Effect of a low-glycemic-index diet during pregnancy on obstetric
outcomes1⫺3
Robert G Moses, Megan Luebcke, Warren S Davis, Keith J Coleman, Linda C Tapsell, Peter Petocz, and
Jennie C Brand-Miller
ABSTRACT
Background: Pregnancy is a condition in which the glycemic index
(GI) may be of particular relevance because maternal glucose is the
main energy substrate for intrauterine growth.
Objective: The aim was to compare the effects of low-GI and
conventional dietary strategies on pregnancy outcomes in healthy
women. Compliance and acceptability were also investigated.
Design: The subjects were assigned alternately to receive dietary
counseling that encouraged either low-GI (LGI) carbohydrate foods
or high-fiber, moderate-to-high GI (HGI) foods and were studied
5 times between 쏝16 wk gestation and delivery. Of the 70 women
who met the inclusion criteria, 62 completed the study (32 in the LGI
and 30 in the HGI groups). Primary outcomes were measures of fetal
size.
Results: The mean diet GI fell significantly in the LGI group but not
in the HGI group. Compared with the LGI group, women in the HGI
group gave birth to infants who were heavier (3408 앐 78 compared
with 3644 앐 90 g; P ҃ 0.051) and had a higher birth centile (48 앐
5 compared with 69 앐 5; P ҃ 0.005), a higher ponderal index (2.62 앐
0.04 compared with 2.74 앐 0.04; P ҃ 0.03), and a higher prevalence
of large-for-gestational age (3.1% compared with 33.3%; P ҃ 0.01).
Women in the LGI group found the diet easier to follow.
Conclusion: Because birth weight and ponderal index may predict
chronic disease in later life, a low-GI diet may favorably influence
long-term outcomes. Am J Clin Nutr 2006;84:807–12.
KEY WORDS Glycemic index, pregnancy, birth weight, pon-
deral index, insulin sensitivity
INTRODUCTION
The clinical significance of the glycemic index (GI) of dietary
carbohydrates has been a subject of debate since the concept was
raised 25 y ago (1). The relation of the GI and, more recently,
dietary glycemic load (GL) to the development and progression
of chronic diseases, particularly those associated with insulin
resistance, remains controversial (2, 3). Although there is no
doubt that a low-GI food choice will ameliorate the postprandial
glucose rise (4), the relevance of the postprandial glucose even in
disorders such as diabetes is still debated (5). This is not unex-
pected given the number of confounding variables in both as-
sessing dietary intake and relating this to the progression of
chronic diseases and specific complications. In addition, the re-
liability of GI values extrapolated from foods tested in healthy
volunteers in one country may be questionable in other countries
and for persons with chronic diseases.
Am J Clin Nutr 2006;84:807–12. Printed in USA. © 2006 American Society for Nutrition
Pregnancy is a physiologic condition in which the GI concept
Downloaded from https://academic.oup.com/ajcn/article/84/4/807/4633214 by guest on 12 April 2021
may be of particular relevance because maternal glucose is the
main energy substrate for intrauterine growth (6). For example,
in pregnancies complicated by diabetes, high concentrations of
glucose, particularly after a meal, give rise to an increased nu-
trient transfer to the fetus and adversely influence birth weight
and pregnancy complications (6, 7).
To date, the validity of the GI concept in pregnancy has been
assessed for only a small number of foods (8), and there is little
knowledge about the effect of a low-GI diet on pregnancy out-
comes. Two published studies in glucose-tolerant women have
yielded conflicting results (9, 10). Moreover, to our knowledge,
no studies have examined the tolerability and sustainability of a
low-GI diet in pregnancy.
The primary aim of this study therefore was to compare the
effects of 2 diets, one designated as low sugar and high fiber with
a moderate-to-high GI (HGI) and one with a low GI (LGI), on
pregnancy outcomes in healthy women. The secondary aims
were to examine the compliance with and the acceptability of a
low-GI diet in pregnancy.
SUBJECTS AND METHODS
The study was conducted in the city of Wollongong, NSW,
Australia, a coastal city with a population of 앒280 000 people
situated 앒50 miles south of Sydney. Healthy pregnant women
were recruited from the antenatal clinic at Wollongong Hospital
and from 2 obstetricians in private practice. They were consid-
ered for this parallel controlled study if they were aged 21– 40 y,
had a singleton pregnancy, were between 12 and 16 wk gestation,
were nonsmokers, and had no more than 1 alcoholic drink each
day. Exclusion criteria were individually assessed by one of us
1
From the Illawarra Area Health Service and Wollongong Hospital, Wol-
longong, NSW, Australia (RGM, ML, WSD, and KJC); Smart Food Center,
University of Wollongong, Wollongong, NSW, Australia (LCT); the De-
partment of Statistics, Macquarie University, Sydney, NSW, Australia (PP);
and the Human Nutrition Unit, University of Sydney, NSW, Australia
(JCB-M).
