Content uploaded by Rachel Bezner Kerr
Author content
All content in this area was uploaded by Rachel Bezner Kerr on Nov 26, 2014
Content may be subject to copyright.
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
90
Food and Nutrition Bulletin, vol. 28, no. 1 © 2007, The United Nations University.
Breastfeeding and mixed feeding practices in Malawi:
Timing, reasons, decision makers, and child health
consequences
Rachel Bezner Kerr, Peter R. Berti, and Marko Chirwa*
Abstract
Background. In order to effectively promote exclusive
breastfeeding, it is important to first understand who
makes child-care and child-feeding decisions, and why
those decisions are made; as in most parts of the world,
exclusive breastfeeding until 6 months of age is uncom-
mon in Malawi.
Objective. To characterize early infant foods in rural
northern Malawi, who the decision-makers are, their
motivation, and the consequences for child growth, in
order to design a more effective program for improved
child health and nutrition.
Methods. In a rural area of northern Malawi, 160
caregivers of children 6 to 48 months of age were asked
to recall the child’s age at introduction of 19 common
early infant foods, who decided to introduce the food,
and why. The heights and weights of the 160 children
were measured.
Results. Sixty-five percent of the children were given
food in their first month, and only 4% of the children
were exclusively breastfed for 6 months. Mzuwula and
dawale (two herbal infusions), water, and porridge were
common early foods. Grandmothers introduced mzuwula
to protect the children from illness; other foods were usu-
ally introduced by mothers or grandmothers in response
to perceived hunger. The early introduction of porridge
and dawale, but not mzuwula, was associated with
worse anthropometric status. Mzuwula, which is not
associated with poor growth, is usually made with boiled
water and given in small amounts. Conversely, porridge,
which is associated with poor child growth, is potentially
contaminated and is served in larger amounts, which
would displace breastmilk.
Conclusions. Promoters of exclusive breastfeeding
should target their messages to appropriate decision
makers and consider targeting foods that are most harm-
ful to child growth.
Key words: Child growth, exclusive breastfeeding,
Malawi, mixed feeding
Introduction
The quality of the diet in the first years of life is a key
factor in children’s health and survival [1]. Mild to
moderate child malnutrition has been estimated to
account for 53% of all child deaths in developing coun-
tries [2]. Appropriate breastfeeding and complementary
feeding methods could potentially halve African infant
mortality rates. Current nutrition policy encourages
mothers to exclusively breastfeed their children for
the first 6 months of life [3], providing the infant with
a nutritionally sufficient, clean, and safe diet. Despite
the recommendations and widespread promotion of
exclusive breastfeeding, it is often not practiced, even
in developing countries where it would be most benefi-
cial. Most recent estimates suggest that one-third of all
infants in sub-Saharan Africa are exclusively breastfed,
an increase from 15% in 1990 [4].
Malawi is a landlocked country in southeastern
Africa with a population of approximately 11 million,
80% of whom live in rural areas. The infant mortality
rate is 114 per 1,000 live births, and the under-five mor-
tality rate is 183 per 1,000 live births [5]. High levels of
child malnutrition have been observed in Malawi for
over two decades [6, 7]. Growth faltering begins soon
after birth, with rapid worsening beginning at approxi-
mately 5 months, continuing through the second year,
when stunting prevalence peaks at 60%, and remain-
Rachel Bezner Kerr is affiliated with the Department of Geo-
graphy, The University of Western Ontario, London, Ontario,
Canada, Peter R. Berti is affiliated with HealthBridge, Ottawa;
Marko Chirwa was affiliated with Soils, Food and Healthy
Communities, Ekwendeni Hospital, Ekwendeni, Malawi.
Please direct queries to the corresponding author: Rachel
Bezner Kerr, Department of Geography, SSC 2409, 1151 Rich-
mond Street N, The University of Western Ontario, London,
Ontario, Canada,N6A 5C2; e-mail: rbeznerkerr@uwo.ca.
These data were presented at the 2005 International Union
of Nutritional Sciences congress in Durban, South Africa.
* Deceased.
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
91
Breastfeeding and mixed feeding practices in Malawi
ing above 50% up to the age of five [7]. About 25% of
Malawian under-five children are underweight; peak
levels occur between 12 and 24 months of age [7]. Food
insecurity is a major problem, with between 70% and
85% of households experiencing food shortages on an
annual basis [8].
Exclusive breastfeeding appears to be uncommon
in Malawi, despite very high rates of breastfeeding
(nonexclusive + exclusive breastfeeding = 95% at 17
months of age [9]) and a national effort to increase
exclusive breastfeeding [9-11]. A confounding factor
and a perceived disincentive to exclusive breastfeeding
is the high prevalence rate of HIV-positive Malawian
pregnant women, estimated at 14% to 20% [12]. One
recent study suggested that HIV-infected Malawian
women were concerned that exclusive breastfeeding
would compromise their health status if they did not
have better access to nutritious foods [13].
