Article

Food and Nutrition Bulletin

2015, Vol. 36(2) 138-153
ª The Author(s) 2015
Identifying the Sociocultural Reprints and permission:
sagepub.com/journalsPermissions.nav
Barriers and Facilitating Factors DOI: 10.1177/0379572115586784
fnb.sagepub.com

to Nutrition-related Behavior
Change: Formative Research
for a Stunting Prevention
Program in Ntchisi, Malawi

Stephen Kodish1, Nancy Aburto2, Mutinta Nseluke Hambayi3,

Caitlin Kennedy1, and Joel Gittelsohn1

Abstract
Background: As the Scaling Up Nutrition (SUN) movement gains momentum globally, more
attention and resources are being given to integrated nutrition interventions. In 2013, the Government
of Malawi, with support from the World Food Programme and partners, initiated such an intervention
in Ntchisi District. Aimed to reduce the prevalence of stunting, the intervention has several com-
ponents, including the provision of a small-quantity, lipid-based nutrient supplement (SQ-LNS) for
children aged 6 to 23 months.
Objective: This paper describes formative research findings derived from a Rapid Assessment
Procedures (RAP) approach to inform the integrated nutrition intervention.
Methods: With a three-phase, emergent research design, this study utilized ethnographic methods
including in-depth interviews, direct meal observations, and full-day child observations. Free lists and
pile sorts were conducted to define food and illness domains through cultural domain analysis. Par-
ticipants included community leaders, caregivers, health surveillance assistants, and children aged 6 to
23 months.
Results: Community members felt that nutrition-related illnesses were less salient and threatening
than other illnesses, and food quality was less important than food quantity. Household food allocation
occurred in predictable patterns and varied by type of household member and season. Considered an
energy-giving food, the SQ-LNS was accepted, but health education and communications tailored to
local understanding of nutrition and health are necessary to ensure its appropriate utilization.

1
Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA
2
United Nations World Food Programme Headquarters, Rome, Italy
3
United Nations World Food Programme Malawi Country Office, Lilongwe, Malawi

Corresponding Author:
Stephen Kodish, Social and Behavioral Interventions Program, Department of International Health, Johns Hopkins Bloomberg
School of Public Health, 615 N. Wolfe St., Baltimore, MD 21205, USA.
Email: skodish1@jhmi.edu
Kodish et al 139

Conclusions: Tailoring a communications strategy to Ntchisi, Malawi could only be done through
formative research to understand the sociocultural factors influencing nutrition-related behaviors. A
RAP approach allowed for a comprehensive understanding of this local environment.

Keywords
ethnographic methods, formative research, small-quantity, lipid-based nutrient supplement (SQ-LNS),
Malawi, prevention of stunting, Scaling Up Nutrition (SUN) movement

