An Integrated Enhanced Infant and Young Child Feeding (IYCF) and Micronutrient Powder Intervention Improved Select IYCF Practices Among Caregivers of Children Aged 12–23 Months in Eastern Uganda

ABSTRACT

Background

There is little evidence of the impact of integrated programs distributing nutrition supplements with behavior change on infant and young child feeding (IYCF) practices.

Objective

We evaluated the impact of an integrated IYCF/micronutrient powder intervention on IYCF practices among caregivers of children aged 12–23 mo in eastern Uganda.

Methods

We used pre-post data from 2 population-based, cross-sectional surveys representative of children aged 12–23 mo in Amuria (intervention) and Soroti (nonintervention) districts (n = 2816). Caregivers were interviewed in June/July at baseline in 2015 and 12 mo after implementation in 2016. We used generalized linear mixed models with cluster as a random effect to calculate the average intervention effect on receiving IYCF counseling, ever breastfed, current breastfeeding, bottle feeding, introducing complementary feeding at age 6 mo, continued breastfeeding at ages 1 and 2 y, minimum meal frequency (MMF), minimum dietary diversity, minimum acceptable diet (MAD), and consumption of food groups the day preceding the survey.

Results

Controlling for child age and sex, household wealth and food security, and caregiver schooling, the intervention was positively associated with having received IYCF counseling by village health team [adjusted prevalence difference-in-difference (APDiD): +51.6%; 95% CI: 44.0%, 59.2%]; timely introduction of complementary feeding (APDiD: +21.7%; 95% CI: 13.4%, 30.1%); having consumed organs or meats (APDiD: +9.0%; 95% CI: 1.4%, 16.6%) or vitamin A–rich fruits or vegetables (APDiD: +17.5%; 95% CI: 4.5%, 30.5%); and MMF (APDiD: +18.6%; 95% CI: 11.2%, 25.9%). The intervention was negatively associated with having consumed grains, roots, or tubers (APDiD: −4.4%; 95% CI: −7.0%, −1.7%) and legumes, nuts, or seeds (APDiD: −15.6%; 95% CI: −26.2%, −5.0%). Prevalences of some IYCF practices were low in Amuria at endline including MAD (19.1%; 95% CI :16.3%, 21.9%).

Conclusions

The intervention had a positive impact on several IYCF practices; however, endline prevalence of some indicators suggests a continued need to improve complementary feeding practices.

This paper examines the impact of an integrated infant and young child feeding (IYCF)/micronutrient powder intervention on IYCF practices among caregivers of children 12–23 mo in Eastern Uganda.

Introduction

The WHO recommends exclusive breastfeeding through age 6 mo, at which point breast milk is no longer sufficient to meet all nutrient needs (1). Soft, semi-soft, or solid complementary foods should be introduced beginning around age 6 mo, along with continued breastfeeding until age 24 mo or older when the child is fully weaned (1). This transition period is associated with malnutrition in settings with poor infant and young child feeding (IYCF) practices. Poor IYCF practices and child malnutrition are prevalent in Uganda. According to the 2016 Uganda Demographic Health Survey (DHS), 8 in 10 children aged 6–9 mo received complementary foods and only 14.6% of children aged 6–23 mo consumed a minimum acceptable diet (MAD) (2). In the same survey, 28.9% of children aged 6–59 mo suffered from stunting and 52.8% had anemia (2).

Traditionally, nutrition counseling has been the primary intervention to improve child feeding practices and nutritional status (3). Beginning in 2011, the WHO issued guidance on home fortification with micronutrient powders (MNPs) to address anemia and micronutrient deficiencies in children aged 6–23 mo in contexts where anemia is ≥20% among children <2 y or children <5 y (4). MNPs are prepackaged combinations of vitamins and minerals that can be mixed into a child's usual food beginning at age 6 mo. To improve IYCF practices and reduce iron deficiency anemia among children aged 6–23 mo, the Uganda Ministry of Health (MoH), in collaboration with the UN World Food Program (WFP) implemented an integrated IYCF/MNP pilot program in Amuria and Katakwi districts in eastern Uganda. The implementing partners planned a program evaluation to understand effectiveness and identify lessons learned for scaling up the intervention in Uganda.

Countries are increasingly adopting MNP programs as a strategy to address micronutrient deficiencies in children. As of 2015, 65 countries had implemented MNP programs (5), and the number of children benefiting from these programs has more than quadrupled from 4 million children in 2014 to 18 million children in 2018 (6). Despite the proliferation of MNP programs, many of which are integrated into existing child nutrition programs, relatively few studies have examined the impact of programs integrating MNP distribution with behavior-change communication on IYCF practices (7–10). MNP theoretically should not displace the normal diet because it is noncaloric and is added to the child's typical foods; however, whether these nutrition supplements influence breastfeeding promotion or feeding diverse, local diets has not been widely explored.

The objective of this analysis was to determine the impact of the pilot program on caregiver IYCF practices in eastern Uganda.

Methods

IYCF and MNP pilot

The year-long IYCF/MNP pilot project started in July 2015 and was implemented in Amuria and Katakwi districts, with support from the implementing partner—Andre Food Consults. The intervention focused on caregiver behavior change to generate demand for MNP and to motivate optimal IYCF practices for children <2 y. Prior to the development of the pilot program, the MoH and MNP Technical Working Group led a situation analysis, MNP package development, and a 20-d MNP acceptability study (11) to create a tailored, evidence-based intervention. This formative research informed the MNP and IYCF behavior-change strategy that was operationalized using different distribution approaches in multiple Ugandan districts piloting MNPs, including Amuria and Katakwi.

MNP was distributed monthly at MoH facilities such as government outposts and health centers and by village health teams (VHTs) in communities at central locations in villages and via home delivery. MNP was distributed free of charge to children aged 6–23 mo according to a schedule of 1 box of 30 sachets every 2 mo. Caregivers also received program materials including MNP ration and adherence cards to promote participation in the pilot and correct MNP dosing. Messages about IYCF and MNP were also spread through a mass media campaign that included radio jingles, brochures, stickers, posters, and partnerships with women's groups.

IYCF counseling and training sessions were carried out in tandem with monthly MNP distribution through interactive group discussion and food-preparation demonstrations by VHTs and MoH staff. The intervention aimed to change typical porridge preparation practices to increase the energy and nutrient density through appropriate consistency (i.e., thick enough to stay on a spoon) and volume. These sessions also promoted feeding a variety of local foods from different food groups at each meal and provided examples of balanced meals to mix with MNP. Dietary diversity education focused on “Go, Glow and Grow” foods—terms used in the preschool curriculum in Uganda to describe the different food groups and what they do for the human body. Additionally, caregivers received information about how frequently children should receive meals and snacks at different ages. Other education topics included anemia, limiting tea consumption to reduce intake of iron-inhibiting compounds, and good hygiene practices to prevent illness.

