Received: 14 May 2018 Revised: 10 June 2019 Accepted: 3 July 2019

| |
DOI: 10.1111/apa.14929

REGULAR ARTICLE

Stunting, recovery from stunting and puberty development in

the MINIMat cohort, Bangladesh

Pernilla Svefors1,2 | Jesmin Pervin3 | Ashraful Islam Khan3 | Anisur Rahman3 | Eva‐
Charlotte Ekström1 | Shams El Arifeen3 | Katarina Ekholm Selling1 | Lars‐Åke Persson1,4

1
International Maternal and Child Health, Department of Women’s and Children’s Interventions in Matlab trial. The birth cohort was followed from
Health, Uppsala University, Uppsala, Sweden 2Center for Epidemiology and
birth to puberty 2001‐2017. Pubertal development according to
Community Medicine, Stockholm, Sweden
3
International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b),
Tanner was self‐assessed. Age at menarche was determined and
Dhaka, Bangladesh in boys, consecutive height measurements were used to
4
Department of Disease Control, London School of Hygiene and Tropical Medicine, ascertain whether pubertal growth spurt had started. The
London, UK
exposures and outcomes were modelled by Cox’s proportional
Correspondence hazards analyses and logistic regression.
Pernilla Svefors, Department of Women’s and Children’s health, Uppsala
University, Uppsala, Sweden. Results: There was no difference in age at menarche between
Email: [email protected] girls that were small or appropriate for gestational age at birth.
Funding information Boys born small for gestational age entered their pubertal
The MINIMat research study was funded by ICDDR,B, United Nations Children's
growth spurt later than those with appropriate weight. Children
Fund (UNICEF), Swedish International Development Cooperation Agency (Sida),
UK Medical Research Council, Swedish Research Council, Department for who were stunted had later pubertal development, age at
International Development (DFID), Japan Society for the Promotion of Science,
menarche and onset of growth spurt than non‐stunted children.
Child Health and Nutrition Research Initiative (CHNRI), Uppsala University and
United States Agency for International Development (USAID). The funders had no Children who recovered from infant or early child‐
role in study design, data collection and analysis, decision to publish or preparation
of the manuscript.
hood stunting had similar pubertal development as non‐stunted
Abstract children. Conclusion: Infant and childhood stunting was
Aim: This paper aimed to analyse the association between small associated with a later pubertal devel‐ opment. Recovery from
for size at birth, stunting, recovery from stunting and pubertal stunting was not associated with earlier puberty in compari‐ son
development in a rural Bangladeshi cohort. with non‐stunted children.

Methods: The participants were 994 girls and 987 boys whose
KEYWORDS
mothers participated in the Maternal and Infant Nutrition
stunting, catch‐up growth, developmental origin of health and disease,
menarche, puberty

Abbreviations: AGA, appropriate for gestational age; BMI, body mass index; HAZ, height‐for‐age z‐scores; icddr,b, International centre for diarrheal disease research, Bangladesh; SD,
standard deviation; SGA, small for gestational age; WHO, World Health Organisation.

This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial License, which permits use, distribution and reproduction in any
medium, provided the original work is properly cited and is not used for commercial purposes.
© 2019 The Authors. Acta Paediatrica published by John Wiley & Sons Ltd on behalf of Foundation Acta Paediatrica

wileyonlinelibrary.com/journal/apa
122 | Acta Paediatrica. 2020;109:122–133.
SVEFORS et al. 123
|
1 | INTRODUCTION
Key notes
Stunting, defined as height measurements below minus two stand‐ ard • No unfavourable association between recovery from
deviations of the World Health Organization (WHO) growth reference stunting in infancy or childhood and early puberty was
median, reflects chronic exposure to a deficient envi‐ ronment and found.
undernutrition. Stunting in early life is linked to a wide range of
adverse consequences for the individual and the formation of human
capital.1 Stunted children have increased risks of morbidity and
mortality and are more likely to have a lower educational achievement The participants in this study were children born to women

and to become short adults. Also, grow‐ ing evidence supports the participating in the Maternal and Infant Nutrition Interventions in
hypothesis that undernutrition during early development might Matlab (MINIMat, isrctn.org identifier: ISRCTN16581394). MINIMat
increase chronic disease risks later in life. Recovery from stunting may was a factorial randomised trial, which provided early prenatal food
prevent several of these negative consequences for short‐ and long‐ and micronutrient supplementation to pregnant women. The pri‐
1
term health and is hence desir‐ able. Paradoxically, there is also mary outcomes were maternal haemoglobin, size at birth and infant
evidence that catch‐up growth, following early life undernutrition, mortality. There were several short‐ and long‐term secondary out‐
might increase the risk of adult non‐communicable diseases. comes that have motivated a follow‐up of the MINIMat birth cohort.
The onset of puberty is an intermediate factor that is associ‐ ated From November 2001 to October 2003 all women within in the icd‐
with early life nutrition as well as future health outcomes. Nutritional dr,b service area who reported to a community health worker that
status is a well‐known determinant of pubertal devel‐ opment. their menstruation was overdue or that they were pregnant were
Younger age at menarche has been reported to be a risk indicator of offered a pregnancy test and invited to participate in the MINIMat
adult chronic diseases in low‐, middle‐ and high‐in‐ come countries, 2,3 trial. In total, 4436 pregnant women participated, resulting in 3625
4 live births. The children whose mothers participated in the initial
including Bangladesh. The mechanisms behind these associations
prenatal intervention study have so far been followed up from the
seem, however, to differentiate between these contexts. In low‐
resource settings with widespread undernutri‐ tion pubertal time of their birth until the typical age of puberty onset. The data
5 collection for the puberty assessment was carried out at two time
development and menarche usually take place at an older age.
points with a 6‐month interval. The first round of data collection was
Epidemiological and adoption studies have indicated that intrauterine
carried out from July 2016 to February 2017 and the second round
growth restriction and early life undernutrition that are followed by
from January to July 2017.
rapid post‐natal weight gain and subsequent height gain may result in
6,7 At enrolment, trained community health workers collected
early puberty and younger age at men‐ arche. In light of this recent
information on household socioeconomic characteristics, paren‐ tal
evidence of adverse outcomes of so‐called catch‐up growth, concerns
education and pregnancy history in the respondents’ homes.
have been raised on the po‐ tential long‐term effects of recovery from
stunting. Further, the association is poorly understood between Anthropometry and puberty assessments were done at clinic visits at

recovery from early growth restriction in children who remain in a local sub‐centres run by the icddr,b in the Matlab area.

