Hyperglycemia and Adverse Pregnancy Outcome Study: Neonatal Glycemia

Boyd E. Metzger, Bengt Persson, Lynn P. Lowe, Alan R. Dyer, J. Kennedy

Cruickshank, Chaicharn Deerochanawong, Henry L. Halliday, Anselm J. Hennis,
Helen Liley, Pak C. Ng, Donald R. Coustan, David R. Hadden, Moshe Hod, Jeremy J.
N. Oats, Elisabeth R. Trimble and for the HAPO Study Cooperative Research Group
Pediatrics 2010;126;e1545; originally published online November 15, 2010;
DOI: 10.1542/peds.2009-2257

The online version of this article, along with updated information and services, is
located on the World Wide Web at:
http://pediatrics.aappublications.org/content/126/6/e1545.full.html

PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

publication, it has been published continuously since 1948. PEDIATRICS is owned,
published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point
Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2010 by the American Academy
of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.

Downloaded from pediatrics.aappublications.org by guest on March 29, 2012

 ARTICLES

Hyperglycemia and Adverse Pregnancy Outcome

Study: Neonatal Glycemia
AUTHORS: Boyd E. Metzger, MD,a Bengt Persson, MD, WHAT’S KNOWN ON THIS SUBJECT: Newborn infants of mothers
PhD,b Lynn P. Lowe, PhD,c Alan R. Dyer, PhD,c J. Kennedy with preexisting overt diabetes mellitus are at risk for
Cruickshank, MD,d Chaicharn Deerochanawong, MD,e hypoglycemia that is associated with fetal hyperinsulinemia.
Henry L. Halliday, MD,f Anselm J. Hennis, MB, BS, PhD,g
Helen Liley, MB, ChB,h Pak C. Ng, MD,i Donald R. Coustan, WHAT THIS STUDY ADDS: For newborn infants of mothers who
MD,j David R. Hadden, MD,k Moshe Hod, MD,l Jeremy J. N.
do not have overt diabetes, neonatal glucose concentrations are
Oats, DM,m and Elisabeth R. Trimble, MD,n for the HAPO
Study Cooperative Research Group
associated with cord blood C-peptide levels but are relatively
stable during the first 5 hours of life.
Departments of aMedicine and cPreventive Medicine, Feinberg
School of Medicine, Northwestern University, Chicago, Illinois;
bDepartment of Pediatrics, Karolinska Institute, Stockholm,

Sweden; dDepartment of Medicine, Central Manchester

University Hospitals and St Mary’s Hospital, University of
Manchester, Manchester, United Kingdom; eDepartment of
Endocrinology, Rajavithi Hospital, Bangkok, Thailand;
abstract +

Departments of fChild Health and nEndocrinology, Queen’s
University, Belfast, United Kingdom; gChronic Disease Research
Centre, University of the West Indies, Bridgetown, Barbados;
hDepartment of Pediatrics, Mater Misericordiae Mothers’
Hospital, University of Queensland, Brisbane, Australia;
iDepartment of Pediatrics, Prince of Wales Hospital, Chinese
OBJECTIVE: The goal was to describe the temporal pattern of neonatal
plasma glucose levels and associations with maternal glucose levels,
cord serum C-peptide levels, and neonatal size and adiposity.
METHODS: A total of 17 094 mothers and infants were included in the

University of Hong Kong, Hong Kong; jDepartment of Obstetrics
and Gynecology, Women and Infants Hospital of Rhode Island
and Warren Alpert Medical School, Brown University,
Providence, Rhode Island; kDepartment of Endocrinology, Royal
Victoria Hospital, Belfast, United Kingdom; lDepartment of
Obstetrics and Gynecology, Helen Schneider Hospital for Women,
Rabin Medical Center, Sackler Faculty of Medicine, Tel Aviv
University, Tel Aviv, Israel; and mDepartment of Obstetrics and
Gynecology, Royal Women’s Hospital, University of Melbourne,
Melbourne, Australia
KEY WORDS
size at birth, neonatal hypoglycemia, cord C-peptide levels,
maternal glucose levels
ABBREVIATIONS
OGTT—oral glucose tolerance test
HAPO—Hyperglycemia and Adverse Pregnancy Outcome
OR—odds ratio
CI—confidence interval
www.pediatrics.org/cgi/doi/doi:10.1542/peds.2009-2257
doi:doi:10.1542/peds.2009-2257
Accepted for publication Aug 11, 2010
Address correspondence to Boyd E. Metzger, MD, Northwestern
University Feinberg School of Medicine, Endocrinology, 645 N
Michigan Ave, Suite 530-22, Chicago, IL 60611. E-mail: bem@
northwestern.edu
PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
Copyright © 2010 by the American Academy of Pediatrics
FINANCIAL DISCLOSURE: The authors have indicated they have
no financial relationships relevant to this article to disclose.
Funded by the National Institutes of Health (NIH).
Hyperglycemia and Adverse Pregnancy Outcome Study (15 centers in 9
countries). Mothers underwent a 75-g, 2-hour, oral glucose tolerance
test (OGTT) at 24 to 32 weeks of gestation. Cord blood and neonatal
blood samples were collected. Biochemical neonatal hypoglycemia
was defined as glucose levels of ⬍10th percentile (2.2 mmol/L). Clini-
cally identified hypoglycemia was ascertained through medical record
review and associations were assessed.
RESULTS: Plasma glucose concentrations were stable during the first
5 hours after birth. Maternal glucose levels were weakly positively
associated with biochemical neonatal hypoglycemia (odds ratios:
1.07–1.14 for 1-SD higher OGTT glucose levels). Frequency of neonatal
hypoglycemia was higher with higher cord C-peptide levels (odds ratio:
11.6 for highest versus lowest C-peptide category). Larger and/or fat-
ter infants were more likely to have hypoglycemia (P ⬍ .001), and
infants with hypoglycemia tended to have a higher frequency of cord
C-peptide levels of ⬎90th percentile.
CONCLUSIONS: Mean neonatal plasma glucose concentrations varied
little in the first 5 hours after birth, which suggests normal postnatal
adjustment. Biochemical and clinical hypoglycemia were weakly re-
lated to maternal OGTT glucose measurements but were strongly
associated with elevated cord serum C-peptide levels. Larger and/or
fatter infants were more likely to develop hypoglycemia and hyperin-
sulinemia. These relationships suggest physiologic relationships be-
tween maternal glycemia and fetal insulin production. Pediatrics 2010;
126:e1545–e1552

