Public Health Nutrition: 8(5), 501–508 DOI: 10.

1079/PHN2005786

Prevalence and determinants of stunting and overweight

in 3-year-old black South African children residing in the
Central Region of Limpopo Province, South Africa
Ramoteme L Mamabolo1,2,*, Marianne Alberts1, Nelia P Steyn2,
Henriette A Delemarre-van de Waal3 and Naomi S Levitt4
1
Medical Sciences Programme, University of Limpopo, Sovenga, South Africa: 2Chronic Diseases of Lifestyle Unit,
Medical Research Council, PO Box 19070, Tygerberg, Cape Town 7505, South Africa: 3Pediatric Endocrinology,
Vrije University Medical Centre, Vrije University, Amsterdam, The Netherlands: 4Department of Medicine,
University of Cape Town, Cape Town, South Africa

Submitted 15 March 2005: Accepted 6 June 2005

Abstract
Objectives: To determine the prevalence of stunting, wasting and overweight and their
determinants in 3-year-old children in the Central Region of Limpopo Province, South
Africa.
Design: Prospective cohort study.
Setting: Rural villages in the Central Region of the Limpopo Province, South Africa.
Subjects: One hundred and sixty-two children who were followed from birth were
included in the study. Anthropometric measurements and sociodemographic
characteristics of the children were recorded.
Results: Height-for-age Z-scores were low, with a high prevalence of stunting (48%).
The children also exhibited a high prevalence of overweight (22%) and obesity (24%).
Thirty-one (19%) children were both stunted and overweight. Gaining more weight
within the first year of life increased the risk of being overweight at 3 years by 2.39
times (95% confidence interval (CI) 1.96–4.18) while having a greater length at 1 year
was protective against stunting (odds ratio (OR) 0.41; 95% CI 0.17–0.97). Having a
mother as a student increased the risk for stunting at 3 years by 18.21 times (95% CI
9.46 –34.74) while having a working mother increased the risk for overweight by
17.87 times (95% CI 8.24 –38.78). All these factors also appeared as risks or as being
Keywords
protective in children who were both overweight and stunted, as did living in a
Stunting
household having nine or more persons (OR 5.72; 95% CI 2.7–12.10). Overweight
Conclusion: The results of this study highlight the importance of evaluating Determinants
anthropometric status in terms of both stunting and overweight. Furthermore, it is Risk factors
important to realise the importance of normal length and weight being attained at Sociodemographic factors
1 year of age, since these in turn predict nutritional status at 3 years of age. Children

The nutritional transition, i.e. the major change in the with poverty, poor nutrition and high prevalences of
nutritional profile of human populations resulting from a infectious diseases such as found in developing countries4,6.
shift from a traditional diet to a Western one, is determined Socio-economic status such as the mother’s education and
by the interplay of economic, demographic, environmen- occupation, household income and health expenditure may
tal and cultural changes occurring in a society and plays a influence stunting levels indirectly7 – 9, whilst a number of
role in the prediction and determination of the nutritional factors may be directly causative, including micronutrient
status of the overall population1 – 3. deficiencies, inadequate protein intake, intrauterine malnu-
Children’s physical growth patterns are among the tools trition, maternal stature and infections5,10.
used to assess a population’s nutritional status. Physical In Africa as a whole, and in Sub-Saharan Africa, 35% and
growth itself is regulated by two major factors: genetics 42% respectively of children under 5 years of age are
and the environment4,5. Although the quantity and quality believed to be stunted11,12. In South Africa, a greater
of food are the main environmental determinants of good proportion of younger children are stunted (21–48%)
nutritional status, children’s physical and psychosocial rather than underweight (8 – 15%)13 – 16. Experience has
surroundings also impact on their growth. shown that these chronically malnourished children are
Stunting (i.e. linear growth faltering) is an important susceptible to increased morbidity and mortality, and are
public health problem for children living in environments the most vulnerable group in communities. Endemic