low‐resource set‐ ting and their pubertal development. This study The children’s weight, length or height were measured at birth,
aimed to anal‐ yse the association between being born small for every month up to 1 year, every 3 months up to 24 months, during
gestational age (SGA), infancy and childhood stunting, recovery from follow‐up surveys at 4.5 and 10 years, and two times at puberty
stunting, and pubertal development, assessed by self‐reported Tanner follow‐up in the age interval 12‐15 years. Most birth anthropome‐
stages, age at menarche and start of puberty growth spurt, in a rural try was measured within 72 hours after birth. Measurements taken up
Bangladeshi cohort. to 30 days after birth were adjusted using a standard deviation score
transformation, based on the assumption that infants remain in the
same relative position in the anthropometric distribution during this
2 | METHODS period. The recumbent length at birth and during infancy (until 1.5
years) was measured with a locally manufactured, collaps‐
The study area, Matlab, is a rural sub‐district 57 km southeast of the ible length board with a precision of 0.1 cm. Maternal weight and
capital Dhaka where relatively widespread child and ma‐ ternal height were measured at around 8 weeks of pregnancy and at fol‐ low‐
undernutrition still prevail. The International Center for Diarrhoeal ups. Uniscale beam scales or SECA electronic scales (UNICEF,
Disease Research, Bangladesh (icddr,b) runs a Health and
Demographic Surveillance System in the area. Community health
workers visit all households and collect demographic and health in‐
formation on a monthly basis.

• We examined the consequences of stunting and recov‐ ery

from stunting for pubertal development in children in
rural Bangladesh.
• Girls that were stunted in infancy and childhood had a
later menarche and boys started their growth spurt later,
as compared to non‐stunted children.
 124 SVEFORS et al.
|

USA; SECA Gmbh & Co), with a precision of 0.01 kg were used when the velocity from 4.5 to 10 years between the two follow‐ups during
measuring children’s weight. Height was measured to the nearest 0.1 puberty were classified as having started the growth spurt. Boys with
cm, using a freestanding stadiometer. Height‐for‐age z‐scores (HAZ) a height velocity not different from the pre‐pubertal velocity and
were calculated from the measured length and height data using the having genital development less than stage four pubertal. Boys with
programme WHOAnthro, from birth to 4.5 years and AnthroPlus at no height increase or a height velocity below the pre ‐pubertal speed
puberty using WHO growth references for 0‐5 and 5‐19 years, and who had reached genital development stage four to five were
respectively. Being stunted was defined as having a HAZ below minus considered to already have passed the growth spurt. Lastly, these
two, and not being stunted was defined as having a HAZ at or above three groups were re‐classified into two groups: not having started
minus two, respectively. The SGA variable is based on weight for the growth spurt or, having started or already passed the growth
gestational age.8 HAZ at birth was not adjusted for gestational age. spurt.
During puberty follow‐up, anthropometry and puberty assess‐ ment The exposure variables were SGA or appropriate for gestational
was done at sub‐centres run by the icddr,b. This was done separately age (AGA), stunting, and recovery from stunting at and between the
for girls and boys by teams of female and male nurses and doctors. following ages: birth, 12, 24 and 54 months. Children with a height‐
Guided by pictures, participants self‐assessed pubertal development; for‐age Z‐score (HAZ) below minus two SD were classified as stunted.
breast, genital and pubic hair development as appro‐ priate, according Children who had been stunted at birth, 12, 24 and 54 months but
to Tanner.9 The female staff interviewed the girls about whether they were not stunted, that is had a HAZ at or above minus two SD, at the
had had their first menstruation and the date at menarche was noted. following age were categorised as recovered from stunting.
When needed, the interview was supported by the mother and by use Differences between population characteristics of included, ex‐
of a local events calendar. Refresher train‐ ing on data collection cluded and non‐participating mothers and children were tested by the
methods including standardization of anthro‐ pometric measurements chi‐square test. The chi‐square test was also used to test the
was conducted periodically for the staff. differences in distribution of background characteristics and the
Written and oral informed consent was obtained from all par‐ adolescents’ pubertal stage and age. The Cox’ proportional hazards
ticipating women in the MINIMat trial and from the parents of the regression model was used to examine the relationship between SGA,
participating children. The Ethical Review Committee at icddr,b in stunting at birth, at 12, 24 and 54 months of age and recov‐ ery from
Bangladesh, approved the original trial, and the Ethical Review stunting on age at menarche. The start of study time was birth. Girs
Committee at icddr,b in Bangladesh and the Regional Ethical Review who had not had the event (menarche) were censored at the time of
Board at Uppsala University, Sweden, approved each follow‐up. last examination. Logistic regression was used to model the same
exposure variables and the binary outcome of whether the boy was
pre‐pubertal or pubertal, that is had started or already completed the
2.1 | Statistical analysis growth spurt. Both crude and adjusted hazards ra‐ tios and odds ratios

An initial dataset was created with children who had height meas‐ were calculated. In the crude analyses, each predictor was included

urements at birth and at 12, 24 and 54 months of age (n = 1930). For one‐by‐one. In the adjusted model, the variables of maternal

descriptive purposes, self‐reported pubertal stages according to education, maternal body mass index (BMI) at recruitment, around
gestational week 8, socioeconomic status of the household
Tanner from the second follow‐up was reported. For the exposure
represented by tertiles of asset scores, parity and preterm birth were
analyses, the outcomes of interest were age at menarche for girls and
included. All analyses were performed using the R statistical software,
onset of the puberty growth spurt for boys, as the validity of
version 3.2.410 and the packages ‘Epi’11and ‘survival’12.
self‐reported pubertal development is typically low. For three of the
girls, the year but not the month of menarche was known and month
June was imputed. In cases (n = 362) when the month but not the date 3 | RESULTS
was known, the value 15 was imputed. For girls who were not
participating in the second pubertal data collec‐ tion (n = 43), the
3.1 | General characteristics
information on menarche from the first round was used. Stage of
pubertal development from the second data collection was used. For From 2002 to 2004, there were 3625 children born to the 4436
descriptive purposes, pubertal stages with less than 15 individuals women enrolled in the MINIMat trial (Figure 1). A total of 845 women
were merged with the next pubertal stage. Median age at menarche were lost during follow‐up before delivery, mainly due to
was calculated with the Kaplan Meier method. Boys with height outmigration, foetal loss or withdrawal of consent. Puberty data were
assessments at 4.5 and 10 years, and two height measurements in the collected from 2307 adolescents. Reasons for losses from follow‐up
puberty follow‐up were included in an assess‐ ment regarding whether after birth up to puberty included outmigration, refusal to participate
they were pre‐pubertal, or had started or completed their growth and death. Out of the children assessed for pubertal development,
spurt. The height velocities between 4.5 and 10 years and between 377 were excluded due to missing height measure‐
the two pubertal measurements were calcu‐ lated (cm/y) and ments at birth, 12, 24 or 54 months resulting in a final sample of 1930
compared. Boys with a height velocity higher than adolescents. A total of 474 boys had height measurements at
SVEFORS et al. 125
|
 TABLE 1 Baseline characteristics of mothers at 8wk of gestation
and households participating in the MINIMat trial, Bangladesh