2
Supported by internal revenue from the Illawarra Diabetes Service and
the University of Sydney. Several items in the food hamper were provided by
Sanitarium Health Foods, Cooranbong, NSW, Australia.
3
Reprints not available. Address correspondence to RG Moses, PO
Box 1958, Wollongong West, NSW Australia 2500. E-mail:
robert.moses@sesiahs.nsw.gov.au.
Received March 5, 2006.
Accepted for publication May 30, 2006.
807
808 MOSES ET AL
(RGM) and included any problem that may have been associated
with glucose metabolism or insulin resistance or interfered with
the ability of the study participant to follow dietary instructions.
Women who were not excluded were assigned alternately to 1
of the 2 diets. A research dietitian saw each woman individually
5 times during the pregnancy. At visit 1 (baseline), a 3-d food
record and diet history were obtained, and height and weight
were obtained. Weight was measured to the nearest 0.1 kg on
floor scales (HD-316, Wedderburn Scales; Tanita Corporation,
Tokyo, Japan) with subjects dressed in light clothes and without
shoes. Height was measured to the nearest 0.1 cm against a wall
with the use of a nonstretchable fiberglass measuring tape
(Gulick II; Country Technology, Inc, Gays Mills, WI). At visit 2,
1 wk later, participants received detailed dietary education tai-
lored for the assigned diet and their individual requirements for
pregnancy. At visits 3 and 4 at 앒22 and 30 wk gestation, respec-
tively, a 24-h diet recall was taken. At visit 5 at 36 wk gestation,
a second 3-d food record and diet history were obtained.
A fasting venous sample was taken within 2 d of visit 1 for
measurement of glucose and insulin, and this was repeated at 28
wk gestation in conjunction with routine glucose tolerance test-
ing (GTT). Homeostasis modeling assessment (HOMA) was
used to estimate insulin resistance by using original linear model
(HOMA1-IR; fasting glucose ҂ fasting insulin/22.5) (11) and
the nonlinear computer model (HOMA2-IR) (12). The computer
model also generated HOMA2-insulin sensitivity (HOMA2-
%S) and HOMA2-␤ cell function (HOMA2-%␤).
Diets
Both diets were compatible with the recommended nutritional
intake in pregnancy (13), aiming for energy intake of 30% fat and
55% carbohydrate, with only the recommended choice of carbo-
hydrate foods varying. No specific or individual recommenda-
tions were made about the intake of total energy, fiber, and fat.
The HGI diet included advice to follow a diet with a high fiber
and low sugar content. Potatoes, wheat-meal bread, and specific
high-fiber breakfast cereals with moderate-to-high GI were rec-
ommended. The LGI diet was based on previously verified
low-GI foods, including pasta and brand-name breads and break-
fast cereals with a high fiber content. During visits, the dietitian
referred to the diets as the “high-fiber, low-sugar” diet or the
“low-GI” diet. Participants were provided with a booklet that
outlined the carbohydrate choices and the food amounts that
constituted one serving. To encourage compliance with both
diets, key foods were provided in a monthly hamper. The dieti-
tian also provided information on the whole diet to ensure energy
and overall nutrient balance and was available for telephone
queries outside of scheduled visits. Study personnel were not
blinded to dietary assignment but were aware of the need for
impartiality and equivalent treatment.
Food intake data for each participant was entered into a cus-
tomized database that incorporated the Australian food-
composition tables and published GI values by using the scale in
which glucose equaled 100 (FOODWORKS PROFESSIONAL,
version 4 2005; Xyris Software, Brisbane, Australia). When nec-
essary, additional GI data were obtained from an online database
(14). Overall dietary GI was calculated as the sum of the
weighted GI of all carbohydrate foods in the diet, with the
weighting proportional to the contribution of each food to total
carbohydrate intake. Because the target diets aimed for similar
carbohydrate content, GL (the product of the GI and the amount
of carbohydrate) was influenced only by differences in GI. Any
differences in obstetric outcomes could therefore be attributed to
the nature of carbohydrate per se (ie, GI) rather than to simulta-
neous changes in both quantity and quality.
