Content uploaded by Paula Krakowiak
Author content
All content in this area was uploaded by Paula Krakowiak on Jan 15, 2014
Content may be subject to copyright.
Maternal Metabolic Conditions and Risk for Autism
and Other Neurodevelopmental Disorders
WHAT’S KNOWN ON THIS SUBJECT: Diabetes during pregnancy
has been associated with general development impairments in
offspring; however, associations between autism and maternal
diabetes have been inconsistent. Few studies have examined
related conditions accompanied by underlying increased insulin
resistance and their association with developmental outcomes.
WHAT THIS STUDY ADDS: This population-based study in young
children provides evidence that maternal metabolic conditions
are a risk factor for autism, developmental delay without autistic
symptoms, and impairments in several domains of development,
particularly expressive language, after adjusting for
sociodemographic and other characteristics.
abstract
OBJECTIVE: We examined whether metabolic conditions (MCs) during
pregnancy (diabetes, hypertension, and obesity) are associated with au-
tism spectrum disorder (ASD), developmental delays (DD), or impair-
ments in specific domains of development in the offspring.
METHODS: Children aged 2 to 5 years (517 ASD, 172 DD, and 315 con-
trols) were enrolled in the CHARGE (Childhood Autism Risks from Ge-
netics and the Environment) study, a population-based, case-control
investigation between January 2003 and June 2010. Eligible children
were born in California, had parents who spoke English or Spanish,
and were living with a biological parent in selected regions of
California. Children’s diagnoses were confirmed by using standardized
assessments. Information regarding maternal conditions was ascertained
from medical records or structured interview with the mother.
RESULTS: All MCs were more prevalent among case mothers compared
with controls. Collectively, these conditions were associated with
a higher likelihood of ASD and DD relative to controls (odds ratio:
1.61 [95% confidence interval: 1.10–2.37; odds ratio: 2.35 [95% confi-
dence interval: 1.43–3.88], respectively). Among ASD cases, children of
women with diabetes had Mullen Scales of Early Learning (MSEL)
expressive language scores 0.4 SD lower than children of mothers
without MCs (P,.01). Among children without ASD, those exposed to
any MC scored lower on all MSEL and Vineland Adaptive Behavior
Scales (VABS) subscales and composites by at least 0.4 SD (P,
.01 for each subscale/composite).
CONCLUSIONS: Maternal MCs may be broadly associated with neuro-
developmental problems in children. With obesity rising steadily, these
results appear to raise serious public health concerns. Pediatrics
2012;129:e1121–e1128
AUTHORS: Paula Krakowiak, MS,
a
,
b
Cheryl K. Walker, MD,
c
Andrew A. Bremer, MD, PhD,
d
Alice S. Baker, BA,
a
Sally
Ozonoff, PhD,
b
,
e
Robin L. Hansen, MD,
b
,
f
and Irva Hertz-
Picciotto, PhD
a
,
b
a
Division of Epidemiology, Department of Public Health Sciences,
Departments of
c
Obstetrics and Gynecology,
e
Psychiatry and
Behavioral Sciences, and
f
Pediatrics, School of Medicine, and
b
MIND (Medical Investigation of Neurodevelopmental Disorders)
Institute, University of California, Davis, California; and
d
Department of Pediatrics, School of Medicine, Vanderbilt
University, Nashville, Tennessee
KEY WORDS
autism, developmental delay, diabetes, epidemiology,
hypertension, obesity
ABBREVIATIONS
ADI-R—Autism Diagnostic Interview, Revised
ADOS—Autism Diagnostic Observation Schedule
ASD—autism spectrum disorder
CI—confidence interval
DD—developmental delays
EEQ—Environmental Exposure Questionnaire
GDM—gestational diabetes
GP—general population
LS—least squares
MC—metabolic condition
MSEL—Mullen Scales of Early Learning
OR—odds ratio
T2D—type 2 diabetes
TD—typical development
VABS—Vineland Adaptive Behavior Scales
Ms Krakowiak developed the concept for this article with
substantial contribution from Drs Walker, Bremer, and Hertz-
Picciotto; Ms Baker extracted medical records data and
contributed extensively to data verification; Ms Krakowiak
performed data validation and statistical analysis and prepared
the manuscript; Dr Ozonoff and Dr Hansen provided input
regarding diagnoses and participated in the interpretation of
findings; and Dr Hertz-Picciotto, as principal investigator,
obtained funding for the CHARGE study and supervised all stages
of the study. All authors participated in the revision of the
manuscript and gave final approval of the version to be
published.
www.pediatrics.org/cgi/doi/10.1542/peds.2011-2583
doi:10.1542/peds.2011-2583
Accepted for publication Dec 19, 2011
Address correspondence to Paula Krakowiak, MS, MIND Institute,
2825 50th St, Sacramento, CA 95817. E-mail: pkrakowiak@ucdavis.
edu
(Continued on last page)
PEDIATRICS Volume 129, Number 5, May 2012 e1121
ARTICLE
Autism spectrum disorders (ASDs) are
neurodevelopmental disorders char-
acterized by impairments in social in-
teraction, communication deficits, and
stereotyped behaviors.
1
Approximately
1 in 110 children has ASD, and the cu-
mulative incidence of this disorder
seems to be increasing.
2,3
Moreover, 1 in
83 children has other developmental
delays (DDs).
