Maternal Exposure to Caffeine and Risk of Congenital Anomalies: A Systematic Review
Author(s): Marilyn L. Browne
Source: Epidemiology, Vol. 17, No. 3 (May, 2006), pp. 324-331
Published by: Lippincott Williams & Wilkins
Stable URL: http://www.jstor.org/stable/20486221 .
Accessed: 05/06/2014 12:17
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
Lippincott Williams & Wilkins is collaborating with JSTOR to digitize, preserve and extend access to
Epidemiology.
http://www.jstor.org
This content downloaded from 169.226.11.193 on Thu, 5 Jun 2014 12:17:54 PM
All use subject to JSTOR Terms and Conditions
REVIEW ARTICLE
Exposure to Caffeine and Risk of
Congenital Anomalies
Maternal
A SystematicReview
Marilyn L. Browne
Background: Caffeine is teratogenicinanimal studieswhen admin
isteredat high concentrations.Previous review articles have con
cluded thatmaternal caffeine consumption does not influencethe
riskof congenital anomalies. These reviewswere narrative rather
thansystematic.The objective of thecurrentsystematicreview is to
provide a critical appraisal of epidemiologic evidence.
database (1966
Methods: A search of theMEDLINE/PUBMED
October 2004) was conducted forall published epidemiologic stud
ies with maternal intake of caffeine as an exposure and major
were abstracted,
malformationsas an outcome. Study characteristics
internalvalidity evaluated, and study findingssummarized.
Results: Twenty-fivepapersmet the initialcriteriafor inclusion,of
which 18were subsequentlyexcluded as a resultof other limitations.
Effect estimates for the remaining7 studieswere generally close to
null. Specific subgroup analyses were summarized across studies
(associations between coffee and cardiovascular malformations,
coffee and oral clefts, and tea and cardiovascularmalformations).
Summarypoint estimatesranged from1.0 to 1.2; theupper limitsof
all confidence intervalswere less than 1.7.
Conclusions: There is no evidence to supporta teratogeniceffectof
caffeine inhumans.Currentepidemiologic evidence is not adequate
to assess the possibility of a small change in risk of congenital
anomalies resultingfrommaternal caffeineconsumption.
(Epidemiology 2006;17: 324-331)
inthebrain,
likelythrough
antagonism
of adenosinereceptors
heart, lungs, and blood vessels. 1"2Based on a recent survey by
theU.S. Department of Agriculture, coffee, soft drinks, and
tea (in that order) are themajor sources of caffeine among
adults.3 Average caffeine intake is estimated to be 164 mg per
day among women 18-34 years and 125 mg per day among
pregnant
women.3In a prospective
cohortstudyconductedin
Connecticut
during1988-1992,caffeineconsumption
during
the firstmonth of pregnancy was reported by 60% of study
participants, with 16% consuming 150 mg or more of caf
feine per day.4
In their review of the animal literature,Nehlig and
effectshave generallybeen
Debry5 note thatteratogenic
observedafteronce-per-dayadministration
of caffeineat
doses much higher than human consumption
levels. How
incombination
withsubstancessuch
ever,whenadministered
as nicotine or alcohol, teratogenic effects of caffeine were
observed at somewhat lower doses. In humans, caffeine and
itsmetabolites easily cross the placenta and reach the fetus.6
in fetal heart rate and placental blood flow have
Decreases
been observed after maternal caffeine ingestion.7'8 Human
studies have also demonstrated increased homocysteine lev
els and decreases in insulin sensitivity after intake of caffeine
or coffee.9 Vascular disruption, increased levels of homocys
teine, and oxidative stress associated with hyperglycemia are
for various congenital malforma
postulated mechanisms
tions.10-12
Given theprevalenceof caffeine
use duringpreg
Caffeine
is a natural component of coffee, tea, and cocoa
products. It is added tomany soft drinks and to certain
and over-the-counter
medications.Caffeine's
prescription
effectsincludecentralnervoussystemstimu
pharmacologic
lation, bronchodilation,
and higher blood
pressure, most
nancy, even a small increase in the risk of congenital anom
alies would have an important effect on public health.
A numberof published reportshave reviewedthe
scientific
literature
maternalcaffeineconsumption
regarding
and congenitalanomalies.5"3'6However, a systematic
re
view of the epidemiologic
literatureon this topic has not been
review,specific
published.In a systematic
well-definedcri
Submitted 23 May
2005;
accepted
19 December
teria are used in searching for papers, selecting studies for
2005.
andOccupationalEpidemiology,
New
From theBureau of Environmental
York StateDepartment
ofHealth,Troy,NY.
This study was
supported by a cooperative
agreement
from the Centers
Disease ControlandPrevention(U50/CCU223184).
u
Supplemental material
for
for this article is available with the online version
of thejournalatwww.epidem.com;clickon "ArticlePlus."
Correspondence:
Marilyn L. Browne,FlaniganSquare, 547 River Street,
Room 200,Troy,NY 12180.E-mail:mlblO(health.state.ny.us.