To evaluate the acceptability of the recommended diet
changes in pregnancy, at their final visit subjects were asked to
score 6 statements on a 5-point Likert scale (1 being “strongly
agree” and 5 being “strongly disagree”). The statements were as
follows: “It was easy to follow the diet recommended during this
study,” “I enjoyed the dietary changes that I made,” “The
changes recommended were affordable,” “My family was ac-
cepting of the changes made to my eating habits,” “The study diet
helped me meet the physical challenges of pregnancy,” “I en-
joyed a wide variety of foods in my eating plan.” In addition,
women were also asked to indicate on a Likert scale how closely
Downloaded from https://academic.oup.com/ajcn/article/84/4/807/4633214 by guest on 12 April 2021
they followed the assigned diet (1 being “all of the time” and 5
being “none of the time”).
Pregnancy care was the responsibility of the obstetric health
care providers and was conducted in accord with standard prac-
tice. Because both diets were within the nutritional guidelines for
pregnancy, the obstetric health care providers were not specifi-
cally informed of the diet assignments. Obstetric outcomes, in-
cluding birth weight, length, head circumference, Apgar score,
and method of delivery, were obtained from the medical record.
For comparison between the 2 groups, the fetal centile was ob-
tained from the Centile Calculator (15) with the use of data from
a white British population. In this way the birth weight was
adjusted for sex, gestational week of delivery, maternal age,
parity, height, and prepregnancy weight by recall. The ponderal
index (in g/cm3҂ 100) of the infant was calculated. The body
mass index (BMI; in kg/m2) of the mother was calculated with the
prepregnancy weight by recall. The Illawarra Area Health Ser-
vice and University of Wollongong Human Research Committee
approved the research, and participants gave written informed
consent.
Statistical analysis
Independent samples t tests were used to compare groups at
baseline and at the final time points. In addition, analysis of
variance was used to compare groups at final visit with baseline
values as covariates. Paired t tests were used to assess changes
during the study period for the subjects as a whole and also within
groups. Pearson’s chi-square test of independence was used to
compare method of delivery and prevalence of infants large for
gestational age (LGA) and small for gestational age (SGA). Mul-
tivariate analysis of variance was used to compare selected ma-
ternal and fetal outcomes between groups. SPSS version 12.0
(SPSS Inc, Chicago, IL) was used for all statistical analyses.
Unless otherwise stated results are expressed as means 앐 SEMs.
Results were considered significant when P 쏝 0.05.
RESULTS
Seventy-seven women were considered for the study, but 7
were not included because they had one of the following condi-
tions: polycystic ovarian disease, Crohn disease, hyperthyroid-
ism, celiac disease, lupus, previous gestational diabetes mellitus
(GDM), and history of food allergy. A total of 70 women were
enrolled with 35 assigned to each group. In the LGI diet group, 2
women withdrew because they were unwilling to follow the diet
and 1 delivered before the final visit at 36 wk gestation. In the
 LOW-GI DIET IN PREGNANCY 809
TABLE 1 TABLE 2
Baseline characteristics and glucose and insulin concentrations at baseline Maternal weight gain and obstetric outcomes1
(visit 1) and at the final visit (visit 5) of subjects who completed the study1
Low-GI High-GI
Low-GI diet High-GI diet diet diet
(n ҃ 32) (n ҃ 30) P2 (n ҃ 32) (n ҃ 30) P

Age at baseline (y) 30.1 앐 0.73 29.6 앐 0.7 0.61 Maternal weight gain (kg)2 11.5 앐 0.53 10.1 앐 0.9 0.16
Weight at baseline (kg) 67 앐 2 73 앐 2 0.09 Method of delivery 0.62
BMI at baseline (kg/m2) 24.4 앐 0.7 26.6 앐 0.9 0.04 Normal vaginal 16 18
Parity at baseline 0.70 앐 0.15 0.87 앐 0.12 0.39 Vacuum 8 6
Fasting glucose (mmol/L) Elective cesarean 3 4
Baseline visit 4.4 앐 0.1 4.3 앐 0.1 0.17 Emergency cesarean 5 2
Final visit 4.1 앐 0.1 4.3 앐 0.1 0.034 Birth weight (g) 3408 앐 78 3644 앐 90 0.051
P4 0.001 0.75 Gestational age at delivery 39.5 앐 0.3 38.9 앐 0.2 0.066
2-h Postload glucose (wk)
(mmol/L) Head circumference (cm) 34.6 앐 0.25 35.1 앐 0.25 0.13
5.4 앐 0.2 5.5 앐 0.2 50.8 앐 0.3 51.1 앐 0.4

Downloaded from https://academic.oup.com/ajcn/article/84/4/807/4633214 by guest on 12 April 2021

Final visit only 0.71 Length (cm) 0.64
Fasting insulin (mU/L) Apgar score 9.1 앐 0.2 9.4 앐 0.1 0.13
Baseline visit 6.6 앐 0.6 7.9 앐 1.0 0.25 Birth centile 48 앐 5 69 앐 5 0.005
Final visit 18.0 앐 4.2 25.4 앐 5.4 0.30 Prevalence of LGA(%)4 3.1 33.3 0.001
HOMA1-IR Prevalence of SGA(%)4 9.4 6.7 0.69
Baseline visit 1.3 앐 0.1 1.6 앐 0.2 0.31 Ponderal index (g/cm3 ҂ 100) 2.62 앐 0.04 2.74 앐 0.04 0.03
Final visit 3.2 앐 0.7 4.9 앐 1.0 0.20 1
GI, glycemic index; LGA, large for gestational age; SGA, small for
HOMA2-IR
gestational age.
