2012;28(6):778-83.
178. Wong VW, Lin A, Russell H. Adopting the new World Health Organization diagnostic criteria for
gestational diabetes: How the prevalence changes in a high-risk region in Australia. Diabetes
Research and Clinical Practice. 2017;129:148-53.
179. Crowther CA, Hiller JE, Moss JR, McPhee AJ, Jeffries WS, Robinson JS. Effect of treatment of
gestational diabetes mellitus on pregnancy outcomes. The New England Journal of Medicine.
2005;352(24):2477-86.
180. Yamamoto JM, Kellett JE, Balsells M, Garcia-Patterson A, Hadar E, Sola I, et al. Gestational
Diabetes Mellitus and Diet: A Systematic Review and Meta-analysis of Randomized Controlled
162
Trials Examining the Impact of Modified Dietary Interventions on Maternal Glucose Control and
Neonatal Birth Weight. Diabetes Care. 2018;41(7):1346-61.
181. Nankervis A, Price S, Conn J. Gestational diabetes mellitus. Australian Journal for General
Practitioners. 2018;47:445-9.
182. Mijatovic-Vukas J, Capling L, Cheng S, Stamatakis E, Louie J, Cheung NW, et al. Associations of
diet and physical activity with risk for gestational diabetes mellitus: A systematic review and
meta-analysis. Nutrients. 2018;10(6):698.
183. Duarte-Gardea MO, Gonzales-Pacheco DM, Reader DM, Thomas AM, Wang SR, Gregory RP, et
al. Academy of Nutrition and Dietetics Gestational Diabetes Evidence-Based Nutrition Practice
Guideline. Journal of the Academy of Nutrition and Dietetics. 2018;118(9):1719-42.
184. Han S, Middleton P, Shepherd E, Van Ryswyk E, Crowther CA. Different types of dietary advice
for women with gestational diabetes mellitus. Cochrane Database of Systematic Reviews.
2017(2).
185. Metzger BE, Buchanan TA, Coustan DR, de Leiva A, Dunger DB, Hadden DR, et al. Summary and
recommendations of the Fifth International Workshop-Conference on gestational diabetes
mellitus. Diabetes Care. 2007;30(Supplement 2):S251.
186. Ribaroff GA, Wastnedge E, Drake AJ, Sharpe RM, Chambers TJG. Animal models of maternal high
fat diet exposure and effects on metabolism in offspring: a meta-regression analysis. Obesity
Reviews. 2017;18(6):673-86.
163
187. Schaefer-Graf UM, Graf K, Kulbacka I, Kjos SL, Dudenhausen J, Vetter K, et al. Maternal lipids as
strong determinants of fetal environment and growth in pregnancies with gestational diabetes
mellitus. Diabetes Care. 2008;31(9):1858-63.
188. Blumer I, Hadar E, Hadden DR, Jovanovič L, Mestman JH, Murad MH, et al. Diabetes and
pregnancy: An endocrine society clinical practice guideline. The Journal of Clinical Endocrinology
and Metabolism. 2013;98(11):4227-49.
189. Practice Bulletin No. 180: Gestational Diabetes Mellitus. Obstetrics and Gynecology.
2017;130(1):e17-e37.
190. Cypryk K, Kamińska P, Kosiński M, Pertyńska-Marczewska M, Lewiński A. A comparison of the
effectiveness, tolerability and safety of high and low carbohydrate diets in women with
gestational diabetes. Endokrynologia Polska. 2007;58(4):314.
191. Major CA, Henry MJ, de Veciana M, Morgan MA. The effects of carbohydrate restriction in
patients with diet-controlled gestational diabetes. Obstetrics and Gynecology. 1998;91(4):600-
4.
192. Brown J, Alwan NA, West J, Brown S, McKinlay CJD, Farrar D, et al. Lifestyle interventions for the
treatment of women with gestational diabetes. Cochrane Database of Systematic Reviews.
2017(5):1-132.
193. Viana LV, Gross JL, Azevedo MJ. Dietary intervention in patients with gestational diabetes
mellitus: A systematic review and meta-analysis of randomized clinical trials on maternal and
newborn outcomes. Diabetes Care. 2014;37(12):3345-55.
164
194. Churchill JA, Berendes HW, Nemore J. Neuropsychological deficits in children of diabetic
mothers. A report from the Collaborative Study of Cerebral Palsy. American Journal of Obstetrics
and Gynecology. 1969;105(2):257.
195. Naeye RL, Chez RA. Effects of maternal acetonuria and low pregnancy weight gain on children's
psychomotor development. American Journal of Obstetrics and Gynecology. 1981;139(2):189.
196. Rizzo T, Metzger BE, Burns K, Burns WJ. Correlations between antepartum maternal metabolism
and intelligence of offspring. The New England Journal of Medicine. 1991;325(13):911-6.
197. Camprubi Robles M, Campoy C, Garcia Fernandez L, Lopez-Pedrosa JM, Rueda R, Martin MJ.
Maternal diabetes and cognitive performance in the offspring: A systematic review and meta-
analysis. Public Library of Science One. 2015;10(11):e0142583.
198. Sussman D, van Eede M, Wong MD, Adamson SL, Henkelman M. Effects of a ketogenic diet
during pregnancy on embryonic growth in the mouse. BMC Pregnancy and Childbirth.
2013;13:109.
199. Grasemann C, Herrmann R, Starschinova J, Gertsen M, Palmert MR, Grasemann H. Effects of
fetal exposure to high-fat diet or maternal hyperglycemia on L-arginine and nitric oxide
metabolism in lung. Nutrition and Diabetes. 2017;7:e244.
200. James RG, III, Alberti KGMM, Davidson MB, DeFronzo RA, The Expert Committee on the D,
Classification of Diabetes M. Report of the Expert Committee on the Diagnosis and Classification
of Diabetes Mellitus. Diabetes Care. 1997;20(7):1183-97.
201. American Diabetes Association. Standards of medical care in diabetes - 2014. Diabetes Care.
2013;37(Supplement 1):S14.
165
202. Guariguata L, Linnenkamp U, Beagley J, Whiting DR, Cho NH. Global estimates of the prevalence
of hyperglycaemia in pregnancy. Diabetes Research and Clinical Practice. 2014;103(2):176-85.
203. Pu J, Zhao B, Wang EJ, Nimbal V, Osmundson S, Kunz L, et al. Racial/ethnic differences in
gestational diabetes prevalence and contribution of common risk factors. Paediatric and
Perinatal Epidemiology. 2015;29(5):436-43.
204. Wilmot EG, Mansell P. Diabetes and pregnancy. Clinical Medicine. 2014;14(6):677-80.
205. Lehnen H, Zechner U, Haaf T. Epigenetics of gestational diabetes mellitus and offspring health:
The time for action is in early stages of life. Molecular Human Reproduction. 2013;19(7):415-22.
206. Moreno-Castilla C, Mauricio D, Hernandez M. Role of Medical Nutrition Therapy in the
management of gestational diabetes mellitus. Current Diabetes Reports. 2016;16(4):1-9.
