HKI presentation for GAAP final technical workshop
1. ENHANCED HOMESTEAD
FOOD PRODUCTION FOR
IMPROVED FOOD
SECURITY AND
NUTRITION IN BURKINA
FASO
Helen Keller International (HKI) and
the International Food Policy Research Institute (IFPRI)
2. E-HFP program in Burkina Faso
Burkina Faso
Eastern region, Gourma
Province, Fada district
Sahel
Water shortages inhibit
having a second cultivation
season
High prevalence of acute
and chronic malnutrition
Food insecurity
Suboptimal maternal and
child nutrition and health
practices
Limited availability of and
access to health services
3. Overall objective of the E-HFP
Program in Burkina Faso
To improve the nutritional status of infants and young
children through a set of production and nutrition
interventions targeted to women with children 3-12
months of age through three primary program impact
pathways
1. Increased availability of micronutrient-rich foods through
household production during the secondary agriculture
season.
2. Income generation through the sale of surplus household
production.
3. Increased knowledge and adoption of optimal nutritional
practices including consumption of micronutrient-rich
foods
4. Program theory framework for HKI’s E-HFP
program in Burkina Faso
Inputs Process Outputs Outcomes Impact
Agriculture and Village Model Establishment Improvements Increased Increased Income Women’s
zoological inputs Farms (VMF) of individual in fruit and availability of empowerment
distributed established farms (40 vegetable micronutrient improved
women per production -rich fruits
village) and
HKI, APRG and Training in vegetables
governmental plant and Women’s assets
structures animal Improvements increased
in small Increased
(Ministries of production
techniques for ruminant and availability of
Health, Improved
poultry food from
Agriculture, master trainers maternal and
production animal origin
Animals, child health
Improvements in
Environment, and and nutrition
Training in Training in household
the Promotion of outcomes
plant and plant and Adoption of consumption
Women, local
authorities and animal animal agriculture
officials) work production production practices
together techniques for techniques for
Village Farm beneficiaries
Leaders (VFL) Improvements in
Beneficiaries
nutrition and
Develop a training received and
feeding practices
strategy in animal; understood
BCC training for children,
and plant agriculture
on ENA pregnant women
production training
practices for and breastfeeding
techniques mothers
master trainers
Develop a
behavior change Improvements in
communication BCC training BCC training Beneficiaries Adoption of care & hygiene
(BCC) strategy on ENA on ENA received and ENA practices for
with regards to practices for practices for understood practices by children, pregnant
Essential village health beneficiaries BCC training beneficiaries women and
Nutrition Actions workers on ENA breastfeeding
(ENA) (VHW) practices mothers
5. Study Design
Longitudinal impact evaluation
Social network census
Operations research
Qualitative research on gender related topics
including ownership and control over
agricultural assets
6. Impact Evaluation
Cluster randomized design
30 intervention villages (~1200 households and 120 village farm
leaders (VFL))
15 “older women leader” villages (OWL)
15 “health committee” villages (HC)
15 control villages (~800 households)
Longitudinal
Baseline Feb-Apr 2010 (target children 3-12 months of age)
Endline Feb-Apr 2012 (target children 21-40 months of age)
Household interview
Male household head and female key respondent including
gender disaggregated modules on asset ownership, agricultural
production, income, household expenditures, knowledge on
nutrition, household food security, dietary diversity, etc.
Anthropometric measures and hemoglobin status of target
children
7. Operations Research and GAAP
Qualitative Research
Operations Research GAAP Qualitative
Research
Random sample Beneficiaries: n=120 Beneficiaries: n=145
of beneficiaries Non-beneficiaries: n=60 Non-beneficiaries: n=75
and non-
beneficiaries
Purposive sample VFL: n=60 VFL: n=60
of key informants OWL: n=30 OWL: n=30
HC: n=30 HC: n=30
Master agriculture trainers: n=18 Land owners: n=30
Master nutrition trainers: n=24 Focus groups: n=24
(12m; 12f)
Data collection May-June 2011 May-June 2011
May-June 2012 May-June 2012
Methods Semi-structured interviews Semi-structured
interviews and focus
8. Key Questions for GAAP
Key Questions Impact Social Qualitativ Analysi
evaluatio network e s
n census research Status
How do women and men view ownership of assets? x On-
going
Did the EHFP program increase women’s ownership of x Complet
assets? e
Did the EHFP program also influence men’s asset holdings? x Complet
e
Were women able to maintain control over the EHFP x x Complet
activities and outputs? e
Did the land agreements and/or project activities influence x Complet
community norms related to women’s land ownership or land e
rights?
