DOI: 10.1111/mcn.

12347

Original Article
Understanding the double burden of malnutrition in food
insecure households in Brazil
Muriel Bauermann Gubert*,‡, Ana Maria Spaniol*, Ana Maria Segall-Corrêa† and
Rafael Pérez-Escamilla‡
*
Department of Nutrition, University of Brasilia, Brasília, Brazil, † Department of Collective Health, University of Campinas, Campinas, Brazil, and ‡ Department of
Chronic Disease Epidemiology, Yale School of Public Health, New Haven, Connecticut, USA

Abstract

Household food insecurity (HFI) has been associated with both obesity among mothers and undernutrition among
children. However, this association has not been well investigated in mother/child pairs living in the same household.
The objective of this study was to examine the relationship of coexistence of maternal overweight and child stunting
with HFI in Brazil. We conducted secondary data analyses of the 2006 Brazilian National Demographic and Health
Survey. We analyzed the nutritional status of 4299 pairs of 15–49-year-olds mothers and their children under 5 years
of age. The double burden of malnutrition (DBM) was defined as the presence of an overweight mother and a
stunted child in the same household. HFI was measured with the Brazilian HFI Measurement Scale. The association
between DBM and HFI was examined with hierarchical multivariable logistic regression analyses. Severe HFI was
associated with DBM after adjusting for macroeconomic and household level socio-economic and demographic var-
iables (Adjusted OR: 2.65 – CI: 1.17–8.53). Findings suggest that policies and programmes targeting HFI are needed
to prevent the coexistence of child chronic undernutrition and maternal overweight/obesity in the same household.
These investments are likely to be highly cost-effective as stunting has been identified as one of the major risk factors
for poor child development and adult overweight/obesity and a strong risk factor for the development of costly
chronic diseases including type 2 diabetes and cardiovascular disease.

Keywords: food insecurity, dual-burden, overweight, stunting, malnutrition, demographic health survey.

Correspondence: Muriel Bauermann Gubert, Department of Nutrition, University of Brasília, Brazil. E-mail: murielgubert@gmail.com

Introduction undernutrition (Ali et al., 2013) and less consistently

with overweight/obesity (Franklin et al., 2012; Kac
Food security is a complex multidimensional process et al., 2012a; Metallinos-Katsaras et al., 2012).
that covers the entire food chain, determined by aspects Studies reveal that women experiencing HFI have a
such as access and biological utilization of food (United higher prevalence of overweight and obesity, especially
Nations, 2015). Household food insecurity (HFI) has those experiencing mild and severe HFI (Franklin
been associated with both undernutrition and obesity et al., 2012; Kac et al., 2012b; Kac et al., 2012a). Over-
(Shamah-Levy et al., 2014), which is expected because weight in these women may be related to short stature
both conditions result from poor eating habits and are caused by energy metabolism abnormalities linked to
likely to be related to the process of nutritional transi- stunting in childhood (Hoffman et al., 2000), enhanced
tion observed in Brazil and worldwide (Ghattas, food reinforcement resulting from earlier food depriva-
2014). In this context, overweight and obesity continue tion and restricted food choice even when food later
to be associated with poverty and other unfavourable becomes plentiful (Lin et al., 2013) or low levels of
social and environmental conditions such as HFI physical activity (To et al. 2014).
(Frongillo & Bernal, 2014), especially among women In children, the risk of stunting increases as the level
(Franklin et al., 2012; Santos, 2013). Among children, of food and nutritional insecurity becomes more severe
HFI has been consistently associated with (Ali et al., 2013; Bernal et al., 2014; Mutisya et al.,

© 2016 John Wiley & Sons Ltd Maternal & Child Nutrition (2017), 13, e12347 1 of 9
2 of 9 M. B. Gubert et al.