*Corresponding author: Email Lesley.mamabolo@mrc.ac.za q The Authors 2005

Downloaded from http:/www.cambridge.org/core. IP address: 36.83.180.13, on 08 Nov 2016 at 21:25:16, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1079/PHN2005786
 502 RL Mamabolo et al.
poverty (as evidenced by poor education, unemployment mounted on a board to the nearest 0.5 cm, with the child
and increased household density) and lack of basic standing on a flat surface against the wall.
resources for nutritional well-being form the basis for The World Health Organization recommends the United
poor nutritional status in these children16 – 19. States’ National Center for Health Statistics percentiles as a
It is now also recognised that, in addition to under- reference for international use24. Following conventional
nutrition in children and pregnant mothers, the nutrition cut-off points of Z-scores, malnutrition in its various forms
transition has led to an increase in overweight and obesity is defined as follows:
in the adult population resulting in an increase in the
. Stunting – height-for-age Z-score (HAZ) below the
prevalence of chronic diseases20. In recent years, however,
median by more than minus two standard deviations
this phenomenon has filtered down to the child population
(, 22SD).
of South Africa and other developing countries undergoing
. Wasting – weight-for-height Z-score (WHZ) below the
the transition13,14,16,21,22. The National Food Consumption
median by ,22SD.
Survey in South African children between 1 and 9 years of
. Underweight – weight-for-age Z-score (WAZ) below the
age reported that the prevalence of combined overweight
median by ,22SD.
and obesity at the national level was as high as that for
. Overweight – either body mass index (BMI) above the
stunting, namely 17.1 and 19.3%, respectively16. The
age-specific reference values recommended by the
present study aimed at determining the prevalence and
International Obesity Task Force (IOTF)25 or WHZ
determinants of stunting and overweight in 3-year-old
above the median by .1SD.
children residing in a semi-rural area in South Africa.
. Obesity – either BMI above the age-specific reference
values recommended by the IOTF25 or WHZ above the
Methods
median by .2SD.
Study area and subjects Changes in SD scores between birth and 1 year, 1 and 3
The study took place in villages in the Central Region of years, and birth and 3 years were calculated. A gain in
Limpopo Province, which is to the north of South Africa. weight or length/height SD score between birth and 1 year
These villages are typical of those found in rural areas of of 0.67SD was taken to indicate clinically significant catch-
the country. They are characterised by poor infrastructure up growth; similarly, a decrease in SD score by , 2 0.67SD
and very high levels of unemployment, reflecting a indicated catch-down growth26.
transition between traditional and modern lifestyles since
many of the younger adults work in neighbouring towns Assessment of sociodemographic factors
and cities and return home for periodic visits. The sociodemographic status of households and factors
The study participants were infants born to women who related to the health of infants were assessed by use of a
were recruited while attending prenatal clinics at nine structured questionnaire developed by the authors23. The
randomly selected local clinics during their third trimester. following were included: the mother’s educational status,
Of the 219 infant participants, 162 could be traced at 3 years occupation, marital status, age and parity, and infant’s age
of age (74% response rate). The fall in number at follow-up of weaning. The housing structures were divided into
was largely due to migration both within the studied three categories: traditional mud houses, brick houses and
villages as well as to outside the study area. The lack of shacks (made of corrugated iron and timber). Over-
proper addresses, formal streets and roads in these villages crowding was assessed by determining the total number of
impeded the process of tracing the mothers and their persons in the household and the number of children less
infants, especially during the early phases of the study. than 5 years of age.
At the age of 3 years or within one month thereof,
trained fieldworkers interviewed the caregivers of the Ethical approval
remaining participants and examined the children’s The University of Limpopo Ethics Committee and the
anthropometric status. The results used herein also Limpopo Province Department of Health Research
involved those collected in the earlier phases of the Committee approved the study. The aim and objectives
study, some of which have been published elsewhere15,23. of the study, as well as a detailed explanation of the
procedures to be employed, were explained to the
Anthropometric measurements mother/caregiver of the selected child both verbally and in
Birth anthropometric measurements were taken from the form of an information sheet. Informed oral consent
hospital records. The children’s weight, height, head was obtained from the child’s mother/caregiver prior to
circumference, mid-arm circumference and abdominal the measurements.
circumference were measured. Weight was measured with
a baby scale (TANITA Baby Scale, model 1380; Tanita Statistical analysis
Corporation, Tokyo, Japan) to the nearest 0.1 kg; height Statistical analyses were done using two statistical
was measured with a non-stretchable tape measure packages: Epi-Info 200227 and SPSS 12.028. Statistical

Downloaded from http:/www.cambridge.org/core. IP address: 36.83.180.13, on 08 Nov 2016 at 21:25:16, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1079/PHN2005786
 Stunting and overweight in 3-year-old black South African children 503
significance was set at P , 0.05. Student’s t-test was used (67%). Consequently, the mother was the primary
for comparison between means, the chi-square test for caregiver for the majority of the children (74%).
comparison between proportions in the various sub- Twenty-six per cent of the children lived in either
groups, and Pearson’s correlations to measure the traditional houses or shacks.
association between two continuous variables. Binary
logistic regression was used to assess the risk factors for Stunting in the children
both stunting and overweight. To take care of any spill- A higher proportion of stunted children were underweight
over effect that might have resulted from children being than non-stunted children (17% vs. 2%) (Table 1). In both
both stunted and overweight, a third group consisting of groups the prevalence of obesity was high (26% and 21%,
these children was created. The infants were thus divided respectively).
into four groups: neither stunted nor overweight, stunted, Stunted 3-year-old children lived in households having
overweight, and both stunted and overweight. Based on a larger number of people (P ¼ 0.008) and more children
these, multinomial regression analysis was conducted to below 5 years of age (P ¼ 0.038) than non-stunted
examine factors associated with the various nutritional children. In addition, the mothers of stunted children
states with reference to the normal children. Only factors were significantly shorter than the mothers of non-stunted
that were significant in the binary logistic regression were children (P ¼ 0.006). Of the anthropometric measure-
entered into the model. ments only head circumference did not differ between
these two groups (Table 2).
Results Compared with non-stunted children, a greater pro-
portion of stunted children were already stunted (49% vs.
Of the 162 children who attended the follow-up at 3 years, 22%; P ¼ 0.001) and wasted (22% vs. 3%; P ¼ 0.001) at 1 year
48% were stunted, 9% were underweight, 1% were wasted, of age; came from families with a large number of people per
18% were overweight and 24% were obese (Table 1). The household (33% vs. 19%; P ¼ 0.017); already had younger
physical characteristics of the children according to their siblings (27% vs. 17%; P ¼ 0.064); came from households
nutritional status, i.e. those who were stunted and those with more children below the age of 5 years (27% vs. 12%;
who were overweight, are presented in Table 2. P ¼ 0.007); had mothers with a primary education (15% vs.
8%; P ¼ 0.132); had mothers who were still students (19% vs.
Sociodemographic characteristics 11%; P ¼ 0.124); were more often cared for by their
Overall 85% of the children were from families with more grandmothers (26% vs. 12%; P ¼ 0.014); and more
than five people per household, 21% already had frequently lived in shacks (21% vs. 12%; P ¼ 0.076).
younger siblings, 19% were from families with three or Table 3 shows the results of the binary logistic
more children below the age of 5 years, and a large regression in the stunted as well as the overweight
proportion (71%) were breast-fed for 12 months or more. children. In the stunted children, living in a household
Most of the mothers were still single (61%) and 88% of with nine or more individuals increased the risk of being
them had secondary education or higher. Ninety per cent stunted eight-fold while having a mother who was a
of the children saw their mothers on a daily basis, but student increased this almost four-fold. Protective factors
67% did not live in the same household as their fathers. against stunting were found to be attaining a greater length
The level of unemployment in the mothers was high and having an increased Ponderal Index by 1 year.

