Part 1: Revised Technical Proposal: in Partnership With
Part 1: Revised Technical Proposal: in Partnership With
Part 1: Revised Technical Proposal: in Partnership With
An Impact Evaluation of the Uganda Multi-Sectoral Food Security and Nutrition Project (UMFSNP)
Date of Submission:
July 22, 2016
Submitted to:
The World Bank
1818 H Street
Washington DC, USA
Submitted by:
Partnership for Child Development (PCD)
Department of Infectious Disease Epidemiology
Imperial College, London
United Kingdom
In partnership with:
1
Table of Contents
Executive Summary............................................................................................................................. 3
1. Background ............................................................................................................................. 4
2. Project Description.................................................................................................................. 5
3. Impact Evaluation ................................................................................................................... 8
4. Methods ................................................................................................................................ 11
5. Analysis Strategy ................................................................................................................... 17
6. Ethical considerations ............................................................................................................... 17
7. Roles & Deliverables ............................................................................................................. 18
2
Executive Summary
More than one third of all child deaths worldwide are attributable to child and maternal
undernutrition, and Uganda is one of the 20 countries in the world with the highest risk. The associated
economic costs are substantial - in Uganda, stunting among children alone is estimated to cost US$899
million annually to the economy (5.6% of GDP). The evidence base highlights the need for multi-
sectoral interventions to address the multiple causes of undernutrition.
In response, a multi-sectoral food security and nutrition project led by the Government of Uganda will
commence in 2015 with financial support provided by the Global Agriculture and Food Security
Program (GAFSP) with the World Bank as the supervising entity. The long-term aim of the project is to
reduce under-nutrition and stunting among children less than 2 years old. Using schools as a platform,
the Uganda Multi-Sectoral Food Security and Nutrition Project (UMFSNP) will introduce a holistic
package of nutrition-specific and nutrition-sensitive interventions targeting different stages of the
lifecycle particularly for women and children, achieved through three types of interventions. School
demonstration gardens will provide an entry point through which parent groups will be formed and
mobilized to increase production of micronutrient-rich foods. Enhanced nutrition services such as
nutrition education and deworming will be delivered in primary schools. Lastly, community nutrition
services for pregnant and lactating women and their infants will be enhanced. Activities detailed in
the Project Implementation Manual will take place in 15 districts over five years.
This Technical Proposal outlines a rigorous Impact Evaluation (IE) that has been designed to evaluate
the program in terms of the Project Development Objectives (PDO) spanning nutrition and agriculture,
as well as identify the pathways through which these outcomes are realized. Several deliverables are
planned to document findings from the IE and inform the implementation of the project.
3
1. Background
More than one third of all child deaths worldwide are attributable to child and maternal
undernutrition, and Uganda is one of the 20 countries in the world with the highest risk.1 The impact
of undernutrition on population health and the associated economic costs are substantial. In Uganda,
stunting among children alone is estimated to cost US$899 million annually to the Uganda economy
(5.6% of GDP).2,3 Therefore it is in the interest of multiple sectors, including business, agriculture,
health and education, to encourage and promote the general sequence of health and development
programmes.4
The period between conception and 2 years of age, known as the “window of opportunity”, is
especially critical for a child’s development, as damages incurred during this period are often difficult
to reverse.5 The costs of undernutrition are amplified by the young age structure in Uganda where
over half of the population is under the age of 15 years, and the high fertility rate of more than 6 births
per childbearing woman. 6 As presented in the UNICEF nutrition conceptual framework, the
determinants of stunting are multiple and require a coordinated approach across sectors.7
In order to reduce undernutrition, studies highlight the need to scale up both nutrition-specific
interventions to address the immediate determinants of stunting (e.g. improving breastfeeding,
complementary feeding, and caregiving practices, and reducing disease burden), and nutrition-
sensitive interventions to address the underlying determinants (e.g. improving food security, school
feeding and caregiving resources, and environmental health). 8 This underscores the need for
interventions that address the whole life course, including maternal nutrition and health9, and that
exploit the synergies between agricultural, health and education. For example, in order to exploit their
full nutrition-enhancing potential, agricultural interventions should also incorporate a component
related to behavioural changes (e.g. dietary diversity, feeding, and hygiene).10
Nutrition-specific and nutrition-sensitive interventions over the course of the lifecycle can help
children reach their full potential. Women of childbearing age with higher nutritional status are more
likely to experience healthier pregnancies, and to have children with higher nutritional status.