Study population

The impact evaluation used pre-post, cross-sectional, population-based surveys representative of children aged 12–23 mo in Amuria (intervention) and Soroti (nonintervention) districts. Although the intervention was implemented in Amuria and Katakwi, the impact evaluation did not include Katakwi due to resource constraints. In presurvey planning, we selected Soroti as the most suitable nonintervention district due to its geographic proximity to Amuria as well as its ethnic, cultural, and socioeconomic characteristics.

The impact evaluation was powered to assess changes in the prevalence of anemia and iron deficiency (12). Changes in IYCF practices were secondary outcomes. The sample size was calculated for 2 sequential, clustered cross-sectional surveys (pre- and postintervention) of equal sample size (13), assuming 60% anemia, 25% iron deficiency, 80% power, 2-sided ɑ of 0.05, an individual response rate of 75%, and a design effect of 1.56 based on the design effect for the rural sample for anemia in children aged 6–59 mo in the 2011 DHS (14). The resultant sample size was 833 children aged 12–23 mo per district for each survey cross-section and included a 30% increase in the number of participants to reflect the estimated prevalence of inflammation and the potential need to exclude or adjust select biomarkers for inflammation in the analysis of primary outcomes (15).

Makerere University led the data collection with oversight from the MoH and WFP. The US CDC provided technical assistance. Baseline and endline data were collected in June–July 2015 and June–July 2016, respectively, 12 mo after program implementation. Similar protocols were followed for both the baseline and endline surveys. Using a multistage cluster-sampling design, we randomly selected 38 clusters without replacement using probability proportional to population size. After completing a census in each selected cluster, we then randomly selected 22 children ages 12–23 mo from each cluster. Clusters and children were sampled without replacement. Because Amuria is rural, we only sampled rural clusters in Soroti; thus, this evaluation is representative only of rural Soroti.

The national census overestimated the number of children aged 12–23 mo in Amuria and Soroti, so fewer than the expected 833 children were residing in the selected clusters. At baseline, 579 of the 637 (90.9%) eligible children in the selected clusters in Amuria and 681 of the 771 (88.3%) eligible children in the selected clusters in Soroti consented to participate. At endline, 761 of the 806 (94.4%) eligible children in the clusters in Amuria and 797 of the 832 (95.8%) eligible children in the clusters in Soroti consented to participate. In total, the combined baseline and endline participation was 92.9% in Amuria and 92.2% in Soroti (n = 2818 children). All children had data on at least 1 IYCF practice outcome. We excluded 2 children (<1%) for missing information on covariates. The final analytic sample included 2816 children.

The School of Biomedical Sciences Higher Degrees, Research and Ethics Committee, College of Health Sciences, Makerere University, granted ethical approval. The Uganda National Council for Science and Technology granted research clearance. Caregivers provided informed consent on behalf of their child(ren) and for themselves. Caregivers received a bar of soap and a medium-sized tub of petroleum jelly as a token of appreciation for participating in the survey. A subsample of children who additionally participated in a test to measure modified relative dose response as part of vitamin A status assessment received an additional long bar of soap and tub of petroleum jelly, a bottle of water, soft drink, and lunch while they waited at the field laboratory for 4 h after dosing for blood draws.

Data collection

The field survey team at baseline and endline participated in an intensive 2-wk training conducted by core survey team members from the CDC, including classroom instruction, demonstrations, role play, and mock interviews. The teams also conducted a 3-d pilot to test survey tools and field procedures. Supervisors observed interviewers during field work.

Enumerators collected data from caregivers using a pretested interviewer-administered questionnaire on the caregiver's experience with the IYCF/MNP pilot, IYCF knowledge and practices, child sociodemographic and health characteristics, child consumption of food groups the day preceding the survey, and household-level data including sociodemographic characteristics, food security, assets, and housing characteristics.

Outcomes

IYCF counseling

Caregivers reported whether they had ever received IYCF counseling from 1) a VHT and 2) facility-based health workers.

IYCF practices

Ever being breastfed was defined as the child having ever received breast milk. Children whose caregivers reported breastfeeding the day or night preceding the survey and those who received breast milk in a different way (e.g., by spoon, cup, or bottle) were classified as currently breastfed. Bottle feeding was defined as having been fed any liquid, including breast milk, from a bottle the day preceding the survey. Continued breastfeeding at ages 1 and 2 y were defined as current breastfeeding among children aged 12–15 mo and children aged 20–23 mo, respectively, per WHO guidelines (16). Timely introduction of complementary foods was defined as the child consuming solid, semi-solid, or soft foods beginning at age 6 mo. We defined minimum dietary diversity (MDD) as having received foods from ≥4 of the following food groups the day preceding the survey: 1) grains, roots, or tubers; 2) legumes, nuts, or seeds; 3) dairy products; 4) flesh foods; 5) eggs; 6) vitamin A–rich fruits or vegetables; and 7) other fruits or vegetables (16). Minimum meal frequency (MMF) was defined as having received solid, semi-solid, or soft foods ≥3 times for breastfed children and ≥4 times for non-breastfed children the day preceding the survey (16). Feeding frequency for non-breastfed children included both milk feeds and solid/semi-solid foods. We defined MAD as having had at least the MDD and the MMF for breastfed children and having had ≥2 milk feeds, MDD not including milk feeds, and MMF for non-breastfed children (16).

We additionally assessed the impact of the pilot on consumption of the following food groups the day preceding the survey: 1) grains, roots, or tubers; 2) legumes, nuts, or seeds; 3) dairy products; 4) organs, meats, or seafood; 5) eggs; 6) vitamin A–rich fruits or vegetables; and 7) other fruits or vegetables. Because the intervention promoted organs and red meat as rich sources of iron, we also divided flesh foods into 1) organs or meats and 2) seafood.

Covariates

Covariates included child sex (female vs. male), child age group (12–17 mo vs. 18–23 mo), caregiver schooling (less than a primary education vs. primary or higher), household wealth tertile, and severe household food insecurity. We created a household wealth index based on housing characteristics, assets, and service amenities using principal components analysis (2). We then categorized household wealth into tertiles. Severely food-insecure households were those who experienced any of the following conditions during the 12 mo preceding the survey: often cut back on meal size or number of meals, ever experienced no food to eat because of lack of resources, had any household member go to sleep hungry at night because there was not enough food, and/or had any household member go an entire day and night without eating because there was not enough food (17).