Table 1 Prevalence (n, %) of different nutritional states in stunted and non-stunted 3-year-old children

Total (n ¼ 162) Stunted (n ¼ 78) Normal (n ¼ 84) P-value

HAZ
22SD to 2SD Stunted 78 (48)
,22SD Normal 84 (52)
WAZ <0.0001
,22SD Under weight 15 (9) 13 (17) 2 (2)
22SD to 2SD Normal 139 (86) 65 (83) 74 (88)
. 2SD Over weight 8 (5) 0 (0) 8 (10)
WHZ 0.327
,22SD Wasted 1 (1) 1 (1) 0 (0)
22SD to 2SD Normal 132 (81) 61 (78) 71 (85)
. 2SD Over weight 29 (18) 16 (21) 13 (15)
BMI 0.057
19.0–24.9 kg m22 Normal 88 (54) 47 (60) 41 (49)
25.0–29.9 kg m22 Over weight 36 (22) 11 (14) 25 (30)
$ 30 kg m22 Obese 38 (24) 20 (26) 18 (21)

HAZ – height-for-age Z-score; SD – standard deviation; WAZ – weight-for-age Z-score; WHZ – weight-for-height Z-score;
BMI – body mass index (based on sex-specific percentile crossing at 18 years).

Downloaded from http:/www.cambridge.org/core. IP address: 36.83.180.13, on 08 Nov 2016 at 21:25:16, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1079/PHN2005786
 504 RL Mamabolo et al.
Table 2 Maternal and household characteristics and anthropometric measurements (mean^standard deviation) of the total study popu-
lation, the stunted and the overweight children

Total Stunted Non-stunted Overweight and Normal weight

(n ¼ 162) (n ¼ 78) (n ¼ 84) P-value* obese (n ¼ 74) (n ¼ 88) P-value*

Maternal and household characteristics

No. in household 7.1 ^ 2.7 7.7 6 2.9 6.6 6 2.3 0.008 7.6 6 2.8 6.7 6 2.5 0.044
No. of children aged ,5 years 1.7 ^ 0.9 1.9 6 1.1 1.6 6 0.8 0.038 1.6 ^ 0.9 1.8 ^ 2.0 0.226
Time of weaning from breast (months) 14.9 ^ 8.2 14.7 ^ 8.4 15.2 ^ 8.0 0.720 15.3 ^ 8.1 14.6 ^ 8.3 0.618
Maternal age (years) 25.5 ^ 6.9 26.4 ^ 7.5 25.8 ^ 7.4 0.633 25.1 ^ 7.1 26.8 ^ 7.6 0.148
Maternal height (m) 1.6 ^ 0.7 1.5 6 0.1 1.6 6 0.1 0.006 1.6 ^ 0.1 1.6 ^ 0.1 0.212
Parity 2.8 ^ 2.1 2.9 ^ 2.1 2.7 ^ 2.2 0.464 2.6 ^ 2.1 3.0 ^ 2.2 0.249
Child measurements
Birth weight (kg) 3.1 ^ 0.6 3.1 ^ 0.6 3.2 ^ 0.6 0.362 3.3 6 0.5 3.0 6 0.6 0.017
Birth length (cm) 49.3 ^ 2.6 48.9 ^ 2.5 49.5 ^ 2.6 0.210 49.7 ^ 2.1 48.9 ^ 2.8 0.108
Birth Ponderal Index (kg m23) 26.3 ^ 4.5 26.4 ^ 4.5 26.2 ^ 4.4 0.841 26.5 ^ 4.6 26.1 ^ 4.3 0.616
1-year weight (kg) 9.4 ^ 1.4 8.9 6 1.4 9.8 6 1.4 <0.0001 9.9 6 1.4 8.9 6 1.3 <0.0001
1-year length (cm) 72.1 ^ 4.0 70.9 6 3.5 73.4 6 4.0 <0.0001 73.1 ^ 3.4 71.3 ^ 4.2 0.010
1-year Ponderal Index (kg m23) 24.8 ^ 4.1 24.6 ^ 4.4 25.1 ^ 3.8 0.561 25.3 ^ 2.7 24.5 ^ 5.0 0.237
3-year weight (kg) 13.8 ^ 2.0 12.7 6 1.5 14.8 6 1.8 <0.0001 15.2 6 1.7 12.6 6 1.3 <0.0001
3-year length (cm) 87.5 ^ 4.2 84.2 6 2.7 90.7 6 2.6 <0.0001 87.6 ^ 4.5 87.5 ^ 3.9 0.915
3-year head circumference (cm) 50.55 ^ 3.3 50.48 ^ 4.5 50.63 ^ 1.3 0.777 51.69 6 4.2 49.59 6 1.83 <0.0001
3-year mid-arm circumference (cm) 16.46 ^ 1.3 16.08 6 1.4 16.81 6 1.0 <0.0001 17.16 6 1.1 15.87 6 1.07 <0.0001
3-year abdominal circumference (cm) 49.83 ^ 3.3 48.74 6 3.3 50.84 6 3.0 <0.0001 51.70 6 3.0 48.26 6 2.66 <0.0001
BMI (kg m21) at 3 years 18.0 ^ 2.3 17.9 ^ 2.2 18.1 ^ 2.4 0.836 19.9 6 1.9 16.5 6 1.0 <0.0001
HAZ at 3 years 22.0 ^ 1.1 22.9 6 0.7 21.2 6 0.6 <0.0001 22.0 ^ 1.1 22.0 ^ 1.0 0.958
WAZ at 3 years 20.5 ^ 1.2 21.2 6 0.9 0.8 6 1.0 <0.0001 0.3 6 0.9 21.2 6 0.8 <0.0001
WHZ at 3 years 0.9 ^ 1.3 0.7 ^ 1.2 1.0 ^ 1.3 0.067 1.8 6 1.0 0.1 6 0.7 <0.0001

BMI – body mass index; HAZ – height-for-age Z-score; WAZ – weight-for-age Z-score; WHZ – weight-for-height Z-score.
* Student t-test for differences between stunted versus non-stunted children, overweight and obese versus normal-weight children.

