Size at Birth, Weight Gain in Infancy and Childhood, and Adult Diabetes Risk in Five Low- or Middle-Income Country Birth Cohorts

Medical Research Council/Wits Developmental Pathways for Health Research Unit, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa.
Diabetes care (Impact Factor: 8.42). 11/2011; 35(1):72-9. DOI: 10.2337/dc11-0456
Source: PubMed
ABSTRACT
We examined associations of birth weight and weight gain in infancy and early childhood with type 2 diabetes (DM) risk in five cohorts from low- and middle-income countries.
Participants were 6,511 young adults from Brazil, Guatemala, India, the Philippines, and South Africa. Exposures were weight at birth, at 24 and 48 months, and adult weight, and conditional weight gain (CWG, deviation from expected weight gain) between these ages. Outcomes were adult fasting glucose, impaired fasting glucose or DM (IFG/DM), and insulin resistance homeostasis model assessment (IR-HOMA, three cohorts).
Birth weight was inversely associated with adult glucose and risk of IFG/DM (odds ratio 0.91[95% CI 0.84-0.99] per SD). Weight at 24 and 48 months and CWG 0-24 and 24-48 months were unrelated to glucose and IFG/DM; however, CWG 48 months-adulthood was positively related to IFG/DM (1.32 [1.22-1.43] per SD). After adjusting for adult waist circumference, birth weight, weight at 24 and 48 months and CWG 0-24 months were inversely associated with glucose and IFG/DM. Birth weight was unrelated to IR-HOMA, whereas greater CWG at 0-24 and 24-48 months and 48 months-adulthood predicted higher IR-HOMA (all P < 0.001). After adjusting for adult waist circumference, birth weight was inversely related to IR-HOMA.
Lower birth weight and accelerated weight gain after 48 months are risk factors for adult glucose intolerance. Accelerated weight gain between 0 and 24 months did not predict glucose intolerance but did predict higher insulin resistance.
Size at Birth, Weight Gain in Infancy
and Childhood, and Adult Diabetes Risk
in Five Low- or Middle-Income Country
Birth Cohorts
SHANE A. NORRIS, PHD
1
CLIVE OSMOND, PHD
2
DENISE GIGANTE, PHD
3
CHRISTOPHER W. KUZAWA, PHD
4
LAKSHMY RAMAKRISHNAN, PHD
5
NANETTE R. LEE, PHD
6
MANUAL RAMIREZ-ZEA, PHD
7
LINDA M. RICHTER, PHD
1
ARYEH D. STEIN, PHD
8
NIKHIL TANDON, MD
9
CAROLINE H.D. FALL, DM
2
THE COHORTS GROUP*
OBJECTIVE dWe examined associations of birth weight and weight g ain in infancy and early
childhood with type 2 diabetes (DM) risk in ve cohorts from low- and mi ddle-i ncome countries.
RESEARCH DESIGN AND METHODSdParticipants were 6,511 young adults from
Brazil, Guatemala, India, the Philippines, and South Africa. Exposures were weight at birt h, at
24 and 48 months, and adult weight, and conditional we ight gain (CWG, deviation from expec-
ted we ight gain) between these ages. Outcomes were adult fasting glucose, impaired fasting
glucose or DM (IFG/DM), and insulin resistance homeostas is model assessment (IR-HOMA,
three cohorts).
RESULTSdBirth weight was inversely associate d with adult glucose and risk of IFG/DM (odds
ratio 0.91[95% CI 0.840.99] per SD). Weig ht at 24 and 48 months and CWG 024 and 2448
months were unrelated to glucose and IFG/DM; however, CWG 48 monthsadulthood was
positively related to IFG/ DM (1.32 [1.221.43] per SD). After adjusti ng for adult waist circum-
ference, birth weight, weight at 24 and 48 months and CWG 024 months were inversely
associated with glucose and IFG/ DM. Birth weight was unrelated to IR -HOMA, whereas greater
CWG at 024 and 2448 months and 48 mo nthsadulthood predicted higher IR-HOMA (all
P , 0.001). After adjusting for adult waist circumference, birth weight was inversely related to
IR-HOMA.
CONCLUSIONSdLower birth weight and accelerated weight gain after 48 months are risk
factors for adult glucose intolerance. Accelerated weight gain between 0 and 24 months did not
predi ct glucose intolerance but did predict higher insulin resistance.
Diabetes Care 35:7279, 2012
R
ecently, Whincup et al. (1) concluded
that birth weight is inversely a s-
sociated with the development
o f type 2 diabetes (DM) and that this
association is strengthened after adjusting
for adult BMI. Studies from high-income
coun tries have sho wn th at rapid weight
gain in childhood or adult life is associated
with an increased incidence of DM
and insulin resistance (2). Therefore, im-
paired fetal grow th and excess postnatal
weight gain are both potential precursors
to adult DM.
Four-fths of all individuals with DM
live in low- and middle-income countries
(LMICs) (3). Many of these countries are
undergoing swift nutritional and eco-
nomic transitions, exposing i ndividuals
to environmental conditions that pro-
mote weight gain. The combination of
early-life undernutrition and overnutri-
tion in adulthood may be fueling the ep-
idemic of DM in LMICs (4).
Few studies have examined childhood
weight gain in relation to adult diabetes in
LMICs. Gestation and the rst 2 postnatal
years (the rst 1,000 days;http://www.
thousanddays.org) are the time when
childrens growth in LMICs falls most rap-
idly below international reference values
(5) and, hence, provide a signicant win-
dow of opportunity for improved infant
survival, cognitive development, and adult
economic status (6,7). A critical public
health question for LMICs is whether pro-
moting ea rly-life weight gain to achieve
improvements in human capital could
have adverse effects on adult diabetes
risk. Data from birth cohorts in the U.K.
and Finland indicate that, as with bi rth
weight, lower weight at 1 year is associated
with an increased risk of DM (8,9). Other
studies have shown that greater weight or
weight gain at this age is associated with an
increased risk of obesi ty ( 10,11), which
could increase diabetes risk.
