Received: 4 February 2017 Revised: 18 July 2017 Accepted: 30 August 2017

DOI: 10.1111/mcn.12527
bs_bs_banner

ORIGINAL ARTICLE

Nutrition, hygiene, and stimulation education to improve

growth, cognitive, language, and motor development among
infants in Uganda: A cluster‐randomized trial
Grace K. M. Muhoozi1,2 | Prudence Atukunda1 | Lien M. Diep3 | Robert Mwadime4 |

Archileo N. Kaaya5 | Anne B. Skaare6 | Tiril Willumsen6 | Ane C. Westerberg7 |

Per O. Iversen1,8

1
Department of Nutrition, Institute of Basic
Medical Sciences, University of Oslo, Oslo, Abstract
Norway Stunting is associated with impaired cognitive and motor function. The effect of an education
2
Department of Human Nutrition and Home intervention including nutrition, stimulation, sanitation, and hygiene on child growth and
Economics, Kyambogo University, Kampala, cognitive/language/motor development, delivered to impoverished mothers in Uganda, was
Uganda
3
assessed. In a community‐based, open cluster‐randomized trial, 511 mother/children dyads aged
Oslo Centre for Biostatistics and
6–8 months were enrolled to an intervention (n = 263) or control (n = 248) group. The primary
Epidemiology, Oslo University Hospital, Oslo,
Norway outcome was change in length‐for‐age z‐score at age 20–24 months. Secondary outcomes
4
SPRING/Uganda (USAID Project), Kampala, included anthropometry and scores on the 2 developmental scales: Bayley Scales of Infant and
Uganda Toddler Development‐III and the Ages and Stages Questionnaire. There was no evidence of a
5
School of Food Technology, Nutrition and difference in mean length‐for‐age z‐score at 20–24 months between the 2 study groups: 0.10,
Bioengineering, Makerere University,
95% CI [−0.17, 0.36], p = .49. The intervention group had higher mean composite development
Kampala, Uganda
6 scores than the controls on Bayley Scales of Infant and Toddler Development‐III, the mean
Institute of Clinical Dentistry, Faculty of
Dentistry, University of Oslo, Oslo, Norway difference being 15.6, 95% CI [10.9, 20.2], p = .0001; 9.9, 95% CI [6.4, 13.2], p = .0001; and
7
Institute of Health Sciences, Kristiania 14.6, 95% CI [10.9, 18.2], p = .0001, for cognitive, language, and motor composite scores, respec-
University College, Oslo, Norway tively. The mean difference in scores from the Ages and Stages Questionnaire were 7.0, 95% CI
8
Department of Haematology, Oslo University [2.9, 11.3], p = .001; 5.9, 95% CI [1.2, 10.3], p = .01; 4.2, 95% CI [1.7, 6.7], p = .001; 8.9, 95% CI
Hospital, Oslo, Norway
[5.3, 12.3], p = .0001; and 4.4, 95% CI [0.0, 8.8], p = .05, for communication, gross motor, fine
Correspondence
motor, problem solving, and personal–social development, respectively. The intervention educa-
Per O. Iversen, Department of Nutrition,
Institute of Basic Medical Sciences, University tion delivered to mothers promoted early development domains in cognitive, language, and motor
of Oslo, P.O. Box 1046 Blindern, 0317 Oslo, development but not linear growth of small children in impoverished rural communities in
Norway. Uganda. Our study showed that child development may be improved with a relatively low cost
Email: [email protected]
intervention strategy. This trial was registered at ClinicalTrials.gov as NCT02098031.

Funding information
KEY W ORDS
Universitetet i Oslo; Throne Holst Foundation;
Throne Holst Foundation and the University of cognitive development, growth, hygiene, infant, nutrition education, Uganda
Oslo

1 | I N T RO D U CT I O N exacerbate poor growth and lead to impaired development in children

(Black et al., 2017; Grantham‐McGregor, Fernald, Kagawa, & Walker,
Globally, more than 165 million children under 5 years of age are 2014). Unhygienic environment and practices in impoverished house-
affected by stunting, indicating chronic undernutrition (Black et al., holds expose children to enteric pathogens and diarrhoea. This com-
2013). The number of affected African children is increasing due to pop- bined with lack of a varied and sufficient supply of food leads to
ulation growth and an almost stagnant prevalence (Onis et al., 2013). growth faltering (Yousafzai, Rasheed, Rizvi, Armstrong, & Bhutta,
Risk factors for stunting include deficient energy and nutrient intake 2014). Stunting is related to long‐term negative effects on cognitive
and infections. Poverty, inadequate care, and lack of stimulation often ability, school completion, and adult productivity (Sudfeld et al., 2015;

Matern Child Nutr. 2017;e12527. wileyonlinelibrary.com/journal/mcn © 2017 John Wiley & Sons Ltd 1 of 11
https://doi.org/10.1111/mcn.12527
2 of 11 bs_bs_banner
MUHOOZI ET AL.

Victora et al., 2008). In the developing world, more than 200 million chil-
dren under 5 years are not developing to their full potential (Black et al., Key messages
2017; Grantham‐McGregor et al., 2014). Reduced school and work per-
formance may further negatively influence the individual's ability to cre- • Child development may be improved with relatively

ate a better life so as to positively impact on societal development. small cost intervention strategies in a low‐resource

Once stunting has occurred, the effects of growth deficits are setting independently of growth deficiencies.