The objective of the research presented in this paper
is to characterize early infant-feeding practices in a
rural Malawi setting, and specifically, to document
the foods that are introduced early to the children,
who the decision makers are for these practices, their
motivation for doing so, and the consequences for child
growth. With this characterization, a more effective
program for the promotion of exclusive breastfeeding
and healthy feeding practices might be designed.
Description of study site
The research was conducted in a rural area of northern
Malawi near the town of Ekwendeni in Mzimba Dis-
trict. Approximately 80% of the population are small-
holder farmers. Maize is the primary staple crop and is
harvested in May and June after the single annual rainy
season. Other important crops include beans, squash,
groundnuts, and sweet potatoes, as well as tobacco as
the major cash crop.
This research was carried out within a larger com-
munity-based research and development project, the
Soils, Food and Healthy Communities Project (SFHC),
based at Ekwendeni Hospital, that began in 2000. SFHC
tries to improve the health of resource-poor households
through participatory research that introduces relay
cropping and intercrops of legumes to improve soil fer-
tility, food security, and nutrition of poor households,
in particular by increasing legume consumption by
young children.
Methods
The overall research approach was interdisciplinary
and multimethod. The research team, composed of
a sociologist, a nutritionist, hospital staff, and farm-
ers, carried out a combination of qualitative and
quantitative research to understand early infant-feed-
ing practices. Early introduction of non-breastmilk
foods to infants was identified as an issue of interest
on the basis of the qualitative research, and specific
questions about these practices were included in the
survey. Twenty-one semistructured interviews were
conducted in 2001 in which key informants (mothers,
older women, and traditional medicine practitioners)
identified by community members were asked about
pregnancy, breastfeeding, early infant feeding, and
general care practices and beliefs. The questions were
developed on the basis of initial test interviews with
informants. Four focus groups were held with groups of
men and women in the villages, using similar questions
as the semistructured interviews. In addition, free lists
of foods eaten by young children were obtained from
28 informants [14]. The team worked in pairs to carry
out the interviews, with one person interviewing and
the other person translating or taking notes. Informed
consent was obtained prior to all interviews. All inter-
views were recorded, and the tapes were transcribed
and translated into English.
Using the qualitative research as a basis for design,
an extensive survey on agriculture and child-care and
child-feeding practices was conducted in February
2002 (during the “hungry season”) with 264 house-
holds. Questions were included about the timing of
the introduction of 19 foods (identified as the most
common early infant foods in the qualitative research),
the reason or reasons that each of these foods was
introduced, and the individual(s) who made the deci-
sion to introduce the food.
The subjects of this study were children 6 to 48
months of age and their primary caregivers (99% of
respondents were the children’s mothers; the remain-
ing respondents were one father, one grandmother,
and one stepmother) from two sets of households:
those in intervention villages and those in control
villages. The original survey design was intended to
compare intervention households with noninterven-
tion households in terms of food security, soil fertility,
child nutritional status, and child-feeding practices.
Intervention-village households were participating
in the SFHC Project; control villages were selected on
the basis of similar socioeconomic and environmental
conditions. Households were recruited to the Project
at village-level meetings organized by the hospital.
Survey participants from intervention-village house-
holds were randomly selected from this group. In
order to reduce selection bias, since other studies have
indicated that more food-secure households join par-
ticipatory agricultural projects [15], and to control for
growth differences based on age, control households
were matched with intervention households on the
basis of two criteria: age of the child and food-security
status (operationalized as the month when self-grown
maize stores are exhausted). The majority of respond-
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
92
R. Bezner Kerr et al.
ents (87%) had a primary-level education, 6.5% had
a secondary-level education, and 6% had no formal
schooling. Most of the respondents (74%) were in
monogamous marriages, 18% were in polygamous
marriages, 6% were divorced, 1% were widowed, and
1% were single. Sixty-three percent of respondents had
one under-five child in the household and 35% had two
under-five children. The average number of people in
the household was five.
In 160 households with a child under 4 years of age,
the primary caregiver was interviewed in the home
about the child’s consumption of 19 specific early infant
foods and was also asked if the child consumed other
foods (fig. 1). In this sample, 112 households (57 con-
trol, 55 intervention) had children 6 to 24 months of
age, and 48 (22 control, 26 intervention) had children
24 to 48 months of age. In the older group, 45 children
were between 24 and 34 months of age and 3 were
more than 34 months of age. For each of the 19 foods,
the caregiver was asked if the food had been intro-
duced to the child, and if so at what age (in months),
the reasons for introducing the food (from a list of 10
choices, including “other”), and who was involved in
deciding when the food should be introduced (a list
was provided of 11 individuals, as well as “other,” “don’t
know,” and no answer). The surveys were pretested in a
neighboring village and revised. In order to minimize
biased responses due to extensive hospital education
about exclusive breastfeeding, no direct questions were
asked about how long the women had breastfed exclu-
sively, The results suggest that asking women indirectly
minimized response bias about breastfeeding practices.
Although the time period for caregivers to recall what
foods were given to their children was long (up to 48
months), exact amounts were not assessed, and initial
qualitative interviews indicated that caregivers were
able to easily recall what foods were initially given to
their children. Thus, measurement bias is anticipated
to be low.