2015). Thus, it is plausible for overweight mothers and analytical sample for this study included the remaining
stunted children to be living together in the same 4299 mother/child pairs.
household (Conde & Monteiro, 2014; Abdullah, 2015). Household food insecurity was assessed with the
Although previous studies have examined the associ- Brazilian HFI Measurement Scale (Escala Brasileira
ation between HFI and the nutritional status of women de Insegurança Alimentar, EBIA). EBIA consists of
and children, the association between HFI and the 14 questions and the score is calculated by the sum of
double burden of malnutrition (DBM) requires further affirmative answers to EBIA items. As recommended,
investigation, especially in settings like Brazil, where households were then classified into four mutually
the nutrition transition is very advanced (Conde & exclusive categories: secure (score = 0), mildly insecure
Monteiro, 2014). The 2006 Brazilian National Demo- (score = 1 to 5), moderately insecure (score = 6 to 9)
graphic and Health Survey (DHS) (Ministerio da and severely insecure (score = 10 to 14) (Instituto
Saude, 2009) provides a unique opportunity for analyz- Brasileiro de Geografia e Estatística., 2014; Segall-
ing this relationship. Therefore, the specific objective of Corrêa et al., 2014).
this study was to examine the association between HFI Each mother and child were grouped into pairs and
severity and risk of DBM in a nationally representative classified in terms of the presence or absence of
cross-sectional sample of Brazilian households. DBM. The pair was considered as having DBM when
the mother had a body mass index equal to or higher
than 25 kg m 2 and the child’s H/A was lower than 2
Subjects and methods z-scores. In our sample, 140 pairs (2.6%) had DBM,
i.e., the coexistence of maternal overweight and child
Secondary data analyses were performed with national undernutrition (low H/A) in the same household.
data from the 2006 Brazilian DHS (Demographic Because of their exceedingly low prevalence, four pairs
Health Survey). The survey was conducted in civilian (0.09%) consisting of a malnourished (underweight)
households randomly selected using complex sampling mother and an overweight child were excluded from
procedures, involving the stratification of Brazilian cen- the analysis. The decision of using maternal overweight
sus tracks (primary sampling units) and the selection of instead of obesity was made based on the low preva-
households within each track (secondary sampling lence of DBM when using obesity as the criteria
units). Households were distributed across all the five (0.9%). We included all children under 5 years old
country’s macro-regions and areas (urban and rural) allowing for the inclusion of more than one child in
(Ministerio da Saude, 2009). the same house.
Our study evaluated the association of interest in a The 2006 Brazilian DHS protocol was approved by
representative sample of women of childbearing age the Research Ethics Committee of the DST/AIDS Re-
(15–49 years) and their children under 5 years of age ferral and Training Center of the Department of Health
who participated in the 2006 Brazilian DHS. The origi- of the State of São Paulo, Brazil. All mothers signed the
nal sample included 15 575 women and 4817 children consent form before the interview.
under 5 living with their mothers. The initial analytical
sample included 4323 mother/child pairs with complete
Statistical analyses
data regarding maternal-child anthropometry and HFI.
Nutritional status was assessed using maternal body Statistical analyses were performed with the Statistical
mass index, and children’s height-for-age (H/A) z-score Package for the Social Sciences (SPSS) software (Ver-
(World Health Organization, 2006). In our study, sion 20.0) using the Complex Samples module. Data
households were excluded if the target children or were weighted for unequal sampling probabilities: a
mother had implausible anthropometric data (children complex three-level design with stratification was ap-
H/A < or > 4 z-scores (excluded 88 cases, 2.0%) and plied to the sampled subjects included in the survey.
maternal IMC less than 13 kg m 2 or higher than The primary units were constituted by Brazilian census
55 kg m 2 (excluded 24 cases, 0.6%). The final sectors. The secondary units were sector clusters, and

© 2016 John Wiley & Sons Ltd Maternal & Child Nutrition (2017), 13, e12347
 Dual-burden malnutrition and food insecurity 3 of 9

the third sampling units were the homes. The sample the risk factors for stunting and maternal obesity to be
was comprised by the universe of civilian households able to compare them against the risk factors for
including favelas or shanty towns and is representative DBM. All statistical analyses used sampling weights
of all five Brazilian geographic regions, including urban to take into account the complex sampling design of
and rural areas. the survey. For the DBM and child stunting analyses,
Descriptive characteristics are presented as percent- the children sampling weights were used, and for the
age and confidence intervals for categorical variables. maternal overweight/obese analyses, the women’s sam-
Pearson’s chi-square test was used to test nutritional pling weights were used.
status (maternal overweight and stunting) and DBM
differences across HFI, socio-economic and demo-
graphic variables (child’s sex, maternal race/skin Results
colour, Brazilian macro-region, source of water, urban
Sample characteristics
or rural area and educational level of the head of the
household) categories. Adjusted odds ratios and their Almost half of the households (48.4%) experienced
corresponding 95% CIs were estimated by logistic HFI during the 3 months preceding the interview.
regression analyses adjusting for sex of the child, Whereas 42.6% of the mothers were overweight or
Brazilian regions, household food security status and obese, 7.2% of their children were stunted. Approxi-
education level of the head of household. mately, a third of the households did not have access
To better understand the association of interest, we to a public water supply and/or to adequate sanitation
conducted hierarchical multivariable logistic regression (septic tank or sanitary sewer). The educational level
analysis including DBM as a binary outcome and HFI of the head of household was below 4 years of schooling
as the key independent variable. Variables with bivari- in 37.3% of households (Table 1).
ate associations at the P < 0.20 level were included in Among the women classified as overweight, 37.0%
the multivariate analyses to minimize the chances of a had short stature (height ≤150 cm) (Gudmundsson
type II error. Models were adjusted first for macroeco- et al., 2005), and 63.0% had normal height. Further-
nomic variables then for household/individual socio- more, 52.7% had normal weight and height. Similarly,
economic and demographic variables and lastly for 4% of overweight children were stunted. Additionally,
both sets of variables. The macroeconomic covariates 92.4% of children with normal height had also normal
included in the final model were Brazilian macro- weight (data not shown in tables).
region and area of residence (urban or rural). The
socio-economic and demographic covariates were
Bivariate analyses
maternal self-reported skin colour/race (black/brown,
yellow/indigenous and white) maternal parity and age A statistically significant association was observed
and educational level of the head of the household. between DBM and both the independent variable and
We also conducted multivariable analyses to identify all covariates analyzed (Table 2). The prevalence of

