DOI: 10.1542/peds.2006-0124
2007;120;e944-e952 Pediatrics
Hiroko Iida, Peggy Auinger, Ronald J. Billings and Michael Weitzman
United States
Association Between Infant Breastfeeding and Early Childhood Caries in the
http://www.pediatrics.org/cgi/content/full/120/4/e944
located on the World Wide Web at:
The online version of this article, along with updated information and services, is
rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Grove Village, Illinois, 60007. Copyright © 2007 by the American Academy of Pediatrics. All
and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk
publication, it has been published continuously since 1948. PEDIATRICS is owned, published,
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly
by on February 24, 2011 www.pediatrics.orgDownloaded from
ARTICLE
Association Between Infant Breastfeeding and Early
Childhood Caries in the United States
Hiroko Iida, DDS, MPHa, Peggy Auinger, MSb, Ronald J. Billings, DDS, MSDa, Michael Weitzman, MDc,d
Departments of aDentistry and bPediatrics, School of Medicine and Dentistry, University of Rochester, Rochester, New York; cDepartment of Pediatrics, School of
Medicine, New York University, New York, New York; dAmerican Academy of Pediatrics Julius B. Richmond Center of Excellence
The authors have indicated they have no financial relationships relevant to this article to disclose.
ABSTRACT
OBJECTIVE. Despite limited epidemiologic evidence, concern has been raised that
breastfeeding and its duration may increase the risk of early childhood caries. The
objective of this study was to assess the potential association of breastfeeding and
other factors with the risk for early childhood caries among young children in the
United States.
METHODS. Data about oral health, infant feeding, and other child and family char-
acteristics among children 2 to 5 years of age (N � 1576) were extracted from the
1999–2002 National Health and Nutrition Examination Survey. The association of
breastfeeding and its duration, as well as other factors that previous research has
found associated with early childhood caries, was examined in bivariate analyses
and by multivariable logistic and Poisson regression analyses.
RESULTS. After adjusting for potential confounders significant in bivariate analyses,
breastfeeding and its duration were not associated with the risk for early childhood
caries. Independent associations with increased risk for early childhood caries were
older child age, poverty, being Mexican American, a dental visit within the last
year, and maternal prenatal smoking. Poverty and being Mexican American also
were independently associated with severe early childhood caries, whereas char-
acteristics that were independently associated with greater decayed and filled
surfaces on primary teeth surfaces were poverty, a dental visit within the last year,
5 years of age, and maternal smoking.
CONCLUSIONS. These data provide no evidence to suggest that breastfeeding or its
duration are independent risk factors for early childhood caries, severe early
childhood caries, or decayed and filled surfaces on primary teeth. In contrast, they
identify poverty, Mexican American ethnic status, and maternal smoking as in-
dependent risk factors for early childhood caries, which highlights the need to
target poor and Mexican American children and those whose mothers smoke for
early preventive dental visits.
www.pediatrics.org/cgi/doi/10.1542/
peds.2006-0124
doi:10.1542/peds.2006-0124
KeyWords
breastfeeding, early childhood caries,
maternal smoking
Abbreviations
ECC—early childhood caries
S-ECC—severe early childhood caries
NHANES—National Health and Nutrition
Examination Survey
dfs—decayed and filled surfaces on
primary teeth
aOR—adjusted odds ratio
CI—confidence interval
IDR—incidence density ratio
FPL—federal poverty level
Accepted for publication Mar 14, 2007
Address correspondence to Hiroko Iida, DDS,
MPH, Bureau of Dental Health, New York State
Department of Health, ESP, Corning Tower,
Room 542, Albany, NY 12237-0619. E-mail:
hiroko.iida@gmail.com
PEDIATRICS (ISSN Numbers: Print, 0031-4005;
Online, 1098-4275). Copyright © 2007 by the
American Academy of Pediatrics
e944 IIDA et al
by on February 24, 2011 www.pediatrics.orgDownloaded from
THE AMERICAN ACADEMY of Pediatrics identifies hu-man milk as the ideal nutrient for infants1 on the
basis of the extensive scientific evidence demonstrating
that breastfeeding and the use of human milk provide
multiple health-related advantages to infants, mothers,
and society.1,2 Breastfeeding is recommended by pedia-
tricians and other health care professionals to be contin-
ued for at least the first year of life and beyond, for as
long as mutually desired by mother and child.1 Pro-
longed and unrestricted breastfeeding, however, has
been reported to be a potential risk factor for early
childhood caries (ECC),3–6 and a recent animal study, the
results of which were recently published in this journal,
found breast milk to be more cariogenic than bovine
milk.7 However, epidemiologic evidence linking infant
breastfeeding and its duration and ECC in children is
very limited. The purpose of this study was to use na-
tionally representative data about children to assess the
potential association of breastfeeding and its duration, as
well as the association of other factors that there is
reason to believe may contribute to ECC, with the risk
for ECC among young children in the United States.