PEDIATRICS Volume 126, Number 6, December 2010 e1545

Downloaded from pediatrics.aappublications.org by guest on March 29, 2012
The objective of the Hyperglycemia and
Adverse Pregnancy Outcome (HAPO)
Study was to clarify the risk of adverse
outcomes associated with degrees of
glucose intolerance during pregnancy
less severe than overt diabetes melli-
tus. Increased size at birth, delivery
through cesarean section, clinical neo-
natal hypoglycemia, and fetal hyperin-
sulinemia were the predefined pri-
mary outcomes, and continuous
relationships of maternal glucose lev-
els below those diagnostic of diabetes
with each of these outcomes were re-
ported.1 These observations provide
strong support for and extend the ma-
ternal hyperglycemia-fetal hyperinsu-
linemia hypothesis proposed by Ped-
ersen2 in 1952.
The primary objective of this report is
to describe the temporal pattern of
neonatal plasma glucose concentra-
tions in the first 5 hours after birth for
⬎17 000 infants of mothers with glu-
cose levels below those that are diag-
nostic of diabetes. We also examine the
association of neonatal hypoglycemia
with maternal glucose levels, fetal in-
sulin levels (cord C-peptide concentra-
tions), and newborn size and adiposity.
METHODS
Approval and Consent
The institutional review boards at all
15 field centers approved the study
protocol. All participants provided
written informed consent. An external
data monitoring committee provided
oversight. Study methods were re-
ported previously.1,3,4 A brief overview
is presented here.
Participants
All pregnant women at each field cen-
ter were eligible to participate unless
they had ⱖ1 exclusion criterion, that
is, age of ⬍18 years, planned delivery
at another hospital, date of last men-
strual period uncertain and no ultra-
sound estimation from 6 to 24 weeks of
e1546 METZGER et al
Downloaded from pediatrics.aappublications.org by guest on March 29, 2012
gestation available, unable to com-
plete the oral glucose tolerance test
(OGTT) by 32 weeks of gestation,
multiple-fetus pregnancy, conception
through gonadotropin ovulation induc-
tion or in vitro fertilization, glucose
testing before recruitment, diagnosis
of diabetes during the current preg-
nancy, diabetes predating the preg-
nancy treated with medication, partic-
ipation in another study that might
interfere with the HAPO Study, known
HIV positivity or hepatitis B or C infec-
tion, previous participation in the HAPO
Study, or inability to converse in the
languages used for field center forms
without the aid of an interpreter.1 Ges-
tational age and expected date of deliv-
ery were determined from the date of
the last menstrual period if the partic-
ipant was certain of the date. If the
participant was uncertain, then the ex-
pected date of delivery was deter-
mined from an ultrasound scan per-
formed between 6 and 24 weeks of
gestation. The final expected date of
delivery also was determined from ul-
trasound scans if (1) gestational dat-
ing from the last menstrual period dif-
fered from ultrasound dating by ⬎5
days in cases in which ultrasonogra-
phy was performed between 6 and 13
weeks or (2) dating differed by ⬎10
days in cases in which ultrasonogra-
phy was performed between 14 and 24
weeks.
Oral Glucose Tolerance Test
All participants underwent a standard
75-g OGTT between 24 and 32 weeks of
gestation (as close to 28 weeks as pos-
sible). Height, weight, and blood pres-
sure were measured at the OGTT visit
by using standardized procedures and
calibrated equipment. Personal and
demographic data were collected by
using standardized questionnaires.
Race/ethnicity was self-identified. A
sample for random plasma glucose
measurement was collected at 34 to 37
weeks of gestation, as a safety mea-
sure to identify patients with hypergly-
cemia above a predefined threshold.
Unblinding
To facilitate timely clinical decisions,
fasting and 2-hour OGTT and random
plasma glucose samples were ana-
lyzed enzymatically at field center lab-
oratories.4 Values were unblinded if
fasting plasma glucose levels were
⬎5.8 mmol/L, 2-hour plasma glucose
levels were ⬎11.1 mmol/L, random
plasma glucose levels were ⱖ8.9
mmol/L, or any plasma glucose level
was ⬍2.5 mmol/L. Otherwise, women,
caregivers, and HAPO Study staff mem-
bers (except laboratory personnel) re-
mained blinded to glucose values. To
avoid center-to-center analytical varia-
tions, aliquots of all OGTT specimens
and neonatal plasma glucose samples
collected for research purposes were
analyzed at the HAPO Study central
laboratory.4
Prenatal Care, Delivery, and
Neonatal Care
Prenatal care, timing of delivery, and
neonatal care, including clinical and
laboratory assessments for neonatal
hypoglycemia, were determined ac-
cording to standard practices at each
center, because participants were not
classified as being at high risk. No field
center arbitrarily delivered infants be-
fore term or routinely performed ce-
sarean delivery at a specified maternal
or gestational age.
According to protocol, non– glucose-
containing fluids were used for oxyto-
cin infusion if labor was induced.
Glucose-containing solutions were not
used during labor or delivery unless
duration exceeded 8 hours or there
was a medical indication. If oral intake
of clear liquids was permitted during
labor, then participants were given
glucose/sucrose-free liquids. Mothers
were permitted to feed their infants
soon after delivery, if desired, because