Characteristics n/n (%)

Maternal age
<20 274/1930 (14.2%)
20‐29 567/1930 (29.4%)
≥30 1089/1930 (56.4%)

Maternal BMI
<18.5 564/1922 (29.3%)
≥18.5 1358/1922 (70.7%)

Maternal education
No education 652/1930 (33.8%)
Formal education 1278/1930 (66.2%)
Parityat enrolment
0 524/1930 (27.2%)
1 534/1930 (27.7%)
2 or more 871/1930 (45.2%)
Perceived income‐expenditure status
Surplus 490/1929 (25.4%)
Breakeven 1050/1929 (54.4%)
Occasional deficit 325/1929 (16.8%)
Constant deficit 57/1929 (3.0%)
Don't know 71/1929 (0.4%)

The characteristics of the households, mothers and children at

birth are presented in Table 1. The participating mothers were short
and slender at recruitment with an average height and weight of
151 cm (SD 5.3 cm) and 48 kg (SD 4.7 kg), respectively, correspond‐
FIGURE 1 Flow chart of children participating in the MINIMat them had more than two children from before.
trial from birth to puberty

3.2 | Birth characteristics and linear growth

4.5 and 10 years and two times, 6 months apart, during the puber‐ tal
follow‐up. The major limiting factor was the 10‐year follow‐up, which At birth, HAZ was low (mean −0.94), 16% were stunted,
included only one of the two‐calendar‐year MINIMat cohort. There
and more than half of the children (62%) were born
were 54 boys with some of these assessments with obvious errors,
SGA. HAZ declined rap‐ idly up to 24 months of age,
resulting in 420 boys included in the analysis of pre‐pubertal or
resulting in a mean HAZ at 12 and 24 months of −1.6
pubertal growth spurt status.
The randomized prenatal interventions were not associated with and −2.0 and a stunting prevalence of 35% and 50%,

any of the determinants of growth included in the analysis (data not respectively (Table 2, Figure 2). Few children
shown). Among mothers of the children not participating in the recovered from stunting during infancy (Figure 2).
puberty follow‐up, there were a slightly higher proportion of young From 24 to 54 months, HAZ in‐ creased (mean HAZ at 54
(<20 years), first‐time mothers, with more than 6 years of education,
months ‐1.6) and the prevalence of stunt‐ ing
and mothers from households belonging to the low‐
decreased to 34% at 54 months of age. At 12 months of
est socioeconomic tertile (data not shown). Out of these, only so‐
age, boys had a higher prevalence of stunting than
cioeconomic status was associated with the outcomes (data not
shown). At birth, non‐participating children were stunted more girls. Later in childhood, the pattern was reversed;
frequently when compared to participating children (19.2% and fewer girls recovered from stunting from 24 to 54
16.3%, respectively, P = <0.05). Other background characteris‐ tics did months and at 54 months girls had a higher prevalence
not differ between participating and non‐participating children.
of stunting than boys (girls 34.8% and boys 29.1%, P <
ing to an average BMI of 20.1 (SD 2.6). Average weight gain from
0.01). The mean HAZ change for children who recovered from
pregnancy week 8 to 30 was 5.4 kg (SD 2.5 kg).
stunting was 1.1 HAZ 0‐12 months, 0.62 HAZ 12‐24 months, and 0.85
Approximately one‐third of the women were illiterate, and half of
HAZ in the age in‐ terval 24‐54 months.
 126 SVEFORS et al.
|

3.3 | Pubertal development dictor variables (data not shown). No differences in puberty char‐
acteristics or age at menarche were found when comparing the
The girls were between 12.4 and 14.5 years of age at the first pu‐ berty
participants and the 377 excluded children who were missing some of
follow‐up (mean age 13.3, SD 0.49), and between 12.9 and 15.0 (mean
the previous anthropometry measurements.
age 13.9, SD 0.49) years at the second round. The boys were between
Table 3 illustrates puberty development at the second round of
12.3 and 14.4 years of age at the first puberty follow‐ up (mean age
data collection, baseline characteristics and stunting status for girls,
13.4, SD 0.47), and between 12.8 and 15.0 (mean age 13.9, SD 0.47)
and Table 4 for boys. All girls had reached Tanner stage two or higher
years at the second round. The age of the participating
in breast development, with a majority of girls being at stage three
(68.5%). For pubic hair development, only two girls were still pre‐
TABLE 2 Anthropometric characteristics of girls and boys
pubertal (0.2%, stage one), most girls were at stage two (42.5%) or
participating in the MINIMat trial from birth to puberty follow‐up
three (49.8%) and none had reached stage five in breast or pubic hair
Girls (n = 987) Boys (n = 987) development. Most boys were at genital stage three (27.1%) or four

Mean (SD) Mean (SD) (72.3%), and 0.6% had reached stage five. Only 0.7% of the boys had
not developed any pubic hair, no boys were at stage two, 47.7% had
Birth weight, g 2654 (374) 2746 (413) Birth length, cm 47.5 (2.04)
reached stage three, 51.4% stage four, and only 0.2% stage five.
48.1 (2.2) HAZa at birth −0.92 (1.1) −0.95 (1.2) HAZ at
Adolescents, who had reached a later stage of puberty, were charac‐
12 mo −1.60 (0.96) −1.68 (1.0) terised by more frequently having mothers with higher education, a
HAZ at 24 mo −2.03 (0.95) −2.03 (1.03) HAZ at 54 mo higher BMI, and belonging to a higher socioeconomic group (Tables 3
b and 4).
−1.67 (0.86) −1.47 (0.91) Height at 10 y (cm) 129.4
The median age at menarche was 13.0 years. A total of 61.8% of
(6.9) 129.4 (5.9) Height at 12 y (cm) 147 (5.92) 146 (8.29) Height at
the girls had reached menarche in the first round and 76.2% in the
13 y (cm) 149 (5.57) 150 (8.48) Height at 14 y (cm) 151 (5.90) 155
second round of data collection. Girls with later menarche were
(8.48) BMI at 12 y 16.91 (2.71) 16.03 (2.41) BMI at 13 y 17.72 (3.07)
characterised by more frequently being stunted and having mothers
16.40 (2.63) BMI at 14 y 18.32 (2.76) 16.96 (2.40)
from a lower socioeconomic group, with less education, and a low
a
Height‐for‐age Z‐score. BMI. The mean recall time, that is time between menarche and data
b
Only half of the cohort was followed at 10 y, n girls464, boys = 466.
collection, was 0.9 years, range 0‐3.6 years.
There was no difference in pubertal development or age at men‐ arche
between SGA and AGA girls (Tables 3 and 5). Boys who were born
SGA, however, came into their pubertal growth spurt later than

FIGURE 2 Stunting prevalence and

recovery from stunting in girls and boys
from birth to adolescence in the MINIMat
adolescents did not differ across the baseline characteristics or pre‐ cohort

SVEFORS et al. 127

|
 )

MI

NI

gn

it

it

n
t

i
P

43

34

0.