Smallholder agricultural interventions to increase and diversify the production of micronutrient rich
foods can promote their accessibility and consumption in communities.
While a body of work exists in this area, there is a lack of studies that employ rigorous methods such
as impact evaluations. For example, an evaluation of a program in Asia highlighted the benefits of
1 Black, Robert E., et al. 2013. "Maternal and child undernutrition and overweight in low-income and middle-income
countries." The Lancet, Volume 382, Issue 9890, p 427 - 451
2 Figure taken from the World Bank project document, p.14. July 2014.
3 Stunting or chronic undernutrition relate to the failure to reach linear growth potential because of inadequate nutrition or
poor health. Chronic undernutrition is associated with long-term malnutrition and poor health and is measured as height for
age that is two z-scores below the international reference standards (WHO 2005). This measure is recognized internationally
as an indicator of long-term undernutrition among young children.
4 Bundy, D. 2011. Rethinking school health: A key component of education for all. World Bank
5 UNICEF. 2013. Improving Child Nutrition. The achievable imperative for global progress. New York. Available at:
http://www.unicef.org/media/files/nutrition_report_2013.pdf
6
Uganda Demographic and Health Survey 2011.
7 UNICEF nutrition conceptual framework.
8 Ruel M, Alderman H. 2013. Nutrition-sensitive interventions and programmes: how can they help to accelerate progress in
improving maternal and child nutrition? The Lancet Maternal and Child Nutrition series; 382: 536–51
9
Black, Robert E., et al. 2013. "Maternal and child undernutrition and overweight in low-income and middle-income
countries." The Lancet, Volume 382, Issue 9890, p 427 - 451
10 Ruel M. 2001. Can Food-Based Strategies Help Reduce Vitamin A and Iron Deficiencies? A Review of Recent Evidence.
4
providing technical assistance and agricultural support to women farmers on nutrition education at
the household level in order to increase food diversity.11 However, the impact pathways were not
defined, nor did the study suggest how the programme could be implemented sustainably. Another
study from Ecuador found that children gained significant knowledge about nutrition behaviours, but
details on the knowledge transfer were not provided.12
The IE will also contribute to the global knowledge base on the effectiveness of schools as a platform
for delivering community-based nutrition interventions.
2. Project Description
2.1 Overview
The project draws on experiences from other multi-sectoral projects that have been implemented in
line with the Uganda Nutrition Action Plan (UNAP) 2011-2016 in particular the parent-led school
gardens program implemented by SNV/Netherlands Development Organization in partnership with
the Embassy of the Kingdom of Netherlands and UNICEF, which found that primary schools can be an
effective platform to mobilize communities to improve agriculture practices. The UMFSNP will be
complemented by other agricultural efforts such as the Agricultural Technology and Agribusiness
Advisory Services (ATAAS) project and the Agriculture Cluster Development Project (ACDP) both
financed by the World Bank, which are designed to raise agricultural productivity and income primarily
through cash and staple crops, but will not address the gap in production and consumption of
micronutrient-rich foods. The project will be implemented through national systems to achieve
ownership, build capacity, and achieve sustainability.
The Project Development Objective (PDO) is to “increase production and consumption of
micronutrient-rich foods and utilization of community-based nutrition services by smallholder
households in project areas.” The objective is expected to be achieved through provision of a set of
nutrition-sensitive and nutrition-specific interventions in targeted communities, primarily by
leveraging schools as a platform for delivery, and with the objective of promoting short-term changes
in high-impact nutrition behaviours and practices known to contribute to stunting reduction in the
medium- and long-term. Schools will serve as an entry point to strengthen linkages between the
community and nutrition services provided by primary school teachers, agricultural extension
services, community health centres and VHTs. 13 In addition, the project will support demand-side
interventions to enhance utilization of VHT delivered community-based nutrition services. Together,
the interventions are holistic and address different stages of the lifecycle, particularly for women and
children. The interventions have several components as follows:
1) Agricultural support for school-based nutrition services: Parent groups (PGs) from communities in
catchment areas of selected primary schools will be formed and mobilized to establish gardens at
primary schools as “agriculture and nutrition classrooms”. Through strengthened linkages
between community groups and strengthened agriculture extension services, PGs, 60% of whom
will be women, will be trained in improved agricultural technologies and practices and will engage
11 Talukder, A. et al (2010). Homestead food production model contributes to improved household food security and
nutrition status of young children and women in poor populations. The Journal of Field Actions: special issue 1.