To clarify relationships between
early-weight dynamics and adult diabetes
risk, we pooled data from ve birth cohort
studies in LMIC s and invest igated associ-
ations of weight at birth, 24 months, 48
months, and young adulthood and con-
ditional weight gain (CWG) between
these ages, with adult-fasting glucose
concentrations, risk of glucose intoler-
ance, and insulin resistance. We hypoth-
esized that lower birth weight and infant
CWG but higher CWG after infancy
woul d predict increase d risk.
cccccccccccccccccccccccccccccccccccccccccccccccc c
From the
1
Medical Research Council/Wits Developmental Pathways for Health Research Unit, Faculty of
Health Sciences, University of the Witwatersrand, Johannesburg, South Africa; the
2
Medical Research
Council Lifecourse Epidemiology Unit, University of Southampton, Southampton, U.K.; the
3
Universidade
Federal de Pelotas, Pelotas, Rio Grande do Sul, Brazil; the
4
Department of Anthropology, Northwestern
University, Evanston, Illinois; the
5
Department of Cardiac Biochemistry, All India Institute of Medical
Sciences, New Delhi, India; the
6
Ofce of Population Studies Foundation, Cebu City, Philippines; the
7
Institute of Nutrition of Central America and Panama, Guatemala City, Guatemala; the
8
Hubert De-
partment of Global Health, Emory University, Atlanta, Georgia; and the
9
Department of Endocrinology, All
India Institute of Medical Sciences, New Delhi, India.
Corresponding author: Shane A. Norris, san@global.co.za.
Received 6 March 2011 and accepted 7 October 2011.
DOI: 10.2337/dc11-0456
*A complete list of the COHORTS group members can be found in the
APPENDIX.
© 2012 by the American Diabetes Association. Readers may use this article as long as the work is properly
cited, the use is educational and not for prot, and the work is not altered. See http://creativecommons.org/
licenses/by-nc-nd/3 .0/ for details.
72 DIABETES CARE, VOLUME 35, JANUARY 2012 care.diabetesjournals.org
Epidemiology/Health Services Research
ORIGINAL ARTICLE
Page 1
RESEARCH DESIGN AND
METHODS
Study populations
The ve cohorts (Table 1) include the
1982 Pelotas Birth Cohort (Brazil) (12), the
Institute of Nutrition of Central America
and Panama Nutrition Trial Coh ort Study
(INTCS, Guatemala) (13), the New Delhi
Birth Cohort Study (India) (14), the Cebu
Longitudinal Health and Nutrition Survey
(CLHNS, the Philippines) (15), a nd the
Birth to Twenty cohort (BTT, South Africa)
(16). All cohort participants were adults
at the time the outcome variables were mea-
sured, with mean (SD) ages of 22.7 (0.4),
32.4 (4.1), 29.2 (1.4), and 21.2 (0.9) years
for Brazil, Guatemala, India, and the
Philippines, respectively, with the excep-
tion of the South Africans, who were aged
15.5 (0.3) years. For all cohorts, appropri-
ate institutional research ethics committ ee
approval was granted, and informed
consent was obtained from participants
or their parents, as appropriate.
Exposures: weights at birth,
24 months, and 48 months
In Brazil, India, and Guatemala, birth
weight was measured by research teams.
In the Philippines, birth weight was
measured by birth attendants using hang-
ing sca les for home births and was ob-
tained from hospital records for hospital
births. In South Africa, birth weight was
measured by birth att endants in the
hospitals, and these were obtained from
hospital birth records. In all sites, post-
natal weights were measured by research
teams using standardized methods.
Weight at 24 months was available at all
sites. Weight at 48 months was available
in three cohorts (Brazil, Guatemala, and
India). In South Africa and the Philippines,
weights obtained at the ne arest possible
age (60 and 102 months, respectively)
were used.
Gestational age
Gestational age was based on the date of
the mothers last menstrual period, except
in the Philippines, where Ballard scores
obtained by clinical assessment of the
newborns neuromuscular and physical
characteristics (17) were used for partic-
ipants with low birth weight.
Adult anthropometry
Adult he ight and we ight were mea-
s ur e d using standardized techniques.
Waist circumference was measured
using a measuring tape to the nearest
0.1 cm at the umbilicus (Guatemala, the
Philippines, and South Africa), narrowest
part of the trunk (Brazil), or midway be-
tween the lower costal margin and supe-
rior iliac crest laterally (India).
Outcomes: adult glucose and
insulin
We considered three adult outc omes as
follows: fasting plasma glucose concen-
tration, the combined preval ence of im-
paired fasting glucose or DM (IFG/DM),
and insulin resistance. In all cohorts , with
the exception of Brazil, the research team
obtained a fasting sample and determined
glucose using site-specic procedu res
(Table 1). In Brazil, random blood sam-
ples were taken, and values were adjusted
for time since the last meal (18). Because
glucometers overestimate glucose con-
centrations in whole venous blood com-
pared with standard laboratory methods
(19,20), we subtracted 0.97 mmol/L from
the values in the Philippines cohort to ob-
tain the best equivalent to venous plasma
as analyzed by a laboratory autoanalyzer
(19). We dened DM as glucose concen-
tration $7.0 mmol/L and IFG as glucose
$6.1 and ,7.0 mmol/L (21). Three sites
(Delhi, the Philippines, and South Africa)
measured insulin (Table 1). Insulin re-
sistance hom eostasis model assessment
(IR-HOMA) was calculated using the
HOMA 2 calculator (22).
Analytic sample
We included 8,746 participants with com-
plete data for weight at birth, 24 months,
48 months, and adulthood (Table 1). Of
these, 6,511 had adult plasma glucose con-
centrations, and6,503 had glucose concen-
trations and waist circumference. Models
for IR-HOMA were based on 3,202 partic-
ipants from India, the Philippines, and
South Africa.