207. Koivusalo SB, Rönö K, Klemetti MM, Roine RP, Lindström J, Erkkola M, et al. Gestational diabetes
mellitus can be prevented by lifestyle intervention: The Finnish gestational diabetes prevention
study (RADIEL): A randomized controlled trial. Diabetes Care. 2016;39(1):24-30.
208. Wang X, Ouyang Y, Liu J, Zhu M, Zhao G, Bao W, et al. Fruit and vegetable consumption and
mortality from all causes, cardiovascular disease, and cancer: Systematic review and dose-
response meta-analysis of prospective cohort studies. The British Medical Journal.
2014;349:g4490.
209. Colberg SR, Sigal RJ, Yardley JE, Riddell MC, Dunstan DW, Dempsey PC, et al. Physical
activity/exercise and diabetes: A position statement of the American Diabetes Association.
Diabetes Care. 2016;39(11):2065-79.
166
210. Stanford KI, Goodyear LJ. Exercise and type 2 diabetes: Molecular mechanisms regulating
glucose uptake in skeletal muscle. Advances in Physiology Education. 2014;38(4):308-14.
211. U.S. Department of Health and Human Services. 2008 Physical Activity Guidelines for Americans.
Washington, DC: U.S. Department of Health and Human Services; 2008.
212. 2018 Physical Activity Guidelines Advisory Committee. 2018 Physical Activity Guidelines
Advisory Committee Scientific Report. Washington, DC; 2018.
213. Löf M, Linköpings u, Institutionen för klinisk och experimentell m, Näringslära,
Hälsouniversitetet. Physical activity pattern and activity energy expenditure in healthy pregnant
and non-pregnant Swedish women. European Journal of Clinical Nutrition. 2011;65(12):1295-
301.
214. Evenson KR, Wen F. Prevalence and correlates of objectively measured physical activity and
sedentary behavior among US pregnant women. Preventive Medicine. 2011;53(1):39-43.
215. Viechtbauer W. Conducting meta-analyses in R with the metafor package. Journal of Statistical
Software. 2010;36(3):1-48.
216. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Multiple outcomes or time‐points within a
study. Chichester, UK: John Wiley & Sons, Ltd; 2009. p. 225-38.
217. Deeks JJ, Altman DG. Effect measures for meta-analysis of trials with binary outcomes. 2008.
218. Nakagawa S, Noble DWA, Senior AM, Lagisz M. Meta-evaluation of meta-analysis: Ten appraisal
questions for biologists. BMC Biology. 2017;15(1):1-14.
167
219. Bao W, Bowers K, Tobias DK, Hu FB, Zhang C. Prepregnancy dietary protein intake, major dietary
protein sources, and the risk of gestational diabetes mellitus: A prospective cohort study.
Diabetes Care. 2013;36(7):2001-8.
220. Bao W, Bowers K, Tobias DK, Olsen SF, Chavarro J, Vaag A, et al. Prepregnancy low-carbohydrate
dietary pattern and risk of gestational diabetes mellitus: A prospective cohort study. The
American Journal of Clinical Nutrition. 2014;99(6):1378-84.
221. Bao W, Tobias DK, Olsen SF, Zhang C. Pre-pregnancy fried food consumption and the risk of
gestational diabetes mellitus: A prospective cohort study. Diabetologia. 2014;57(12):2485-91.
222. Bao W, Tobias DK, Hu FB, Chavarro JE, Zhang C. Pre-pregnancy potato consumption and risk of
gestational diabetes mellitus: prospective cohort study. The British Medical Journal.
2016;352:h6898.
223. Bowers K, Yeung E, Williams MA, Qi L, Tobias DK, Hu FB, et al. A prospective study of
prepregnancy dietary iron intake and risk for gestational diabetes mellitus. Diabetes Care.
2011;34(7):1557-63.
224. Bowers K, Tobias DK, Yeung E, Hu FB, Zhang C. A prospective study of prepregnancy dietary fat
intake and risk of gestational diabetes. The American Journal of Clinical Nutrition.
2012;95(2):446-53.
225. Chen L, Hu FB, Yeung E, Willett W, Zhang C. Prospective study of pre-gravid sugar-sweetened
beverage consumption and the risk of gestational diabetes mellitus. Diabetes Care.
2009;32(12):2236-41.
168
226. Chen L, Hu FB, Yeung E, Tobias DK, Willett WC, Zhang C. Prepregnancy consumption of fruits and
fruit juices and the risk of gestational diabetes mellitus: A prospective cohort study. Diabetes
Care. 2012;35(5):1079-82.
227. Solomon CG, Willett WC, Carey VJ, Rich-Edwards J, Hunter DJ, Colditz GA, et al. A prospective
study of pregravid determinants of gestational diabetes mellitus. The Journal of the American
Medical Association. 1997;278(13):1078-83.
228. Tobias DK, Zhang C, Chavarro J, Bowers K, Rich-Edwards J, Rosner B, et al. Prepregnancy
adherence to dietary patterns and lower risk of gestational diabetes mellitus. The American
Journal of Clinical Nutrition. 2012;96(2):289-95.
229. Zhang C, Liu S, Solomon CG, Hu FB. Dietary fiber intake, dietary glycemic load, and the risk for
gestational diabetes mellitus. Diabetes Care. 2006;29(10):2223-30.
230. Zhang C, Schulze MB, Solomon CG, Hu FB. A prospective study of dietary patterns, meat intake
and the risk of gestational diabetes mellitus. Diabetologia. 2006;49(11):2604-13.
231. Zhang C, Solomon CG, Manson JE, Hu FB. A prospective study of pregravid physical activity and
sedentary behaviors in relation to the risk for gestational diabetes mellitus. Archives of internal
medicine. 2006;166(5):543-8.
232. Zhang C, Tobias DK, Chavarro JE, Bao W, Wang D, Ley SH, et al. Adherence to healthy lifestyle
and risk of gestational diabetes mellitus: Prospective cohort study. The British Medical Journal.
2014;349:1-11.
233. Adeney KL, Williams MA, Schiff MA, Qiu C, Sorensen TK. Coffee consumption and the risk of
gestational diabetes mellitus. Acta Obstetricia et Gynecologica Scandinavica. 2007;86(2):161-6.
169
234. Badon SE, Wartko PD, Qiu C, Sorensen TK, Williams MA, Enquobahrie DA. Leisure time physical
activity and gestational diabetes mellitus in the Omega Study. Medicine and Science in Sports
and Exercise. 2016;48(6):1044-52.
235. Dempsey JC, Sorensen TK, Williams MA, Lee I, Miller RS, Dashow EE, et al. Prospective study of
gestational diabetes mellitus risk in relation to maternal recreational physical activity before
and during pregnancy. American Journal of Epidemiology. 2004;159(7):663-70.
236. Osorio-Yanez C, Qiu CF, Gelaye B, Enquobahrie DA, Williams MA. Risk of gestational diabetes
mellitus in relation to maternal dietary calcium intake. Public Health Nutrition. 2017;20(6):1082-
9.
237. Qiu C, Frederick IO, Zhang C, Sorensen TK, Enquobahrie DA, Williams MA. Risk of gestational
diabetes mellitus in relation to maternal egg and cholesterol intake. American Journal of
Epidemiology. 2011;173(6):649-58.