4
Language and cognitive
delays are also seen in a majority of
children with ASD,
5
suggesting that com-
mon exposures may be contributing to
the pathology of both of these devel-
opmental disorders. To date, the etiology
of ASD is unknown; however, several
studies suggest that its pathogenesis
most likely begins in utero.
6–8
Similarly,
the causes of cognitive impairment re-
main unknown for most children.
9
An
association between general develop-
mental impairments and maternal di-
abetes has been previously observed.
Studies involving women with diabetes
found correlations between gestational
measures of maternal lipid and glucose
metabolism and poorer performance of
the offspring on standardized IQ tests
10
and motor development assessments.
11,12
Dionne et al
13
reported significant ex-
pressive language impairments in young
children born to mothers with gestational
diabetes (GDM) compared with children
of women without diabetes. Moreover, 2
small studies
14,15
demonstrated mild
deficits in recognition memory per-
formance in infants of mothers with
diabetes, suggesting aberrations in
hippocampal function. Abnormalities
within the limbic system have also
been documented in children with ASD,
and language deficits are among the
core features of this disorder. However,
the association between ASD and ma-
ternal diabetes has not been consis-
tently reported in population-based
studies,
16–18
highlighting the need for
further investigation.
Insulin resistance and chronic inflam-
mation in type 2 diabetes (T2D) and
related conditions, including obesity
and hypertension, have been well es-
tablished.
19–21
In addition, because
sensitivity to insulin naturally decrea-
ses during gestation, women with
impaired glucose tolerance before
pregnancy may develop GDM when
their insulin production becomes in-
sufficient to maintain euglycemia.
22,23
In the United States, nearly 60% of
women of childbearing age (20–39
years) are overweight, one-third are
obese, and 16% have metabolic syn-
drome.
24,25
Moreover, recent studies
found that 1.1% of US pregnancies
were complicated by chronic hyper-
tension,
26
and in California, 1.3% of
pregnant women had T2D and another
7.4% had GDM.
27
To date, no studies
involving humans have examined the
relationship between these metabolic
conditions (MCs) collectively and de-
velopmental outcomes in children. The
aims of this study were to describe the
prevalence of diabetes (T2D/GDM), hy-
pertension, and obesity during the in-
dex pregnancy in mothers of children
with ASD, DD, and typical development
(TD) and to investigate whether these
conditions were associated with impair-
ments in specific developmental domains
in the offspring.
METHODS
Study Population
This study was conducted by using data
from the CHARGE (Childhood Autism Risks
from Genetics and the Environment)
study, an ongoing, population-based,
case-control study.
28
Participants were
selected from 3 strata: ASD, DD with-
out ASD, and general population (GP).
Eligible children were between the ages
of 24 and 60 months, born in California,
living with at least 1 biological parent
who spoke English or Spanish, and
residing in the catchment areas of
aspecified list of regional centers in
California. Children with major motor
and sensory impairments (eg, blindness,
deafness) that would preclude a valid
developmental assessment were ex-
cluded. No other exclusions were
made on the basis of genetics or
family phenotype. Children with ASD
or DD were identified through re-
gional centers, providers/clinics, self-
referrals, and general public outreach.
GP children were identified from state
birth files, and a stratified random
sample was generated by frequency-
matching to a projected distribution
of ASD cases on age, gender, and re-
gional center catchment area. Children
with DD were not frequency-matched to
either group. The CHARGE study pro-
tocol was approved by institutional
review boards of the University of Cal-
ifornia in Davis and Los Angeles and the
State of California Committee for the
Protection of Human Subjects. Written
informed consent was obtained before
participation.
Diagnostic Validation
All children referred for the study with
a diagnosis of autism/ASD were reeval-
uated with the Autism Diagnostic Inter-
view, Revised (ADI-R)
29
and the Autism
Diagnostic Observation Schedule (ADOS)
30
by trained clinicians at the UC Davis
MIND (Medical Investigation of Neuro-
developmental Disorders) Institute to
confirm the diagnosis by using criteria
described by Risi et al.
31
The Social Com-
munication Questionnaire,
32
designed to
screen for ASD, was administered to
parents of DD and GP children; children
with scores above the ASD cutoff ($15)
were assessed with the ADOS and ADI-R,
and reclassified to ASD if criteria were
satisfied.
Mullen Scales of Early Learning (MSEL)
33
and Vineland Adaptive Behavior Scales
(VABS)
34
were administered to all chil-
dren to determine cognitive and adap-
tive development, respectively, and
diagnostic groups for children without
ASD were defined on the basis of these
assessments. The DD group consisted
e1122 KRAKOWIAK et al
of children with composite scores ,70
on the MSEL and/or VABS. The TD group
only included GP children with no pre-
vious diagnosis of ASD or DD, Social
Communication Questionnaire score
,15, and composite scores $70 on
both the MSEL and VABS. All CHARGE
study clinical assessment personnel
had attained research reliability on the
developmental assessments they ad-
ministered (ADI-R, ADOS, MSEL, and
VABS). Bilingual study staff were avail-
able to administer informed consent
and all instruments/questionnaires in
Spanish.
Maternal Conditions and Potential
Confounders
Demographic and medical informa-
tion was obtained from the CHARGE
Environmental Exposure Questionnaire
(EEQ; available for 97.6% of partic-
ipants), bir th files, and medical records
(available for 57.7% of participants).