Background interventions available to prevent stunting, nutri-

tion programs commonly focus on promoting
Chronic malnutrition is estimated to affect 165
optimal complementary feeding practices for
million children globally, with lifelong and inter-
children aged 6 to 23 months. In a systematic
generational consequences for the health of indi-
review of behavior change interventions to pro-
viduals, communities, and countries.1 In recent
mote complementary feeding10, researchers con-
years, attention has focused on preventing stunt-
cluded that effective interventions are derived
ing, an indicator of chronic undernutrition.2 This
from well-planned and thorough formative
attention has been channeled into the Scaling Up
research to inform intervention design and imple-
Nutrition (SUN) movement, to which 50 countries mentation. This type of research is carried out
were committed as of mid-2014.3,4 With unprece-
prior to implementation of intervention, with the
dented political commitments in place and donor
goal of collecting detailed information to gain an
money more available than once existed for mater-
in-depth understanding of both the target popula-
nal and child nutrition, interventions to address this
tion and the local context where the intervention
issue are being rolled out in an increasing number
will take place.11 Understanding the array of
of low- and middle-income settings.
sociocultural and context-specific determinants
There is a strong scientific case that interven-
that impact key behaviors related to nutrition is
tion strategies designed to improve child nutrition important for intervention design.10,12-15
should be implemented during the first 1,000
Currently, there is no universally agreed-upon
days of life—from conception until 2 years.5
methodology for conducting the requisite forma-
Thus, SUN activities in any particular country are
tive research for undernutrition interventions.16
often composed of a combination of nutrition-
However, due to the relatively recent attention
specific and nutrition-sensitive interventions6 that
to specialized nutritious foods, such as micro-
include 13 interventions recommended to have
nutrient powders (MNP) and small-quantity,
the most substantial impact on maternal and child
lipid-based nutrient supplements (SQ-LNS), in
health.7 Many of those suggested interventions nutrition interventions that aim to enhance the qual-
require some level of behavior change by the
ity of complementary foods of young children,
maternal caregiver and other household mem-
there have been a growing number of formative
bers. Whether that means going to the health cen-
research studies described in the literature.16-22 The
ter for supplementation (e.g., vitamin A, iron,
methodologies of these studies vary greatly, mak-
zinc, multiple micronutrients) or adhering to
ing comparisons among them difficult and leaving
health promotions and education (e.g., exclu-
researchers without a consistent basis on which to
sively breastfeeding from 0 to 5 months, comple-
model new formative research studies that include
mentary feeding from 6 to 23 months, hygiene specialized nutritious foods.
promotion), SUN activities are largely behavior
change interventions.
There is ample evidence to support the use of
behavior change interventions to improve key Objective
health-related behaviors and child nutrition out- Our objective in this formative research was to
comes.7-9 Given the range of behavior change inform the design and implementation of a
140
Government of Malawi nutrition intervention that
aims to prevent stunting during the first 1,000 days
of life through a series of interventions, including
the provision of a small-quantity (20-g), lipid-
based nutrient supplement (SQ-LNS) to be con-
sumed by every child from 6 to 23 months of age
once a day either directly out of its sachet or for-
tified in a complementary food. The SQ-LNS used
for this research was Nutributter, made by Nutriset
(Maulanay, France). Each reference to ‘‘SQ-LNS’’
hereafter refers to this particular small-quantity
(20-g) product unless otherwise specified. Our
research questions were as follows:
 What cultural perceptions and household
behaviors exist in relation to childhood ill-
nesses, concepts of healthy growth and
development, and food utilization?
 How can an ethnomedical model, which
describes a local body of knowledge about
a specific illness or group of illnesses23, be
developed for salient nutrition-related ill-
nesses in this setting?
 What are community members’ attitudes
toward an SQ-LNS that will be introduced
as part of the integrated nutrition interven-
tion in this setting?
In answering these questions, we sought to
also pilot a methodological approach developed
for the World Food Programme that could serve
as a template for formative research in other inter-
ventions with specialized nutritious foods in other
settings.
Methods
Study Setting
Data were collected from February until May
2013 in all seven traditional authorities (geo-
graphic areas) of Ntchisi District, which lies in
the Central Region of Malawi and is home to
approximately 250,000 residents. Nearly 70% of
both men and women engage in agricultural
activities as their primary livelihood. Only 6.4%
of the population has access to electricity.
Ntchisi children under 5 years of age suffer
from a high disease burden. A malaria-endemic
area, the Central Region has a mortality rate
Food and Nutrition Bulletin 36(2)
among children under five of 129 per 1,000 live
births, as well as the highest prevalence of anemia
(64%) and diarrhea (20%) in this age group in the
country.24 Where two-thirds (66.3%) of the pop-
ulation reported suffering from food shortage dur-
ing the year, Ntchisi has a 58.1% prevalence of
stunting among children under five, the second
highest in the country among 31 districts.25
Although nearly all children under 2 years of
age are breastfed (99%), a smaller percentage is
exclusively breastfed until 6 months (71%) or
continuously breastfed until 2 years (76.8%).
Among all children aged 6 to 23 months surveyed
in this region, only 30.2% received food from
more than four food groups in the previous
24 hours, and just 52.9% met minimum meal fre-
quency standards. Complementary foods are
composed primarily of grains, which were con-
sumed more than any other food group by chil-
dren aged 6 to 23 months.24
Overall Study Design
The formative research design was based on a Rapid
Assessment Procedures (RAP) approach.26,27 The
design drew from a RAP manual that was
recently developed for the World Food Pro-
gramme to use during programs that utilize spe-
cialized nutritious foods. This manual, which is
being piloted in various World Food Programme
program contexts, has four sections: 1) situational
analysis, 2) selection of an appropriate specia-
lized nutritious food for a given setting, 3) intro-
ducing the specialized nutritious food to the
community/social marketing, and 4) monitoring
and evaluation. The tool uses mixed methods to
aid practitioners in successfully choosing, intro-
ducing, and monitoring specialized nutritious
foods as part of integrated nutrition programs (see
Note 1).
Drawing from sections 2 and 3 of that manual,
this formative research followed a three-phase,
emergent design that was iterative in nature.
Designed to be exploratory, without precon-
ceived hypotheses but instead with an open-
ended, flexible, and inductive approach to data
collection and analysis28, this study used mixed
methods to collect multiple forms of data. Phase 1
explored the food, illness, and health domains of
Kodish et al
caregivers, as well as household food preparation
and feeding practices. Phase 2 collected ethno-
graphic data and built off phase 1 findings to
develop an ethnomedical model of nutrition-
related illness through cultural domain analysis
for illustrating cultural perceptions of the Chewa
medical system. This phase also introduced an
SQ-LNS to households as part of an 8-week feed-
ing trial. Phase 3 was designed to capture care-
giver experiences during the trial and develop
community-developed communications strate-
gies for the effective promotion of the SQ-LNS.
Further details of the study design are described
elsewhere (see Note 1).
Sampling
We recruited participants through a two-tiered,
purposive sampling strategy.29 To ensure the
study collected data from all seven traditional
authorities in Ntchisi, the district nutrition team
first identified an equal number of representative
villages in each area. Once these villages were
identified and village chiefs provided permission
for community participation, information-rich
participants who met the specific criteria described
below were recruited for participation.28,30
In-depth interviews, direct observations, and full-day
child observations. Community leaders, caregivers
of children aged 6 to 23 months, and households
with children aged 6 to 23 months were purpo-
sively sampled.29 To identify community leaders,
health surveillance assistants, who were paid gov-
ernment employees working at the 11 Ntchisi
health centers, helped identify village chiefs who
could speak not only about themselves but also
about their communities with deep insight.30
Working together, the health surveillance
assistants and community leaders identified care-
givers of children aged 6 to 23 months who could
offer rich information about both the Chewa cul-
ture of central Malawi and young child health, as
well as represent a diverse interview sample stra-
tified by sex (21 women and 5 men), location of
residence within the district (equal representation
from all seven traditional authorities of the dis-
trict), and number (range, one to nine; median,
three) and age (youngest child aged 6 to
141
Table 1. Summary of Ethnographic Methods Used in
This Study.
Method Description
In-depth One-on-one, open-ended,
interviews exploratory interview with an
informant
Pile sorts Respondents are asked to sort
cards, each containing the name
of an item (e.g., an illness), into
piles so that items in a pile are
more similar to each other than
they are to items in separate piles
Free lists Simple listing activity to elicit items
specific to the cultural domain of
a participant (e.g., illnesses and
foods)
Direct Midday and evening household meal
observations observations using an
observation guide to record
events of food preparation,
household feeding, and food
allocation
Full-day child A form of direct observation during
observations which a child aged 6 to 23
months is watched throughout
the entire day to record his or
her eating behaviors
23 months) of children31,32 to ensure inclusion
of different perspectives. Final sample sizes
were based on the amount of data needed to
reach data saturation, i.e., the point at which
additional data collection no longer generated
new understanding.33,34
Pile sorts and free lists. Caregivers were purpo-
sively sampled, as described above.31,32 Data
were collected using guidelines proposed by
Weller and Romney35, who suggest that 20 to
30 free list informants are sufficient for coherent
domains, such as causes of illnesses, and 30 to
40 pile sort informants will produce reliable data.
The caregivers who free listed were different
from those who pile sorted.
Data Collection
Tables 1 and 2 provide an overview of the meth-
ods and sample sizes described below. Data were
collected in all seven traditional authorities of
Ntchisi during each phase.
142 Food and Nutrition Bulletin 36(2)

Table 2. Summary of Data Collection Efforts by Phase of Formative Research.

Phase Method Type of participant No. of participants

1 In-depth interviews Community leaders 7

In-depth interviews Caregivers 11
Mealtime direct observations Householdsa 19
Free lists Caregivers 42
Full-day child observations Children 6–23 mob 7
2 Pile sorts Caregivers 37
Mealtime direct observations Householdsa 17
Free lists Caregivers 22
Full-day child observations Children 6–23 mob 7
3 Mealtime direct observations Householdsa 19
In-depth interviews Caregivers 15
In-depth interviews Health Surveillance Assistants 7
Full-day child observations Children 6–23 mob 7
Total sample 167c
a
The same 19 households were observed in each phase, except that 2 households were not observed in phase 2 because of
impassable roads during the rainy season.
b
The same seven children were observed in each phase.
c
This total represents the number of different participants in all three phases of the research, not merely the sum of each row
due to repeated observations of the same participants.