What trade-offs were women required to make in order to x On-
participate in the EHFP program? going
Did exposure to nutrition education diffused through village x Complet
health committee members (HC) increase knowledge and e
uptake of new practices as compared to that diffused
through older women leaders (OWL) or vice versa?
9. Did the EHFP program increase women’s
and/or men’s ownership of assets?
10. Ownership of assets: Household
durables and agricultural assets
35 9
8 ***
30 7
6
25 5
4
20 3 ***
2
15 1
0
10
5
0
Men's Men's Women's Women's
durables durables durables durables
baseline endline baseline endline
Treatment Control Treatment Control
Note: Comparison is to a control group that did not receive any program services. All estimates controlled for baseline
age, sex, clustering, and attrition. All values are coefficient (SE). *** p <0.01
11. Ownership of assets: livestock
25
**
20
15
10
***
5
0
Men's small livestock Men's small livestock Women's small Women's small
baseline endline livestock baseline livestock endline
Treatment Control
Note: Comparison is to a control group that did not receive any program services. All estimates controlled for baseline
age, sex, clustering, and attrition. All values are coefficient (SE). * *p<0.05, *** p <0.01
12. Were women able to maintain control over
the EHFP activities and outputs?
13. Control over EHFP activities and
outputs: Gardens, seeds, and
vegetables
Operations research: Round Operations research: Round
1 (2011) 2 (2012)
100% 100%
90%
90%
80%
70% 80%
60%
50% 70%
40% 60%
30%
20% 50%
10% 40%
0%
30%
20%
10%
0%
Responsible Owns land Makes Manages
for care of for garden decisions revenue
garden on produce generated
Beneficiary Husband Joint Beneficiary Husband Joint
14. Control over EHFP activities and
outputs: Chickens
Operations research: Round Operations research: Round
1 (2011) 2 (2012)
60% 60%
50% 50%
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%
Allowed to sell Keeps income from Responsible for Keeps income from
chickens chickens decicions on chickens chickens
Beneficiary Husband Beneficiary Husband
15. Control over EHFP activities and
outputs: Goats
Operations research: Round Operations research: Round
1 (2011) 2 (2012)
80% 80%
70% 70%
60% 60%
50% 50%
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%
Who makes decisions Who makes decisions Who keeps income
Beneficiary Husband Joint Beneficiary Husband Joint
16. Did the land agreements and/or project activities
influence community norms related to women’s
land ownership or land rights?
17. Community norms related to women’s
land ownership and land rights: Land
acquisition
Land for agricultural purposes is
primarily obtained through inheritance
and gifts.
In general, men obtain land through
inheritance.
Women generally obtain land through
marriage/widowhood or through gifts.
18. Community norms related to women’s land
ownership and land rights: Obstacles to owning
land
Respondents in both beneficiary villages (56%m-63%f) and non-
beneficiary villages (46%m-51%f) reported obstacles to
women’s ability to own land, mainly due to traditional / social
barriers
The most commonly cited ways to improve women’s ability to
own land were to:
Discontinue traditional customs and practices
Sensitize stakeholder’s about women’s ability to own land
Grant pieces of land to women
Respondents in both beneficiary villages (36%m-40%f) and non-
beneficiary villages (24%m-36%f) reported obstacles to
women’s ability to use land, mainly due to lack of inputs such
as seeds, fertilizers or tools and lack of rainfall as well as
traditional practices.
The most commonly cited ways to improve women’s ability to use
land were to:
Provision of inputs
Sensitize stakeholder’s about women’s ability to own land
Grant pieces of land to women
19. Community norms related to women’s land
ownership and land rights: Perceived changes in
women’s ability to own and use land by both
women and men
Women Men
HC OWL All Control HC OWL All Control
(n=70) (n=75) (n=145) (n=75) (n=58) (n=60) (n=118) (n=63)
Change in 46 (66) 49 (65) 95 (62) 11 (15) 32 (55) 36 (60) 68 (57) 14 (22)
opinion
about who
can own
and use
land
Change in 18 (26) 15 (20) 33 (23) 1 (1) 16 (28) 15 (25) 31 (26) 2 (3)
ability to
own land
Change in 29 (41) 32 (43) 61 (42) 3 (4) 27 (47) 21 (35) 48 (41) 1 (2)
ability to
use land
20. Trade-offs required for participation
in the EHFP program
Only 11% of beneficiaries (13/118) stated that
taking care of the garden interferes with their
other activities (e.g. outside work and domestic
tasks).