more permanent, especially after 36 months (Victora et al., 2008). • Education about nutrition, hygiene, sanitation, and
The first 2 years of life are often denoted the “critical window” for stimulation to impoverished mothers, markedly
growth as well as for cognitive and social development (UBOS. Uganda improved cognitive, language, and motor development
Demographic and Health Survey, 2011). Therefore, support and guid- of children between 6 and 24 months of age in rural
ance given during that period to caregivers focusing on optimal feeding Uganda.
are key for good child health. • Stunting prevalence was not improved by education to
Communication aimed at behavioural change at individual, house- mothers about nutrition, hygiene, sanitation, and
hold, and community levels is common in prevention and management stimulation.
of undernutrition. Culturally accepted nutrition education interven-
tions, emphasizing timely, appropriate, and responsive feeding,
improve linear growth and may reduce the impact of poverty on when the included children were aged between 6 and 8 months. Chil-
growth (Ayhan, Suskan, & Yildirim, 1996; Vazir et al., 2013). dren were evaluated at the end of the intervention when they were
Undernutrition is common in rural Uganda (Kikafunda, Agaba, & 12–16 months of age and 8 months later when they were 20–
Bambona, 2014), and stunting affects about one third of children 24 months of age. The study was conducted in Kabale and Kisoro dis-
under 5 years, and it is even higher (45–50%) in the south‐western part tricts in south‐western Uganda because of the high prevalence of
of the country (Engebretsen, Tylleskar, Wamani, Karamagi, & stunting (UBOS. Uganda Demographic and Health Survey, 2011).
Tumwine, 2008). There is a lack of adequately designed studies, for Town centres were excluded to minimize differences in socio‐eco-
example, randomized controlled trials, evaluating effects of nutrition nomic status and feeding practices. The people living in the study area
education interventions in impoverished Ugandan communities, and are predominantly small‐scale farmers. Both districts are densely pop-
child development is rarely addressed. Two intervention studies mainly ulated and are made up of several subcounties, each consisting of 18–
targeted parental behaviour and child stimulation but with minor focus 25 villages.
on nutrition education and child development (Morris et al., 2012; The study was reviewed by the Makerere University School of
Singla, Kumbakumba, & Aboud, 2015). Public Health, Higher Degrees Research and Ethics Committee and
The purpose of the present study was to assess the effects of a approved by the Uganda National Council for Science and Technology
nutrition education intervention, delivered in group meetings to and the Norwegian Regional Committee for Medical and Health
impoverished mothers, on child growth and cognitive development in Research Ethics. The consent form was translated into the local lan-
rural south‐western Uganda. The primary outcome was difference in guage, and all participants gave written or thumb‐printed, informed
linear growth when the children were 20–24 months old. The main sec- consent. Additional methodological information is provided in
ondary outcomes were cognitive, language, motor, and personal–social Supporting Information.
development as well as problem solving abilities. We specifically
targeted mothers of 6–8 months children because (a) complementary
feeding is recommended to start at 6 months, (b) this age is most suscep-
2.2 | Intervention
tible to poor linear growth, and (c) availability of age‐appropriate scales The intervention was delivered by an education team of 4 trained per-
for assessments of cognitive function and other developmental sons (2 female and 2 male bachelor graduates in nutrition) to 26 groups
domains. The intervention consisted of educating mothers to increase of 4–12 mothers/group. Children who did not have a mother as care-
dietary diversity to improve nutrient intake as well as continued giver were recruited with a grandmother. A village health team (VHT)
breastfeeding. In addition, we focused on hygiene and sanitation as well leader or a mother leader was selected by consensus in the groups.
as child stimulation. Hygiene and sanitation were included because The intervention was delivered at three main sessions with each group
water quality, sanitation, and hygiene (i.e., the WASH initiative) may during a period of 6 months. These sessions followed strict guidelines
be important for reducing stunting (Cumming & Cairncross, 2016). agreed upon prior to implementation. The quality of intervention was
monitored on site by the first author who attended all the sessions.
The VHT leader/mother leader was responsible for two main activities:
(a) organize monthly meetings to practice preparation of dishes dem-
2 | METHODS
onstrated to the groups, feed the children, revisit, and discuss knowl-
edge about feeding and hygiene emphasized in the intervention; and
2.1 | Study overview
(b) conduct of monthly follow‐up meetings with the mothers in their
We performed a two‐armed (intervention and control) open, cluster‐ homes to assist and encourage them to adhere to the intervention.
randomized trial with recruitment between October 24, 2013, and The VHT leader/mother leader attended all the three main education
February 16, 2014. The intervention period lasted 6 months, starting sessions with the mothers.
MUHOOZI ET AL. bs_bs_banner
3 of 11

The intervention delivery strategy revolved about two behaviour group allocation. The second team of four persons (two nutrition grad-
change techniques: providing information and prompt practice (dem- uates and two child development graduates) blinded to group alloca-
onstrations). A nutrition education curriculum was based on the 10 tion, collected child development data at baseline (6–8 months) and
guiding principles of complementary feeding of breastfed children anthropometry and child development data when the children were
(PAHO/WHO, 2003). We formulated recipes and demonstrated their 12–16 and 20–24 months. All questionnaires and assessments were
cooking using locally available foods of good quality protein with administered in the local language. The second team also administered
emphasis on animal protein. The mothers were encouraged to have a a knowledge assessment questionnaire from a subsample of mothers
kitchen garden with vegetables and domestic animals (chicken/rab- in both study groups at the end of the study period (i.e., when the chil-
bits), which could provide cheap animal protein (see Supporting dren were 20‐ to 24‐months old).
Information). A baseline data questionnaire was administered to the mothers
With regard to hygiene, the education team emphasized hand through interview and included the following: (a) socio‐demographic
washing before feeding as well as use of clean utensils during food characteristics; (b) morbidity, that is, if their child had any illness in
preparation and feeding. Sharing of spoons with the child was discour- the previous 2 weeks and/or if the child was sick at the time of data
aged to prevent infections. The education team demonstrated the use collection; and (c) child feeding practices. The questions were adapted
of tooth brush, which was distributed to all household members. from validated tools (PAHO, 2004).
Moreover, the education team emphasized the need to take ill children The socio‐economic status of each household was obtained using
to hospital for medical attention and to increase the feeding frequency the “Simple Poverty Scorecard for Uganda” (Schreiner, 2011). This tool
during and after illness episodes. has 10 indicators, which are strongly associated with poverty and are
The team highlighted the importance of play in the cognitive, lan- sensitive to changes in poverty status. Each indicator is given a score,
guage, and motor development of a child. Together with the mothers, and these scores are added to give a total poverty score ranging from
specific play activities and toys that could be useful in developing each 0 (high likelihood of poverty) to 100 (least likelihood of poverty).
of the development domains were identified. The mothers were Sanitation was evaluated by observation and recording pres-
encouraged to engage in play activities and use play toys. The stimula- ence = 0 or absence = 1 of (a) stagnant water in the compound, (b)
tion intervention was based on social‐cognitive learning theory, where human faeces around the house, (c) animal droppings, and (d) litter in
mothers were taught the benefits of the stimulation practices the compound; condition or absence of (a) plate stand (drying rack),
(Bandura, 2001). The education team together with the mothers iden- (b) bath shelter, and (c) latrine were scored as good = 3, fair = 2, poor = 1,
tified and resolved any hindrances and barriers to the stimulation inter- and absent = 0. All scores were combined to make the composite var-
vention. For cognitive development, mothers were encouraged to use, iable (household sanitation) with scores ranging from 0 to 4 (poor), 5–8
name, and identify body parts to facilitate the children's understanding (fair), and 9–13 (good).
of their daily routine in relation to the body. Other activities included The Household Food Insecurity Access Scale, without any modifi-
hiding favourite items for children to find, demonstrating screwing cations, was used to determine household food insecurity (Coates,
and unscrewing bottles, and imaginary play such as pretending to feed Swindale, & Bilinsky, 2007; Webb et al., 2006) at the three time points.
a doll when given a cup and spoon. For language development, the Responses determined the Household Food Insecurity Access Preva-
team emphasized the use of imitation, role‐playing games, songs, and lence status indicator as a proxy of household food insecurity (Coates
music. Mothers were encouraged to talk to the children, call them by et al., 2007). Each household was then classified into either food
their names, and respond to them by word and gesturing. In addition, secure or mildly, moderately, and severely food insecure.
mothers were advised to mention household and personal items while The child dietary diversity score (CDDS) adapted from the House-
pointing at them, naming domestic animals and imitating words and hold Diversity Score tool was used to determine the quality of the diet
actions. For motor development, the children were encouraged to pick (Swindale & Bilinsky, 2006). The tool has a range of 0–8 food groups
up items with their fingers, holding a pencil to scribble, throwing a ball, (grains, roots, or tubers; vitamin A‐rich plant foods; other fruits or veg-
crawling or walking upstairs, and kicking a ball. Mothers were advised etables; meat, poultry, fish, and seafood; eggs; pulses/legumes/nuts;
to hold infants by hand and assist them to walk. milk and milk products; and foods cooked in oil/fat). If the child con-
Each of the three main sessions lasted 6–8 hours. Details of the sumed any of the foods the previous 24 hr that belonged to any of
intervention procedures are given in Supporting Information. these groups, it would be scored 1, and these were summed up to give
To ensure compliance of mothers to the intervention groups, a fol- the CDDS, which was then categorized as low (scores) 0–3 or high (≥4)
low‐up assessment form (Supporting Information) with grading from 1 (WHO/UNICEF, 2007). The CDDS has good sensitivity and specificity
to 10 points was scored based on specific observed activities, reports and correlates significantly with weight‐ and length‐for‐age z‐scores
from VHT leaders/mother leaders, and interaction. (Moursi et al., 2008; Steyn, Nel, Nantel, Kennedy, & Labadarios, 2006).
Nutritional status was evaluated using weight, length, head cir-
cumference (HC), and mid upper arm circumference (MUAC) following
2.3 | Data collection and analyses standard procedures and calibrations as recommended (WHO, 2006).
Two trained field‐teams independently collected data. The first team Weight (to the nearest 0.1 kg) was measured with a Seca‐scale model
(12 persons in 6 pairs, “advanced level” and bachelor degree graduates 881 (Hamburg, Germany), whereas recumbent length was measured
who were familiar with the study area) collected baseline characteris- (to the nearest 0.1 cm) with a length board (Seca, SO114530). HC
tics and anthropometry from both study groups and were blinded to was measured with a nonstretchable tape (Seca, S0145630 PAC‐50)
4 of 11 bs_bs_banner
MUHOOZI ET AL.