The EEQ is a structured telephone-
administered interview with the bi-
ological mother and incl udes questions
about demographic characteristics,
maternal medical history, and various
environmental exposures. Trained study
staff extracted data from medical
records.
The primary MC of interest was maternal
T2D or GDM in theindex pregnancy. Other
conditions of interest werehypertension
and obesity, defined as BMI $30, with
onset before the index pregnancy. BMI
(kilograms per meter squared) was
calculated by using the height and pre-
pregnancy weight recorded in the
medical records (when available) or
from the EEQ. BMI obtained from self-
reported measurements was validated
against BMI calculated from medical
record measurements in a subset of 346
(34.5%) women for whom both data
sources were available. The intraclass
correlation coefficient between these
BMI calculations was 0.912, indicating
strong agreement.
Diabetes and hypertension (with or
without preeclampsia) were consid-
ered present if they were noted on the
medical history form in the prenatal
medical record (when available) or if
mothers answered “yes”to “During
this [index] pregnancy were you ever
told by a physician or nurse that you
had gestational diabetes?”or “At any
time before you became pregnant with
[index child], were you ever told by
a doctor that you had [diabetes, high
blood pressure]?”in the EEQ. Discrep-
ancies between medical records and
self-report were verified. Self-reported
diabetes and hypertension were vali-
dated in a subset of 560 (55.8%) women.
Agreement between self-report and
medical records was excellent for di-
abetes (k= 0.79) and fair for hy-
pertension (k= 0.38). Women who
misreported having hypertension were
more likely to be multiparous and to
have a history of gestationally induced
hypertension. Because obesity, hyper-
tension, and diabetes (T2D/GDM) are
closely linked clinical manifestations of
insulin resistance, an “any metabolic
condition”variable was created as an-
other proxy measure of insulin resis-
tance. For all analyses, these predictor
variables were categorized as follows:
(1) had condition of interest; (2) did not
have condition of interest but had an-
other MC or was overweight (BMI $25);
and (3) did not have any MCs and had
aBMI,25. For each MC variable, we
present results for level 1 versus level
3 to maximize the contrast and sim-
plify the interpretation of findings.
Covariates selected a priori included
mother’s age at delivery, race/ethnicity
(non-Hispanic white, other non-Hispanic,
or Hispanic), education level (high school
or less, some college, or bachelor de-
gree or higher), delivery payer (govern-
ment program or private insurance),
calendar time defined as number of
years from the first participant’senroll-
ment date, the child’s gender and age in
years at study enrollment, and catch-
ment area; the latter 3 factors were
frequency-matching variables. An in-
dicator for known chromosomal/genetic
(eg, trisomy 21, Angelman syndrome),
metabolic/mitochondrial (eg, Leigh syn-
drome, carnitine deficiency), or neu-
rologic (eg, cerebral palsy, epilepsy,
hydrocephalus) disorders was cre-
ated by using parent-report data from
the child’s medical history completed
by a study physician.
For this study, 1004 children (517 ASD,
172 DD, and 315 TD), consisting of 964
singletons and 20 sibling pairs, were
included from a pool of 1317 partic-
ipants who completed a clinic visit be-
tween January 2003 and June 2010.
Excluded were 245 children who either
did not complete the necessary assess-
ments to confirm diagnosis or did not
satisfy the diagnostic criteria for the
groups considered in this study, and 68
children for whom data regarding ma-
ternal diabetes, hypertension, and obe-
sity were unavailable.
Statistical Analyses
All analyses were performed by using
SAS version 9.2 (SAS Institute Inc, Cary,
NC). A directed acyclic graph
35
(Fig 1)
was constructed to represent the un-
derlying causal relationships and used
as guidance in evaluating associations
among predictors, covariates, and
outcomes. All variables with arrows
pointing to the predictor (obesity, hy-
pertension, or T2D/GDM) and the out-
come (neurodevelopmental disorder)
were defined as potential confounders
and evaluated further. To determine
whether MCs during pregnancy were
associated with an increased risk of
having a child with ASD or DD relative
to TD, 4 multinomial logistic regression
models were fitted and corrected for
family clusters, 1 for each condition
and 1 for the “any metabolic condition”
predictor. Odds ratios (ORs) and 95%
confidence intervals (CIs) were used as
ARTICLE
PEDIATRICS Volume 129, Number 5, May 2012 e1123
estimates of relative risk. Final models
were adjusted for mother’s age at de-
livery,race/ethnicity, education, payer,
calendar time, child’s age and gen-
der, and catchment area.
To evaluate children’s developmental
scores in association with maternal
MCs, linear regression models cor-
rected for family clusters were fitted.
Age-standardized scores from MSEL
visual reception, fine motor, receptive
language, and expressive language
subscales (mean 6SD: 50 610) and
from VABS communication, socializa-
tion, and motor skills domains (mean
6SD: 100 615) were examined; MSEL
and VABS composite scores (mean 6
SD: 100 615) were also considered.
Covariates in these models included
mother’s age, race/ethnicity, education,
payer, calendar time, child’s age and
gender, and catchment area. These
models compared mean assessment
scores of children whose mothers had
an MC of interest (diabetes, hyperten-
sion, obesity, or any MC) versus those
whose mothers did not have any of
these MCs and had a BMI ,25; com-
parisons were done separately among
children with and those without ASD.