Phase 1 (exploring cultural perceptions and behaviors
of food and illness). Phase 1 used in-depth inter-
views and free lists (e.g., ‘‘List all of the illnesses
that young children suffer from in this commu-
nity’’) to explore the cultural domains of commu-
nity members. Direct observations of midday and
evening meals were used to understand household
food preparation and feeding practices. Full-day
child observations documented the dietary beha-
vior of children aged 6 to 23 months in this setting
from 8:00 A.M. until sundown.
Phase 2 (developing an ethnomedical model of
nutrition-related illness). Phase 2 collected further
cultural data on food and illness domains using
free lists of illness causes (e.g., ‘‘Please list all of
the different causes of kunyentchera (wasting)’’
to finalize construction of an ethnomedical model
of nutrition-related illness. Unconstrained pile
sorts of salient foods for young children were
completed to determine local food groupings.35,36
A second round of direct meal observations and
full-day child observations was conducted with
the same households observed in phase 1 with
special attention to infant and young child feed-
ing practices after the introduction of the
SQ-LNS.
Phase 3 (understanding caregiver experiences with an
SQ-LNS). Phase 3 used in-depth interviews to elu-
cidate caregivers’ experiences with the SQ-LNS,
specifically identifying barriers and facilitating
factors to its appropriate use. For triangulation,
in-depth interviews with health surveillance
assistants were also conducted to corroborate and
help explain findings from phases 1 and 2. A third
round of meal observations and full-day child
observations were completed with the same
households from the previous two phases to mini-
mize reactive behaviors and to observe feeding
practices on three separate days over the
3-month study period.
Data Analysis
In-depth interview data. Textual data were trans-
lated and transcribed verbatim from Chichewa
digital recordings into written English, maintain-
ing local terminology.37 Atlas.ti version 7.0 com-
puter software (Scientific Software Development,
Berlin, Germany) was used for data management
and coding. We inductively analyzed the textual
data drawing from the procedures of Grounded
Theory.38 The process began with line-by-line
coding until strong analytic directions were
Kodish et al 143

Table 3. Guiding Analytic Categories for Textual Data Analysis of Caregiver Interviews.a

Category Number Phase 1 analytic categories Phase 3 analytic categories

1 Culture and context Culture and context

2 Illnesses in the community Lives of young children aged 6–23 mo
3 Meal preparation and eating/feeding Caregiver experiences with SQ-LNS
4 Food sharing Child’s response to SQ-LNS
5 Developing a nutrition program Sharing of SQ-LNS
6 Seasonality Caregiver perceptions of SQ-LNS
7 Potential barriers to SQ-LNS usage Effective promotion of SQ-LNS
8 Facilitating factors to SQ-LNS usage Specific message development
9 –– Extra information for program design

SQ-LNS, small-quantity, lipid-based nutrient supplement.

a
Phase 1 interviews explored sociocultural characteristics in general, whereas phase 3 interviews focused more specifically on
caregiver experiences with the SQ-LNS during a small home-feeding trial.

identified and a codebook was developed (Table 3).
Using that framework, focused coding was then
applied to the data set to synthesize and explain
larger segments of text. Upon development of ana-
lytic categories, axial coding reassembled the data
to give coherence to the emerging analysis.39 Con-
stant comparisons were made throughout the cod-
ing process33, and memos, or analytic notes, were
written to draw comparisons among codes, cate-
gories, and quotations from text.38
Direct observation data. Both qualitative (descrip-
tive data related to types and amounts of foods
consumed) and quantitative (frequencies of giv-
ing or receiving food) information was recorded
with standardized forms. Observations were
recorded at a minimum every 5 minutes and more
frequently when events related to the research
questions occurred.40 The research team aggre-
gated the descriptive field notes from the obser-
vations, applied focused codes, and drew out key
themes salient to the research questions.38
Free list data. Anthropac, version 4.98 (Analytic
Technologies, Lexington, KY, USA), was used to
analyze lists of foods, illnesses, and causes of
illness. Items of salience (S)—a statistic account-
ing for rank and frequency—were generated, as
well as a rank-ordered identification of the most
commonly mentioned items.41 These items of sal-
ience were merged with findings from interview
data to construct an ethnomedical model of
nutrition-related illness.
Pile sort data. After the participants sorted food
cards into piles, the numbers identifying the cards
being coded for analysis were entered into the
software and aggregate proximities were calcu-
lated, yielding 18 18, item-by-item matrices
of food items with cells indicating the proportion
of times two items appeared in the same pile
across participants. The aggregate proximity
matrices were then analyzed using multidimen-
sional scaling (MDS) as recommended by Weller
and Romney.35 Stress, taken as a goodness-of-fit
statistic in MDS analyses, was calculated for each
aggregate proximity matrix.
Ethical Approval
The study protocol was approved by the Institu-
tional Review Board of the Johns Hopkins
Bloomberg School of Public Health and the
Malawi National Health Sciences Research Com-
mittee. Adult participants provided informed oral
consent for themselves and assent on behalf of
child participants.
Results
Cultural Perceptions of Health, Foods,
and Illnesses
In Ntchisi, caregivers and community leaders
shared similar overall perceptions related to ill-
ness, food, and nutritional status. The data exhibit
a strong fit to the cultural consensus model42,
144
Figure 1. Multidimensional scaling map of Chewa young child food items (n ¼ 18).
supporting the assertion that, despite some indi-
vidual differences, the majority culture of Ntchisi
is relatively homogenous in these domains.
Caregivers defined ‘‘healthy growth and
development’’ (kukula mwa thanzi) as a child
who is happy, not getting sick frequently, gaining
weight, and eating a diet of ‘‘different food
groups’’ (zakudya za magulu or zakudya za
kasintchasintha). Consuming ‘‘different food
groups’’ was a salient theme and is considered
important for child health. Caregivers categorized
foods into four general categories which they
labeled as ‘‘fruits,’’ ‘‘vegetables,’’ ‘‘energy-giv-
ing’’ (e.g., porridge, groundnuts, SQ-LNS), and
‘‘rare’’ (e.g., meat, fish) (Figure 1 and Table 4).
When asked to describe challenges that they
face in the village, community members explained
that childhood illness is not considered a major
challenge in comparison with more formidable
barriers such as ‘‘lack of money,’’ ‘‘hunger,’’ and
‘‘long distance to the hospital.’’ Cultural domain
analysis results illustrate that the three illnesses
most salient to community members were
malaria (S ¼ 0.91), cough (S ¼ 0.72), and diar-
rhea (S ¼ 0.71). Caregivers explained that there
is stigma attached to many illnesses afflicting
young children, especially those that are
Food and Nutrition Bulletin 36(2)
nutrition-related, because they are feared to be
an indication of poor caring practices. This
stigma results in caregivers sometimes avoiding
health clinics and instead visiting traditional hea-
lers. Community members invoked both natura-
listic and personalistic causes in explaining
episodes of illness; they just as often pointed to
a creator who is responsible for illness as they
discussed poor health-related behaviors, such as
poor handwashing.
Understanding Prevention and Treatment
of Salient Nutrition-related Illnesses
Nutrition-related illnesses are significantly less
salient and are perceived to be less severe than
other childhood illnesses, especially other ill-
nesses related to fever (kutentha thupi), such as
malaria (S ¼ 0.91). Community members
described four manifestations of nutrition-related
illness: wasting (kunyenthchera, S ¼ 0.33), edema
(kutupikana/kutupa, S ¼ 0.32), lack of blood in
the body/anemia (kuchepa kwa magazi nthupi,
S ¼ 0.21), and stunted growth (kupilipizika/kupi-
nimbira, S ¼ 0.04 (Table 5). Community leaders
and caregivers explained that stunted growth is a
natural occurrence.
Kodish et al 145