Anthropometric data were collected from 405 chil-
dren in a central village location (e.g., primary school)
by trained research assistants following standardized
procedures and using calibrated equipment. Nude
weights were measured by an electronic TANITA
“Baby and Mommy Scale” (model 1582; lb/kg version)
and recorded to the nearest 0.01 kg. A 100-cm length
board (Perspective Enterprises) was used to record
the length of children from sampled intervention and
control households.
The research protocol was reviewed and approved
by the Malawi National Research Council and by the
Cornell University Committee on Human Subjects.
Informed consent was obtained orally from each adult
caregiver prior to the survey and collection of anthro-
pometric data.
Data management and analysis
Qualitative data were analyzed for trends, key concepts,
and practices by data analysis techniques described by
Miles and Huberman [16] and Patton [17]. A coding
scheme was developed using different themes based
FIG. 1. Flow chart of sampling methods and number of participants, February 2002 survey
!NTHROPOMETRICS
N
INTERVENTIONCHILDREN
CONTROLCHILDREN
!NTHROPOMETRICSAND
INTRODUCTIONOFFOODDATA
N
)NTRODUCTIONOFFOOD
N
2ANDOMSELECTIONOFHOUSEHOLDSFROMINTERVENTIONVILLAGES
HADrCHILDYR
IFUNDERCHILDTHEYOUNGESTWASENROLLED
!MERGEDSAMPLEOFCHILDREN
FROMHOUSEHOLDS)NVILLAGES
3ELECTIONOFHOUSESFROMCONTROLVILLAGES
MATCHEDWITHINTERVENTIONCHILDRENBYAGEAND
FOODSECURITYSTATUSGIVINGCHILDREN
nMO
N
nMO
N
nMO
N
INTERVENTIONCHILDREN
CONTROLCHILDREN
INTERVENTIONCHILDREN
CONTROLCHILDREN
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
93
Breastfeeding and mixed feeding practices in Malawi
on the interview schedules. This code list was modi-
fied during the data analysis process as new themes
emerged. A set of definitions for each code was also
developed and modified throughout the analytical
process. After coding, the data were read over and
examined for emerging themes and trends. Inductive
analysis was used to identify indigenous food concepts,
feeding practices, and key social roles in child feeding.
Follow-up interviews and focus group discussions were
conducted to better understand these concepts, prac-
tices, and roles. Identification of sensitive topics in the
qualitative research allowed the team to carefully word
survey questions to avoid biased answers.
The quantitative data, i.e., the anthropometric data
and the data from the survey regarding age at intro-
duction of the food, reasons for introducing the food,
and the decision makers, were entered in Excel and
analyzed using SAS (version 8). For each month of
age, the percentage of children given a food for the first
time was calculated as (number of children first given
the food at age X or younger) ÷ (number of children
whose age was X or more). The effect of household
food security and socioeconomic status (SES) on age
at introduction of each food was tested by multino-
mial logistic regression (age at introduction vs. month
household ran out of or expected to run out of maize,
where ages at introduction are classified as < 1 month,
1-4 months, 4-6 months, and > 6 months or never [i.e.,
the child was > 6 months of age but had not been given
the food]). The reasons for the introduction and the
decision makers were tabulated, and cross-tabulations
between “reasons” and “decision makers” were also
generated.
The effect of early introduction of individual foods
was tested by comparing growth status with age at
introduction of the three key early infant foods. The
children’s heights and weights were converted to
height-for-age, weight-for-age, and weight-for-height
z-scores (HAZ, WAZ, and WHZ, respectively) by
calculating the difference between the child’s measure
and the age-matched mean measure of a reference
population, and then dividing by the standard devia-
tion of a reference population [18]. In SAS, a general
linear model was used where the dependent variable
was HAZ, WAZ, or WHZ and the independent vari-
able was age at introduction of a food (categorized as
described above). The model included children from
6 to 48 months of age for whom we had both anthro-
pometric data and data on the timing of introduction
of early infant food. “Porridge” includes the earliest
introduction of any type of porridge: porridge with
unrefined maize flour (mgaiwa) and porridge with milk
and maize flour (chintuwe) (table 1). “Dawale” includes
dawale, “dawale water,” and “dawale porridge,” and the
earliest age of introduction of any of the three types of
dawale was used. The model tested for effects of the
child’s sex and age, the month at which the household
ran out of or expected to run out of maize (as an indica-
tor of food security), and an SES index. The SES index
consisted of a rating for housing, ownership of mate-
rial goods, and maternal education. Housing materials
were ranked from low to high for the following mate-
rials: sticks, thatch or grass, unfired brick, clay tiles,
fired brick, and iron sheets. The material goods were
oxcart, wheelbarrow, radio, plow, motorcycle, ridger
(a soil-tilling implement), mosquito net, bicycle, sofa
or armchair, table and chairs, and tobacco press. The
index assigned one point for each item owned.