Least squares (LS) means and SEs were
used to measure association between
MCs and developmental scores. The de-
velopmental domains examined were
selected a priori because of their bi-
ological relevance to MCs, and adjust-
ment for multiple comparisons was not
performed. In the non-ASD subset, DD
and TD groups were combined because
the mean differences in developmental
scores of children born to mothers
with MCs compared with no MCs were
similar in magnitude between these di-
agnostic groups. In addition, the combined
sample provided more power to detect
an association, through not only a larger
sample but also the full range of variation
in cognitive and adaptive scores.
36
RESULTS
Mothers of children with ASD were
similar to controls in terms of race/
ethnicity, education, and delivery payer
(Table 1). Mothers of DD children were
more likely to be Hispanic and to have
lower education and less likely to have
had private insurance compared with
controls. Multiparas in both case groups
weremorelikelytohaveahistoryofGDM
compared with controls. As expected,
because of frequency-matching, the ASD
and TD groups were similar with respect
to child’s age and gender; the DD group,
which was not matched, had a higher
proportion of girls compared with the
other 2 groups. Discrepancies between
the ASD and TD groups regarding the Los
Angeles catchment area arose because
of delayed implementation of the re-
cruitment protocol for controls in the
initial year of the study at the Los Angeles
site, leading to some imbalance in the
case-control ratio.
The proportions of T2D/GDM in the ASD
(9.3%) and DD (11.6%) groups were
higher compared with controls (6.4%);
after adjustment for covariates, moth-
ers with diabetes were 2.3 times more
likely to have a child with DD (OR: 2.33
[95% CI: 1.08–5.05]), but the associa-
tion between diabetes and ASD did not
reach statistical significance (Table 2).
The prevalence of hypertension was
low in all groups but more common
among case mothers than controls
(ASD: 3.7%; DD: 3.5%; TD: 1.3%); after
adjusting for covariates, the associa-
tion between hypertension and ASD or
DD was not significant. The risk of
having a child with ASD or DD, relative
FIGURE 1
Directed acyclic graph. All relationships represented in this graphwere based on review of the literature, with the exception of the frequency-matching variables
(child’s age and gender, and catchment area). Solid arrows indicate stronger associations, and dashed arrows denote weaker associations.
e1124 KRAKOWIAK et al
to TD, was significantly increased among
obese women (ASD, OR: 1.67 [95% CI:
1.10–2.56]; DD, OR: 2.08 [95% CI: 1.20–
3.61]); .20% of case mothers were
obese compared with 14.3% of controls.
The prevalence of any MC was higher in
the ASD (28.6%) and DD (34.9%) groups
compared with controls (19.4%), with
respective adjusted ORs of 1.61 (95% CI:
1.10–2.37) and 2.35 (95% CI: 1.43–3.88).
Analyses restricted to children without
known genetic, metabolic, or neurologic
disorders conferred similar or slightly
stronger associations (data not shown).
Within the ASD group, children of
mothers with diabetes performed 0.37
SD lower on the MSEL expressive lan-
guage scale compared with children of
nondiabetic mothers (P= .01; Table 3);
MSEL receptive language and VABS
communication scores were also lower
among children of diabetic mothers,
with differences approaching statistical
significance. No significant differences
in MSEL or VABS scores were observed
regarding MCs collectively among chil-
dren with ASD.
Among children without ASD, MSEL re-
ceptive and expressive language scores
were ∼0.5 SD lower among children of
mothers with diabetes compared with
nondiabetic mothers (P= .03 for both
subscales; Table 4); MSEL composite
scores were borderline lower among
children of mothers with diabetes.
VABS socialization scores were 0.49 SD
lower among children from diabetic
pregnancies (P=0.01). The presence of
any MC was associated with lower
scores on all MSEL subscales and
composite (visual: –0.51 SD [P=0.01];
motor: –0.53 SD [P=0.01]; receptive:
–0.50 SD [P= .004]; expressive: –0.59 SD
[P= .0004]; composite: –0.65 SD [P=
.001]) and all VABS domains and com-
posite (communication: –0.43 SD [P=
.01]; socialization: –0.50 SD [P=0.0004];
motor: –0.39 SD [P=0.04]; composite: –
0.51 SD [P=0.005]). Findings from re-
stricted analyses were nearly identical
(data not shown).