12 FAO (1997). Nutrition education for the public. Discussion papers of the FAO expert consultation. FAO food and
5
in demonstration and replication of these to increase production of micronutrient-rich foods. The
participation of PGs will be facilitated by two lead farmers (LFs) selected by the community taking
into consideration their agricultural skills, available land and interest in sharing improved
technology. The LFs will undertake community-based multiplication and replication of
micronutrient-rich planting materials to increase availability of these products at the community.
In some districts, these activities will also be supported by local community-based organizations
(CBO) These activities correspond with Sub-components 1.1 and 1.3 of the PAD.
2) Enhancing nutrition services delivered through primary schools: School and community-based
nutrition activities will be strengthened through the establishment of a School Nutrition
Committee, provision of nutrition education (promotion of optimal nutrition and hygiene
practices), school-based deworming for all school children, and weekly iron folic acid (IFA) tablet
supplementation for female students in primary four and above. These activities correspond with
Sub-component 1.2 of the PAD.
3) Enhanced nutrition services through Village Health Teams (VHTs): Nutrition services will be
delivered through schools and the community including monthly delivery of community
mobilization on nutrition, nutrition behaviour change communications, and monthly growth
monitoring and promotion of children under 24 months; provision of IFA supplements to pregnant
or lactating women, deworming to pregnant women, and zinc supplements for children 6-59
months.
2.2 Implementation
The UMFSNP will be implemented in 15 of the 111 districts in Uganda based on a number of pre-
established criteria, including high levels of stunting and low levels of adequate dietary diversity. These
districts are located in four geographical regions and six ecological zones (see Table 1). Each has an
average of 348,000 residents for a total population of 7 million.
The intervention will commence in 2015 in 5 districts with the greatest nutritional deficits14 (Phase 1),
and will be introduced in 10 of the remaining 15 additional districts in Year 2 depending on their
readiness to implement (Phase 2). The Ugandan decentralization policy introduced in 1997 as
rendered the Chief Administrative Officer and District Directorates as virtually responsible for all
activities and operational fund management within the district. Some districts are better staffed and
able to plan, manage and coordinate than others, and as such, readiness to implement was a key
factor in the selection process.
14
Levels of stunting are higher and diet diversity are lower in the Phase 1 districts as compared with the Phase 2
districts (PAD, Annex 7).
6
West Nile Nebbi,
Yumbe,
Arua,
Maracha,
Schools in the districts will be selected for the project based on a two-stage process. In the first stage,
eligible schools will be identified to participate in the project based on criteria including: (a) Rural or
peri-urban; (b) Government aided schools implementing the Universal Primary Education (UPE)
program; (c) Presence of head teacher and agriculture teacher; and (d) Unqualified school audit for
the past financial audit. At Stage II, the schools that met stage I criteria will be requested to fill out an
application form with the following criteria to identify school-level ownership and readiness for
implementation: (a) Presence of a functional School Management Committee (SMC); (b) An existing,
or commitment to establish a functional sub-committee of the SMC, a "School Nutrition Committee",
which will include representation of the school administration; (c) Existence of at least one half acre
of arable and conflict-free land with available water; (d) Organized PGs willing to participate in school
level nutrition programs, including time and labour commitments; and the establishment of a project
bank account.
Data from the Ministry of Education for these districts indicate that there is an average of 130 public
primary schools per district. Districts will identify a list of schools that meet all of the eligibility criteria.
These lists will be reviewed and confirmed by the project team. The Project Implementation Manual
budgets for up to 100 schools per district to participate in the project. The actual figure of schools
participating per district will be updated during project design consultations.
In total, the project is expected to have 1.14 million primary beneficiaries mainly from smallholder
farming families. In total, it is estimated that approximately 15 districts, 5,400 VHT members, 1,500
primary school demonstration gardens, 3,000 LFs and 3,000 PGs (consisting of an estimated 45,000
parents) will be supported over the life of the project. More than 60 percent of the PG members will
be women. The 5 districts included in phase 1 comprise of about 380,000 primary beneficiaries, 75,000
pregnant and lactating women, 125,000 children 0 to 23 months, and 180,000 people in households
of LFs and PGs that are expected to benefit.
7
3. Impact Evaluation
3.1 Overview
A rigorous impact evaluation (IE) will be carried out to assess the impact of the multi-sectoral
intervention on key outcomes as well as to identify pathways through which the project activities
translate into the outcome-level changes realised. The IE will be limited to the Phase 1 districts to
permit as much time as possible for outcomes to accrue over the project period. In addition, the
evidence emerging from the IE will help identify lessons learned to inform the implementation of the
project in the Phase 2 districts.