238. Qiu C, Zhang C, Gelaye B, Enquobahrie DA, Frederick IO, Williams MA. Gestational diabetes
mellitus in relation to maternal dietary heme iron and nonheme iron intake. Diabetes Care.
2011;34(7):1564-9.
239. Rudra CB, Williams MA, Lee IM, Miller RS, Sorensen TK. Perceived exertion in physical activity
and risk of gestational diabetes mellitus. Epidemiology. 2006;17(1):31-7.
240. Gresham E, Collins CE, Mishra GD, Byles JE, Hure AJ. Diet quality before or during pregnancy and
the relationship with pregnancy and birth outcomes: The Australian Longitudinal Study on
Women's Health. Public Health Nutrition 2016;19(16):2975-83.
170
241. Schoenaker DAJM, Soedamah-Muthu SS, Mishra GD. Quantifying the mediating effect of body
mass index on the relation between a Mediterranean diet and development of maternal
pregnancy complications: The Australian Longitudinal Study on Women's Health. The American
Journal of Clinical Nutrition. 2016;104(3):638-45.
242. Schoenaker DA, Soedamah-Muthu SS, Callaway LK, Mishra GD. Pre-pregnancy dietary patterns
and risk of gestational diabetes mellitus: Results from an Australian population-based
prospective cohort study. Diabetologia. 2015;58(12):2726-35.
243. van der Ploeg HP, van Poppel MN, Chey T, Bauman AE, Brown WJ. The role of pre-pregnancy
physical activity and sedentary behaviour in the development of gestational diabetes mellitus.
Journal of Science and Medicine in Sport. 2011;14(2):149-52.
244. Oken E, Ning Y, Rifas-Shiman SL, Radesky JS, Rich-Edwards JW, Gillman MW. Associations of
physical activity and inactivity before and during pregnancy with glucose tolerance. Obstetrics
and Gynecology. 2006;108(5):1200-7.
245. Radesky JS, Oken E, Rifas-Shiman SL, Kleinman KP, Rich-Edwards JW, Gillman MW. Diet during
early pregnancy and development of gestational diabetes. Paediatric and Perinatal
Epidemiology. 2008;22(1):47-59.
246. Baptiste-Roberts K, Ghosh P, Nicholson WK. Pregravid physical activity, dietary intake, and
glucose intolerance during pregnancy. Journal of Women's Health. 2011;20(12):1847-51.
247. Dye TD, Knox KL, Artal R, Aubry RH, Wojtowycz MA. Physical activity, obesity, and diabetes in
pregnancy. American Journal of Epidemiology. 1997;146(11):961-5.
171
248. Putnam KF, Mueller LA, Magann EF, Thagard A, Johnson AM, Ounpraseuth ST, et al. Evaluating
effects of self-reported domestic physical activity on pregnancy and neonatal outcomes in "stay
at home" military wives. Military Medicine. 2013;178(8):893-8.
249. Harrison C, Lombard C, Teede H. Understanding health behaviours in a cohort of pregnant
women at risk of gestational diabetes mellitus: an observational study. BJOG: An International
Journal of Obstetrics & Gynaecology. 2012;119(6):731-8.
250. Chasan-Taber L, Silveira M, Lynch KE, Pekow P, Braun B, Manson JE, et al. Physical activity before
and during pregnancy and risk of abnormal glucose tolerance among Hispanic women. Diabetes
and Metabolism. 2014;40(1):67-75.
251. Chasan-Taber L, Schmidt MD, Pekow P, Sternfeld B, Manson JE, Solomon CG, et al. Physical
activity and gestational diabetes mellitus among Hispanic women. Journal of Women's Health.
2008;17(6):999-1008.
252. Behboudi-Gandevani S, Safary K, Moghaddam-Banaem L, Lamyian M, Goshtasbi A, Alian-
Moghaddam N. The relationship between maternal serum iron and zinc levels and their
nutritional intakes in early pregnancy with gestational diabetes. Biological Trace Element
Research. 2013;154(1):7-13.
253. Hinkle SN, Laughon SK, Catov JM, Olsen J, Bech BH. First trimester coffee and tea intake and risk
of gestational diabetes mellitus: A study within a national birth cohort. BJOG: An International
Journal of Obstetrics and Gynaecology. 2015;122(3):420-8.
254. Currie L, Woolcott C, Fell D, Armson B, Dodds L. The association between Physical Activity and
Maternal and Neonatal Outcomes: A Prospective Cohort. Maternal and Child Health Journal.
2014;18(8):1823-30.
172
255. Iqbal R, Rafique G, Badruddin S, Qureshi R, Cue R, Gray-Donald K. Increased body fat percentage
and physical inactivity are independent predictors of gestational diabetes mellitus in South
Asian women. European Journal of Clinical Nutrition. 2007;61(6):736-42.
256. Morkrid K, Jenum AK, Berntsen S, Sletner L, Richardsen KR, Vangen S, et al. Objectively recorded
physical activity and the association with gestational diabetes. Scandinavian Journal of Medicine
& Science in Sports. 2014;24(5):e389-97.
257. Dominguez LJ, Martinez-Gonzalez MA, Basterra-Gortari FJ, Gea A, Barbagallo M, Bes-Rastrollo
M. Fast food consumption and gestational diabetes incidence in the SUN project. Public Library
of Science One. 2014;9(9):e106627.
258. Karamanos B, Thanopoulou A, Anastasiou E, Assaad-Khalil S, Albache N, Bachaoui M, et al.
Relation of the Mediterranean diet with the incidence of gestational diabetes. European Journal
of Clinical Nutrition. 2014;68(1):8-13.
259. Qiu C, Frederick IO, Zhang C, Sorensen TK, Enquobahrie DA, Williams MA. Original contribution:
Risk of gestational diabetes mellitus in relation to maternal egg and cholesterol intake. American
Journal of Epidemiology. 2011;173(6):649-58.
260. Wei B, Shanshan L, Chavarro JE, Tobias DK, Yeyi Z, Hu FB, et al. Low carbohydrate-diet scores
and long-term risk of type 2 diabetes among women with a history of gestational diabetes
mellitus: A prospective cohort study. Diabetes Care. 2016;39(1):43-9.
261. Radd-Vagenas S, Kouris-Blazos A, Singh MF, Flood VM. Evolution of Mediterranean diets and
cuisine: Concepts and definitions. Asia Pacific Journal of Clinical Nutrition. 2017;26(5):749-63.
173
262. Schoenaker DAJM, Mishra GD, Callaway LK, Soedamah-Muthu SS. The role of energy, nutrients,
foods, and dietary patterns in the development of gestational diabetes mellitus: A systematic
review of observational studies. Diabetes Care. 2016;39(1):16-23.
263. Donazar-Ezcurra M, Lopez-del Burgo C, Bes-Rastrollo M. Primary prevention of gestational
diabetes mellitus through nutritional factors: A systematic review. BMC Pregnancy and
Childbirth. 2017;17(1).