a pilot study showed no difference in
plasma glucose concentrations ac-
cording to feeding status.5 Neonatal
anthropometric measurements were
obtained by using standardized meth-
ods.6 Medical records were abstracted
to obtain data regarding the prenatal,
labor and delivery, postpartum, and
newborn courses.
Cord Serum C-peptide and Plasma
Glucose Samples
Cord blood samples for assessment of
fetal ␤-cell function (C-peptide levels)
and measurement of glucose levels
were collected as soon as possible af-
ter clamping of the cord. Cord serum
levels of C-peptide (secreted in
equimolar concentrations with insu-
lin) were used as the index of fetal
␤-cell function, rather than insulin lev-
els, because degradation of insulin is
increased in the presence of even
slight hemolysis, ⬃15% of cord sam-
ples showed detectable hemolysis,
and the concentrations of C-peptide
are not altered by hemolysis.7
The C-peptide samples were collected
into plain phlebotomy tubes. The glucose
samples (in sodium fluoride-containing
tubes) were placed in ice water, trans-
ported to the laboratory, and separated
within 60 minutes after collection (me-
dian separation time: 45 minutes). Ali-
quots were frozen and analyzed later at
the central laboratory by using a Vitros
750 analyzer (Ortho Clinical Diagnostics,
Rochester, NY) for glucose and an Au-
todelfia instrument (Perkin Elmer, Bos-
ton, MA) for C-peptide.4 The Wales Exter-
nal Quality Assessment Scheme used by
the central laboratory ensured accept-
able accuracy of glucose measurements
at 2.0 mmol/L of ⫾6%. The functional
sensitivity of the C-peptide assay was 0.2
␮g/L.4
Neonatal Glucose Samples
At 1 to 2 hours after birth, capillary
blood was collected, through heel
PEDIATRICS Volume 126, Number 6, December 2010
Downloaded from pediatrics.aappublications.org by guest on March 29, 2012
prick, into a microtube containing so-
dium fluoride, immediately placed in
ice water, transferred to the labora-
tory, centrifuged and separated within
60 minutes (median separation time:
30 minutes), and frozen for sub-
sequent glucose analysis at the cen-
tral laboratory. For determination of
whether neonatal plasma glucose con-
centrations decrease progressively in
healthy newborns, rather than show-
ing an early postdelivery plateau, 5% of
infants were selected randomly before
delivery to have a second sample
drawn 4 hours after birth.
Outcomes
Biochemical Neonatal Hypoglycemia
Central laboratory–measured plas-
ma glucose concentrations of ⬍2.2
mmol/L, representing the 10th percen-
tile of the 17 094 HAPO Study values,
were used to define biochemical hypo-
glycemia. The 5th percentile was 1.9
mmol/L.
Clinical Neonatal Hypoglycemia
This was defined by ⱖ1 clinical crite-
rion, that is, a notation of neonatal hy-
poglycemia in the medical record and
symptoms or treatment with glucose
infusion or laboratory-reported glu-
cose levels of ⱕ1.7 mmol/L in the first
24 hours after birth or ⱕ2.5 mmol/L
after the first 24 hours.8 Meter mea-
surements of glucose were not in-
cluded in this determination. Data on
clinical hypoglycemia were available
for 23 227 infants (excluding fetal
deaths).
Birth Weight of ⬎90th Percentile
The 90th percentile values for gesta-
tional age (30 – 44 weeks) were deter-
mined for 8 newborn gender/ethnicity
groups (white, other, black, Hispanic,
or Asian), with adjustment for gesta-
tional age, field center, and parity (0, 1,
or ⱖ2), by using quantile regression.
ARTICLES
Percent Body Fat of ⬎90th Percentile
Fat mass was calculated from birth
weight, length, and flank skinfold thick-
ness, according to the equation used
by Catalano et al.9 Percent body fat was
then calculated as 100 ⫻ fat mass/
birth weight. Percent body fat of ⬎90th
percentile for gestational age (36 – 44
weeks only) was defined by using the
same methods as described above for
birth weight of ⬎90th percentile.
Birth Weight of ⬍10th Percentile
The 10th percentiles were calculated
by using similar methods as for birth
weight of ⬎90th percentile.
Cord C-peptide Levels of ⬎90th
Percentile
The 90th percentile for the total HAPO
Study cohort (1.7 ␮g/L) was used.
Statistical Analyses
Descriptive statistics included means
and SDs for continuous variables and
numbers and proportions for categor-
ical variables. For analyses of associa-
tions of maternal glycemia with bio-
chemical hypoglycemia and cord
serum C-peptide levels with both bio-
chemical and clinical hypoglycemia,
glucose and cord C-peptide measure-
ments were considered as both cate-
gorical and continuous variables in
multivariate logistic regression analy-
ses. In categorical analyses, each mea-
sure of glycemia and cord C-peptide
levels was divided into 7 categories.1,6
as indicated in Tables 2 and 3.
For glucose measurements as a con-
tinuous variable, odds ratios (ORs)
were calculated for each plasma glu-
cose measurement higher by 1 SD.
Squared terms were added to assess
whether the logarithm of the odds of
biochemical hypoglycemia was lin-
early related to glucose levels. Interac-
tions of glucose measurements as a
continuous variable with field center,
BMI, age, height, and mean arterial