and MUAC was measured with a nonstretchable tape (Seca, S0145620 about 1.5 hr. Data quality was maintained by daily checks for com-
MUAC, Child 11.5 Red/PAC‐50) at the midpoint between the pleteness. The second author (clinical psychologist) supervised the
acromion and the olecranon. The date of birth was obtained from the administration of tests, the interview styles, and the data records,
child health card. For children without health card (8.3%), a record of and the first author (nutritionist) supervised all anthropometric assess-
events was used to determine the approximate date of birth. Anthro- ments. For consistency, we studied the guidelines and rules of using
pometric data were converted to z‐scores, length‐for‐age (LAZ), anthropometric tools and child development scales before each
weight‐for‐age (WAZ), weight‐for‐length (WLZ), mid upper arm cir- assessment time.
cumference (MUACZ), and head circumference (HCZ), using the In case of illness, the child would not be assessed and the house-
Anthro (version 3.2.2) software, a nutritional assessment tool based hold would be revisited. A knowledge assessment questionnaire was
on the WHO standards. Undernutrition was defined as a z‐score below administered to a subsample of mothers in both study groups evaluat-
minus two SD from the median of the WHO reference standards for ing knowledge on nutrition, child stimulation, and hygiene practices.
LAZ (stunting), WAZ (underweight), or WLZ (wasting; WHO, 2006).
Interobserver Pearson's correlation coefficients for reliability between
the assessors based on measurements of 20 randomly selected chil-
2.4 | Design and statistical analyses
dren were excellent, ranging between 0.91 and 0.98 for all anthropo- Proportionate sampling was used to obtain 10 subcounties (6 out of 19
metric measurements. in Kabale and 4 out of 14 in Kisoro) to participate in the study. We
Child development was determined by the Bayley Scales of Infant used a three‐stage procedure to obtain households for the study. First,
and Toddler Development‐III (BSID‐III) using the three subscales of by simple random sampling, three subcounties in Kabale were allo-
cognitive, language, and motor functions (Fernald et al., 2009). The cated to the intervention group and the other three to the control
Bayley scales provide comprehensive development measures for chil- group. Similarly, in Kisoro, two subcounties were allocated to the inter-
dren up to 42 months. The BSID‐III has been adapted for appropriate vention and the other two to the control group. Second, all the villages
use among children in rural Uganda (Singla et al., 2015) and in similar in each participating subcounty (intervention or control) were listed
settings (Hamadani, Huda, Khatun, & Grantham‐McGregor, 2006). alphabetically and assigned numbers in an ascending order. By use of
The BSID‐III scales were translated to the local language and back‐ computer‐generated random numbers, villages to whose assigned
translated to English. Unfamiliar items in the stimulus and picture number matched with the random numbers were chosen. The inter-
booklets were replaced with familiar objects in the Ugandan context; vention villages did not share common geographical boundaries with
for example, apples were replaced with tomatoes and a vacuum control villages to minimize contamination of the intervention contents
cleaner with a mop. Replacement items were chosen based on their between the two study groups. Third, by complete enumeration, all
size, colour, and shape to maintain functional equivalence with the consenting households with children aged 6–8 months within a partic-
original stimuli. The raw scores were converted to composite scores ipating village were recruited to the study. If a household had more
accordingly using BSID‐III conversion tables. Infant development was than one eligible child, the youngest was selected, and in the case of
also assessed using the Ages and Stages Questionnaire (ASQ; twins, we randomly selected one for evaluation. On the basis of the
Kerstjens et al., 2009). The scores obtained from each of the five 2002 population housing census and the fertility rate, we expected
developmental domains were calculated according to a scale from 0 an average of 6 mother–child dyads per 150 households in every vil-
(worst) to 60 (best). Both tools were used because we did not include lage to have a child between 6 and 8 months old. Households were
the social–emotional scale of BSID‐III. Thus, ASQ, a caregiver report, excluded if the child had (a) congenital malformation(s), (b) a physical
was used to assess the social–emotional abilities of the child and to disorder that would influence growth or preclude anthropometric mea-
evaluate a wide range of adaptive behaviours not obtained with the surements or influence nutrient intake, or (c) been diagnosed with a
BSID‐III. The administration of the tests was performed by personnel mental or brain illness as reported by the mother or a health worker.
fluent in English and the local language. Feedback on administration The study personnel collecting the data and analysing the study out-
and data records were reviewed at the end of each assessment day. comes were blinded to group allocation.
A 6‐day intensive training session on use of the BSID‐III and ASQ Sample size calculation was based on the primary outcome (LAZ at
was conducted by a clinical psychologist (second author). The scales 20–24 months). The mean ± SD for LAZ is 0.0 ± 1.0 in a healthy pop-
were pretested on 8 boys and 12 girls from Kabale who were not part ulation. We defined a difference of 0.3 SD LAZ at 20–24 months
of the study. Internal consistency of the scales was good as shown by between the intervention and control group as clinically relevant, cor-
Chronbach's α = 0.78 for BSID III and Chronbach's α = 0.83 for ASQ. responding to about half a percentile in LAZ (Ong, Ahmed, Emmett,
Interobservation agreement between the child assessment team was Preece, & Dunger, 2000). To detect a change of 0.3 SD in LAZ at
good indicated by an intraclass correlation coefficient of 0.75 20–24 months with a significance level of 5% and a power of 80%,
(p = .0001) for BSID‐III and 0.79 (p = .0001) for ASQ. 176 children were required per group. Fifty‐one children per sub-
Child assessments were performed in hired special rooms in the county (cluster) were included presuming 10 subcounties as clusters
villages so that there were no interruptions and interference during and an intracluster correlation of 0.01 (Campbell, Donner, & Klar,
assessment. We also had a mobile tent in cases where rooms were 2007). We included 511 mother–children pairs, and the assessment
not available. The mothers were given money for transportation to was by intention to treat.
the test sites. We started assessment using BSID‐III followed by the We used a four‐level model to compare the intervention with the
ASQ and finally growth assessment. The total duration per child was control group at baseline, 12–16 and at 20–24 months. The
MUHOOZI ET AL. bs_bs_banner
5 of 11