Table 4. Description of Clusters Based on Multidimensional Scaling of Young Child Food Items.

Chichewa food
term English food equivalent Caregiver description of cluster

FRUITS
Yembe Mango ‘‘These are fruits that are found in this community and that children like
Gwafa Guava to eat.’’
Lalanje Orange Female caregiver, age 45
Nthochi Banana
ENERGY-GIVING
Dowe Green maize ‘‘These are foods that are so important to a child’s health and growth.’’
Mbatata Sweet potato Female caregiver, age 45
Nsima Cooked maize flour
Phala Porridge ‘‘These make you full and give you good health.’’
Phala la nsinjiro Porridge with groundnut Female caregiver, age 23
flour
SQ-LNS SQ-LNS
Chiponde Chiponde
Mtedza/sawa Groundnut
RARE
Nyama Meat ‘‘It is because these are rarely eaten since most of the households cannot
Nsomba Fish afford to buy them.’’
Female caregiver, age 27
VEGETABLES
Nkhwani Pumpkin leaves ‘‘These are vegetables and they make children healthy.’’
Bonongwe Amaranthus hybridus Female caregiver, age 28
Mpiru Mustard leaves
Chisoso/kazota Blackjack/Bidens pilosa ‘‘These are vegetables that we get from our gardens and are easily
found.’’
Female caregiver, age 40

Table 5. Nutrition-related Illnesses in Order of Salience to Caregivers.

English illness Rank order among

Chichewa nutrition-related illness term equivalent/explanation Salience all illnesses listed

Kunyentchera Wasting 0.328 5/65

Kutupikana/kutupa Edema/swelling 0.315 4/65
Kuchepa kwa magazi mthupi Anemia 0.210 7/65
Kupilipizika/kupinimbira Stunted growth 0.043a 13/65
a
A rank order of 13/65 may seem high, but the salience statistic illustrates the very low importance that caregivers give to their
children’s stunted growth.

Well people are different from each other . . . one
can be tall and the other can be short . . . it’s just
by nature that the person is like that . . . it’s just how
God created the child.
–Female caregiver of six children, in-depth
interview
We generated an ethnomedical model of
causes, illnesses, and treatments for these ill-
nesses from the data (Figure 2). The model
illustrates that a ‘‘lack of food in the body’’ is the
perceived primary cause of nutrition-related ill-
ness, highlighting a focus on dietary quantity, not
quality. The least salient nutrition-related illness,
stunted growth, is considered locally to stem
from ‘‘a lack of food in the body,’’ ‘‘genetic
inheritance,’’ and, less commonly, ‘‘frequent ill-
ness.’’ Compared with edema, wasting, and ane-
mia, there is less consensus among caregivers
around the causes of stunted growth.
146
Figure 2. Ethnomedical model of nutrition-related illness in Ntchisi. *Nsima is cooked maize flour. **Chiponde is
a locally produced ready-to-use therapeutic food (RUTF) used for the treatment of acute malnutrition in Ntchisi.
Caregivers do not hold distinct concepts of
prevention and treatment in relation to nutrition-
related illnesses as clearly as they do with other
illnesses, such as malaria and diarrhea, on which
previous extensive health promotion efforts in
Ntchisi have focused their attention. For those
illnesses, prevention is commonly understood in
concrete terms, such as using an insecticide-
treated bednet for malaria, or practicing hygienic
hand-washing behavior for diarrhea. For nutrition-
related illnesses, concepts of prevention focus on
dietary quantity (e.g., ‘‘lack of food in the body’’),
whereas those of treatment highlight the perceived
importance of dietary quality (e.g., ‘‘breastfeed-
ing,’’ ‘‘balanced diet,’’ ‘‘drinking milk,’’ ‘‘fruits
and vegetables,’’ and ‘‘eating vegetables’’).
Intra- and Interhousehold Food Allocation
Patterns and Behavior
In Ntchisi, food is allocated in predictable pat-
terns that vary according to the time of year and
Food and Nutrition Bulletin 36(2)
type of household member. Direct observations
revealed that caregivers favor children aged 6 to
23 months in terms of the food they receive, the
time they receive it, and the individual attention
they are given during meal times. During non-meal
times, these children are given extra maize-based
porridge, usually at mid-morning and mid-
afternoon. During midday and early evening meal
times, they eat with their caregivers. In many
instances, these children eat before the rest of the
family in the evening due to early bedtimes. Data
indicate that this caregiver favoritism provides
additional and more frequent food rations to chil-
dren aged 6 to 23 months throughout the day. When
possible, caregivers add special complementary
foods, such as soybeans or groundnuts, to children’s
diets, a strategy promoted by the health surveillance
assistants and government. Forced hand-feeding of
children aged 6 to 23 months was commonly seen
during meal observations.
Intrahousehold food sharing—eating from
shared plates, caregiver-to-young-child feeding,
Kodish et al 147

Table 6. Projected Food and SQ-LNS Sharing Practices by Season.a

Seasons and sharing

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Season Lean Harvest Moderate food supply Lean

Hot rainy Cold Hot dry Hot rainy

Wedding and initiation ceremonies, so more sharing

at community level and less sharing at household
level
Sharing of Theft from others’ Theft of food at household level
food fields and houses to sell for money to buy alcohol
due to hunger directly after harvest