Statistical analyses

We used an intent-to-treat difference-in-difference (DiD) modeling approach to assess the impact of the IYCF/MNP intervention on indicators of IYCF practice (15, 18). The DiD method quantifies the additive impact of the IYCF/MNP intervention in Amuria (the intervention district) relative to Soroti (the nonintervention district) between 2015 and 2016 net of differences due to district-level differences or secular trends (15). We used generalized linear mixed models with cluster as a random effect to calculate the average intervention effect, presented as adjusted prevalence DiDs (APDiDs). The base models included a district variable (Amuria vs. Soroti) to account for the fixed effects of intervention assignment and district, a variable for time [post-(2016) vs. pre-(2015) pilot implementation], a district × time interaction term, sex, and age group (15). Adjusted models for all outcomes additionally controlled for household wealth, severe household food insecurity, and caregiver schooling.

Additionally, among children in Amuria at endline, we assessed coverage of IYCF counseling and IYCF practices by intensity of intervention coverage. We categorized program intensity as follows: 1) low VHT contact and low MNP coverage (reference group), 2) high VHT contact only, 3) high MNP coverage only, and 4) high VHT contact and high MNP coverage. High contact with VHT was defined as the caregiver seeking or receiving services from a VHT ≥1 time/mo. We defined high MNP coverage as having reported receiving at least ≥60 sachets every 6 mo during the pilot year—the minimum recommended dose per the 2011 WHO guideline (19). We calculated the expected number of sachets received based on the child's age at the time of data collection, assuming the children began receiving MNP as soon as they were eligible (i.e., at age 6 mo or the start of the intervention, whichever came later). Using multivariable logistic regression, we regressed program intensity on each outcome, adjusting sex, age group, caregiver schooling, and household wealth and food security.

We analyzed the data using SAS version 9.4 PROC MIXED procedures (SAS Institute, Inc.) and applied sampling weights to account for complex sampling design. We considered 2-sided P values < 0.05 to be statistically significant.

Results

Child sociodemographic and health characteristics, IYCF practices, and program coverage indicators at baseline and endline are presented in Table 1. In total, 94.3% of caregivers were the mothers of the index child (data not shown). At baseline, 47.5% (95% CI: 44.9%, 50.1%) of the participants were female and 53.4% (95% CI: 50.3%, 56.5%) were aged 18–23 mo (data not shown). Fifty-seven percent (95% CI: 52.9%, 62.1%) of children in Amuria and 64.0% (95% CI: 61.2%, 66.7%) of children in Soroti were currently breastfed. Few children at baseline consumed an MAD [4.8% (95% CI: 2.7%, 7.0%) in Amuria and 10.7% (95% CI: 8.4%, 13.1%) in Soroti].

TABLE 1.

Select characteristics of children aged 12–23 mo at baseline (2015) and endline (2016) in Amuria (intervention) and Soroti (nonintervention) districts: infant and young child feeding/micronutrient powder intervention surveys, Uganda¹

	Amuria (intervention district)				Soroti (nonintervention district)
	Baseline, 2015		Endline, 2016		Baseline, 2015		Endline, 2016
	n	% (95% CI)	n	% (95% CI)	n	% (95% CI)	n	% (95% CI)
Female, %	579	46.5 (42.8, 50.1)	760	47.2 (43.7, 50.8)	680	48.5 (44.7, 52.4)	797	46.9 (43.4, 50.4)
Age group, %	579		760		680		797
12–17 mo		45.8 (40.8, 50.7)		60.3 (56.8, 63.8)		47.4 (43.7, 51.0)		59.7 (56.3, 63.1)
18–23 mo		54.2 (49.3, 59.2)		39.7 (36.2, 43.2)		52.6 (49.0, 56.3)		40.3 (36.9, 43.7)
Household wealth tertile,2 %	579		760		680		797
Poorest		38.2 (32.3, 44.1)		34.3 (31.1, 37.6)		29.3 (25.0, 33.6)		30.0 (26.8, 33.1)
Middle		34.5 (30.1, 39.0)		34.9 (31.5, 38.3)		32.4 (28.7, 36.0)		33.1 (29.9, 36.4)
Wealthiest		27.3 (22.0, 32.6)		30.8 (27.7, 33.9)		38.4 (34.3, 42.5)		36.9 (33.7, 40.1)
Caregiver has less than a primary education, %	579	73.1 (68.1, 78.0)	760	70.8 (67.6, 73.9)	680	71.6 (67.4, 75.8)	797	70.3 (67.1, 73.4)
Severe household food insecurity,3 %	579	4.0 (2.0, 5.9)	760	12.8 (10.4, 15.1)	680	2.1 (0.8, 3.3)	797	11.3 (9.1, 13.5)
High contact with VHT,4 %	579	20.6 (16.3, 24.8)	760	64.4 (60.0, 68.7)	680	17.6 (12.3, 23.0)	797	49.2 (43.8, 54.6)
Ever received MNP, %	—	—5	722	95.3 (93.8, 96.8)	—	—5	797	0.5 (0.0, 1.0)6
High MNP sachet coverage,7 %	—	—	760	53.4 (50.0, 56.8)	—	—	797	0.1 (0.0, 0.4)6
Recent MNP intake,8 %	—	—	688	65.3 (61.7, 68.8)	—	—	4	0.0 (0.0, 0.0)6
Intensity of program coverage,4,7 %	—		760		—		—
Low VHT contact and low MNP coverage		—		30.1 (27.0, 33.9)		—		80.3 (77.7, 82.9)
High VHT contact only		—		16.4 (13.8, 19.1)		—		19.7 (17.1, 22.3)
High MNP coverage only		—		30.7 (27.4, 33.9)		—		0.1 (0.0, 0.4)6
High VHT contact and high MNP coverage		—		22.8 (19.9, 25.7)		—		0.0 (0.0, 0.0)6
VHT ever spoke with caregiver about how to feed the child, %	577	15.9 (13.0, 18.9)	760	63.0 (59.7, 66.4)	680	15.6 (11.2, 20.0)	797	11.5 (9.4, 13.7)
Health center staff ever spoke to caregiver about how to feed the child, %	577	65.7 (60.5, 70.9)	760	54.2 (50.6, 57.8)	679	68.9 (64.8, 73.1)	797	58.1 (54.7, 61.5)
Ever breastfed, %	578	99.0 (98.2, 99.7)	760	94.2 (92.5, 95.9)	679	98.4 (97.4, 99.4)	795	94.6 (93.0, 96.2)
Current breastfeeding, %	579	57.5 (52.9, 62.1)	760	53.0 (49.5, 56.5)	680	64.0 (61.2, 66.7)	796	60.9 (57.6, 64.3)
Bottle feeding, %	577	1.0 (0.1, 2.0)6	760	1.7 (0.8, 2.6)	678	1.5 (0.7, 2.3)	796	1.8 (0.9, 2.7)
Continued breastfeeding at age 1 y,9 %	164	91.5 (87.1, 95.8)	224	92.4 (88.9, 95.9)	203	95.6 (92.7, 98.4)	224	89.7 (85.6, 93.9)
Continued breastfeeding at age 2 y,10 %	200	28.5 (22.7, 34.3)	253	20.6 (15.7, 25.4)	219	33.8 (27.9, 39.7)	263	29.3 (23.8, 34.7)
Introduced complementary foods at age 6 mo, %	579	42.0 (37.7, 46.2)	760	66.6 (63.3, 69.9)	680	57.1 (52.3, 61.8)	797	59.3 (55.9, 62.8)
Prior day food consumption, %
Grains, roots, or tubers	578	99.1 (98.4, 99.8)	760	95.8 (94.4, 97.2)	680	96.9 (95.3, 98.6)	796	97.9 (96.9, 98.9)
Legumes, nuts, or seeds	578	72.8 (66.6, 79.1)	760	59.7 (56.3, 63.2)	680	61.6 (55.3, 67.9)	796	64.4 (61.2, 67.7)
Dairy	577	7.6 (5.0, 10.2)	760	6.6 (4.8, 8.3)	680	6.8 (4.6, 8.9)	794	3.8 (2.5, 5.1)
Flesh foods (organs, meats, or seafood)	578	47.2 (41.2, 53.3)	760	52.8 (49.2, 56.3)	680	55.7 (51.6, 59.8)	795	55.8 (52.5, 59.2)
Organs or meats	578	12.1 (8.1, 16.1)	760	24.3 (21.3, 27.4)	680	24.6 (19.8, 29.3)	795	27.5 (24.4, 30.7)
Seafood	578	37.7 (32.8, 42.6)	760	32.1 (28.9, 35.3)	680	36.0 (32.1, 40.0)	795	33.5 (30.2, 36.7)
Eggs	578	3.1 (1.2, 5.0)	760	8.4 (6.5, 10.4)	680	5.0 (3.2, 6.8)	795	6.5 (4.8, 8.3)
Vitamin A–rich fruits or vegetables	578	31.7 (24.8, 38.5)	760	77.1 (74.5, 79.7)	680	58.8 (52.5, 65.2)	795	84.7 (82.3, 87.0)
Other fruits and vegetables	579	0.3 (0.0, 0.8)6	760	0.1 (0.0, 0.4)6	680	0.0 (0.0, 0.0)6	797	0.5 (0.0, 1.0)6
Minimum dietary diversity,11 %	579	18.3 (13.6, 23.0)	760	59.9 (56.5, 63.4)	680	26.6 (22.5, 30.8)	797	61.6 (58.3, 65.0)
Minimum meal frequency,12 %	579	16.4 (12.6, 20.2)	760	27.1 (24.0, 30.3)	680	33.8 (30.3, 37.4)	797	25.3 (22.4, 28.3)
Minimum acceptable diet,13 %	579	4.8 (2.7, 7.0)	760	19.1 (16.3, 21.9)	680	10.7 (8.4, 13.1)	797	18.9 (16.3, 21.6)