Copyright
(C 2006 by Lippincott Williams
& Wilkins
ISSN: 1044-3983/06/1703-0324
DOI: 10.1097/01.ede.0000208476.36988.44
324
inclusion,criticallyappraisingstudy
methods,and summa
The methods for this systematic
rizingstudyfindings.17
review were based onMulrow's andWeed's
guidelines.17"18
The objectives of this review are to summarize and
evaluatemethodologicalaspectsof epidemiologicstudiesof
maternal caffeine exposure and risk of congenital anomalies.
A critical appraisal of currently available epidemiologic ev
idence is used to suggest recommendations for future re
search.
Epidemiology
* Volume
This content downloaded from 169.226.11.193 on Thu, 5 Jun 2014 12:17:54 PM
All use subject to JSTOR Terms and Conditions
17, Number
3, May
2006
Epidemiology * Volume
Maternal Exposure to Caffeine and Risk of Congenital Anomalies
17, Number 3, May 2006
METHODS
ingwere
Search Strategy and Selection Criteria
Studies were
identified through a
search of
the
included in an assessment of internal validity. No
was made toexcludestudiesbased onmethodologic
attempt
quality.
Narrative Summary and Meta-analysis of Study
MEDLINE/PUBMED database (1966-October 2004). A
free-text
search limitedto human studieswas conducted Findings
Summaryodds ratioswere calculatedforstudieswith
using the terms "caffeine or caffeinated or coffee or tea" and
or defects."The bibliographies the same type of exposure (caffeine, coffee, or tea) and the
"anomaliesormalformations
of original reports and review articles were searched for
additional published studies. No language restriction was
imposed; case reports, studies based on a small case group (n
< 10), and unpublished manuscripts were not included in the
search. Only one report describing data for any given study
weremultiplereports,
was selected;when there
population
the study with themost recent and complete data was
se
were examinedto determine
lected.The fullstudyreports
same type of malformation. Results pertaining to the same
exposure and outcome combination were available fora small
number of outcomes and mostly forcoffee intake.A narrative
is also presented.
summary
(CIs)
Odds ratios(ORs) and 95% confidenceintervals
were calculated for several studies that did not reportmea
sures of association. Odds ratios forhigh exposure (defined as
3+ servings per day formost studies) versus no exposure
eligibility for inclusion when citation titles or abstracts did
across studiesusing thegeneralvariance
were summarized
based method describedby Petitti.2When available, ad
atmoderate to high levels of exposure and forcertain types of
congenital anomalies rather than all anomalies. In keeping
with the goal of assessing the adequacy of the published
epidemiologic literature, the criteria used to include studies
for review were as follows:
of effect
A generalvariance-basedtestof homogeneity20
homogeneity(Q-test
estimateswas performed;significant
detail.
not containsufficient
wouldmost likelyoccur
effects
of caffeine
Teratogenic
* Caffeine or caffeinated beverage consumption was exam
ined as a risk factor formajor congenital anomalies, defined
as "those that have surgical, medical, or serious cosmetic
importance.",19
*Effectestimates
were presentedforone ormore categories
of congenital anomalies classified by organ system or
subtypes within an organ system; studies with analysis
were excluded.
combined"
limitedto "allmalformations
* Caffeine exposure assessment included intake from one or
beverages
more caffeinated
beverages(becausecaffeinated
are themain sources of caffeine exposure and of greatest
relevanceinassessingriskto thegeneralpopulation).
* Effectestimates
were presentedformoderatetohighex
posure compared with minimal or no exposure. Moderate
to high exposure has customarily been defined as 300 or
more mg of caffeine or 3 or more servings of a caffeinated
beverage. Low exposure is defined here as less than 100 mg
caffeine per day or less than one serving per day. Studies
that combined all levels of exposure, grouped moderate to
high exposures in the referent group, or only contrasted
mean levels of exposure were excluded.
Because bias toward the null value would be expected
to result from extensively grouped exposure or outcome
categories such as "any exposure" or "all malformations," the
use of these groupings ismost problematic in interpretingnull
Studies excludedbased on groupedexposureor
findings.
outcome are summarized in the appendix available with the
online version of this article.
Appraisal of Study Methods
For all studies meeting inclusion criteria, the full study
reportwas examined; study characteristics and results were
abstracted using a standard format. Evaluation of selection
bias, recall bias, misclassification of exposure, and confound
justed odds ratioswere used in themeta-analysis calculations.
P <
0.05) was a contraindication formeta-analysis. Hetero
geneity was also assessed by the I2 statistic.2 A fixed-effects
model was used to estimate the variance of the summary odds
was low; theDerSimonian
ratiowhen studyheterogeneity
and Laird random-effectsmethod was used to estimate the
variance of the summary odds ratio in the presence of
was used
RevMan 4.2 software22
moderateheterogeneity.20
to calculate and graph meta-analysis
results.
RESULTS
Literature Search
Of these,
retrieved
304 citations.
The searchstrategy
therewere 21 original research reports describing exposure to
a source of caffeine during pregnancy and risk of congenital
anomalies. One was a study presented in a letter to the
editor;23 the original study report published in French was
translated to English for this review.24 Three additional cita
tionswere foundin thereference
listsof publishedreview
articles25-27 and one additional study forwhich results were
published in a textbook was identified,28 resulting in 25
selectedstudies.
excluded.
were subsequently
Eighteenof thesestudies
from medications
Three
assessed
caffeine exposure
only,2526'29 and one evaluated risk of minor anomalies.27 An
additional 14 studies were excluded because they did not
evaluate the riskofmoderate to high exposure compared with
low or no exposure (n = 11),30 0- or they presented results
only for all malformations combined (n = 3).24,41i42 (A
summary description of these 14 studies is provided in the
electronic appendix.) Six study reports by Tikkanen and
Heinonen33-3743 all referred to the same study population.