Highest sharing of food at Highest sharing of food

household level at community level

Sharing of Highest sharing of SQ- Highest sharing of SQ- Moderate sharing of SQ-LNS at both
SQ-LNS LNS at household level LNS at community level household and community levels

SQ-LNS consumption
Other may be irregular due to
availability of other
foods
SQ-LNS, small-quantity, lipid-based nutrient supplement.
a
Because the formative research was conducted during the lean season only (January–April), data from the May–December
calendar were derived from descriptive self-report.

and child-to-child food sharing—is influenced by
many factors. Caregivers explained that sharing
patterns vary by season as well as by the age and
sex of household members. We observed exten-
sive eating from shared plates during the lean sea-
son from January to April when food stocks are
carefully being conserved until the harvest. After
the harvest (May to July), husbands reportedly
steal food from the household to sell for money
to buy alcohol. During this same time, caregivers
explained that less intrahousehold food sharing
occurs because an abundance of food allows for
individual portions during meal times (Table 6).
Similarly, interhousehold food sharing—food
sharing among households, barter of food for other
foods or goods, food sharing at community gather-
ings, and theft—also varies by season. Caregivers
explained that these practices peak during and
soon after the harvest when there is an abundance
of food for barter. Also, they described interhouse-
hold food sharing to be common practice during
times of celebration, such as initiations and
weddings. However, during the lean season, inter-
household theft of food occurs from both fields
and households. We observed children of similar
ages who tend to live, play, and eat together—a
situation that facilitates food sharing among chil-
dren of similar age and sex in village settings.
Impressions of the SQ-LNS
Caregivers had positive impressions of the
SQ-LNS, and children aged 6 to 23 months were
highly accepting of its flavor, a driver of product
sharing and overuse that occurred within and
between households. Community members sug-
gested that sustained compliance will be more fea-
sible with development of culturally appropriate
packaging that includes the sachet language in
Chichewa as well as locally developed images and
clearer instructions for use. Most people asked for
larger quantities of the SQ-LNS, comparing it to
the locally produced ready-to-use therapeutic food
(RUTF) Chiponde, which is already being used
148
Figure 3. Comparison of sizes of the locally produced
ready-to-use therapeutic food (RUTF), Chiponde
(92 g), and the small-quantity, lipid-based nutrient
supplement (SQ-LNS) (20 g).
in Ntchisi for the treatment of acute malnutri-
tion and has a weight of 92 g (Figure 3). Even
after the research team explained the reasons for the
SQ-LNS, confusion about its purpose vis-à-vis that
of Chiponde existed among caregivers.
I think it [SQ-LNS] is a medicine . . . because let’s
say if a child is suffering from malnutrition and they
start taking this, then they can get better, and the
child can become healthy.
–Female caregiver of child aged 12 months,
in-depth interview
When asked directly for their perceptions of
the SQ-LNS, 10 caregivers said that the product
was a medicine, 3 that it was a food, and 5 that it
had qualities of both a food and a medicine. They
either called the product by its name, referred to it
as Chiponde, or called it ‘‘chakudya,’’ Chichewa
for ‘‘food.’’ Nine of 12 caregivers indicated that
they would pay a nominal price (mean, 56 MWK/
0.14 USD; median, 35MWK/0.09 USD) for a
sachet of the product if it was introduced through
a market-based system, citing their children’s
health as the reason for this willingness.
Discussion
This paper describes a RAP approach that pro-
duced a holistic account of perceptions of illness,
food, and health; household food utilization
patterns; and SQ-LNS acceptability during
Food and Nutrition Bulletin 36(2)
formative research within the SUN movement.
These findings, which shed light on the sociocul-
tural context of Ntchisi, Malawi, should be used
as key considerations in the design, implementa-
tion, and monitoring of interventions.
Triangulating in-depth interview, free list, and
pile sort data from the RAP methodology allowed
us to construct an ethnomedical model of causes,
illnesses, and treatments for nutrition-related ill-
nesses in Ntchisi (Figure 2). Program staff can
use this model to develop health education mate-
rials that directly address community perceptions
of the types and causes of nutrition-related
illnesses. Distinct concepts of prevention and treat-
ment are not as clearly defined for nutrition-related
illnesses as they are for other illnesses. Health pro-
motion teams could thus consider drawing on cul-
tural metaphors from local bodies of knowledge
related to other illnesses and applying them to pre-
vention of nutritional illness with an SQ-LNS;
such an approach may resonate with community
members.43 For example, health messaging could
compare the consumption of an SQ-LNS for pre-
vention of chronic malnutrition to sleeping under
an insecticide-treated bednet for protection against
malaria. Such messaging would be most effective
if it built off the existing bodies of knowledge
around nutrition-related illness, while not trying
to deconstruct them into entirely new cognitive
frameworks unfamiliar to community members.
It is far easier to introduce, reinforce, or build on
existing frameworks than it is to try to change
well-established perceptions.44
Findings from cultural domain analysis high-
lighted community members’ low perceived
threat of nutrition-related illnesses in general, and
child stunting in particular. This finding may
prove to be a major barrier to messaging that reso-
nates with caregivers and thus compliance with the
SQ-LNS. Developing messaging with consider-
ation of this finding and aligned with behavior
change theory will be important for effective
development of Behavior Change Communica-
tion (BCC) in this setting. The Persuasive Health
Message (PHM) framework (Figure 4), which
includes elements from several well-known
behavioral models, would be especially useful for
message development.45 To develop messages
within this framework, Witte46 recommends the
Kodish et al 149

CONSTANTS TRANSIENTS
Threat
-Suscepbility
-Severity Message Goals

Salient Beliefs
Efficacy
-Response Efficacy
-Self-Efficacy Salient Referents
Persuasive
Cues Message
-Message Culture
-Source
Environment
Audience Profile
-Demographics Preferences
-Psychographics
-Customs, Values