¹Values % (95% CIs) unless otherwise indicated; n values are unweighted. All estimates account for weighting and complex sampling design. MNP, micronutrient powder; VHT, village health team.

²Household wealth index was based on a principal components analysis of household assets, housing characteristics, and service amenities, per the guidelines for the Uganda Demographic and Health Survey, and divided household wealth into tertiles (2).

³Severely food-insecure households were those who experienced any of the following conditions during the 12 mo preceding the survey: often cut back on meal size or number of meals; ever experienced no food to eat because of lack of resources; had any household member going to sleep hungry at night because there was not enough food; and/or had any household member go an entire day and night without eating because there was not enough food (17).

⁴High contact with VHT defined as seeking or getting services ≥1 time/mo.

⁵The results from the baseline survey found only 2 children in total reported receiving MNP during the 3 mo preceding the survey. Of those, only 1 reported ever consuming foods mixed with MNP.

⁶Interpret with caution. Estimates may be unstable (numerator n <10).

⁷High MNP sachet coverage was defined as having reported receiving at least 60 sachets (2 boxes) every 6 mo during the 12-mo pilot—the minimum recommended dose per the 2011 WHO guideline (19). We calculated the expected number of sachets received based on the child's age at the time of data collection, assuming the child began receiving MNP as soon as he/she was eligible (i.e., at age 6 mo or the start of the intervention, whichever came later).

⁸Recent MNP intake defined as consumed MNP during the 2 wk preceding the endline survey.

⁹Among children 12–15 mo (16).

¹⁰Among children 20–23 mo (16).

¹¹We defined minimum dietary diversity as having received foods from ≥4 of the following food groups the day preceding the survey: 1) grains, roots, or tubers; 2) legumes, nuts, or seeds; 3) dairy products; 4) flesh foods; 5) eggs; 6) vitamin A–rich fruits or vegetables; and 7) other fruits or vegetables (16).

¹²Minimum meal frequency was defined as having received solid, semi-solid, or soft foods (meals and/or snacks) at least 3 times for breastfed children and 4 times for non-breastfed children the day preceding the survey (16). Feeding frequency for non-breastfed children included both milk feeds and solid/semi-solid foods.

¹³We defined minimum acceptable diet as having had at least the minimum dietary diversity and the minimum meal frequency for breastfed children and having had at least 2 milk feeds, minimum dietary diversity not including milk feeds, and minimum meal frequency for non-breastfed children (16).

Intervention coverage was high in Amuria. At endline, 95.3% (95% CI: 93.8%, 96.8%) of children in Amuria had ever received MNP compared with <1% of children in Soroti. Sixty-five percent (95% CI: 61.7%, 68.8%) of children in the intervention district consumed MNP during the 2 wk preceding the survey compared with 4 children in the nonintervention district. At endline, 63.0% (95% CI: 59.7%, 66.4%) of caregivers in Amuria reported receiving IYCF counseling from a VHT compared with 11.5% (95% CI: 9.4%, 13.7%) of caregivers in Soroti. Among children in Amuria at endline, 30.1% (95% CI: 27.0%, 33.9%) had both low VHT contact and low MNP coverage, 16.4% (95% CI: 13.8%, 19.1%) had high VHT contact only, 30.7% (95% CI: 27.4%, 33.9%) had high MNP coverage only, and 22.8% (95% CI: 19.9%, 25.7%) had both high VHT contact and high MNP coverage.

IYCF counseling coverage

The intervention was associated with +51.6% (95% CI: 44.0%, 59.2%) prevalence of having received IYCF counseling by a VHT (Table 2). The intervention was not associated with IYCF counseling by health center staff.