e1547
pressure also were examined. Be-
cause of the large sample size in the
HAPO Study and the large number of
infants with biochemical hypoglyce-
mia, squared terms and interaction
terms were considered statistically
significant for this outcome only for P
⬍ .001. Cord C-peptide levels also were
examined as a continuous variable
with both linear and squared terms for
each outcome. Because the squared
term was significant for both biochem-
ical (P ⬍ .001) and clinical (P ⬍ .05)
hypoglycemia, only results of categor-
ical analyses are reported.
Two logistic models (models I and II)
were fit for each outcome, with model I
including adjustment for field center
(data not shown). Model II included ad-
ditional adjustment for multiple pre-
specified potential confounders, in-
cluding maternal age, BMI, height,
gestational age and mean arterial pres-
sure at the OGTT, hospitalization before
delivery, smoking status, and alcohol use
and family history of diabetes.
The ␹2 test was used for comparisons
of proportions and the 2-sample t test
for comparisons of mean cord
C-peptide levels for subjects with and
without hypoglycemia, in subgroups
defined according to birth weight and
percent body fat. All analyses were
conducted with SAS 9.1 (SAS Institute,
Cary, NC) or Stata 10.0 (Stata Corp, Col-
lege Station, TX).
RESULTS
Pregnancies with glucose values
blinded, birth weight recorded, gesta-
tional age determined, and obstetric
and neonatal records available (N ⫽
23 316) were included in the initial re-
port.1 Results presented in Table 1 are
for the 17 094 of those pregnancies
with central laboratory– determined,
1- to 2-hour neonatal plasma glucose
levels. Maternal and neonatal charac-
teristics of the participants in the en-
e1548 METZGER et al
Downloaded from pediatrics.aappublications.org by guest on March 29, 2012
TABLE 1 Characteristics of Study Participants With Measurements of Neonatal Glucose Levels
No. of Participants Finding
Maternal characteristics
Age, mean ⫾ SD, y 17 094 29.1 ⫾ 5.8
BMI, mean ⫾ SD, kg/m2a 17 094 27.5 ⫾ 5.0
Mean arterial pressure, mean ⫾ SD, mm Hga 17 094 80.5 ⫾ 8.2
Plasma glucose level, mean ⫾ SD, mmol/La
Fasting 17 094 4.5 ⫾ 0.4
1h 17 094 7.5 ⫾ 1.7
2h 17 094 6.2 ⫾ 1.3
Neonatal characteristics
Gestational age at delivery, mean ⫾ SD, wk 17 094 39.4 ⫾ 1.6
Birth weight, mean ⫾ SD, g 17 086 3298 ⫾ 494
Cord serum C-peptide level, mean ⫾ SD, ␮g/L 16 429 1.00 ⫾ 0.59
Cord plasma glucose level, mean ⫾ SD, mmol/L 16 416 4.5 ⫾ 1.1
First neonatal glucose level, mean ⫾ SD, mmol/L 3.2 ⫾ 0.9
Second neonatal glucose level, mean ⫾ SD, mmol/L
17 094
574 3.3 ⫾ 0.8
Obstetric outcomes, n (%)
Primary cesarean sectionb 15 285 2557 (16.7)
Preeclampsiab 17 094 713 (4.2)
Newborn outcomes, n (%)
Birth weight of ⬎90th percentile 17 086 1647 (9.6)
Percent body fat of ⬎90th percentile 14 823 1454 (9.8)
Birth weight of ⬍10th percentile 17 086 1618 (9.5)
Cord C-peptide level of ⬎90th percentile 16 429 1346 (8.2)
Biochemical neonatal hypoglycemia 17 094 1554 (9.1)
Clinical neonatal hypoglycemia 17 094 313 (1.8)
Premature deliveryb 17 094 1007 (5.9)
Shoulder dystocia/birth injury 17 094 203 (1.2)
Hyperbilirubinemia 17 094 1401 (8.2)
Intensive neonatal careb 17 094 1213 (7.1)
aMaternal BMI, mean arterial pressure, and plasma glucose level were measured at the OGTT visit.
bThese outcomes were found more frequently in the subset of neonates for whom HAPO Study samples for neonatal glucose
measurements were not collected (P ⬍ .001).
tire cohort1 and of those described 15-minute intervals of collection time
here were similar (Table 1). (45–180 minutes) after delivery. For a
By using the protocol for intravenous subset of 574 randomly selected neo-
fluid administration during labor and nates, a second heel-stick sample was
delivery described above, glucose was collected. Second-sample collection
administered in only 7% of cases. Glu- times were divided into 5 periods.
Forty-nine samples were collected be-
cose concentrations in plasma from
tween 120 and 180 minutes after deliv-
mixed arterial and venous cord blood
ery. Times for the other samples are
at birth and from heel-stick capillary
blood samples from neonates are presented in 30-minute intervals from
181 to 300 minutes after delivery. Over-
shown in Fig 1. Cord blood was col-
lected as soon after birth as possible all, plasma glucose concentrations
(median time: 2.0 minutes [interquar- varied by no more than 0.2 mmol/L dur-
tile range: 1.0 –5.0 minutes]). The ing the first 5 hours after delivery. Simi-
mean ⫾ SD cord plasma glucose con- lar results were noted in comparison of
centration was 4.5 ⫾ 1.1 mmol/L. values for the first and second samples
Among the 16 613 neonates with a de- for the subgroup of 544 neonates for
livery time recorded, 430 had the first whom 2 samples were collected by de-
sample collected at ⱕ45 minutes sign and for whom times from delivery
(mean plasma glucose concentration: were available (data not shown).
3.3 mmol/L). Values for mean plasma We found a modest positive correla-
glucose concentrations are plotted at tion between cord plasma glucose con-
 ARTICLES