The evaluation will define a set of treatment and control schools and surrounding parishes matched
on baseline characteristics such as ethnic group composition, school attendance and level of
agricultural infrastructure. Treatment schools and the parishes in which they are located will be
selected from the pool of schools that will receive the project in the 5 Phase 1 districts. Control schools
and the parishes they are located in will be selected from an adjacent district with a comparable
sociodemographic and agricultural profile and which will not be participating in the project. The
baseline IE survey will be conducted in Year 1 of the project while an end-line IE survey will be
conducted in Year 4 of the project. In addition, a midline IE survey will be conducted in Year 2, while
Focus Group Discussions (FGD) and Key Informant Interviews (KII) will be carried out in project areas
during the course of the project. The baseline and end-line surveys will include indicators aligned with
the GAFSP global indicators to allow for cross-country comparisons with other projects. More
information about the Data Collection Plan including sample size calculations can be found in Section
4.
The IE will be complemented by the project’s M&E plan through which output-level data will regularly
be collected from districts. Frequent correspondence and information sharing between the PCD-IDCL
evaluation team and M&E focal point in the Project Coordination Unit (PCU) will be critical to ensure
that the IE and M&E are aligned and complementary, and to assess the contribution of activities to
realized outcomes. Given the geographic spread of the project, previous interventions carried out in
the districts, and the decentralized nature of the project, significant heterogeneity in project
implementation is expected and must be accounted for in the interpretation of results.
What is the impact of the project on the production of micronutrient-rich foods year round at
the community level?
What is the impact of school and community-based nutrition activities on anaemia prevalence
among school-aged children?
To what extent did project activities contribute to changes in nutrition knowledge and
behaviours primarily for pregnant and lactating women, caregivers of children 0-23 months
as well as for school-age children?
8
What barriers to adoption of improved nutrition behaviours/practices are addressed by the
project? Which activities were most successful in facilitating the adoption of desired nutrition
behaviours/practices?
The evaluation also presents an opportunity to assess the viability of the project in terms of the extent
to which targeted PGs and households in project areas adopt new agricultural practices and
technologies.
Figure 1 illustrates the impact pathways of the project from outputs to impacts with causal links noted
by arrows. For example, knowledge about production of micronutrient-rich foods through the
demonstration gardens may lead to replication in the community through knowledge sharing and
direct support from LFs. Additional assumptions that are critical to the success of the project are:
Districts, VHTs and schools will have the capacity and resources to implement the project
activities;
Targeted individuals will have time to allocate to the activities introduced by the project;
Women’s workload resulting from participating in the demonstration garden intervention will
not compromise their role as caregivers;
Improved knowledge and skills can increase agricultural production all year round;
There is adequate agricultural potential and water to produce micronutrient-rich foods in the
target communities;
Referral higher-level nutrition and health services are accessible in target communities.
9
FIGURE 1: Impact pathways of the project – from outputs to impacts
10
The number of households and communities sampled for the IE must be in accordance with the
expected changes over the project period to ensure for the identification of treatment effects, as well
as to contain costs to the degree possible. Sample size calculations based on the expected treatment
effects outlined in the PAD are described in Section 4.
The timing of data collection in relation to project activities and the agricultural season is also critical.
According to the timeline set by the IE team, the baseline survey should occur as soon as possible after
the research protocol is approved. Follow-up surveys should be conducted in the same season to allow
for easier comparison over time, although seasonal variation can be accounted for with data from the
comparison group.
Lastly, the IE, in particular the qualitative information collected from FGD and KII, can be a source of
objective information to identify what is working and not working about the project, and help validate
the M&E data, which will be provided from the districts. Objective information is critical to inform the
project implementation and ensure its success. As such, it is critical for the PCU and project team to
respect negative findings as well as positive ones, as well as the IE team to not only report findings but
provide suggestions for improvement and learning.
4. Methods
Based on the methods described in this proposal, the final methodology is detailed in the Research
Protocol.
Household and school surveys would be implemented at all of the time points and for all districts. The
questionnaires for the IE districts, however, will be more in-depth. The school survey will be fielded
to an administrator in each school of the selected parishes. An anaemia sub-subsidy will be undertaken
of school children in grades P4-P7. Haemoglobin concentration will be tested amongst both boys and
girls in school.
The survey data will complement information collected through FGD and KIIs, as well as the project
M&E. TABLE 3 illustrates the relation between the collection of data from these three sources.