264. Rhee JJ, Cho E, Willett WC. Energy adjustment of nutrient intakes is preferable to adjustment
using body weight and physical activity in epidemiological analyses. Public Health Nutrition
2014;17(5):1054-60.
265. Welsh JA, Lundeen EA, Stein AD. The sugar-sweetened beverage wars: Public health and the
role of the beverage industry. Current Opinion in Endocrinology, Diabetes and Obesity.
2013;20(5):401-6.
266. Brand-Miller JC, Barclay AW. Declining consumption of added sugars and sugar-sweetened
beverages in Australia: a challenge for obesity prevention. The American Journal of Clinical
Nutrition. 2017;105(4):854-63.
267. Colchero MA, Rivera-Dommarco J, Popkin BM, Ng SW. In Mexico, evidence of sustained
consumer response two years after implementing a sugar-sweetened beverage tax. Health
Affairs. 2017;36(3):564.
268. ABS. Australian Health Survey: Nutrition First Results - Foods and Nutrients, 2011-12 2015
[27th April 2018]. Available from:
http://www.abs.gov.au/ausstats/[email protected]/lookup/4364.0.55.007main+features7102011-12.
174
269. Collison KS, Zaidi MZ, Subhani SN, Al-Rubeaan K, Shoukri M, Al-Mohanna FA. Sugar-sweetened
carbonated beverage consumption correlates with BMI, waist circumference, and poor dietary
choices in school children. BMC Public Health. 2010;10(1):234-.
270. Esposito K, Maiorino MI, Bellastella G, Chiodini P, Panagiotakos D, Giugliano D. A journey into a
Mediterranean diet and type 2 diabetes: A systematic review with meta-analyses. The British
Medical Journal. 2015;5(8):e008222.
271. Micha R, Wallace SK, Mozaffarian D. Red and processed meat consumption and risk of incident
coronary heart disease, stroke, and diabetes mellitus: A systematic review and meta-analysis.
Circulation Research. 2010;121(21):2271-83.
272. Feskens EJM, Sluik D, Woudenbergh vGJ. Meat consumption, diabetes and its complications.
Current Diabetes Reports. 2013;13(2):298-306.
273. Kahn BB, Flier JS. Obesity and insulin resistance. The Journal of Clinical Investigation.
2000;106(4):473-81.
274. Jansson N, Nilsfelt A, Gellerstedt M, Wennergren M, Rossander-Hulthén L, Powell TL, et al.
Maternal hormones linking maternal body mass index and dietary intake to birth weight. The
American Journal of Clinical Nutrition. 2008;87(6):1743-9.
275. Arem H, Moore SC, Patel A, Hartge P, Berrington De Gonzalez A, Visvanathan K, et al. Leisure
time physical activity and mortality: A detailed pooled analysis of the dose-response
relationship. The Journal of the American Medical Association Internal Medicine.
2015;175(6):959-67.
276. Australia SM. Position Statement: Exercise in Pregnancy and the Postpartum Period. 2016. p. 9.
175
277. Evenson KR, Barakat R, Brown WJ, Dargent-Molina P, Haruna M, Mikkelsen EM, et al. Guidelines
for physical activity during pregnancy: Comparisons from around the world. Los Angeles, CA:
SAGE Publications; 2014. p. 102-21.
278. Bennie JA, Pedisic Z, Van Uffelen JGZ, Gale J, Banting LK, Vergeer I, et al. The descriptive
epidemiology of total physical activity, muscle-strengthening exercises and sedentary behaviour
among Australian adults: Results from the national nutrition and physical activity survey. BMC
Public Health. 2016;16(1):73.
279. O'Donovan G, Lee IM, Hamer M, Stamatakis E. Association of "Weekend Warrior" and Other
Leisure Time Physical Activity Patterns With Risks for All-Cause, Cardiovascular Disease, and
Cancer Mortality. JAMA Internal Medicine. 2017;177(3):335-42.
280. Chastin SFM, De Craemer M, De Cocker K, Powell L, Van Cauwenberg J, Dall P, et al. How does
light-intensity physical activity associate with adult cardiometabolic health and mortality?
Systematic review with meta-analysis of experimental and observational studies. British Journal
of Sports Medicine. 2018:1-8.
281. Cordero Y, Mottola MF, Vargas J, Blanco M, Barakat R. Exercise Is Associated with a Reduction
in Gestational Diabetes Mellitus. Medicine and Science in Sports and Exercise. 2015;47(7):1328-
33.
282. Bennett WL, Liu SH, Yeh HC, Nicholson WK, Gunderson EP, Lewis CE, et al. Changes in weight
and health behaviors after pregnancies complicated by gestational diabetes mellitus: the
CARDIA study. Obesity (Silver Spring). 2013;21(6):1269-75.
283. Deierlein AL, Siega-Riz AM, Evenson KR. Physical activity during pregnancy and risk of
hyperglycemia. Journal of Women's Health. 2012;21(7):769-75.
176
284. Saldana TM, Siega-Riz AM, Adair LS. Effect of macronutrient intake on the development of
glucose intolerance during pregnancy. The American Journal of Clinical Nutrition.
2004;79(3):479-86.
285. He JR, Yuan MY, Chen NN, Lu JH, Hu CY, Mai WB, et al. Maternal dietary patterns and gestational
diabetes mellitus: A large prospective cohort study in China. British Journal of Nutrition.
2015;113(8):1292-300.
286. Higgins JPT GS. Cochrane handbook for systematic reviews of interventions: The Cochrane
Collaboration; 2011. Available from: Available from www.handbook.cochrane.org.
287. Hardy K, Brand-Miller J, Brown KD, Thomas MG, Copeland L. The importance of dietary
carbohydrate in human evolution. The Quarterly Review of Biology. 2015;90(3):251-68.
288. Elia M, Folmer P, Schlatmann A, Goren A, Austin S. Carbohydrate, fat, and protein metabolism
in muscle and in the whole body after mixed meal ingestion. Metabolism: Clinical and
Experimental. 1988;37(6):542-51.
289. Somogyi M. and Weichselbaum T.E. Ketone-sparing effects of glucose. The Journal of Biological
Biochemistry. 1942;145:567-74.
290. Australian Dietary Guidelines. NHMRC (2013): Australian dietary guidelines Canberra: National
Health and Medical Research Council; 2013 [Available from:
https://www.nhmrc.gov.au/guidelines-publications/n55.
291. Food and Agriculture Organization. Global trends in production and consumption of
carbohydrate foods 2018 [Available from:
http://www.fao.org/docrep/w8079e/w8079e0g.htm.
177
292. UN. World population prospects: The 2017 revision. NY, USA: UN; 2017.
293. Bhupathiraju SN, Hu FB. Epidemiology of obesity and diabetes and their cardiovascular
complications. Circulation Research. 2016;118(11):1723-35.
294. Sato J, Kanazawa AMDP, Makita S, Hatae C, Komiya K, Shimizu T, et al. A Randomized controlled
trial of 130 g/day low-carbohydrate diet in type 2 diabetes with poor glycemic control. Clinical
Nutrition. 2016;36(4):992-1000.