FIGURE 1
Temporal patterns of cord plasma glucose and neonatal plasma glucose (NG) concentrations.

centrations and neonatal plasma glu- Table 2 shows the unadjusted frequen- poglycemia tended to increase with in-
cose concentrations (r ⫽ 0.256; P ⬍ cies and fully adjusted model II ORs and creasing maternal glucose levels.
.001). As indicated in Table 1, cord 95% confidence intervals (CIs) for as- Adjusted ORs differed significantly
plasma glucose concentrations gener- sociations of maternal OGTT glucose from 1.0 only for the highest category
ally were not increased (4.5 ⫾ 1.1 measurements with biochemical neo- of fasting plasma glucose levels, the
mmol/L). natal hypoglycemia. Frequency of hy- second highest category of 1-hour
plasma glucose levels, and the 5 high-
TABLE 2 Relationship Between Maternal OGTT Glucose Measurements and Biochemical Neonatal est categories of 2-hour plasma glu-
Hypoglycemia cose levels. For glucose levels as a con-
N n (%) OR (95% CI)a tinuous variable in model II, there were
Fasting plasma glucose level weak but positive associations with
⬍4.2 mmol/L 3103 254 (8.2) 1.00
biochemical hypoglycemia (ORs of
4.2–4.4 mmol/L 5555 490 (8.8) 1.06 (0.90–1.24)
4.5–4.7 mmol/L 4520 386 (8.5) 1.01 (0.85–1.19) 1.07–1.14 for glucose levels higher by 1
4.8–4.9 mmol/L 1960 200 (10.2) 1.17 (0.95–1.44) SD). We previously reported relatively
5.0–5.2 mmol/L 1339 141 (10.5) 1.12 (0.89–1.41)
weak associations of OGTT glucose
5.3–5.5 mmol/L 480 54 (11.3) 1.11 (0.80–1.54)
ⱖ5.6 mmol/L 137 29 (21.2) 2.68 (1.71–4.20) measurements with clinical neonatal
Continuous variableb 17 094 1554 (9.1) 1.07 (1.01–1.14) hypoglycemia.1
1-h plasma glucose level
⬍5.8 mmol/L 3102 273 (8.8) 1.00 Frequency of biochemical neonatal hy-
5.9–7.3 mmol/L 5490 451 (8.2) 0.90 (0.77–1.06) poglycemia increased progressively
7.4–8.6 mmol/L 4406 404 (9.2) 0.99 (0.84–1.17) from 5.5% for the lowest category of
8.7–9.5 mmol/L 2042 196 (9.6) 1.05 (0.85–1.28)
9.6–10.7 mmol/L 1377 142 (10.3) 1.12 (0.89–1.40) C-peptide levels (ⱕ0.5 ␮g/L) to 36.9%
10.8–11.7 mmol/L 493 66 (13.4) 1.47 (1.09–1.99) for the highest category (ⱖ3.1 ␮g/L)
ⱖ11.8 mmol/L 184 22 (12.0) 1.26 (0.78–2.03) (Table 3). ORs for biochemical hypogly-
Continuous variableb 17 094 1554 (9.1) 1.07 (1.01–1.13)
2-h plasma glucose level cemia were higher in each category,
⬍5.0 mmol/L 3108 245 (7.9) 1.00 compared with those for clinical hypo-
5.1–6.0 mmol/L 5436 437 (8.0) 1.03 (0.87–1.22) glycemia, and reached 11.61 in the
6.1–6.9 mmol/L 4352 417 (9.6) 1.24 (1.04–1.47)
7.0–7.7 mmol/L 2231 221 (9.9) 1.27 (1.04–1.55) highest category. ORs for clinical hypo-
7.8–8.7 mmol/L 1281 142 (11.1) 1.46 (1.16–1.84) glycemia were increased in the high-
8.8–9.8 mmol/L 513 69 (13.5) 1.74 (1.29–2.34) est 3 categories and reached 5.26.
ⱖ9.9 mmol/L 173 23 (13.3) 1.74 (1.08–2.80)
Continuous variableb 17 094 1554 (9.1) 1.14 (1.07–1.20) Table 4 shows the association of neo-
a Adjusted (model II) for field center, maternal age, BMI, mean arterial pressure, height, gestational age at the OGTT, parity, natal anthropometric outcomes with
smoking, alcohol use, and hospitalization before delivery, family history of diabetes mellitus, and infant’s gender.
b OR for a 1-SD increase in the glucose level (0.4 mmol/L for fasting plasma glucose levels, 1.7 mmol/L for 1-hour plasma
the frequency of biochemical and clin-
glucose levels, and 1.3 mmol/L for 2-hour plasma glucose levels). ical neonatal hypoglycemia. When in-