Qualitative interviews, participant observation, and
focus group discussions suggested that housing materi-
als and ownership of goods were good SES indicators.
Other studies in Malawi have included ownership of
goods and housing type as good proxy indicators of
SES [19]. Maternal education was included because
other studies in Malawi and elsewhere have found that
maternal education can have a significant effect on
children’s nutritional status [11]. The independent vari-
ables are treated as fixed effects in this model. There
was good concordance between individual indicators
and SES points.
Results
In the results presented in this paper, there are no
differences between intervention and control villages
(data not shown), and therefore the data are presented
for the villages combined.
Perceptions of breastfeeding
Informants felt that the first milk (colostrum) was
good for babies, as well as breastmilk, but exclusive
breastfeeding was not widely practiced. They said that
breastmilk protects a baby from diseases, helps their
bowels develop, and gives them energy. All women
TABLE 1. Definitions of local food terms based on qualita-
tive interviews
Term Definition
Mzuwula Herbal infusion made from the leaves of
specific tree species found in the area. The
infusion is made from crushed leaves mixed
with boiled or cold water
Dawale Herbal infusion made from the roots of
a specific tree species found in the area.
Sometimes other leaves are crushed and
added to the infusion. Sometimes fed to
infant as an infusion, sometimes added to
porridge to make a very thin porridge.
Chinthipu Very thin, watery, maize porridge
Mgaiwa Unrefined maize flour
Chintuwe Porridge with milk and maize flour
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
94
R. Bezner Kerr et al.
interviewed named breastmilk as the primary source of
food for babies, but several other foods could be given
to a baby if it cried. The crying of a baby was seen as a
sign of hunger and an indication that the baby was not
getting enough food from breastmilk. Many women
said that breastmilk was given until the baby cried after
breastfeeding, at which time a thin maize porridge was
given. Some informants said that certain babies are
“born hungry” and may have to be fed porridge at a
very early age, or gripe water or different herbal infu-
sions, discussed below, in some cases. The mothers
said that the most important reason for introducing
porridge was that the baby was crying.
Age at introduction
The age at introduction of the 19 foods is summarized
in figure 2 in a cumulative frequency plot. The percent-
age of children who have been given a food by the end
of each month of age is shown. Sixty-five percent of the
children are given some type of food in the first month.
By 6 months, 96% of the children have been given some
type of food (see the “Any food” line in fig. 2).
There are three notable patterns in the introduction
of foods. First, mzuwula (an infusion made with the
leaves of a local tree) is introduced to 50% of the chil-
dren in the first month and then to only an additional
10% over the next 17 months. Mzuwula is an infusion
of water and pounded leaves and roots from a par-
ticular tree species. Approximately 80% of the children
who were given any food in the first month were given
mzuwula. Second, plain water or water with dawale,
another kind of root water, is introduced to 20% of the
children in the first month and then to an additional
10% per month for the next 6 months. Third, porridge
with chinthipu (thin porridge with white maize flour)
or porridge with dawale is introduced to 10% of chil-
dren in the second month and to 70% of children by
the sixth month.
The exact amount given to children varied with the
substance. In the qualitative interviews, informants
were asked to estimate the amounts given to children.
They indicated that mzuwula was typically given
in small amounts (e.g., a teaspoon or about 5 mL),
whereas porridge, water, or dawale could be given in
amounts ranging from 25 to 500 mL. Specific amounts
were not measured.
There was no relationship between age at introduc-
tion of each food, anthropometric measurements, and
household food security or SES. There was a significant
relationship between total SES points and the introduc-
tion of dawale. Households with a higher SES index
were more likely to introduce dawale to children after
the age of 6 months. The reasons for this difference
may be linked to a local perception that food-insecure
women produce insufficient breastmilk [20]. These
perceptions and links to breastfeeding patterns are
being further studied through qualitative research.
Reported reasons for introduction
The caregivers’ reported reasons that the various foods
were introduced to the children are summarized in
table 2. A commonly reported reason for introducing
FIG. 2. Age at introduction of foods to Ekwendeni children: cumulative
frequency.
Dawale includes “dawale,” “dawale water,” and “dawale porridge.” Dawale is a root that
is cut up and added to water, which is boiled and then strained and given as water or
added to porridge. Porridge includes chinthipu (made from maize and water), mgaiwa
(made from fermented, unrefined maize), and chinthuwe (made from milk and maize
flour). Note that the percentage of children given the food fluctuates after the age of 6
to 9 months. This is a sampling artifact (fewer children are sampled at older ages); in
reality there would be a plateau at the percentage of children who ever eat a food. There
are other foods or food groups not shown that never exceed 10% cumulative frequency
(gripe water, chindongwa [a groundnut-maize bread], soft drinks, sugar water, and milk)
or 20% cumulative frequency (infant formula)
0
10
20
30
40
50
60
70
80
90
100
%
Age (mo)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 7 18
Any food
Porridge
Water
Mzuwula
Dawale
Vegetable soup
Tea
Fruit
Nsima
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
95
Breastfeeding and mixed feeding practices in Malawi
many foods was that the child was hungry or
crying. The most common reason reported as
“other” was that the child was thirsty. However,
mzuwula was given in 84% of cases to “protect”
the child from illness believed to be caused by
“promiscuity” of the mother or father (even
within marriage) within 1 year of birth, or in
some instances by promiscuity of anyone in the
village [20].