subcounty, village, child, and children within villages were the random On the basis of a questionnaire given at the end of the study,
intercepts, and the three time points and group (intervention or con- the mothers in the intervention group performed significantly better
trol) were the random slope and fixed variables in the model. Unstruc- in nutrition knowledge than the control group (results in Table S1).
tured or exchangeable variance–covariance structure was used for the A significantly higher proportion of mothers in the intervention group
random part at child level, and the models were fitted via the restricted were able to identify foods and their nutrients correctly than the
maximum likelihood method. We fitted data by the maximum likeli- control group; 84% versus 50% were cleaning the child's mouth,
hood method and used a log likelihood ratio test to determine the and 51% versus 6% described preparation of silverfish very well for
overall effect of the intervention for the entire study period. Values the child. These data suggest that the mothers of the intervention
are given as mean (SD or 95% CI) differences between the two groups. children had acquired the desired level of knowledge from the
We analysed the mean slope (the average change per 6 months) within intervention.
each group to assess the effect of the intervention over time. We per- The means ± SD of the CDDS in the intervention group were
formed the statistical analyses with Stata/SE version 14 and SPSS ver- 3.3 ± 1.7, 3.9 ± 1.2, and 4.1 ± 1.1 at 6–8 months (baseline), 12–
sion 22. All analyses were adjusted for cluster and individual levels of 16 months, and 20–24 months, respectively. The corresponding values
sampling. Significance was set at p < .05 for the primary outcome. for the control group were 2.8 ± 1.6, 3.5 ± 1.2, and 3.4 ± 1.1 at 6–
Due to multiple testing regarding the secondary outcomes, we used 8 months, 12–16 months, and 20–24 months, respectively. There
a Bonferroni correction of 0.05/12, and therefore, significance for was no significant difference in mean CDDS between the two study
the secondary outcomes was set at 0.004. groups at the three study time points; the estimated difference
between the slopes being 0.28, 95% CI [−0.24, 0.81], p = .29. There
was, however, an average increase per 6 months in CDDS of 0.86,
95% CI [0.46, 1.27], p = .0001, in the intervention group compared
3 | RESULTS
to 0.58, 95% CI [0.16, 1.00], p = .007, in the control group (i.e., differ-
ence in control group not significant after Bonferroni correction).
3.1 | Enrolment and baseline characteristics Household food insecurity affected the intervention and control
We enrolled 511 households with children between 6 and 8 months, groups equally. About 84.4%, 83.8%, and 76.4% of the households in
which were assigned to either the intervention (41 villages, n = 263) the intervention group suffered mild to severe household food insecu-
or the control (41 villages, n = 248) group (Figure 1). Anthropometric rity at baseline, at 12–16 months, and at 20–24 months, respectively.
data collection and child development testing were performed at base- The corresponding values in the control group were 85.9%, 89.3%,
line, end of the intervention period when the children were 12– and 80.0%.
16 months and when the children were 20–24 months. Eighteen chil-
dren in the intervention and 21 in the control group were lost to fol-
3.2 | Anthropometric measures during the study
low‐up at the end of the intervention period. Two children in the
intervention and three children in the control group died. We did not
period
register any loss to follow‐up during the subsequent follow‐up period. There was no evidence of a difference in linear growth, measured as
Thus, at the age of 20–24 months, 243 children in the intervention mean LAZ, at 20–24 months between the intervention and control
group and 224 children in the control group could be analysed. There group (Table 2): 0.10, 95% CI [−0.17, 0.36], p = .49. There was, how-
were no differences (p = .40) between those who completed the study ever, a decline in mean LAZ slopes during the study period in both
and those who did not. study groups, indicating linear growth faltering (Table S2). Stunting
There were no apparent differences between the two study was found in 142 children (49.3%) in the intervention and 146
groups in the socio‐demographic characteristics (Table 1). The preva- (50.7%) children in the control group (p > .05) at the end of the study
lence of stunting, underweight, and wasting at baseline, expressed as period.
z‐scores <2 SD from the median of the reference population, was sim- With regard to the other anthropometric indicators (WAZ, WLZ,
ilar in both study groups. About one third of all the children were sick MUACZ, and HCZ), there were no significant differences between the
at baseline, with equal numbers in both study groups. The most com- intervention and control group, neither at baseline nor at 12–16 months
mon illnesses were cough and common cold, diarrhoea, suspected or at 20–24 months (Table 2). We observed no significant change in the
malaria, and eye infections. Nine children in both study groups had mean slopes for WAZ and WLZ among the intervention and control chil-
not started complementary feeding, and most children (97%) were still dren (Table S2). The LAZ, MUACZ, and HCZ mean slopes decreased sig-
breastfeeding. In general, the household sanitation was fair although a nificantly in both study groups, and the mean slopes did not differ
substantial proportion lacked basic sanitary facilities such as latrines. between intervention and control groups for any of the anthropometric
Furthermore, the mean poverty scores for households in both study indicators (p > .05; Table S2). The effect sizes of the anthropometric
groups were about 50. measures at 20–24 months are shown in Table S3.
Most (84%) mothers of children in the intervention group
attended the three sessions during the intervention period. Using the
follow‐up assessment form, in the first follow‐up, 11 groups scored
3.3 | Child development during the study period
(range) 2–5, and the other 15 groups scored 6–9. In the second fol- There was a significant increase in mean cognitive and motor compos-
low‐up, five groups scored 4–5 and 21 groups scored 6–10. ite scores measured by BSID‐III from baseline to 20–24 months in the
6 of 11 bs_bs_banner
MUHOOZI ET AL.