PEDIATRICS Volume 126, Number 6, December 2010 e1549

Downloaded from pediatrics.aappublications.org by guest on March 29, 2012
TABLE 3 Relationship Between Cord Serum C-Peptide Levels and Neonatal Hypoglycemia fants with values in ⱕ90th percentile
Cord Serum C-Peptide Level N n (%) OR (95% CI)a (P ⬍ .001). The frequencies of clinical
Biochemical neonatal hypoglycemiab hypoglycemia differed significantly be-
ⱕ0.5 ␮g/L 2409 133 (5.5) 1.00
tween small-for-gestational age neo-
0.6–0.8 ␮g/L 5454 317 (5.8) 1.08 (0.88–1.34)
0.9–1.2 ␮g/L 4875 448 (9.2) 1.77 (1.44–2.17) nates (birth weights in ⬍10th percen-
1.3–1.5 ␮g/L 1701 223 (13.1) 2.63 (2.09–3.30) tile for gestational age) and those with
1.6–2.1 ␮g/L 1344 190 (14.1) 2.97 (2.33–3.77) birth weights in ⱖ10th percentile;
2.2–3.0 ␮g/L 459 109 (23.7) 5.86 (4.40–7.81)
ⱖ3.1 ␮g/L 187 69 (36.9) 11.61 (8.12–16.60) however, there was no difference in
Clinical neonatal hypoglycemiac the frequencies of biochemical hypo-
ⱕ0.5 ␮g/L 2915 60 (2.1) 1.00 glycemia between these 2 groups.
0.6–0.8 ␮g/L 6537 87 (1.3) 0.70 (0.50–0.98)
0.9–1.2 ␮g/L 5900 88 (1.5) 0.82 (0.59–1.16) Table 5 shows the frequency of ele-
1.3–1.5 ␮g/L 2077 42 (2.0) 1.16 (0.77–1.75) vated cord serum C-peptide levels and
1.6–2.1 ␮g/L 1639 58 (3.5) 1.97 (1.35–2.89)
2.2–3.0 ␮g/L 571 25 (4.4) 2.50 (1.52–4.11)
mean cord C-peptide concentrations
ⱖ3.1 ␮g/L 244 18 (7.4) 5.26 (2.96–9.34) for infants with anthropometric out-
a Adjusted (model II) for field center, maternal age, BMI, mean arterial pressure, mean arterial pressure squared, height, comes, with or without biochemical or
gestational age at the OGTT, parity, smoking, alcohol use, hospitalization before delivery, family history of diabetes mellitus,
and infant’s gender.
clinical neonatal hypoglycemia. For
b Biochemical neonatal hypoglycemia indicates neonatal glucose concentrations of ⬍10th percentile for the study popula- each neonatal anthropometric pheno-
tion (⬍2.2 mmol/L).
c Clinical neonatal hypoglycemia indicates that there was a notation of neonatal hypoglycemia in the medical record and
type, infants with hypoglycemia tended
symptoms or treatment with a glucose infusion or there was a laboratory report of a glucose level of ⱕ1.7 mmol/L in the to have a higher frequency of cord
first 24 hours after birth or ⱖ2.5 mmol/L after the first 24 hours. C-peptide levels in ⬎90th percentile
and a higher mean cord C-peptide con-
TABLE 4 Association of Neonatal Hypoglycemia With Neonatal Anthropometric Measurements centration, compared with those with-
n/N (%) out hypoglycemia.
Biochemical Neonatal Clinical Neonatal
Hypoglycemia Hypoglycemia DISCUSSION
Birth weight Neonatal hypoglycemia in association
⬎90th percentile 202/1647 (12.3) 75/2219 (3.4)
ⱕ90th percentile 1349/15 439 (8.7) 394/20 963 (1.9) with fetal hyperinsulinemia repre-
P ⬍ .001 P ⬍ .001 sents the crux of the Pedersen hypoth-
⬍10th percentile 162/1618 (10.0) 102/2253 (4.5) esis,2 and the concept has been vali-
ⱖ10th percentile 1389/15 468 (9.0) 367/20 929 (1.8)
P ⫽ .169 P ⬍ .001 dated repeatedly for offspring of
Percent body fat mothers with preexisting diabetes10 or
⬎90th percentile 176/1454 (12.1) 39/1892 (2.1) gestational diabetes.11 Our data indi-
ⱕ90th percentile 1081/13 369 (8.1) 208/17 435 (1.2)
P ⬍ .001 P ⫽ .001 cate that the relationship spans the
range of maternal glycemia below the
level of overt diabetes mellitus. Also, in
fants were larger or fatter (birth age), both biochemical and clinical the newborn period, after the initial de-
weight or percent of newborn body fat neonatal hypoglycemia occurred with crease in glucose concentrations that
in ⬎90th percentile for gestational greater frequency, compared with in- follows the cessation of mother-fetus de-

TABLE 5 Neonatal Anthropometric Measurements, Neonatal Hypoglycemia, and Cord Serum C-Peptide Levels
Outcome Biochemical Neonatal Hypoglycemia Clinical Neonatal Hypoglycemia
Present Absent P Present Absent P
Birth weight of ⬎90th percentile
C-peptide level of ⬎90th percentile, n/N (%) 77/199 (38.7) 235/1401 (16.8) ⬍.001 19/60 (31.7) 365/1895 (19.3) .017
C-peptide concentration, mean, ␮g/L 1.77 1.23 ⬍.001 1.68 1.30 .017
Percent body fat of ⬎90th percentile
C-peptide level of ⬎90th percentile, n/N (%) 57/169 (33.7) 198/1248 (15.9) ⬍.001 11/33 (33.3) 306/1678 (18.2) .027
C-peptide concentration, mean, ␮g/L 1.69 1.21 ⬍.001 1.58 1.28 .017
Birth weight of ⬍10th percentile
C-peptide level of ⬎90th percentile, n/N (%) 14/148 (9.5) 35/1371 (2.6) ⬍.001 8/79 (10.1) 54/1752 (3.1) .001
C-peptide concentration, mean, ␮g/L 1.06 0.79 .001 0.94 0.81 .095