Baseline visit 0.8 앐 0.1 1.0 앐 0.1 0.27 2
Weight gain difference between visit 5 and visit 1.
Final visit 2.1 앐 0.4 3.0 앐 0.6 0.23 3
x៮ 앐 SEM (all such values).
HOMA2-%␤ 4
LGA 쏜90th centile for birth weight; SGA 쏝10th centile for birth
Baseline visit 108 앐 7 127 앐 7 0.07
weight.
Final visit 237 앐 34 275 앐 43 0.64
HOMA2-%S
Baseline visit 155 앐 15 134 앐 12 0.29
Final visit 110 앐 33 69 앐 8 0.30
1
women in the HGI group were born at an earlier week of gestation
The baseline visit was between 12 and 16 wk gestation; the final visit
and tended to be heavier than the infants of the women in the LGI
was at 36 wk gestation. GI, glycemic index; HOMA1, homeostasis modeling
assessment with original linear modeling; HOMA2, homeostasis modeling
diet group. To adjust for the known effects on birth weight of
assessment with nonlinear computer modeling; IR, insulin resistance; ␤, ␤ maternal age, weight, height, ethnicity, and parity, as well as
cell function; S, insulin sensitivity. infant sex and gestational age, birth centiles were calculated. The
2
When only baseline data or only final visit data are reported, P repre- mean centile of the infants of women in the HGI diet group (69 앐
sents a between-group comparison. When both baseline and final visit data 5) was significantly higher than those in the LGI diet group (48 앐
are reported, P for baseline represents a between-group comparison and P for 5; P ҃ 0.005). Significant differences were observed between the
final visit represents a comparison between the low-GI group and the high-GI groups in the prevalence of infants who were LGA (defined as
group, adjusted for differences at baseline. 쏜90th centile for birth weight), but no difference was observed
3
x៮ 앐 SEM (all such values). in the rate of SGA (쏝10th centile; Table 2). Ponderal index, an
4
Within-group comparison (baseline compared with final visit). Com-
estimate of infant body fatness, was also significantly higher in
parison of groups at baseline used independent samples t tests; comparison of
groups at final visit used independent samples t tests and ANOVA with
the HGI group, both before and after adjustment for baseline
baseline value as covariate. Comparisons from baseline to final visit used maternal BMI (P ҃ 0.030 and P ҃ 0.027, respectively). No
paired t tests. significant differences were observed in the method of delivery.
GDM was diagnosed in one woman from the HGI group.
Maternal glucose and insulin concentrations at the 28-wk GTT
HGI diet group, 1 woman was unwilling to follow the diet, 1 was and measures of glucose homeostasis at week 28 are shown in
lost to follow-up, and 3 had miscarriages. Data analysis was Table 1. Compared with baseline (visit 1), HOMA-IR increased
based on the 32 women in the LGI diet group and 30 women in markedly in both groups (HOMA1-IR, P 쏝 0.001; HOMA2-IR,
the HGI group who finished the study. P 쏝 0.001), whereas indirect measures of ␤ cell function in-
Women who were assigned to the HGI diet group had a slightly creased (HOMA2-%␤; P 쏝 0.001) and insulin sensitivity de-
higher BMI (P ҃ 0.04) and higher HOMA2-␤ cell function (P ҃ creased (HOMA2-%S; P ҃ 0.009). However, no effect of diet
0.07) than did women in the LGI diet group (Table 1). No other composition was observed on changes in glucose homeostasis.