TABLE 3: Collection of monitoring and evaluation data for the project by project year
2015 2016 2017 2018 2019
11
Household and school surveys X X X
Focus groups; Key informant X X X X
interviews; Rural rapid appraisals
Monitoring (undertaken by project X X X X X
team)
The qualitative information will be especially useful to address research questions related to project
implementation and impact pathways, as well as to validate some measures collected through the
survey and the M&E.
4.2 Sample size calculation
The sample size calculation found that a minimum total of 108 parishes with 25 households in each
parish, that would represent the cluster, evenly split across treatment and control arms of the IE,
should be selected. These households must include at least one child 6-23 months. To ensure sufficient
sample to detect statistically significant differences, the IE technical assistance team suggests that 54
parishes representing clusters are sampled for each IE arm with 25 households from each parish.
TABLE 4 presents the recommended sample size figures for the main category and sub-categories. As
mentioned above according to sample size calculation, a total of 54 parishes from each arm of the IE
with 25 households from each parish leading to a total sample of 1350 households in each arm will be
sampled. In addition, 324 households will be sampled for each arm for each of the 3 categories
described in Table 4 i.e. 0-5 months ;24-59 months and P4-P7 children to measure other indicators
besides the PDO indicator (Minimum Dietary Diversity), such as infant feeding practices,
complimentary foods, Minimum Meal Frequency, stunting and anaemia. Thus the total sample is 2322
households for each arm with 43 [(25) +(6*3)] households per parish. Some of the 1350 households
may include an infant 0-5 months, and they will be surveyed as well. Sampling of more parishes and
households would lend greater confidence to detecting impact in the key indicators. For a sub-sample
of these households, haemoglobin concentration would be tested for mothers and children ages 0-24
using Hemocue.
12
6-23 mo 1350 1350
0-5 324 324
24-59 mo 324 324
P4-P7 children 324 324
The 54 schools and surrounding parishes for the treatment arm could be evenly selected from the 5
Phase 1 districts such that each district has minimum 10 sampled parishes. A similar approach could
be taken for the comparison districts; such that minimum 10 parishes are selected from each of 5
adjacent districts. TABLE 5 identifies an adjacent district for each of the 5 Phase 1 districts.
Alternatively, one comparison district could be selected for two treatment districts if there is
comparability and cost savings are apparent. For example, Arua could serve as a comparison for
Maracha and Nebbi, instead of Koboko and Zombo. Once the comparison districts are identified,
candidate comparison schools can be identified using the two sets of eligibility criteria applied in the
5 Phase 1 districts to identify project schools. From the pool of candidate comparison schools, a set of
schools and their surrounding parishes for the IE would be identified through matching with the
treatment schools and parishes using EMIS data and district provided information.
TABLE 5: Treatment and comparison districts
Treatment districts: Comparison districts:
Maracha Koboko
Bushenyi Sheema
Namutumba Kaliru
Nebbi Zombo
Ntungamo Rukungiri or Mbara
Following the baseline survey and the baseline indicators will be reviewed to estimate the minimum
detectable effects (MDEs) over the period of the impact evaluation and compared with the target
values presented in PAD.
Accurate age determination may be challenging and is a critical factor in this study. Thus the PCD-IDCL
team will support the data collection team at each survey data collection point to carefully identify
households with a child whose age falls within the specified range. A random sample of at least 50
households from each parish should then be drawn from the sampling frame to respond to the survey.
In the Community Connector project, VHTs were instrumental in identifying women with young
infants in the project districts. For the IE, VHTs would be consulted to identify the sampling frame.
A list of all households in the selected parishes that include a household with a child 0 to 23 months
of age would be drawn and enumerated, and will constitute the sampling frame for households. A
13
similar approach will be used for the midterm and end-line surveys, that will draw a fresh sample as
the baseline sample will age beyond the 0-23 month age bracket over the period. Similarly, a list of
households with a child of 24-59 months of age and school age will be also drawn for each parish and
will constitute the sample frame for these categories.
While additional time and effort is required, the enumeration of households and random sampling
will enhance the rigour of the study. Other studies have not used enumeration, but this may introduce
biases into the study. Households that are available and sampled may differ systematically than
households that are not in terms of socioeconomic status and health outcomes, thus providing a
skewed assessment.
Quantitative data – will be collected using a single digital questionnaire, administered face-to-face to
mothers/ caregivers of children 0-23 months in home settings. The data collection tool designed in
English will be translated into local languages. For the IE districts these languages will include: Alur for
Nebbi and Zombo districts; Lugbara and Kakwa for Maracha and Kaboko, respectively; Runyankole for
Bushenyi, Shema, Ntungamo and Rukungiri; and Lusoga for Namutumba and Kaliro districts. Data will
be collected simultaneously in the IE districts (5 intervention and 5 control districts) by trained
Enumerators.