Despite some overlap, all were selected because each con
tributedquantitative information for different cardiovascular
malformation categories. However, 5 of the 6 reports33-37
were ultimately excluded from the full review because expo
sure was categorized as "any versus none." The Boston
University Slone Epidemiology Center Birth Defects Study
? 2006
Williams
&Wilkins
Lippincott
This content downloaded from 169.226.11.193 on Thu, 5 Jun 2014 12:17:54 PM
All use subject to JSTOR Terms and Conditions
325
Browne
Epidemiology * Volume 17, Number 3, May 2006
contributed data for 2 reports; both are included because each
examinedcongenitalanomalies fromdifferent
organ sys
The inclusion
ofpreviouspregnancies
byMcDonald et
a148posed a numberofmethodologicproblems.First,the
tems.44'45
contribution of more than one observation (pregnancy) by a
Methodology of IncludedStudies
For the7 studiesretainedforanalysis,publication
dates
ranged from 1974 to 1993.28,43-48 Four were conducted in the
United States or Canada and 3 in European or Scandinavian
countries. Six were case-control studies28'43-47 and one was
a retrospective
cohortstudy.48
and re
Studycharacteristics
sults are summarized in Table 1.
Response rateswere generally high and comparable for
cases and controls, so nonresponse was not an obvious source
of distortion. Rosenberg et a144 and Werler et a145 used an
affected(malformed)
controlgroup.This approachreduces
the potential for response bias or recall bias thatmay occur
when a nonmalformed control group is used. In each of the
included studies, time to interview was relatively short and
similar for cases and controls (generally less than 1year after
the index birth) with the exception of theMcDonald
et al
study.49 It is possible that recall bias may have contributed to
elevated risk estimates in the studies by Fedrick,46McDonald
et al,48 and Ferencz et al.28 Inconsistent findings for various
et al argue
types ofmajor congenital anomalies inMcDonald
against an important influence of recall bias in that study.
However, increased reportingof high tea consumption among
case mothers living inmedium and high incidence areas but
not in low incidence areas in the Fedrick study46 could have
been the result of heightened concerns about risk factors
among women inmedium and high incidence areas or, as the
authorsuggested,
becauseof anobservedcorrelation
between
water softness and anencephaly. The estimates in Table 1
summarize the findings for the entire study area.
Only 3 of the 7 studies evaluated multiple sources of
caffeine exposure28'43'44 and only one accounted for changes
in caffeine intake during early pregnancy. Ferencz et a128
examined average daily caffeinated beverage intake during
the period from 3 months before through 3 months after the
lastmenstrualperiod,whereas theotherstudiesevaluated
intake
duringpregnancy
only.Poormeasurement
of exposure
is a particular concern in studies of caffeine consumption
duringpregnancy.50
Control of confounding was
limited in some of the
includedstudies.Fedrick46selected control subjectsby
matching on several variables but did not account formatch
ing in the analysis. Only crude comparisons of caffeine
exposure and outcomes were available for the studies by
TikkanenandHeinonen43and Ferenczet al.28Confounding
from factors includingmaternal age and smoking was assessed
in the4 remaining
Statistical
adjustment
by
studies.44'45'47'48
et a148 reduced the excess relative risk associated
McDonald
with 3 ormore cups of coffee per day by 24%, 25%, and 71%
for heart anomalies, oral clefts, and neural tube defects,
respectively
(basedon cruderiskestimatescalculatedby the
presentauthor).Because positiveconfounding
was observed
et al and not in the other 3
only in the study by McDonald
studies,44,4547 it is uncertain whether uncontrolled confound
ing contributed to positive associations
Fedrick46 and Ferencz et al.28
326
in the studies by
study
motherviolates theassumption
of independent
obser
vations in standardstatisticaltests.Second, inaccuraciesin
outcomeand exposureassessment
were likely.For previous
pregnancies,
ascertainment
of congenital
anomaliesreliedon
self-report
and exposureassessmentrequiredrecallof expo
sure that occurred years in the past. McDonald
et al used
separateregression
models toestimateeffectsforcurrent
and
previouspregnancies.
The authorsstatedthattheregression
coefficients
forpreviouspregnancies
were not significantly
different from those for current pregnancies, and they pre
sented odds ratios based on a weighted average of the 2
models.48
Results of Included Studies
Narrative Summary
The odds ratios for the 7 included studies were gener
ally close to the null value. Selected elevations in riskwere
observed in 3 studies. In a study conducted in England and
Wales,
Fedrick46 noted an elevated risk of anencephaly
among women who reported drinking 3 or more cups of tea
per day (odds ratio = 1.6, 95% confidence interval = 1.1
2.4). McDonald
et a148 found an adjusted odds ratio of 1.5
(CI = 1.1-2.2)
for risk of cardiovascular malformations
associated with consumption of 3 or more cups of coffee per
day among women inMontreal. Although the authors of the
InfantStudy28statedthat"no case
Baltimore-Washington
control differences were noted" in caffeinated beverage in
take, crude odds ratios (calculated by this author) support a
dose-response
trend for coffee intake and cardiovascular
malformations; when intake is categorized as 3 or more cups
of coffee per day, the OR was 1.3 (1.0-1.6). An elevated
odds ratio for cardiovascular malformations was also ob
served for 7 or more cups of tea per day (1.7; 1.1-2.8); no
dose-response pattern was present.