significant differences were observed in the baseline character- The reported dietary intake assessed by a 3-d food record is
istics of the women. All women were white. shown in Table 3. At the baseline visit, no significant differences
Maternal weight gain (from visit 1 to visit 5) and obstetric were observed between the 2 groups in energy, macronutrient
outcomes are shown in Table 2. No significant difference were intakes, or diet GI. At the final visit, both groups had increased
observed in weight gain between the 2 groups. For the women in their fiber intake (by 앒5 g/d), but only the LGI diet group had
the LGI diet group, there were 16 male infants, whereas for the reduced diet GI (by 7 units; P 쏝 0.001) and intake of saturated fat
women in the HGI group there were 19 male infants. Although (P ҃ 0.006). Women in the HGI diet group had significantly
not reaching individual statistical significance, the infants of reduced their intake of polyunsaturated fat (by 앒17%; P ҃
810 MOSES ET AL

TABLE 3 TABLE 4
Reported dietary intake assessed by 3-d food record at baseline and the Ratings of acceptability for the 2 diets1
final visit1
LGI diet HGI diet P
High-GI diet
I adhered well to the dietary 2.1 앐 0.12 2.0 앐 0.1 0.766
Low-GI diet (n ҃ 30; 28
instructions.
(n ҃ 32) at final visit) P 2
It was easy to follow the 1.6 앐 0.1 1.9 앐 0.1 0.048
Energy (kJ) diet recommended during
Baseline visit 8540 앐 2803 9004 앐 350 0.30 this study.
Final visit 8740 앐 310 9020 앐 370 0.77 I enjoyed the dietary 1.6 앐 0.1 1.7 앐 0.1 0.541
Protein (% of energy) changes that I made.
Baseline visit 18.7 앐 0.6 17.4 앐 0.6 0.13 The changes recommended 1.6 앐 0.1 1.6 앐 0.1 0.646
Final visit 19.5 앐 0.6 19.6 앐 0.6 0.73 were affordable.
Carbohydrate (% of energy) My family was accepting of 1.8 앐 0.1 1.8 앐 0.1 0.677
Baseline visit 45.6 앐 1.0 47.1 앐 1.0 0.30 the changes made to my
Final visit 46.3 앐 0.9 46.4 앐 0.9 0.87 eating habits.
1.9 앐 0.2 1.9 앐 0.2

Downloaded from https://academic.oup.com/ajcn/article/84/4/807/4633214 by guest on 12 April 2021

Fat (% of energy) The study diet helped me 0.971
Baseline visit 31.3 앐 1.0 31.1 앐 0.8 0.88 meet the physical
Final visit 29.5 앐 0.9 29.2 앐 1.1 0.92 challenges of pregnancy.
MUFA (% of energy) I enjoyed a wide variety of 1.7 앐 0.1 1.7 앐 0.1 0.856
Baseline visit 11.1 앐 0.5 11.1 앐 0.3 0.97 foods in my eating plan.
Final visit 10.6 앐 0.5 10.5 앐 0.5 0.90 1
LGI, low glycemic index; HGI, high glycemic index.
PUFA (% of energy) 2
x៮ 앐 SEM (all such values).
Baseline visit 4.7 앐 0.2 4.6 앐 0.2 0.70
Final visit 5.3 앐 0.3 3.8 앐 0.2 0.001
P4 0.13 0.008 diets apart from the response to question 2. Women in the LGI
Saturated fatty acids (% of diet group were more likely to agree that their recommended diet
energy)
was easy to follow (P 쏝 0.048).
Baseline visit 12.6 앐 0.5 12.5 앐 0.5 0.94
Final visit 11.0 앐 0.5 12.1 앐 0.6 0.09
Alcohol (% of energy) DISCUSSION
Baseline visit 0.1 앐 0.1 0.1 앐 0.05 0.91
Final visit 0.2 앐 0.1 0.2 앐 0.1 0.88 The primary aim of this study was to assess effects of inclusion
Fiber (g) of high- or low-GI carbohydrate-rich foods on the outcomes of
Baseline visit 22.9 앐 1.1 23.3 앐 1.2 0.78 pregnancy. For this purpose the principal findings were indica-
Final visit 27.4 앐 1.5 28.0 앐 1.1 0.82 tors of fetal size. The infants of women eating the HGI diet tended
GI to be heavier at birth than the infants of women eating the LGI
Baseline visit 57 앐 1 59 앐 1 0.051 carbohydrates (P ҃ 0.051). Adjusted birth weight centiles were
Final visit 51 앐 1 58 앐 1 쏝0.001
used, and the infants of women in the HGI diet group were
P4 쏝0.001 0.62
significantly larger than those in the LGI diet group and had a
1
MUFA, monounsaturated fatty acid; PUFA, polyunsaturated fatty higher amount of body fat as estimated by the ponderal index.