Key messages
• In Brazil, severe household food insecurity was associated with the double burden malnutrition in the same
household.
• Life course policies and programmes targeting household food insecurity are needed to prevent the double burden
malnutrition.
• Household food insecurity should be included as a key evaluation indicator of maternal-child nutrition programmes.
• Prospective research is needed to better understand the mechanisms explaining the simultaneous risk for maternal
overweight/obesity and child stunting and to identify modifiable factors that affect this relationship.

© 2016 John Wiley & Sons Ltd Maternal & Child Nutrition (2017), 13, e12347
4 of 9 M. B. Gubert et al.

Table 1. Description of the study population. Brazilian National Demo- DBM was associated with HFI severity level following
graphic and Health Survey 2006 a dose-response pattern, and it was more prevalent in
Variables n % households: (1) with a male child; (2) located in the
Northern region; and (3) where the head had lower
Double burden of malnutrition in mother/child pairs educational levels. The prevalence of stunting was also
No 4159 97.4
associated with the degree of HFI severity following a
Yes (overweight/stunting) 140 2.6
Child’s nutritional status * dose–response relationship, and it was also higher for
Height/age male child, children living in the North of Brazil and
Stunting 392 7.2
among children living in household where the head
Normal 3907 92.8
Body mass index/age had lower educational levels. There were no associa-
Thinness 114 1.9 tions between HFI and any of the socio-demographic
Normal 3991 94.7 covariates with maternal overweight.
Overweight 194 3.4
Maternal nutritional status
The crude odds for DBM were 4.56 times higher
Underweight 192 4.6 among households with severe food insecurity com-
Normal weight 2315 52.8 pared with their food secure counterparts. DBM was
Overweight 1170 28.8
also more likely to occur in households with a male
Obesity 622 13.8
Sex of the child child (OR: 1.86). Households in the Northern were al-
Male 2202 52.2 most three times more likely than those in the Midwest
Female 2097 47.8
region (OR: 2.84) to experience DBM. Similar associa-
Maternal race/skin colour
White 1438 32.9 tions were found with regard to risk of DBM among
Black or brown 2588 61.5 lower in comparison with higher maternal education
Yellow or indigenous 224 5.6 level (OR: 2.97) (Table 3).
Macro-region
Northern 962 10.7
In order to understand if including more than one
Northeast 862 28.6 child per household affected the findings, we replicated
Southeast 847 41.1 the analyses using only the youngest child followed by
Southern 785 12.1
using only the oldest one in the 12.5% of households
Midwest 843 7.5
Household food security status† where siblings were included. The association between
Food security 2096 51.6 DBM and severe food insecurity remained, but it
Mild food insecurity 1184 28.5
tended to be underestimated for the youngest child
Moderate food insecurity 661 14.5
Severe food insecurity 358 5.4 (OR: 2.45 for severe insecurity compared with those
Adequate sanitation with food security) and overestimated for the oldest
Yes 2118 66.3 child (OR: 5.16) (data not shown in tables).
No 1669 33.7
Adequate water supply
The crude odds for stunting were 2.58 times higher
Yes 2956 69.5 among households with severe food insecurity com-
No 1308 30.5 pared with their food secure counterparts. Stunting was
Household location
also more likely to occur in male children (OR: 1.47)
Urban 2819 80.9
Rural 1480 19.1 and in children living in households located in the
Educational level of the head of the household Northern (OR: 2.64) compared with those in the Mid-
4 years of education or less 1744 37.3
west region. The crude odds for stunting were three times
5 to 8 years of education 1279 29.9
9 years of education or more 1153 32.8 higher if the head of household had less than 4 years of
education compared with those with 4 years or more.
*This variable refers to the nutritional status of all the children
under five included in the study. †As assessed by the Brazilian
Household Food Insecurity Measurement Scale (Escala Brasileira Multivariable analyses
de Medida Domiciliar de Insegurança Alimentar, EBIA). Raw sam-
ple sizes reported, sampling weights used to compute prevalence The associations detected in the bivariate analyses
estimates. remained significant even after adjustment for each of

