Content uploaded by Lars-Åke Persson
Author content
All content in this area was uploaded by Lars-Åke Persson
Content may be subject to copyright.
Risk of childhood undernutrition related to
small-for-gestational age and preterm birth
in low- and middle-income countries
Parul Christian,
1
* Sun Eun Lee,
1
Moira Donahue Angel,
1
Linda S Adair,
2
Shams E Arifeen,
3
Per Ashorn,
4
Fernando C Barros,
5
Caroline HD Fall,
6
Wafaie W Fawzi,
7
Wei Hao,
8
Gang Hu,
9
Jean H Humphrey,
1
Lieven Huybregts,
10,11
Charu V Joglekar,
12
Simon K Kariuki,
13,14
Patrick Kolsteren,
10,11
Ghattu V Krishnaveni,
15
Enqing Liu,
16
Reynaldo Martorell,
8
David Osrin,
17
Lars-Ake Persson,
18
Usha Ramakrishnan,
8
Linda Richter,
19
Dominique Roberfroid,
10
Ayesha Sania,
20
Feiko O Ter Kuile,
14,21,22
James Tielsch,
23
Cesar G Victora,
24
Chittaranjan S Yajnik,
12
Hong Yan,
25
Lingxia Zeng
25
and Robert E Black
1
1
Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD, USA,
2
Department of Nutrition, University of North Carolina, Chapel Hill, NC, USA,
3
International Centre for Diarrhoeal Disease
Research, Dhaka, Bangladesh,
4
Department of International Health, University of Tampere, School of Medicine, Tampere, Finland,
5
Programa de Po
´
s-graduac¸a
˜
o em Sau
´
de e Comportamento, Universidade Cato
´
lica de Pelotas, Pelotas, Brazil,
6
MRC Lifecourse
Epidemiology Unit, Southampton General Hospital, Southampton, UK,
7
Departments of Global Health and Population, Nutrition,
and Epidemiology, Harvard School of Public Health, Boston, MA, USA,
8
Hubert Department of Global Health, Rollins School of
Public Health, Emory University, Atlanta, GA, USA,
9
Pennington Biomedical Research Center, Baton Rouge, LA, USA,
10
Woman
and Child Health Research Center, Department of Public Health, Institute of Tropical Medicine, Nationalestraat 155, 2000
Antwerpen, Belgium,
11
Department of Food Safety and Food Quality, Ghent University, Sint-Pietersnieuwstraat 25, B 9000 Ghent,
Belgium,
12
Diabetes Unit, King Edward Memorial Hospital and Research Centre, Pune, India,
13
Center for Global Health Research,
KEMRI, Kisumu, Kenya,
14
KEMRI/CDC Research and Public Health Collaboration, Kisumu, Kenya,
15
Epidemiology Research Unit,
CSI Holdsworth Memorial Hospital, Mysore, India,
16
Tianjin Women‘s and Children‘s Health Center, Tianjin, China,
17
Institute for
Global Health, UCL Institute of Child Health, London, UK,
18
International Maternal and Child Health, Uppsala University, Uppsala,
Sweden,
19
Human Sciences Research Council and the Developmental Pathways for Health Research Programme, University of the
Witwatersrand, Johannesburg, South Africa,
20
Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA,
21
Child and Reproductive Health Group, Liverpool School of Tropical Medicine, Liverpool UK,
22
Centers for Disease Control and
Prevention, Kenya,
23
Department of Global Health, George Washington University, School of Public Health and Health Services,
Washington, DC, USA,
24
Programa de Po
´
s-graduac¸a
˜
o em Epidemiologia, Universidade Federal de Pelotas, Pelotas, Brazil and
25
Department of Public Health, Xi’an Jiaotong University College of Medicine, PO Box 46, Xi’an, Shaanxi 710061, China
*Corresponding author. Johns Hopkins Bloomberg School of Public Health, 615. N Wolfe St, Room E2541, Baltimore,
MD 21205, USA. E-mail: pchristi@jhsph.edu
Accepted 14 May 2013
Background Low- and middle-income countries continue to experience a large
burden of stunting; 148 million children were estimated to be
stunted, around 30–40% of all children in 2011. In many of these
countries, foetal growth restriction (FGR) is common, as is subse-
quent growth faltering in the first 2 years. Although there is agree-
ment that stunting involves both prenatal and postnatal growth
failure, the extent to which FGR contributes to stunting and
other indicators of nutritional status is uncertain.
Methods Using extant longitudinal birth cohorts (n ¼ 19) with data on birth-
weight, gestational age and child anthropometry (12–60 months),
we estimated study-specific and pooled risk estimates of stunting,
wasting and underweight by small-for-gestational age (SGA) and
preterm birth.
Results We grouped children according to four combinations of SGA and
gestational age: adequate size-for-gestational age (AGA) and pre-
term; SGA and term; SGA and preterm; and AGA and term (the
Published by Oxford University Press on behalf of the International Epidemiological Association
ß The Author 2013; all rights reserved.
International Journal of Epidemiology 2013;1–16
doi:10.1093/ije/dyt109
1
Int. J. Epidemiol. Advance Access published August 6, 2013
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
reference group). Relative to AGA and term, the OR (95% confi-
dence interval) for stunting associated with AGA and preterm, SGA
and term, and SGA and preterm was 1.93 (1.71, 2.18), 2.43 (2.22,
2.66) and 4.51 (3.42, 5.93), respectively. A similar magnitude of risk
was also observed for wasting and underweight. Low birthweight
was associated with 2.5–3.5-fold higher odds of wasting, stunting
and underweight. The population attributable risk for overall SGA
for outcomes of childhood stunting and wasting was 20% and 30%,
respectively.
Conclusions This analysis estimates that childhood undernutrition may have its
origins in the foetal period, suggesting a need to intervene early,
ideally during pregnancy, with interventions known to reduce FGR
and preterm birth.
Keywords Foetal growth restriction, preterm birth, stunting, wasting,
childhood
Introduction
Childhood undernutrition marked by stunting (height
for age <2 z-scores) is very common, affecting 164.8
million (22.7%) children globally, 148 million of whom
live in low- and middle-income countries (LMIC).
1
Additionally, 8% or 52 million children are severely
wasted.
1
Childhood undernutrition is well known to
increase the risk of short-term mortality.
2,3
In addition,
height at 24 months of age is predictive of adult
height, and childhood stunting is linked with cognitive
deficits, less schooling and potential for income gener-
ation and employment, all of which contribute to
reduced human capital in developing countries.
4
The timing and pattern of growth faltering in the
first 2 years of life is well established, with height for
age z-scores in LMIC declining soon after birth to a
nadir of 1.75 to <2 z-scores by 24 months of age,
and little if any subsequent catch-up growth being
evident up to 5 years of age.
5
Numerous factors,
including inappropriate breastfeeding and infant and
young child feeding practices, lack of adequate quality
and amount of complementary foods, infection
and other environmental exposures are known to
contribute to this pattern of growth faltering in
under-resourced settings. The first 1000 days of life
(conception through 24 months of age; www.
thousanddays.org), during which critical human
growth and development occur, is well recognized as
a life-stage continuum between the foetal period and
infancy and early childhood. Thus, foetal growth and
birthweight as its culmination, are likely to influence
childhood growth, and stunting in children may have
prenatal origins.
6
For example, high rates of low
birthweight (LBW, birthweight <2.5 kg) and child-
hood stunting tend to co-exist in many settings.
LBW is estimated at 16% in developing countries,
with rates higher in Asia than in Africa.
2
Underlying
biological contributors to LBW include preterm birth
(gestational age <37 weeks) and foetal growth re-
striction (FGR), which is usually described as small-
for-gestational age (SGA), defined as a sex-specific
birthweight below the 10th percentile for gestational
age of a reference standard. In 2010, the Child Health
Epidemiology Reference Group (CHERG) estimated
11.1% of all live births, or 14.9 million, to be preterm,
worldwide.
7
The 2010 estimate of prevalence of SGA
at full term (537 weeks) ranges from 5.5% of live
births in Eastern Asia to 40.3% in Southern Asia.
8
LBW and stunting are positively correlated in ecolo-
gical country-level analyses
9
but few birth cohort stu-
dies have examined the associations between foetal
growth and childhood nutritional status. Specifically,
the effects of stunting and wasting related to SGA
and preterm have not been previously estimated.
SGA and preterm are important to examine separately
as their aetiologies are known to differ, as may their
relative contribution to childhood undernutrition.
The primary aim of this analysis was to estimate the
risk of stunting, wasting and underweight in children
12–60 months of age, and when available at exactly
24 months of age, related to combinations of preterm
and SGA categories and LBW, using data from birth
cohorts and longitudinal studies.
Methods
Identification of birth cohorts
We conducted a detailed search for extant datasets
from birth cohorts with anthropometric data in early
childhood in LMIC. Studies were identified through
Medline search and other known sources including:
(i) the concurrent CHERG (www.cherg.org) working
group, (ii) a meta-analysis of multiple micronutrient
supplementation trials on birth outcomes,
10
(iii) the
Consortium of Health Oriented Research in
Transitioning Societies (COHORTS) profile
11
and
2
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
(iv) the principal investigator (Parul Christian) of the
CHERG Prenatal origins working group.