0<

3.

1(

3/

5.

4/

41

1.

(
)

7.

8
(

3/

62

5.

4/

74

9.

(
1

0.

0<

0.

1(

3/

7.

4/

61

0.

(
)

3.

3/

01

7.

4/

61

7.

(
)

7.

3/

71

5.

4/

61

3.

(
2

0.

3.

1(

3/

2.

2
9

4/

31

2.

(
)

7.

3/

62

8.

4/

25

0.

6(

2.

0.

3/

51

1.

6(

4/

69

1.

6(

0.

3/

51

8.

4/

91

6.

(
5

2.

0.

1(

3/

4.

1(

4/

17

8.

1(

%
0.

3/

72

6.

4/

24

2.

(
1

0.

0<

7.

3/

92

7.

4/

14

1.

(
)

3.

3/

3.

4/

15

1.

(
1

0.

0<

7.

3/

32

7.

6(

4/

23

0.

1(

7.

6(

3/

5.

6(

4/

23
)

6.

3(
)

7.

1(

3/

8.

4/

31

5.

(
1

0.

0<

7.

3/

32

0.

4/

15

5.

(
)

3/

1.

4/

02

9.

2(
)

3.

3/

9.

4/

12

6.

(
u

it

C(
u

ht
r

it

ir

tc
a

la

ta

bu

P
b

1
3

9/

12
4

9.

9/

96
1

4/

61

0.

4.

1/

34
1

4/

25

6.

1/

62

1
1

4/

18
1

0.

0<

9.

1/

24
1

4/

41
)

5.

1/

55
1

4/

89

6.

1/

27
1

4/

41

0.

2.

1/

93
1

4/

72

8.

1/

92
1
9

4/

25

4.

6.

1/

49
9

4/

71

4.

1/

57
1

4/

57

0.

9.

(
9

1/

51
1

4/

64

1.

1/

45
1

4/

71

0.

0<

2.

1(

1/

2
1

4/

42

8.

1/

76
1
1

4/

42

0.

0<

9.

1/

31

4/

51

3.

1/

9
1

4/

61

8.

1(

1/

03
1

4/

51
1

0.

0<

1.

6(

1/

11

4/

21

4.

1/

4
1

4/

25

5.

1/

55

dr

bu

fl

r
s

i
t

B
41

8
6

36

21

a
)

I
n

e
C

i
a

g
)

7,

64
6/

42

8.

82
1/

15

4.

1‐

1.

1(

3.

31
)

3.

6(

64
6/

24

2.

6(

82
1/

77

1.

1‐

9.

1(

0.

31
)

7.

64
6/

52

3.

82
1/

85

4.

1‐

1.

1(

3.

31
)

5.

3
(

64
6/

01
2

0.

82
1/

55

3.

1‐

0.

1(

1.

31
)

8.

64
6/

81

7.

1(

82
1/

51

9.

1‐

6.

1(

7.

21
)

2.

64
6/

10

4.

82
1/

44

5.

1‐

1.

1(

3.

31
)

8.

6(

64
6/

34

6.

6(

82
1/

48

1.

1‐

9.

1(

0.

31
)

0.

6(

64
6/

83

5.

6(

82
1/

08

2.

1‐
0.

1(

1.

31
)

9.

64
6/

85
2

5.

82
1/

84

2.

1‐

0.

1(

1.

31
)

6.

1(

64
6/

01

1.

82
1/

72

3.

1‐

0.

1(

1.

31
1

4.

64
6/

93
5

9.

82
1/

01

1.

1‐

0.

1(

1.

31
)

0.

1(

64
6/

11

0.

82
1/

69
)

0.

64
6/

35

0.

82
1/

23

1‐
0

it

m
)

8.

6(

64
6/

60

1.

82
1/

95

1.

1‐

9.

1(

0.

31
)

6.

64
6/

63

1.

82
1/

4.

1‐

6.

1(

9.

21
)

6.

64
6/

02

8.

82
1/

56

5.

1‐

1.

1(

3.

31
o

2‐

it

m
)

9.

64
6/

92

0.

82
1/

23

0.

1‐

8.

1(

9.
21
)

6.

64
6/

32

9.

82
1/

5.

1‐

6.

1(

8.

21
)

6.

64
6/

333

0.

82
1/

19

4.

1‐

1.

1(

3.

31
a

gA

EL

AT
tr

c
G
A
M

it

la

ta

M
n

it

N
n

it

la

F
s

li

tr

tl

W
ts

L
e

ld
di

M
ts

gi

H
I

M
B

la

ta

M
5.

1<
5.

1≥
e

ht
r

iB
A

GS
A

GA
ht

tn
u

tS
ht

tn
u

to

N
e

de
hc
a

R
o

N
s

Y
o

it

tn
u

to

N
o

21

1‐

R
o

tn
u

tS
o

it

tn
u

s
to

N
o

R
o

2‐

21
o

tn
u

tS

128 SVEFORS et al.

| those with an appropriate weight at birth (Table 6). Girls and boys

t 2

n 9
24‐54 months to experience In this study, we followed a recovery from stunting and
an earlier menarche or birth cohort from rural pubertal develop‐
e 4/

55

pubertal growth Bangladesh and

1

7
o

.
l
2

) 6
e
9

v 1
%

e
6.

d 6

i 3
P
(

spurt than adolescents who

)

3
%

never had been stunted

4
7.

9 6(

(Tables 5 and 6).

/

0
(
3/

2
43

5
)

%
8

.
9.

examined the association

9 9

ment. A difference in
4 1(

between longitudinal growth

pubertal maturation
3
9

4 4/

restriction,
between children who
89

9 4

/ /

)
9
2

3
9

34 )

8.

) 1(

%
who were stunted at 12, 24
and 54 months were less 4 | DISCUSSION
7

likely to be in a
2 )

4
%
(

higher Tanner stage (Tables

1 3

0.

3 and 4) and reach menarche

0< 8

)
(Table 5) or
enter pubertal growth spurt
%

7.