© 2016 John Wiley & Sons Ltd Maternal & Child Nutrition (2017), 13, e12347
 Dual-burden malnutrition and food insecurity 5 of 9

Table 2. Maternal overweight, children stunting and double burden of malnutrition according to associated factors. Brazilian National Demographic and
Health Survey 2006

Maternal overweight Children stunted Double burden of malnutrition

Variables Yes No P-value* Yes No P-value* Yes No P-value*

(%) (%) (%) (%) (%) (%)

Sex of the child

Male 42.0 58.0 0.640 8.5 91.5 0.028 3.3 96.7 0.021
Female 43.3 56.7 5.9 94.1 1.8 98.2
Maternal race/skin colour
White 44.3 55.7 0.218 6.6 93.4 0.577 1.8 98.2 0.363
Black or brown 42.9 57.1 7.7 92.3 2.8 97.2
Yellow or indigenous 31.5 68.5 5.8 94.2 3.2 96.8
Macro-region
Northern 37.5 62.5 0.317 14.5 85.5 0.004 5.0 95.0 0.056
Northeast 40.9 59.1 6.3 93.7 2.3 97.7
Southeast 44.8 55.2 6.0 94.0 2.0 98.0
Southern 45.1 54.9 8.2 91.8 3.2 96.8
Midwest 40.4 59.6 6.1 93.9 1.8 98.2
Household food security status†
Food secure 44.6 55.4 0.371 5.7 94.3 0.011 1.9 98.1 0.001
Mild food insecure 41.4 58.6 7.7 92.3 2.4 97.6
Moderate food insecure 37.4 62.6 9.7 90.3 2.8 97.2
Severe food insecure 44.9 55.1 13.4 86.6 8.3 91.7
Adequate sanitation
Yes 44.8 55.2 0.080 6.0 94.0 0.222 2.0 98.0 0.234
No 39.7 60.3 7.8 92.2 2.9 97.1
Adequate water supply
Yes 44.5 55.5 0.069 6.8 93.2 0.415 2.7 97.3 0.277
No 38.2 61.8 8.0 92.0 2.0 98.0
Household location
Urban 42.6 57.4 0.954 7.2 92.8 0.777 2.5 97.5 0.577
Rural 42.8 57.2 7.6 92.4 2.9 97.1
Educational level of the head of the
household
4 years of education or less 38.9 61.1 0.078 10.0 90.0 0.001 3.4 96.6 0.019
5 to 8 years of education 46.5 53.5 8.3 91.7 3.1 96.9
9 years of education or more 43.8 56.2 3.6 6.4 1.2 98.8

Numbers in bold indicate statistically significant estimates. *P-value referring to the chi-square test with statistical significance set at P < 0.05. Anal-
yses were performed using sampling weights. †As assessed by the Brazilian Household Food Insecurity Measurement Scale (Escala Brasileira de
Medida Domiciliar de Insegurança Alimentar. EBIA).

the variables under study (Table 3). The risk of stunting and households located in the Northern (OR: 2.98)
was higher for male than female children, for those living compared with Midwest region became stronger.
in the Northern compared with those living in the Midwest DBM was more likely to occur among heads of house-
region, and for pairs living in households where the head hold with lower compared with higher educational level
of household had low educational level. Multivariable (OR:2.54) (Table 3).
analyses confirmed a lack of association between mater- Hierarchical multivariable analyses (Table 4)
nal overweight and any of the socio-economic variables. showed that the association between DBM and severe
The adjusted odds for DBM were 3.33 times higher HFI remained after controlling for macroeconomic
among households with severe food insecurity com- (OR: 4.14) and socio-economic–demographic (OR:
pared with their food secure counterparts. Moreover, 2.65) variables, and when both sets of covariates were
the association of DBM with male child (OR: 1.96) included (OR: 3.64).

© 2016 John Wiley & Sons Ltd Maternal & Child Nutrition (2017), 13, e12347
6 of 9 M. B. Gubert et al.

Table 3. Crude and adjusted odds ratio for maternal overweight, child stunting and double burden of malnutrition for mother/child pairs according to
associated factors. Brazilian National Demographic and Health Survey 2006

Presence of maternal overweight Presence of child stunting Presence of double burden of

malnutrition

Associated factors OR (95%CI) Adjusted OR OR (95%CI) Adjusted OR OR (95%CI) Adjusted OR