Literature review (data inputs)
We developed five categories of Mesh and title/ab-
stract (tiab) terms to identify studies with both
birth and childhood (12–60 months of age) anthropo-
metric data in LMIC: exposure, outcome, age group at
follow up, study type and countries. Within each cat-
egory, all search terms were combined with ‘OR’. All
categories were combined as follows: [Exposure] AND
[Outcome] AND [Age group at follow up] AND
[Study type] AND [LMIC]. The exposure category
consisted of Mesh terms: ’birthweight’, ‘infant, low
birthweight’, ‘gestational age’, ‘f(o)etal growth retard-
ation’, ‘Infant’, ‘premature’, ‘premature birth’, ‘obstet-
ric labor, premature’, ‘Infant, small for gestational
age’, ‘pregnancy’, ‘pregnancy complications’ and simi-
lar tiab terms including ‘appropriate for gestational
age’ and ‘small for gestational age’. The outcome cat-
egory consisted of Mesh terms: ‘growth’, ‘growth and
development’, ‘child development’, ‘body weight’,
‘body size’, ‘body mass index’, ‘body height’, ‘body
composition’, ‘anthropometry’ and related tiab terms
such as ‘stunting’. The age at follow up category
included ‘infant’, ‘child’ and ‘child, preschool’ Mesh
and tiab terms. The study type category included
Mesh terms: ’retrospective studies’, ‘prospective stu-
dies’, ‘longitudinal studies’, ‘epidemiological studies’,
‘case-control studies’, and ‘follow up studies’. LMIC
were included according to the country classification
by UN Millennium Development Goal (MDG) regional
groupings. We limited the search to human studies
published between 1 January 1990 and 8 June 2011.
Language was not restricted if the abstract was avail-
able in English.
Titles of articles identified through Medline search
were retrieved and evenly divided into three groups in
order of publication date. Three PhD students at the
Johns Hopkins Bloomberg School of Public Health
each screened 700 titles and abstracts for eligibility
and coded them as potentially relevant or irrelevant.
Irrelevant studies were filtered out, and full texts of
potentially relevant articles were reviewed for eligibil-
ity. Unique cohorts were identified and further inves-
tigated for eligibility through confirmation of several
publications. Studies with questionable eligibility
were evaluated by the PI of the CHERG working
group (P.C.). Inclusion criteria were: birth cohorts or
studies with birth data with valid birthweights and
gestational age measurements (last menstrual
period, ultrasound, Ballard, Capurro, or Dubowitz
methods); and anthropometric measures of children
with a follow up of at least 12 months or at least one
measurement between 12 and 60 months of age. We
chose to include a minimum follow up of 12 months
because linear growth faltering is rapid through the
1st year of life. Because growth faltering continues
even through 24 months of age and stabilizes
somewhat after this, we also examined the risk of
stunting and wasting in a smaller group of studies
in which measurement of anthropometry was done
at exactly 24 months of age. Exclusion criteria were
HIV infection in either mothers or babies, genetic dis-
orders in children and small sample size (n < 200) of
the original study. PIs of all eligible cohorts were con-
tacted and invited to collaborate. PIs who agreed to
participate were given the option of conducting a
standard set of analyses independently and sending
the results to the CHERG working group or to grant
permission to use their dataset to conduct the
analysis.
Data analysis
Data analysis was conducted for each cohort and es-
timates were used for meta-analysis.
We used SGA, preterm (<37 weeks of gestation)
and LBW (<2.5 kg) for prenatal exposures. SGA was
defined as less than the 10th percentile for gestational
age using the US population-based standard of
Alexander et al.
12
Children whose birthweights were
measured more than 72 h after birth and those whose
gestational ages at delivery were equal to or more
than 44 weeks were excluded. We also excluded
implausible combinations of gestational age and
birthweights based on the method reported by
Alexander et al.
12
Outcomes included length/height for age standard
deviation score (HAZ), weight for length/height
(WHZ) and weight for age (WAZ), calculated accord-
ing to the 2006 WHO child growth standards.
13
Stunting, wasting and underweight were defined as
HAZ, WHZ and WAZ less than 2 z-scores. Child age
was defined as the number of months from date of
birth to date of anthropometric measurement. For
children with longitudinal anthropometric measure-
ments between 12 and 60 months of age, the last
visit with complete anthropometry was used.
We estimated odds ratios (OR) and 95% confidence
intervals (CI) for stunting, wasting and underweight
associated with LBW, SGA and preterm. ORs were
estimated instead of relative risks because incident
stunting or wasting was not known in many studies,
only a single measurement of anthropometry was
available in some studies and, in those which had
multiple measurements, we used only the last avail-
able measurement. We then evaluated the independ-
ent contributions of AGA and preterm, SGA and term,
and SGA and preterm—compared with AGA and term
as the reference category—to childhood undernutri-
tion. Interaction between preterm and SGA was
tested in the datasets provided to the CHERG working
group. To examine a dose-response relationship, we
performed an additional analysis estimating ORs of
stunting associated with birthweight categories
(52.5 kg and <2.5 kg) in infants born SGA and
term, compared with infants born AGA and term.
We also examined ORs for stunting associated with
RISK OF CHILDHOOD UNDERNUTRITION 3
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
sub-categories of gestational age (34 to 37 weeks, 32
to 34 weeks and <32 weeks) compared with the ref-
erence group, AGA and term infants. All analyses
were conducted for children 12 to 60 months of age
and a subset of children at 24 months of age. Child
age was adjusted for in all analyses except for the
analysis with children followed up at exactly 24
months.
Meta-analysis for overall (LMIC) and
regional estimation
We used ORs and standard errors/CIs from each study
to make regional and overall estimates using meta-
analysis with random-effects models. The
DerSimonian and Laird random effects method was
used to incorporate an estimate of between-study
variation (heterogeneity) in the weighting.
14
Between-study heterogeneity was quantified using
the I
2
statistic (expressed as %) and Cochrane’s Q
(significance level <0.05). Forest plots were used to
display individual and pooled estimates. Sensitivity
analysis was done by running models adjusted for
child’s age and sex, twins, infection, child and mater-
nal interventions, parity, socioeconomic status, mater-
nal education and maternal infection, using the 10
datasets provided to the CHERG working group to
examine whether confounding was a concern.
Stratified analyses by maternal age (<20 years, 20–
29 years and 530 years), parity (0, 1–3 and 5 4),
child sex (male/female), cohort years (1970–80s,
1990s and 2000s) (data not shown) and preterm cate-
gories (<32 weeks, 32–34 weeks and 34–37 weeks)
were performed to examine heterogeneity by sub-
groups. Funnel plots (not shown) were examined to
assess publication bias that may affect pooled esti-
mates. A forest plot using a random-effects model
was used to present estimates of ORs for childhood
stunting at 12 to 60 months by birthweight categories
in children born SGA and term, with a reference
group of children born AGA and term. All statistical
analyses were performed with Stata software, version
12 (Stata Corp, College Station, TX).
Population attributable risk
In order to estimate population attributable risk
(PAR) which would be over-estimated using OR and
high prevalence rates of exposures, we derived relative
risks (RRs) and 95% CIs from adjusted ORs and their
95% CIs by employing the method proposed by Zhang
and Yu.
15
This approximation is applied when the
prevalence of a condition in the study population is
high. The validity of this method was examined by
comparing with RRs estimated by Poisson regression
among 10 available studies. For stunting, the approxi-
mated mean RR for SGA and term was almost iden-
tical (RR ¼ 1.53) in the 10 studies when compared
with that derived using the regression analysis
(RR ¼ 1.54). We calculated regional and LMIC PAR
using Levin’s formula:
16
p(RR 1)/(1 þ p(RR 1))
and 95% CIs of PAR were estimated using the delta
method.
17
Population attributable burden was calcu-
lated by multiplying the PAR by the estimated
number of children stunted or wasted. Regional and
LMIC prevalence of size for gestational age and term
combinations were based on the year 2010 extracted
from the concurrent CHERG working group.
8
Estimated regional and LMIC prevalence and
number of stunted and wasted children were based
on the year 2011.
1
We derived pooled RRs from the
two China studies and applied them to Eastern Asia
with the following caveats: only one study had an
estimate for stunting and wasting associated with
AGA and preterm and was considered as representa-
tive; and neither study had estimates for wasting
associated with SGA and preterm, because no chil-
dren were wasted in this category, thus null associ-
ations were used. We calculated pooled RRs from the
remaining Asian cohort studies and applied them to
Southern, Southeastern and Western Asia. RRs for
Sub-Saharan Africa and Latin America were applied
to all African regions and Latin America and
Caribbean, respectively. Overall (LMIC) RRs were
applied to Oceania and Caucasus and Central Asia
regions.
Results
Our search yielded over 2000 articles that were re-
viewed by title and abstract, and 1880 were excluded
according to general criteria such as study not from
LMIC, in an HIV population or being a review paper
(Figure 1). Of the remaining 230 articles, 179 were
excluded because of missing exposure or outcome
variables, and identification of redundant cohorts.