DISCUSSION
In this study, we observed that diabetes,
hypertension, and obesity were more
common among mothers of children
with ASD and DD compared with con-
trols. Furthermore, diabetes, in par-
ticular, was associated with statistically
significantly greater deficits in expres-
sive language among children with ASD,
although the magnitude of the deficits
was relatively small. Among children
without ASD, MCs collectively were
TABLE 1 Characteristics of the Study Sample According to Diagnostic Group: CHARGE Study,
January 2003–June 2010 (N= 1004)
Characteristic ASD DD TD ASD
Versus TD
DD
Versus TD
n%n%n% P P
Race/ethnicity .49 .002
White, non-Hispanic 304 58.8 82 47.7 199 63.2
Hispanic (any race) 131 25.3 60 34.9 68 21.6
Other race, non-Hispanic 82 15.9 30 17.4 48 15.2
Education .10 ,.001
High school or less 78 15.1 50 29.1 47 14.9
Some college 208 40.2 66 38.4 104 33.0
Bachelor degree or higher 231 44.7 56 32.5 164 52.1
Delivery payer .12 ,.001
Government program 84 16.3 50 29.1 39 12.4
Private insurance 433 83.7 122 70.9 276 87.6
Multipara 281 54.4 111 64.5 176 55.9 .63 .06
Maternal parent with diabetes 111 22.0 39 23.1 64 20.8 .63 .41
Missing 13 3 7
GDM in previous pregnancy
(multiparas only)
19 6.8 6 5.4 5 2.8 .06 .28
BMI (kg/m
2
) .05 .003
,18.5 (underweight) 19 3.7 6 3.5 10 3.2
18.5–24.9 (normal) 269 52.0 65 37.8 172 54.6
25.0–29.9 (overweight) 118 22.8 60 34.9 88 27.9
$30.0 (obese) 111 21.5 41 23.8 45 14.3
Hypertension before pregnancy 19 3.7 6 3.5 4 1.3 .03 .11
Diabetes in index pregnancy
Type 1 1 0.2 1 0.6 0 0.0 .99 .36
T2D 4 0.8 1 0.6 1 0.3 .65 .30
GDM 44 8.5 19 11.0 19 6.1 .18 .16
T2D or GDM 48 9.3 20 11.6 20 6.4 .13 .10
Any type 49 9.5 21 12.2 20 6.4 .11 .07
Took antidiabetic medication 10 1.9 10 5.8 5 1.6 .69 .01
Hypertension, obesity, or
diabetes (T2D or GDM)
148 28.6 60 34.9 61 19.4 .003 ,.001
Child’s gender, male
a
436 85.8 121 68.0 256 81.3 .08 .001
Child with a known genetic,
metabolic, or neurologic disorder
12 2.3 56 32.9 0 0.0 ,.001 ,.001
Missing 1 2
Regional Center catchment
area at enrollment
a
,.001 .01
Alta, Far Northern, and Redwood Coast 185 35.8 87 50.6 132 41.9
North Bay 70 13.6 18 10.4 48 15.2
East Bay, San Andreas, and Golden Gate 88 17.0 17 9.9 63 20.0
Valley Mountain, Central Valley, and Kern 88 17.0 38 22.1 51 16.2
All Los Angeles, Orange, San Diego,
Tri-counties, and Inland
86 16.6 12 7.0 21 6.7
Continuous variables Mean SD Mean SD Mean SD PP
Age at delivery, y 31.11 5.48 30.81 6.52 31.08 5.65 .94 .63
Child’s age at study enrollment, y
a
3.65 0.80 3.79 0.76 3.54 0.80 .06 .001
Time from earliest consent date, y 3.16 1.93 4.26 1.81 3.92 1.77 ,.001 .05
a
Children from the GP were frequency matched on age, gender, and catchment area (regional centers) to ASD cases.
ARTICLE
PEDIATRICS Volume 129, Number 5, May 2012 e1125
associated with impairments in visual
reception, motor skills, and receptive and
expressive language, as well as adaptive
communication and socialization.
Our findings relating diabetes to im-
pairments in cognitive and language
development areconsistent withsome
10,
13,16
but not all previous studies.
17,18
Hultman et al
17
included numerous ma-
ternal and pregnancy characteristics
in their multivariable model; as such,
the temporal and causal interrelation-
ships among these risk factors were
effectively ignored. Hence, it is plau-
sible that the association with ASD
may have been attenuated as a con-
sequence of including complications
downstream of diabetes. Furthermore,
although Dodds et al
18
reported higher
proportions of diabetes (preexisting
and GDM) among mothers of ASD chil-
dren compared with controls in un-
adjusted analyses that only neared
statistical significance, key confounders
were not considered. Interestingly, pre-
pregnancy obesity ($90 kg) and ex-
cessive weight gain ($18 kg) during
pregnancy were significantly associ-
ated with ASD; however, the final model
also included intermediary maternal
characteristics (eg, labor type) poten-
tially on the causal pathway between
these risk factors and the outcome, thus
limiting the interpretability of these
findings. Nevertheless, obesity is a sig-
nificant risk factor for both hyperten-
sion and diabetes (T2D and GDM) and is
characterized by increased insulin re-
sistance and chronic inflammation, as
are the other 2 MCs we examined.
19,20,22
Therefore, in our study, we constructed
a causal diagram to investigate the in-
terrelationships among maternal MCs,
covariates, and the outcome. We also
compared the risk of an adverse out-
come in children whose mothers had
a given condition (eg, diabetes) relative
to those whose mothers had neither
that condition nor the risk factors for
it (eg, no hypertension and with BMI
TABLE 2 OR for Autism/ASD or Other Delays in Relation to Diabetes and Related Conditions:
CHARGE Study, January 2003–June 2010 (N= 1004)
Conditions in Index
Pregnancy
ASD DD TD ASD Versus TD DD Versus TD
n%n%n%OR
a
95% CI OR
a
95% CI
Diabetes
b
48 9.3 20 11.6 20 6.4 1.52 0.82–2.83 2.33 1.08–5.05
Hypertension 19 3.7 6 3.5 4 1.3 2.84 0.94–8.56 3.58 0.93–13.78
Obesity 111 21.5 41 23.8 45 14.3 1.67 1.10–2.56 2.08 1.20–3.61
Any MC(s) 148 28.6 60 34.9 61 19.4 1.61 1.10–2.37 2.35 1.43–3.88
a
Adjusted for mother’s age at delivery, race/ethnicity, education level, delivery payer, calendar time, child’s age at enrollment
and gender, and catchment area. Comparison group had no hypertension or diabetes (T2D or GDM) and also had BMI ,25;
this group included 267 in ASD, 64 in DD, and 172 in TD groups.
b
T2D or GDM only.