None of the beneficiary women interviewed
thought that taking care of their chickens
interfered with their other activities.
18% of women (16/89) stated that there were
costs to working at the VMF including having to
neglect their domestic work, taking care of their
children, not being able to go to the market, and
that it takes time to go and work at the VMF.
21. Did exposure to nutrition education diffused through village health
committee members (HC) increase knowledge and uptake of new
practices as compared to that diffused through older women leaders
(OWL) or vice versa?
22. Impact of the EHFP program on
nutrition knowledge: IYCF practices
Children < 6 Begin Begin
Give breast months of age giving giving
Give
milk within should not liquids semi-solid
colostru
the first drink any other than foods at 6
m to
hour after liquids other breast milk months of
children
birth than breast at 6 months age
milk of age
N=1,138 N=1,144 N=1,129 N=1,142 N=1,149
OWL
0.16*** 0.092*** 0.23** 0.13** 0.13**
villages
(0.054) (0.029) (0.094) (0.063) (0.055)
HC villages 0.17*** 0.080*** 0.23*** 0.19*** 0.17***
(0.052) (0.029) (0.078) (0.059) (0.059)
p-value 0.006 0.009 0.006 0.007 0.011
Note: Comparison is to a control group that did not receive any program services. Estimates
controlled for baseline age, sex, clustering, and attrition. ** p < 0.05, *** p<0.01
23. Impact of the EHFP program on health-
related knowledge: Hand-washing practices
50%
**
45%
40%
35%
30% Control villages
25%
Older women leader
20%
villages
15% Health committee
10% villages
5%
0%
Before feeding a Before feeding a
child, baseline child, endline
Note: Comparison is to a control group that did not receive any program services. Estimates
controlled for baseline age, sex, clustering, and attrition. ** p < 0.05
24. Impact of the EHFP program on IYCF practices:
Breastfeeding practices among children 3-12
months of age at endline
100%
90%
80%
70% **
60%
50%
40%
30%
20%
10%
0%
Child ever breastfed
Initiated breastfeeding < 1 hbreastfeeding, children < 6 mo children <
Exclusively after birth
Predominately breastfed,
Control villages Older women leader villages
Health committee villages
Note: Comparison is to a control group that did not receive any program services. Estimates
controlled for baseline age, sex, clustering, and attrition. ** p < 0.05
25. Impact of the EHFP program on IYCF practices:
Dietary diversity among children 3-12 months of age
at baseline
25%
* Met minimum
dietary
20%
diversity
requirement
15% n=691
Older women
0.12*
10% leaders
(0.070)
Health committee 0.098
5% (0.077)
p-value 0.14
Note: Comparison is to a control group that did
not receive any program services. All
0% estimates controlled for baseline
age, sex, clustering, and attrition. All values
Met minimum dietary Met minimum dietary are coefficient (SE). * p<0.10
diversity requirement at diversity requirement at
baseline endline
Control villages Older women leader villages
Health committee villages
26. Impact of the EHFP program on infant and young
child feeding practices: Intake of iron-rich foods
among children 3-12 months of age at baseline
70%
60% ** Iron-rich
foods
50%
n=662
40% Older women
0.15**
leaders
30%
(0.072)
20% Health
0.023
committee
10%
(0.090)
Note: Comparison is to a control group that did
0% p-value 0.13
not receive any program services. All
estimates controlled for baseline
Had iron-rich foods at Had iron-rich foods at age, sex, clustering, and attrition. All values
are coefficient (SE). * *p<0.05
baseline endline
Control villages Older women leader villages
Health committee villages
27. Impact of the EHFP program on nutritional status
of children: Hemoglobin among children 3-12
months of age at baseline
10.0
9.8 *
Hemoglobin
9.6 (g/dL)
n=1144
9.4
Older women
0.24
leaders
9.2
(0.31)
Health committee 0.49*
9.0
(0.27)
8.8 p-value 0.19
Note: Comparison is to a control group that did
not receive any program services. All
estimates controlled for baseline
8.6 age, sex, clustering, and attrition. All values
are coefficient (SE). * p<0.10
Baseline hemoglobin (g/dL) Endline hemoglobin (g/dL)
Control villages Older women leader villages
Health committee villages
28. Change in hemoglobin from baseline to endline
among children 3-5.9 months of age at baseline
10.0
9.8 ** Hemoglobin
(g/dL)
9.6 n=449
Older women
0.044
9.4 leaders
(0.32)
9.2 Health committee 0.76**
(0.30)
9.0 p-value 0.043
Note: Comparison is to a control group that did
not receive any program services. All
8.8 estimates controlled for baseline
age, sex, clustering, and attrition. All values
are coefficient (SE). * *p<0.05
8.6
Baseline hemoglobin (g/dL) Endline hemoglobin (g/dL)