Qualitative data – will be collected during focus group discussions and key informant interviews. FGDs
will be conducted with mothers or caregivers of children 0-23 months; PGs; and school children.
Questions in the FGD will focus on knowledge, behaviours and practices, in particular facilitators and
barriers to behaviour change. KIIs will be conducted with district officials; school headmasters or
teachers; VHTs; and lead farmers. The KIIs will focus on project management and implementation,
institutional capacity and costing. In both the FGDs and KIIs, knowledge about the UMFSNP and how
it differs from other projects will be assessed.
Quality Control- PCD-IDCL shall adopt three main strategies to ensure high standards of the data
collection process and the ultimate quality of the data collected. First, there will be a constant and
random back-checking of field enumerators’ work by the field supervisors to make sure that the
quality of the data is not compromised in any way. A minimum 5% of all interviews conducted by each
interviewer are to be ‘back-checked’ by the supervisor. This will involve returning to a respondent who
will have been interviewed in order to re-interview and verify key points of the questionnaires, and to
check that the standard protocol had been correctly applied. Secondly, PCD-IDCL will also have a
monitoring team on the field to review the work of the supervisors and the interviewers to ensure
their adherence to protocols in the field. This will ensure that enumerators are conducting the survey
according to lay down standards and quality assurance mechanisms. Since the CAPI system will also
allow the project coordinator to have access to the data to review whilst interviewers are still on the
field, cases with issues can be rectified through interviewers’ call-backs. The third and most significant
is the role of the quality control assistants. The quality control assistants will provide further quality
control as they will randomly select a percentage of each field enumerator’s output and re-interview
14
respondents on selected screener questions for the purpose of cross-validation and to provide
feedback to the field enumerators. The entire quality control process will be supervised by IDCL in
Kampala and by the lead IE expert at the Centre for Health Policy at Imperial College London.
- Changes in infant and young child feeding practices (0-23 months) assessed by indicators such as
exclusive breastfeeding rates among children 0-5 months; and minimum dietary diversity (MDD),
minimum meal frequency (MMF) and minimum acceptable diet (MAD) among children 6-23
months. MDD is a PDO indicator and is the key outcome measure of the impact evaluation.
- Changes in stunting prevalence among children 0-23 months
- Changes in production of micronutrient-rich crops namely green leafy vegetables, yellow/orange
vitamin A rich fruits and vegetables such as carrots, pumpkins, etc, at household level (PDO
indicator)
Explanatory indicators – these are optional but may be important in providing the basis for success or
failure of the project. The majority of the explanatory factors will be addressed in the qualitative
assessments.
- Morbidity indicators such as the two week prevalence of common childhood illness
- Primary health care and services such as immunisation coverage of for children 0-23 months,
deworming and IFA supplementation among school going children, participation in GMP activities
- Equity – assessed by socioeconomic status of households covered by the project
- Household food security
- Gender dynamics especially the time and decision dynamics for child care between men and
women in households.
15
Diet - the key indicators of dietary diversity will be assessed at the individual level. Individual dietary
diversity scores (IDDS), which is a measure of the diversity of food groups contained in the diet
consumed by children 6-23 months will be developed. Dietary diversity will be assessed based on
seven food groups namely: cereals, pulses and oils, meats, eggs, milk, vitamin A rich fruits and
vegetables, and other fruits and vegetables.15 The key indicator for the project is minimum dietary
diversity (MDD) defined as the proportion of children who will have received foods from at least four
food groups the previous day of the assessment. Additional indicators to be assessed will include:
minimum meal frequency (MMF) defined as children 6-23 months who will have had three meals (if
breastfeeding) or four meals (if not breastfeeding); and minimum acceptable diet (MAD). Minimum
acceptable diet is the combination of children who had minimum acceptable diet diversity and those
who had minimum meal frequency. In additional household total nutrient intake will be computed for
children using Optifood linear programing software to describe changes attributable to the project.