Data management and statistical
methods
Fasting glucose and IR-HOMA had
skewed distributions and were log-
transformed. Measures of weight were
converted into sex-specic SD scores us-
ing the World Health Organization Growth
Standards and further standardized within
each site to create variables with a mean
of0andaSDof1.Weimputed48-month
SD scores for pa rticipants in South Africa
and the Philippines, assuming a linear
change in the SD score from 24 to 60 or
102 months, respectively. Small for ges-
tational age (SGA) was dened as a birth
weight below the age and sex-speci
c 10th
percentile of the birth weight distribution
published by Williams et al. (23).
To focus on weight gain in the three
intervals of 024 months, 2448 months,
and 48 months adulthood, we used
CWG variables. These repre sent how
much a child has exceeded or fallen short
of expected weight at a particular age,
given his or her earlier weight trajectory
and th e weight trajectories of the who le
population. They are calculated sep-
arately for ea ch site and sex as the stan-
dardized residual from a linear regression
model in which all weights measured up
to the beginning of the interval were used
as predictors. By construction, birth
weight and the three CWG measures are
all uncorrelated (24,25).
Anal yses were conducted using SPSS
18.0 software (IBM SPSS, Armonk, NY).
We used x
2
and t tests to assess dif-
ferences between the included and
excluded participants. Associations be-
tween weight and CWG in early life (ex-
posures) and adult outcomes were
assessed within each of the 10 combina-
tionsofsiteandsex,andtheninpooled
data, adjusted for site, sex, and adult age.
Additional models were adjusted se-
quentially for adult waist circumference,
BMI, and height. We used linear regres-
sion for continuous outcomes and logis-
tic regression for binary outcomes. We
tested for heterogeneity across the 10
site-sex strata using F tests from nested
linear models and x
2
tests based on the
difference in deviance from nested logistic
regression models. Because the study in
Guatemala was an interve ntion trial, we
checked for heterogeneity of effects o n
outcomes between the intervention a nd
control arms of the trial by testing for inter-
actions between early-life weights and in-
tervention group.
RESULTS
Characteristics of participants
in the ve cohorts
Particip ants included in the analysis sam-
ple were similar to those excluded with
respect to sex, birth weight, infant weight,
48-month weight, adult weight, and adult
waist circumference (all P . 0.10). Indian
participants had th e lo west birth weights
and the highest prevalence of SGA births
(Table 2) . Brazilian participants had the
highest birth weights and adult heights.
South African partic ipants were the heavi-
est at 24 and 48 months. The preval ence
of IFG/DM was highest in India and low-
est in the Philippi nes.
care. diabetesjournals.org DIABETES CARE, VOLUME 35, JANUARY 2012 73
Norris an d Associates
Page 2
Table 1dCharacteristics of the ve cohorts, glucose and insulin sample collection, and measurement protocols
Cohort Design
Cohort inception
year
Initial cohort
sample
Cohort target
sample* Cohort description
Gluco se blood collection
and measurement
Insulin blood collection
and measurement
Pelotas Birth
Cohort
(Brazil)
Prospective
cohort
1982 5,914 4,387; 70.3% Children born in the cit ys
maternity hospital (.99%
of all births) in 198 2. All
social economic groups
included
Random nger-prick
capillary whole-blood
sample analyzed by glucometer
(Accu-Check Advantage,
Roche Ltd)
Not collected
INTCS
(Guatemala)
Community
trial
19691977 2,393 354; 50.8% Intervention trial of a
high-energy and protein
supplement in women,
and children ,7 years in
1969 and born during
19691977 in four villages
Fasting nger-prick capillary
plasma sample analyzed by
autoanalyzer using standard
enzymatic methods
Not collected
New Delhi
Birth
Cohort
Study
(India)
Prospective
cohort
196972 8,181 1,284; 96.9% Babies born to an identied
population of married
women living in a dened
area of Delhi. Primarily
middle-class sample
Fasting venous plasma sample
analyzed by autoanalyzer using
standard enzymatic methods
Fasting venous plasma sample
and specic insulin determined
by radioimmunoassay
(Coat-a-Count Insul in Kit,
Diagnostic Products Corp.,
Los Angeles, CA)
CLHNS (the
Philippines)
Prospective
cohort
19831984 3,080 2,080; 78.0% Pregnant women living in
33 randomly sel ected
neighborhoods; 75%
urban. All social economic
groups included
Fasting venous whole-blood
sample analyzed using a
glucometer (OneTouch,
Johnson & Johnson Ltd.)
Fasting venous plasma samples
and specic insulin analyzed
by Siemens Centaur XP clinical
chemistry
BTT (South
Africa)
Prospective
cohort
1990 3,273 641; 60.1% Babies born to pregnant
women living in a dened
urban geographic area in
Johannesburg. Predomin antly
poor, black sample
Fasting venous plasma sample
analyzed by an autoanalyzer
using standard enzymatic
methods
Fasting venous plasma sample
and insulin (cross-)reactivity
with proinsulin 8.5%) analyzed
by Immulite (Siemens
Chemiluminescent Technology)
*Participants who had birth weight, weight in infancy, and weight in 48 months and percentage of target sample with glucose measurements.
74 DIABETES CARE, VOLUME 35, JANUARY 2012 care.diabetesjournals.org
Infant and child growth and adult diabetes risk
Page 3
Table 2dCharacteristics of the study sample by sex and cohort location
Cohort N
Early life Adult
SGA
(%)
Weight (kg)
Age
(years)
Weight
(kg)
Height
(cm)
BMI
(kg/m
2
)
Waist
circ.