TABLE 2.

APDiD of IYCF counseling coverage among caregivers of children aged 12–23 mo between intervention (Amuria) and nonintervention (Soroti) districts: IYCF/micronutrient powder intervention baseline survey 2015 and endline survey 2016, Amuria and Soroti districts, Uganda¹

	APDiD (95% CI)	P
Received IYCF counseling by VHT worker
Base model	51.6 (44.0, 59.2)	<0.0001
Adjusted model	51.6 (44.0, 59.2)	<0.0001
Received IYCF counseling by health center staff
Base model	−0.9 (−8.9, 7.1)	0.8
Adjusted model	−1.0 (−9.0, 6.9)	0.8

¹Unweighted sample sizes are 2814 (counseling by VHT) and 2813 (counseling by health center staff). Base model estimates are the difference in the prevalence (95% CI) for exposure to the IYCF/MNP intervention controlling for fixed effects of district (Amuria vs. Soroti), year (2016 vs. 2015), child sex, and age group (12–17 mo vs. 18–23 mo). The adjusted model additionally controls for household wealth tertile, severe household food insecurity, and caregiver schooling level (less than primary education vs. primary or higher). CIs account for weighting and complex sampling design. APDiD, adjusted prevalence difference-in-difference; IYCF, infant and young child feeding; MNP, micronutrient powder; VHT, village health team.

Program impact on IYCF practices

The intervention was associated with +21.7% (95% CI: 13.4%, 30.1%; P < 0.0001) prevalence of timely introduction of complementary foods (Table 3). There was no impact on ever breastfeeding, current breastfeeding, continued breastfeeding at age 1 or 2 y, or bottle feeding.

TABLE 3.

APDiD of IYCF practices among caregivers of children aged 12–23 mo between intervention (Amuria) and nonintervention (Soroti) districts: IYCF/micronutrient powder intervention baseline survey 2015 and endline survey 2016, Amuria and Soroti districts, Uganda¹

	APDiD (95% CI)	P
Ever breastfed
Base model	−1.0 (−3.8, 1.7)	0.5
Adjusted model	−0.9 (−3.7, 1.8)	0.5
Current breastfeeding
Base model	−2.6 (−9.5, 4.3)	0.5
Adjusted model	−2.3 (−9.1, 4.4)	0.5
Bottle feeding
Base model	0.4 (−1.6, 2.4)	0.7
Adjusted model	0.3 (−1.6, 2.3)	0.7
Continued breastfeeding at age 1 y2
Base model	7.0 (−0.3, 14.4)	0.06
Adjusted model	7.1 (−0.3, 14.4)	0.06
Continued breastfeeding at age 2 y3
Base model	−3.4 (−14.9, 8.0)	0.6
Adjusted model	−3.6 (−15.1, 7.8)	0.5
Introduced complementary foods at age 6 mo
Base model	22.1 (13.7, 30.5)	<0.0001
Adjusted model	21.7 (13.4, 30.1)	<0.0001
Consumed grains, roots, or tubers
Base model	−4.3 (−7.0, −1.6)	0.002
Adjusted model	−4.4 (−7.0, −1.7)	0.001
Consumed legumes, nuts, or seeds
Base model	−15.2 (−25.9, −4.5)	0.005
Adjusted model	−15.6 (−26.2, −5.0)	0.004
Consumed dairy
Base model	1.8 (−2.2, 5.9)	0.4
Adjusted model	1.8 (−2.4, 5.9)	0.4
Consumed flesh foods (organs, meats, or seafood)
Base model	5.8 (−3.9, 15.4)	0.2
Adjusted model	5.3 (−4.3, 15.0)	0.3
Consumed organs or meats
Base model	9.5 (1.8, 17.2)	0.01
Adjusted model	9.0 (1.4, 16.6)	0.02
Consumed seafood
Base model	−3.0 (−12.1, 6.0)	0.5
Adjusted model	−3.1 (−12.2, 6.0)	0.5
Consumed eggs
Base model	3.7 (−0.2, 7.6)	0.06
Adjusted model	3.5 (−0.3, 7.3)	0.07
Consumed vitamin A–rich fruits and vegetables
Base model	17.7 (4.7, 30.7)	0.008
Adjusted model	17.5 (4.5, 30.5)	0.008
Consumed other fruits or vegetables
Base model	−0.7 (−1.4, 0.02)	0.06
Adjusted model	−0.7 (−1.4, 0.01)	0.05
Consumed minimum dietary diversity
Base model	6.8 (−1.5, 15.0)	0.1
Adjusted model	6.0 (−2.2, 14.1)	0.1
Consumed minimum meal frequency
Base model	18.9 (11.4, 26.3)	<0.0001
Adjusted model	18.6 (11.2, 25.9)	<0.0001
Consumed minimum acceptable diet
Base model	5.9 (0.3, 11.6)	0.04
Adjusted model	5.6 (0.02, 11.2)	0.05

¹Unweighted sample sizes are 815 (continued breastfeeding at age 1 y), 935 (continued breastfeeding at age 2 y), 2811 (bottle feeding; consume dairy), 2812 (ever breastfed), 2813 (consumed organs, meats, or seafood; consumed organs or meats; consumed seafood; consumed eggs; consumed vitamin A–rich fruits or vegetables), 2814 (consumed grains, roots, or tubers; consumed legumes, nuts, or seeds), 2815 (current breastfeeding), and 2816 (complementary foods introduced at age 6 mo; consumed other fruits or vegetables; consumed minimum dietary diversity; consumed minimum meal frequency; consumed minimum acceptable diet). Base model estimates are the difference in the prevalence (95% CI) for exposure to the MNP intervention controlling for fixed effects of district (Amuria vs. Soroti), year (2016 vs. 2015), child sex and age group (12–17 mo vs. 18–23 mo). The adjusted model additionally controls for household wealth tertile, severe household food insecurity, and caregiver schooling level (less than primary education vs. primary or higher). CIs account for weighting and complex sampling design. APDiD, adjusted prevalence difference-in-difference; IYCF, infant and young child feeding; MNP, micronutrient powder.

²Among children 12–15 mo (16).

³Among children 20–23 mo (16).

The intervention was associated with select indicators of food intake the day preceding the survey including +9.0% prevalence of organ or meat consumption (95% CI: 1.4%, 16.6%; P = 0.02) and +17.5% consumption of vitamin A–rich fruits and vegetables (95% CI: 4.5%, 30.5%; P = 0.008). The intervention was associated with a lower prevalence of reported consumption of grains, roots, or tubers (APDiD: −4.4%; 95% CI: −7.0%, −1.7%; P = 0.001) and legumes, nuts, or seeds (APDiD: −15.6%; 95% CI: −26.2%, −5.0%; P = 0.004). There was no impact on reported consumption of dairy, flesh foods, eggs, or seafood.