After this exclusion we had 51 unique cohorts,
which, with 8 others from identified known sources,
gave us a pool of 59 cohorts which we assessed for
their eligibility for inclusion in this analysis. A total
of 33 cohorts were deemed eligible and 19 datasets
were ultimately analysed. Fourteen studies were
excluded because they were deemed ineligible upon fur-
ther examination (n ¼ 4), or because the PI did not
respond (n ¼ 7) or declined participation (n ¼ 3).
Studies included in the analysis were conducted be-
tween 1970 and 2007. Ten of the datasets were ana-
lysed by CHERG researchers, whereas nine
investigators conducted their own analyses using the
analytic protocol and programs provided to them.
Weight and gestational age at birth were available
for 58 317 out of a total 66 074 live births. Of the
children who were followed (n ¼ 54 611), anthropom-
etry was available in 44 374 (81.2%) children who
were included in the analysis (Supplementary Figure
1, available as Supplementary data at IJE online).
Birthweight and gestational age across the 19 stu-
dies are shown in Table 1. Mean birthweight ranged
from 2.63 kg in India to 3.43 kg in China. Mean ges-
tational age was less variable at 39.0 weeks,
4
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
although two cohorts recorded a lower mean at 38.1
weeks. LBW prevalence was highly variable, ranging
from 1.0% in Tianjin, China, to 31.0% in Nepal, with a
similar variability in the prevalence of SGA ranging
from 6.1% in China to 64.5% in India. Preterm birth
ranged from 1.1% in China to 18.5% in Malawi. The
prevalence of SGA and term was higher in South and
Southeastern Asia than in East Asia, whereas SGA
and preterm was not common across the cohorts, sug-
gesting different underlying causes of preterm birth
and foetal growth restriction. The burden of stunting
was high in many cohorts, ranging between 50 and
70%, although older cohorts appeared to have a
higher prevalence (Table 2). Wasting was less
2110 Articles identified through
database search
8 Cohorts identified through
known sources
230 Full-text articles reviewed
59 Cohorts assessed for eligibility
33 Principal Investigators invited to participate
19 Data inputs included in meta-analysis
10 datasets analysed by CHERG researchers
9 datasets analysed by Principal Investigators
14 Cohorts excluded
4 were not eligible
7 did not respond
1 declined to participate
2 interested but did not send data or
analysis by deadline
26 Cohorts excluded
6 did not collect valid gestational age or
birth weight data
2 included only term or LBW babies
10 did not follow up at 12 to 60 months
3 included only HIV+ mothers or babies
5 other reasons*
1880 Excluded based on titles or
abstracts using general criteria
(Animal study;developed country;
did not have study birth or child
anthropometry; HIV population;review
article)
179 Articles excluded based on
inclusion criteria
34 Missing valid birthweight or
gestational age measurements
22 Missing anthropometry
123 Redundant cohorts
51 Unique cohorts identified
*Other reasons include genec disorders (n = 3), and small sample size (<200) (n = 2)
Figure 1 Flow diagram of literature search for cohorts
RISK OF CHILDHOOD UNDERNUTRITION
5
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
Table 1 Birthweight and gestational age characteristics by study cohort
Study N
Birthweight,
kg mean (SD)
Gestational age,
wks mean (SD)
LBW
a
(%)
SGA
b
(%)
Preterm
c
(%)
AGA
Preterm
(%)
SGA
Term (%)
SGA
Preterm (%)
Nepal (Janakpur Trial)
18
909 2.79 (0.42) 38.9 (1.7) 21.1 50.5 6.7 4.2 48.0 2.5
Nepal (Newborn Washing Study, Sarlahi)
19
1134 2.69 (0.45) 38.9 (2.3) 31.0 53.3 16.0 11.6 48.9 4.4
Bangladesh (MINIMat)
20
2499 2.70 (0.40)
d
38.8 (1.7)
d
29.3 58.9 7.4 4.6 56.1 2.8
India (Pune Maternal Nutrition Study)
21
674 2.63 (0.38) 39.0 (1.7) 31.6 64.5 9.5 5.9 61.0 3.6
India (Parthenon Study, Mysore)
22
585 2.87 (0.44) 38.9 (1.6) 17.8 43.9 6.5 4.8 42.2 1.7
China (Shaanxi)
23
1261 3.18 (0.41) 39.9 (1.5) 3.6 18.3 3.5 3.2 18.0 0.3
China (Tianjin)
24
9289 3.43 (0.44) 39.3 (1.2) 1.0 6.1 1.1 0.9 6.0 0.2
Philippines (Cebu)
25
2414 3.00 (0.42) 38.8 (2.1) 10.6 24.4 16.5 14.3 22.1 2.2
South Africa (BT20)
26
1656 3.09 (0.51) 38.1 (1.9) 9.9 16.4 11.7 10.0 14.7 1.6
Burkina Faso (MISAME)
27–29
1261 2.94 (0.42) 38.9 (2.1) 13.1 31.3 10.5 8.3 29.2 2.1
Zimbabwe (ZVITAMBO)
30
5733 3.01 (0.44) 39.0 (1.4) 11.7 30.2 6.1 3.9 28.0 2.2
Malawi (LCSS)
31
146 3.00 (0.51) 39.1 (3.1) 10.3 22.6 18.5 17.8 21.9 0.7
Tanzania (Dar es-Salam)
32
6098 3.14 (0.50) 39.6 (2.8) 6.7 19.2 14.7 13.9 18.4 0.8
Kenya (ABCP)
33
947 3.11 (0.43) 39.0 (1.1) 6.9 21.8 2.1 1.7 21.3 0.4
Brazil (Pelotas 1982)
34
3571 3.25 (0.51)
e
39.4 (1.8)
e
5.9 16.5 5.4 4.4 15.6 0.9
Brazil (Pelotas 1993)
35
1220 3.17 (0.53) 38.1 (1.6) 10.2 14.4 10.5 7.4 22.1 3.5
Brazil (Pelotas 2004)
36
3600 3.17 (0.53)
f
38.6 (2.3)
f
9.1 14.6 13.9 11.8 12.6 2.0
Guatemala (INCAP)
37
601 3.02 (0.47) 39.4 (2.4) 9.3 33.4 10.8 10.0 32.6 0.8
Mexico (Cuernavaca)
38
776 3.23 (0.45) 38.7 (1.8) 4.8 9.2 9.0 8.9 9.0 0.1
LBW, low birthweight; SGA, small-for-gestational age.
a
LBW defined as birthweight <2.5 kg.
b
SGA defined as size <10th percentile for gestational age.
c
Preterm defined as delivery at gestational age <37 weeks.
d
Estimated using n ¼ 2735.
e
Estimated using n ¼ 3948.
f
Estimated using n ¼ 3830.
6 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
Table 2 Child age and anthropometric measures by study cohort
Study N
Age, mo.
mean (SD)
HAZ
mean (SD)
WHZ
mean (SD)
WAZ
mean (SD)
Stunting
(%)
Wasting
(%)
Underweight
(%)
Nepal (Janakpur Trial)
18
909 30.3 (4.2) 2.23 (1.10) 0.63 (1.04) 1.68 (1.02) 58.0 7.6 37.2
Nepal (Newborn Washing Study, Sarlahi)
19
1134 38.7 (11.1) 2.50 (1.29) 1.01 (1.00) 2.14 (0.97) 62.5 11.8 57.3
Bangladesh (MINIMat)
20
2499 54.6 (1.8) 1.55 (0.92) 1.31 (0.84) 1.80 (0.87) 31.3 19.5 40.3
India (Pune Maternal Nutrition Study)
21
674 59.2 (3.0) 1.61 (0.87) 1.34 (0.81) 1.84 (0.84) 32.3 21.8 42.0
India (Parthenon Study, Mysore)
22
585 59.5 (0.7) 0.90 (0.90) 1.40 (0.90) 1.40 (0.90) 9.9 25.0 26.2
China (Shaanxi)
23
1261 29.6 (3.3) 0.89 (1.00) 0.16 (0.89) 0.35 (0.84) 12.2 1.1 2.5
China (Tianjin)
24
9289 48.7 (7.0) 0.19 (0.96) 0.15 (0.97) 0.23 (0.95) 1.1 0.9 0.6
Philippines (Cebu)
25
2414 23.5 (2.1) 2.56 (1.18) 0.50 (1.00) 1.70 (1.07) 67.4 6.8 36.8
South Africa (BT20)
26
1656 29.7 (16.7) 1.04 (1.25) 0.30 (1.38) 0.32 (1.20) 20.8 4.4 7.1
Burkina Faso (MISAME)
27–29
1261 22.0 (9.9) 1.75 (1.28) 0.74 (1.17) 1.46 (1.09) 41.6 12.5 28.6
Zimbabwe (ZVITAMBO)
30
5733 16.0 (5.3) 1.23 (1.19) 0.11 (1.15) 0.52 (1.09) 23.8 3.3 8.2
Malawi (LCSS)
31
146 56.7 (11.0) 2.37 (1.12) 0.13 (1.12) 1.36 (1.04) 58.2 3.4 21.9
Tanzania (Dar es-Salam)
32
6098 13.7 (2.7) 1.14 (1.20) 0.23 (1.21) 0.72 (1.18) 21.7 7.1 12.6
Kenya (ABCP)
33
947 34.0 (14.8) 2.27 (1.59) 0.36 (1.50) 1.00 (1.29) 56.1 6.4 18.8
Brazil (Pelotas 1982)
34
3571 43.2 (3.7) 0.59 (1.08) 0.57 (0.97) 0.04 (1.03) 9.5 0.3 2.0
Brazil (Pelotas 1993)
35
1220 54.0 (3.6) 0.20 (1.12) 0.54 (1.15) 0.25 (1.17) 5.2 0.4 2.3
Brazil (Pelotas 2004)
36
3600 49.5 (1.7) 0.16 (1.05) 0.70 (1.17) 0.37 (1.19) 3.6 0.6 1.6
Guatemala (INCAP)
37
601 44.8 (13.8) 2.62 (1.01) 0.16 (0.89) 1.44 (0.88) 73.4 1.2 24.6
Mexico (Cuernavaca)
38
776 48.7 (5.8) 0.50 (0.90) 0.10 (1.00) 0.20 (1.00) 5.0 0.8 2.1
HAZ, height-for-age z-score; WHZ, weight-for-height z-score; WAZ, weight-for-age z-score.