e1550 METZGER et al
Downloaded from pediatrics.aappublications.org by guest on March 29, 2012

livery of glucose, plasma glucose con-
centrations remain relatively stable.
These data confirm that the vast majority
of infants born of healthy mothers who
do not have overt diabetes mellitus suc-
cessfully make the early transition to the
metabolism of endogenously derived nu-
trient substrates.
On the basis of comparisons for a rel-
atively small number of infants of nor-
mal mothers or mothers with type 1
diabetes mellitus, Pedersen12 con-
cluded many years ago that most in-
fants achieve stable glucose concen-
trations within a few hours after birth
and remain free of hypoglycemia. Our
results from measurements of plasma
glucose levels obtained through heel
stick in the first several hours after
birth from a large cohort support
those suggestions convincingly.
For women with preexisting diabetes,
efforts generally are made to avoid
maternal hyperglycemia during labor
and delivery, through concern that hy-
perglycemia might stimulate fetal in-
sulin secretion, which in turn might in-
crease the risk of hypoglycemia in the
early neonatal period.13 HAPO Study par-
ticipants did not have overt diabetes,
they did not receive intravenous glucose
administration during labor in 93% of
cases, and they generally had normal
cord plasma glucose concentrations at
delivery. In this setting, we found a posi-
tive correlation between cord glucose
concentrations and 1- to 2-hour neonatal
glucose concentrations.
Findings of higher frequencies of bio-
chemical and clinical hypoglycemia
among the larger and fatter newborns
who also had higher cord serum
C-peptide levels are consistent with the
Pedersen hypothesis and the putative
role of fetal hyperinsulinemia. It is
known that the risk of neonatal hypogly-
cemia is increased for infants born
small for gestational age,14 but it is
thought that factors other than hyperin-
sulinemia account for the disturbance in
PEDIATRICS Volume 126, Number 6, December 2010
Downloaded from pediatrics.aappublications.org by guest on March 29, 2012
glucose homeostasis. Our data suggest
that this concept may require further
evaluation. We did find a lower mean
cord serum C-peptide concentration for
infants with birth weights of ⬍10th per-
centile; however, among those infants
we also found a higher C-peptide concen-
tration and a higher rate of C-peptide val-
ues of ⬎90th percentile for infants with
hypoglycemia, compared with infants
who did not develop biochemical or clin-
ical hypoglycemia.
An important strength of the HAPO Study
is derived from the rigorous protocol
that was followed to minimize glycolysis
of the neonatal glucose samples. The
⬎17 000 glucose measurements that
were performed for infants of mothers
without overt diabetes mellitus repre-
sent a major resource that, in the future,
might be used to derive a consensus def-
inition of hypoglycemia in early postnatal
life, regarding which there is longstand-
ing debate.15–19
There also are some potential limita-
tions. We could not justify extending the
protocol for repeated collection of blood
samples for glucose measurements be-
yond the early neonatal period in this
population of overall healthy newborns.
The HAPO Study was an observational
study of normal, rather than high-risk,
mothers and infants. There was some
potential bias in the sample used for the
neonatal glucose measurements re-
ported here, because the protocol-
defined collection of heel stick samples
was omitted more frequently by caregiv-
ers if cord blood was not collected, if in-
fants were born through cesarean deliv-
ery, preterm, or from mothers with
preeclampsia, or if infants were admit-
ted to the NICU. Although this might lead
to an underestimate of the frequency of
biochemical hypoglycemia, no bias
would be expected in clinical assess-
ments for neonatal hypoglycemia.
Overall, the frequency of a diagnosis of
clinical neonatal hypoglycemia was
low in the HAPO Study cohort. There is
ARTICLES
controversy regarding the definition of
clinically significant neonatal hypogly-
cemia, and differences in ascertain-
ment are likely to be present among
clinicians. Nevertheless, associations
of maternal glycemia from the OGTT at
28 weeks of gestation, cord serum
C-peptide levels, and neonatal anthro-
pometric measurements with clinical
neonatal hypoglycemia were similar to
the associations found with biochemi-
cally defined hypoglycemia.
CONCLUSIONS
Our data from plasma glucose measure-
ments for ⬎17 000 infants of mothers
without diabetes provide useful clinical
information. Our findings of strong con-
tinuous associations between maternal
OGTT measurements and cord C-peptide
levels and between cord C-peptide levels
and neonatal hypoglycemia and exces-
sive size at birth are consistent with and
extend the Pedersen hypothesis. The fact
that these relationships extend across
the entire range of maternal glycemia
suggests a physiologic relationship be-
tween maternal glycemia and fetal insu-
lin production.
ACKNOWLEDGMENTS
The study was funded by the National In-
stitute of Child Health and Human Devel-
opment and the National Institute of Dia-
betes, Digestive, and Kidney Diseases
(grants R01-HD34242 and R01-HD34243),
by the National Center for Research Re-
sources (grants M01-RR00048 and M01-
RR00080), and by the American Diabetes