295. Peterson CM, Jovanovic-Peterson L. Percentage of carbohydrate and glycemic response to
breakfast, lunch, and dinner in women with gestational diabetes. Diabetes. 1991;40(2):172-4.
296. Hernandez TL, Van Pelt RE, Anderson MA, Daniels LJ, West NA, Donahoo WT, et al. A higher-
complex carbohydrate diet in gestational diabetes mellitus achieves glucose targets and lowers
postprandial lipids: a randomized crossover study. Diabetes Care. 2014;37(5):1254-62.
297. Potter JM, Reckless JPD, Cullen DR. Diurnal variations in blood intermediary metabolites in mild
gestational diabetic patients and the effect of a carbohydrate-restricted diet. Diabetologia.
1982;22(2):68-72.
298. Young BC, Ecker JL. Fetal macrosomia and shoulder dystocia in women with gestational
diabetes: Risks amenable to treatment? Current Diabetes Reports. 2013;13(1):12-8.
299. Mulla WR. Carbohydrate content in the GDM diet: Two Views: View 2: Low-carbohydrate diets
should remain the initial therapy for gestational diabetes. Diabetes Spectrum. 2016;29(2):89-
91.
300. Wong VW, Jalaludin B. Gestational diabetes mellitus: Who requires insulin therapy? Australian
and New Zealand Journal of Obstetrics and Gynaecology. 2011;51(5):432-6.
178
301. Blumer I, Hadar E, Hadden DR, Jovanovic L, Mestman JH, Murad MH, et al. Diabetes and
pregnancy: An Endocrine Society clinical practice guideline. Diabetes Technology and
Therapeutics. 2015;17:S67-S8.
302. Adam-Perrot A, Clifton P, Brouns F. Low-carbohydrate diets: nutritional and physiological
aspects. Obesity Reviews. 2006;7(1):49-58.
303. Kizirian NV, Markovic TP, Muirhead R, Brodie S, Garnett SP, Louie JC, et al. Macronutrient
balance and dietary glycemic index in pregnancy predict neonatal body composition. Nutrients.
2016:1-13.
304. Louie JCY, Markovic TP, Ross GP, Foote D, Brand-Miller JC. Higher glycemic load diet is associated
with poorer nutrient intake in women with gestational diabetes mellitus. Nutrition Research.
2013;33(4):259-65.
305. Metzger, Boyd E. International Association of Diabetes and Pregnancy Study Groups Consensus
Panel. International Association of Diabetes and Pregnancy Study Groups recommendations on
the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care. 2010;33(3):676-
82.
306. Hackett AF, Appleton DR, Rugg-Gunn AJ, Eastoe JE. Some influences on the measurement of
food intake during a dietary survey of adolescents. Human Nutrition Applied Nutrition.
1985;39(3):167.
307. Urbain P, Bertz H. Monitoring for compliance with a ketogenic diet: What is the best time of day
to test for urinary ketosis? Nutrition and Metabolism. 2016;13(1):1-6.
179
308. Higham R, Tharmanathan P, Birks Y. Use and reporting of restricted randomization: A review.
Journal of Evaluation in Clinical Practice. 2015;21(6):1205-11.
309. Sedgwick P. Treatment allocation in trials: block randomisation. The British Medical Journal.
2014;348(mar28 1):g2409-g.
310. IOM. Dietary reference intakes for energy, carbohydrate, fiber, fat, fatty acids, cholesterol,
protein, and amino acids. Washington, D.C: National Academies Press; 2005.
311. Atkinson FS, Foster-Powell K, Brand-Miller JC. International tables of glycemic index and lycemic
load values: 2008. Diabetes Care. 2008;31(12):2281-3.
312. Dobbins TA, Sullivan EA, Roberts CL, Simpson JM. Australian national birthweight percentiles by
sex and gestational age, 1998-2007. The Medical journal of Australia. 2012;197(5):291-4.
313. Beeby PJ, Bhutap T, Taylor LK. New South Wales population-based birthweight percentile charts.
Journal of paediatrics and child health. 1996;32(6):512-8.
314. National Center for Health Statistics and National Center for Chronic Disease Prevention and
Health Promotion. CDC Growth charts: United States - Head circumference-for-age percentiles;
Boys, birth-36 months 2000 [Available from:
https://www.cdc.gov/growthcharts/data/set1/chart09.pdf.
315. National Center for Health Statistics and National Center for Chronic Disease Prevention and
Health Promotion. CDC Growth charts: United States - Head circumference-for-age percentiles;
Girls, birth-36 months 2000 [Available from:
https://www.cdc.gov/growthcharts/data/set1clinical/cj41l020.pdf.
180
316. IOM, National Research Council Committee to Reexamine, IOM Pregnancy Weight Guidelines.
The National Academies Collection: Reports funded by National Institutes of Health. In:
Rasmussen KM, Yaktine AL, editors. Weight Gain During Pregnancy: Reexamining the
Guidelines. Washington (DC): National Academies Press (US) National Academy of Sciences.;
2009.
317. Gin H, Vambergue A, Vasseur C, Rigalleau V, Dufour P, Roques A, et al. Blood ketone monitoring:
A comparison between gestational diabetes and non-diabetic pregnant women. Diabetes and
Metabolism. 2006;32(6):592-7.
318. Brandt I. Head circumference and brain development: Growth retardation during intrauterine
malnutrition and catch-up growth mechanisms. Klinische Wochenschrift. 1981;59(17):995-
1007.
319. Frisch S, Zittermann A, Berthold HK, Götting C, Kuhn J, Kleesiek K, et al. A randomized controlled
trial on the efficacy of carbohydrate-reduced or fat-reduced diets in patients attending a
telemedically guided weight loss program. Cardiovascular Diabetology. 2009;8(1):36.
320. Pichon L, Huneau JF, Fromentin G, Tome D. A high-protein, high-fat, carbohydrate-free diet
reduces energy intake, hepatic lipogenesis, and adiposity in rats. The Journal of Nutrition.
2006;136(5):1256-60.
321. Knopp RH, Magee MS, Raisys V, Benedetti T. Metabolic effects of hypocaloric diets in
management of gestational diabetes. Diabetes. 1991;40 Suppl 2(Supplement_2):165-71.
322. Rae A, Bond D, Evans S, North F, Roberman B, Walters B. A randomised controlled trial of dietary
energy restriction in the management of obese women with gestational diabetes. The Australian
& New Zealand Journal of Obstetrics & Gynaecology. 2000;40(4):416-22.
181
323. Anderson GD, Ahokas RA, Lipshitz J, Dilts JPV. Effect of maternal dietary restriction during
pregnancy on maternal weight gain and fetal birth weight in the rat. The Journal of Nutrition.
1980;110(5):883-90.
324. Gilmore LA, Butte NF, Ravussin E, Han H, Burton JH, Redman LM. Energy Intake and Energy
Expenditure for Determining Excess Weight Gain in Pregnant Women. Obstetrics and
Gynecology. 2016;127(5):884-92.