RISK OF CHILDHOOD UNDERNUTRITION 7
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
common across regions, with the exception of South
Asia.
ORs for conventional exposure categories including
LBW, preterm and SGA (Table 3) showed that LBW
was associated with an overall increased odds of 2.92
(95% CI: 2.56, 3.33) for stunting.
Regional and overall LMIC risk estimates for stunt-
ing, wasting and underweight at 12–60 months of age
are presented by SGA and preterm categories (Table 4,
Figure 2). Relative to AGA and term birth, the ORs
(95% CI) for stunting associated with AGA and pre-
term, SGA and term and SGA and preterm were 1.93
(1.71, 2.18), 2.43 (2.22, 2.66) and 4.51 (3.42, 5.93),
respectively (Table 4), suggesting an additive risk in
the presence of both conditions. A similar magnitude
of risk was also observed for wasting and under-
weight among children. None of the SGA and preterm
interaction terms had P-value of <0.05 (data not
shown). There was limited regional variation in
these risk estimates, although there was a trend of
preterm alone being associated with a somewhat
lower risk of undernutrition in Southern/Eastern
Asia than in Africa and Latin America. In Latin
America the risk for wasting among SGA and preterm
was high at 21.57 (8.18, 56.90) given a low prevalence
of wasting (<0.5%) in the referent category.
SGA and term infants without LBW had lower odds
of stunting (OR ¼ 1.92, 95% CI: 1.75, 2.11) vs SGA or
term accompanied by LBW (OR ¼ 3.00, 95% CI: 2.36,
3.81) (Figure 3). AGA but extremely preterm (<32
weeks), very preterm (32–34 weeks) and moderate-
to-late preterm (34–37 weeks) had ORs (95% CI) of
3.93 (2.79, 5.54), 2.78 (1.75, 4.40) and 1.81 (1.63,
2.02), respectively (Figure 4).
Adjustment for confounders beyond child age,
including sex, parity, twinning, infection, interven-
tions, socioeconomic status, maternal education, pre-
natal intervention and infection did not change the
risk estimates (Supplementary Table S1, available as
Supplementary data at IJE online). ORs for the com-
binations of size for gestational age and preterm birth,
using anthropometry at exactly age 24 months as the
outcome, were also estimated (Supplementary Table
S2, available as Supplementary data at IJE online).
PAR was estimated for childhood stunting and
wasting for each of the risk categories (Table 5).
The PAR for SGA and term for stunting and wasting
was 0.16 (0.12, 0.19) and 0.24 (0.21, 0.26), respect-
ively. The prevalences of SGA and AGA preterm are
small, largely because of the low prevalence of pre-
term ranging from 1 to 3%. The combined PAR related
to overall SGA for stunting was 0.20, and that for
wasting was 0.30. PAR for LBW was estimated at
0.12 (0.09, 0.16) for stunting and 0.18 (0.14, 0.23)
for wasting (data not shown). The attributable
burden related to overall SGA for stunting was
Table 3 Regional and overall odds ratios for childhood undernutrition at 12 to 60 months among babies born low
birthweight, small-for-gestational age, and preterm
LBW
a
SGA
b
Preterm
c
Regions [number of studies] N
e
OR
d
(95% CI) OR
d
(95% CI) OR
d
(95% CI)
Stunting
Southern/Eastern Asia [8] 18 765 2.64 (2.30, 3.02) 2.13 (1.85, 2.46) 1.34 (1.15, 1.55)
Sub-Saharan Africa [6] 15 841 2.77 (2.12, 3.62) 2.32 (2.10, 2.56) 1.98 (1.72, 2.27)
Latin America [5] 9768 3.67 (2.96, 4.56) 3.07 (2.58, 3.66) 1.79 (1.25, 2.55)
LMIC [19] 44 374 2.92 (2.56, 3.33) 2.32 (2.12, 2.54) 1.69 (1.48, 1.93)
Wasting
Southern/Eastern Asia [8] 18 765 2.42 (2.03, 2.88) 2.46 (2.15, 2.81) 1.20 (0.85, 1.69)
Sub-Saharan Africa [6] 15 841 2.48 (1.89, 3.25) 2.18 (1.89, 2.52) 1.76 (1.46, 2.12)
Latin America [5] 9768 7.48 (3.79, 14.80) 3.78 (1.85, 7.75) 3.78 (1.93, 7.39)
LMIC [19] 44 374 2.68 (2.23, 3.21) 2.36 (2.14, 2.60) 1.55 (1.21, 1.97)
Underweight
Southern/Eastern Asia [8] 18 765 3.28 (2.86, 3.77) 3.14 (2.55, 3.87) 1.27 (1.10, 1.46)
Sub-Saharan Africa [6] 15 841 3.48 (3.06, 3.97) 2.60 (2.26, 2.99) 2.05 (1.68, 2.51)
Latin America [5] 9768 4.56 (3.14, 6.64) 3.78 (2.46, 5.80) 2.14 (1.56, 2.93)
LMIC [19] 44 374 3.48 (3.14, 3.87) 2.96 (2.61, 3.36) 1.66 (1.42, 1.95)
SGA, small-for-gestational age; LBW, low birthweight; LMIC, low- and middle-income countries.
a
LBW defined as <2.5 kg measured within 72 h of birth, compared with 52.5 kg.
b
SGA defined as size <10th percentile for gestational age, compared with 510th percentile for gestational age.
c
Preterm defined as delivery at gestational age <37 weeks, compared with gestational age 537 weeks.
d
Adjusted for child age.
e
Maximum number.
8 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
about 32 million, whereas that for wasting was esti-
mated at 15 million (Table 6).
Discussion
Our analysis estimated the odds ratios of childhood
(12–60 months) stunting, wasting and underweight
associated with SGA and preterm birth using data
from 19 birth cohorts from LMICs. Relative to AGA
and term, SGA and term birth was associated with a
2.4 and AGA and preterm birth with a 1.9 increased
odds of stunting. The odds ratio was increased to 4.5
in births both SGA and preterm. Similar associations
were observed for wasting and underweight. In LMIC,
the attributable burden of stunting and wasting
among children due to SGA was estimated at 32
and 15 million, respectively. The observational
nature of these data does not permit causal inference,
but is consistent with foetal origins of both childhood
stunting and wasting.
We saw limited variation in the association between
birth exposures and childhood undernutrition by
region, which we consider to be biological. The risk
of undernutrition associated with being born small or
too soon was comparable across populations and re-
gions, despite the large variation in the prevalence of
both SGA and preterm birth, largely reflecting the
common underlying causes in these settings of
either foetal growth restriction or preterm birth.
Latin America as a region had the lowest rates of
both exposures, but the highest risk association, al-
though the number of cohorts and sample sizes of the
studies were small, yielding somewhat unstable esti-
mates. Two studies contributed data from China,
23,24
which showed extremely low prevalence rates of both
the exposure and the outcomes, with comparable or
even lower rates of these seen in developed countries.