TABLE 3 Assessment Scores of Children of Mothers With and Without Diabetes (T2D or GDM),
a
Stratified According to ASD Status
Assessment ASD (n= 315) No ASD (n= 276)
Diabetes No Conditions P
b
Diabetes No Conditions P
b
LS Mean SE LS Mean SE LS Mean SE LS Mean SE
MSEL
Visual reception T score 27.98 2.62 28.29 1.04 .91 44.35 2.94 46.63 1.25 .47
Fine motor T score 26.16 1.94 27.81 1.00 .39 39.89 2.49 44.33 1.24 .10
Receptive language T score 22.88 1.71 25.98 0.86 .07 36.95 2.31 42.36 1.11 .03
Expressive language T score 21.51 1.36 25.19 0.84 .01 36.99 2.27 42.23 1.10 .03
Composite Standard score 56.16 3.00 59.53 1.45 .26 81.90 3.99 90.12 1.95 .06
VABS
Communication standard score 61.74 2.19 66.07 1.24 .05 87.08 3.25 92.53 1.55 .12
Socialization standard score 64.43 1.82 66.52 1.04 .25 87.84 2.70 95.13 1.43 .01
Motor skills standard score 74.86 2.57 74.37 1.57 .85 88.50 4.00 92.05 1.84 .40
Composite standard score 60.82 1.95 62.89 1.21 .28 85.36 3.68 91.69 1.78 .11
a
Diabetes group includes mothers with T2D or GDM; no conditions group consists of mothers with no diabetes (T2D or GDM)
or hypertension and who have BMI,25; this comparison group consisted of 267 in ASD and 236 in non-A SD strata.
b
Adjusted for mother’s age at delivery,race/ethnicity, education level, delivery payer, calendar time, child’sage at enrollment
and gender, and catchment area.
TABLE 4 Assessment Scores of Children of Mothers With and Without any MCs,
a
Stratified According
to ASD Status
Assessment ASD (n= 315) No ASD (n=276)
Any MC No Conditions P
b
Any MC No Conditions P
b
LS Mean SE LS Mean SE LS Mean SE LS Mean SE
MSEL
Visual reception T score 26.98 1.41 28.31 1.03 0.37 41.57 1.72 46.66 1.25 .01
Fine motor T score 27.11 1.18 27.79 1.00 0.58 38.98 1.72 44.32 1.24 .01
Receptive language T score 24.68 1.09 25.92 0.86 0.28 37.34 1.48 42.32 1.11 .004
Expressive language T score 23.72 0.96 25.16 0.84 0.16 36.33 1.38 42.21 1.10 ,.001
Composite Standard score 57.59 1.79 59.49 1.44 0.31 80.35 2.61 90.11 1.94 .001
VABS
Communication standard score 63.71 1.47 66.04 1.23 0.13 85.87 2.11 92.38 1.55 .01
Socialization standard score 67.16 1.32 66.50 1.04 0.62 87.63 1.78 95.10 1.43 ,.001
Motor skills standard score 75.00 1.83 74.35 1.57 0.73 86.15 2.51 92.06 1.84 .04
Composite standard score 62.62 1.55 62.87 1.20 0.86 84.09 2.32 91.68 1.77 .005
a
Any MCs group includes mothers with diabetes (T2D or GDM), hypertension, or a BMI $30; no conditions group consists of
mothers with no diabetes (T2D or GDM) or hypertension and who have a BMI ,25; this comparison group consisted of 267 in
ASD and 236 in non-ASD strata.
b
Adjusted for mother’s age at delivery,race/ethnicity, education level, delivery payer, calendar time, child’sage at enrollment
and gender, and catchment area.
e1126 KRAKOWIAK et al
,25). This approach was applied to
account for increased insulin resistance
due to other MCs.
Reliance on self-reported medical con-
ditions was a limitation of this study.
However, in the 56% of participants for
whom medical records were avail-
able, we found the 2 sources to be in
good agreement. Thus, despite this
limitation, we can have confidence in
our results. Furthermore, although
biological measurements (eg, glucose,
insulin, lipids, immune biomarkers)
before and during pregnancy would
have been ideal, we chose conditions
(T2D/GDM, hypertension, and obesity)
highly indicative of increased insulin
resistance as proxy measures of dys-
regulated metabolism and chronic in-
flammation because we lacked these
biological measurements for most of
the participants.
Nonetheless, our study offers several
strengths. First, it includes a represen-
tative population of cases and controls
with well-defined and consistently
applied diagnoses; the lack of these de-
sign characteristics was a limitation in
previously published population-based
studies investigating maternal risk fac-
tors and ASD.
16–18
Secondly, whereas pre-
vious studies have examined maternal
T2D or GDM in relation to neurodevel-
opmental disorders or developmental
measures, to our knowledge, this is the
first to also examine a broader group of
conditions highly predictive of insulin
resistance and these forms of diabetes.
In a diabetic and possibly prediabetic
pregnancy, poorly regulated maternal
glucose can result in adverse fetal de-
velopment. Prolonged fetal exposure to
elevatedglucoselevelsresultsin chronic
fetal hyperinsulinemia, which in turn
triggers the fetus to increase oxygen
consumption and metabolism, inducing
chronic intrauterine tissue hypoxia.