(Table 6). There was no

6

tendency among
(

3/

adolescents who recovered

32

)
from stunting between 0‐12,
% 12‐24 or
)

assessed as stunting,
8.

1
%

(
were born small or AGA was
%

small were less likely to have

4

.
4

(
expected, children who were
seen only in boys, where entered the puberty growth stunted in infancy and
boys born
a 3 )
spurt. As
53
childhood had
h

c
4
% had a similar implemented in an plex cohort study
area with a over long time.
9/

later pubertal
i
4.

puberty
96

1
5
bu

Experienced field
P
b 2
1(

2 4/

development, that is development as health and

workers, study
4

d 51 6

girls had later demographic

r

their peers who

1/
n
62 i
1

a
b
2

1
3
1

0.
4/

8
menarche and boys never had been surveillance system
4

and a well‐
81
0<

had started their stunted,

9/

12
4

growth spurt later independently of developed research

%

9.

3
as compared to non‐ when during infancy infrastructure. This
stunted children. or childhood the re‐ setting made it
6(

P
9

Children who possible to run a

6

covery had
)

1/

8
%

recovered from occurred. com‐

01

) 7.

0
1

%
.

stunting, however, The MINIMat cohort

2
2

9. (
h
4/
7 9
97 t

study was
1 6

1/

t c n e ) s

frequent education, and status was

g
r t
n e
%

the observed associated with

e
6. G

refresher
r

5
m

puberty
o

p f

training, slight over‐

7 e

development
o
(
t
l
82
1/ a

e 02

representation and therefore

ir

including
p
v

po‐
e

anthropometry of a lower
d

standardization socioeconomic
s )

4
a 1.

3
e

exercises. group among

r
1‐

9.

B
2

There were non‐

1(

0.

participants
31

small and
)

statistically was not con‐

1.

significant tradictory. Of
(

64
6/

02
2

s
.

r differences in the background

l

characteristics
e

back‐ ground
r ie

i
41 )
M‐

o n

characteristics that differed

%
a

m
l
1.

4
4 p

%
5‐

4
3

82
a

K
between between
1/
2
o

participating
02
5 g .
2
t
.

)
n

8 I it
n g

and non‐
n

6
e n
u
(
i
a
t

i dr

participating
a
3 s
) o

4 6. c

m 3 c

children.
9 )

1‐ a
/

% 3.

6
3 s
4.

Younger, more
is
4
5 1( y

la

tentially a
36 4 n
5.
(

educated
31 a
64 t
6/

confounding
39 a
t
2

nurses and women may factor. The

e

)
A

la
medical doctors spend time out differences
were small,
n

collected data of the study

%

2.

adjust‐
it

a
4

during
t

area more
2 ,

s
(

3
82
e
1/

pregnancy, at
13

frequently, for
4

birth, and in the example in their

o

ll

follow‐up
a

maternal home.
)

participating
m
0.

periods during The household and non‐

% S

1‐
6
,
7.
.

2 A

infancy, participating
3

wealth
1(

1 G

childhood and classification children, only

(
8. S
;

21

3 ht

socioeconomic
r

i
4

ad‐ olescence. reflects partly

b )
9
t

/
a
%
)

The data
8 C

other
5
(
% e
2

6. g
m
21

collectors
0

conditions than
A

( la

received
64

maternal
6/

o
3
a d
it
e
a
e 31

h
ments were analyses, and it unlikely that they influenced findings. There significant differences in of stunting at later, be‐
done in the is therefore
9

the main were small but

4

9
(

the occurrence birth, but not e

G n v tS

differences menarche The date of

)

onset at two information

o

) y r y
u o

r t

date and menarche time points, close to the

c pp b
d

in
A
d
e S
e
e

those with was increasing event. The

v e

R
u
t

background
o
A

n a

missing collected by the accuracy of

c
i l
G

characteristi informa‐
t e
u

self‐ probability recall of

r 4 A

n
c

2 la
:

cs or linear
t
o
d Ca
0

reported of collecting menarche

g
s
a

C a ,
n

tween non‐ growth

(

date of the
t

d 3
o

it S

e
a

participating between
o 2 e b
tn iv

3 u e (
g
%
t m r

2 a

and
s 1
bb
4 .
E
A
n
to 5‐

participating
L
N 42 e
o o

v
B

children.
m m
e
n

‐
A 4 4

tion on exact
a

There were
5 o
5
T
A r t

girls with a day.

M g t

no
o ‐
o
n

r i e

complete
d 4
d
f t t

e 9
m
e
S
r i
/
tn
u t
,
e

SVEFORS et al. 129

|

TABLE 4 Self‐reported pubertal development stages in relation to maternal characteristics, birth size, previous stunting and recovery from
stunting among boys in the MINIMat cohort, Bangladesh

Boys

Genital development 3 4 and 5a Pubic hair development 2 and 3b 4 and 5c

All P
455/882 (51.6%) Maternal education 643/882 (72.9%) 427/882 (48.4%)
239/882 (27.1%) P

No education 94/239 (39.3) 200/643 (31.1%) 0.02 145/427 (31.9%) 149/455 (34.9%) 0.34 Formal education 145/239 (60.7%) 443/643 (68.9%)
310/427 (68.1%) 278/455 (65.1%) Wealth index tertiles

Lowest 97/239 (40.6%) 200/643 (31.1%) <0.01 156/427 (36.5%) 141/455 (31.0%) <0.01 Middle 99/239 (41.4%) 236/643 (36.7%) 174/427 (40.7%)
161/455 (35.4%) Highest 43/239 (18.0%) 207/643 (32.2%) 97/427 (22.7%) 152/455 (33.6%) Maternal BMI

<18.5 89/239 (37.2%) 166/643 (26.0%) <0.01 143/427 (33.6%) 112/455 (24.8%) <0.01 ≥18.5 150/239 (62.8%) 473/643

(74.0%) 283/427 (66.4%) 340455 (75.2%) Birth size

SGA 154/239 (64.4%) 349/642 (54.3%) <0.01 263/427 (61.6%) 240/455 (52.7%) <0.01 AGA 85/239 (35.6%) 294/642 (45.7%) 164/427 (38.4%)
215/455 (47.3%) Stunted at birth 44/239 (18.4%) 93/642 (14.5%) 0.15 80/427(18.7%) 57/455 (12.5%) 0.01 Not stunted at birth 195/239 (81.6%)
550/643 (85.5%) 347/427 (81.3%) 398/455 (87.5%) Stunting and recovery from stunting 0‐12 mo

Not stunted at 0 or 12 mo 108/239 (45.2%) 424/642 (65.9%) <0.01 214/427 (50.1%) 318/455 (69.9%)
<0.01