acid; GI, glycemic index. Women who consumed the LGI diet did not have an increased
2
For the baseline visit, comparison was between groups. For the final number of infants who were either SGA or LGA, but infants in
visit, comparison was between the low-GI group and the high-GI group,
the HGI diet group were far more likely to be LGA (33.3% of
adjusted for differences at baseline.
3
x៮ 앐 SEM (all such values).
infants were 쏜90th centile) than in the LGI diet group (3.1% of
4
Within-group comparison (baseline compared with final visit). Com- infants). The calculation of birth centiles was made by using
parison of groups at baseline used independent samples t tests; comparison of internationally recognized centile calculators that are based on a
groups at final visit used independent samples t tests and ANOVA with British population. Although our local population is likely to be
baseline value as covariate. Comparisons from baseline to final visit used slightly different than the British population, this would not have
paired t tests. influenced the magnitude of the difference between the 2 groups.
In nonpregnant, healthy subjects, mixed meals based on
low-GI foods lead to a reduction in the postprandial glycemia
0.008). No difference was observed in the contribution of car- (16, 17), and this is likely to be relevant in pregnancy. For preg-
bohydrate to total energy in both groups (앒46%). The analysis of nant women there is a continuum of risk of adverse pregnancy
the diet histories produced similar findings (data not shown). outcomes related to increasing maternal glucose concentrations
The average GI for both groups was also determined by anal- after GTT (18). In the clinical context, women with GDM who
ysis of 24-h recalls at the intermediate visits 3 and 4. Women in have their management based on the 1-h rather than the 2-h
the LGI diet group had a significantly lower average GI at both postprandial glucose concentration have better obstetric out-
visits than did women in the HGI diet group (P 쏝 0.001 for both). comes (19). For pregnant women an association is recognized
At visits 3, 4, and 5, women in the LGI diet group had an average between the 1-h postprandial glucose concentration and fetal
GI 쏝 55, whereas at each of these visits women in the HGI diet adiposity as estimated by the fetal abdominal circumference
group had an average GI 쏜 55. (20). The higher ponderal index found in the infants of women
Responses in relation to diet acceptability are summarized in who consumed HGI carbohydrates may suggest an adverse in-
Table 4. No differences were observed in the ratings for both fluence of this kind of diet on fetal outcomes.
 LOW-GI DIET IN PREGNANCY
No significant differences were observed in obstetric out-
comes between the 2 groups with respect to the method of de-
livery or the rate of GDM (only 1 case). GDM is the most com-
mon medical problem during pregnancy, and in the Wollongong
area it has an incidence of 7.2% (21). Women who develop GDM
were found to consume fewer low-GI carbohydrate foods than
women who remain glucose tolerant (22). The current study was
not powered to look at the effect of a LGI diet on the incidence of
GDM. However, the favorable effects of a LGI diet on fetal
outcomes could lead to speculation about reducing the incidence
of GDM.
In 1998 Clapp (23) reported on a longitudinal study of 12
women who were recruited before pregnancy and followed
through to delivery. The women agreed to follow a diet that
provided 55– 60% carbohydrate. Initially, this diet was com-
posed of foods with a low GI but at 8 wk of gestation they were
randomly assigned to either continue with the low-GI diet or to
follow a high-GI diet for the duration of pregnancy. For women
on a high-GI diet, the glucose responses to a standard meal
progressively increased during pregnancy, whereas for women
who consumed the low-GI diet the glucose responses did not
change. Remarkably, the women who consumed the high-GI diet
all had a infant that was LGA and with a mean weight of 1000 g
more than the mean weight for women who consumed the low-GI
diet. The number of women in this study was relatively small, and
an exercise program was a confounding variable.