A dose-response relationship was generally lacking in
theRosenberg et a144 study findings (only slight support for a
positive trend for cleft lip with or without cleft palate).
However, there is some evidence of invertedU-shaped trends
for inguinal hernia, cardiovascular malformations (CVMs), and
neural tube defects, and an inverse trendfor isolated cleftpalate.
Odds ratios (calculated by this author) fordaily caffeine intake
of none, 1-199 mg, 200-399 mg, and 400 mg ormore were 1.0,
1.6, 1.5, and 1.2 for inguinal hernia; 1.0, 1.8, 1.7, and 1.2 for
CVMs; 1.0, 1.5, 0.9, and 0.6 forneural tube defects; and 1.0, 1.0,
0.8, and 0.5 for isolated cleft palate, respectively. These esti
mates were imprecise as a result of small numbers in the
andhighexposurecategories.
nonexposed
Elevations in relative risk were observed for some of
the anomaly categories other than cardiovascular malforma
tions examined byMcDonald
et al.48 Evidence of a relation
ship between coffee consumption and various congenital
anomalies was absent in the following studies: Kurppa et a147
(oral clefts, skeletal, cardiovascular, or central nervous sys
tem anomalies), Werler et a145 (gastroschisis), and Tikkanen
and Heinonen43 (hypoplastic leftheart syndrome).
? 2006
Lippincott
Williams
&Wilkins
This content downloaded from 169.226.11.193 on Thu, 5 Jun 2014 12:17:54 PM
All use subject to JSTOR Terms and Conditions
Maternal Exposure to Caffeine and Risk of Congenital Anomalies
17, Number 3, May 2006
Epidemiology *Volume
(
0 to
)
0)0
0
c 0 0
0~~~~0
0~~~~~~~
LI
-
~~~0
v 0
o(
~
c
Ch -
o
>0 '
~~~~
0~~~~~~~~~~~
Z0L
0
0 .C))
0
~~~~~~
-0
.~~~~
00
+0
w~
,-
~ ~
0 0
4-
~~~~~~~~
0Cd0
0l
0
(0
0
00)(0o
0
0
o0o0
~
~~~~~0
~~~0~
it
0
~
0~~~~~'~~~~>
0)0000
O.,
,00V.
)
0
)0
0
w
0
.
Cs
A
(
0
rt(
m 00
(1F'
-
0
ON
00 0
0 0 - 0~~~~~~~~~~~~~~~~~~~~~~~~~~~1
0
~~~~~~~~~~~w
o ~~~~~~~~~~
0
v
~~
~~~~~~~
0
~~~~~~
'g
0
(0W
0),
~
(0
a!
+0
~~~~~
>~~~~~~~~~
C)
~
N
0
0
Q)
~~~~~-0
(0
C,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0
w
+
NO ~~~
0) u
)
O
000000
-, -,
0
0 0(,
0
N
-- -
--e
l
0
0
0
0
0
0
0
0
-
-
O
0)r-0
S
00~~~~~~~
(00)
cn
0~~~00
0)
>
0z
~0
C
(0
o
-Z
o =
0~.
-o00
4-
E
0)
4u.
o+-~ 0=
00oo
0
0
-o
~~~
C
0)
U ~~~~ 0~~~~~~~~~~~
~
0
4 -o
0
0
*
0)0
~~~~~~~~~0
rA0
0
N~~~~~~~~~~~~~~~~0
((0~~~~~~~~~~~~~~~~~~~~l
v)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
o(0
.
00)(0 -O
Cu
.
0
cl
(C
~~
0
S
0) ~~~~~
(0
~
0
>~~~~~~~~~
o
U
la.)
~
(0
oS~o
u~~0~-0)
C-~~~~~~~~~~~~~~~(,0
o ~~~~~~~~~~N
'
4
00,
.
(
0~~~~~~~~~~~~0
0~~~~~~~~:
0
0
(O
0,
(0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~'
0
0.
00)0u
0~~~'Cl-N-Cl
Nt
cn
0~~~~~~~~O
0
C
E
2
-~~~~~~~~~~~~~~~0
z 00
E
E
)0~C1~
IC
Zo
u
CA
~
~
~
~
0
0
0~~~~~~~~~~~~~~~~~~~~~-0
00
(n0
cnC
~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
(0
r
N
oo~~(0 ~
0~~~~~
0
This content downloaded from 169.226.11.193 on Thu, 5 Jun 2014 12:17:54 PM
All use subject to JSTOR Terms and Conditions
0*
0)
5
E
0 00
cd
000)00
Epidemiology * Volume 17, Number 3, May 2006
Browne
o--60
2
ct
~4-
~~~~~~~ ~
~
00
0004
~
~
Q
0
~
~
~
r
C
C)O~
~~~
~~~
r- 00~~~~~~
0-0r
0~~ 0
0.50
N
CZ~
E~
~~~
.
v)0v)CO
0 ~~~~~
40
-
~
700
-.