Association. Support also was provided
to local field centers by Diabetes UK
(grant RD04/0002756), Kaiser Perma-
nente Medical Center, KK Women’s and
Children’s Hospital, Mater Mothers’ Hos-
pital, Novo Nordisk, the Myre Sim Fund of
the Royal College of Physicians of Edin-
burgh, and the Howard and Carol Ber-
nick Family Foundation. A complete list of
authors in the HAPO Study Cooperative
Research Group is available in the Sup-
plemental Appendix.
e1551
REFERENCES
1. HAPO Study Cooperative Research Group.
Hyperglycemia and adverse pregnancy out-
comes. N Engl J Med. 2008;358(19):
1991–2002
2. Pedersen J. Diabetes and Pregnancy: Blood
Sugar of Newborn Infants. Copenhagen,
Denmark: Danish Science Press; 1952:230
3. HAPO Study Cooperative Research Group.
The Hyperglycemia and Adverse Pregnancy
Outcome (HAPO) Study. Int J Gynaecol Ob-
stet. 2002;78(1):69 –77
4. Nesbitt GS, Smye M, Sheridan B, Lappin TRJ,
Trimble ER; HAPO Study Cooperative Re-
search Group. Integration of local and cen-
tral laboratory functions in a worldwide
multicentre study: experience from the Hy-
perglycemia and Adverse Pregnancy Out-
come (HAPO) Study. Clin Trials. 2006;3(4):
397– 407
5. Sweet DG, Hadden D, Halliday HL. The effect
of feeding on the neonatal blood glucose
level at 1-hour of age. Early Hum Dev. 1999;
55(1):63– 66
6. HAPO Study Cooperative Research Group.
Hyperglycemia and Adverse Pregnancy Out-
come (HAPO) Study: associations with neo-
natal anthropometrics. Diabetes. 2009;
58(2):453– 459
e1552 METZGER et al
Downloaded from pediatrics.aappublications.org by guest on March 29, 2012
7. O’Rahilly S, Burnett MA, Smith RF, Darley JH,
Turner RC. Haemolysis affects insulin but
not C-peptide immunoassay. Diabetologia.
1987;30(6):394 –396
8. Alkalay AL, Sarnat HB, Flores-Sarnat L,
Elashoff JD, Farber SJ, Simmons CF. Popu-
lation meta-analysis of low plasma glucose
thresholds in full-term normal newborns.
Am J Perinatol. 2006;23(2):115–119
9. Catalano PM, Thomas AJ, Avallone DA, Amini
SB. Anthropometric estimation of neonatal
body composition. Am J Obstet Gynecol.
1995;173(4):1176 –1181
10. Sosenko IR, Kitzmiller JL, Loo SW, Blix P,
Rubenstein AH, Gabbay KH. Infant of the dia-
betic mother: correlation of increased
C-peptide levels with macrosomia and hypo-
glycemia. N Engl J Med. 1979;301(16):
859 – 862
11. Silverman BL, Rizzo T, Green OC, et al. Long-
term prospective evaluation of offspring of
diabetic mothers. Diabetes. 1991;40(suppl
2):121–125
12. Pedersen J. The Pregnant Diabetic and Her
Newborn: Problems and Management. Bal-
timore, MD: Williams & Wilkins: ; 1977
13. Kitzmiller JL, Jovanovic L, Bown F, Coustan D,
Reader DM, eds. Managing Preexisting Diabe-
tes and Pregnancy: Technical Reviews and
Consensus Recommendations for Care. Alex-
andria, VA: American Diabetes Association;
2008
14. Holtrop PC. The frequency of hypoglycemia
in full-term large and small for gestational
age newborns. Am J Perinatol. 1993;10(2):
150 –154
15. Cornblath M, Schwartz R. Disorders of Car-
bohydrate Metabolism in Infancy. 3rd ed.
Boston, MA: Blackwell Scientific; 1991
16. Srinivasan G, Pildes RS, Cattamachi G, Voora
S, Lilien LD. Plasma glucose values in nor-
mal neonates: a new look. J Pediatr. 1986;
109(1):114 –117
17. Heck LJ, Erenberg A. Serum glucose in term
neonates during the first 48 hours. J Pedi-
atr. 1987;110(1):119 –122
18. Cornblath M, Hawdon JM, Williams AF, et al.
Controversies regarding definition of neona-
tal hypoglycemia: suggested operational
thresholds. Pediatrics. 2000;105(5):
1141–1145
19. Rozance PJ, Hay WW. Hypoglycemia in new-
born infants: features associated with ad-
verse outcomes. Biol Neonate. 2006;90(2):
74 – 86
 Hyperglycemia and Adverse Pregnancy Outcome Study: Neonatal Glycemia
Boyd E. Metzger, Bengt Persson, Lynn P. Lowe, Alan R. Dyer, J. Kennedy
Cruickshank, Chaicharn Deerochanawong, Henry L. Halliday, Anselm J. Hennis,
Helen Liley, Pak C. Ng, Donald R. Coustan, David R. Hadden, Moshe Hod, Jeremy J.
N. Oats, Elisabeth R. Trimble and for the HAPO Study Cooperative Research Group
Pediatrics 2010;126;e1545; originally published online November 15, 2010;
DOI: 10.1542/peds.2009-2257
Updated Information & including high resolution figures, can be found at:
Services http://pediatrics.aappublications.org/content/126/6/e1545.full.
html
Supplementary Material Supplementary material can be found at:
http://pediatrics.aappublications.org/content/suppl/2010/11/05
/peds.2009-2257.DC1.html
References This article cites 15 articles, 2 of which can be accessed free
at:
http://pediatrics.aappublications.org/content/126/6/e1545.full.
html#ref-list-1
Permissions & Licensing Information about reproducing this article in parts (figures,
tables) or in its entirety can be found online at:
http://pediatrics.aappublications.org/site/misc/Permissions.xht
ml
Reprints Information about ordering reprints can be found online:
http://pediatrics.aappublications.org/site/misc/reprints.xhtml

PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

publication, it has been published continuously since 1948. PEDIATRICS is owned, published,
and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk
Grove Village, Illinois, 60007. Copyright © 2010 by the American Academy of Pediatrics. All
rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.

Downloaded from pediatrics.aappublications.org by guest on March 29, 2012