Decision makers
The individuals who decided when a given food
was to be introduced are summarized in table 3.
Most often the caregiver (in 99% of cases the
mother) was the decision maker, but the mother-
in-law and occasionally the father-in-law were
also important decision makers, particularly
with regard to foods that were given to protect
the child.
Cross-tabulations of “reasons × decision
makers” revealed that water and porridge were
given in 80% of cases because the child was
perceived by the mother or mother-in-law to
be hungry and crying. However, in 78% of cases
when mzuwula was introduced, the mother-
in-law made the decision and it was given to
protect the child from the disease associated
with promiscuity. Young women noted in focus
groups that their mothers-in-law have tremen-
dous influence over all child-care and feeding
activities [20]. A crying baby is perceived as a
sign of poor child care, and a grandmother can
even remove a child from the home if she feels
the child is not getting enough food. Vigorous
discussions with older women indicated a very
strong belief that breastmilk is not adequate for
young babies, particularly during the “hungry
season” from December to March, which is
also the period reported by informants to have
the highest level of illness from water-borne
diseases.
Relationship between early infant-feeding practices
and anthropometric measurements
We tested the relationship between anthropo-
metric measurements of 160 children 6 to 48
months of age (of whom 112 were less than 24
months old) and the timing of introduction
of four food types during infancy. The model
tested the z-score vs. the age at introduction
of water, mzuwula, porridge, or dawale, while
controlling for the child’s age and sex and the age
at which the household ran out of or expected
to run out of maize, as a general indicator of
food security and SES. Age at introduction was
classified as < 1 month, 1 to 3 months, 4 to 6
months, or > 6 months or never). The models
TABLE 2. Number of times various reasons were given by mothers for feeding particular foods to their children (n = 157)
a
Reason
Water
Maize porridge
Mzuwula
Vegetable soup
Fermented,
unrefined maize
porridge (mgaiwa)
Fruit
Tea
Nsima
Dawale
Dawale water
Dawale porridge
Formula
Gripe water
Milk
Sweet potato
Maize bread
Soft drink
Milk and maize
porridge
Sugar water
Advised to feed 4 2 1 1 4 1 1 3 1 1 1 0 0 1 0 1 0 0 0
Child ill 3 0 2 0 0 2 1 0 1 0 0 0 3 0 0 1 0 1 0
Child crying or hungry 102 103 9 26 40 21 29 28 19 16 15 14 4 5 6 1 6 8 1
Child’s interest in food 1 1 0 4 2 23 19 16 0 0 0 5 0 0 5 6 2 0 1
Mother ill 2 1 0 2 2 2 1 2 0 0 0 0 0 0 0 0 0 0 0
Not enough breastmilk 9 15 0 0 5 1 4 3 0 3 1 0 0 4 0 0 1 3 0
To protect child 2 1 72 16 0 12 6 1 15 10 6 0 11 3 0 0 0 0 0
To strengthen child 2 17 6 27 30 19 5 16 7 11 6 6 0 4 2 1 3 0 0
Mother couldn’t breastfeed 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0
Other 22 0 0 0 1 2 0 1 0 2 1 0 0 0 0 2 1 0 2
No. of children given food 138 125 86 73 71 68 62 55 41 35 25 23 18 14 13 12 11 9 4
a. Up to three reasons were given per food.
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
96
R. Bezner Kerr et al.
for WHZ (p = .002, r
2
= 0.21) and WAZ
(p = .009, r
2
= 0.19) were significant. The
age at introduction of water, porridge, and
dawale, but not mzuwula, was associated
with WHZ and WAZ (p < .02 in all cases
except WHZ–dawale, where p = .08). The
results for porridge and dawale are depicted
in figure 3; the early introduction of dawale
or porridge is associated with a z-score
disadvantage of 0.5 to 1.5. The relation-
ship with age at introduction of water is
ambiguous, with early (< 1 month) and late
(> 6 months) introduction of water having
an advantage of approximately 0.7 z-score
over intermediate introduction.
Discussion
Giving an infant foods other than breast-
milk during the first 6 months of life
displaces the more nutritious breastmilk
from the infant’s diet and risks introducing
diarrhea-causing pathogens to the child.
It is acknowledged that the results of this
study cannot be generalized beyond the
study area, since the study population is in
an area with an active primary health care
program and was not randomly selected. In
addition, the cultural reasons for specific
feeding practices found in the Ekwendeni
area are not applicable to the entire country.