Nutritional status - Weight, height (or recumbent length in children under two years of age), and mid-
upper arm circumference (MUAC) will be measured in sampled mothers/caregivers and children. It
will not be possible to collect anthropometric measurements in triplicate because of the digital data
collection but due emphasis will be made during training and the recorded measures will be shouted
out to the entire team of Enumerators in order to minimize error. The World Health Organization
(WHO, 2006) growth standards will be used to determine children’s z-scores for height-for-age,
weight-for-age, weight-for-height, BMI, and MUAC. Body mass index (BMI) will be calculated for all
non-pregnant mothers while only MUAC will be assessed for both pregnant and non-pregnant women.
Instruments, particularly scales (electronic scales will be use), will be validated regularly, and staff will
perform the anthropometric measurements according to standard and standardized techniques.
Anaemia status- Blood samples will be collected from the children (6-23 months) and mothers to
determine haemoglobin levels using hemocue analysers (H301). A hemocue is an instant digital test
and will be carried out by trained Biomedical Technologists in the field. Haemoglobin concentration
to assess anaemia prevalence will be determined using a HemoCue on all caregivers and children.
Analysis will be based on national and WHO cut-offs.
Agricultural production- household agricultural activities and production of MNR foods will be
assessed. Household food security status will be assessed using food consumption scores (FCS) and
the household hunger scales (HHS) approaches.
Morbidity- Morbidity from common childhood illness like acute respiratory infections (ARI), malaria,
diarrhoea, measles, skin and eye infections will be assessed over a two-week recall period. In addition,
coverage of the essential primary care services such as immunization, vitamin supplementation and
deworming among infants and young children, GMP, and environmental and domestic sanitation
factors such as latrine and safe water coverage will be assessed.
15
WHO Indicators for assessing infant and young child feeding practices part 2: measurements.
16
The household survey will be the main source of information for the IE. The instrument includes a
wide range of questions covering agricultural production including of micronutrient-rich foods,
agricultural investment, technologies and practices, diversity of food produced and consumed and
utilization of agricultural inputs. Caregivers with a child 6-23 months will be asked additional questions
regarding infant feeding using 24-hour recall using validated measures to assess diet diversity.
Caregivers with a child less than 6 months of age will be asked about other infant feeding practices
including breastfeeding. All caregivers will also be asked to provide information retrospectively about
the pregnancy of the focus child. More information on the specific indicators are described in Section
4.4.
In addition, a school survey will provide information about ongoing activities, programs and
infrastructure that may or may not be part of the GAFSP. The school survey will capture information
on critical factors such as the state of WASH facilities, and health and nutrition education. Blood
samples of P4-P7 school children will be conducted at the same time as the school surveys to minimize
data collection costs. Permissions from the households will be solicited beforehand.
5. Analysis Strategy
A mixed-methods strategy will be undertaken to conduct the impact evaluation. The main analyses
will focus on the PDO indicators listed in Table 4 using the baseline and endline surveys. The treatment
effect of the intervention will be estimated as per the equation below where Yi represents the PDO
indicator, Tc represents treatment status at the community level, and community (c), household (hh).
Depending on the level of the PDO indicator, child-, mother-, or district-level (i) covariates associated
with the dependent variables. The inclusion of covariates in the impact estimations can enhance the
precision of the treatment effect.
𝑌𝑖 = 𝛽0 + 𝑇𝑐 𝛽𝑇 + ∑ 𝛽𝑐 𝑋𝑐 + ∑ 𝛽ℎ 𝑋ℎ + ∑ 𝛽𝑖 𝑋𝑖
𝑐
ℎℎ 𝑖
In addition to the main impact estimations, the evaluation team will undertake analyses to address
the other research questions. Analyses of lower level variables (e.g. inputs and outputs) can help
identify the pathways through which impacts are or are not realized at the school and community
levels, and can be especially important to inform the scale-up of the program. In addition, sub-group
analyses will be conducted for populations of interest, for example girls and women, to assess
differential impacts and pathways.
6. Ethical considerations
Ethical considerations in studies involving human subjects are always of paramount importance. The
ethics approval for this study will be processed through IDCL as per the Uganda National Council for
Science and Technology and will also be reviewed by PCD/ICL ethical review board. The IRBs will
review any modifications to the protocol or reportable new information about the study.
A written informed consent will be sought from study participants (caregivers a), such that each
participant understands that participation in the research is voluntary, and that they are free to
withdraw participation at any time, even after consent is granted. Prior to enrollment, study
17
participants will be provided with information on the study objectives and design, and confidentiality
issues will be emphasized. After providing the information, they will be given the opportunity to ask
questions and enroll in the study, or not. Caregivers who may want to discuss their participation with
family members will be given an opportunity to do so and to enroll at another visit by the study
Research Assistants. If participants are unable to read, a witness to the consent process will sign the
consent form. All participants will receive a copy of the consent form to keep. The consent form will
also discuss the possibility of data being used in future research.