(95%CI)* (95%CI)* (95%CI)*

Child’s sex
Female 1 1 1 1 1 1
Male 0.95 (0.76–1.18) 0.95 (0.76–1.18) 1.47 (1.04–2.09) 1.54 (1.06–2.24) 1.86 (1.09–3.17) 1.96 (1.14–3.38)
Race/skin colour
White 1 1 1 1 1 1
Black/brown 0.94 (0.75–1.18) 1.03 (0.81–1.30) 1.17 (0.79–1.74) 1.06 (0.67–1.68) 1.55 (0.81–2.94) 1.61 (0.80–3.24)
Yellow/indigenous 0.58 (0.29–1.17) 0.58 (0.29–1.17) 0.86 (0.39–1.90) 0.75 (0.35–1.60) 1.74 (0.56–5.36) 1.64 (0.58–4.69)
Macro-region
Midwest 1 1 1 1 1 1
Northern 0.89 (0.65–1.22) 0.91 (0.65–1.25) 2.64 (1.70–4.08) 2.36 (1.51–3.70) 2.84 (1.44–5.63) 2.98 (1.38–6.45)
Northeast 1.02 (0.72–1.45) 1.08 (0.75–1.55) 1.04 (0.64–1.68) 0.95 (0.59–1.53) 1.28 (0.62–2.66) 1.33 (0.60–2.92)
Southeast 1.20 (0.89–1.62) 1.23 (0.91–1.66) 0.99 (0.53–1.85) 1.09 (0.58–2.03) 1.10 (0.47–2.58) 1.37 (0.57–3.29)
Southern 1.21 (0.90–1.63) 1.23 (0.91–1.67) 1.39 (0.86–2.23) 1.47 (0.92–2.36) 1.76 (0.82–3.79) 2.20 (0.99–4.88)
Household food security status†
Food secure 1 1 1 1 1 1
Mild food insecure 0.88 (0.66–1.16) 0.92 (0.69–1.24) 1.39 (0.85–2.3) 1.23 (0.73–2.09) 1.26 (0.61–2.61) 1.10 (0.51–2.36)
Moderate food insecure 0.74 (0.51–1.08) 0.84 (0.57–1.23) 1.79 (1.09–2.93) 1.40 (0.82–2.37) 1.47 (0.73–2.94) 1.16 (0.56–2.41)
Severe food insecure 1.01 (0.64–1.61) 1.26 (0.79–2.03) 2.58 (1.44–4.62) 1.75 (0.96–3.18) 4.56 (2.11–9.82) 3.33 (1.41–7.84)
Adequate water supply
Yes 1 1 1 1 1 1
No 0.77 (0.58–1.02) 0.81 (0.59–1.01) 1.19 (0.79–1.79) 1.17 (0.74–1.86) 0.73 (0.41–1.30) 0.69 (0.39–1.24)
Household location
Urban 1 1 1 1 1 1
Rural 1.01 (0.70–1.45) 1.10 (0.77–1.57) 1.06 (0.70–1.62) 0.82 (0.55–1.22) 1.17 (0.67–2.03) 0.94 (0.56–1.58)
Educational level of the head of
the household
4 years of education or below 0.82 (0.62–1.07) 0.85 (0.65–1.12) 3.01 (1.80–5.04) 2.71 (1.48–4.94) 2.97 (1.37–6.43) 2.54 (1.10–5.89)
5 to 8 years 1.12 (0.84–1.48) 1.13 (0.84–1.52) 2.45 (1.39–4.23) 2.30 (1.25–4.24) 2.71 (1.20–6.13) 2.53 (1.14–5.65)
9 years of education or more 1 1 1 1 1 1

Numbers in bold indicate statistically significant estimates. *OR, odds ratio; CI, confidence interval; Adjusted OR, adjusted odds ratio; logistic regres-
sion adjusting for sex of the child, macro-region, food security status, household location and educational level of the head of the household. Analyses
were performed using survey weights. †As assessed by the Brazilian Household Food Insecurity Measurement Scale (Escala Brasileira de Medida
Domiciliar de Insegurança Alimentar. EBIA).

Discussion Household food insecurity as a risk factor for double

burden of malnutrition
The results of this study show that the coexistence of child
undernutrition and maternal overweight (characterized Our study confirmed our primary hypothesis that severe
here as DBM) in the same household in Brazil is not as HFI was significantly associated with the DBM in Brazil.
frequent as in other countries (Lee et al., 2012; Freire Even after controlling for covariates related to DBM,
et al., 2014; Sarmiento et al., 2014). This low prevalence severe HFI remained as the main factor associated with
of DBM was found in a previous study and was expected DBM. Previous studies showed an association between
given the very advanced stage of the nutrition transition in economic variables and the existence of DBM (Lee
Brazil, where women are experiencing weight gain from et al., 2012), but the association between DBM and food
one generation to the next that exceeds the height gain insecurity had not been explored in previous studies
rate (Conde & Monteiro, 2014). emphasizing the innovative nature of our findings.