Control villages Older women leader villages
Health committee villages
29. Impact of the EHFP program on nutritional status
of children: HAZ among children 3-12 months of
age at baseline
0.0
-0.2 Baseline HAZ Endline HAZ
-0.4
-0.6
-0.8
-1.0
-1.2
-1.4
-1.6
-1.8
-2.0
Control villages Older women leader villages Health committee villages
30. Impact of the E-HFP program on nutritional status
of children: Wasting among children 3-12 months
of age at baseline
35%
30%
25%
20%
Control villages
15% Older women leader villages
Health committee villages
10%
5%
0%
Baseline prevalence of Endline prevalence of
wasting wasting
31. Summary
Key Question Summary
Did the EHFP program increase Yes, the E-HFP program had a positive impact
women’s ownership of assets? on women's ownership of agricultural assets
and small livestock.
Did the EHFP program also influence Yes, the E-HFP program had a negative
men’s asset holdings? impact on men's ownership of agricultural
assets which was about equal to the positive
impact on women’s ownership of agricultural
assets. The program had a positive impact on
men’s ownership of small livestock which was
larger than that for women.
Were women able to maintain control Yes, especially in regards to the garden
over the EHFP activities and outputs? activities. The vast majority of women were
primarily responsible for decisions related to
what to grow in the garden and were able to
keep the income generated from the sale of
the produce.
32. Summary
Key Question Summary
Did the land agreements and/or project Yes, in beneficiary villages some change was
activities influence community norms noted in people’s opinions about who could
related to women’s land ownership or own and use land. In addition both men and
land rights? women in beneficiary villages reported that
women’s ability to own and use land had
actually changed in the past two years whereas
this was rarely reported in control villages.
What trade-offs were women required Only about 11% of respondents reported that
to make in order to participate in the taking care of their gardens affected their other
EHFP program? activities. 18% said that working at the VMF
had costs (e.g. time and neglect of domestic
work).
Did exposure to nutrition education There do seem to be some differences
diffused through village health although there is not yet a clear pattern of
committee members (HC) increase effects by method of dissemination. Analysis of
knowledge and uptake of new practices related data is on-going.
as compared to that diffused through
older women leaders (OWL) or vice
versa?
Editor's Notes
Addis Ababa -January 9, 2012
Physical capital through project inputs and productsFinancial capital through increased revenue from household gardens Social capital through village model farms and behavior change communication strategyHuman capital through agriculture and nutrition training and improved knowledge and adoption of best practices in agriculture and nutrition and subsequent improved nutritional status
Impact estimates: Men = 3.67 (1.71); n=1380, p=0.036 Women = 2.75 (0.78); n=1380, p=0.001
Round 1: 85% of beneficiaries (compared to 4% of non-beneficiaries) had a home garden and for 91% of these women this was new since joining the program.The vast majority of the women credit the program with increasing the increasing production of vegetables and 74% also thought the program had increased production of chickens.
Nearly half of respondents in all groups report that men generally inherit land when wife passes; unless children are mature enough to inherit itMore than half of respondents in all groups report that women do not inherit land from their husband after his passing due to tradition and related inheritance and usage rules (depends on whether there are children, age of woman, etc.)
To discontinue practices respondents nearly all thought that village chiefs/leaders would need to take the lead.For sensitization and granting land, respondents had more varied opinions stating that traditional leaders, government and local authorities as well as non-governmental organizations should be involved.with the greater proportion of women in control (26%) as compared to intervention villages (7%) stating this as an obstacle to women’s ability to use landFor provision of inputs and sensitizationrespondents nearly all thought that government or non-governmental organizations would need to take the lead.For granting land, respondents had more varied opinions stating that traditional leaders, husbands, government and local authorities as well as non-governmental organizations should be involved.