NC
C)
Cj
0)
)
C
EI
E
O O15
to000
0
' V- 0
cl 0
UC '
2
~~~~~
Cu
U
0
E.
0~~~~~50U0~~~~~~~~~~~~~CZZ
0
to
~~Cu
~~~~~
CZ
~~~~~~~~~~~0U
S00
0Cz0
U0-OC
U CZ
0
0
CO
0
00
0
CZ
0)~~~~~~~~~~~~~~~~~~~~~~~>VC
C0~~~~~~~~~~~EC
0
to
C)
0 ~~~~~~~~~~~~~~~~~~~~~~~.
0i6
-
U-
--
0
0
CO
0)
U
C
Uu
cO
0U)0
00
0
C)
~
~
U~~~~~~0
0)0~~~~~~~00U
0
0CO
Cu
) C)c
Cd
C
r_ I. 00
:3
U
~0C
~
~
~
~
~
~
~
~
~
~
~
O
C
This content downloaded from 169.226.11.193 on Thu, 5 Jun 2014 12:17:54 PM
All use subject to JSTOR Terms and Conditions
c
Epidemiology * Volume
Matemal Exposure to Caffeine and Risk of Congenital Anomalies
17, Number 3, May 2006
Quantitative Synthesis
Odds ratios for single types of exposure (caffeine,
coffee, or tea) and specific malformations were summarized
throughmeta-analysis methods. There were overlapping ex
combinations for coffee and neural tube
posure-outcome
defects (2 studies), coffee and oral clefts (3 studies), coffee
and cardiovascular malformations (4 studies), and tea and
cardiovascular malformations (2 studies).
Summary odds ratios were calculated for risk of high
exposure, defined as 3 ormore servings per day (4 ormore for
Kurppa et a147 estimates) versus no exposure (Fig. 1). A test
of homogeneity was of borderline statistical significance for
studies of coffee and neural tube defects; therefore, a sum
mary estimate was not calculated. Although theQ-test was
not statistically significant, 2 of the 4 risk estimates for the
association between coffee and cardiovascular malformations
were below the null value and 2 were above the null value;
the 12 statistic (52%) indicated moderate heterogeneity. A
summary estimate using a random-effects variance model is
presented with reservation. Slight elevations were observed
for associations between coffee intake and cardiovascular
malformations (OR = 1.1; CI = 0.9-1.5) and coffee intake
and oral clefts (1.2; 0.9-1.6), but not for the association
between tea and cardiovascular malformations (1.0; 0.9-1.2).
There is limited ability to evaluate study heterogeneity
given the few studies contributing to each summary estimate.
Even so, recall bias is a possible source of heterogeneity in
the estimates for the association between coffee and cardio
vascular malformations. The Rosenberg et a144 study took
place before the 1981 U.S. Federal Drug Administration
andCVMs
Coffee
andNTDs
Coffee
198244
Rosenberg
Kurppa198347
McDonald199248
199328
Ferencez
Total
a
-
-
198244
Rosenberg
McDonald199248
andoralclefts Rosenberg
Coffee
198244 47.4
Kurppa198347
McDonald199248
Total
Tea andCVMs
weight
Analysis
OR (95%CI)
%
OR and95%CI
Study
1.3
_
(0.9-1.9)
Rosenberg
198244
199328
Ferencz
Total
0.2
0.5
1
2
20.2
12.1
23.5
44.2
100.0
0.9(0.6-1.3)
0.8(0.4-1.4)
1.5(1.1-2.2)
1.3(1.0-1.6)
1.1(0.9-1.5)
31.7
0.4
68.3
(0.2-1.0)
1.1(0.7-1.8)
33.8
18.7
100.0
1.0(0.6-1.6)
1.4(0.7-2.7)
1.2(0.9-1.6)
26.4
73.6
100.0
1.1(0.7-1.6)
1.0(0.9-1.2)
1.0(0.9-1.2)
5
FIGURE 1. Meta-analysis results for consumption of specific
caffeinated beverages (high consumption vs. none) and se
lected congenital anomalies. High consumption defined as 4+
cups per day for theKurppa et al study47 and 3+ cups per day
for each of the other studies. Test forheterogeneity forcoffee
and CVMs: P = 0.10; 12= 52%. Test for heterogeneity for
coffee and NTDs: P = 0.05; 12= 73%. Test forheterogeneity
for coffee and oral clefts: P = 0.65; 12= 0%. Test for hetero
geneity for tea and CVMs: P = 0.92; 12= 0%.
(FDA) advisory51 to limit caffeine intake during pregnancy.
In addition, the use of affected controls was expected to
reduce recall bias. Kurppa et a147 stated that public concern
about harmful effects of caffeine did not exist in Finland at
the time of their study. For these reasons, recall bias was
unlikely for the 2 studies with effect estimates below the null
value. The 2 studies with risk estimates above the null value
took place after the FDA advisory and may have been more
vulnerable to recall bias.