© 2016 John Wiley & Sons Ltd Maternal & Child Nutrition (2017), 13, e12347
 Dual-burden malnutrition and food insecurity 7 of 9

Table 4. Crude odds ratio for double burden of malnutrition according to food security and adjusted for groups of factors related to food insecurity and
maternal nutritional status

Household food security status

OR for double burden of malnutrition Secure Mild food Moderate food Severe food
insecurity insecurity insecurity

Crude OR (95%CI) 1 1.26 (0.61–2.61) 1.47 (0.73–2.94) 4.56 (2.11–9.82)

Adjusted OR – macroeconomic variables (95%CI)* 1 1.29 (0.61–2.70) 1.45 (0.74–2.84) 4.14 (1.80–9.54)
Adjusted OR – socio-economic and demographic variables 1 1.02 (0.47–2.23) 1.04 (0.48–2.22) 2.65 (1.17–6.04)
(95%CI)†
Adjusted OR (95%CI)‡ 1 1.11 (0.52–2.40) 1.22 (0.59–2.56) 3.64 (1.55–8.53)

Adjusted OR, OR adjusted odds by logistic regression. All analyses were performed using survey weights. *Food security adjusted for macroeco-
nomic variables: household location and administrative macro-region. †Food security adjusted for socio-economic and demographic variables:
race/skin colour, maternal age, maternal parity and educational level of the head of the household. ‡Food security adjusted for macroeconomic,
socio-economic and demographic variables.

Food and nutritional insecurity, especially when our study is that we did not have enough statistical
reaching severe levels, has been shown to be a key factor power to test for interactions between HFI and each
associated with maternal overweight (Franklin et al., one of the covariates as there were only 140 pairs with
2012; Santos, 2013) and child stunting (Ali et al., 2013; DBM.
Felisbino-Mendes et al., 2014). HFI is determined by
socio-economic factors and leads to other vulnerabilities
Other factors associated with the double burden of
related to food access, such as overweight in adult
malnutrition
women and malnutrition in children, a condition that in-
fluences their growth, cognitive development and over- In addition to HFI, sex (male), Northern region and
all health (Gundersen & Ziliak, 2015). The association lower educational level of the head of the household
between HFI and DBM may be related to low-dietary were also associated with DBM.
quality including low food variety and high-energy den- The difference in the prevalence of DBM among the
sity (Shamah-Levy et al., 2014). Maternal stunting dur- Brazilian macro-regions reflects the well-known
ing childhood may also underlie the association country’s social and economic inequalities. The highest
between severe HFI and DBM in the same household prevalence of DBM was found in the Northern region,
(Felisbino-Mendes et al., 2014). Mothers having experi- which has the worst socio-economic indicators and the
enced stunting early in life and who remain stunted have highest prevalence of HFI (Instituto Brasileiro de
a higher chance of developing overweight in adulthood Geografia e Estatística., 2014) and stunting (Instituto
perhaps because of epigenetic factors (Reinhardt & Brasileiro de Geografia e Estatística, 2010; Ministerio
Fanzo, 2014) that were not examined in our study. da Saude, 2009) and a high prevalence of overweight
We acknowledge as a limitation of our study that among women (46.7%) (Instituto Brasileiro de
HFI measured in the last 3 months preceding the survey Geografia e Estatística, 2010).
may have been different from when stunting in children The results of the present study show that the risk of
and maternal obesity developed earlier in life. Thus, DBM is higher in households where the head has lower
prospective studies are needed to better understand education levels. This was expected, as educational
the nature of the HFI–DBM association detected in level is a strong social determinant of health. It is also
our study. However, it is important to acknowledge that important to recognize that the annual increment in
previous prospective studies have shown an association obesity rates among Brazilian women between the
between excessive weight gain and HFI among adults years of 1974–2009 was always higher for the bottom
(Cheung et al., 2015) and between stunting and HFI household income quintile compared with the highest
in children (Saha et al., 2009). Another limitation of (Conde & Monteiro, 2014). For stunting, as expected,

© 2016 John Wiley & Sons Ltd Maternal & Child Nutrition (2017), 13, e12347
8 of 9 M. B. Gubert et al.

the prevalence is higher among the poorest households Acknowledgment

(Instituto Brasileiro de Geografia e Estatística, 2010).
The authors thank the Brazilian Institute of Geography
Comparison of risk factors for maternal overweight, and Statistics (IBGE) for collecting and making the
child stunting and double burden of malnutrition data available for these analyses.