Overweight in the children mother daily were less likely to be overweight at 3 years of
Using the reference values of the IOTF25 it was found that age (Table 3).
46% of the children were overweight and obese. A large
proportion of these children (40%) were also stunted. Various nutritional states of the children
As seen in Table 2, there were more people living in the From the multinomial regression analysis, being stunted
households of the overweight compared with the normal- was associated with having a mother who was a student,
weight children (P ¼ 0.044). The overweight children’s while attaining greater length at 1 year was a protective
anthropometric measurements were significantly different factor. Having a working mother and being overweight at
from those of the normal-weight children except for birth 1 year were risk factors for being overweight at 3 years. On
length, birth and 1-year Ponderal Index; and length and the other hand, being stunted and overweight was
HAZ at 3 years of age (Table 2). associated with having a large number of individuals in
When comparing overweight children with normal- the household; having a mother who was employed or a
weight children, a smaller proportion of the overweight student; and having a greater weight at 1 year. The only
children were wasted at birth (27% vs. 57%; P , 0.0001); factor found to be protective against stunting and
stunted at 1 year (26% vs. 41%; P ¼ 0.013); underweight at 1 overweight was having a greater length at 1 year (Table 4).
year (6% vs. 16%; P ¼ 0.019); and had siblings (14% vs.
26%; P ¼ 0.038). A higher proportion of overweight Catch-up growth
children had a working mother (23% vs. 13%; P ¼ 0.063) The study also considered children who showed either
and were mainly from households of greater household catch-up or catch-down growth for both weight and
density (34% vs. 20%; P ¼ 0.027) compared with normal- length from birth (Table 5). No significant gender
weight children. differences were observed in the children with respect to
Children having a low WHZ at birth were six times more the various growth patterns followed.
likely to be overweight at 3 years of age, while those with a With respect to the change in SD scores for HAZ, a high
higher weight at 1 year were seven times more likely to be proportion of children who showed catch-down growth
overweight at 3 years of age. Having an employed mother between birth and 1 year had no electricity in the home
was also a risk factor for being overweight in children. It (x 2 ¼ 3.76, P ¼ 0.001) and were born to older mothers
was three times higher if the mother worked or was a (x 2 ¼ 10.10, P ¼ 0.039). Catch-down growth between 1
student. However, the risk was increased to six-fold if the and 3 years was associated with lack of electricity
mother was younger than 20 years old and eight-fold (x 2 ¼ 6.39, P ¼ 0.041) in addition to more individuals in
higher if the mother was the primary caregiver. Children the household (x 2 ¼ 9.25, P ¼ 0.05), being born to a
who attained a greater length by 1 year and saw the mother with low parity (i.e. three children or less)