37
Further biological responses may result
in fetal iron deficiency.
38
Both fetal hyp-
oxia and iron deficiency can profoundly
affect neurodevelopment in humans,
including alterations in myelination
and cortical connectivity and aberra-
tions in hippocampal neurons.
39
Fetal
iron deficiency has also been associ-
ated with reduced recognition memory
as well as behavioral and develop-
mental problems.
40
In addition, in-
creased maternal levels of cytokine
interleukin-6, which can cross the pla-
centa, have been shown to disrupt
normal fetal brain development in ex-
perimental studies involving animal
models, resulting in onset of seizures,
impairments in spatial learning, increa-
ses in hypothalamic-pituitary activity,
disturbances in cholinergic input to
the hippocampus, and reduced neu-
rogenesis in adult hippocampus.
41,42
Increased levels of this interleukin
and other proinflammatory cytokines
are also produced in the presence of
MCssuchasdiabetesandobesity.
CONCLUSIONS
The prevalence of obesity and diabetes
among US women of childbearing age is
34% and 8.7%, respectively.
24,27
Our
findings raise concerns that these
maternal conditions may be associated
with neurodevelopmental problems in
children and therefore could have se-
rious public health implications.
REFERENCES
1. American Psychiatric Association. Diagnos-
tic and Statistical Manual of Mental Dis-
orders.4thed,textrevision.Washington,DC:
American Psychiatric Association; 2000
2. Autism and Developmental Disabilities
Monitoring Network Surveillance Year 2006
Principal Investigators; Centers for Disease
Control and Prevention (CDC). Prevalence
of autism spectrum disorders—Autism and
Developmental Disabilities Monitoring Net-
work, United States, 2006. MMWR Surveill
Summ. 2009;58(10):1–20
3. Hertz-Picciotto I, Delwiche L. The rise in
autism and the role of age at diagnosis.
Epidemiology. 2009;20(1):84–90
4. Bhasin TK, Brocksen S, Avchen RN, Van
Naarden Braun K. Prevalence of four de-
velopmental disabilities among children
aged 8 years—Metropolitan Atlanta De-
velopmental Disabilities Surveillance Pro-
gram, 1996 and 2000. MMWR Surveill Summ.
2006;55(1):1–9
5. Matson JL, Shoemaker M. Intellectual dis-
ability and its relationship to autism spec-
trum disorders. Res Dev Disabil. 2009;30(6):
1107–1114
6. Newschaffer CJ, Croen LA, Daniels J, et al.
The epidemiology of autism spectrum dis-
orders. Annu Rev Public Health. 2007;28:
235–258
7. Pardo CA, Eberhart CG. The neurobiology of
autism. Brain Pathol. 2007;17(4):434–447
8. Gardener H, Spiegelman D, Buka SL. Pre-
natal risk factors for autism: comprehen-
sive meta-analysis. Br J Psychiatry. 2009;
195(1):7–14
9. Yeargin-Allsopp M, Murphy CC, Cordero JF,
Decouflé P, Hollowell JG. Reported biomedical
causes and associated medical conditions
for mental retardation among 10-year-old
children, metropolitan Atlanta, 1985 to 1987.
Dev Med Child Neurol. 1997;39(3):142–149
10. Rizzo TA, Metzger BE, Dooley SL, Cho NH. Early
malnutrition and child neurobehavioral
development: insights from the study of
children of diabetic mothers. Child Dev.
1997;68(1):26–38
11. Ratzon N, Greenbaum C, Dulitzky M, Ornoy
A. Comparison of the motor development of
school-age children born to mothers with
and without diabetes mellitus. Phys Occup
Ther Pediatr. 2000;20(1):43–57
12. Rizzo TA, Dooley SL, Metzger BE, Cho NH,
Ogata ES, Silverman BL. Prenatal and
perinatal influences on long-term psycho-
motor development in offspring of diabetic
mothers. Am J Obstet Gynecol. 1995;173(6):
1753–1758
13. Dionne G, Boivin M, Séguin JR, Pérusse D,
Tremblay RE. Gestational diabetes hinders
language development in offspring. Pedi-
atrics. 2008;122(5). Available at: www.pedi-
atrics.org/cgi/content/full/122/5/e1073
14. DeBoer T, Wewerka S, Bauer PJ, Georgieff
MK, Nelson CA. Explicit memory perfor-
mance in infants of diabetic mothers at 1
ARTICLE
PEDIATRICS Volume 129, Number 5, May 2012 e1127
year of age. Dev Med Child Neurol. 2005;47
(8):525–531
15. Deregnier RA, Nelson CA, Thomas KM,
Wewerka S, Georgieff MK. Neurophysiologic
evaluation of auditory recognition memory in
healthy newborn infants and infants of di-
abetic mothers. JPediatr. 2000;137(6):777–784
16. Leonard H, de Klerk N, Bourke J, Bower C.
Maternal health in pregnancy and intellec-
tual disability in the offspring: a popul ation-
based study. Ann Epidemiol. 2006;16(6):
448–454
17. Hultman CM, Sparén P, Cnattingius S. Peri-
natal risk factors for infantile autism. Epi-
demiology. 2002;13(4):417–423
18. Dodds L, Fell DB, Shea S, Armson BA, Allen AC,
BrysonS.Theroleofprenatal,obstetricand
neonatal factors in the development of au-
tism. JAutismDevDisord. 2011;41(7):891–902
19. Olefsky JM, Glass CK. Macrophages, in-
flammation, and insulin resistance. Annu
Rev Physiol. 2010;72:219–246
20. Ferrannini E, Haffner SM, Stern MP. Essen-
tial hypertension: an insulin-resistant state.