(cm)
Glucose
concentra tion
(mmol/L)*
Insulin
concentration
(pmol/L)* IFG (%)
Diabetes
(%)
IR-
HOMA*Birth
24
months
48
months
Male
Brazil 1,58 7 15 3.3 11.4 15.8 22.7 72.2 173.7 23.9 90.0 5.4 N/A 15.1 3.9 N/A
(0.5) (1.6) (2.3) (0.4) (14.2) (7.0) (4.2) (10.3) (5.0; 5.9)
Guatemala 86 30 3.2 10.1 14.0 31.4 65.1 163.6 24.3 85.6 5.2 N/A 3.5 0.0 N/A
(0.5) (1.0) (1.2) (1.1) (10.7 (6.2) (3.3) (8.6) (4.9; 5.4)
India 729 41 2.9 10.4 14.0 29.1 71.8 169.6 24.9 90.3 5.4 39.0 17.6 3.6 0.90
(0.4) (1.3) (1.6) (1.3) (14.0) (6.2) (4.3) (12.0) (4.9; 5.9) (18 .0; 66.0) (0.40; 1.40)
Philippines 862 26 3.0 10.1 12.8 21.5 56.1 163.1 21.1 72.2 5.7 36.6 0.6 0.3 0.80
(0.4) (1.1) (1.8) (0.4) (9.2 ) (5.8) (3.0) (7.4) (5.4; 6.0) (27.0; 52.8) (0.60; 1.20)
South
Africa 195 15 3.1 11.6 16.0 15.6 54.4 166.7 19.5 69.6 5.1 33.8 3.1 0.0 0.60
(0.5) (1.8) (2.0) (0.3) (11.5) (8.2) (3.3) (8.6) (4.8; 5.5) (20.6; 54.5) (0.40; 1.00)
Total 3,459
Female
Brazil 1,49 6 14 3.2 10.8 15.3 22.7 60.8 160.9 23.5 74.9 5.2 N/A 9.9 3.3 N/A
(0.5) (1.6) (2.4) (0.4) (12.6) (6.2) (4.6) (10.4) (4.8; 5.7)
Guatemala 94 28 3.0 9.6 13.4 31.2 60.6 151.0 26.6 91.6 5.0 N/A 2.1 1.1 N/A
(0.5) (1.1) (1.5) (1.2) (11.0) (5.3) (4.5) (10.7) (4.7; 5.4)
India 511 39 2.8 9.8 13.3 29.2 59.4 154.9 24.6 79.7 5.3 32.4 11.0 3.1 0.7
(0.4) (1.3) (1.6) (1.4) (13.1) (5.6) (5.0) (12.2) (4.8; 5.8) (13 .8; 60.0) (0.30; 1.30)
Philippines 761 22 3.0 9.5 12.7 21.3 46.4 151.1 20.3 67.9 5.5 45.6 0.9 0.0 1.00
(0.4) (1.1) (1.5) (0.8) (8.1 ) (5.4) (3.2) (7.5) (5.2; 5.8) (33.0; 61.8) (0.70; 1.40)
South
Africa 190 13 3.0 11.2 15.5 15.6 54.8 158.5 21.8 70.7 5.0 54.2 2.6 0.0 1.00
(0.5) (1.5) (1.9) (0.2) (10.4) (5.9) (3.9) (8.6) (4.7; 5.2) (34.9; 79.4) (0.60; 1.48)
Total 3,052
Data are mean (SD) unless otherwise stated. N/A, not available. *Median (IQR) for logged variables. circ., circumference
care. diabetesjournals.org DIABETES CARE, VOLUME 35, JANUARY 2012 75
Norris an d Associates
Page 4
Weight in early life and adult
glucose concentrations and
prevalence of IFG/DM
Table 3 shows the cross-sectional analyses
for associations of early weights with adult
glucose concentrations and prevalence of
IFG/DM and includes P values for the tests
for site-sex heterogeneity. There was mini-
mal evidence of site-sex heterogeneity, and
consequently, all cohort data were pooled.
There was no evidence of heterogeneity be-
tween trial groups in Guatemala; therefore,
both groups were considered together.
Birth weight was inversely associated with
adult fasting glucose and with t he pre-
valence of IFG/DM. Fasting glucose de -
creased by 0.025 SD and the risk of
developing IFG/DM was reduced by 9%
per SD (;0.5 kg) increase in birth weight.
There were no associations between 24- or
48-month weight and glucose or IFG/DM
(Table 3).
Weights a t birth, 24 months, and 48
months were positively correla ted with
BMI (all P , 0.001) and waist circumfer-
ence (all P , 0.001). Adult BMI and waist
circumference were positively related to
glucose concentrations (both P , 0.001).
After adjusting for adult waist circumfer-
ence, the inverse associations of birth
weight with plasma glucose and IFG/DM
were strengthened (Table 3), and there
were inverse associations between 24-
and 48-month weight and these out-
comes. These ndings were little changed
after further adjusting for adult BMI and
height (data not shown).
CWG and adult glucose
concentrations and IFG/DM
The longitudinal analyses f ound minimal
evidence of site-s ex heterogeneity, a nd
the da ta were pooled. CWGs b etween
birth and 48 months were not associated
with fasting glucose or the prevalence
of IFG/DM (Table 4). CWG between
48 months and adulth ood was strongly
positively associated with adult glucose
and IFG/DM. In models further adjusting
for waist circumference, birth weight and
CWG b irth48 months were inversely
associated with fasting glucose and/or
IFG/DM; whereas CWG 2448 months
was not a ssoci ated. These ndings were
unaltered with adjustment for adult BMI
and height (data not shown).
Weight in early life and adult
insulin resistance
There was minimal evid ence of site-sex
heterogeneity in the associations betwee n
weight in early life and IR-HOMA (Table 3).
In pooled models, birth weight was un-
related to IR-HOM A, whereas 24- and
48-month weights were positively a s s o -
ciated with IR-HOMA. IR-HOMA was
positively associated with adult BMI and
waist circumference (both P , 0.001). Af-
ter adjus ting for adult waist circumfer-
ence, the associations of birth weight and
24- and 48-month weight with adult in-
sulin resistance were inversed (Table 3).
CWG and adult insulin resistance
There was one signicant P value for site-
sex heterogeneity (CWG f rom 48 month s
to adulthood; Table 4). This was ex-
plained by heterogeneity in the associa-
tion between CWG from 48 months to
adulthood and a dult w aist circumference.
This association was strongest in India,
where waist circumference rose by 10.2
(SE 0.3) cm in wom en and 9.9 (0.3) cm
in men per SD increase in CWG com-
pared with 4.9 (0.2) and 4.6 (0.2) cm in
the Philippines, and 5.8 (0.5) and 5.0
(0.5) cm in South Africa. The h eterogene-
ity was much reduced with adjustment
for adult waist circumference (Table 4).