Downloaded from http:/www.cambridge.org/core. IP address: 36.83.180.13, on 08 Nov 2016 at 21:25:16, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1079/PHN2005786
 Stunting and overweight in 3-year-old black South African children 505
Table 3 Binary logistic regression analysis on both stunted and between birth and 1 year were born to older mothers
overweight children at age 3 years: odds ratio (95% confidence (x 2 ¼ 11.80, P ¼ 0.019). Catch-down growth between 1
interval)
and 3 years was associated with households having
Parameter Stunted (n ¼ 47) Overweight (n ¼ 43) electricity (x 2 ¼ 7.54, P ¼ 0.023), while in those who
Birth weight 0.77 (0.16– 3.65) 3.24 (0.87 – 8.96) showed catch-up from birth to 1 years, most were born to
Birth length 0.87 (0.21– 5.21) 0.83 (0.24 – 3.28) mothers with low parity (i.e. three children or less)
Birth Ponderal Index 1.08 (0.96– 1.21) 0.95 (0.82 – 1.09)
Birth HAZ
(x 2 ¼ 6.08, P ¼ 0.048).
, 2 2SD 0.55 (0.13– 2.39) 4.03 (0.12 – 14.2)
Normal 1.00 1.00 Discussion
Birth WHZ
, – 2SD 0.73 (0.24– 2.22) 5.93 (1.82 – 9.38)
Normal 1.00 1.00 The major findings of this study were that: (1) half of the
1-year weight 3.06 (0.45– 8.01) 6.92 (2.03 – 16.23) children were stunted (48%), whereas less than 10% were
1-year length 0.71 (0.59– 0.84) 0.25 (0.06 – 0.97)
1-year Ponderal Index 0.81 (0.69– 0.95) 0.29 (0.08 – 1.08) underweight and few were wasted; (2) just under half were
1-year HAZ overweight (22%) and obese (24%); and (3) just under a fifth
, 2 2SD 2.00 (0.50– 8.11) 1.32 (0.23 – 7.78)
were both overweight and stunted (19%). The frequency of
Normal 1.00 1.00
1-year WHZ stunting reported herein is considerably higher than the
. 2SD 0.25 (0.04– 1.57) 4.28 (0.38 – 7.92) national rates in children aged 1–3 years as well as rates
Normal 1.00 1.00
Stunted at 3 years
reported for Limpopo Province in which this study took
Yes 0.64 (0.19 – 2.16) place29. This finding may imply a greater degree of poverty
No 1.00 with poorer infrastructure in the study region than in other
Overweight at 3 years
Yes 1.91 (0.69– 5.33) regions of South Africa. As stunting was already present in
No 1.00 35% of the study children at 1 year of age15, the mechanisms
Maternal age underlying this are more likely to be chronic than acute.
, 20 years 0.72 (0.18– 2.88) 6.17 (1.41 – 7.13)
. 30 years 0.71 (0.18– 2.79) 1.87 (0.56 – 6.24) The highest risk for stunting amongst the sociodemo-
20 – 29 years 1.00 1.00 graphic factors evaluated was household size, suggesting
Maternal length 0.60 (0.24– 4.35) 2.65 (0.61 – 5.32)
that there is more competition for available food when the
Parity
4þ 1.60 (0.54– 4.71) 0.77 (0.17 – 3.51) household is large30. Another factor which conferred a
0–3 1.00 1.00 high risk was having a student mother, which may be
Marital status
Married 0.66 (0.20– 2.19) 1.13 (0.21 – 6.21)
explained by the resulting reduced family income and
Single 1.00 1.00 individual care and attention given to the child, as he/she
Maternal education has to be left with another carer during the day31. The
Primary 1.69 (0.31– 9.31) 0.97 (0.16 – 6.04)
Secondary and higher 1.00 1.00 congruence of these findings with previous reports8,9,32
Maternal occupation highlights their relevance to the development of stunting
Working 3.83 (0.93– 5.74) 3.35 (2.88 – 9.06) in developing countries. Assessment of length at 1 year of
Student 3.87 (1.16– 12.87) 3.61 (0.63 – 7.57)
Unemployed 1.00 1.00 age may be important for health providers since this can
Main caregiver be used to identify stunted children.
Mother 1.03 (0.21– 4.92) 8.13 (1.23 – 13.94)
In common with other studies, the stunted children in
Other 1.00 1.00
House type this cohort had significantly shorter mothers than those
Brick 0.89 (0.27– 3.01) 0.52 (0.14 – 1.19) who were not stunted5,8,10. Although this may constitute
Other 1.00 1.00
No. in household
an underlying genetic factor, it is more likely that the
5–8 3.06 (0.71– 13.24) 0.92 (0.14 – 5.93) intergenerational effect reflects long-standing maternal
9þ 8.43 (1.64– 13.21) 1.61 (0.20 – 12.72) undernutrition and possibly even maternal restraint to
0–4 1.00 1.00
No. of children aged , 5 years growth during their own in utero period.
3þ 1.06 (0.22– 5.15) 2.62 (0.63 – 10.85) A relatively small proportion of the stunted children
0–2 1.00 1.00 (17%) were underweight, whilst 31% were overweight
Child sees mother daily
Yes 0.62 (0.07– 5.79) 0.03 (0.002– 0.54) when defined by WHZ .1SD and 46% when using the
No 1.00 1.00 IOTF cut-offs25, implying that stunting is associated with
Father in same household
both under- and overnutrition. Indeed the high prevalence
Yes 0.44 (0.07– 2.79) 1.13 (0.31 – 4.12)
No 1.00 1.00 of overweight associated with stunting raises many
questions about the aetiology and outcomes of both
HAZ – height-for-age Z-score; SD – standard deviation; WHZ – weight-
for-height Z-score. under- and overnutrition. The phenomenon of childhood
obesity accompanying stunting has been reported in other
(x 2 ¼ 8.91, P ¼ 0.012) and being overweight at 3 years developing countries as well as in developed ones13,33.
(x 2 ¼ 6.39, P ¼ 0.041). The observation in this cohort that just under half of the
With respect to change in SD scores for WHZ, a high children were overweight and/or obese is of great
proportion of children who showed catch-up growth concern. Despite the prevalence of overweight being

Downloaded from http:/www.cambridge.org/core. IP address: 36.83.180.13, on 08 Nov 2016 at 21:25:16, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1079/PHN2005786
 506 RL Mamabolo et al.
Table 4 Multinomial logistic regression of some risk factors on child nutritional status: odds ratio (95% confidence interval)

Parameter Stunted (n ¼ 47) Overweight (n ¼ 43) Overweight and stunted (n ¼ 31)

Birth WHZ 0.81 (0.22–3.05) 0.37 (0.09– 1.57) 0.59 (0.13–2.74)

1-year weight 3.71 (0.48–8.37) 2.39 (1.96– 4.18) 8.97 (1.09–13.51)
1-year length 0.41 (0.17–0.97) 0.53 (0.09– 3.15) 0.14 (0.02–0.87)
1-year Ponderal Index 0.52 (0.24–1.10) 0.72 (0.13– 3.95) 0.21 (0.04–1.08)
Maternal age
, 20 years 0.65 (0.09–4.41) 2.20 (0.34– 14.29) 1.04 (0.13–8.56)
. 30 years 0.79 (0.19–3.22) 0.45 (0.09– 2.23) 0.83 (0.16–4.40)
20–29 years 1.00 1.00 1.00
Maternal occupation
Working 3.23 (0.46–5.42) 17.87 (8.24– 38.78) 10.97 (1.41–14.89)
Student 18.21 (9.46–34.74) 4.13 (0.39– 9.89) 7.45 (3.20–17.34)
Unemployed 1.00 1.00 1.00
Main caregiver
Mother 0.93 (0.11–8.31 3.74 (0.29– 8.71) 4.43 (0.38–15.34)
Other 1.00 1.00 1.00
Household size
5 –8 3.64 (0.46–8.74) 5.35 (0.72– 9.81) 5.23 (0.31–8.69)
9þ 6.95 (0.62–18.12) 7.61 (0.70– 12.21) 5.72 (2.70–12.10)
0 –4 1.00 1.00 1.00
Child sees mother daily
Yes 0.05 (0.01–2.81) 0.19 (0.02– 0.38) 0.08 (0.01–4.57)
No 1.00 1.00 1.00

WHZ – weight-for-height Z-score.