Close or equal to 90% of all respondents report nochangesin men’s ability to own land over the past 2 yearsOver 90% of men and women in control villages report no changes in women’s ability to own land. In beneficiary villages however, some change is reported (see next slide), due to land grants by HKI or husbands, and establishment of (community) gardens for women.The changes that were cited in regards to changes in women’s ability to own land primarily had to do with women in intervention villages being granted land by their husbands or HKI, that they now have community gardensThe changes that were mentioned related to changes in women’s ability to use land primarily revolved around increased women’s access to land due to transfers from men and advocacy and to support provided to women in terms of inputs and equipment. Men in beneficiary villages also cited the production trainings provided by HKI as increasing women’s capacity to use land and that the women were now using the land at the VMF for production.
The most notable change in knowledge related to important times to wash hands was the change related to washing hands before feeding children. At endline a little less than half of the caregivers interviewed living in HC villages correctly stated this time for hand washing whereas only about 30% of caregivers living in control villages and OWL villages correctly identified this important time to wash hands. There was about a 12% difference in the change proportion from baseline to endline between caregivers living in HC villages as compared to those living in control villages who correctly stated this. About 65% of the caregivers interviewed also knew that hands should be washed before eating. Very few caregivers mentioned that it was also important to wash their hands after using the toilet (about 10%) or after cleaning a child who had defecated (about 5%). This is an area that could be improved in future programs as these are common routes through which children get exposed to infectious agents.
Nearly all children who were between the ages of 3 and 12 months at the time of the endline evaluation had been breastfed, as was seen among the children included in the baseline study. A greater proportion of caregivers in all types of villages reported having initiated breastfeeding within the first hour of birth as compared to those interviewed at baseline, and caregivers from both types of treatment villages were more likely than those in the control villages to report having done this ideal practice. The difference between OWL villages and control villages was about 16% and was statistically significant. The proportion of caregivers who had exclusively or predominately breastfed their children < 6 months of age in the previous 24 hours was similar across the control and treatment villages.
Very few children met their minimum dietary diversity requirements (4 out of 7 food groups) at baseline. At endline, although the percentage of children that met this requirement increased in all groups, there were still only 17% of children who had met the minimum requirements across the sample of children who had both baseline and endline data. At endline, children in the two treatment groups were almost twice as likely to have met their requirements for minimum dietary diversity. However, only the difference in the change from baseline to endline between the OWL treatment group and the control group was statistically significant (p< 0.10).
At baseline only about 10% of the children between the ages of 6 and 12 months had eaten iron-rich foods during the 24 hours prior to the baseline survey. At the time of the endline survey when these children were between 21 and 40 months of age more than half of them had eaten an iron-rich food during the 24 hours before participating in the endline survey. Children living in OWL villages were the most likely to have eaten an iron-rich food in the previous 24 hours and the change in proportion from baseline to endline was about 15% greater among children living in OWL villages as compared to those living in control villages.
After two years of participating in the E-HFP program, children living in HC treatment villages had a higher mean hemoglobin concentration than those living in control villages. The difference in change from baseline to endline was about 0.5 g/L between the control group and the health committee treatment group.
The impact of the E-HFP program on improving hemoglobin concentration was even more evident among children who were 3-5.9 months of age at the time of the baseline study. Again children living in HC villages had a significantly higher mean hemoglobin concentration as compared to those living in control villages. Among these younger children the difference in the change of hemoglobin concentration over the two years of program implementation was 0.76 g/dL between the control group and the HC treatment group.
As is common among young children at-risk for nutritional deficiencies children’s growth was already faltering at 3-12 months of age (at basleine) and declined as children got older. However, there were no significant differences between the control and the treatment groups in the change in HAZ over time. Furthermore, there were no significant differences between the treatment groups in the change in the prevalence of stunting, WAZ scores or the prevalence of underweight among children who were 3-12 months of age at baseline.
The improvements in the WHZ scores and decline in the prevalence of wasting are noteworthy. However, again there were no statistically significant differences between the control group and either of the treatment groups for changes in the prevalence of wasting or for the change in WHZ scores from baseline to endline. The changes in prevalence for the three groups was (control = -15%, older women leaders = -17% and health committee –23%). It is possible that we did not have enough power to detect differences between the treatment and control groups for this outcome.