Figure 4. The Persuasive Message Framework. Source: Witte46

following three steps: 1) identify salient commu-
nity beliefs and specific behavioral goals, 2)
develop an audience profile, and 3) construct a
persuasive message that is framed to fit within
local bodies of understanding. This formative
work has already yielded rich information for
steps 1 and 2 in this process and provides further
information (e.g., nutrition-related illness risk per-
ception, self-efficacy toward using SQ-LNS, and
various cues for action) that can be used to develop
persuasive messages in the Ntchisi context.
Tailoring health education curricula and BCC
that address the determinants of home fortifica-
tion at multiple behavioral levels through varied
communication channels, as has been effective in
other similar programs, is also necessary in this
setting.47 Promotion of the SQ-LNS should link
the desired behaviors to more salient community
challenges; for example, the SQ-LNS could fight
off other childhood illnesses and thus reduce trips
to the hospital and cut healthcare costs for a
household. BCC efforts also need to sensitize the
community to the benefits of a 20-g SQ-LNS
when the perceived foremost cause of nutrition-
related illnesses in this setting is related to dietary
quantity (‘‘a lack of food in the body’’).
The RAP approach also highlighted factors
that may facilitate successful programming with
an SQ-LNS used for prevention. This study found
high acceptability of the SQ-LNS, similar to pre-
vious research.48-53 Modification of packaging
for specific cultural contexts based on formative
work has been successfully achieved in other set-
tings54 and should be considered in interventions
that utilize specialized nutritious foods, both in
Malawi and elsewhere. The sweet flavor of this
product, in settings such as Ntchisi where mono-
tonous dietary intake prevails, provides an incen-
tive for caregivers to fortify their children’s
complementary foods with it regardless of the
health benefits, potentially reducing the practice
of traditional forced hand-feeding due to poor
child appetite stemming from dietary monotony.
The high acceptability of the SQ-LNS was
reportedly one primary reason for its overuse (see
Note 2). To overcome this barrier to smooth pro-
gram implementation, a RAP approach can eluci-
date season-specific food-sharing practices that
may help to predict when and how SQ-LNS overuse
may also occur (Table 6), enabling more targeted
communications to community members of differ-
ent age groups and sexes. Positioning the SQ-LNS
within the already-existing cultural food domain of
caregivers during BCC and using emic terminology
to appeal to a local, rather than a biomedical, con-
cept of ‘‘healthy growth and development’’ may be
150
more familiar and thus likely to resonate with care-
takers. Also, using existing or preferred communi-
cation methods by local authorities (e.g., village
chiefs and health surveillance assistants) to deliver
salient, local messages to entire household units
will be important during a BCC campaign.13
By identifying local dietary practices, the find-
ings from the RAP built upon health behaviors
within the community for BCC development and
tailoring. For example, during the postharvest
period in Ntchisi, caregivers indicate that they
enrich flour-based maize porridge with soybeans
or groundnuts, a similar practice to home fortifi-
cation of complementary foods with an SQ-LNS.
Because this local practice already exists and is
readily understood to be a healthful way to feed a
child, BCC messaging should piggyback on this
existing behavior and related perceptions. By
contrast, the program may benefit by distancing
itself from other existing behaviors. For example,
targeted messages to concerned caregivers should
clarify the distinction between consumption of
Chiponde, the ready-to-use therapeutic food
(RUTF) used for treatment of people living with
HIV and malnourished children, and an SQ-LNS
used for prevention, with care not to create undue
stigma for those accessing such community-based
management of acute malnutrition programs.
This study has some limitations. There was
probably reactivity as a result of data collectors
observing household food preparation and eating
behaviors of community members. We employed
repeated observations of the same households in
order to minimize the negative manifestations of
reactivity and documented frequency of reactive
behaviors, as has been done in previous work.55
Also, although some meaning may have been lost
during the process of translation from Chichewa
into English, we used verbatim translations with
emic language to maintain local words and mean-
ings, as well as using a triangulation of both
methods and participants to help ensure data cred-
ibility.56,57 We also employed member checking,
a process in which we took key findings back to
the communities themselves in order to verify and
confirm the accuracy of the interpretations made
by the research team.56,58
Formative research has increasingly informed
international nutrition programming in recent
Food and Nutrition Bulletin 36(2)
years, and researchers and practitioners are begin-
ning to embrace this pre-implementation phase as
critical to successful design and implementation
of BCC programs.12,17,21 Just as a new monitor-
ing manual for home fortification has been devel-
oped for the intervention nutrition community59,
so too should one be developed to guide forma-
tive research and translation of findings into pro-
gram design. For the next steps of this program, a
BCC team should utilize the ethnomedical model
of nutrition-related illness to develop health mes-
sages that are aligned with local understanding
and that utilize the emic terminology of Ntchisi.
Table 6, which outlines projected SQ-LNS shar-
ing based on current food-sharing patterns,
should serve as a starting point for developing a
BCC strategy that aims to reduce the likelihood of
product sharing. Finally, momentum stemming
from the positive community feedback and high
acceptability of the SQ-LNS can be harnessed dur-
ing early social mobilization efforts. Considering
the complexities around home fortification, ethno-
graphic formative research is an important step in
thoughtful nutrition intervention design.
Acknowledgments
The authors gratefully acknowledge the Ntchisi com-
munity for participating in this study. Thanks also go
for the invaluable data collection efforts by our local
team, including Yamikani Kunashe, Evelyne Kam-
wendo, Ndapile Bwanausi, Chifundo Nsanjama, and
Eric Maonga. We would also like to sincerely thank