subsequent food availability for the child in the develop-

Table 5 Percentage of children by change in standard deviation
(SD) score at various time frames
ment of overweight in this setting40,41.
Previous studies on feeding practices have reported a
Change in SD score dose-dependent protective effect of breast milk against
Catch-up No change Catch-down obesity. Although most children in the present study were
breast-fed for a year or more, a phenomenon well
HAZ
0 –1 years 16.5 25.2 58.3 documented in South Africa19,42,43, only 10% of the
1 –3 years 13.1 41.5 45.4 children in this cohort were exclusively breast-fed for the
0 –3 years 4.2 29.2 66.7 first 3 months of life and 4% for 6 months15; hence it is
WHZ
0 –1 years 74.1 17.3 7.9 difficult to comment on the role of exclusive breast-
1 –3 years 44.6 37.7 17.7 feeding in the aetiology of overweight and/or obesity.
0 –3 years 80.6 14.6 4.9 However, the majority of infants in this study were
HAZ – height-for-age Z-score; WHZ – weight-for-height Z-score. introduced to complementary foods at a very early age
(between 0 and 3 months). Most of these foods were high
comparable to that of other developing countries, the in energy and carbohydrates and poor in protein and
obesity levels are higher34, even in comparison with other essential nutrients required for linear growth. Hence it is
studies in South African children, which range from 7.5 to possible to understand why obesity and stunting could be
20.1%16,21. The wide variation in South African prevalence the outcomes15,18,42,44.
rates is possibly due to inequalities in socio-economic Baughcum et al.37 reported that a high percentage of
status, since it has been shown that some provinces, mothers with a poor education did not consider their
including Limpopo, are amongst the poorest. Limpopo overweight children to be overweight. It is believed that
Province has a high prevalence of illiteracy, high rates of mothers have their own culturally based and influenced
unemployment, poor sanitation facilities, low levels of definitions of obesity36. Thus with the cultural belief that
electrification and high household density, which have all ‘bigger is healthier’ in the Pedi group studied36,38, the
been documented as being related to stunting35. mother of a lean healthy baby might feel that her child is not
Previous research has shown that overweight is an getting sufficient food and, as such, change her feeding
outcome of sociodemographic status, maternal feeding practices. This may result in her overfeeding the child, a
practices and cultural beliefs as to what is normal1,9,36 – 39. phenomenon also reported by Kagamimori et al.45.
‘Young’ mothers, ‘working’ mothers and ‘primary care- As opposed to adults, where obesity is an immediate
giver’ mothers all conferred an increased risk for having an risk factor for most chronic diseases, in children it rarely
overweight child. However, having a working mother was poses an immediate and serious health threat. In effect, its
the only factor that remained significantly associated with short-term consequences are often psychosocial and
overweight in the multinomial analysis, thus emphasising include outcomes such as low self-esteem, poor peer
the likely importance of increased family income and acceptance and low participation in social and sporting