325. Ramírez-López MT, Vázquez M, Lomazzo E, Hofmann C, Blanco RN, Alén F, et al. A moderate
diet restriction during pregnancy alters the levels of endocannabinoids and endocannabinoid-
related lipids in the hypothalamus, hippocampus and olfactory bulb of rat offspring in a sex-
specific manner. Public Library of Science One. 2017;12(3):e0174307-e.
326. Zinn C, Rush A, Johnson R. Assessing the nutrient intake of a low-carbohydrate, high-fat (LCHF)
diet: A hypothetical case study design. The British Medical Journal. 2018;8(2):1-7.
327. Guess N. Dietary intake in people consuming a reduced-carbohydrate diet in the National Diet
and Nutrition Survey. Journal of Human Nutrition and Dietetics. 2017;30(3):360-8.
328. Desrosiers TA, Siega-Riz AM, Mosley BS, Meyer RE. Low carbohydrate diets may increase risk of
neural tube defects. Birth Defects Research. 2018;110(11):901-9.
329. Hernandez TL, Van Pelt RE, Anderson MA, Reece MS, Reynolds RM, de la Houssaye BA, et al.
Women with gestational diabetes mellitus randomized to a higher-complex carbohydrate/low-
fat diet manifest lower adipose tissue insulin resistance, inflammation, glucose, and free fatty
acids: A pilot study. Diabetes Care. 2016;39(1):39-42.
182
330. Romon M, Nuttens MC, Vambergue A, Verier-Mine O, Biausque S, Lemaire C, et al. Higher
carbohydrate intake is associated with decreased incidence of newborn macrosomia in women
with gestational diabetes. Journal of the American Dietetic Association. 2001;101(8):897-902.
331. Moore VM, Davies MJ, Willson KJ, Worsley A, Robinson JS. Dietary composition of pregnant
women is related to size of the baby at birth. The Journal of Nutrition. 2004;134(7):1820-6.
332. McKenzie KM, Dissanayake HU, McMullan R, Caterson ID, Celermajer DS, Gordon A, et al.
Quantity and quality of carbohydrate intake during pregnancy, newborn body fatness and
cardiac autonomic control: Conferred cardiovascular risk? Nutrients. 2017;9(12):1-12.
333. Jung CH, Choi KM. Impact of high-carbohydrate diet on metabolic parameters in patients with
type 2 diabetes. Nutrients. 2017;9(4):1-21.
334. Rebholz SL, Burke KT, Yang Q, Tso P, Woollett LA. Dietary fat impacts fetal growth and
metabolism: uptake of chylomicron remnant core lipids by the placenta. American Journal of
Physiology. 2011;301(2):E416-E25.
335. Taylor M, Galanis E. Food safety during pregnancy. Canadian Family Physician. 2010;56(8):750-
1.
336. Health Q. Healthy food access basket: Average costs by food category in 2014: Queensland
Government 2016 [Available from: https://www.health.qld.gov.au/research-
reports/reports/public-health/food-nutrition/access/category#milk.
337. Mani I, Dwarkanath P, Thomas T, Thomas A, Kurpad AV. Maternal fat and fatty acid intake and
birth outcomes in a South Indian population. International Journal of Epidemiology.
2016;45(2):523-31.
183
338. Bell JD, Margen S, Calloway DH. Ketosis, weight loss, uric acid, and nitrogen balance in obese
women fed single nutrients at low caloric levels. Metabolism: Clinical and Experimental.
1969;18(3):193-208.
339. Calloway DH. Dietary components that yield energy. Environmental Biology and Medicine.
1972;1(3):175-86.
340. Vice E, Privette JD, Hickner RC, Barakat HA. Ketone body metabolism in lean and obese women.
Metabolism: Clinical and Experimental. 2005;54(11):1542-5.
341. Hanson PG, Johnson RE, Zaharko DS. Correlation between ketone body and free fatty acid
concentrations in the plasma during early starvation in man. Metabolism: Clinical and
Experimental. 1965;14(10):1037-40.
342. Sivan E, Boden G. Free fatty acids, insulin resistance, and pregnancy. Current Diabetes Reports.
2003;3(4):319-22.
343. Schaefer-Graf UM, Graf K, Kulbacka I, Kjos SL, Dudenhausen J, Vetter K, et al. Maternal lipids as
strong determinants of fetal environment and growth in pregnancies with gestational diabetes
mellitus. Diabetes Care. 2008;31(9):1858-63.
344. Ivanovic DM, Leiva BP, Pérez HT, Olivares MG, Dıaz NS, Urrutia MaSC, et al. Head size and
intelligence, learning, nutritional status and brain development: Head, IQ, learning, nutrition
and brain. Neuropsychologia. 2004;42(8):1118-31.
345. Gale CR, O'Callaghan FJ, Bredow M, Martyn CN, Avon Longitudinal Study of P, Children Study T.
The influence of head growth in fetal life, infancy, and childhood on intelligence at the ages of
4 and 8 years. Pediatrics. 2006;118(4):1486.
184
346. Veena SR, Krishnaveni GV, Fall CH. Newborn size and body composition as predictors of insulin
resistance and diabetes in the parents: Parthenon Birth Cohort Study, Mysore, India. Diabetes
Care. 2012;35(9):1884-90.
347. Yokoyama Y, Sugimoto M, Ooki S. Analysis of factors affecting birthweight, birth length and head
circumference: study of Japanese triplets. Twin Research and Human Genetics. 2005;8(6):657-
63.
348. Bouthoorn SH, Frank JvL, Anita CSH-K, Moll HA, Tiemeier H, Hofman A, et al. Head circumference
of infants born to mothers with different educational Levels; The generation R study. Public
Library of Science One 2012;7(6):e39798.
349. Pedersen M, von Stedingk H, Botsivali M, Agramunt S, Alexander J, Brunborg G, et al. Birth
weight, head circumference, and prenatal exposure to acrylamide from maternal diet: The
European prospective mother-child study (NewGeneris). Environmental Health Perspectives.
2012;120(12):1739-45.
350. Xiao L, Ding G, Vinturache A, Xu J, Ding Y, Guo J, et al. Associations of maternal pre-pregnancy
body mass index and gestational weight gain with birth outcomes in Shanghai, China. Scientific
Reports. 2017;7:41073.
351. Obel C, Hedegaard M, Henriksen TB, Secher NJ, Olsen J. Stressful life events in pregnancy and
head circumference at birth. Developmental Medicine and Child Neurology. 2003;45(12):802-6.
352. Rhodes ET, Pawlak DB, Takoudes TC, Ebbeling CB, Feldman HA, Lovesky MM, et al. Effects of a
low–glycemic load diet in overweight and obese pregnant women: A pilot randomized
controlled trial. The American Journal of Clinical Nutrition. 2010;92(6):1306-15.
185
353. Moses RG, Luebcke M, Davis WS, Coleman KJ, Tapsell LC, Petocz P, et al. Effect of a low-glycemic-
index diet during pregnancy on obstetric outcomes. The American Journal of Clinical Nutrition.
2006;84(4):807-12.
354. Millichap JG. Head Circumference and Neurocognitive Outcomes. Pediatric Neurology Briefs.
2015;29(7):54.