DISCUSSION
Studies with extensive grouping by exposure and out
come were not selected for review. Classifying caffeine
intake as "any versus none" is problematic because grouping
low exposures with high exposures could hide a true effect
thathappens only at high levels of exposure. Different types
of congenital anomalies have diverse etiologies; grouping all
anomalies as a single outcome can mask an effect thatoccurs
for only particular categories of anomalies.52 With a few
exceptions, the studies excluded for these reasons did not find
associations between caffeine intake and congenital anoma
lies. In the few instances inwhich a positive association was
observed,24'30 selection bias and confounding were possible
alternative explanations.
The 7 included studies were well-conducted studies and
provide the best available epidemiologic evidence about caf
feine teratogenicity. Inaccuracies in exposure assessment and
the relatively small number of specific exposure-outcome
relationships that have been studied limit conclusions based
on these data. Meta-analysis was possible for associations
between coffee and oral clefts, coffee and cardiovascular
malformations, and tea and cardiovascular malformations.
Summary estimates could not be calculated for total caffeine
intake (versus coffee or tea consumption) because only one of
the 7 included studies reported findings for caffeine exposure
frommultiple sources. The few exposure-outcome combina
tions suitable forquantitative synthesis emphasize the relative
scarcity of published research on this topic. The possibility of
publication bias is a concern in interpreting the results re
stricted to published reports.Findings of littleor no increased
risk in early studies of this topicmay have reduced intereston
the part of both investigators and editors in publishing addi
tional studies with null results.
The results of the studies included in this review sug
gest that a large increase in the risk of congenital anomalies
is unlikely to result from consumption of caffeinated bever
ages during pregnancy. The 10% to 20% excess risk associ
ated with coffee intake suggested by the summary estimates
for cardiovascular malformations and oral clefts would be
important if the relative risks reflected a true effect of coffee
consumption. However, these small increases might also be
explained by recall bias or confounding. Conversely, sources
of nondifferential exposure misclassification were likely to
have attenuated risk estimates. Therefore, these slight in
creases and thenull results forother anomaly categories may
warrant furtherattention. The few available studies for each
anomaly category do not provide adequate negative evidence
regarding risk.
? 2006
Williams
&Wilkins
Lippincott
This content downloaded from 169.226.11.193 on Thu, 5 Jun 2014 12:17:54 PM
All use subject to JSTOR Terms and Conditions
329
*Volume
17,
3,May
2006
Epidemiology
Number
Browne
Misclassification of exposure is a special concern in
studies of caffeine exposure. Wide variations in the caffeine
content of a beverage serving53 and changes in caffeine
difficulties.
duringpregnancy54
are2 prominent
consumption
in caffeine
metabo
differences
In addition,interindividual
lism represent a source of error inmeasuring caffeine expo
sure. The rate of caffeine clearance among smokers is ap
proximately twice that in nonsmokers as a result ofmetabolic
GrossoandBracken50
by cigarette
smoke.55'56
enzymeinduction
of sourcesofmis
recently
provideda detaileddescription
classification
inmeasuringcaffeineintakeduringpregnancy.
as
measurement
of caffeine
biomarkers
The authors
proposed
methods.
of questionnaire-based
a remedyto thedeficiencies
Unfortunately, prospective studymethods are not practical in
studiesof congenitalanomalies.However,validationstudies
measured exposure information
comparingprospectively
estimatescollectedafterdelivery
with questionnaire-based
in retrospective
uncertainty
would be helpfulin estimating
study results and in guiding development of improved expo
methods.
sureassessment
A listof recommendations
forfuture
researchfollows:
* Collect information on multiple sources of caffeine. To
separate any effect of caffeine from that of other compo
nentsof coffee,tea,and soda,analyzeexposure-outcome
beveragesinaddi
forindividualcaffeinated
relationships
tion to total caffeine.
*Avoid extensivegroupingof different
de
malformations;
finemalformation outcome categories thatare "embryolog
icallyor pathogenetically
meaningful."52
*Throughvalidationstudies,identify
methods
questionnaire
thatcould improve exposure assessment accuracy. For exam
ple, questions on brew strength(tea and coffee), portion size,
and changes in caffeine consumption during the period start
through
thefirsttrimester.
ingbeforepregnancy
* In the analysis, form categories of exposure
that allow
assessmentof dose-response;avoid groupinghighexpo
sures with low exposures.
* Assess potential effectmodification by smoking status and
other factors that affect caffeine metabolism.
*To reduceconcernsaboutspuriousfindings
from
multiple
to
analyses, use a priori relationships and mechanisms
guide the choice of outcome categories and potential effect
modifiers. However, exploratory analyses in the absence of
mechanistic hypotheses, when reported as such, are also
useful in guiding subsequent research efforts.
specific categories of congenital anomalies
exposuretocaffeine.
M ENTS
ACKNOWLEDG
The author thanks
AprilAustin,Erin Bell, Charlotte
Druschel,Lenore Gensburg,and Paul Romittifor helpful
comments
on earlierdraftsof this
manuscript.
REFERENCES
1. Daly
JW.Mechanism
of action of caffeine. In: Garattini S, ed. Caffeine,
Coffee, and Health. New York: Raven Press, Ltd; 1993:97-150.