Downloaded from http:/www.cambridge.org/core. IP address: 36.83.180.13, on 08 Nov 2016 at 21:25:16, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1079/PHN2005786
 Stunting and overweight in 3-year-old black South African children 507
activities46. Available evidence suggests that the influence References
of childhood obesity on adult status rises with age, with a
low predictive value in children below 2 years of age, a 1 Frongillo EA Jr, de Onis M, Hanson KMP. Socioeconomic and
demographic factors are associated with worldwide patterns
significant predictor at 3–5 years, and an even a stronger
of stunting and wasting of children. Journal of Nutrition
one at older ages47,48. 1997; 127: 2302 –9.
The risk factors for the coexistence of stunting and 2 Bourne LT, Lambert EV, Steyn K. Where does the black
overweight in the children in this study were having an population of South Africa stand on the nutrition transition?
Public Health Nutrition 2001; 5: 157 – 62.
increased weight and shorter length at 1 year, having more
3 Monteiro CA, Conde WL, Popkin BM. Is obesity replacing or
than nine persons in the household, and having a working adding to undernutrition? Evidence from different social
mother or a mother who was a student. The association classes in Brazil. Public Health Nutrition 2002; 5: 105 – 12.
with increased weight and shorter length is likely due to the 4 Adair LS, Guilkey DK. Age specific determinants of stunting
in Filipino children. Journal of Nutrition 1997; 127: 314– 20.
quality of the diet consumed (rich in carbohydrates and 5 Saleemi MA, Ashraf RN, Mellander L, Zaman S. Determinants
poor in proteins)18,43,44,49,50. It is likely that the children of stunting at 6, 12, 24 and 60 months and postnatal linear
have a sub-minimum protein intake, which results in failure growth in Pakistani children. Acta Paediatrica 2001; 90:
of linear growth and the high stunting levels already 1304– 8.
6 Vella V, Tomkins A, Borghesi A, Migliori GB, Oryem VY.
evident at 1 year. Should the diet subsequently remain a Determinants of stunting and recovery from stunting in
high-energy, high-carbohydrate one, it is likely that excess Northwest Uganda. International Journal of Epidemiology
weight gain will take place in addition to sub-optimal linear 1994; 23: 782– 6.
growth; with the latter still not being optimal. 7 Zamiliah MM, Kandian M, Lin KG, Siong TE. Socioeconomic
profile and nutritional status of children in rubber
The seemingly contradictory association between smallholdings. Asia Pacific Journal of Clinical Nutrition
maternal occupation and the combination of stunting and 2002; 11: 133 – 41.
overweight is likely due to an oversimplification of the 8 Delpeuch F, Traissac P, Martin-Prevel Y, Massamba JP,
interrelationship between the various demographic and Maire B. Economic crisis and malnutrition: socioeconomic
determinants of anthropometric status of preschool children
household factors studied as markers of poverty. and their mothers in an African urban area. Public Health
According to Dietz51, there are four ‘critical periods’ for Nutrition 2000; 3: 39 –47.
development of obesity in children: gestation, early 9 Ukwuani FA, Suchindran CM. Implications of women’s
infancy, period of adiposity rebound (5 –7 years) and work for child nutritional status in sub-Saharan Africa: a
case study of Nigeria. Social Science & Medicine 2003; 56:
adolescence. However, it has been shown that the age 2109– 21.
group 0–3 years is the most vulnerable to stunting, which 10 Hernandez-Diaz S, Peterson KE, Dixit S, Hernandez B,
may increase risks of being overweight later in Parra S, Barquera S, et al. Association of maternal short
stature with stunting in Mexican children: common genes vs
life15,16,41,52. Catch-up (catch-down) growth is a phenom-
common environment. European Journal of Clinical
enon characterised by an increased (decreased) growth Nutrition 1999; 53: 938 – 45.
velocity in height and/or weight following the removal 11 de Onis M, Frongillo EA, Blossner M. Is malnutrition
(addition) of some constraints on normal growth. Rapid declining? An analysis of changes in levels of child
malnutrition since 1980. Bulletin of the World Health
catch-up growth in weight during infancy has been linked Organization 2000; 78: 1222 –33.
to risk factors for overweight and obesity26,53,54. We have 12 Hautvast JLA, Tolboom JJM, Kafwembe EM, Musonda RM,
previously shown that stunting and overweight in this Mwanakasale V, van Staveren WA, et al. Severe linear growth
group of children starts from an early age, 0–1 year15. retardation in rural Zambian children: the influence of
biological variables. American Journal of Clinical Nutrition
The results from this study highlight the importance of 2000; 71: 550– 9.
evaluating anthropometric status in terms of both stunting 13 Faber M, Jogessar VB, Benade AJS. Nutritional status and
and overweight as well as the combination of these two dietary intakes of children aged 2 –5 years and their
factors. In the past the focus has been on one or the other caregivers in a rural South African community. Inter-
national Journal of Food Sciences and Nutrition 2001; 52:
and the combination has remained largely uninvestigated. 401 –11.
It is important that healthcare providers realise the 14 Monyeki KD, van Lenthe FJ, Steyn NP. Obesity; does it occur
importance of normal weight and length being attained in African children in a rural community in South Africa?
at 1 year of age, since these in turn predict nutritional International Journal of Epidemiology 1999; 28: 287 – 92.
15 Mamabolo RL, Alberts M, Mbenyane GX, Steyn NP,
status at 3 years of age. Nthangeni NG, Delemarre-van de Waal HA, et al. Feeding
practices and growth of infants from birth to 12 months in
Acknowledgements the central region of the Limpopo Province of South Africa.
Nutrition 2004; 20: 327– 33.
16 Steyn NP, Labadarios D, Maunder E, Nel J, Lombard C;
This study was supported by grants from the Institute for Directors of the National Food Consumption Survey.
Research in Extramural Medicine (EMGO), Vrije Univer- Secondary anthropometric data analysis of the National
sity, Amsterdam (The Netherlands). We would like to Food Consumption Survey in South Africa: the double
burden. Nutrition 2005; 21: 4 – 13.
thank Ms MA Makwela and Mr RP Mamabolo (fieldwor-
17 Patel DN, Pettifor JM. Malnutrition in South Africa.
kers), and also the nursing staff at the nine clinics, for their South African Journal of Food Science and Nutrition 1992;
assistance in conducting this study. 4: 22.