355. Nowicki E, Siega-Riz A-M, Herring A, He K, Stuebe A, Olshan A. Predictors of measurement error
in energy intake during pregnancy. American Journal of Epidemiology. 2011;173(5):560-8.
356. Winkvist A, Persson V, Hartini TN. Underreporting of energy intake is less common among
pregnant women in Indonesia. Public Health Nutrition. 2002;5(4):523-9.
357. Peuchant E, Brun J-L, Rigalleau V, Dubourg L, Thomas M-J, Daniel J-Y, et al. Oxidative and
antioxidative status in pregnant women with either gestational or type 1 diabetes. Clinical
Biochemistry. 2004;37(4):293-8.
358. Peters SJ, LeBlanc PJ. Metabolic aspects of low carbohydrate diets and exercise. Nutrition and
Metabolism. 2004;1:7.
359. Goffinet L, Barrea T, Beauloye V, Lysy PA. Blood versus urine ketone monitoring in a pediatric
cohort of patients with type 1 diabetes: A crossover study. Therapeutic Advances in
Endocrinology and Metabolism. 2017;8(1-2):3-13.
360. ABS. 2016 Census QuickStats - Campbelltown (C) (NSW) 2018 [updated 17th August 2018; cited
2018 26th August 2018]. Available from:
http://quickstats.censusdata.abs.gov.au/census_services/getproduct/census/2016/quickstat/L
GA11500.
186
361. ABS. 2016 Census QuickStats - Inner West (A) (NSW) 2018 [updated 17th August 2018. Available
from:
http://quickstats.censusdata.abs.gov.au/census_services/getproduct/census/2016/quickstat/L
GA14170?opendocument.
362. Hoffman L, Nolan C, Wilson JD, Oats JJ, Simmons D. Gestational diabetes mellitus - management
guidelines. The Australasian Diabetes in Pregnancy Society. The Medical Journal of Australia.
1998;169(2):93.
363. Nankervis A MH, Moses R, Ross GP, Callaway L, Porter C, Jeffries W, Boorman C, De Vries B,
McElduff A. ADIPS Consensus Guidelines for the testing and diagnosis of hyperglycaemia in
pregnancy in Australia and New Zealand. Sydney: Australasian Diabetes in Pregnancy Society;
2014.
364. Breij LM, Steegers-Theunissen RPM, Briceno D, Hokken-Koelega ACS. Maternal and fetal
determinants of neonatal body composition. Hormone Research in Paediatrics. 2015;84(6):388-
95.
365. WHO. The WHO child growth standards - weight-for-age, birth to 6 month (boys): WHO;
[Available from: https://www.who.int/childgrowth/standards/cht_wfa_boys_p_0_6.pdf?ua=1.
366. WHO. The WHO child growth standards - weight-for-age percentiles, birth to 6 month (girls):
WHO; [Available from: https://www.who.int/childgrowth/standards/chts_wfa_girls_p/en/.
367. Firth D. Bias reduction of maximum likelihood estimates. Biometrika. 1993;80(1):27-38.
368. Faucher MA, Barger MK. Gestational weight gain in obese women by class of obesity and select
maternal/newborn outcomes: A systematic review. Women and Birth. 2015;28(3):e70-e9.
187
369. Spanou L, Dalakleidi K, Zarkogianni K, Papadimitriou A, Nikita K, Vasileiou V, et al. Ketonemia
and ketonuria in gestational diabetes mellitus. Hormones (Athens, Greece). 2015;14(4):644-50.
370. Robinson HL, Barrett HL, Foxcroft K, Callaway LK, Dekker Nitert M. Prevalence of maternal
urinary ketones in pregnancy in overweight and obese women. Obstetric Medicine.
2018;11(2):79-82.
371. Sridhar SB, Xu F, Hedderson MM. Trimester-specific gestational weight gain and infant size for
gestational age. Public Library of Science One. 2016;11(7):e0159500.
372. Yee LM, Cheng YW, Inturrisi M, Caughey AB. Gestational weight loss and perinatal outcomes in
overweight and obese women subsequent to diagnosis of gestational diabetes mellitus. Obesity
(Silver Spring). 2013;21(12):E770-E4.
373. Catalano PM, Shankar K. Obesity and pregnancy: Mechanisms of short term and long term
adverse consequences for mother and child. The British Medical Journal. 2017;356:1-16.
374. Fields DAP, Krishnan SMD, Wisniewski ABP. Sex differences in body composition early in life.
Gender Medicine. 2009;6(2):369-75.
375. Villar J, Puglia FA, Fenton TR, Cheikh Ismail L, Staines-Urias E, Giuliani F, et al. Body composition
at birth and its relationship with neonatal anthropometric ratios: The newborn body
composition study of the INTERGROWTH-21st project. Pediatric Research. 2017;82(2):305-16.
376. Bejar LM, Reyes OA, Garcia-Perea MD. Electronic 12-Hour dietary recall (e-12HR): Comparison
of a mobile phone app for dietary intake assessment with a food frequency questionnaire and
four dietary records. Journal of Medical Internet Research mHealth and uHealth.
2018;6(6):e10409.
188
377. Ortega RM, Perez-Rodrigo C, Lopez-Sobaler AM. Dietary assessment methods: Dietary records.
Nutricion Hospitalaria. 2015;31:38-45.
378. Knutson KL. Impact of sleep and sleep loss on glucose homeostasis and appetite regulation.
Sleep Medicine Clinics. 2007;2(2):187-97.
379. Morselli L, Leproult R, Balbo M, Spiegel K. Role of sleep duration in the regulation of glucose
metabolism and appetite. Best Practice and Research: Clinical Endocrinology and Metabolism.
2010;24(5):687-702.
380. Sedov IDM, Cameron EEM, Madigan SP, Tomfohr-Madsen LMP. Sleep quality during pregnancy:
A meta-analysis. Sleep Medicine Reviews. 2017;38:168-76.
381. Lauderdale D.S KKL, Yan L.L., Liu K, Rathouz P.J. Sleep duration: how well do self-reports reflect
objective measures? The CARDIA Sleep Study. Epidemiology. 2008;19(6):838-45.
382. Bird SR, Hawley JA. Update on the effects of physical activity on insulin sensitivity in humans.
The British Medical Journal Open Sport and Exercise Medicine. 2017;2(1):e000143-e.
383. Committee on Practice, Bulletins-Obstetrics. Practice Bulletin No. 137: Gestational diabetes
mellitus. Obstetrics and Gynecology. 2013;122(2 Pt 1):406-16.
384. Hajianfar H, Esmaillzadeh A, Feizi A, Shahshahan Z, Azadbakht L. Major maternal dietary
patterns during early pregnancy and their association with neonatal anthropometric
measurement. BioMed Research International. 2018;2018:4692193-.
385. Hjertholm KG, Iversen PO, Holmboe-Ottesen G, Mdala I, Munthali A, Maleta K, et al. Maternal
dietary intake during pregnancy and its association to birth size in rural Malawi: A cross-sectional
study. Maternal and Child Nutrition. 2017;14(1):e12433.
189
386. Koski KG, Lanoue L, Young SN. Restriction of maternal dietary carbohydrate decreases fetal brain
indoles and glycogen in rats. The Journal of Nutrition. 1993;123(1):42-51.