Recovered 0‐12 mo 6/239 (2.5%) 28/642 (4.4%) 16/427 (3.7%) 18/455 (4.0%) Stunted at 12 mo 125/239 (52.3%) 191/642 (29.7%)
197/427 (46.1%) 119/455 (26.2%) Stunting and recovery from stunting 12‐24m

Not stunted at 12 or 24 mo 74/239 (30.9%) 336/642 (54.3%) <0.01 158/427 (37.0%) 252/455 (55.4%)
<0.01

Recovered 12‐24 mo 10/239 (4.2%) 19/642 (3.0%) 11/427(2.6%) 18/455 (4.0%) Stunted at 24 mo 155/239 (64.9%) 288/642 (44.8%)
258/427 (60.4%) 185/455 (40.7%) Stunting and recovery from stunting 24‐54 mo

Not stunted at 24 or 54 mo 80/239 (33.5%) 350/642 (54.4%) <0.01 164/427 (38.4%) 266/455 (58.5%)
<0.01

Recovered 24‐54 mo 55/239 (23.0%) 153/642 (23.8%) 106/427 (24.8%) 102/455 (22.4%) Stunted at 54 mo 104/239 (43.5%)
140/642 (21.8%) 157/427 (36.8%) 87/455 (19.1%)

Abbreviations: AGA, Appropriate for Gestational Age at birth; SGA, Small for Gestational Age at birth.
a
Stage 4 (638/882, 72.3%), Stage 5 (5/882, 0.6%).
b
Stage 2 (6/882, 0.68 %), Stage 3 (421/882, 47.7%).
c
Stage 4 (453/882, 51.4%), Stage 5 (2/882, 0.2%).

date depends mainly on the interval of recall, that is the time be‐ tween probably low. The time‐to‐event analyses employed are appropriate
13
the date of menarche and the time of data collection. In our study, for this kind of data, where some girls have not yet had their first
the mean interval of recall was 0.9 years, thus the recall errors were menstruation.
 The analysis of start of growth spurt was done only in the sub‐ set of spurt was mainly based on the height measurements and did there‐
boys, who had anthropometric measurement at 10 years of age. The fore not suffer from the reporting bias that influences the self‐as‐
10‐year follow‐up was carried out in one of the sessment of pubertal stage.
two‐calendar‐year birth cohort. The selection should therefore not Studies on the reliability of self‐assessment of pubertal have found
suffer from any selection bias related to the exposures of interest in reasonable agreement between self‐assessment and exam‐ ination by
this analysis. The classification of boys in relation to pubertal growth a trained clinician in some studies,14 whereas others have
130 SVEFORS et al.
|

TABLE 5 Size at birth, stunting and recovery from stunting in boys. On average, girls start to develop secondary sexual charac‐
relation to the hazards ratios for girls having reached menarche in the teristics about 1 year before boys9 and their pubertal growth spurt
MINIMat cohort. Crude and adjusted Cox’s proportional hazard
starts 2 years ahead of boys: Growth velocity during puberty is at its
analyses n = 987
highest in girls at breast stage three and in boys at genital stage four.
Crudea Adjustedb Due to the fact that it is probable that boys overestimated their
pubertal stage, height measurements from the follow‐ups were used
HR 95% CI HR 95% CI
to estimate pubertal development.
Size at birth
The research question ‘Does recovery from stunting come with
AGA ref ref any adverse future effects?’ inspired the analysis in this paper. The
SGA 1.00 0.87‐1.16 1.06 0.90‐1.23 Stunted at birth terminology regarding recovery from stunting and catch‐up growth is
problematic, as there are no commonly accepted definitions.
Not stunted at birth 0.93 0.77‐ 1.14 0.94 0.76‐1.27 ties above normal after a period of slow growth.
Similar studies in high‐income settings have This type of ‘catch‐ up growth’ was not prevalent
Stunted at birth
ref ref used the terminology ‘catch‐up’, describing in our cohort. Recovery from stunting
children with rapid linear growth, with veloci‐
Stunting and recovery from stunting 0‐12 mo was in this paper defined as having a HAZ score above minus two,
Not stunted at 0 or 12 mo 1.05 0.77‐ 1.44 1.08 0.79‐1.49 0.70 0.60‐0.83 0.71 median age at menarche in our sample was 13.0
Recovered years, which was substantially lower than a
0‐12 mo 0.60‐0.84 previous estimate of 15.8 years in Bangladesh in
Stunted at 197616, but the same as in a study from 1996
12 mo with no extra criteria. However, the children (me‐ dian age 13.0)17 and slightly higher than the
ref ref
that were classified as ‘recovered’ from stunting estimate in a study from
did display a substantial increase in HAZ. The
Stunting and recovery from stunting 12‐24 mo 2009 (median age 12.8 years)18. These assessments were all carried
Not stunted at 12 or 24 mo 0.92 0.62‐1.39 0.97 0.64‐ 1.46 0.68 0.59‐0.79 0.75 much from age at menarche in several high‐
Recovered income countries, such as Canada, 12.9
12‐24 mo 0.64‐0.88 years,19and Sweden, 13.2 years5. During the last
Stunted at century, age at menarche has decreased in most
24 mo high‐income countries and the same downward
out in rural areas. The median age at menarche
ref ref secular
of 13.0 years from our study does not differ

Stunting and recovery from stunting 24‐54 mo trend is now also observed in low‐ and middle‐income countries. As

Not stunted at 24 or 54 mo 1.00 0.83‐1.20 1.06 0.87‐1.29 0.54 0.46‐0.64 0.54 reported in our and other studies from rural
Recovered Bangladesh is not fully explained, considering
24‐54 mo 0.45‐0.64 the large proportion of adolescents that still is
Stunted at affected by undernutrition.
54 mo the decline in age at menarche mainly has been In our cohort, more than 60% of the children
ref ref attributed to general improvements in nutrition were born SGA,
and health, the relatively low age at men‐ arche
Abbreviation: AGA, Appropriate weight for Gestational Age at birth; earlier or later pubertal development stage. Studies on the reliability
SGA, Small for Gestational Age of self‐assessment have also reported that girls, especially younger
a
Bivariable analysis of each row characteristic separately bAnalysis for
girls, tend to underestimate their development stage, whereas boys,
each row characteristic with adjustments made for preterm birth,
maternal parity, height, BMI, education and socioeco‐ nomic status especially older boys, tend to overestimate. This could possibly be due
when giving birth. to expectations to be in a similar puberty stage as their peers.15 These
biases were most probably also present in our study, where boys
generally reported being in a later puberty stage than girls, de‐ spite
found considerable inconsistencies.15 There seems to be an agree‐ the fact that girls generally enter puberty at a younger age than
ment that self‐assessment is appropriate in larger epidemiological and length for age at birth was almost one standard deviation below
studies in order to examine the relative distribution of being in an the median of the WHO growth reference. Linear growth faltered
 dramatically up to 24 months of age when half of the children were boys. Boys had a higher stunting prevalence during infancy and at
stunted. Very few recovered from stunting during infancy. About one‐ puberty, whereas girls had a higher prevalence of stunting at 4.5
fifth of the children recovered from stunting after 2 years, but at 4.5 years. This may be explained by factors such as differences in foetal
years one‐third of the children were still stunted. This growth pattern growth,22 variation in susceptibility to exposures during early
is consistent with established growth trajectories in South Asia and development23 as well as gender inequalities in the society starting
reported patterns of stunting and recovery from stunting from both already in childhood.24 Another factor to be considered is a different
Africa and Asia.20,21 timing of puberty growth spurt onset among girls and boys in our
The prevalence of stunting differed significantly between girls and study sample as compared to the WHO reference population.
SVEFORS et al. 131
|