Table 4 Regional and overall odds ratios for childhood undernutrition at 12 to 60 months of age, by adequacy of size for
gestational age and preterm birth using AGA and term as the reference category
AGA and preterm
a,b
SGA and term
a,b
SGA and preterm
a,b
Regions [number of studies] N
d
OR (95% CI)
c
OR (95% CI)
c
OR (95% CI)
c
Stunting
Southern/Eastern Asia [8] 18 765 1.56 (1.31, 1.87) 2.25 (2.03, 2.50) 3.63 (2.50, 5.28)
Sub-Saharan Africa [6] 15 841 2.13 (1.87, 2.42) 2.36 (2.09, 2.67) 5.95 (3.84, 9.22)
Latin America [5] 9768 2.21 (1.59, 3.08) 3.34 (2.78, 4.02) 4.31 (2.22, 8.36)
LMIC [19] 44 374 1.93 (1.71, 2.18) 2.43 (2.22, 2.66) 4.51 (3.42, 5.93)
Wasting
Southern/Eastern Asia [8] 18 765 1.65 (1.03, 2.64) 2.58 (2.22, 2.98) 3.50 (2.25, 5.42)
Sub-Saharan Africa [6] 15 841 2.20 (1.54, 3.13) 2.42 (2.07, 2.82) 3.44 (2.21, 5.34)
Latin America [5] 9768 3.28 (1.04, 10.29) 3.73 (1.92, 7.21) 21.57 (8.18, 56.90)
LMIC [19] 44 374 1.96 (1.46, 2.63) 2.52 (2.27, 2.80) 4.19 (2.90, 6.05)
Underweight
Southern/Eastern Asia [8] 18 765 1.58 (1.33, 1.88) 3.27 (2.67, 4.00) 4.57 (3.43, 6.08)
Sub-Saharan Africa [6] 15 841 2.36 (2.01, 2.77) 2.77 (2.36, 3.24) 6.10 (4.48, 8.30)
Latin America [5] 9768 3.05 (2.06, 4.52) 4.46 (2.82, 7.06) 6.67 (3.56, 12.52)
LMIC [19] 44 374 2.07 (1.76, 2.44) 3.17 (2.78, 3.62) 5.35 (4.39, 6.53)
AGA, adequate-for-gestational age; SGA, small-for-gestational age; LMIC, low- and middle-income countries.
a
AGA defined as size 510th percentile for gestational age; SGA defined as size <10th percentile for gestational age.
b
Term defined as delivery at gestational age 537 weeks; preterm defined as delivery at gestational age <37 weeks.
c
Adjusted for child age.
d
Maximum number.
Figure 2 Regional and LMIC odds ratios for childhood
stunting at 12 to 60 months of age by size for gestational
age and preterm birth. Reference group is children born
AGA and term; LMIC, low- and middle-income countries
RISK OF CHILDHOOD UNDERNUTRITION
9
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
In general, the studies from Africa had younger ages
at follow up, which may have attenuated the
relationship.
Recently, preterm has been shown to be associated
with a higher risk of neonatal mortality, with relative
risks ranging from 6 to 9, vs SGA alone, which was
consistently associated with a 3-fold increased risk.
39
This seems plausible, as prematurity is one of the
three largest causes of neonatal mortality world-
wide.
40
In contrast, the longer-term growth conse-
quences we report appear to indicate a stronger
association between SGA and stunting than between
preterm and stunting. Additionally, both SGA and
preterm birth categories showed a dose-response rela-
tionship with stunting; SGA with low birthweight vs.
not, and extremely or very preterm birth vs moderate
or late preterm, were associated with a higher risk of
stunting in childhood. This indicates that foetal
growth restriction as assessed by SGA, and in the ab-
sence of low birthweight carries a lower, but still
raised, risk of later stunting, suggesting a need for
programmes to move toward measurement of SGA
in addition to birthweight and preterm birth to iden-
tify newborns at risk.
Although there are many children who are growth
restricted despite not being low birthweight and/or
preterm, the co-occurrence of both preterm and SGA
was remarkably low (1–2%), suggesting different
aetiologies of SGA and preterm birth. The lower
prevalence of the two conditions may also in part be
due to the higher mortality in this group. Generally,
preterm babies are less likely to be SGA because they
have less time for growth and therefore growth re-
striction, especially as foetal growth is accelerated in
the last few weeks of pregnancy.
Our data analysis and approach have several
strengths and some limitations. Across 19 identified
cohorts, we had a total sample of 44 374. We had
good representation from different regions of the
world. Deriving regional estimates from a limited
number of countries may have its disadvantages for
some regions, but the lack of heterogeneity in the risk
estimates assures us that the relationship we observed
was biologically plausible. Additionally, our sensitivity
NOTE: Weights are from random-effects analysis
Birthweight ≥ 2.5 kg, SGA Term
Nepal (Janakpur Trial)
Nepal (Newborn Washing Study, Sarlahi)
China (Tianjin)
Philippines (Cebu)
South Africa (BT20)
Malawi (LCSS)
Burkina Faso (MISAME)
Zimbabwe (ZVITAMBO)
Kenya (ABCP)
Guatemala (INCAP)
D+L Subtotal (I
2
= 0.0%, P = 0.460)
Birthweight <2.5 kg, SGA Term
Nepal (Janakpur Trial)
Nepal (Newborn Washing Study, Sarlahi)
China (Tianjin)
Philippines (Cebu)
South Africa (BT20)
Malawi (LCSS)
Burkina Faso (MISAME)
Zimbabwe (ZVITAMBO)
Kenya (ABCP)
Guatemala (INCAP)
D+L Subtotal (I
2
= 56.8%, P = 0.013)
Study
1.74 (1.28, 2.36)
1.49 (1.09, 2.03)
2.15 (1.14, 4.05)
1.73 (1.34, 2.23)
1.72 (1.17, 2.52)
1.93 (0.74, 5.00)
2.12 (1.58, 2.85)
2.06 (1.77, 2.41)
1.93 (1.32, 2.82)
3.09 (1.86, 5.13)
1.92 (1.75, 2.11)
3.70 (2.42, 5.66)
2.21 (1.58, 3.09)
4.04 (0.97, 16.91)
4.61 (2.82, 7.53)
2.89 (1.78, 4.69)
2.06 (0.37, 11.62)
2.60 (1.70, 3.96)
3.83 (3.12, 4.70)
1.40 (0.80, 2.45)
5.04 (1.75, 14.51)
3.00 (2.36, 3.81)
ES (95% CI)
9.54
9.34
2.24
13.85
6.13
0.99
10.45
37.67
6.27
3.51
100.00
12.75
15.08
2.47
11.24
11.39
1.76
12.82
18.46
9.90
4.13
100.00
%
Weight
(D+L)
1 3
Figure 3 Random-effects model forest plot of odds ratios for childhood stunting at 12 to 60 months by birthweight
categories in children born SGA and term. Reference group is children born AGA and term. Includes 10 cohorts,
using data provided to CHERG working group
10 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
analysis revealed no evidence of strong confounding
and showed a dose-response relationship, indicating
possible causal associations. One limitation, as ex-
pected in longitudinal studies, was the loss-to-follow
up, especially due to mortality, which may result in a
potential survival bias. A higher mortality in the SGA
and especially in the preterm birth categories probably
resulted in an underestimation of ORs rather than an
overestimate, making our estimates of the risk rela-
tionship conservative. Other types of losses and miss-
ing values may have either attenuated or inflated the
ORs, but we had limited information to determine
whether the loss-to-follow-up differed by exposure
categories. Another limitation was the lack of data
on birth length in several cohorts, which we antici-
pate would be more strongly correlated with later
stunting than SGA.
41
In a study among West
Javanese infants, neonatal weight and length in mul-
tivariable analyses were the strongest positive pre-
dictors of nutritional status of infants and the
strongest negative predictors of increases in weight
and length during infancy.
42
This analysis will be
NOTE: Weights are from random-effects analysis
.