In the pooled analysis, ther e were
positive asso ciations of CWG with IR-
HOMA in all three periods (Table 4). After
further adjusting for adult waist circum-
ference or for BMI and height (data not
shown), the CWG variables in early life
were not associated with insulin resistance.
SGA and the associations of weight
gain with outcomes
We examined interactions of SGA ( as a
binary variable) and birth weight (as a
continuous variable) with each CWG var-
iable. There were no signicant interactions
with CWG at any age (all
P values . 0.06).
This indicates that CWG in infancy and
childhood had similar associations with
adult outcomes across the range of birth
weights, whether or not the individual was
SGA at birth.
CONCLUSIONSdIn a pooled anal-
ysis of ve cohorts f rom LMICs, lower
birth weight was associated with higher
adult glucose conc entrations and an in-
creased risk of glucose intolerance. Weight
at 24 months and 48 months and CWG be-
tween birth and 48 months we re unrelated
to glucose concentrations and IFG/DM.
In contrast, CWG between 48 months and
adulthood was strongly and positively
related to both outcomes. Adult waist cir-
cumference was positively associated with
all early weights, as well as with adult glu-
cose concentration and IFG/DM. After
Table 3dCross-sectional analyses of the associations of weight at birth, and at ages 24 and 48 months with adult fasting glucose concentration, presence of IFG/DM, and insulin
resistance
Weight (SD-score)
Fasting glucose (SD score) IFG/DM IR-HOMA (SD score)
B (95% CI) PP
het
OR (95% CI) PP
het
B(95%CI) PP
het
Adjusted for sex, site and adult age
Birth 20.025 (20.049 to 0.000) 0.05 0.18 0.913 (0.8440.988) 0.02 0.50 20.022 (20.056 to 0.011) 0.19 0.97
Age 24 months 0.000 (20.024 to 0.024) 0.99 0.17 0.949 (0.8771.027) 0.19 0.03 0.086 (0.0520.119) ,0.001 0.07
Age 48 months 0.008 (20.016 to 0.032) 0.51 0.21 0.955 (0.8821.033) 0.25 0.14 0.140 (0.1070.174) ,0.001 0.15
Additionally adjusted for adult waist circumference
Birth 20.039 (20.063 to 20.015) 0.002 0.16 0.882 (0.8140.955) 0.002 0.27 20.081 (20.111 to 20.050) ,0.001 0.92
Age 24 months 20.032 (20.057 to 20.006) 0.01 0.23 0.876 (0.8070.952) 0.002 0.03 20.057 (20.089 to 20.025) ,0.001 0.84
Age 48 months 20.039 (20.065 to 20.012) 0.004 0.22 0.840 (0.7710.916) ,0.001 0.39 20.075 (20.110 to 20.040) ,0.001 0.91
B, regression coefcient; OR, odds ratio; P, P value for the association between early life weight and the glucose or insulin outcome; P
het
= P value for site-sex heterogeneity in these associations among the ve cohorts.
76 DIABETES CARE, VOLUME 35, JANUARY 2012 care.diabetesjournals.org
Infant and child growth and adult diabetes risk
Page 5
adjusting for adult waist circumference,
birth weight, infant weig ht, and infant
CWG were all inversely associated with
glucose and prevalence of IFG/DM. Birth
weight showed no association with insulin
resistance (estimated in three cohorts, Ta-
ble 3), whereas greater CWG in any period
was associated with hig her insulin resis-
tance. After adjusting for waist circumfer-
enc e, in sulin resistance, as with glucose
and IFG/DM, was inversely a ssociated
with birth weight (Table 3). SGA status
did not modify the associations between
CWG and outcomes.
Birth weight
The inverse associations of birth weight
with adult glucose concentrations and
risk of IFG/DM were consistent with pre-
vious studie s and with a recent systematic
review, mainly from high-income popu-
lations (1). Th e size o f the effect on I FG/
DM (9% [95% CI 116] reduction in risk
per SD incr ease in birth weight of ;500 g)
was similar to th at reported for DM alone
in the systematic review (25 [1930] re-
duction in risk per kilogram of birth
weight). Our ndings are compatible
with the hypothesis t hat environmental
factors that in uence fetal growth (e.g.,
maternal size and nutritional status)
have long-term effects on glucose homeo-
stasis. Insulin resistance later in life was
not related to birth weight unadjusted,
but was inversely associated with birth
weight after adjusting for adul t adiposity.
Infant and childhood weight and
weight gain
There were no associations of 24- or
48-month weight, or CWG through 48
months, with adult glucose or IFG/DM. In
contrast, CWG between 48 months and
adulthood was strongly associated with
these outcomes. This suggests that infancy
and early childhood may be an important
window of opportunity to promote weight
gain in LMIC populations to improve sur-
vival and adult human capital without
exacerbating adult DM risk. However, the
ndings for insulin resistance suggest that
this conclusion still needs to be treated
with caution. Insulin resistance was pos-
itively related to infant weight and CWG,
and it is possible that this will result in a
higher risk of diabetes at older ages. The
strong association between CWG from 48
months to adulthood is consistent with
other studies (9,14) and suggests that ac-
celerated weight gain or upward crossing
of weight percentiles during childhood
should be avoided.