Nonetheless, some broader implications
can be drawn from this research, because
despite this active program, low rates of
exclusive breastfeeding were found, and
similar practices have been noted in other
regions of southern Africa. The health
consequences of not practicing exclusive
breastfeeding have been demonstrated
in numerous studies in numerous set-
tings [21], and feeding anything other than
breastmilk is discouraged by Ekwendeni
Hospital. However, despite an ongoing
exclusive breastfeeding promotion cam-
paign and recognition of the hospital as
a baby-friendly hospital since 2000, as in
many settings throughout the world, exclu-
sive breastfeeding is rarely practiced in the
surveyed population. Sixty-five percent
of infants were given some food in their
first month, and only 4% of children were
reported to be exclusively breastfed for 6
months.
Typically, promotion of exclusive breast-
feeding is targeted to the mothers in health
clinic education sessions. We found that
TABLE 3. Individuals who decided when a food was to be introduced to the child’s diet
Decision maker
Total no. of decisions
Water
Porridge with
chinthipu
Mzuwula
Vegetable soup
Porridge with unre-
fined flour (mgaiwa)
Fruit
Tea
Nsima
Root water (dawale)
Dawale water
Dawale porridge
Formula
Gripe water
Milk
Sweet potato
Corn bread
(chindongwa)
Soft drink
Porridge with
chintuwe
Sugar water
Mother 627 100 96 15 69 63 56 56 50 16 12 12 22 7 11 11 11 10 7 3
Father-in-law or mother-in-law 206 31 28 63 3 7 5 1 4 23 21 12 1 1 1 0 1 2 2 0
Spouse 34 2 2 0 3 0 5 4 3 1 0 0 1 9 2 1 0 0 1 0
Parents 9 1 1 2 1 0 0 0 1 0 2 1 0 0 0 0 0 0 0 0
Grandparent 8 0 1 3 0 0 0 0 0 2 0 1 0 1 0 0 0 0 0 0
Sister or brother 5 0 0 0 1 0 3 0 0 0 0 0 1 0 0 0 0 0 0 0
Sister-in-law or brother-in-law 4 0 1 2 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0
Child 3 0 0 0 0 0 1 0 2 0 0 0 0 0 0 0 0 0 0 0
Aunt or uncle 2 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Niece or nephew 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
No answer 16 2 1 2 2 2 1 0 2 0 1 1 0 0 0 1 0 0 0 1
No. of children given food 138 125 86 73 71 68 62 55 41 35 25 23 18 14 13 12 11 9 4
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
97
Breastfeeding and mixed feeding practices in Malawi
such promotion may be inappropriate or inadequate,
since the mother does not always decide what a child
eats and often it is the mother-in-law who decides.
Mzuwula is given to “protect” children against an ill-
ness believed to be caused by promiscuity [20]. This
belief is firmly held by many villagers. Is it therefore
necessary to promote the reduction of the practice of
feeding mzuwula to infants? Conventional exclusive
breastfeeding policy would certainly discourage giving
infants mzuwula. However, mzuwula is often prepared
with boiled water (and therefore is likely to be sterile),
and it is given infrequently and in small amounts (and
thus displaces little breastmilk). Perhaps not surpris-
ingly, then, its early introduction is not related to child
growth. Furthermore, it is tied in to a very strongly held
belief system, the behavior is deeply engrained, and
it would probably be very difficult to change. Project
staff initiated a nutrition program with villagers that
attempts to promote healthy ways of using mzuwula,
such as always boiling the water prior to preparation,
or bathing the child in mzuwula rather than feeding it
to the child. Thus, the Project does not attack strongly
held traditional beliefs while trying to improve child-
feeding practices.
Porridge and dawale, on the other hand, are given to
the child because the decision maker believes the child
is hungry or thirsty. Implicitly tied to this idea is the
notion that mothers produce insufficient breastmilk
for babies, a belief found in other parts of southern
Africa [22]. Porridge is prepared in the morning and
then allowed to cool and is served throughout the day,
and thus is a potential source of pathogens [23]. It has
limited nutritional value but is filling, and it is given
to the child regularly and in amounts that are likely
to displace breastmilk. Further, its early introduction
is related to poor growth in this population (fig. 3).
Anthropometric status is not significantly associated
with “age at introduction of any food” (or, to say it in
a different way, months of exclusive breastfeeding),
but is significantly associated only with the age at
introduction of those specific foods that are of par-
ticularly low nutritional quality and are potentially
contaminated, i.e., porridge and dawale.
There is no concordance between the timing of
introduction of the four types of food discussed here.
Many caregivers who introduce mzuwula early do not
introduce dawale early, and vice versa, and therefore
there can be differences in the relationship between
age at introduction and anthropometric measurements
for the different foods. Although the relationships may
indicate causality, there is no evidence from this dataset
to support a causal relationship. The early introduction
of porridge and dawale may be associated with other
harmful but unobserved behaviors, or it may be a
general marker of food insecurity, maternal education,
or some other macro-level variable. It also may be an
example of reverse causality if children who are grow-
ing poorly are given infant foods earlier by concerned
caregivers. If that is the case, the early introduction
of non-breastmilk foods did not improve growth in
these children, who up to 4 years later were still lagging
behind their peers. Future research in the region, using
longitudinal rather than cross-sectional sampling, will
address this question.