512 assessed for eligibility

1 excluded
(presenting with a
congenital disorder)

511enrolled

511 randomized

263 allocated to intervention 248 allocated to control

18 lost to 21 lost to
follow-up follow-up
2 died 3 died

12-16 months (end of 12-16 months (end of

intervention) intervention)
243 analysed 224 analysed

20-24 months (end of follow- 20-24 months (end of

up) follow-up)
243 analysed 224 analysed

FIGURE 1 Flowchart of trial progress

intervention but not among the control children (Table 3). Although the control group during the study period. The effect sizes of from the
mean slope for cognitive development increased significantly over BSID‐III and ASQ at 20–24 months are shown in Table S3.
time in the intervention group, there was no significant mean slope
change among the controls. The mean slope for motor development
increased significantly over time in the intervention but decreased sig- 4 | DISCUSSION
nificantly in the control group (Table S4). In contrast, the mean com-
posite scores for language decreased significantly in both groups We found that a decline in linear growth rates occurred during the
during the study period (Table 3), as did the mean slopes. The decrease study period in both groups, and we could not detect a significant dif-
was more pronounced among the controls. All mean scores in the ference between the two groups at 20–24 months. Similarly, the WAZ,
three domains were significantly higher in the intervention group than WLZ, MUACZ, and HCZ remained similar between the two groups
the control group (Table 3). during the study period. In contrast to these anthropometric results,
We also assessed child development using the ASQ scales the intervention clearly improved child development domains. Notably,
(Table 4). Whereas there were no significant differences in mean com- at 20–24 months, the children in the intervention group scored mark-
munication scores between the two study groups at baseline and at edly better on all three subscales of the BSID‐III scale and on the five
12–16 months, the children in the intervention group scored signifi- developmental domains of the ASQ scale, compared with the control
cantly higher than the controls at 20–24 months. A similar pattern group. In the intervention group, the mean slopes for the development
was noted for gross and fine motor as well as the personal–social scores consistently indicated a more favourable change in the various
development domains. With regard to the problem solving domain, subscales and domains, as measured with both scales.
the intervention group performed significantly better at 12–16 months The age group of 6–8 months was identified for intervention
and at 20–24 months. The mean slopes for communication, fine motor, because complementary feeding is recommended to start at 6 months.
problem solving, and personal–social developments decreased signifi- This is also the age at which growth faltering is common, often due to
cantly more in the control compared with the intervention group over low nutrient density of the foods complementing breast milk
time (Table S4). The mean slope for gross motor development (Shrimpton et al., 2001). In addition, infections, such as diarrhoea
increased significantly more in the intervention compared with the resulting from poor food hygiene, may lead to increased nutrient
MUHOOZI ET AL. bs_bs_banner
7 of 11

TABLE 1 Study population characteristics at baselinea

Characteristic Intervention (n = 263) Control (n = 248)

Children
Males 139 (52.9) 123 (49.6)

Females 124 (47.1) 125 (50.4)

Age at inclusion (months; mean ± SD) 7.4 ± 0.8 7.3 ± 0.9

Stuntingb 55 (20.9) 70 (28.0)
b
Underweight 25 (9.5) 36 (14.5)
b
Wasting 12 (4.6) 12 (4.8)

Breastfeeding frequency
≥8 times/day 170 (64.6) 172 (69.4)
<8 times/day 93 (35.4) 76 (30.6)

Started complementary feeding

Yes 254 (96.6) 236 (95.2)
No 9 (3.4) 9 (4.8)

Illness at baseline
Yes 94 (35.7) 71 (28.6)
No 169 (64.3) 177 (71.4)

Maternal data (mean ± SD)

Maternal education (years) 4.9 ± 2.8 4.9 ± 2.8
Maternal age (years) 26.1 ± 5.8 26.8 ± 6.3
Number of children per mother 3.4 ± 2.2 3.3 ± 2.2

Household data (mean ± SD)

Household head age (years) 31.4 ± 7.9 33.4 ± 10.7
Household head education (years) 6.4 ± 3.1 5.9 ± 3.1
Household size 5.5 ± 2·1 5.5 ± 2.1
Household poverty score 47.8 ± 11.7 47.6 ± 11.4
Sanitation composite score 7.2 ± 1.9 7.3 ± 1.9
a
Values are n (%) unless otherwise specified.
b
z‐score values are <−2 SD of the median reference group.

losses, increased nutrient needs, and poor appetite (Vazir et al., 2013). shown) and this would lead to low micronutrients in their diets, possi-
It has also been shown that this age period is appropriate for interven- bly in part due to limited availability of affordable animal protein food
tions to reduce stunting (Victora, de Onis, Hallal, Blossner, & choices and poverty constraints. This, together with the observed low
Shrimpton, 2010). CDDS, might have limited improvements in growth of the children.
Our nutrition education intervention was specifically designed for Moreover, because the CDDS do not measure the amount or fre-
impoverished, rural women of low socio‐economic status, with few quency of food eaten, we do not know the actual differences in the
years of education, having heavy workloads, and who could not easily dietary intakes between the intervention and control groups.
access or even afford “already‐made” infant foods. We therefore opted The lack of significant differences in the anthropometric mea-
for an intervention strategy in which they could use the available and sures between the two groups could be attributed to the lack of
accessible food to improve complementary diets of their children. capacity to access the foods recommended, as our intervention pro-
Moreover, we strongly underscored the importance of hygiene in han- vided information and education and not actual foods. Although
dling the food and included oral/dental aspects. mothers in the intervention group seemed to have acquired the nec-
The CDDS was comparable and not significantly different essary knowledge about infant and young feeding (Table S1), the
between the two groups. Because both groups were equally affected observed household food insecurity limited their ability to provide
by household food insecurity, this could explain the lack of significant quality and quantity food to the children. In line with this, many
difference in the CDDS between groups. However, the significant mothers in the intervention group reported lack of money to buy
increase in CDDS within the intervention group shows that there foods rich in protein and micronutrients required for adequate child-
was increase in dietary diversity for some children. Still, we noted that hood growth. Maternal education, which was low (5 years) in both
the provision of sources of animal protein foods was still low (data not groups, could have affected the linear growth as observed in a study
8 of 11 bs_bs_banner
MUHOOZI ET AL.