Table 4dLongitudinal analyses of the associations of weight at birth and conditional weight gain from birth to 24 months, 24 to 48 months, and 48 months to adulthood, with adult
fasting glucose concentration, presence of IFG/DM, and insulin resistance
CWG (SD-score)
Fasting glucose (SD score) IFG/DM IR-HOMA (SD score)
B(95%CI) PP
het
OR (95% CI) PP
het
B(95%CI) PP
het
Adjusted for sex, site and adult age
Birth 20.025 (20.049 to 20.000) 0.05 0.18 0.913 (0.8440.988) 0.02 0.50 20.022 (20.056 to 0.011) 0.19 0.97
Birth24 months 0.010 (20.014 to 0.034) 0.41 0.26 0.982 (0.9081.062) 0.65 0.04 0.100 (0.0660.133) ,0.001 0.04
2448 months 0.014 (20. 010 to 0.038) 0.26 0.65 0.9 98 (0.9231.080) 0.97 0.29 0.110 (0.0770.144) ,0.001 0.16
48 monthsadulthood 0.096 (0.072 0.120) ,0.001 0 .25 1.321 (1.2201.430) ,0. 001 0.88 0.355 (0.3230.387) ,0.001 0.001
Additionally adjusted for adult waist circ umference
Birth 20.039 (20.063 to 20.015) 0.002 0.16 0.882 (0.8140.955) 0. 002 0.27 20.081 (20.111 to 20.050) ,0.001 0.92
Birth24 months 20.016 (20.041 to 0.009) 0.20 0.37 0.919 (0.8470.997) 0.04 0.07 20.026 (20.058 to 0.006) 0.11 0.89
2448 months 20.016 (20.041 to 0.010) 0.23 0.78 0.921 (0.8480.999) 0.05 0.60 20.030 (20.063 to 0.003) 0.07 0.99
All conditional weight variables were included in regr ession models si multaneously. B, regression coefcient; OR, odds ratio; P, P value for the association between CWG and the gluco se or insulin outcome; P
het
= P value
for site-sex heterogeneity in these associations among th e ve cohorts.
care. diabetesjournals.org DIABETES CARE, VOLUME 35, JANUARY 2012 77
Norris an d Associates
Page 6
Interpretation of the ndings after
adjustment for adult adiposity
Infant weight, 48-month weight, and
CWGs were correlated with adult weight
and with adult BMI and waist circumfer-
ence. Thus, weight at these ages, or some
component(s) of it, makes a contribution
to adult adiposity and a strong risk factor
for diabetes. Without adjustment for
adult waist circumference, we foun d no
association of 24- and 48-month wei ght/
CWG with adult glucose and IFG/DM.
After adjusting for adult waist circumfer-
ence, we found inverse associations of
birth and/or 24-month weight, and CWG
between birth and 24 months, with all
three outcomes. Our interpretatio n is that
for any g iven adul t waist circumference,
higher birth weight and/or infant weight
and weight gain are associated with a
lower adult insulin resistance and DM
risk. This suggests that there is a compo-
nent of early weight/weight gain that is
not associated with larger adult waist cir-
cumference and that may be protective
against later disease. An example may be
lean tissue or muscle mass. Our data sug-
gest that the positive associations of 24- and
48-month weight, and CWG at all ages,
with adult insulin resistance are driven by
the component(s) associated with larger
adult waist circumference.
Strengths and limitations of the
study
The major strength of this study was the
prospective serial measurements of
weight from a large number of individuals
in LMICs. Our choice of time points was
limited due to differing ages of follow-up
in the ve sites; measurements earlier in
infancy and later in childhood would have
been valuable. Interpolation was required in
two cohorts to estimate 48-month weight;
however, weight gain between the end of
infancy and the onset of puberty tends to be
linear.
Another limitation was loss to follow-
up, especially in the older (India and
Guatemala) cohorts. However, comparison
of the analysis sample with the original full
cohorts showed that their early weights
were similar. Bias would be introduced only
if the associations between early size and
glucose tolerance differ between those who
were and were not included in the analysis.
Additional limitations were heteroge-
neity in the age o f the participants among
the ve cohorts and the methods used for
measuring birth wei ght, gestational age,
and plasma glucose co ncen trations. Even
though three cohorts used one approach
and two cohorts used another to measure
birth weight, both methods are acc ept-
able means of obtaining birth weight.
Only one of the ve cohorts determined
gestational age in an alternative way to
the other four cohorts, and again used a
recognized method.
We had only single plasma glucose
concentrations (no gl ucose tolerance test
data). The site in Brazil collected non-
fasting blood glucose but validated the
equation to correct these values to a fast-
ing state (18). The differences in glucose
measurement s between whole blood and
plasma are well dened (22) . The differ-
ences between labo ratory and glucometer
measurements are less well known, but
laboratory and glucometer values in ve-
nous and capillary samples, and in whole
blood and plasma, have been compared
extensively in the literature, including
direct comparisons of the methods
used in our study (20). Despite these dif-
ferences in methodologies we are struck by
the consistency of results across the ve
sites. This speaks to the robustness of our
ndings.
In conclusion, lower birth weight is a
risk f actor for glucose intolerance and has
important implications for LMICs, where
poor birth outcomes are common.
Grea ter CWG between 48 months and
adolescence/adulthoo d (1532 years) is
also a risk factor for glucose intolerance,
providing more evidence that upward
crossing of weight percentiles after 48
months should be avoided in LMICs.
Our analysis showed no increased risk
of IFG/DM associated with greater infant
and early childhood weight gain, which
suggests that it may be possible to pro-
mote weight gain at this stage of life t o
accrue benets for survival, growth falter-
ing, and human capital, without i ncreas-
ing adult diabetes risk. However, the
cohorts are still young, and the associations
of above-average weight gain in infancy
with increased adult waist circumference
and insulin resistance may predict a risk
of diabetes in the future.
AcknowledgmentsdCOHORTS is supported
by Wellcome Trust (U.K.) and the Bill and
Melinda Gates Foundation. Funding for
the individual cohorts was as follows: INTCS
(Guatemala)dU.S. National Institutes of Health
and U.S. National Science Foundation; Pelotas
Birth Cohort (Brazil)dWellcome Trust; New
Delhi Birth Cohort Study (India)dIndian Coun-
cil of Medical Research, U.S. National Center for
Health Statistics, Medical Research Council
(U.K.), and British Heart Foundation; BTT
(South Africa)dWellcome Trust, Human Sci-
ences Research Council, South African Medical
Research Council, South-African Netherlands
Programme on Alternative Development,
Anglo American Chairmans Fund, and Univer-
sity of the Witwatersrand; and CLHNS (the
Philippines)dU.S. National Institutes of Health.