2. Mandel HG. Update on caffeine consumption, disposition and action.
Food Chem Toxicol. 2002;40:1231-1234.
3. Frary CD, Johnson RK, Wang MQ. Food sources and intakes of caffeine
in the diets of persons in theUnited States. J Am Diet Assoc. 2005;105:
110-113.
4. Dlugosz
caffeine con
L, Belanger K, Hellenbrand K, et al. Maternal
sumption and spontaneous abortion: a prospective cohort study. Epide
miology. 1996;7:250-255.
con
5. Nehlig A, Debry G. Potential teratogenic and neurodevelopmental
sequences of coffee and caffeine exposure: a review on human and
animal data. Neurotoxicol
Teratol. 1994;16:531-543.
6. Scott WJ, Jr.Caffeine-induced
limb malformations: description of mal
formations and quantitation of placental
427-435.
7. Kirkinen
placental
P, Jouppila P, Koivula
and fetal blood flow
SUMMARY
330
transfer. Teratology.
1983;28:
A, et al. The effect of caffeine on
in human pregnancy. Am J Obstet
1983;147:939-942.
Gynecol.
8. Salvador HS, Koos BJ. Effects of regular and decaffeinated coffee on
fetal breathing and heart rate. Am J Obstet Gynecol.
1989; 160:1043
1047.
B. Coffee consumption and human health?
T, Halvorsen
for effects of coffee consump
beneficial or detrimental??Mechanisms
tion on different risk factors for cardiovascular
disease and type 2
9. Ranheim
diabetes mellitus. Mol Nutr Food
Res. 2005;49:274-284.
Fetal vascular disruptions: mechanisms
and some result
ing birth defects. Pediatr Ann. 1981;10:219-233.
11. Hobbs CA, Cleves MA, Melnyk S, et al. Congenital heart defects and
10. Van Allen MI.
abnormal maternal biomarkers of methionine
olism. Am J Clin Nutr. 2005;81:147-153.
12. Loeken MR.
and homocysteine metab
Free radicals and birth defects. J Matern
Fetal Neonatal
Med.
2004;15:6-14.
13. Golding J.Reproduction and caffeine consumption?a
literature review.
1995;43:1-14.
Early Hum Dev.
14. Christian MS, Brent RL. Teratogen update: evaluation of the reproduc
tive and developmental risks of caffeine. Teratology. 2001;64:51-78.
15. Levit?n A, Cowan L. A review of the literature relating caffeine
to their risk of reproductive hazards. Food
consumption by women
Chem Toxicol. 2002;40:1271-1310.
16. Nawrot P, Jordan S, Eastwood
J, et al. Effects of caffeine on human
health. Food Addit Contam. 2003;20:1-30.
17. Mulrow CD. The medical review article: state of the science. Ann Intern
Med.
1987;106:485-488.
DL. Methodologie
Inst. 1997;89:6-7.
19. Rasmussen
SA, Olney
18. Weed
The 7 studies included in this review do not provide
adequate evidence on which to base conclusions about the
safety of maternal caffeine consumption during pregnancy.
There is no evidence that caffeine intake causes a large
increase in the risk of various types of congenital anomalies,
but there is greater uncertainty about small elevations in risk.
Given the relatively high prevalence of maternal caffeine
exposure, even a small increase in the risk of congenital
anomalies would have an important effect on public health.
Large study populations and improved exposure assessment
methods would be necessary to rule out small risks for
after maternal
classification
guidelines
for review papers. J Nati
Cancer
for case
RS, Holmes
LB, et al. Guidelines
for the National Birth Defects Prevention Study. Birth
Defects Res A Clin Mol Teratol. 2003;67:193-201.
Decision
20. Petitti DB. Meta-Analysis,
Analysis, and Cost-Effectiveness
2nd ed. New
Analysis: Methods for Quantitative Synthesis inMedicine,
York: Oxford University Press; 2000.
21. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analy
sis. Stat Med. 2002;21:1539-1558.
22. Review Manager
(RevMan)
[computer program]. Version 4.2 forWin
dows. Oxford: The Cochrane Collaboration;
2003.
23. Lechat MF, Borlee
I, Bouckaert A, et al. Caffeine study. Science.
1980;207:1296-1297.
24. Borlee I, Lechat MF, Bouckaert A, et al. Le cafe, facteur de risque
1978;97:279-284.
pendant la grossesse? Louvain Med.
( 2006
Williams
&Wilkins
Lippincott
This content downloaded from 169.226.11.193 on Thu, 5 Jun 2014 12:17:54 PM
All use subject to JSTOR Terms and Conditions
Epidemiology * Volume
17, Number 3, May 2006
25. Nelson MM, Forfar JO. Associations between drugs administered during
pregnancy and congenital abnormalities of the fetus.BMJ. 1971;1:523-527.
26. Heinonen
in Preg
OP, Slone D, Shapiro S. Birth Defects and Drugs
nancy. Littleton: Publishing Sciences Group; 1977.
27. Termes K, Blackard C. Maternal alcohol consumption, birth weight, and
minor physical anomalies. Am J Obstet Gynecol.
1980;138:774-780.