MATERIALS ANDMETHODS
Data from the 1999–2002 National Health and Nutrition
Examination Survey (NHANES),8 a cross-sectional sur-
vey conducted by the National Center for Health Statis-
tics, Centers for Disease Control and Prevention, were
analyzed for 1576 children 2 to 5 years of age with
information on both infant feeding and oral health. The
NHANES includes a household interview with informa-
tion regarding numerous aspects of children’s diet, nu-
trition and oral health behavior, and family socioeco-
nomic characteristics completed by the person most
knowledgeable about the child, usually the mother. Re-
sults of dental examinations also are included as part of
this survey.
Infant Breastfeeding Data
Information about infant feeding was obtained from par-
ents/guardians of children during an in-person inter-
view based on retrospective recall. The definition of the
various breastfeeding categories used in this study was
based on the schema developed by the Interagency
Group for Action on Breastfeeding in 1988.9 Breastfeed-
ing and its duration (overall, full, and exclusive) were
examined by using the following criteria: whether the
child was ever breastfed (history of breastfeeding), the
age when the child completely stopped breastfeeding or
being fed breast milk (overall breastfeeding duration),
the age when the child was first fed something other
than breast milk or water (exclusive breastfeeding dura-
tion), and the age when the child was first fed formula,
milk, or solid foods on a daily basis (full breastfeeding
duration). Children who consumed only breast milk,
with or without the consumption of water, were in-
cluded in the assessment of exclusive breastfeeding du-
ration although, most conservatively, exclusive breast-
feeding often refers to the period when infants do not
consume anything other than breast milk. This decision
was based on limitations of the NHANES data and the
fact that there are no data to suggest that the consump-
tion of water is a risk factor for dental caries.
ECC, Severe ECC, and Decayed or Filled Primary Tooth Surfaces
Count Data
A dental examination was performed for children aged
�2 years by a trained and calibrated dentist in a medical
examination center. Dental caries, that is, decayed or
filled primary tooth surfaces (dfs), was assessed by
means of a visual/tactile examination without radio-
graphs.10 The American Academy of Pediatric Dentistry
defines ECC and severe ECC (S-ECC) as follows11: ECC is
the presence of �1 decayed, missing (because of caries),
or filled primary tooth surfaces (dmfs) in any primary
tooth in a child �71 months of age. S-ECC is defined as
any sign of smooth-surface caries in children�3 years of
age; �1 cavitated, missing (because of caries), or filled
smooth surface in primary maxillary anterior teeth from
ages 3 through 5 years; or the presence of �1 decayed,
missing (because of caries), or filled primary tooth sur-
faces of�4 at age 3 years,�5 at age 4 years, or�6 at age
5 years. The reasons for missing primary teeth were not
identified in the NHANES data. Hence, in this study,
ECC refers to the presence of any dfs on any primary
tooth and S-ECC refers to the presence of dfs on any
maxillary incisor in children 2 to 5 years of age. The total
dfs count was used as a measure of disease severity.
Other Variables Investigated
Other potential associations with ECC, S-ECC, and dfs
count were analyzed, including birth weight, age, gen-
der, race/ethnicity, poverty status, maternal age at
child’s birth, maternal history of smoking during preg-
nancy, history of admission to a NICU, and time since
last dental visit. Information about these factors was
obtained through the household interview. As discussed
further in “Discussion,” each of these factors was in-
cluded because of previous research indicating their as-
sociation with dental caries or breastfeeding.12–17
Statistical Analyses
The prevalence of children with ECC, S-ECC, and mean
dfs count was calculated to assess the association of
caries with the history and duration of breastfeeding
(overall, full, and exclusive) and selected other factors.