J Cardiovasc Pharmacol. 1990;15(suppl 5):
S18–S25
21. Sweet IR, Gilbert M, Maloney E, Hockenbery
DM, Schwartz MW, Kim F. Endothelial in-
flammation induced by excess glucose is
associated with cytosolic glucose 6-phos-
phate but not increased mitochondrial res-
piration. Diabetologia. 2009;52(5):921–931
22. Zavalza-Gómez AB, Anaya-Prado R, Rincón-
Sánchez AR, Mora-Martínez JM. Adipokines
and insulin resistance during pregnancy.
Diabetes Res Clin Pract. 2008;80(1):8–15
23. American Diabetes Association. Diagnosis
and classification of diabetes mellitus. Di-
abetes Care. 2009;32(32 suppl 1):S62–S67
24. Flegal KM, Carroll MD, Ogden CL, Curtin LR.
Prevalence and trends in obesity among US
adults, 1999-2008. JAMA. 2010;303(3):235–
241
25. Ervin RB. Prevalence of metabolic syn-
drome among adults 20 years of age and
over, by sex, age, race and ethnicity, and
body mass index: United States, 2003-2006.
Natl Health Stat Report. May 2009;(13):1–7
26. Martin JA, Hamilton BE, Sutton PD, et al.
Births: final data for 2007. Natl Vital Stat
Rep. 2010;58(24):1–85
27. Lawrence JM, Contreras R, Chen W, Sacks
DA. Trends in the prevalence of preexisting
diabetes and gestational diabetes mellitus
among a racially/ethnically diverse pop-
ulation of pregnant women, 1999-2005. Di-
abetes Care. 2008;31(5):899–904
28. Hertz-Picciotto I, Croen LA, Hansen R, Jones
CR, van de Water J, Pessah IN. The CHARGE
study: an epidemiologic investigation of
genetic and environmental factors con-
tributing to autism. Environ Health Per-
spect. 2006;114(7):1119–1125
29. Le Couteur A, Lord C, Rutter M. Autism Di-
agnostic Interview–Revised (ADI-R).Los
Angeles, CA: Western Psychological Serv-
ices; 2003
30. Lord C, Risi S, Lambrecht L, et al. The autism
diagnostic observation schedule-generic:
a standard measure of social and commu-
nication deficits associated with the spec-
trum of autism. J Autism Dev Disord. 2000;30
(3):205–223
31. Risi S, Lord C, Gotham K, et al. Combining
information from multiple sources in the
diagnosis of autism spectrum disorders.
J Am Acad Child Adolesc Psychiatry. 2006;
45(9):1094–1103
32. Rutter M, Bailey A, Berument SK, Lord C,
Pickles A. Social Communication Question-
naire (SCQ). Los Angeles, CA: Western Psy-
chological Services; 2003
33. Mullen EM. Mullen Scales of Early Learning.
Circle Pines, MN: American Guidance Serv-
ices, Inc; 1995
34. Sparrow SS, Balla DA, Cicchetti DV. Vineland
Adaptive Behavior Scales Interview Edition
Expanded Form Manual. Circle Pines, MN:
American Guidance Services, Inc; 1984
35. Greenland S, Pearl J, Robins JM. Causal
diagrams for epidemiologic research. Epi-
demiology. 1999;10(1):37–48
36. Winship C, Radbill L. Sampling weights and
regression analysis. Sociol Methods Res.
1994;23(2):230–257
37. Eidelman AI, Samueloff A. The pathophysi-
ology of the fetus of the diabetic mother.
Semin Perinatol. 2002;26(3):232–236
38. Georgieff MK. The role of iron in neuro-
development: fetal iron deficiency and the
developing hippocampus. Biochem Soc
Trans. 2008;36(pt 6):1267–1271
39. Georgieff MK. The effect of maternal di-
abetes during pregnancy on the neuro-
development of offspring. Minn Med. 2006;
89(3):44–47
40. Lozoff B, Georgieff MK. Iron deficiency and
brain development. Semin Pediatr Neurol.
2006;13(3):158–165
41. Jonakait GM. The effects of maternal in-
flammation on neuronal development: pos-
sible mechanisms. Int J Dev Neurosci.2007;
25(7):415–425
42. Boksa P. Effects of prenatal infection on
brain development and behavior: a review
of findings from animal models. Brain
Behav Immun. 2010;24(6):881–897
(Continued from first page)
PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
Copyright © 2012 by the American Academy of Pediatrics
FINANCIAL DISCLOSURE: Dr Hansen receives grant support from Autism Speaks; the other authors have indicated they have no financial relationships relevant to
this article to disclose.
FUNDING: This research was supported by the National Institutes of Health (P01 ES11269 and R01 ES015359), the US Environmental Protection Agency through the
Science to Achieve Results program (R829388 and R833292), and by the MIND Institute, University of California Davis. Funded by the National Institutes of Health
(NIH).
e1128 KRAKOWIAK et al