AGA & gestational age 34 to 37 wks
Nepal (Janakpur Trial)
Nepal (Newborn Washing Study, Sarlahi)
Bangladesh (MINIMat)
India (Pune Maternal Nutrition Study)
India (Parthenon Study, Mysore)
China (Shaanxi)
Philippines (Cebu)
South Africa (BT20)
Malawi (LCSS)
Burkina Faso (MISAME)
Zimbabwe (ZVITAMBO)
Tanzania (Dar es-Salam)
Kenya (ABCP)
Brazil (Pelotas 1982)
Brazil (Pelotas 1993)
Brazil (Pelotas 2004)
Guatemala (INCAP)
Mexico (Cuernavaca)
Subtotal (
I
2
= 0.0%, P = 0.511)
AGA & gestational age 32 to 34 wks
Nepal (Newborn Washing Study, Sarlahi)
Bangladesh (MINIMat)
India (Pune Maternal Nutrition Study)
Philippines (Cebu)
South Africa (BT20)
Malawi (LCSS)
Burkina Faso (MISAME)
Zimbabwe (ZVITAMBO)
Tanzania (Dar es-Salam)
Brazil (Pelotas 1982)
Brazil (Pelotas 1993)
Brazil (Pelotas 2004)
Guatemala (INCAP)
Mexico (Cuernavaca)
Subtotal (
I
2
= 61.6%, P = 0.001)
AGA & gestational age <32 wks
Nepal (Janakpur Trial)
Bangladesh (MINIMat)
Philippines (Cebu)
South Africa (BT20)
Malawi (LCSS)
Burkina Faso (MISAME)
Zimbabwe (ZVITAMBO)
Tanzania (Dar es-Salam)
Brazil (Pelotas 1982)
Brazil (Pelotas 1993)
Brazil (Pelotas 2004)
Subtotal (
I
2
= 0.0%, P = 0.561)
Study
1.61 (0.79, 3.28)
1.85 (1.20, 2.85)
1.05 (0.62, 1.76)
1.47 (0.61, 3.56)
1.15 (0.25, 5.25)
1.08 (0.37, 3.11)
1.46 (1.12, 1.92)
1.78 (1.17, 2.69)
1.76 (0.59, 5.22)
2.33 (1.41, 3.83)
1.76 (1.28, 2.40)
2.19 (1.81, 2.64)
1.79 (0.60, 5.29)
1.75 (1.03, 2.96)
3.16 (1.44, 6.92)
1.96 (1.10, 3.49)
1.35 (0.65, 2.81)
1.16 (0.34, 3.95)
1.81 (1.63, 2.02)
2.44 (0.48, 12.26)
3.35 (0.89, 12.60)
4.61 (1.34, 15.87)
1.33 (0.65, 2.73)
6.01 (1.99, 18.20)
0.50 (0.09, 2.87)
5.00 (1.72, 14.56)
3.45 (1.10, 10.82)
1.69 (1.13, 2.52)
1.92 (0.24, 15.72)
2.94 (0.94, 9.19)
11.43 (5.44, 24.04)
1.06 (0.36, 3.18)
2.39 (0.29, 19.63)
2.78 (1.75, 4.40)
2.29 (0.20, 25.60)
10.45 (2.01, 54.31)
5.07 (0.63, 40.89)
2.39 (1.01, 5.69)
1.12 (0.08, 16.18)
1.65 (0.47, 5.74)
4.51 (0.27, 75.59)
4.22 (2.54, 7.00)
11.95 (2.39, 59.87)
8.34 (2.16, 32.16)
3.61 (1.08, 12.08)
3.93 (2.79, 5.54)
ES (95% CI)
2.25
6.10
4.24
1.46
0.50
1.02
15.59
6.66
0.96
4.59
11.56
31.78
0.97
4.12
1.86
3.43
2.13
0.76
100.00
5.06
6.36
6.83
10.16
7.57
4.55
7.80
7.37
12.24
3.53
7.38
9.98
7.64
3.52
100.00
2.01
4.32
2.69
15.65
1.64
7.54
1.48
45.64
4.52
6.44
8.06
100.00
%
Weight
1 4
Figure 4 Random-effects model forest plot of odds ratios for childhood stunting at 12 to 60 months by gestational age
categories in children born AGA and preterm. Reference group is children born AGA and term
RISK OF CHILDHOOD UNDERNUTRITION
11
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
Table 5 Population attributable risk for childhood stunting and wasting at 12–60 months of age due to SGA and preterm birth categories, overall and by MDG
regions
SGA and term
a,b
SGA and preterm
a,b
AGA and preterm
a,b
MDG-regions [number of studies]
Prevalence
(%)
c
RR PAR (95% CI)
Prevalence
(%)
c
RR PAR (95% CI)
Prevalence
(%)
c
RR PAR (95% CI)
Stunting
Caucasus and Central Asia [19]
d
12.9 1.76 0.09 (0.07, 0.11) 2.1 2.68 0.03 (0.02, 0.05) 7.3 1.49 0.03 (0.02, 0.05)
Eastern Asia [2] 5.3 2.10 0.05 (0.03, 0.08) 1.7 7.27 0.09 (0.03, 0.16) 5.8 0.99 0.00 (0.05, 0.05)
Southeastern Asia [6] 21.2 1.51 0.10 (0.06, 0.14) 3.0 1.85 0.03 (0.01, 0.04) 10.6 1.21 0.02 (0.01, 0.03)
Southern Asia [6] 41.5 1.51 0.18 (0.11, 0.24) 2.9 1.85 0.02 (0.01, 0.04) 10.3 1.21 0.02 (0.01, 0.03)
Western Asia [6] 19.6 1.51 0.09 (0.05, 0.13) 2.2 1.85 0.02 (0.01, 0.03) 7.7 1.21 0.02 (0.01, 0.02)
Oceania [19]
d
19.4 1.76 0.13 (0.10, 0.16) 1.6 2.68 0.03 (0.02, 0.04) 5.7 1.49 0.03 (0.02, 0.04)
Northern Africa [6] 8.5 1.65 0.05 (0.03, 0.08) 1.2 2.98 0.02 (0.02, 0.03) 6.2 1.62 0.04 (0.02, 0.05)
Sub-Saharan Africa [6] 23.5 1.65 0.13 (0.08, 0.19) 2.0 2.98 0.04 (0.03, 0.04) 10.3 1.62 0.06 (0.04, 0.08)
Latin America and the Caribbean [5] 10.7 2.45 0.13 (0.03, 0.24) 1.8 3.84 0.05 (0.01, 0.09) 6.7 1.88 0.06 (0.00, 0.11)
LMIC [19] 24.7 1.76 0.16 (0.12, 0.19) 2.3 2.68 0.04 (0.02, 0.05) 9.1 1.49 0.04 (0.03, 0.06)
Wasting
Caucasus and Central Asia [19]
d
12.9 2.25 0.14 (0.12, 0.16) 2.1 3.55 0.05 (0.03, 0.07) 7.3 1.88 0.06 (0.03, 0.09)
Eastern Asia [2] 5.3 2.55 0.08 (0.02, 0.14) 1.7 1.00 0.00 5.8 2.61 0.08 (0.09, 0.26)
Southeastern Asia [6] 21.2 2.19 0.20 (0.17, 0.24) 3.0 2.88 0.05 (0.02, 0.08) 10.6 1.53 0.05 (0.02, 0.12)
Southern Asia [6] 41.5 2.19 0.33 (0.28, 0.38) 2.9 2.88 0.05 (0.02, 0.08) 10.3 1.53 0.05 (0.02, 0.12)
Western Asia [6] 19.6 2.19 0.19 (0.16, 0.22) 2.2 2.88 0.04 (0.02, 0.06) 7.7 1.53 0.04 (0.01, 0.09)
Oceania [19]
d
19.4 2.25 0.19 (0.17, 0.22) 1.6 3.55 0.04 (0.02, 0.06) 5.7 1.88 0.05 (0.02, 0.07)
Northern Africa [6] 8.5 2.26 0.10 (0.07, 0.12) 1.2 3.09 0.02 (0.01, 0.04) 6.2 2.11 0.06 (0.03, 0.10)
Sub-Saharan Africa [6] 23.5 2.26 0.23 (0.19, 0.27) 2.0 3.09 0.04 (0.02, 0.06) 10.3 2.11 0.10 (0.05, 0.15)
Latin America and the Caribbean [5] 10.7 3.69 0.22 (0.07, 0.38) 1.8 20.79 0.26 (0.08, 0.44) 6.7 3.20 0.13 (0.06, 0.31)
LMIC [19] 24.7 2.25 0.24 (0.21, 0.26) 2.3 3.55 0.06 (0.03, 0.08) 9.1 1.88 0.07 (0.03, 0.11)
AGA, adequate-for-gestational age; SGA, small-for-gestational age; PAR, population attributable risk; LMIC, low- and middle-income countries.
a
AGA defined as size 510th percentile for gestational age; SGA defined as size <10th percentile for gestational age.
b
Term defined as delivery at gestational age 537 weeks; preterm defined as delivery at gestational age <37 weeks.
c
Estimates of prevalence extracted from the concurrent CHERG working group.
8
d
No study from this region; RR for all LMIC used for PAR calculation.
12 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
pursued separately in cohorts which have data on
birth length. Also, conditional growth analysis
would have enabled us to account for the contribution
of different periods of postnatal growth.
Nutritional interventions during pregnancy have
been shown to be beneficial for foetal growth, but
less so for increasing gestational age, except perhaps
for prenatal zinc.
43
Recent meta-analyses of food (cal-
orie and protein) supplementation during pregnancy
showed a reduction of 34% in the risk of SGA
(RR ¼ 0.68, 95% CI: 0.51, 0.92),
44
and meta-analyses
of randomized controlled trials of daily prenatal mul-
tiple micronutrient supplementation (without food)
also show a significant reduction in SGA (pooled
RR ¼ 0.83, 95% CI: 0.73, 0.95).
45
Combining these nu-
tritional intervention approaches in settings where
maternal undernutrition is high, and multiple micro-
nutrient deficiencies common,
46
would not only be
important for addressing the huge burden of foetal
growth restriction, but may also result in improved
childhood nutritional status. Prenatal food supple-
mentation has been shown to reduce stunting in chil-
dren,
47
but not with multiple micronutrients,
47,18
although weight and circumferential measurements
improved at 2 years of age.
18
Integration of prenatal
and preschool balanced protein-calories with other
public health programmes in a study in India was
associated with improved height and healthier profiles
of cardiovascular disease risk.
48
Pre-pregnancy maternal height and BMI are strong
determinants of low birthweight and SGA, in addition
to maternal weight gain during pregnancy.
49
An in-
verse association between maternal height and child
stunting has been found in a pooled analysis of DHS
data,
50
although the data were cross-sectional.
51
Pre-
pregnancy maternal height, an indicator of long-term
nutritional status and other exposures, is linked to
uterine volume and is a well-known predictor of
foetal growth and birth size.