387. Simpson SJ, Raubenheimer D. Obesity: The protein leverage hypothesis. Obesity Reviews.
2005;6(2):133-42.
388. Bhatia BD, Banerjee D, Agarwal DK, Agarwal KN. Fetal growth: relationship with maternal dietary
intakes. The Indian Journal of Pediatrics. 1983;50(2):113-20.
389. Krasnow SM, Nguyen MLT, Marks DL. Increased maternal fat consumption during pregnancy
alters body composition in neonatal mice. American Journal of Physiology. 2011;301(6):E1243-
E53.
390. Jones HN, Woollett LA, Barbour N, Prasad PD, Powell TL, Jansson T. High-fat diet before and
during pregnancy causes marked up-regulation of placental nutrient transport and fetal
overgrowth in C57/BL6 mice. FASEB Journal. 2009;23(1):271-8.
391. Gustafsson HC, Kuzava SE, Werner EA, Monk C. Maternal dietary fat intake during pregnancy is
associated with infant temperament. Developmental Psychobiology. 2016;58(4):528-35.
392. Ljubomir M, Ivan H, Zorica G, Mirjana B, Aleksandra N. Biochemical and physiological
characteristics of neonates born to mothers with diabetes during gestation. Journal of Medical
Biochemistry. 2012;31(1):47-52.
393. Lain SJD, Bentley JPM, Wiley VP, Roberts CLD, Jack MM, Wilcken BP, et al. Association between
borderline neonatal thyroid-stimulating hormone concentrations and educational and
developmental outcomes: a population-based record-linkage study. The Lancet Diabetes and
Endocrinology. 2016;4(9):756-65.
190
394. Becker DV, Braverman LE, Delange F, Dunn JT, Franklyn JA, Hollowell JG, et al. Iodine
supplementation for pregnancy and lactation - United States and Canada: recommendations of
the American Thyroid Association. Thyroid. 2006;16(10):949-51.
395. Heinemann L. Finger pricking and pain: A never ending story. Journal of Diabetes Science and
Technology. 2008;2(5):919-21.
396. Lin T, Mayzel Y, Bahartan K. The accuracy of a non-invasive glucose monitoring device does not
depend on clinical characteristics of people with type 2 diabetes mellitus. Journal of Drug
Assessment. 2018;7(1):1-7.
397. Newton CA, Raskin P. Diabetic ketoacidosis in type 1 and type 2 diabetes mellitus: Clinical and
biochemical differences. Archives of internal medicine. 2004;164(17):1925-31.
191
Appendices
___________________________________________________
Figure A1: Assessing the risk of publication bias using funnel plots for different meta-analyses. lnOR, natural log odds ratio
Figure A1a. Any type of PA in early pregnancy versus none (n studies = 9, z = -0.65, p = 0.52).
Figure A1b. Pre-pregnancy LTPA high versus none reported in MET.hr/week (n studies = 6, z = -2.96, p = 0.003).
Figure A1c. Pre-pregnancy LTPA high versus none levels reported in hr/week, (n studies = 4, z = -2.34, p = 0.02). Due to insufficient number of studies reporting on early pregnancy.
Table A1. Natural odds ratio (lnOR) values before back-transformation correspond to Figures 3A, 3B, 4A, 4B, 5 and 6.
Study lnOR 95 % Confidence Intervals Lower Upper
Figu
re 3
A
Badon et al. 2016 (234) -0.64 -1.18 -0.11
Chasan-Taber et al. 2008 (251) -0.22 -1.43 0.99
Chasan-Taber et al. 2014 (250) -0.24 -1.12 0.65
Currie et al. 2014 (254) -0.51 -1.37 0.35
Dempsey et al. 2004 (235) -1.76 -2.69 -0.83
Morkrid et al. 2014 (256) -0.37 -0.70 -0.05
Solomon et al. 1997 (217) -0.04 -0.30 0.21
Van der Ploeg et al. 2011 (233) 0.08 -0.35 0.50
Zhang et al. 2006 (221) -0.36 -0.54 -0.19
Oken et al. 2006 (234) -0.58 -1.12 -0.04
Rudra et al. 2006 (229) -0.46 -1.34 0.42
OVERALL -0.36 -0.57 -0.16
Figu
re 3
B
Badon et al. 2016 (224) -0.56 -1.04 -0.08
Chasan-Taber et al. 2008 (241) -0.36 -1.33 0.61
Chasan-Taber et al. 2014 (240) -0.37 -1.26 0.52
Currie et al. 2014 (244) -0.58 -1.49 0.33
Dempsey et al. 2004 (225) -0.35 -1.04 0.35
Dye et al. 1997 (237) 0.00 -0.22 0.22
Morkrid et al. 2014 (246) -0.30 -0.67 0.07
Oken et al. 2006 (234) -0.11 -0.70 0.49
OVERALL -0.24 -0.45 -0.03
Figu
re 4
A
Badon et al. 2016 (224) -0.64 -1.18 -0.11
Chasan-Taber et al. 2008 (241) 0.74 -0.43 1.92
Chasan-Taber et al. 2014 (240) 0.23 -0.63 1.09
Rudra et al. 2006 (229) -1.71 -2.64 -0.77
Dempsey et al. 2004 (225) -0.37 -0.70 -0.05
Morkrid et al. 2014 (246) -1.97 -2.86 -1.07
Solomon et al. 1997 (217) -0.04 -0.30 0.21
Van der Ploeg et al. 2011 (233) 0.20 -0.22 0.62
Zhang et al. 2006 (221) -0.36 -0.53 -0.19
Oken et al. 2006 (234) -0.58 -1.12 -0.04
OVERALL -0.43 -0.86 0.00
Study
lnOR
95 % Confidence Intervals
Lower Upper Fi
gure
4B
Badon et al. 2016 (224) -0.67 -1.16 -0.18
Chasan-Taber et al. 2008 (241) -0.36 -1.50 0.79
Chasan-Taber et al. 2014 (240) 0.07 -0.68 0.82
Dempsey et al. 2004 (225) -0.63 -1.39 0.13
Oken et al. 2006 (234) -0.11 -0.70 0.49
OVERALL -0.37 -0.70 -0.04
Figu
re 5
Badon et al. 2014 (224) -0.64 -1.18 -0.11
Dempsey et al. 2004 (225) -1.71 -2.64 -0.77
Rudra et al. 2006 (229) -1.97 -2.86 -1.07
Solomon et al. 1997 (217) -0.04 -0.30 0.21
Van der Ploeg et al. 2011 (233) 0.20 -0.22 0.62
Zhang et al. 2006 (221) -0.36 -0.53 -0.19
OVERALL -0.66 -1.32 -0.00
Figu
re 6
Badon et al. 2014 (224) -0.64 -1.18 -0.11
Dempsey et al. 2004 (225) -1.71 -2.64 -0.77
Morkrid et al. 2014 (246) -0.37 -0.70 -0.05
Solomon et al. 1997 (217) -0.27 -0.76 0.22
OVERALL -0.62 -1.09 -0.14