TABLE 6 Size at birth, stunting and recovery regression analyses Characteristic Mean age Size at birth
from stunting in relation to the odds ratios for
Started
boys having started the growth spurt in the
spurt? Crudea Adjustedb No Yes OR 95% CI OR
MINIMat cohort. Crude and adjusted logistic
95% CI

AGA 13.8 22 153 ref ref SGA 13.9 58 187 0.46 0.27‐0.79 0.47 0.24‐0.92 Stunted at birth

Not stunted at birth birth

13.8 59 289 ref ref 13.9 21 50 0.49 0.27‐0.87 0.31 0.13‐0.74
Stunted at

Stunting and recovery from stunting 0‐12 mo

Not stunted at 0 or 12 mo 12 mo
13.8 36 201 ref ref 13.9 5 14 0.50 0.17‐1.48 0.28 0.07‐1.15 13.9 38 121 0.57
Recovered
0‐12 mo
0.34‐0.95 0.45 0.23‐0.89
Stunted at

Stunting and recovery from stunting 12‐24 mo

Not stunted at 12 or 24 mo 24 mo
13.8 22 144 ref ref 13.9 2 13 0.99 0.21‐4.70 1.19 0.13‐11.0 13.9 50 158 0.48
Recovered
12‐24 mo
0.28‐0.84 0.54 0.27‐1.05
Stunted at

Stunting and recovery from stunting 24‐54 mo

Not stunted at 24 or 54 mo 54 mo
13.9 23 155 ref ref 13.8 17 70 0.61 0.31‐1.22 0.91 0.40‐2.07 13.9 34 98 0.43
Recovered
24‐54 mo
0.24‐0.77 0.36 0.17‐0.78
Stunted at

Abbreviations: AGA, Appropriate weight for Gestational Age at birth; SGA, Small for Gestational
Age.
a
Bivariable analysis of each row characteristic separately.
b
Analysis for each row characteristic with adjustments made for preterm birth, maternal parity,
height, BMI, education and socioeconomic status when giving birth.

Age at puberty onset and linear growth restriction can be viewed association of being born SGA and early puberty in girls. The reasons
as two outcomes of nutritional cues, an older age at puberty onset of the discrepancy between our findings and earlier evidence might be
prolongs the childhood growth phase and results in a taller height at that the causes of intrauterine growth restric‐ tion and low
the start of puberty. As expected, children who were stunted in birthweight are complex and vary between different
childhood had reached a lower pubertal stage and had an older age at contexts. High‐income settings, where most studies on SGA and pu‐
menarche. berty have been carried out, have a prevalence of SGA from 4% to 8%
Earlier epidemiological and adoption studies have reported as‐ and smoking is the most prominent risk factor.27 In our setting, we had
sociations of being born SGA with a younger age at puberty onset, an almost tenfold higher prevalence of SGA, and smoking was
especially in girls.25,26 extremely rare among women. Another important factor that
In our sample, we found evidence that SGA boys differed in pu‐ bertal distinguishes the South Asian context is the relatively small maternal
development as compared to AGA boys. We did not, however, find any size, contributing to small size at birth.28
 There were gender differences infant and child growth tra‐ growth trajectories after birth. Another explanation can be the dif‐
jectories, occurrence of stunting and recovery from stunting. The ferent outcomes representing puberty development for girls and boys,
observed gender differences in the association between SGA and as these occur at different stages of the pubertal development,
pubertal development may therefore reflect the gender‐specific menarche (relatively late) and growth spurt (very early).
132 SVEFORS et al.
|

Many studies that have reported associations between prenatal The authors have no conflicts of interest to declare.
growth, post‐natal growth during infancy and childhood and pu‐ ORCID