The intervention was also associated with +18.6% (95% CI: 11.2%, 25.9%; P < 0.0001) prevalence of consuming MMF and +5.6% (95% CI: 0.02%, 11.2%; P = 0.05) prevalence of consuming the MAD. Although several core IYCF indicators improved, only 19.1% (95% CI: 16.3%, 21.9%) of children in the intervention district had an MAD at endline.

IYCF counseling and IYCF practices by intensity of program coverage

Intensity of program coverage was significantly associated with some of the program outcomes. Relative to low VHT contact/low MNP coverage, high VHT contact only, high MNP coverage only, and high VHT contact/high MNP coverage were associated with increased odds of having received IYCF counseling by a VHT [adjusted OR (aOR): 2.7; 95% CI: 1.6, 4.5; P = 0.0003 for high VHT contact only; aOR: 1.6; 95% CI: 1.1, 2.4; P = 0.02 for high MNP coverage only; and aOR: 9.0; 95% CI: 5.1, 15.8; P < 0.0001 for high VHT contact/high MNP coverage] (Table 4). High VHT contact/high MNP coverage was associated with 2 times higher odds of having received IYCF counseling by health center staff (95% CI: 1.2, 3.3; P = 0.007).

TABLE 4.

IYCF counseling coverage and IYCF practices by intensity of program coverage at endline (2016) in Amuria (intervention) district: IYCF/micronutrient powder intervention survey, Uganda¹

	Adjusted OR (95% CI)	P
Received IYCF counseling by VHT
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	2.7 (1.6, 4.5)	0.0003
High MNP coverage only	1.6 (1.1, 2.4)	0.02
High VHT contact + high MNP coverage	9.0 (5.1, 15.8)	<0.0001
Received IYCF counseling by health center staff
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.3 (0.8, 2.1)	0.3
High MNP coverage only	1.2 (0.9, 1.7)	0.2
High VHT contact + high MNP coverage	2.0 (1.2, 3.3)	0.007
Ever breastfed
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	0.7 (0.3, 1.4)	0.3
High MNP coverage only	2.6 (0.9, 7.4)	0.07
High VHT contact + high MNP coverage	1.7 (0.7, 4.1)	0.2
Current breastfeeding
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.1 (0.6, 1.9)	0.8
High MNP coverage only	1.9 (1.2, 3.0)	0.01
High VHT contact + high MNP coverage	2.4 (1.5, 3.9)	0.0005
Bottle feeding
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	0.5 (0.04, 5.1)	0.5
High MNP coverage only	1.0 (0.2, 4.4)	0.9
High VHT contact + high MNP coverage	1.1 (0.3, 4.2)	0.8
Continued breastfeeding at age 1 y2
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.3 (0.2, 9.0)	0.8
High MNP coverage only	1.9 (0.4, 8.0)	0.4
High VHT contact + high MNP coverage	6.5 (1.0, 47.7)	0.06
Continued breastfeeding at age 2 y3
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	0.7 (0.3, 1.3)	0.2
High MNP coverage only	0.9 (0.4, 2.4)	0.8
High VHT contact + high MNP coverage	1.4 (0.6, 3.7)	0.4
Introduced complementary foods at age 6 mo
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	0.6 (0.4, 1.0)	0.06
High MNP coverage only	1.4 (0.9, 2.3)	0.1
High VHT contact + high MNP coverage	0.8 (0.5, 1.3)	0.4
Consumed grains, roots, or tubers
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	0.9 (0.2, 3.3)	0.8
High MNP coverage only	0.6 (0.3, 1.3)	0.2
High VHT contact + high MNP coverage	1.0 (0.3, 3.6)	0.9
Consumed legumes, nuts, or seeds
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.1 (0.7, 1.7)	0.7
High MNP coverage only	1.0 (0.6, 1.6)	0.9
High VHT contact + high MNP coverage	0.8 (0.5, 1.3)	0.4
Consumed dairy
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.6 (0.7, 3.6)	0.2
High MNP coverage only	0.3 (0.1, 0.9)	0.04
High VHT contact + high MNP coverage	1.0 (0.4, 2.2)	0.9
Consumed flesh foods (organs, meats, or seafood)
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	0.9 (0.5, 1.5)	0.7
High MNP coverage only	1.2 (0.8, 1.8)	0.4
High VHT contact + high MNP coverage	1.2 (0.8, 1.7)	0.5
Consumed organs or meats
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.2 (0.7, 2.3)	0.4
High MNP coverage only	1.0 (0.6, 1.7)	0.9
High VHT contact + high MNP coverage	1.1 (0.6, 1.8)	0.8
Consumed seafood
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	0.6 (0.3, 1.1)	0.09
High MNP coverage only	1.1 (0.7, 1.7)	0.7
High VHT contact + high MNP coverage	1.0 (0.6, 1.6)	0.9
Consumed eggs
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	2.6 (1.0, 6.7)	0.04
High MNP coverage only	1.9 (0.9, 3.7)	0.07
High VHT contact + high MNP coverage	2.3 (1.1, 4.9)	0.02
Consumed vitamin A–rich fruits and vegetables
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.2 (0.7, 2.2)	0.4
High MNP coverage only	1.1 (0.6, 2.0)	0.8
High VHT contact + high MNP coverage	1.2 (0.6, 2.2)	0.6
Consumed minimum dietary diversity
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.2 (0.8, 1.8)	0.3
High MNP coverage only	1.1 (0.7, 1.7)	0.5
High VHT contact + high MNP coverage	1.2 (0.8, 1.8)	0.5
Consumed minimum meal frequency
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.2 (0.7, 2.2)	0.5
High MNP coverage only	1.5 (0.9, 2.4)	0.08
High VHT contact + high MNP coverage	1.9 (1.2, 3.0)	0.0008
Consumed minimum acceptable diet
Low VHT contact + low MNP coverage	Ref	—
High VHT contact only	1.7 (0.8, 3.4)	0.1
High MNP coverage only	1.6 (1.0, 2.6)	0.07
High VHT contact + high MNP coverage	2.0 (1.1, 3.8)	0.02