�2 tests, t tests, and analyses of variance were performed
to assess statistical significance. Logistic regression mod-
els were performed with the outcomes of ECC and S-
ECC (any versus none) to assess the independent asso-
ciation of breastfeeding while controlling for potential
confounders found in the bivariate (unadjusted) analy-
PEDIATRICS Volume 120, Number 4, October 2007 e945
by on February 24, 2011 www.pediatrics.orgDownloaded from
ses (P � .10). Similarly, Poisson regression models were
analyzed with dfs count as the outcome. Adjusted odds
ratios (aORs) and incidence density ratios (IDRs) were
calculated for each, respectively. The IDR is the ratio of
the incidence rate among exposed to that of unexposed
children. An IDR of unity indicates that a covariate is not
associated with the dependent variable (dfs count). All of
the potential confounders included in the models were
categorical variables; therefore, each AOR and IDR com-
pares the association of a particular category to a refer-
ence category for each covariate. Interactions with
breastfeeding were included in secondary analyses;
however, results are based on the models with only the
main effects. Because the NHANES uses a complex, mul-
tistage sampling design, SUDAAN software (Research
Triangle Institute, Research Triangle Park, NC)18 was
used to estimate appropriate variances for all of the
analyses, including bivariate analyses and multivariable
logistic and Poisson regressions.19 Results were weighted
to be representative of 2- to 5-year-olds in the United
States according to the sampling weights that are pro-
vided by the NHANES.19
RESULTS
Not shown in the tables are the following findings about
sociodemographic characteristics of children according
to breastfeeding category: children reported to have
been breastfed were more likely to be nonblack, living at
or above 200% of the federal poverty level (FPL), and
born to older mothers who were less likely to report
smoking during pregnancy (P � .001 for each). Those
exclusively breastfed for �9 months were more likely to
be normal birth weight and Mexican American, whereas
children who were breastfed overall for a year or longer
were more likely to be Mexican American and born to
mothers who did not smoke during pregnancy (P � .05
for each). The majority of breastfed children (�75%)
were introduced to something other than breast milk or
water by age 3 to 6 months.
Sample characteristics and the results of bivariate
analyses are shown in Table 1. Overall, 27.5% of 2- to
5-year-old-children had ECC (any dfs), and 10% had
S-ECC (any dfs on maxillary incisors). Approximately
60% of children were reported as having ever been
breastfed, and overall such children had lower rates of
ECC and S-ECC compared with those never breastfed.
Children breastfed overall for �1 year were more likely
to experience ECC than children who were breastfed for
�1 year (32.8% vs 22.5%; P � .01; data not shown in
tables), whereas there was no statistically significant dif-
ference in rates of ECC between those exclusively
breastfed for �9 months compared with those exclu-
sively breastfed for �9 months (19.5% vs 25.4%; P �
.36; data not shown in tables).
ECC rates increased with age, whereas S-ECC rates
did not. Family income and child race/ethnicity both
were associated with ECC and S-ECC, with rates of both
highest among those living below the FPL and lowest
among those living at �200% of the FPL. Mexican
American children had the highest rates of both ECC and
S-ECC, followed by non-Hispanic black children and
non-Hispanic white children, with children of other
race/ethnicities having the lowest rates. Prenatal mater-
nal smoking was associated with increased rates of ECC
but not with rates of S-ECC. Overall, 36.3% of children
who had a dental visit in the last year had ECC, whereas
the percentage of children who had ECC and did not
have a dental visit within the last year was only 18.5%.
In logistic regression analyses conducted to identify
factors independently associated with ECC (Table 2), a
history of ever having been breastfed was not associated
with rates of ECC. In contrast, independent risks for ECC
were increased child age, Mexican American ethnicity,
living below the FPL, maternal prenatal smoking, and
having had a dental visit within the last year.
Factors independently associated with an increased
risk for S-ECC are shown in Table 3 and include Mexican
American ethnicity and living at �200% of the FPL,
whereas a history of ever having been breastfed was not
associated with rates of S-ECC. Table 4 shows results of
multivariable analyses that demonstrate that breastfeed-
ing duration, whether overall, full, or exclusive, is not
associated with either reduced or increased risk of ECC
or S-ECC.
Overall, the mean dfs count was 2.4. Poisson regres-
sion models that included the child’s birth weight, age,
race/ethnicity, poverty status, maternal age at child’s
birth, maternal smoking during pregnancy, and time
since last dental visit were used to assess the association
between infant breastfeeding and dfs counts (Table 5).
History of breastfeeding or breastfeeding duration of any
type again were not significantly associated with DFS
counts, whereas being 5 years old, living below the FPL,
maternal smoking during pregnancy, and having a den-
tal visit in the past year each were independently asso-
ciated with increased dfs counts. Children’s birth weight,
race/ethnicity, and maternal age at child’s birth were not
associated with increased numbers of caries.
Additional regression analyses were performed to at-
tempt to better explicate the relationships between
breastfeeding status and other factors investigated for
associations with ECC, as well as for interaction effects.