52
In many LMIC settings
the intergenerational cycle of growth failure links
small maternal size to the mother’s size at birth and
growth in childhood and adolescence. Thus, there is
evidence that factors that well precede the pregnancy
are strong and robust predictors of childhood under-
nutrition. Preconceptional interventions for improving
maternal pre-pregnancy BMI, and interventions that
can influence linear growth and attained adult height
are therefore likely to benefit outcomes of foetal
growth.
It is well recognized that postnatal growth tracks,
although conventionally ‘catch-up’, is believed to
occur among about 50% of children to achieve a
normal pattern of growth and to meet genetic poten-
tial, at least in well-nourished environments.
53
In
undernourished settings, our analyses allowed us to
estimate the extent of tracking between the prenatal
and postnatal periods. Factors and practices contribut-
ing to growth faltering in the first few years of life
include: inadequate exclusive breastfeeding; fre-
quency, energy density and micronutrient levels of
complementary foods; and infectious morbidity. In
many resource-poor settings, the basic and underlying
causes of both foetal and childhood undernutrition
are common and need to be uniformly addressed to
improve growth and development in the first 1000
days of life.
Our analysis revealed an independent relationship,
stronger for foetal growth restriction than preterm,
with childhood stunting and wasting across regions,
suggesting that child growth and nutritional status
may be strongly linked to foetal life and in part pre-
natal in origin, suggesting a need to intervene during
an earlier life stage with the focus on pregnancy nu-
trition.
52
Addressing the global problem of childhood
stunting requires a life-stage approach, with interven-
tions targeting the pregnancy period even as efforts
Table 6 Population attributable burden
a
for childhood stunting and wasting by low birthweight, SGA and preterm birth
LBW SGA Preterm
MDG-region Stunting Wasting Stunting Wasting Stunting Wasting
Caucasus and Central Asia 78 199 26 128 172 679 57 045 96 794 33 493
Eastern Asia 328 108 184 427 1 117 785 152 087 701 034 169 982
Southeastern Asia 1 043 320 599 275 1 825 529 1 332 796 686 975 558 854
Southern Asia 93 42 401 5 971 181 13 604 176 10 677 979 3 069 295 2 907 789
Western Asia 281 569 84 066 482 464 183 505 150 424 63 847
Oceania 33 936 13 005 61 888 23 516 21 730 8 815
Northern Africa 149 313 86 705 255 232 144 422 204 120 106 770
Sub-Saharan Africa 5 897 206 2 203 385 9 492 042 3 535 281 5 450 969 1 884 386
Latin America and the Caribbean 670 169 259 594 1 149 626 386 248 650 828 309 578
Pooled (LMIC) 20 290 635 9 508 034 32 062 280 14 994 233 13 217 176 6 671 370
SGA, small-for-gestational age, defined as size <10th percentile for gestational age. LBW, defined as <2.5 kg measured within 72 h
of birth; LMIC, low- and middle-income countries.
a
Attributable burden calculated using estimated number of children stunted or wasted by MDG region, 2011.
RISK OF CHILDHOOD UNDERNUTRITION 13
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
for improving breastfeeding and complementary feed-
ing practices in the first 2 years of life are continued.
The need for such a life-course approach to interven-
ing is reflected in the emphasis on the first 1000 days.
Supplementary Data
Supplementary data are available at IJE online.
Funding
This analytical work was supported by the Bill and
Melinda Gates Foundation [810-2054], Seattle,
Washington, USA, to the US Fund for UNICEF to
support the work of the Child Health Epidemiology
Reference Group. Financial support for analysis was
offered to investigators (CV, JH) through a subcon-
tract mechanism administered by the US Fund for
UNICEF.
Acknowledgements
We thank the following co-investigators and contribu-
tors to the original studies:
Lotta Alho,
I
Bernard Chasekwa,
II
Michael Dibley,
III
Jacqueline Hill,
IV
Yongli Lang,
V
Hermann Lanou,
VI
Kenneth Maleta,
VII
Dharma Manandhar,
VIII,IX
Bernard Nahlen,
X–XIII
Laetitia Nikie
`
ma,
XIV
Anja
Terlouw,
XII,XIII,XV
Laeticia Celine Toe,
XVI
Anjana
Vaidya,
XVII
Sargoor Veena
XVIII
and Ni Zhou
V
.We
also thank Myra Shapiro
XIX
for assistance with the
literature review and Lee Wu
XIX
for managing the
datasets and codebooks.
I
University of Tampere, School of Medicine, Tampere,
Finland,
II
Zvitambo Project, Harare, Zimbabwe,
III
International School of Public Health, University of
Sydney, New South Wales, Australia,
IV
Medical
Research Council Epidemiology Resource Centre,
Southampton General Hospital, Southampton, UK
V
Tianjin Women‘s and Children‘s Health Center,
Tianjin, China,
VI
Institut de Recherche en Sciences
de la Sante
´
, Ouagadougou, Burkina Faso,
VII
Community Health, College of Medicine,
University of Tampere, Tampere, Finland,
VIII
Mother
and Infant Research Activities (MIRA), Kathmandu,
Nepal,
IX
Deptartment of Pediatrics, Kathmandu
Medical College, Kathmandu, Nepal,
X
Centers for
Disease Control and Prevention, Atlanta, GA, USA,
XI
President’s Malaria Initiative, Washington, DC,
USA,
XII
KEMRI/CDC Research and Public Health
Collaboration, Kisumu, Kenya,
XIII
Centers for Disease
Control and Prevention, Kenya
XIV
Nutrition Unit, Institut de Recherche en Sciences
de la Sante
´
, Ouagadougou, Burkina Faso,
XV
Child and
Reproductive Health Group, Liverpool School of
Tropical Medicine, Liverpool, UK,
XVI
Centre Muraz,
Burkina Faso
XVII
Institute for Global Health, UCL Institute of Child
Health, London, UK,
XVIII
Epidemiology Research Unit,
CSI Holdsworth Memorial Hospital, Mysore, India
and
XIX
Department of International Health,
Bloomberg School of Public Health, Johns Hopkins
University, Baltimore, MD, USA
Conflict of interest: None declared.
KEY MESSAGES
The extent to which stunting, common in low- and middle-income countries, is associated with foetal
growth restriction and preterm birth—two causes of low birth weight—remains to be established.
A meta-analysis of 19 longitudinal birth cohorts revealed small-for-gestational age (SGA) and pre-
term to be each associated with a 2.4 and 1.9 times increased odds of stunting, with both conditions
being associated with a 4.5 times the risk.
The population attributable risk for overall SGA for outcomes of childhood stunting and wasting was
20% and 30%, respectively.
Childhood undernutrition is high in many settings; our analysis reveals that it may have its origins in
part in the foetal period suggesting the need for early life interventions, especially during pregnancy.
References
1
United Nations Children’s Fund, World Health
Organization, World Bank. UNICEF-WHO-World Bank
Joint Child Malnutrition Estimates. New York, Geneva,
Washington, DC: UNICEF, WHO, World Bank, 2012.
2
Black RE, Allen LH, Bhutta ZA et al. Maternal and child
undernutrition: global and regional exposures and health
consequences. Lancet 2008;371:243–60.
3
Pelletier DL. The relationship between child anthropom-
etry and mortality in developing countries: implications
for policy, programs and future research. J Nutr 1994;124:
S2047–S81.
4
Victora CG, Adair L, Fall C et al. Maternal and child
undernutrition: consequences for adult health and
human capital. Lancet 2008;371:340–57.
5
Victora CG, de Onis M, Hallal PC, Blossner M,
Shrimpton R. Worldwide timing of growth faltering:
14 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
revisiting implications for interventions. Pediatrics 2010;
125:e473–80.
6
Christian P. Prenatal origins of malnutrition. In:
Kalhan SC, Prentice AM, Yajnik CS (eds). Emerging
Societies—Coexistence of Childhood Malnutrition and Obesity.
Vol 63, Vevey, Switzerland: S Karger Ag, 2009.
7
Blencowe H, Cousens S, Oestergaard MZ et al. National,
regional, and worldwide estimates of preterm birth rates
in the year 2010 with time trends since 1990 for selected
countries: a systematic analysis and implications. Lancet
2012;379:2162–72.
8
Lee ACC, Katz J, Blencowe J et al. Born too small: na-
tional and regional estimates of term and preterm small-
for-gestational-age in 138 low-middle income countries
in 2010. Lancet Global Health 2013 (in press).
9
United Nations Subcommittee on Nutrition. Third Report
on the World Nutrition Situation: A Report Compiled From
Information Available to the ACC/SCN. Geneva,
Switzerland: United Nations, 1997.
10
Fall CH, Fisher DJ, Osmond C, Margetts BM. Multiple
micronutrient supplementation during pregnancy in
low-income countries: a meta-analysis of effects on
birth size and length of gestation. Food Nutr Bull 2009;
30:S533–46.
11
Richter LM, Victora CG, Hallal PC et al. Cohort profile: the
consortium of health-orientated research in transitioning
societies. Int J Epidemiol 2012;41:621–26.
12
Alexander GR, Himes JH, Kaufman RB, Mor J, Kogan M.
A United States national reference for fetal growth. Obstet
Gynecol 1996;87:163–68.