Scholl et al (10) took a different perspective. They proposed
that eating foods with a low GI may lead to lower maternal
glucose concentrations and less fuel for the fetus to develop a
normal pattern of growth. Women who consumed a low-GI diet
had a birth weight that was 쏜100 g lower and also had twice the
incidence of SGA infants. However, when dietary GL was con-
sidered, no association was observed with birth weight or the risk
of a SGA infant. The strength of this study was that it involved
쏜1000 women, but they were all from a particularly underpriv-
ileged area. Approximately 50% of the carbohydrate eaten came
from refined sugars, and, because sweet drinks and confectionery
often have a relatively low GI, it is possible that the lower birth
weights for women with the lowest GI diet may have also been
influenced by a poor-quality diet overall. In our study, the overall
quality of both diets was good, with food and nutritional intakes
in accordance with recommendations.
Secondary aims of the present study were to assess the toler-
ability and sustainability of a LGI diet. The 2 groups of women
were well matched for initial macronutrient intake. The GI was
similar at baseline and comparable to the median GI of American
diets (57 on the glucose scale) (24). Mean dietary GI fell signif-
icantly in women in LGI diet group at visits 3, 4, and 5 (the final
visit) compared with their baseline result and with the results of
women in the HGI diet group. This finding indicated that the LGI
diet was sustainable at least for the duration of the pregnancy.
Because both diets were similar in carbohydrate content (앒46%
energy or 앒250 g), only differences in the GI influenced dietary
GL (the product of GI ҂ amount of carbohydrate). Finally, al-
though the diets were virtually equal in terms of subjective ad-
herence, enjoyment, affordability, and family acceptability,
women who consumed the LGI diet were more likely to report
that their diet was easy to follow. This finding should be inter-
preted cautiously.
The strengths of this study include the study design, large
sample size, high continuation rate, high compliance, and the
811
detailed and repeated ascertainment of dietary measurements.
Together these factors increase the reliability and sensitivity of
the data. A particular strength was extensive knowledge of the GI
of the individual Australian foods (25). Importantly, in previous
studies we confirmed that sample menus representative of the 2
diets produced differential day-long glucose and insulin post-
prandial responses as predicted by their calculated GI (17). A
further strength was that the subjects were free-living women
who represent an important target for early intervention. A po-
tential weakness of the study may have been the alternate assign-
ment of women to one of the study groups.
The study had limitations. Although a similar macronutrient
intake was targeted in both groups, subjects in the LGI group
reduced their intake of saturated fat and the diet GI, whereas
subjects in the HGI group reduced their intake of polyunsaturated
Downloaded from https://academic.oup.com/ajcn/article/84/4/807/4633214 by guest on 12 April 2021
fat. These differences may reflect inherent characteristics of the
diets that might also operate outside the research setting. Lack of
blinding of subjects and investigators to diet assignment can
introduce a source of bias. Maternal weight gain and most mea-
sures of diet acceptance, however, were similar with both diets,
which suggested no overt bias toward any one diet.
In summary, infants of women instructed to consume low-GI
carbohydrate foods during pregnancy were of normal size but
were smaller and had less body fat than did the women whose
dietary GI did not change during pregnancy. Because birth
weight and ponderal index predict long-term risk of obesity and
chronic disease (26), a low-GI diet in pregnancy may favorably
influence long-term outcomes. Adequately powered studies of
healthy pregnant women are required to address the possibility
that consumption of low-GI carbohydrates during pregnancy
could reduce the risk of GDM.
RGM and JCB-M conceived and designed the study, analyzed and inter-
preted the findings, and contributed to the writing of the manuscript. LCT
contributed to the design and conduct of the study, the interpretation of the
findings, and the writing of the final manuscript. PP analyzed the findings and
contributed to their interpretation. ML, WSD, and KJC conducted the study
and contributed to the interpretation of the findings and the writing of the final
manuscript.
JCB-M is a coauthor of The Low GI Diet (New York, NY: Marlowe and
Co, 2005) and a coauthor of The New Glucose Revolution book series (New
York, NY: Marlowe and Co; Sydney, Australia). None of the other authors
had any potential conflict of interests relevant to the conduct of this research.
REFERENCES
1. Jenkins DJ, Wolever TM, Taylor RH, et al. Glycemic index of foods: a
physiological basis for carbohydrate exchange. Am J Clin Nutr 1981;
34:362– 6.
2. Willett W, Manson J, Liu S. Glycemic index, glycemic load, and risk of
type 2 diabetes. Am J Clin Nutr 2002;76:274S– 80S.
3. Pi-Sunyer FX. Glycemic index and disease. Am J Clin Nutr 2002;76:
290S– 8S.
4. Brand-Miller J, Thomas M, Swan V, Ahmad Z, Petocz P, Colagiuri S.
Physiological validation of the concept of glycemic load in lean young
adults. J Nutr 2003;133:2728 –32.