28. Ferencz C, Rubin JD, Loffredo CA, et al. Epidemiology
of Congenital
Heart Disease:
the Baltimore-Washington
Infant Study, 1981-89.
Mount Kisco, NY: Futura Publishing Co; 1993.
29. Samren EB, van Duijn CM, Christiaens GC, et al. Antiepileptic drug
in the offspring. Ann
regimens and major congenital abnormalities
Neurol.
1999;46:739-746.
30. Furuhashi N, Sato S, Suzuki M, et al. Effects of caffeine ingestion during
pregnancy. Gynecol Obstet Invest. 1985;19:187-191.
31. Niebyl JR, Blake DA, Rocco LE, et al. Lack of maternal metabolic,
endocrine, and environmental influences in the etiology of cleft lipwith
or without cleft palate. Cleft Palate J. 1985;22:20-28.
32. Adams MM, Mulinare J,Dooley K. Risk factors for conotruncal cardiac
defects inAtlanta. J Am Coll Cardiol.
1989;14:432-442.
33. Tikkanen
J,Heinonen OP. Risk factors for cardiovascular malformations
in Finland. Eur J Epidemiol.
1990;6:348-356.
J,Heinonen OP. Maternal exposure to chemical and physical
in the off
factors during pregnancy and cardiovascular malformations
spring. Teratology. 1991;43:591-600.
35. Tikkanen J,Heinonen OP. Risk factors for ventricular septal defect in
34. Tikkanen
Finland. Public Health.
1991;105:99-112.
OP. Risk factors for atrial septal defect. Eur J
1992;8:509-515.
Epidemiol.
37. Tikkanen J,Heinonen OP. Risk factors for conal malformations of the
36. Tikkanen
J,Heinonen
heart. Eur J Epidemiol.
1992;8:48-57.
38. Torfs CP, Christianson RE. Maternal risk factors and major associated
defects in infants with Down
1999; 10:264
syndrome. Epidemiology.
270.
39. Fixier DE, Threlkeld N. Prenatal exposures and congenital heart defects
in Down syndrome infants. Teratology.
1998;58:6-12.
40. Natsume N, Kawai T, Ogi N, et al. Maternal risk factors in cleft lip and
palate: case control study. Br J Oral Maxillofac
Surg. 2000;38:23-25.
41. Linn S, Schoenbaum
SC, Monson RR, et al. No association between
coffee consumption and adverse outcomes of pregnancy. N Engl JMed.
1982;306:141-145.
Maternal Exposure to Caffeine and Risk of Congenital Anomalies
42. Olsen
J,Overvad K, Frische G. Coffee consumption, birthweight, and
1991;2:370-374.
reproductive failures. Epidemiology.
43. Tikkanen J,Heinonen OP. Risk factors for hypoplastic left heart syn
drome. Teratology. 1994;50:112-117.
44. Rosenberg L, Mitchell AA, Shapiro S, et al. Selected birth defects
relation to caffeine-containing
JAMA.
1982;247:1429
beverages.
1432.
in
45. Werler MM, Mitchell AA, Shapiro S. Demographic,
reproductive, med
ical, and environmental factors in relation to gastroschisis. Teratology.
1992;45:353-360.
tea drinking: evidence for a
46. Fedrick J. Anencephalus
and maternal
1974;67:356-360.
possible association. Proc R Soc Med.
47. Kurppa K, Holmberg PC, Kuosma E, et al. Coffee consumption during
a nationwide case
pregnancy and selected congenital malformations:
control study. Am J Public Health.
1983;73:1397-1399.
AD, Armstrong BG, Sloan M. Cigarette, alcohol, and coffee
48. McDonald
consumption
91-93.
and
congenital
defects. Am J Public
Health.
1992;82:
and
AD, McDonald
JC, Armstrong B, et al. Occupation
1987;44:521-526.
pregnancy outcome. Br J Ind Med.
50. Grosso LM, Bracken MB. Caffeine metabolism, genetics, and perinatal
49. McDonald
outcomes: a review of exposure assessment considerations during preg
nancy. Ann Epidemiol.
2005;15:460-466.
51. US Department of Health and Human Services. Caffeine and Pregnancy.
(FDA) 81-1081. Public Health Service, Food and Drug Administration;
1981.
et al. Interpretation of recurring
52. Khoury MJ, James LM, Flanders WD,
weak associations
obtained from epidemiologic
studies of suspected
human teratogens. Teratology. 1992;46:69-77.
53. Bracken MB, Triche E, Grosso L, et al. Heterogeneity
in assessing
self-reports of caffeine exposure:
implications for studies of health
effects. Epidemiology.
2002;13:165-171.
54. Pirie PL, Lando H, Curry SJ, et al. Tobacco, alcohol, and caffeine use
and cessation in early pregnancy. Am J Prev Med. 2000;18:54-61.
55. Burg AW.
Physiological
disposition of caffeine. Drug Metab
1975;4:199-228.
56. Landi MT, Sinha R, Lang NP et al. Human cytochrome P4501A2.
SciPubl.
1999;173-195.
? 2006 LippincottWilliams & Wilkins
This content downloaded from 169.226.11.193 on Thu, 5 Jun 2014 12:17:54 PM
All use subject to JSTOR Terms and Conditions
Rev.
IARC
331