Participants’ privacy and confidentiality will be strictly observed and all research-associated risks will
be minimized. In order to protect participants’ privacy, all questionnaires will be administered in a
private setting. To protect participants’ confidentiality, all completed consent forms will be stored in
a locked space. All electronic data will be entered into a password-protected database only accessible
by authorized personnel. Information on identity of participants will never be revealed to or discussed
with third parties.
Potential research-associated risks include breach of confidentiality, discomfort with answering some
sensitive questions, and slight risks associated with finger pricks to assess anaemia. Every effort will
be made to prevent a loss of confidentiality, as described above. Research staff will be specifically
trained to administer the questionnaires and will recognize that some individuals may feel
uncomfortable discussing certain topics especially on household feeding practices and income. They
will let participants know that they may refuse to answer any question.
Standard safety precautions will be taken to do finger pricks and will be performed only by trained
research staff. Participants who will be found to have severe anaemia (haemoglobin < 8 g/dL) or
malnutrition (BMI < 18.5 in adults or children with visible severe wasting, or with oedema, or MUAC <
125 mm in children 12-23 months) will be promptly counselled and referred to a health facility for
treatment.
18
Final baseline June 2017
Briefing on key findings on baseline report July 2017
Midline Survey TBD
Midline Report TBD
A draft baseline report will be prepared as soon as possible following the data collection that will focus
on key indicators including PDO indicators. A more detailed baseline report will be completed in early
2017 that will assess the comparability of the treatment and control arms, and highlight how potential
pathways may be affected by project activities. Accompanying briefings will be developed and
presented to stakeholders with a focus on findings that can inform the project implementation and
scale-up to Phase 2 districts.
The Midline Report will present findings from the M&E data review, the midline survey as well as the
qualitative data collection through the FGDs and KIIs in the 5 districts. The final IE report will be the
main deliverable of the project, and will be completed in early 2020. This report will focus on the
baseline and end-line data from the 5 Phase 1 districts (well as well as data collected from the
comparison districts) to estimate the impact of the GAFSP on key outcome-level over the period using
analytic methods described in Section 5. In addition, the impact pathways triggered by the project
activities will be described. The briefing on key findings from the final IE report will be prepared for
stakeholders of the UMFSNP in Uganda, as well as the GAFSP coordinating committee. The briefing
will focus on findings from the final IE report, and will also draw on findings from the implementation
report.
A capacity statement of both institutions is presented below. Their combined expertise will support
the provision of technical assistance, analysis and interpretation of results, and dissemination of
results.
Partnership for Child Development. Since its creation in 1992, PCD has grown to become a global
consortium of over 100 civil society organizations, academic institutions and technical experts with a
streamlined Coordinating Centre based at Imperial College London. All of PCD’s work is underpinned
with strong in-country collaboration and consultations with key stakeholders. This aspect ensures
informed programme designs, identification of key issues to be addressed, stakeholder buy in and
commitment of resources. Through global partnerships with the World Bank, WFP and the African
Union’s New Partnership for Africa’s Development (NEPAD) amongst others, PCD has been providing
direct, evidence-based and context-specific support to governments to strengthen school health and
nutrition policies and programs, sourced with local agricultural production, in 20 countries globally.
PCD has expertise in monitoring and evaluation assistance, particularly impact assessments and is
currently engaged in impact evaluation in several African countries. Governments are directly
benefiting from this practical knowledge based products for more effective programming of their food
and nutrition security programmes and policies.
PCD will be the lead organization contracted by the World Bank to carry out the evaluation. Dr. Lesley
Drake will be the Principal Investigator for this study. Additional expertise will be provided by Research
Associate in Food Policy (Samrat Singh), Senior Research Advisor (Dr. Meena Fernandes), Senior
Nutrition Consultant (Prof. Josephine Kiamba), Public Health IE specialist (Dr Ranjeeta Thomas), Health
Economist (Dr Lesong Conteh) and Senior Programme Managers (Elodie Yard and Iain Gardiner).
19
To carry out the evaluation, PCD will be supported by IDCL, which has a strong in-country experience
of conducting impact evaluation for Government departments and development partners. IDCL has
substantial experience in undertaking surveys and training especially in the area of agriculture,
nutrition and public health. Mr. Julius Twinamasiko who has been closely involved with the IE since its
inception will be leading the management of IDCL for this project and will coordinate fieldwork and
data management.
20