No potential c onicts of interest relevant to
this article were repor ted.
The COHORT S group designed the re-
search. S.A.N. and C.H.D.F. conducted the
research, analyzed data, wrote the manuscript,
read a nd approved the nal manuscript, and
had primary responsibility for nal content.
C.O. conducted the research, analyzed data ,
read a nd approved the nal manuscript, and
had primary responsibility for nal content.
D.G., C.W.K., L.R., N.R.L., M.R.-Z., and
L.M.R. conducted the research and read and
approved the nal manuscript. A.D.S. con-
ducted the research, wrote the manuscr ipt,
and read and approved the nal manuscript.
N.T . conducted the research and read and
approved the nal manuscript.
The authors thank the following colleagues
from each site. Pelotas Birth Cohort (Brazil):
Rosangela Lima; CLHNS (the Philippines):
Sororro Gultiano, Josephine A vila, Lorna
Perez, and Thomas McDade; New Delhi Birth
Cohort Study (India): Shanti Ghosh, I.M.
Moriyama, Vinod Kapani, Rajeshwari Verma,
Bhaskar Singh, Arti Mishra, K.D. Gupta,
K. Belwal, Dileep Gupta, and Shikha Sinha;
INTCS (Guatemala): Rafa el Flores, Usha
Ramakrishnan, Kathryn Yount, Ruben
Grajeda, Paul Melgar, Humberto Mendez,
Luis Fernando Ramirez, Jere Behrman, John
Hoddinott, Agnes Quisumbing, Alexis Murphy,
and John Maluccio; BTT (South Africa): John
Pettifor and Noel Cameron.
APPENDIXdThe COHOR TS group
members are as follows: C esar G. Victora,
Pedro C. Hallal, Fernando C. Barros,
and Bernardo L. Horta (Universidade
Federal de Pelotas, Brazil); Reynaldo
Martorell (Hubert Department of Global
Health, Rollins School of Public Health,
Emory University, Atlanta, Georgia);
Santosh K. Bhargava (Sunderla l Jain
Hospital, New Delhi, India); Harshpal
Singh Sachdev (Sitaram Bhartia Institute of
Science and Research, New Delhi, India);
Linda Adair (University of North Carolina
at Chapel Hill, Chapel Hill, North Caro-
lina); Judith B orja (Ofce of Population
Studies, University of San Carlos, Cebu
City, Philippines); Darren Dahly (Uni-
versity of Leeds, Leeds, U.K.); and Mathew
Mainwaring (Developmental Pathways
for Health Research Unit, Department of
Paediatrics, Faculty of Health Sciences, Uni-
versity of the Witwatersrand, Johannesburg,
South Africa).
78 DIABETES CARE, VOLUME 35, JANUARY 2012 care.diabetesjournals.org
Infant and child growth and adult diabetes risk
Page 7
References
1. WhincupPH,KayeSJ,OwenCG,etal.Birth
weight and risk of type 2 diabetes: a sys-
tematic review. JAMA 2008;300:28862897
2. Nobili V, Al isi A, Panera N, Agostoni C.
Low birth weight and catch-up-growth as-
sociated with metabolic syndrome: a ten
year systematic review. Pediatr Endocrinol
Rev 2008;6:241247
3. International Diabetes Federation. IDF
Diabetes Atlas. 4th e d. Brussels, Belgium,
IDF, 2009
4. Gluckman PD, Hanson MA. Developmental
origins of disease paradigm: a mechanistic
and evolutionary perspective. Pediatr Res
2004;56:311317
5. Victora CG, de Onis M, Hallal PC,
Blössner M, Shrimpton R. Worldwide
timing of growth faltering: revisiting im-
plications f or interventions. Pedi atrics
2010;125:e473e480
6. Martorell R, Horta BL, Adair LS, et al.;
Consortium on Health Orientated Re-
s earch in Transitional Societies Group.
Weight gain in the rst two years of life is
an important predictor of schooling out-
comes in pooled analyses from ve birth
cohorts from low- and middle-income
countries. J Nutr 2010;140:348354
7. Hoddinott J, Maluccio JA, Behrman JR,
Flores R, Martor ell R. Effect of a nutrition
intervention during early childhood on
economic productivity in Guatemala n
adults. Lancet 2008;371:411416
8. Hales CN, Barker DJ, Clark PM, et al. Fetal
and infa nt growth a nd impaired glucose
tole rance at age 64. BMJ 1991;303:1019
1022
9. Eriksson JG, Forsé n T, Tuomilehto J,
Osmond C, Barker DJ. Early a diposity
rebound in childhood and risk of Type 2
diabetes in adult life. Diabetologia 2003;
46:190194
10. Baird J, Fi sher D, Lucas P, Kleijnen J,
Roberts H, Law C. Being big or grow ing
fast: syste matic review of size and growth
in infanc y and later obesity. BMJ 2005;
331:929
11. Stettler N, Iotova V. Early growth patterns
and long-term obesity risk . Curr Opin
Clin Nutr Metab Care 2010;13:294
299
12. Victora CG, Barros FC. Cohort prole: the
1982 Pelotas (Brazil) birth cohort study.
Int J Epidemiol 2006;35:237242
13. Stein AD, Melgar P, Ho ddinott J, Martorell
R. Co hort Prole: the Insti tute of Nutrition
of Central America and Panama (INCAP)
Nutrition Trial Cohort Study. Int J Epi-
demiol 2008;37:716720
14. Bhargava SK, Sachdev HS, Fall CH, et al.
Relation of serial changes in childhood
body-mass index to impaired glucose tol-
erance in young adulthood. N Engl J Med
2004;350:865875
15. Adair LS. Size at birth and growth trajec-
tories to young adulthood. Am J Hum Biol
2007;19:327337
16. Richte r L, Norris SA, Pettifor J, Yach D,
Cameron N. Cohort Prole: Mandelas
children: the 1990 Birth to Twenty study
in South Africa. Int J Epide miol 2007;36:
504
511
17. Ballard JL, Novak KK, Driver M. A sim-
plied score for assessment of fetal mat-
uration of newly born infants. J Pediatr
1979;95:769774
18. Horta BL, Gigante DP, Victora CG, Barros
FC, Oliveira I, Silveira V. [Early determi-
nants of random blo od glucose among
adults of the 1982 birth cohort, Pe lotas,
Southern Brazil]. Rev Saude Publica 2008;
42(Suppl. 2):93100
19. Kumar G, Sng BL, Kumar S. Correlation of
capillary a nd venous blood glucometr y
with laboratory determination. Prehosp
Emerg Care 2004;8:378383
20. Petersen JR, Graves DF, Tacker DH,
Okorodudu AO, Mohammad AA, Cardenas
VJ Jr. Comparison of POCT and central
laboratory blood glucos e resul ts using
arterial, capillary, and venous samples from
MICU patients on a tight glycemic protocol.