There were no socio-demographic covariates associ-

ated with maternal overweight. Previous studies had Source of founding
shown that in Brazil women’s overweight is not associ-
ated with severe HFI (Velasquez-Melendez et al., 2011) This work was supported by the Brazilian National Coun-
and its prevalence is similar across income quintiles cil of Technological and Scientific Development/CNPq
(45.0% in the bottom and 47.4% in the upper quintile (M.B.G. grant number 456699/2013-9, 232569/2014-2).
so that it reads upper quintile) (Instituto Brasileiro de
Geografia e Estatística, 2010). Conflicts of interest
The covariates associated with stunting were the same
as those for DBM: child sex (male), region (Northern) The authors declare that they have no conflicts of
and lower education level of the head of household. interest.
These data lead us to hypothesize that perhaps stunting
of the index child may reflect maternal stunting since
childhood which in turn is a risk factor for maternal over-
Contributions
weight (Gluckman et al., 2011). A study using data from
MBG, RP-E and AMS-C designed the research; MBG
54 Demographic Health Surveys conducted in different
and AMS performed statistical analysis; RP-E and
countries between 1991 and 2009 had previously shown
AMS-C supplied technical assistance and advice;
that maternal overweight is the principal predictor for
MBG, AMS, RP-E and AMS-C wrote the paper, and
DBM, but when maternal overweight is held constant,
MBG had primary responsibility for the final content.
child stunting becomes a positive predictor of DBM prev-
All authors read and approved the final manuscript.
alence (Dieffenbach & Stein, 2012). In our study, child
stunting seems to be the key-risk factor for the occurrence
of DBM. It is possible that the well-documented intergen- References
erational transmission of obesity risk from mother to chil-
Abdullah A. (2015) The double burden of undernutrition and
dren through epigenetic mechanisms (Gluckman et al., overnutrition in developing countries: an update. Current
2011) helps explains these findings. As the maternal-child Obesity 4, 337–49.
obesity cycle has been proven to be hard to break Ali D., Saha K.K., Nguyen P.H., Diressie M.T., Ruel M.T.,
(Gluckman et al., 2011), the coexistence of maternal over- Menon P. et al. (2013) Household food insecurity is associ-
ated with higher child undernutrition in Bangladesh,
weight and child stunting in the same household can be
Ethiopia, and Vietnam, but the effect is not mediated by
reflecting this kind of epigenetic accommodation. child dietary diversity. Journal of Nutrition 143, 2015–21.
Bernal J., Frongillo E.A., Herrera H.A. & Rivera J.A. (2014)
Food insecurity in children but not in their mothers is asso-
Conclusion ciated with altered activities, school absenteeism, and
stunting. Journal of Nutrition 144, 1619–26.
In Brazil, severe HFI is associated with the occurrence Cheung H.C., Shen A., Oo S., Tilahun H., Cohen M.J. &
of DBM. Future longitudinal studies are needed to Berkovitz S.A. (2015) Food insecurity and body mass index:
understand the factors and complex pathways linking a longitudinal mixed methods study, Chelsea, Massachusetts,
2009–2013. Preventing Chronic Disease 12 (E125), 1–13.
HFI with DBM risk. These studies can also help to
Conde W.L. & Monteiro C.A. (2014) Nutrition transition
understand how early childhood chronic malnutrition and double burden of undernutrition and excess of weight
interact with maternal overweight/obesity to increase in Brazil. American Journal of Clinical Nutrition 100,
the long-term risk of obesity in the offspring. 1617S–22S.

© 2016 John Wiley & Sons Ltd Maternal & Child Nutrition (2017), 13, e12347
 Dual-burden malnutrition and food insecurity 9 of 9