Downloaded from http:/www.cambridge.org/core. IP address: 36.83.180.13, on 08 Nov 2016 at 21:25:16, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1079/PHN2005786
 508 RL Mamabolo et al.
18 Steyn NP, Badenhorst CJ, Nel JH, Ladzani R. Breast-feeding childhood. Journal of Developmental and Behavioral
and weaning practices of Pedi mothers and the dietary Pediatrics 2001; 22: 391 – 408.
intakes of their preschool children. South African Journal of 37 Baughcum AE, Chamberlin LA, Deeks CM, Powers SW,
Food Science and Nutrition 1993; 5: 10 –13. Whitaker RC. Maternal perceptions of overweight preschool
19 Dannhauser A, Bester CJ, Joubert G, Badenhorst PN, Slabber children. Pediatrics 2000; 106: 1380– 5.
M, Badenhorst AM, et al. Nutritional status of preschool 38 Kruger R, Gericke GJ. A qualitative exploration of rural
children in informal settlement areas near Bloemfontein, feeding and weaning practices, knowledge and attitudes on
South Africa. Public Health Nutrition 2000; 3: 303 – 12. nutrition. Public Health Nutrition 2003; 6: 217 –23.
20 Fourie J, Steyn K. Chronic Diseases of Lifestyle in South 39 Faith MS, Keller KL, Matz P, Johnson SL, Lewis R, Jorge MA,
Africa. MRC Technical Report. Cape Town: Medical et al. Project Grow-2-Gether: a study of the genetic and
Research Council, 1995. environmental influences on child eating and obesity. Twin
21 Bourne LT, Langenhoven ML, Steyn K, Jooste PL, Laubscher Research 2002; 5: 472– 5.
JA, Bourne DE. Nutritional status of 3 – 6 year-old children in 40 Engel PL. Influences of mother’s and father’s income on
the Cape Peninsula. The BRISK Study. East African Medical child’s nutritional status in Guatemala. Social Science &
Journal 1994; 7: 695 – 702. Medicine 1993; 37: 1303– 12.
22 Yanovski JA, Yanovski SZ. Treatment of pediatric and 41 Popkin B, Richards M, Monteiro C. Stunting is associated
adolescent obesity. Journal of the American Medical with overweight in children of four nations that are
Association 2003; 289: 1851 –3. undergoing the nutrition transition. Journal of Nutrition
23 Mamabolo RL, Alberts M, Steyn NP, Delemarre-van de Waal 1996; 126: 3009– 16.
HA, Nthangeni NG, Levitt NS. Evaluation of the effectiveness 42 Delport SD, Becker PJ, Bergh A. Growth, feeding practices
of iron and folate supplementation during pregnancy in a and infections in black infants. South African Medical
rural area of Limpopo province. South African Journal of Journal 1997; 87: 57– 61.
Clinical Nutrition 2004; 17: 15 –21. 43 Steyn NP, Nel JH, Kunnene E, Tichelaar HY, Oelofse A,
24 Hamill PV, Dritz TA, Johnson CL, Reed RB, Roche AF. NCHS Prinsloo JF, et al. Differences between underweight and
Growth Curves for Children Birth – 18 years. DEHW normal weight rural preschool children in terms of infant
Publications No. (PHS) 78-1650, Series 11, No. 165. feeding practices and socio-economic factors. South African
Washington, DC: DEHW Publishers, 1977. Medical Journal 1998; 5: 641 – 6.
25 Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a 44 Ladzani R, Steyn NP, Nel JH. Infant feeding practices of Pedi
standard definition for child overweight and obesity women in six semi-rural areas of Northern Province. South
worldwide: international survey. British Medical Journal African Journal of Epidemiology and Infection 1998; 13:
2000; 320: 1 – 6. 63 –5.
26 Ong KK, Ahmed ML, Emmett PM, Preece MA, Dunger
45 Kagamimori S, Yamagami T, Sokejima S, Numata N,
DB, The ALSPAC Study Team. Association between
Handa K, Nanri S, et al. The relationship between
postnatal catch-up growth and obesity in childhood:
lifestyle, social characteristics and obesity in 3-year-old
prospective cohort study. British Medical Journal 2000;
Japanese children. Child: Care, Health and Development
320: 967– 71.
1999; 25: 235 –46.
27 Centers for Disease Control and Prevention (CDC). Epi-Info
46 Dietz WH. Health consequences of obesity in youth:
2002, A Database and Statistics Program for Public Health
childhood predictors of adult disease. Pediatrics 1998;
Professionals. Atlanta, GA: CDC, 2003.
101: 518– 25.
28 SPSS Inc. SPSS for Windows Base System User’s Guide,
Release 12.0. Chicago, IL: SPSS Inc., 2003. 47 Serdula MK, Ivery D, Coates RJ, Freedman DS, Williamson
29 Labadarios D, Steyn N, Maunder E, MacIntyre U, Swart R, DF, Byers T. Do obese children become obese adults? A
Gericke G, et al. The National Food Consumption Survey review of the literature. Preventive Medicine 1993; 22:
(NFCS): Children aged 1 –9 years, South Africa, 1999. 167 – 77.
Stellenbosch: Stellenbosch University Press, 2000. 48 Whitaker RC, Wright JA, Pepe MS, Siedel KD, Dietz WH.
30 James WPT, Ferro-Luzzi A, Sette S, Mascie-Taylor CGN. The Predicting obesity in young adulthood from childhood and
potential use of maternal size in priority setting when parental obesity. New England Journal of Medicine 1997;
combating childhood malnutrition. European Journal of 337: 869– 73.
Clinical Nutrition 1999; 53: 112 – 9. 49 Steyn NP, Badenhorst CJ, Nel JH. The nutritional status of
31 Lei DLM, Chaves SP, Paes AT, Escuder MM, Ribeiro AB, Freire Pedi preschool children in two rural areas of Lebowa.
RD, et al. Risk of linear growth retardation during the first South African Journal of Food Science Nutrition 1992; 4:
two years of life: a new approach. European Journal of 24 –8.
Clinical Nutrition 1999; 53: 456 – 60. 50 Badenhorst CJ, Steyn NP, Jooste PL. Nutritional status of Pedi
32 Shah SM, Selwyn BJ, Luby S, Merchant A, Bano R. Prevalence school children aged 6 –14 years in two rural areas in
and correlates of stunting among children in rural Pakistan. Lebowa: a comprehensive nutritional survey of dietary
Pediatrics International 2003; 45: 49 –53. intake, anthropometric and clinical measurements. South
33 Sanchez-Castillo CP, Lara JL, Villa AR, Aguirre J, Escobar M, African Journal of Food Science and Nutrition 1993; 5:
Gutierrez H, et al. Unusually high prevalence rates of obesity 112 – 5.
in four Mexican rural communities. European Journal of 51 Dietz WH. Critical periods in childhood for the development
Clinical Nutrition 2001; 55: 833 – 40. of obesity. American Journal of Clinical Nutrition 1994; 59:
34 Martorell R, Khan LK, Hughes ML, Grummer-Strawn LM. 955 – 9.
Overweight and obesity in preschool children from 52 Popkin BM. The nutrition transition and obesity in the
developing countries. International Journal of Obesity and developing world. Journal of Nutrition 2001; 131: 871S – 3S.
Related Metabolic Disorders 2001; 24: 959 – 67. 53 Law CM, Shiell AW, Newsome CA, Syddall HE, Shinebourne
35 Zere E, McIntyre D. Inequities in under-five child EA, Fayers PM, et al. Fetal, infant, and chzildhood growth
malnutrition in South Africa. International Journal for and adult blood pressure: a longitudinal study from birth to
Equity in Health 2003; 2: 7. 22 years of age. Circulation 2002; 105: 1088 –92.
36 Baughcum AE, Powers SW, Johnson SB, Chamberlin LA, 54 Cameron N, Pettifor J, De Wet T, Norris S. The relationship of
Deeks CM, Jain A, et al. Maternal feeding practices and rapid weight gain in infancy to obesity and skeletal maturity
beliefs and their relationships to overweight in early in childhood. Obesity Research 2003; 1: 457 – 60.

Downloaded from http:/www.cambridge.org/core. IP address: 36.83.180.13, on 08 Nov 2016 at 21:25:16, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
http://dx.doi.org/10.1079/PHN2005786