Breastfeeding was associated with a 40% reduced risk
for ECC (aOR: 0.6; 95% confidence interval [CI]: 0.4–
0.9; data not shown in tables) when poverty status,
maternal age at child’s birth, and maternal prenatal
smoking were excluded from the full model used in
Table 2 and similarly was associated with a 40% de-
crease in the risk for S-ECC (aOR: 0.6; 95% CI: 0.4–0.9;
data not shown in tables) when poverty status and ma-
ternal age at child’s birth were removed from the full
model used in Table 3. When accounting for interactions
e946 IIDA et al
by on February 24, 2011 www.pediatrics.orgDownloaded from
TABLE 1 Distribution of US Children Aged 2 to 5 Years by Study Variables for ECC and S-ECC (Unadjusted Analysis)
Variable ECC S-ECC
Sample
Size
Prevalence, % P Sample
Size
Prevalence, % P
Overall 1576 27.5 1503 10.0
History of breastfeeding 1568 .03 1495 .02
Ever 939 24.9 899 8.2
Never 629 32.3 596 13.5
Overall breastfeeding duration 1562 .01 1489 .02
�1 y 216 32.8 210 11.9
6 mo to�1 y 274 20.6 265 5.4
4 to�6 mo 105 24.9 97 9.6
1 to�4 mo 233 22.2 222 6.0
�1 mo 105 26.6 99 11.6
0 629 32.3 596 13.5
Exclusive breastfeeding duration, mo 1561 .054 1488 .049
�9 57 19.5 55 7.2
6 to�9 166 29.5 160 8.5
3 to�6 311 23.6 299 6.9
�3 398 24.8 378 9.3
0 629 32.3 596 13.5
Full breastfeeding duration, mo 1562 .04 1489 .03
�9 55 25.5 53 9.6
6 to�9 179 27.2 174 7.5
3 to�6 304 24.8 286 6.4
�3 395 23.8 380 9.8
0 629 32.3 596 13.5
Birth weight, g 1517 .63 1448 .02
Very low (�1500) 24 17.3 23 2.2
Low (1500–2499) 124 29.2 118 9.1
Normal (�2500) 1369 27.6 1307 10.1
Age, y �.001 .22
2 495 10.9 487 7.6
3 381 20.9 370 10.1
4 362 34.4 347 10.8
5 338 44.3 299 11.7
Gender .49 .99
Male 793 28.7 751 10.0
Female 783 26.3 752 10.0
Race/ethnicity �.001 �.001
Non-Hispanic white 489 25.3 477 7.9
Non-Hispanic black 456 31.8 427 13.9
Mexican American 478 41.9 458 17.9
Other 153 17.3 141 7.2
Poverty status, % FPL 1408 �.001 1342 �.001
�100 511 41.3 478 18.6
100 to�200 395 27.9 377 11.2
�200 502 17.2 487 4.4
Maternal age at child’s birth, y 1570 �.001 1498 .01
�19 253 38.5 237 18.5
20–29 839 29.3 804 10.1
�30 478 21.6 457 7.4
Maternal smoking during pregnancy 1563 .01 1491 .21
Yes 226 38.4 217 14.1
No 1337 25.3 1274 9.1
NICU 1568 .73 1495 .29
Yes 175 28.7 165 13.2
No 1393 27.3 1330 9.4
Time since last dental visit 1508 �.001 1445 .26
�1 y 672 36.3 619 11.3
�1 y 836 18.5 826 8.6
ECC includes any DFS score (�1), and S-ECC includes any DFS score (�1) on maxillary incisors.
Source: NHANES, 1999–2002.8
PEDIATRICS Volume 120, Number 4, October 2007 e947
by on February 24, 2011 www.pediatrics.orgDownloaded from
of breastfeeding and other factors included in the full
model, breastfed Mexican American children were at
greater risk for ECC than non-Hispanic white children
who were never breastfed (aOR: 2.1; 95% CI: 1.1–3.8),
and breastfed children living below the FPL were also
more likely to experience ECC than children living in
families at �200% of the FPL who had never been
breastfed (aOR: 3.2; 95% CI: 1.4–7.3).
DISCUSSION
Although Bowen and Lawrence,7 using a desalivated rat
model, reported recently that human breast milk was
more cariogenic than bovine milk, epidemiologic data on
breastfeeding and caries risk are quite limited. Breast-
feeding and dental caries among children aged 2 through
5 years was studied previously using data from 1988 to
1994, and no association was found.16 In the current
study, using more recent data and more detailed cate-
gorization of breastfeeding duration and type, the poten-
tial association of the duration of exclusive breastfeeding
and breastfeeding accompanied by additional supple-
mental feedings that potentially contained sucrose were
investigated. The findings indicate that infant breast-
feeding and its duration, whether overall, full, or exclu-
sive, is not associated with any increased risk for ECC or
S-ECC. In contrast, poverty, Mexican American ethnic sta-
tus, and maternal smoking during pregnancy were each
found to be independently associated with ECC. Several
previous studies reported findings similar to those reported
in this article concerning the association between various
aspects of breastfeeding and ECC.20–23 Although breastfeed-
ing was not found to be associated with either an increased
or decreased risk of ECC, decreased family income and
prenatal maternal smoking, both strongly associated with
decreased rates of breastfeeding as demonstrated in previ-