the World Food Programme head of suboffice in
Ntchisi, Trust Mlambo, and the Government of
Malawi, Ntchisi nutritionist James Mtonga for their
on-the-ground help and unwavering support through
the project. Finally, a special thanks to former Boston
Consulting Group team leader Patrick Cleary, who pro-
vided much in-country support during the planning of
the formative research.
Authors’ Contributions
Stephen Kodish and Joel Gittelsohn led the design, data
collection, and data analysis of the formative research.
Nancy Aburto, Mutinta Nseluke Hambayi, and Caitlin
Kennedy supported the overall project with continual
guidance, feedback, and support. Stephen Kodish
drafted the first version of the manuscript. Nancy
Aburto, Mutinta Nseluke Hambayi, Caitlin Kennedy,
and Joel Gittelsohn critically reviewed and revised the
Kodish et al
subsequent drafts prior to finalization. All authors read
and approved the final submission.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest
with respect to the research, authorship, and/or publi-
cation of this article.
Funding
The author(s) disclosed receipt of the following finan-
cial support for the research, authorship, and/or publi-
cation of this article: Stephen Kodish and Joel
Gittelsohn were consultants to the World Food Pro-
gramme, funded by the Children’s Investment Fund
Foundation, and received support for the conduct of
the research.
Notes
1. Kodish S. Gittelsohn J. Rapid Assessment Proce-
dures to develop and evaluate specialized nutritious
food programs: Applying qualitative research skills
v3.5. Copy in possession of the World Food Pro-
gramme: 1–109.
2. Kodish S, Aburto NJ, Dibari F, Nseluke Hambayi
M, Gittelsohn J. Patterns of household SQ-LNS uti-
lization in rural Malawi and Mozambique: Implica-
tions for interventions with specialized nutritious
foods. Matern Child Nutr (under review).
References
1. Black RE, Victora CG, Walker SP, Bhutta ZA,
Christian P, de Onis M, Ezzati M, Grantham-
McGregor S, Katz J, Martorell R, Uauy R; Mater-
nal and Child Nutrition Study Group. Maternal and
child undernutrition and overweight in low-
income and middle-income countries. Lancet
2013;382:427-51.
2. Maternal and Child Nutrition Study Group. Mater-
nal and Child Nutrition: Building momentum for
impact. Lancet 2013;382:372-5.
3. Bezanson K, Isenman P. Scaling Up Nutrition: A
framework for action. Food Nutr Bull 2010;31:
178-86.
4. Scaling Up Nutrition. A framework for action.
2014. Available at: http://scalingupnutrition.org/
sun-countries. Accessed 7 April 2015.
5. Bhutta ZA, Das JK, Rizvi A, Gaffey MF, Walker
N, Horton S, Webb N, Lartey A, Black RE.
Evidence-based interventions for improvement of
151
maternal and child nutrition: What can be done and
at what cost? Lancet 2013;382:452-77.
6. Ruel MT, Alderman H; Maternal and Child Nutri-
tion Study Group. Nutrition-sensitive interven-
tions and programmes: How can they help to
accelerate progress in improving maternal and
child nutrition? Lancet 2013;382:536-51.
7. Bhutta ZA, Ahmed T, Black RE, Cousens S,
Dewey K, Giugliani E, Haider BA, Kirkwood B,
Morris SS, Sachdev HPS, Shekar M. What works?
Interventions for maternal and child undernutrition
and survival. Lancet 2008;371:417-40.
8. Dewey KG, Adu-Afarwuah S. Systematic review
of the efficacy and effectiveness of complemen-
tary feeding interventions in developing countries.
Matern Child Nutr 2008;4:24-85.
9. Shi L, Zhang J. Recent evidence of the effective-
ness of educational interventions for improving
complementary feeding practices in developing
countries. J Trop Pediat 2011;57:91-8.
10. Fabrizio CS, van Liere M, Pelto G. Identifying
determinants of effective complementary feeding
behavior change interventions in developing coun-
tries. Matern Child Nutr 2014;10:575-92.
11. Gittelsohn J, Evans M, Story M, Davis SM, Met-
calfe L, Helitzer DL, Clay TE. Multisite formative
assessment for the Pathways study to prevent obe-
sity in American Indian schoolchildren. Am J Clin
Nutr 1999;69 S:767S-72S.
12. Bentley ME, Johnson SL, Wasser H, Creed-Kana-
shiro H, Shroff M, Fernandez Rao S, Cunningham
M. Formative research methods for designing cul-
turally appropriate, integrated child nutrition and
development interventions: An overview. Ann N
Y Acad Sci 2014;1308:54-67.
13. Gittelsohn J, Vastine AE. Sociocultural and house-
hold factors impacting on the selection, allocation,
and consumption of animal-source foods: Current
knowledge and application. J Nutr 2003;133:
S4036-41.
14. Kodish S, Rah JH, Kraemer K, de Pee S, Gittel-
sohn J. Understanding low usage of micronutrient
powder in the Kakuma Refugee Camp, Kenya:
Findings from a qualitative study. Food Nutr Bull
2011;32:292-303.
15. Kodish S, Monterrosa E. Formative research at the
forefront: Understanding context and behavior to
design culturally appropriate, nutrition interven-
tions. Sight Life 2013:27(2):16-20.
152
16. Tripp K, Perrine CG, de Campos P, Knieriemen M,
Hartz R, Ali F, Jefferds MD, Kupka R. Formative
research for the development of a market-based
home fortification programme for young children
in Niger. Matern Child Nutr 2011;7(suppl 3):
82-95.
17. Pelto GH, Armar-Klemesu M, Siekmann J, Scho-
field D. The focused ethnographic study ‘assessing
the behavioral and local market environment for
improving the diets of infants and young children 6
to 23 months old’ and its use in three countries.
Matern Child Nutr 2013;9(suppl 1):35-46.
18. Segre J, Winnard K, Hailu Abrha T, Abebe Y,
Shilane D, Lapping K. Willingness to pay for
lipid-based nutrient supplements for young chil-
dren in four urban sites of Ethiopia. Matern Child
Nutr 2012. doi: 10.1111/mcn.12022.
19. Paul KH, Muti M, Chasekwa B, Mbuya MNN,
Madzima RC, Humphrey JH, Stoltzfus RJ. Com-
plementary feeding messages that target cultural
barriers enhance both the use of lipid-based nutri-
ent supplements and underlying feeding practices
to improve infant diets in rural Zimbabwe. Matern
Child Nutr 2010;8:225-38.
20. Young SL, Blanco I, Hernandez-Cordero S, Pelto
GH, Neufeld M. Organoleptic properties, ease of
use, and perceived health effects are determinants
of acceptability of micronutrient supplements
among poor Mexican women. J Nutr 2010;140:
605-11.
21. Locks LM, Pandey PR, Osei AK, Spiro DS, Adhi-
kari DP, Haselow NJ, Quinn VJ, Nielsen JN. Using
formative research to design a context-specific
behaviour change strategy to improve infant and
young child feeding practices and nutrition in
Nepal. Matern Child Nutr 2013. doi: 10.1111/
mcn.12032.
22. Jefferds MED, Ogange L, Owuor M, Cruz K, Per-
son B, Obure A, Suchdev PS, Ruth LJ. Formative
research exploring acceptability, utilization, and