¹Unweighted sample sizes are 224 (continued breastfeeding at age 1 y), 253 (continued breastfeeding at age 2 y), and 760 (received IYCF counseling from VHT; received IYCF counseling from health center staff; ever breastfed; current breastfeeding; bottle feeding; introduced complementary foods at age 6 mo; consumed grains, roots, or tubers; consumed legumes, nuts, or seeds; consumed dairy; consumed organs, meats, or seafood; consumed organs or meats; consumed seafood; consumed eggs; consumed vitamin A–rich fruits or vegetables; minimum dietary diversity; minimum meal frequency; minimum acceptable diet). Model estimates are adjusted ORs and 95% CIs for intensity of program exposure (ref: low VHT contact and low MNP coverage), adjusting for child sex and age group (12–17 mo vs. 18–23 mo), household wealth tertile, severe household food insecurity, and caregiver schooling level (less than primary education vs. primary or higher). CIs account for weighting and complex sampling design. High contact with VHT defined as seeking or getting services ≥1 time/mo. High MNP sachet coverage was defined as having reported receiving at least 60 sachets (2 boxes) every 6 mo during the 12-mo pilot—the minimum recommended dose per the 2011 WHO guideline (WHO 2011). We calculated the expected number of sachets received based on the child's age at the time of data collection, assuming the child began receiving MNP as soon as he/she was eligible (i.e., at age 6 mo or the start of the intervention, whichever came later). Because the model for “consumed other fruits and vegetables” would not converge, it is not included in this table. IYCF, infant and young child feeding; MNP, micronutrient powder; Ref, reference; VHT, village health team.

²Among children 12–15 mo (16).

³Among children 20–23 mo (16).

High VHT contact/high MNP coverage was associated with 2.4 times higher odds of current breastfeeding relative to those with low VHT contact/low MNP coverage (95% CI: 1.5, 3.9; P = 0.0005). Both high VHT contact and high VHT contact/high MNP coverage were positively associated with odds of egg consumption (aOR: 2.6; 95% CI: 1.0, 6.7; P = 0.04; and aOR: 2.3; 95% CI: 1.1, 4.9; P = 0.02, respectively). High VHT contact/high MNP coverage was associated with 2 times higher odds of consuming the MMF (aOR:1.9; 95% CI: 1.2, 3.0; P = 0.0008) and MAD (aOR: 2.0; 95% CI: 1.1, 3.8; P = 0.02).

Discussion

Using an intent-to-treat DiD approach, we modeled the impact of an integrated IYCF/MNP pilot intervention on IYCF practices among caregivers of children aged 12–23 mo in eastern Uganda. The intervention was positively associated with having received IYCF counseling by a VHT, timely introduction of complementary feeding, having consumed organs or meats and vitamin A–rich fruits or vegetables, and MMF. The intervention was negatively associated with having consumed grains, roots, or tubers and legumes, nuts, or seeds. Among children in the intervention district at endline, the highest intensity of program coverage (high VHT contact/high MNP coverage) was associated with increased odds of having received IYCF counseling by a VHT, IYCF counseling by health center staff, current breastfeeding, egg consumption, MMF, and MAD relative to low intensity of program coverage (low VHT contact/low MNP sachet coverage). Our findings suggest that VHT effectively delivered IYCF messaging, MNP distribution did not appear to interfere with breastfeeding or complementary feeding practices, and that the integrated IYCF/MNP program had a positive impact on several key IYCF practices.

Based on endline survey data and quarterly monitoring reports from the implementing partner, the intervention had strong implementation and high adoption and adherence among participants compared with similar interventions in other settings (20). At endline, coverage of IYCF counseling by VHT was >5 times higher in Amuria than in Soroti (63.0% vs. 11.5%). We found no evidence of MNP stock-outs in program monitoring reports, and only 2 caregivers reported limited MNP supply (15). Nearly all children surveyed in Amuria (95.3%) had received MNP by endline, and more than half (53.4%) reported receiving the 2011 WHO-recommended dose during the pilot year (15). Reported recent MNP intake was also high with 65.3% of caregivers in Amuria reporting that their child had consumed MNP in the 2 wk preceding the endline survey. Nearly 70% of children in Amuria had high VHT contact, high MNP coverage, or high VHT contact/high MNP coverage. Because <1% of caregivers in the nonintervention district reported that their child ever received MNP, we found little evidence that the intervention crossed over into Soroti.

Our findings suggest that VHTs were able to effectively deliver community-based IYCF education and counseling to caregivers. The intervention was associated with >50% prevalence of having received IYCF counseling by a VHT. Prior to the intervention, MoH health centers most intensively counseled on early initiation of breastfeeding and exclusive breastfeeding. Community-based organizations, nongovernmental organizations, and other partners also supported IYCF activities. Thus, the intervention introduced a more comprehensive package of IYCF education to child nutrition activities, acknowledging this evaluation did not include the 0- to 11-mo age range to evaluate effectiveness of this package among younger children. The evolving MNP distribution scheme might also help explain our findings. At the start of the pilot, the intervention relied on existing MoH outreach sites. After the first project quarter, based on lower-than-desired program coverage, the implementing partner created new outreach sites, and VHTs delivered IYCF messaging and MNP to caregivers who missed scheduled distribution at outreach sites (20). Research has shown that community-based MNP programs tend to have higher coverage (21). A systematic review and meta-analysis of 83 studies found that community health worker home visits were associated with increased early initiation of breastfeeding (OR: 1.50; 95% CI: 1.12, 1.99) and exclusive breastfeeding (OR: 4.42; 95% CI: 2.28, 8.56) but not with MDD or MMF (22). However, it is unclear whether additional outreach sites and home delivery by VHTs would be feasible without support from an implementing partner or financially sustainable at scale without additional resources (20).

The relation between high-intensity program coverage (high VHT contact/high MNP coverage) and odds of having received IYCF counseling and select breastfeeding and complementary feeding indicators suggests that there may be synergy by integrating MNP as part of an IYCF intervention. Although IYCF counseling alone might be sufficient to improve child feeding practices in some contexts (23, 24), integrated IYCF/MNP programs might increase demand for nutrition education and counseling. For example, where MNP is viewed as an incentive for mothers to participate in nutrition counseling, there might be increased delivery of IYCF messaging (25). Community-based delivery of MNP could also increase hands-on learning opportunities for caregivers (26). Given that an estimated 74% of MNP programs are linked to IYCF programs (27), synergy from integrating infant and child feeding activities has the potential to strengthen child nutrition programs.

High levels of breastfeeding in this population suggest that the ability to improve breastfeeding indicators among children >6 mo may be limited; however, this evaluation supports the assumption that integrated IYCF/MNP programs are not likely to discourage breastfeeding. Except for a marginally significant improvement in continued breastfeeding at age 1 y, the intervention did not have an impact on any breastfeeding indicators. Almost all children at baseline had been breastfed at some point since their birth and >90% had continued breastfeeding at age 1 y, leaving little room for improvement. Similarly, bottle feeding was very infrequently practiced at baseline. The 2016 Uganda DHS reported similar levels of ever breastfeeding (98%) both nationally and in the Teso region where Amuria and Soroti are located (2). To our knowledge, only 1 study evaluating integrated IYCF/MNP programs and IYCF practices reported significant impacts on breastfeeding. An evaluation of an integrated IYCF/MNP pilot intervention for children aged 6–23 mo in Nepal reported that children who consumed MNP had higher odds of continued breastfeeding at age 2 y (7). Others reported no significant effects on breastfeeding indicators (8–10). Existing evidence suggests that these integrated IYCF/MNP programs do not undermine breastfeeding practices.