TABLE 2 Child growth during the study perioda

Intervention (n = 240–263)b Control (n = 212–248)b Intergroup differencec p valued Overall p valuee

Age of child (months) Length‐for‐age z‐scores

Baseline (6–8) −1.07 ± 1.15 −1.20 ± 1.24 0.13 [−0.14, 0.39] .35 .09
12–16 −1.74 ± 0.97 −1.69 ± 1.10 −0.05 [−0.32, 0.21] .70
20–24 −2.15 ± 1.00 −2.25 ± 1.10 0.10 [−0.17, 0.36] .49
Weight‐for‐age z‐scores
Baseline (6–8) −0.63 ± 1.10 −0.72 ± 1.13 0.09 [−0.20, 0.36] .56 .60
12–16 −0.99 ± 1.06 −1.05 ± 1.10 0.06 [−0.23, 0.33] .72
20–24 −0.87 ± 0.87 −0.88 ± 0.87 0.01 [−0.27, 0.29] .96
Weight‐for‐length z‐scores
Baseline (6–8) 0.12 ± 1.22 0.16 ± 1.26 −0.04 [−0.32, 0.22] .74 .25
12–16 −0.24 ± 1.14 −0.34 ± 1.13 0.10 [−0.17, 0.37] .48
20–24 0.31 ± 0.86 0.36 ± 0.84 −0.05 [−0.33, 0.22] .70
Mid upper arm circumference z‐scores
Baseline (6–8) 0.29 ± 1.05 0.27 ± 0.98 0.02 [−0.21, 0.24] .91 .84
12–16 −0.21 ± 0.85 −0.25 ± 0.83 0.04 [−0.20, 0.26] .79
20–24 −0.16 ± 0.77 −0.15 ± 0.77 −0.01 [−0.25, 0.21] .87
Head circumference z‐scores
Baseline (6–8) 0.68 ± 1.10 0.57 ± 1.18 0.11 (−0.10–0.32) .32 .60
12–16 0.22 ± 0.91 0.05 ± 1.33 0.17 (−0.06–0.38) .14
20–24 0.39 ± 0.85 0.33 ± 0.91 0.06 (−0.16–0.27) .60
a
Values are means ± SD.
b
The variation in n is due to missing data.
c
Mean differences (95% CI), estimates were adjusted for clustering.
d
The p value is for the difference between the two study groups at each study time point.
e
Overall p value is for the overall effect of intervention obtained from the log likelihood ratio test.

TABLE 3 Mean composite scores derived from the Bayley Scales of Infant and Toddler Development‐III scalesa

Intervention (n = 243–245)b Control (n = 212–220)b Intergroup differencec p valued Overall p valuee

Age of child (months) Cognitive composite scores

Baseline (6–8) 102.1 ± 12.9 103.4 ± 13.8 −1.4 [−6.1, 3.3] .56 .0001
12–16 110.6 ± 13.1 103.1 ± 12.2 7.5 [2.9, 12.2] .002
20–24 114.9 ± 21.3 99.3 ± 17.1 15.6 [10.9, 20.2] .0001
Language composite scores
Baseline (6–8) 103.5 ± 14.4 100.2 ± 14.1 3.3 [−0.1, 6.7] .06 .0001
12–16 93.7 ± 8.3 93.3 ± 7.6 0.4 [−3.0, 3.8] .82
20–24 98.3 ± 14.3 88.4 ± 9.1 9.9 [6.4, 13.2] .0001
Motor composite scores
Baseline (6–8) 104.9 ± 13.8 104.4 ± 14.7 0.5 [−3.2, 4.1] .81 .0001
12–16 104.9 ± 12.4 98.2 ± 11.1 6.7 [3.3, 10.6] .0001
20–24 113.7 ± 18.9 99.1 ± 14.3 14.6 [10.9, 18.2] .0001
a
Values are means ± SD; mean differences (95% CI) of composite scores. Estimates were adjusted for clustering.
b
The variation in n is due to missing data as some children would not complete all tests.
c
Mean differences (95% CI), estimates were adjusted for clustering
d
The p value is for the difference between the two study groups at each study point.
e
Overall p value is for the overall effect of intervention obtained from the log likelihood ratio test.

of five African countries (Amugsi, Dimbuene, Kimani‐Murage, Mberu, likely compromise the safety and nutritional quality of the prepared
& Ezeh, 2017). Further, the heavy work schedule of the mothers may food.
have limited their time to prepare the food as instructed, and this Although sanitation was generally fair and about the same in both
together with poor sanitary conditions and lack of clean water would groups at baseline, our emphasis on the WASH initiative in the
MUHOOZI ET AL. bs_bs_banner
9 of 11

TABLE 4 Mean scores derived from the Ages and Stages Questionnaire (ASQ)a
Intervention Control (n = 243–245)b Intergroup differencec (n = 212–220)b p valued Overall p valuee

Age of child (months) Communication scores

Baseline (6–8) 47.3 ± 12.1 47.1 ± 11.9 0.2 [−4.0, 4.5] .90 .00001
12–16 35.1 ± 13.6 36.9 ± 12.9 −1.8 [−6.0, 2.5] .41
20–24 40.8 ± 14.5 33.8 ± 15.3 7.0 [2.9, 11.3] .001
Gross motor scores
Baseline (6–8) 47.5 ± 12.8 46.7 ± 13.0 0.8 [−3.7, 5.3] .73 .0009
12–16 44.7 ± −15.3 43.2 ± −15.1 1.5 [−3.1, 5.9] .54
20–24 52.8 ± 10.3 46.9 ± −13.8 5.9 [1.2, 10.3] .01
Fine motor scores
Baseline (6–8) 53.8 ± 11.7 54.6 ± 8.8 −0.8 [−3.2, 1.7] .56 .0005
12–16 45.5 ± 12.1 42.7 ± 12.1 2.8 [0.4, 5.3] .03
20–24 44.6 ± 9.9 40.4 ± 11.5 4.2 [1.7, 6.7] .001
Problem solving scores
Baseline (6–8) 52.4 ± 12.7 51.9 ± 11.5 0.5 [−3.0, 4.0] .78 .00001
12–16 46.0 ± 12.1 41.3 ± 13.4 4.7 [1.1, 8.1] .01
20–24 49.5 ± 11.7 40.6 ± 13.1 8.9 [5.3, 12.3] .0001
Personal–social development scores
Baseline (6–8) 50.6 ± 11.9 49.9 ± 13.1 0.7 [−3.8, 5.0] .79 .034
12–16 38.3 ± 15.1 34.5 ± 14.0 3.8 [−0.7, 8.2] .10
20–24 41.0 ± 11.3 36.6 ± 11.1 4.4 [0.0, 8.8] .05
a
Values are means ± SD; mean differences (95% CI) of ASQ scores. Estimates were adjusted for clustering.
b
The variation in n is due to missing data as some children would not complete all the tests.
c
Mean differences (95% CI), estimates were adjusted for clustering.
d
The p value is for the difference between the two study groups at each study point.
e
Overall p value is for the overall effect of intervention obtained from the log likelihood ratio test.