promotion in order to develop a micronutrient
powder (Sprinkles) intervention among Luo fami-
lies in western Kenya. Food Nutr Bull 2010;
31(suppl 2): S179-85.
23. Kleinman A. Patients and healers in the context of
culture. Berkeley, CA, USA: University of
California Press, 1980.
24. Demographic and Health Survey (DHS). Malawi
Demographic and Health Survey 2010. Available
Food and Nutrition Bulletin 36(2)
at: http://dhsprogram.com/pubs/pdf/FR247/
FR247.pdf. Accessed 7 April 2015.
25. National Statistical Office. Integrated Household
Survey 2010–2011: Household Socio-economic
Characteristics Report. Zomba, Malawi: National
Statistical Office, 2012.
26. Gittelsohn J, Pelto PJ, Bentley ME, Bhattacharyya
K, Jensen JL. Rapid Assessment Procedures
(RAP) — Ethnographic methods to investigate
women’s health. Boston, MA, USA: International
Nutrition Foundation, 1998.
27. Scrimshaw SCM, Hurtado E. Rapid Assessment
Procedures for nutrition and primary health care:
Anthropological approaches to improving pro-
gramme effectiveness. Tokyo: United Nations
University Press, 1987.
28. Creswell JW. Qualitative inquiry and research
design: Choosing among five approaches, 2nd
ed. Thousand Oaks, CA, USA: Sage Publications,
2007.
29. Maxwell J. Qualitative research design: An inter-
active approach, 2nd ed. Thousand Oaks, CA,
USA: Sage Publications, 2005.
30. Bernard HR. Research methods in anthropology,
2nd ed. Thousand Oaks, CA, USA: Sage Publica-
tions, 2006.
31. Patton MQ. Qualitative evaluation and research
methods, 3 rd ed. Thousand Oaks, CA, USA: Sage
Publications, 2002.
32. Sandelowski M. Combining qualitative and quan-
titative sampling, data collection, and analysis
techniques in mixed-method studies. Res Nurs
Health 2000;23:246-55.
33. Glaser B, Strauss A. The discovery of grounded
theory: Strategies for qualitative research. New
York: Aldine de Gruyter, 1967.
34. Morgan D. Focus groups as qualitative research,
2nd ed. Thousand Oaks, CA, USA: Sage Publica-
tions, 1997.
35. Weller SC, Romney AK. Structured interviewing.
Newbury Park, CA, USA: Sage Publications, 1988.
36. Brieger WR. Food groups in cultural perspective.
Trop Doct 1985;15:42-3.
37. Harris M. History and significance of the emic/
etic distinction. Annu Rev Anthropol 1976;5:
329-50.
38. Charmaz K. Constructing grounded theory: A
practical guide through qualitative analysis. Lon-
don: Sage Publications, 2006.
Kodish et al
39. Strauss A, Corbin J. Basics of qualitative research:
Grounded theory procedures and techniques. New-
bury Park, CA, USA: Sage Publications, 1990.
40. Harvey SA, Olortegui MP, Leonstini E, Winch PJ.
‘‘They’ll change what they’re doing if they know
you’re watching’’: Measuring reactivity in health
behavior because of an observer’s presence — a
case from the Peruvian Amazon. Field Methods
2009;21:3-21.
41. Quinlan M. Considerations for collecting freelists
in the field: Examples from ethnobotany. Field
Methods 2005;17:219-34.
42. Weller SC. Cultural consensus theory: Applica-
tions and frequently asked questions. Field Meth-
ods 2007;19:339-68.
43. Nichter M. Education by appropriate analogy. In:
Anthropology and international health: South
Asian case studies. Dordrecht, Netherlands:
Springer, 1989:287-306.
44. Atkin CK. Perspectives on health campaign per-
suasion strategies. Paper presented at: The Annual
Meeting of the Speech Communication Associa-
tion; 1993; Miami, FL.
45. Witte K. Preventing AIDS through persuasive com-
munications: A framework for constructing effec-
tive, culturally-specific, preventative health
messages. Int Intercult Comm Annu 1992;16:67-86.
46. Witte K. Fishing for success: Using the persuasive
health message framework to generate effective
campaign messages. In: Maibach E, Louiselle
Parrott R, eds. Designing health messages:
Approaches from communication theory and pub-
lic health practice. Thousand Oaks, CA, USA:
Sage Publications, 1995:145-166.
47. Sanghvi T, Jimerson A, Hajeebhoy N, Zewale M,
Nguyen GH. Tailoring communication strategies
to improve infant and young child feeding prac-
tices in different country settings. Food Nutr Bull
2013;34: S169-80.
48. Adu-Afarquah S, Lartey A, Zeilani M, Dewey KG.
Acceptability of lipid-based nutrient supplements
(LNS) among Ghanaian infants and pregnant or
lactating women. Matern Child Nutr 2011;7:
344-56.
49. Flax VL, Thakwalakwa C, Phuka J, Ashorn U,
Cheung YB, Maleta K, Ashorn P. Malawian moth-
ers’ attitudes towards the use of two supplemen-
tary foods for moderately malnourished children.
Appetite 2009;53:195-202.
153
50. Hess SY, Bado L, Aaron GJ, Ouedraogo J, Zeilani
M, Brown KH. Acceptability of zinc-fortified,
lipid-based nutrient supplements (SQ-LNS) pre-
pared for young children in Burkina Faso. Matern
Child Nutr 2010;7:357-67.
51. Lesorogol C, Jean-Louis S, Green J, Iannotti L.
Preventative lipid-based nutrient supplements
(SQ-LNS) and young child feeding practices:
Findings from qualitative research in Haiti [pub-
lished online May 1, 2014]. Matern Child Nutr.
doi:10.1111/mcn.12122
52. Matias SL, Chaparro CM, Perez-Exposito AB,
Development Bank, Peerson JM, Dewey KG.
Acceptability of a lipid-based nutrition supple-
ment among Guatemalan infants and young chil-
dren. Washington, DC: FHI360/FANTA-2 Bridge,
2011.
53. Phuka JC, Ashorn U, Ashorn P, Zeilani M, Cheung
YB, Dewey KG, Manary M, Maleta K. Accept-
ability of three novel lipid-based nutrient supple-
ments among Malawian infants and their
caregivers. Matern Child Nutr 2011;7:368-77.
54. Centers for Disease Control and Prevention and
Nutriset. Formative assessment to design the
packaging of a lipid-based nutrient supplement for
a home fortification program to improve the nutri-
tional status of young children in the Democratic
Republic of Congo (DRC). Report prepared for
UNICEF and Ministry of Health in DRC 2013:
1-46. Available at: http://www.hftag.org/
resource/drc-kulabora-formative-assessment-on-
packaging-aug-2013-final-pdf/.
55. Gittelsohn J, Shankar AV, Ram R, Gnywali T,
West KP. Estimating reactivity and its effects in
direct observation studies of health behavior. Hum
Organ 1997;56:182-9.
56. Lincoln YS, Guba EG. Naturalistic inquiry. Bev-
erly Hills, CA, USA: Sage Publications, 1985.
57. Miles MB, Huberman AM. Qualitative data anal-
ysis, 2nd ed. Thousand Oaks, CA, USA: Sage Pub-
lications, 1994.
58. Glesne C, Peshkin A. Becoming qualitative
researchers: An introduction. White Plains, NY,
USA: Longman, 1992.
59. Jefferds MED, Flores-Ayala R. Introducing a new
monitoring manual for home fortification and
strengthening capacity to monitor nutrition inter-
ventions [published online May 1, 2014]. Matern
Child Nutr. doi:10.1111/mcn.12118