Dieffenbach S. & Stein A.D. (2012) Stunted child/overweight coexistence of child stunting and maternal overweight in
mother pairs represent a statistical artifact, not a distinct Guatemala. Economics and Human Biology 10, 232–41.
entity. Journal of Nutrition 142, 771–3. Lin H., Carr K.A., Fletcher K.D. & Epstein L.H. (2013) Food
Felisbino-Mendes M.S., Villamor E. & Velasquez-Melendez reinforcement partially mediates the effect of socioeco-
G. (2014) Association of maternal and child nutritional nomic status on body mass index. Obesity 21, 1307–12.
status in Brazil: a population based cross-sectional study. Metallinos-Katsaras E., Must A. & Gorman K. (2012) A longi-
PLoS ONE 9, e87486. tudinal study of food insecurity on obesity in preschool
Franklin B., Jones A., Love D., Puckett S., Macklin J. & children. Journal of the Academy of Nutrition and Dietetics
White-Means V. (2012) Exploring mediators of food insecu- 112, 1949–58.
rity and obesity: a review of recent literature. Journal of Ministerio da Saude (Ministry of Health) (2009) Pesquisa
Community Health 37, 253–64. Nacional de Demografia e Saúde da Criança e da Mulher –
Freire W.B., Silva-Jaramillo K.M., Ramirez-Luzuriaga M.J., PNDS 2006: dimensões do processo reprodutivo e da saúde
Belmont P. & Waters W.F. (2014) The double burden of un- da criança. Ministério da Saúde: Rio de Janeiro.
dernutrition and excess body weight in Ecuador. American Mutisya M., Kandala N.B., Ngware M.W. & Kabiru C.W.
Journal of Clinical Nutrition 100, 1636S–43S. (2015) Household food (in)security and nutritional status
Frongillo E.A. & Bernal J. (2014) Understanding the coexis- of urban poor children aged 6 to 23 months in Kenya.
tence of food insecurity and obesity. Current Pediatric 2, BMC Public Health 15, 1052.
284–290. United Nations. (2015) World Food Program. Available at:
Ghattas H. (2014) Food Security and Nutrition in the Context of http://www.wfp.org/food-security (Accessed 20 September
the Nutrition Transition. Food and Agriculture Organization 2015).
of the United Nations: Rome. Reinhardt K. & Fanzo J. (2014) Addressing chronic malnutri-
Gluckman P.D., Hanson V. & Low F.M. (2011) The role of tion through multi-sectoral, sustainable approaches: a review
developmental plasticity and epigenetics in human health. of the causes and consequences. Frontiers nutrition 1, 13.
Birth Defects Research. Part C, Embryo Today : Reviews Saha K.K., Frongillo E.A., Alam D.S., Arifeen S.E., Persson L.
93, 12–8. A. & Rasmussen K.M. (2009) Household food security is
Gundersen C. & Ziliak J.P. (2015) Food insecurity and health associated with growth of infants and young children in rural
outcomes. Health Affairs 34, 1830–9. Bangladesh. Public Health Nutrition 12, 1556–62.
Hoffman D.J., Sawaya A.L., Verreschi I., Tucker K.L. & Santos L.M. (2013) Obesity, poverty, and food insecurity in
Roberts S.B. (2000) Why are nutritionally stunted children Brazilian males and females. Cadernos de Saúde Pública
at increased risk of obesity? Studies of metabolic rate and 29, 237–9.
fat oxidation in shantytown children from Sao Paulo, Brazil. Sarmiento O.L., Parra D.C., Gonzalez S.A., Gonzalez-Casanova
American Journal of Clinical Nutrition 72, 702–7. I., Forero A.Y. & Garcia J. (2014) The dual burden of
Instituto Brasileiro de Geografia e Estatística (2010) Pesquisa malnutrition in Colombia. American Journal of Clinical
de Orçamentos Familiares–POF 2008/2009. Antropometria e Nutrition 100, 1628S–35S.
estado nutricional de crianças, adolescentes e adultos no Brasil. Shamah-Levy T., Mundo-Rosas V. & Rivera-Dommarco J.A.
Instituto Brasileiro de Geografia e Estatística: Rio de Janeiro. (2014) Magnitude of food insecurity in Mexico: its relation-
Instituto Brasileiro de Geografia e Estatística (2014) Pesquisa ship with nutritional status and socioeconomic factors.
Nacional por Amostra de Domicílios – PNAD. Segurança Salud Pública de México 56, s79–85.
Alimentar 2013. Instituto Brasileiro de Geografia e Segall-Corrêa A.M., Marin-León L., Melgar-Quiñonez H. &
Estatística: Rio de Janeiro. Pérez-Escamilla R. (2014) Refinement of the Brazilian
Kac G., Schlussel M.M., Perez-Escamilla R., Velasquez- Household Food Insecurity Measurement Scale: recommen-
Melendez G. & Silva A.A. (2012a) Household food insecurity dation for a 14-item EBIA. Revista de Nutrição 27, 241–251.
is not associated with BMI for age or weight for height among To Q.G., Frongillo E.A., Gallegos D. & Moore J.B. (2014)
Brazilian children aged 0–60 months. PLoS ONE 7, e45747. Household food insecurity is associated with less Physical
Kac G., Velasquez-Melendez G., Schlussel M.M., Segall- Activity among children and adults in the U.S. population.
Corrêa A.M., Silva A.A. & Perez-Escamilla R. (2012b) Journal of Nutrition 144, 1797–1802.
Severe food insecurity is associated with obesity among Velasquez-Melendez G., Schlussel M.M., Brito A.S., Silva A.
Brazilian adolescent females. Public Health Nutrition 15, A., Lopes-Filho J.D. & Kac G. (2011) Mild but not light or
1854–60. severe food insecurity is associated with obesity among
Gudmundsson S., Henningsson A.C. & Lindqvist P. (2005) Brazilian women. Journal of Nutrition 141, 898–902.
Correlation of birth injury with maternal height and World Health Organization (2006) WHO Child Growth Stan-
birthweight. BJOG 112, 764–7. dards: Length/height-for-age, Weight-for-age, Weight-for-
Lee J., Houser R.F., Must A., de Fulladolsa P.P. & Bermudez length, Weight-for-height and Body Mass Index-for-age:
O.I. (2012) Socioeconomic disparities and the familial Methods and Development. WHO: Geneva.

© 2016 John Wiley & Sons Ltd Maternal & Child Nutrition (2017), 13, e12347
Copyright of Maternal & Child Nutrition is the property of Wiley-Blackwell and its content
may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for
individual use.