intervention lacked social support, especially in the provision of quality maternal stature, and health during and after pregnancy together with
and quantity of water. The WASH initiative embedded in the UN Sus- birth weight all impact on neonatal and infant nutrition (Abu‐Saad &
tainable Development Goal 6 is important if other goals and targets are Fraser, 2010).
to be achieved. It has been identified as a unit in the health pillar of Importantly, our intervention led to significant improvements in
nurturing care to promote childhood development (Black et al., cognitive, language, and motor development of the children when
2017). Undernutrition has been linked to poor WASH via enteric path- tested by BSID‐III and ASQ. The intervention group scored better in
ogens and infections in low‐ and middle‐income countries (Cumming & development domains compared to the controls. Our findings suggest
Cairncross, 2016; Nabwera, Fulford, Moore, & Prentice, 2017). that the benefits of the intervention on development may not be
The present findings on growth are in line with other studies try- linked to linear growth. A systematic review of nutrition and stimula-
ing to improve dietary intake and growth through education but did tion interventions (Grantham‐McGregor et al., 2014) reported that
not find significant differences in linear growth or prevalence of although stimulation interventions consistently benefited child devel-
stunting among children (Bhandari et al., 2001; Santos et al., 2001). opment, nutrition intervention usually benefited growth and some-
Whereas one Indian study found no impact on linear growth using times development. This supports the findings of the current study
home educational counselling (Bhandari et al., 2001), another study on the part of stimulation. From our study, we cannot be sure if
from the same researchers (Bhandari et al., 2004) found a small but nutrition education had any impact on development. It has been
significant increase in length. A Peruvian study reported that nutrition shown that undernutrition in early life may negatively affect
education had a significant effect on linear growth at 18 months of neurodevelopmental processes and therefore, structural and func-
age (Penny et al., 2005). The latter two studies, however, had longer tional development of the brain, children's experiences, and behaviour
period of intervention, more frequent contact with the study groups, (Prado & Dewey, 2014). However, experience‐dependent processes
and were also performed and supported by the existing health and brought about by enhanced responsiveness and child stimulation in
nutrition services. In addition, they were integrated with child and the intervention group could have led to new neuronal growth and
local government programs. The Ugandan study (Singla et al., 2015) hence buffer the potential negative impacts of undernutrition (Prado
emphasized more sessions of parenting with a diet component, but & Dewey, 2014). The development changes were observed to persist
that did not improve linear growth. In the present study, there were beyond intervention period partly because we were teaching forward
three main education and two follow‐up sessions. Growth is moder- and the skills would persist and improve as the children grew older.
ated by interplay of many factors such as morbidity, food insecurity, We also noted that there could be a combined effect of the various
10 of 11 bs_bs_banner
MUHOOZI ET AL.

intervention activities on development domains, as observed in other CONFLICTS OF INTERES T

studies (Yousafzai et al., 2014). Cognitive development had a larger The authors declare that they have no conflicts of interest.
effect size compared to motor and language development. Possibly,
child play matched with the caregiver's aspirations for future school
CONT RIB UT IONS
achievement for the child may in part explain this preferential cogni-
tive improvement (Vazir et al., 2013). Child stimulation involves rather GKMM, PA, ACW, and POI designed the study; GKMM and PA con-
ducted the study and collected the data; all authors analysed the data;
simple doable actions that are easily integrated into daily care roles by
GKMM and POI wrote the paper; all authors read and approved the
everyone in the home, including men. The decline in language domain
(receptive and expressive combined) in both groups, although the final manuscript.

intervention group later improved at 20–24 months, was mainly

affected by children performing poorer on the expressive subscale. ORCID
This could be attributed to lack of play materials that facilitate speech Per O. Iversen http://orcid.org/0000-0003-2977-9153
and language development. The simplest language stimulation activi-
ties for children included talking, singing, storytelling, and later on RE FE RE NC ES
reading to the children. With the heavy workloads, the mothers prob- Abu‐Saad, K., & Fraser, D. (2010). Maternal nutrition and birth outcomes.
ably could get adequate time to engage the children in language stim- Epidemiologic Reviews, 32, 5–25.

ulation. On the other hand, there are a variety of simple play materials Amugsi, D. A., Dimbuene, Z. T., Kimani‐Murage, E. W., Mberu, B., & Ezeh,
A. C. (2017). Differential effects of dietary diversity and maternal char-
for cognitive and motor development that mothers could avail to chil-
acteristics on linear growth of children aged 6–59 months in sub‐
dren to play with while they were busy. The BSID III results are in line Saharan Africa: A multi‐country analysis. Public Health Nutrition, 20,
with the results of the studies performed in similar settings (Cromwell 1029–1045.
et al., 2014; Hamadani et al., 2006; Rademeyer, Jacklin, & Maxeke, Ayhan, H., Suskan, E., & Yildirim, S. (1996). The effect of nursing or rampant
2013; Singla et al., 2015). caries on height, body weight and head circumference. Journal of Clini-
cal Pediatric Dentistry, 20, 209–212.
This study has some methodological weaknesses. Our study may
Bandura, A. (2001). Social cognitive theory: An agentic perspective. Annual
have been underpowered due to the small number of clusters.
Review of Psychology, 52, 1–26.
Although we examined the CDDS, we have no estimate of their actual
Bhandari, N., Bahl, R., Nayyar, B., Khokhar, P., Rohde, J. E., & Bhan, M. K.
dietary intakes. We also lack data on birthweight and maternal weight
(2001). Food supplementation with encouragement to feed it to infants
gain during pregnancy, which could have interfered with our findings from 4 to 12 months of age has a small impact on weight gain. Journal of
because they are linked to long‐term nutritional manifestations. Due Nutrition, 131, 1946–1951.
to the rural setting with challenging logistics, we could not collect Bhandari, N., Mazumder, S., Bahl, R., Martines, J., Black, R. E., & Bhan, M.
K. (2004). An educational intervention to promote appropriate com-
nutrient biomarkers or measures of body composition. Similarly, we
plementary feeding practices and physical growth in infants and
could not apply objective, neuropsychological tests. Moreover, we young children in rural Haryana, India. Journal of Nutrition, 134,
were unable to examine other factors that could have diluted the 2342–2348.
effect of the intervention, such as other sources of information (e.g., Black, M. M., Walker, S. P., Fernald, L. C., Andersen, C. T., DiGirolamo, A. M.,
traditional knowledge and practices) as well as seasonality and its influ- Lu, C., … Bhutta, Z. A. (2017). Early childhood development coming of
age: Science through the life course. Lancet, 389, 77–90.
ence on food diversity. The intervention could have been improved
with a higher frequency of contact with the mothers groups. The main Black, R. E., Victora, C. G., Walker, S. P., Bhutta, Z. A., Christian, P., de
Onis, M., … Uauy, R. (2013). Maternal and child undernutrition and
strengths of the study were the long follow‐up time and low rates of overweight in low‐income and middle‐income countries. Lancet, 382,
loss to follow‐up, the inclusion of the WASH in nutrition messaging, 427–451.
the use of two independent developmental tools, and the empower- Campbell, M. J., Donner, A., & Klar, N. (2007). Developments in cluster
ment and collaboration between mothers to increase focus on child randomized trials and Statistics in Medicine. Statistics in Medicine,
26, 2–19.
health and development.
Coates, J., Swindale, A., & Bilinsky, P. (2007). Household food insecurity
access scale (HFIAS) for measurement of household food access: Indicator
guide (v. 3). Washington, D.C.: Food and Nutrition Technical Assistance
5 | C O N CL U S I O N Project.
Cromwell, E. A., Dube, Q., Cole, S. R., Chirambo, C., Dow, A. E., Heyderman,
Our educational intervention did not improve growth in children, how- R. S., & Van Rie, A. (2014). Validity of US norms for the Bayley Scales of
ever, cognitive and motor development in the intervention group Infant Development‐III in Malawian children. European Journal of Paedi-
atric Neurology, 18, 223–230.
improved significantly. More studies are needed to identify appropri-
ate community‐based interventions to reduce the level of stunting. Cumming, O., & Cairncross, S. (2016). Can water, sanitation and hygiene
help eliminate stunting? Current evidence and policy implications.
Child development interventions are worth investing in to benefit child
Maternal & Child Nutrition, 12(Suppl 1), 91–105.
development in poor rural settings of Uganda.
Engebretsen, I. M., Tylleskar, T., Wamani, H., Karamagi, C., & Tumwine, J. K.
(2008). Determinants of infant growth in Eastern Uganda: A commu-
ACKNOWLEDGMENTS nity‐based cross‐sectional study. BMC Public Health 22, 8‐418.
Grantham‐McGregor, S., Cheung, Y. B., Cueto, S., Glewwe, P., Richter,
We thank the participating communities for allowing us to conduct this L., & Strupp, B. (2007). Developmental potential in the first 5 years
study. for children in developing countries. Lancet, 369, 60–70.
MUHOOZI ET AL. bs_bs_banner
11 of 11