A recent study that compared infant-feeding patterns
and child growth in Ghana, Peru, and India found that
there was no significant difference in the risk of death
between exclusively breastfed and predominantly
breastfed infants, and that nonbreastfed infants had
a significantly greater risk of death than those who
were exclusively or predominantly breastfed [24]. The
authors of this study concluded that the risk of not
breastfeeding needs to be taken into account when
advising HIV-infected mothers about their infant-feed-
ing options, and that rather than focusing on exclusive
breastfeeding in areas where predominant breastfeed-
ing is the norm, continued high rates of breastfeeding
should be encouraged. Our findings support the idea
of promoting predominant breastfeeding and the need
to identify specific infant-feeding practices that may
be particularly problematic, such as early introduction
of porridge. Breastfeeding provides other important
non-growth-related benefits to both mother and child,
such as reduced child infections and delayed maternal
fertility postpartum.
Whether or not the relationships are causal (or
reverse-causal), the biological plausibility of the rela-
tionships and the importance of discouraging early
introduction of foods, encouraging exclusive breast-
feeding or predominant breastfeeding, and, in par-
ticular, discouraging early introduction of porridge
and dawale should be priorities in nutrition programs
in Ekwendeni. The next step in this project will be to
use these findings to develop an appropriate exclusive
breastfeeding campaign and to monitor the success
of the campaign in terms of changing behavior and
improving child health. An additional complicating
factor in Malawi is the recommendation of exclusive
FIG. 3. Least-square means of weight-for-height and weight-
for-age in children according to age at first introduction of
porridge or dawale
n
n
n
:SCORE
!GEATINTRODUCTIONOFDAWALEORPORRIDGEMO
n n
7EIGHT
FORAGE
7EIGHT
FORHEIGHT
0ORRIDGE
$AWALE
0ORRIDGE
$AWALE
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
98
R. Bezner Kerr et al.
breastfeeding in the face of HIV infection rates. The
World Health Organization continues to recommend
exclusive breastfeeding unless replacement feeding
is “acceptable, feasible, affordable, sustainable and
safe,” which is not the case for most households in the
Malawian context. Nonetheless, the current recom-
mendation of the Ekwendeni Hospital staff for mothers
who are HIV-positive is exclusive breastfeeding for 6
months followed by abrupt weaning. The recommen-
dation is based on studies indicating that this method
reduces the transmission of HIV [25]. Informants in
follow-up qualitative research have expressed confu-
sion about what breastfeeding method is the best for
their babies in the face of conflicting messages. Abrupt
weaning is against current common practice and
suggests that breastmilk becomes unsafe. Given the
prevalence of herbal remedies such as mzuwula that are
used to protect infants, mothers may consider provid-
ing different herbal remedies to mitigate the dangers
of infected breastmilk.
Because extensive promotion of exclusive breastfeed-
ing appears to have had limited success in this region,
we are launching a new phase of participatory research
with households to encourage exclusive breastfeeding
through small group discussions involving all house-
hold members who influence child nutrition, includ-
ing grandmothers, mothers, and fathers. These small
groups will try to use a problem-solving approach to
encourage exclusive breastfeeding. If successful, the
model (first, find the reasons why exclusive breast-
feeding is not practiced; second, find out who has the
decision-making power to change the practice; and
third, consider which particular foods should be tar-
geted for reduction) may be used elsewhere in Malawi
and in other countries where conventional exclusive
breastfeeding promotional methods have not been
successful.
Acknowledgments
We could not have done this research without the sup-
port of many others: Tanya Trevors, Laifolo Dakishoni,
Lizzie Shumba, Rodgers Msachi, Solomon Mkumbwa,
Angela Shonga, Keston Ndlovu, the survey enumera-
tors, and all members of the Farmer Research Team
and the Nutrition Research Team. Ekwendeni Hospital
staff provided helpful logistical support. David Ryan
carried out very helpful data entry and analysis. David
Pelletier provided useful feedback throughout the
research. Finally, we thank the participating members
of the Soils, Food and Healthy Communities Project
in Ekwendeni, who made this work possible. This
work was carried out with the aid of a grant from the
International Development Research Centre, Ottawa,
Canada, and Presbyterian World Service and Develop-
ment, Toronto, Canada.
References
1. UNICEF. Strategy for improved nutrition of chil-
dren and women in developing countries. New York:
UNICEF, 1990.
2. Caulfield LE, de Onis M, Blossner M, Black RE. Under
-
nutrition as an underlying cause of child deaths associ-
ated with diarrhea, pneumonia, malaria, and measles.
Am J Clin Nutr 2004;80:193–8.
3. Fifty-fourth World Health Assembly. Infant and young
child nutrition. World Health Assembly 54.2, Agenda
item 13.1,18 May 2001.
4. UNICEF. Progress for children: A report card on
nutrition. No. 4, May 2006. New York: UNICEF, 2006.
Available at: http://www.unicef.org/progressforchildren/
2006n4/. Accessed 10 December 2006.
5. UNICEF. State of the world’s children 2005. New York:
UNICEF, 2005.