Clin Chim Acta 2008;396:1013
21. World Health Organization. Deniti on,
Diagnosis and Classicatio n of Diabetes
Mellitus and its Complications. Report of a
WHO Consultatio n. Part 1: Diagnosis and
Classication of Diabetes Mellitus.Geneva,
Switzerland, WH O, 1999
22. Matthe ws DR, Hosker JP, Rudenski AS,
Naylor BA, Treacher DF, Turner RC.
Homeostasis model assessment: insuli n
resistance and beta-cell function from
fasting plasma glucose and insulin con-
centrations in man. Diabetologia 1985;28:
412419
23. Williams RL, Creasy RK, Cunningham GC,
Hawes WE, Norris FD, Tashiro M. Fetal
growth and perinatal viability in California.
Obstet Gynecol 1982;59:624632
24. Keijzer-Veen MG, Euser AM, van Montfoort
N, Dekker FW, Vandenbroucke JP, Van
Houwelingen HC. A regression model with
unexplained residuals was preferred in the
analysis of the fetal origins of a dult dis-
eases hypothesis. J Clin Epidemiol 2005;
58:13201324
25. Adair LS, Martorell R, Stein AD, et al.;
COHORTS group. Size at birth,weight gain
in infancy and childhood, and adult blood
pressure in 5 low- and middle-income-
country cohorts: when does weight gain
matter? Am J Clin Nutr 2009;89:1383
1392
care. diabetesjournals.org DIABETES CARE, VOLUME 35, JANUARY 2012 79
Norris an d Associates
Page 8
  • Source
    • "As noted by Sunguya et al. (2014), Bchildren born with low weight or those who succumb to undernutrition in their early childhood have a high risk of early adulthood obesity and NCDs, including diabetes and heart diseases.^ Hence the importance of taking an holistic view of nutrition, particularly in countries undergoing a rapid economic, dietary and nutrition transition (Uauy et al. 2011; Norris et al. 2012). However, there are three main problems relating to this particular metric of (over)nutrition: a) very few developing countries systematically monitor trends in child overweight, b) there is as yet no empirically-documented intervention that has prevented an increase in overweight and obesity at a population-wide level anywhere in the world (which also represents an important opportunity for innovation and better documentation of potentially impactful practices), and c) there is a strong statistical inverse association over the long run between poverty reduction per capita in developing countries and the rise of child overweight – that is, while a policy of support for agriculture in poorer economies, for example, may be associated with a decline in stunting over time, that same policy may be associated with an increase in child obesity (Webb and Block 2012). "
    [Show abstract] [Hide abstract] ABSTRACT: Attention to nutrition continues to grow. The recent surge in interest has included widening agreement on two major issues: first, nutrition goals cannot be achieved through targeted actions alone; nutrition-sensitive interventions are needed as well. Second, the multiple actions required to address all forms of malnutrition through the lifecycle cannot be proxied by a single target or metric. Although the Millennium Development Goals included one concrete measure of nutrition (children underweight), the post-2015 Sustainable Development Goals will include multiple measures that better inform a diversity of policy and programming actions. This suggests a need for improved understanding of how multiple forms of malnutrition are linked, how public investments may affect one form of malnutrition but possibly not others, and how best to measure progress on multiple nutrition fronts, including through nutrition-sensitive actions, such as investments in agriculture. This paper proposes a composite index that highlights the state of nutrition across six separate nutrition goals endorsed by the World Health Assembly in 2012, allowing for ranking (comparison among countries) and monitoring of change (within countries) over time. Establishing an index that captures gains or losses in nutrition across all six goals simultaneously highlights the complexity of nutrition problems and required solutions. Such an index can be used to track progress towards goals set for 2025, but also support dialogue on the individual index components and how investments should be prioritized for maximum impact.
    Full-text · Article · Apr 2015 · Food Security
  • Source
    • "Only less than 10% of genetically susceptible individuals may present subsequently with T1DM. This validates the noticeable role of environmental factors like viruses (Bell et al., 1996; Froguel and Velho, 1999), bacteria (Herold et al., 2011), diet (Hu, 2011; Katz, 2014), and weight gain (Colditz et al., 1995; Norris et al., 2012) in the onset and progression of the disease. p0070 Type 2 diabetes mellitus (T2DM) which takes into account about 90À95% of all diabetics, is relatively different from T1DM in terms of etiology and the time of onset of the clinical phenotype, where T1DM appears mostly in infants and juveniles while T2DM generally occurs among middle-aged and older adults. "
    Full-text · Article · Jan 2015
  • Source
    • "Only less than 10% of genetically susceptible individuals may present subsequently with T1DM. This validates the noticeable role of environmental factors like viruses (Bell et al., 1996; Froguel and Velho, 1999), bacteria (Herold et al., 2011), diet (Hu, 2011; Katz, 2014), and weight gain (Colditz et al., 1995; Norris et al., 2012) in the onset and progression of the disease. p0070 Type 2 diabetes mellitus (T2DM) which takes into account about 90À95% of all diabetics, is relatively different from T1DM in terms of etiology and the time of onset of the clinical phenotype, where T1DM appears mostly in infants and juveniles while T2DM generally occurs among middle-aged and older adults. "
    Full-text · Chapter · Jan 2015
Show more