berty.26,29 In an adoption study from Sweden, the girls with the most Pernilla Svefors https://orcid.org/0000-0003-2612-2973 Lars‐
pronounced stunting and fastest weight gain and catch‐up growth had
Åke Persson https://orcid.org/0000-0003-0710-7954
the lowest age at menarche.29 In our analysis, children who re‐ covered
from stunting had a similar puberty development as children who
were never stunted, irrespective of when during infancy or child‐ hood
the recovery occurred. That is, we did not find any evidence of adverse REFERENCES
effects of recovery from stunting in infancy or childhood in terms of
1. Victora CG, Adair L, Fall C, et al. Maternal and child undernutri‐ tion:
very early pubertal development. In contrast to most of the above‐ consequences for adult health and human capital. Lancet
mentioned epidemiological and adoption studies, the adoles‐ cents in 2008;371(9609):340‐357.
this cohort were born into and remained in an underpriv‐ ileged 2. Kivimäki M, Lawlor DA, Smith GD, et al. Association of age at men‐
setting. The nutrition transition, characterised by increased access to arche with cardiovascular risk factors, vascular structure, and func‐
tion in adulthood: the cardiovascular risk in young Finns study. Am J
highly processed foods and sugar‐sweetened beverages and less
Clin Nutr. 2008;87(6):1876‐1882.
physical activity, which rapidly takes place in many low‐ to middle‐ 3. Stöckl D, Meisinger C, Peters A, et al. Age at menarche and its as‐
income settings, is not yet fully present in our rural study area, as sociation with the metabolic syndrome and its components: results
reflected by the low occurrence of overweight in our sample.30 Thus, from the KORA F4 study. PLoS ONE 2011;6(10):e26076.
4. Akter S, Jesmin S, Islam M, et al. Association of age at menarche with
the results of our study are mainly representative for other rural areas
metabolic syndrome and its components in rural Bangladeshi women.
in Bangladesh and neighbouring countries in South Asia where Nutr Metab. 2012;9(1):99.
children are born small and become and stay short and light. 5. Parent A‐S, Teilmann G, Juul A, Skakkebaek NE, Toppari J,
The reported associations of puberty onset with adult chronic Bourguignon J‐P. The timing of normal puberty and the age limits of
sexual precocity: variations around the world, secular trends, and
diseases do not necessarily reflect causal pathways.2 Puberty onset
changes after migration. Endocr Rev. 2003;24(5):668‐693.
and adult chronic risk may, however, share common developmental 6. Adair LS. Size at birth predicts age at menarche. Pediatrics
origins. Age at puberty may therefore be seen as an early indicator for 2001;107(4):1‐9.
the interactions of genetics and environmental conditions possi‐ 7. Proos L, Gustafsson J. Is early puberty triggered by catch‐up growth
bly affecting future adult health. following undernutrition? Int J Environ Res Public Health
2012;9(12):1791‐1809.
8. Alexander GR, Himes JH, Kaufman RB, Mor J, Kogan M. A United
States National reference for fetal growth. Obstet Gynecol.
5 | CONCLUSION 1996;87(2):163‐168.
9. Tanner JM, Whitehouse RH. Clinical longitudinal standards for height,
weight, height velocity, and stages of puberty. Arch Dis Childhood
In this study, a birth cohort from rural Bangladesh with children hav‐
1976; 51(170): 170–179.
ing substantial prenatal, infancy and childhood growth restriction was 10. Team RC. R: A language and environment for stastical computing;
followed into puberty. Stunted children were less biologically mature 2017.
than their non‐stunted peers. In contrast to earlier stud‐ ies in high‐ 11. Carstensen B, Plummer M, Laara E, Hills M.Package “Epi”. cranr‐
projectorg2017.
income settings or on children adopted to high‐income countries, we
12. Terneau TM, Lumley T.Package ‘survival’. cranr‐projectorg2017. 13.
could not demonstrate any unfavourable association between Koo MM, Rohan TE. Accuracy of short‐term recall of age at men‐ arche.
recovery from stunting in infancy or childhood and very early puberty Ann Human Biol. 2009;24(1):61‐64.
development. Those recovering from stunting in this context had 14. Rabbani A, Noorian S, Fallah JS, Setoudeh A, Sayarifard F, Abbasi F.
similar puberty development as children who had never been stunted. Reliability of pubertal self assessment method: an Iranian study. Iran J
Pediatr. 2013;23(3):327‐332.
Considering stunting’s devastating consequences for the individual
15. Rasmussen AR, Wohlfahrt‐Veje C, Tefre de Renzy‐Martin K, et al.
and society, the prevention and treatment of stunt‐ ing should be Validity of self‐assessment of pubertal maturation. Pediatrics
highly prioritized in national and global public health agendas. 2015;135(1):86‐93.
16. Chowdhury AKMA, Huffman SL, Curlin GT. Malnutrition, menarche,
and marriage in rural Bangladesh. Social Biol. 1977;24(4):316‐325. 17.
ACKNOWLEDGEMENTS Chowdhury S, Shahabuddin AKM, Seal AJ, et al. Nutritional status and age
at menarche in a rural area of Bangladesh. Ann Human Biol.
We thank the participants and their families in Matlab for their con‐ 2000;27(3):249‐256.
tinuing involvement in the MINIMat trial, and the field‐team mem‐ 18. Rah JH, Shamim AA, Arju UT, Labrique AB, Rashid M, Christian P. Age
of onset, nutritional determinants, and seasonal variations in men‐ arche in
bers and data management staff for their excellent work.
rural Bangladesh. J Health Popul Nutr. 2009;27(6):802‐807.
19. Harris MA, Prior JC, Koehoorn M. Age at menarche in the Canadian
population: secular trends and relationship to adulthood BMI. J
CONFLICT OF INTEREST
 Adolesc Health 2008;43(6):548‐554. Transitions to stunted and non‐stunted status both occur from birth
20. Teivaanmäki T, Cheung YB, Kortekangas E, Maleta K, Ashorn P. to 15‐years‐of‐age in Malawi. Acta Paediatr. 2015;104(12): 1278‐
1285.
SVEFORS et al. 133
|

21. Adair LS. Filipino children exhibit catch‐up growth from age 2 to 12
years. J Nutr. 1999;129(6):1140‐1148.
22. Brown ZA, Schalekamp‐Timmermans S, Tiemeier HW. Fetal sex
specific differences in human placentation: a prospective cohort
study. Placenta 2014;35(6):359‐364.
23. van Abeelen AFM, de Rooij SR, Osmond C, et al. The sex ‐specific
effects of famine on the association between placental size and later
hypertension. Placenta 2011;32(9):694‐698.
24. Khatun M, Stenlund H, Hörnell A. BRAC initiative towards pro‐ moting
gender and social equity in health: a longitudinal study of child growth
in Matlab, Bangladesh. Public Health Nutr. 2004;7(08): 1071‐1079.
25. Ibanez L, de Zegher F. Puberty and prenatal growth. Mol Cell Endocrinol.
2006;5:22‐25.
26. Persson I, Ahlsson F, Ewald U, et al. Influence of perinatal factors on the
onset of puberty in boys and girls: implications for inter‐ pretation of
link with risk of long term diseases. Am J Epidemiol. 1999;150(7):747‐
55.
27. Kjøbli E, Bach R, Skogseth H, Jacobsen GW. The Scandinavian Small‐
for‐Gestational Age (SGA) pregnancy and birth cohort – A source to
continual insight into fetal growth restriction and long
term physical and neurodevelopmental health in mother and off‐
spring. Norsk Epidemiologi. 2016;26(1–2):1‐10.
28. Svefors P, Rahman A, Ekström E‐C, et al. Stunted at 10 years. Linear
growth trajectories and stunting from birth to pre‐adolescence in a
rural Bangladeshi cohort. PLoS ONE 2016;11(3):e0149700 ‐
e0149718.
29. Proos LA, Hofvander Y, Wennqvist K, Tuvemo T. A longitudinal study
on anthropometric and clinical development of Indian chil‐ dren adopted in
Sweden, Upsala. J Med Sci. 1992;97(1):93‐106.
30. Khan AI, Hawkesworth S, Ekström E‐C, et al. Effects of exclusive
breastfeeding intervention on child growth and body composition:
the MINIMat trial, Bangladesh. Acta Paediatr. 2013;102(8):815‐823.

How to cite this article: Svefors P, Pervin J, Islam Khan A, et al.

Stunting, recovery from stunting and puberty development in
the MINIMat cohort, Bangladesh. Acta Paediatr.
2020;109:122–133. https://doi.org/10.1111/apa.14929