The intervention had a positive impact on the proportion of children consuming the MMF and receiving timely introduction of complementary foods and a marginally significant impact on MAD. The studies that have examined the impact of integrated IYCF/MNP interventions on IYCF practices have also reported a positive impact on complementary feeding practices. Among children aged 6–23 mo, there were reported increases in MMF (Nepal) (7), MDD (Nepal and Burkina Faso) (7, 10), consuming ≥4 food groups (Madagascar) (8), and MAD (Nepal and Burkina Faso) (7, 10), recognizing that the Nepal and Madagascar studies did not have control groups. MNP distribution at age 6 mo might encourage timely introduction of complementary feeding since MNPs are mixed into soft, semi-solid, or solid foods and not liquids like breast milk. The impacts of integrated IYCF/MNP interventions on timely introduction of complementary feeding are equivocal. Among children aged 6–23 mo in Nepal, having received MNP from a community health worker was associated with a 6% higher prevalence of starting complementary foods at age 6 mo (9). However, there was a nearly 18-percentage-point decrease in timely introduction of complementary foods among children aged 6–23 mo in Burkina Faso (10). Although MNPs are different from other supplements such as corn soy blend that are meant to improve both macro- and micronutrient intakes, several studies from southern and western Africa have assessed the impact of complementary food supplements on IYCF practices among children <2 y. These studies reported increases in the proportion of women waiting until 6 mo to introduce complementary foods, feeding frequency, MMF, MDD, and MAD (28–30).

The intervention was associated with a higher prevalence of reported consumption of both vitamin A–rich fruits or vegetables and organs or meats. In addition to promoting dietary diversity, behavior-change education focused specifically on these food groups as good sources of nutrients to help prevent vitamin A deficiency, iron deficiency, and anemia. Although reported consumption of vitamin A–rich fruits or vegetables increased substantially in both Amuria and Soroti between baseline and endline, increases in Amuria were significantly larger. At endline, reported consumption of vitamin A–rich fruits or vegetables in both districts was higher than national averages in the 2016 Uganda DHS (49.9% of breastfeeding children and 54.5% of non-breastfeeding children) (2).

The intervention was also associated with a lower prevalence of reported consumption of grains, roots or tubers and legumes, nuts, or seeds. Diets in Uganda are composed mainly of staple foods such as cassava, rice, and potatoes. Decreased consumption of staple foods is expected with concurrent increases in dietary diversity. For example, in populations undergoing nutrition transition, a decreasing proportion of the diet comes from staple foods as diets include more nonstaples (31). Although this population is not undergoing nutrition transition per se, Amuria saw small but significant displacement of staple foods by vitamin A–rich fruits or vegetables and organs or meats. Reductions in reported consumption of legumes, nuts, or seeds are concerning because this food group is a low-cost source of iron relative to flesh foods. Future research might explore how nutrition messaging could promote specific nutrient-dense food groups without displacing other healthful foods such as legumes.

Our findings suggest a continued need to improve complementary feeding practices in eastern Uganda. While several IYCF practices improved in our intervention, it is important to note that only 19.1% of children aged 12–23 mo in Amuria had an MAD at endline—similar to the Teso region in which the study districts are located (20.2%) and slightly higher than the national-level estimate (14.6%) in 2016 (2). Further, data were collected during the first harvest season, which may correspond with wider availability of and increased access to foods relative to the lean season. Thus, the reported prevalences of IYCF practices might be lower at other times of year. Strengthening elements of IYCF counseling, such as responsive feeding, might further improve complementary feeding behaviors (32). Further, improvements in reported consumption in our evaluation are somewhat surprising considering there was a 9-percentage-point increase in the prevalence of severe household food security across both districts between baseline and endline (data not shown). Although our models adjusted for household food security, it is not clear how caregivers were able to improve feeding practices despite decreases in food access and availability.

This analysis had several strengths and limitations. The impact evaluation had a quasi-experimental design, using cross-sectional, pre-post design including a nonintervention district. We adjusted for a range of potential confounders that might have affected the trajectory of change in the 2 districts, such as severe household food insecurity. Caregivers reported their IYCF practices, so responses might be subject to social desirability or recall bias. In the meat food group, we were unable to distinguish iron-rich meats like beef from iron-fair meats like poultry.

Using an intent-to-treat DiD approach, we found that an integrated IYCF/MNP pilot intervention had a positive impact on receipt of IYCF counseling by a VHT and on select IYCF practices among caregivers of children aged 12–23 mo in eastern Uganda. Further, among those in the intervention community, higher intensity of program coverage was positively associated with select IYCF practices, suggesting synergy between the MNP and nutrition education and counseling components of the intervention. Our findings suggest that VHT effectively delivered IYCF messaging and that MNP distribution did not appear to interfere with breastfeeding or complementary feeding practices. However, the low prevalence of some IYCF practices at endline suggests a continued need to improve complementary feeding practices.

Notes

The Government of Uganda, Ministry of Health (MoH) and World Food Program (WFP) Uganda Country Office supported the implementation of the pilot intervention. WFP Uganda funded an external agency (Makerere University) to conduct the evaluation surveys using contributions from a Canadian International Development Agency (CIDA) grant number 10017474 for WFP's Nutrition Strengthening Plan “Providing the Right Food at the Right Time at the Right Place.” The US Centers for Disease Control and Prevention provided technical assistance for this survey through a Memorandum of Agreement with the WFP. The MoH and WFP supported the program implementation indirectly via supervisory visits but were not directly involved in the census, participant sampling, gathering informed consent of participants, or data collection.

Author disclosures: The authors report no conflicts of interest.

The findings and conclusions in this manuscript are those of the authors and do not necessarily represent the official position of the CDC, Makerere University, World Food Program, or the Ministry of Health of Uganda.

Abbreviations used: aOR, adjusted OR; APDiD, adjusted prevalence difference-in-difference; DHS, Demographic and Health Survey; DiD, difference-in-difference; IYCF, infant and young child feeding; MAD, minimum acceptable diet; MDD, minimum dietary diversity; MMF, minimum meal frequency; MNP, micronutrient powder; MoH, Ministry of Health; VHT, village health team; WFP, UN World Food Program.