6. NSO/Macro International. Malawi Demographic and
Health Survey 1992. Zomba: Malawi National Statistical
Office (NSO) and ORC Macro, 1994.
7. NSO/Macro International. Malawi Demographic and
Health Survey 2000. Calverton, Md, USA, and Zomba,
Malawi: National Statistical Office (NSO) and ORC
Macro, 2001.
8. Bezner Kerr R. Informal labor and social relations in
northern Malawi: The theoretical challenges and impli-
cations of ganyu labor for food security. Rural Sociol
2005;70:67–187.
9. Centre for Social Research. Malawi social indicators
survey 1995: A survey into the state of health, nutri-
tion, water and sanitation and education of children in
Malawi. Zomba: Centre for Social Research, 1996.
10. Shrestha RM. Breast-feeding and weaning practices in
urban areas of Malawi. Lilongwe: Ministry of Health/
UNICEF, 1989.
11. Vaahtera M, Kulmala T, Hietanen A, Ndekha M, Cull
-
inan T, Salin ML, Ashorn P. Breastfeeding and comple-
mentary feeding practices in rural Malawi. Acta Paediatr
2001;90:328–32.
12. UNAIDS. UNAIDS/WHO AIDS epidemic update.
Available at: http://www.unaids.org/epi/2005/doc/
report_pdf.asp. Accessed 10 December 2006.
13. Bentley ME, Corneli AL, Piwoz E, Moses A, Nkhoma J,
Carlton Tohill BC, Ahmed Y, Adair L, Jamieson DJ, van
der Horst C. Perceptions of the role of maternal nutri-
tion in HIV-positive breast-feeding women in Malawi.
J Nutr 2005;135:945–9.
14. Blum L, Pelto PJ, Pelto GH, Kuhnlein HV. Community
assessment of natural food sources of vitamin A: Guide-
lines for an ethnographic protocol. Ottawa and Boston,
MA, USA: International Nutrition Foundation for
Developing Countries and International Development
Research Centre, 1997.
15. Humphries S, Gonzales J, Jiminez J, Sierra F. Searching
for sustainable land use practices in Honduras: Lessons
Delivered by Publishing Technology to: Guest User IP: 162.218.208.135 on: Tue, 14 Oct 2014 01:18:24
Copyright (c) Nevin Scrimshaw International Nutrition Foundation. All rights reserved.
99
Breastfeeding and mixed feeding practices in Malawi
from a programme of participatory research with
hillside farmers. Agricultural Research and Extension
Network Paper 2000:104.
16. Miles MB, Huberman AM. Qualitative data analysis: An
expanded sourcebook. Thousand Oaks, CA, USA: Sage
Publications, 1994.
17. Patton MQ. Qualitative evaluation and research meth
-
ods. Newbury Park, Calif, USA: Sage Publications,
1990.
18. Hamill PVV, Drizid TA, Johnson CL, Reed RB, Roche
AF, Moore WM. NCHS growth curves
for children
birth–18 years. Vital and Health Statistics Series
11, No.
165. DHEW Publication No. (PHS) 78-1650. Hyattsville,
MD, USA: US Department of Health, Education, and
Welfare, Public Health Service, 1977.
19. Smale M, Phiri ADK. Institutional change and discon
-
tinuities in farmers’ use of hybrid maize seed and ferti-
lizer in Malawi: Findings from the 1996–97 CIMMYT
MoALD survey. CIMMYT Economics Working Paper
98-01. Mexico City: CIMMYT, 1998.
20. Bezner Kerr R. Contested knowledge and disputed
practice: Maize and groundnut seeds and child feeding
in northern Malawi. PhD Dissertation, Department of
Development Sociology, Cornell University, Ithaca, NY,
USA, 2006.
21. WHO Collaborative Study Team on the Role of Breast
-
feeding on the Prevention of Infant Mortality. Effect
of breastfeeding on infant and child mortality due to
infectious diseases in less developed countries: A pooled
analysis. Lancet 2000;355:451–5.
22. Sibeko L, Dhansay MA, Charlton KE, Johns T, Gray-
Donald K. Beliefs, attitudes, and practices of breastfeed-
ing mothers from a periurban community in South
Africa. J Hum Lact 2005;21:31–8.
23. Simango C. Isolation of Escherichia coli in foods. Cent
Afr J Med 1995;41:181–5.
24. Bahl R, Frost C, Kirkwood BR, Edmond K, Martines K,
Martines J, Bhandari N, Arthur P. Infant feeding pat-
terns and risks of death and hospitalization in the first
half of infancy: Multicentre cohort study. Bull World
Health Organ 2005;83:418–26.
25. Iliff PJ, Piwoz EG, Tavengwa NV, Zunguza CD, Marinda
ET, Nathoo KJ, Moulton LH, Ward BJ, Humphrey JH;
ZVITAMBO study group. Early exclusive breastfeeding
reduces the risk of postnatal HIV-1 transmission and
increases HIV-free survival. AIDS 2005;19:699–708.