5. American Diabetes Association. Postprandial blood glucose. Diabetes
Care 2001;24:775– 8.
6. Combs C, Gunderson E, Kitzmiller J, et al. Relationship of fetal mac-
rosomia to maternal postprandial glucose control during pregnancy.
Diabetes Care 1992;15:1251–7.
7. Jovanovic-Peterson L, Peterson C, Reed G, et al. Maternal postprandial
glucose levels and infant birth weight: the Diabetes in Early Pregnancy
Study. Am J Obstet Gynecol 1991;164:103–11.
8. Lock D, Bar-Eyal A, Voet H, Madar Z. Glycemic indices of various
812 MOSES ET AL
foods given to pregnant diabetic subjects. Obstet Gynecol 1988;71:
180 –3.
9. Clapp J III. Diet, exercise, and feto-placental growth. Arch Gynecol
Obstet 1997;261:101–7.
10. Scholl T, Chen X, Khoo C, Lenders C. The dietary glycemic index during
pregnancy: influence on infant birth weight, fetal growth and biomarkers
of carbohydrate metabolism. Am J Epidemiol 2004;159:467–74.
11. Matthews DR, Hosker J, Redenski A, Naylor B, Treacher D, Turner R.
Homeostasis model assessment: insulin resistance and beta-cell function
from fasting plasma glucose and insulin concentrations in man. Diabe-
tologia 1985;28:412–9.
12. Wallace T, Levy J, Matthews DR. Use and abuse of HOMA modelling.
Diabetes Care 2004;27:1487–95.
13. Commonwealth Department of Health and Aged Care. Australian guide
to healthy eating. Canberra, Australia: Australian Government Publish-
ing Service, 1998.
14. University of Sydney. GI database. Version current 1 July 2005. Internet:
http://www.glycemicindex.com (accessed 15 August 2005).
15. Gestation Network. Centile calculator. Internet: http://www.gestation.
net (accessed 12 September 2005).
16. Wolever T, Yang M, Zeng X, Atkinson F, Brand-Miller J. Food glycemic
index, as given in glycemic index tables, is a significant determinant of
glycemic responses elicited by composite breakfast meals. Am J Clin
Nutr 2006;83:1306 –12.
17. McMillan-Price J, Petocz P, Atkinson F, et al. Comparison of 4 diets of
varying glycemic load on weight loss and cardiovascular risk reduction
in overweight and obese young adults: a randomised controlled trial.
Arch Intern Med 2006;166;1466 –75.
18. Moses R, Calvert D. Pregnancy outcomes in women without gestational
diabetes mellitus related to the maternal glucose level. Is there a contin-
uum of risk? Diabetes Care 1995;18:1527–33.
19. Moses R, Lucas E, Knights S. Gestational diabetes mellitus. At what
time should the postprandial glucose level be monitored? Aust N Z J
Obstet Gynaecol 1999;39:458 – 60.
20. Parretti E, Mecacci F, Papini M, et al. Third-trimester maternal glucose
levels from diurnal profiles in non-diabetic pregnancies: correlation with
sonographic parameters of fetal growth. Diabetes Care 2001;24:1319 –
23.
21. Moses R, Griffiths R, McPherson S. The incidence of gestational dia-
betes in the Illawarra area of New South Wales. Aust N Z J Obstet
Gynaecol 1994;34:425–7.
22. Gillen L, Tapsell L, Martin G, Daniells S, Knights S, Moses R. The type
and frequency of consumption of carbohydrate rich foods may play a role
in the clinical expression of insulin resistance during pregnancy. Nutr
Diet 59;59:135-43.
23. Clapp J III. Effect of dietary carbohydrate on the glucose and insulin
Downloaded from https://academic.oup.com/ajcn/article/84/4/807/4633214 by guest on 12 April 2021
response to mixed caloric intake and exercise in both nonpregnant and
pregnant women. Diabetes Care 1998;21(suppl):107–12.
24. Ma Y, Olendzki B, Chiriboga D, et al. Association between dietary
carbohydrates and body weight. Am J Epidemiol 2005;161:359 – 67.
25. Foster-Powell K, Holt SH, Brand-Miller JC. International table of gly-
cemic index and glycemic load values: 2002. Am J Clin Nutr 2002;76:
5–56.
26. Krassas G, Tzotzas T. Do obese children become obese adults: childhood
predictors of adult disease. Pediatr Endocrinol Rev 2004;1:455–9.