13
World Health Organization Multicentre Growth Reference
Study Group. WHO Child Growth Standards based on
length/height, weight and age. Acta Paediatr Suppl 2006;
450:76–85.
14
Deeks JJ, Altman DG, Bradburn MJ. Statistical methods
for examining heterogeneity and combining results from
several studies in meta-analysis. In: Egger M,
Davey Smith G, Altman DG (eds). Systematic Reviews in
Health Care: Meta-Analysis in Context. 2nd edn. London:
BMJ Books, 2001.
15
Zhang J, Yu KF. What‘s the relative risk? A method of
correcting the odds ratio in cohort studies of common
outcomes. JAMA 1998;280:1690–91.
16
Levin ML. The occurrence of lung cancer in man. Acta
Unio Int Contra Cancrum 1953;9:531–41.
17
Rao CR. Linear statistical inference and its application. New
York: John Wiley, 1965.
18
Vaidya A, Saville N, Shrestha BP, Costello AM,
Manandhar DS, Osrin D. Effects of antenatal multiple
micronutrient supplementation on children‘s weight and
size at 2 years of age in Nepal: follow up of a double-
blind randomised controlled trial. Lancet 2008;371:
492–99.
19
Tielsch JM, Darmstadt GL, Mullany LC et al. Impact of
newborn skin-cleansing with chlorhexidine on neonatal
mortality in southern Nepal: a community-based, cluster-
randomized trial. Pediatric. 2007;119:e330–40.
20
Persson L-A
˚
, Arifeen S, Ekstrom E-C et al. Effects of pre-
natal micronutrient and early food supplementation on
maternal hemoglobin, birthweight, and infant mortality
among children in Bangladesh: the MINIMat randomized
trial. JAMA 2012;307:2050–59.
21
Joglekar CV, Fall CH, Deshpande VU et al. Newborn size,
infant and childhood growth, and body composition and
cardiovascular disease risk factors at the age of 6 years:
the Pune Maternal Nutrition Study. Int J Obes (Lond)
2007;31:1534–44.
22
Krishnaveni GV, Hill JC, Veena SR et al. Truncal adiposity
is present at birth and in early childhood in South Indian
children. Indian Pediatr 2005;42:527–38.
23
Zeng L, Dibley MJ, Cheng Y et al. Impact of micronutrient
supplementation during pregnancy on birthweight, dur-
ation of gestation, and perinatal mortality in rural west-
ern China: double blind cluster randomised controlled
trial. BMJ 2008;337:a2001.
24
Zhang X, Liu E, Tian Z et al . High birthweight and over-
weight or obesity among Chinese children 3–6 years old.
Prev Me. 2009;49:172–78.
25
Adair LS, Guilkey DK. Age-specific determinants of stunt-
ing in Filipino children. J Nutr 1997;127:314–20.
26
Richter L, Norris S, Pettifor J, Yach D, Cameron N. Cohort
Profile: Mandela‘s children: the 1990 Birth to Twenty
study in South Africa. Int J Epidemiol 2007;36:504–11.
27
Huybregts L, Roberfroid D, Lanou H et al. Prenatal food
supplementation fortified with multiple micronutrients
increases birth length: a randomized controlled trial in
rural Burkina Faso. Am J Clin Nutr 2009;90:1593–600.
28
Roberfroid D, Huybregts L, Lanou H et al. Effects of ma-
ternal multiple micronutrient supplementation on fetal
growth: a double-blind randomized controlled trial in
rural Burkina Faso. Am J Clin Nutr 2008;88:1330–40.
29
Roberfroid D, Huybregts L, Lanou H et al. Impact of pre-
natal multiple micronutrients on survival and growth
during infancy: a randomized controlled trial. Am J Clin
Nutr 2012;95:916–24.
30
Malaba LC, Iliff PJ, Nathoo KJ et al. Effect of postpartum
maternal or neonatal vitamin A supplementation on
infant mortality among infants born to HIV-negative
mothers in Zimbabwe. Am J Clin Nutr 2005;81:454–60.
31
Maleta K, Virtanen SM, Espo M, Kulmala T, Ashorn P.
Childhood malnutrition and its predictors in rural
Malawi. Paediatr Perinat Epidemiol 2003;17:384–90.
32
Fawzi WW, Msamanga GI, Urassa W et al. Vitamins and
perinatal outcomes among HIV-negative women in
Tanzania. N Engl J Med 2007;356:1423–31.
33
ter Kuile FO, Terlouw DJ, Kariuki SK et al. Impact of
permethrin-treated bed nets on malaria, anemia, and
growth in infants in an area of intense perennial malaria
transmission in western Kenya. Am J Trop Med Hyg 2003;
68:68–77.
34
Victora CG, Barros FC. Cohort Profile: the 1982 Pelotas
(Brazil) birth cohort study. Int J Epidemiol 2006;35:237–42.
35
Victora CG, Hallal PC, Araujo CL, Menezes AM, Wells JC,
Barros FC. Cohort Profile: the 1993 Pelotas (Brazil) birth
cohort study. Int J Epidemiol 2008;37:704–09.
36
Santos IS, Barros AJ, Matijasevich A, Dominques MR,
Barros FC, Victora CG. Cohort Profile: the 2004 Pelotas
(Brazil) birth cohort study. Int J Epidemiol 2011;40:
1461–68.
37
Stein AD, Melgar P, Hoddinott J, Martorell R. Cohort
Profile: the Institute of Nutrition of Central America
and Panama (INCAP) Nutrition Trial Cohort Study. Int
J Epidemiol 2008;37:716–20.
38
Ramakrishnan U, Stein AD, Parra-Cabrera S et al. Effects
of docosahexaenoic acid supplementation during preg-
nancy on gestational age and size at birth: randomized,
double-blind, placebo-controlled trial in Mexico. Food Nutr
Bull 2010;31:S108–16.
39
Katz J, Lee ACC, Kozuki N et al. Born too small or too
soon: a pooled analysis of mortality risk among preterm
RISK OF CHILDHOOD UNDERNUTRITION
15
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
and small-for-gestational-age infants in low- and middle-
income countries. Lancet 2013 (In press).
40
Lawn JE, Cousens S, Zupan J. 4 million neonatal deaths:
When? Where? Why? Lancet 2005;365:891–900.
41
Schroeder DG. Malnutrition. In: Semba RD, Bloem M
(eds). Nutrition and Health in Developing Countries.
2nd edn. Totowa, NJ: Springer, 2008.
42
Schmidt MK, Muslimatun S, West CE, Schultink W,
Gross R, Haitvast JG. Nutritional status and linear
growth of Indonesian infants in West Java are deter-
mined more by prenatal environment than by postnatal
factors. J Nutr 2002;132:2202–07.
43
Chaffee BW, King JC. Effect of zinc supplementation on
pregnancy and infant outcomes: a systematic review.
Paediatr Perinat Epidemiol 2012;26:118–37.
44
Imdad A, Bhutta ZA. Maternal nutrition and birth out-
comes. Effect of balanced protein-energy supplementa-
tion. Paediatr Perinat Epidemiol 2012;26:178–90.
45
Ramakrishnan U, Grant FK, Goldenberg T, Bui V,
Imdad A, Bhutta ZA. Effect of multiple micronutrient sup-
plementation on pregnancy and infant outcomes: a sys-
tematic review. Paediatr Perinat Epidemiol 2012;26:153–67.
46
Jiang T, Christian P, Khatry SK, Wu L, West KP Jr.
Micronutrient deficiencies in early pregnancy are
common, concurrent, and vary by season among rural
Nepali pregnant women. J Nutr 2005;135:1106–12.
47
Khan AI, Kabir I, Ekstrom E-C et al. Effects of prenatal
food and micronutrient supplementation on child growth
from birth to 54 months of age: a randomized trial in
Bangladesh. Nutr J 2011;10:134.
48
Kinra S, Rameshwar Sarma KV, Ghafoorunissa et al.
Effect of integration of supplemental nutrition with
public health programmes in pregnancy and early child-
hood on cardiovascular risk in rural Indian adolescents:
long term follow up of Hyderabad nutrition trial. BMJ
2008;337:a605.
49
Kramer MS. The epidemiology of adverse pregnancy out-
comes: an overview. J Nutr 2003;133:1592S–96S.
50
Ozaltin E, Hill K, Subramanian SV. Association of mater-
nal stature with offspring mortality, underweight, and
stunting in low- to middle-income countries. JAMA
2010;303:1507–16.
51
Christian P. Maternal height and risk of child mortality
and undernutrition. Editorial. JAMA 2010;303:1539–49.
52
Christian P. Nutrition and maternal mortality in develop-
ing countries. In: Lammi-Keefe CJ, Couch SC,
Philipson EH (eds). Handbook of Nutrition and Pregnancy.
Totowa, NJ: Humana Press, 2008.
53
Demerath EW, Choh AC, Czerwinski SA et al. Genetic and
environmental influences on infant weight and weight
change: the Fels Longitudinal Study. Am J Hum Biol
2007;19:692–702.
16 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
at Uppsala Universitetsbibliotek on August 12, 2013http://ije.oxfordjournals.org/Downloaded from