Fernald, L. C. H., Kariger, P., Engle, P., & Raikes, A. (2009, 2009). In BANK Shrimpton, R., Victora, C. G., de Onis, M., Lima, R. C., Blossner, M., &
TW (Ed.), Examining early child development in low‐income countries: A Clugston, G. (2001). Worldwide timing of growth faltering: Implications
toolkit for the assessment of children in the first five years of life. Washing- for nutritional interventions. Pediatrics, 107, E75.
ton, D.C: The World Bank. p. 68–9 Singla, D. R., Kumbakumba, E., & Aboud, F. E. (2015). Effects of a parenting
Grantham‐McGregor, S. M., Fernald, L. C., Kagawa, R. M., & Walker, S. intervention to address maternal psychological wellbeing and child
(2014). Effects of integrated child development and nutrition interven- development and growth in rural Uganda: A community‐based, cluster
tions on child development and nutritional status. Annals of the New randomised trial. The Lancet Global Health, 3, e458–e469.
York Academy of Sciences, 1308, 11–32. Steyn, N. P., Nel, J. H., Nantel, G., Kennedy, G., & Labadarios, D. (2006).
Hamadani, J. D., Huda, S. N., Khatun, F., & Grantham‐McGregor, S. M. Food variety and dietary diversity scores in children: Are they good
(2006). Psychosocial stimulation improves the development of under- indicators of dietary adequacy? Public Health Nutrition, 9, 644–650.
nourished children in rural Bangladesh. Journal of Nutrition, 136, Sudfeld, C. R., McCoy, D. C., Danaei, G., Fink, G., Ezzati, M., Andrews, K. G.,
2645–2652. & Fawzi, W. W. (2015). Linear growth and child development in low‐
Kerstjens, J., Bos, A., ten Vergert, E., de Meer, G., Butcher, P., & Reijneveld, and middle income countries: A meta‐analysis. Pediatrics, 135, e1266–
S. (2009). Support for the global feasibility of the Ages and Stages e1275.
Questionnaire as developmental screener. Early Human Development, Swindale, A., & Bilinsky, P. (2006). Household dietary diversity score (HDDS)
85, 433–437. for measurement of household food access: Indicator guide (v.2). Washing-
Kikafunda, J. K., Agaba, E., & Bambona, A. (2014). Malnutrition amidst ton, D.C: Food and Nutrition Technical Assistance Project, Academy for
plenty: An assessment of factors responsible for persistent high levels Educational Development.
of childhood stunting in food secure Western Uganda. AJFAND, 14, UBOS. Uganda Demographic and Health Survey. (2011). Kampala, Uganda:
9288–9313. UBOS and Calverton( pp. 2012). Maryland: ICF International Inc.
Morris, J., Jones, L., Berrino, A., Jordans, M. J., Okema, L., & Crow, C. (2012). Vazir, S., Engle, P., Balakrishna, N., Griffiths, P. L., Johnson, S. L., Creed‐
Does combining infant stimulation with emergency feeding improve Kanashiro, H., … Bentley, M. E. (2013). Cluster‐randomized trial on
psychosocial outcomes for displaced mothers and babies? A controlled complementary and responsive feeding education to caregivers found
evaluation from northern Uganda. American Journal of Orthopsychology, improved dietary intake, growth and development among rural Indian
82, 349–357. toddlers. Maternal & Child Nutrition, 9, 99–117.
Moursi, M. M., Arimond, M., Dewey, K. G., Treche, S., Ruel, M. T., & Victora, C. G., Adair, L., Fall, C., Hallal, P. C., Martorell, R., Richter, L., &
Delpeuch, F. (2008). Dietary diversity is a good predictor of the micro- Sachdev, H. S. (2008). Maternal and child undernutrition: Conse-
nutrient density of the diet of 6‐ to 23‐month‐old children in quences for adult health and human capital. Lancet, 371, 340–357.
Madagascar. Journal of Nutrition, 138, 2448–2453.
Victora, C. G., de Onis, M., Hallal, P. C., Blossner, M., & Shrimpton, R. (2010).
Nabwera, H. M., Fulford, A. J., Moore, S. E., & Prentice, A. M. (2017). Worldwide timing of growth faltering: Revisiting implications for inter-
Growth faltering in rural Gambian children after four decades of inter- ventions. Pediatrics, 125, e473–e480.
ventions: A retrospective cohort study. The Lancet Global Health, 5,
Webb, P., Coates, J., Frongillo, E. A., Rogers, B. L., Swindale, A., & Bilinsky, P.
e208–e216.
(2006). Measuring household food insecurity: Why it's so important and
Ong, K. K., Ahmed, M. L., Emmett, P. M., Preece, M. A., & Dunger, D. B. yet so difficult to do. Journal of Nutrition, 5, 1404S–1408S.
(2000). Association between postnatal catch‐up growth and obesity in
WHO. World Health Organization. (2006). Child Growth Standards:
childhood: Prospective cohort study. BMJ, 320, 967–971.
Length/height‐for‐age, weight‐for‐age, weight‐for‐length, weight‐for‐
Onis, M., Dewey, K. G., Borghi, E., Onyango, A. W., Blossner, M., Daelmans, height and body mass index for‐age: Methods and development
B., … Branca, F. (2013). The World Health Organization's global target 2006. http://www.who.int/childgrowth/standards/Technical_report.
for reducing childhood stunting by 2025: Rationale and proposed pdf (accessed June 7, 2015).
actions. Maternal & Child Nutition, 9(Suppl 2), 6–26.
WHO/UNICEF. (2007). Indicators for assessing infant and young child feeding
PAHO. (2004). In Cataloguing‐in‐Publication PHL (Ed.), ProPAN: Process for practices Part 1 definitions. Washington DC: World Health Organization.
the promotion of child feeding. Washington, D.C: PAHO.
Yousafzai, A. K., Rasheed, M. A., Rizvi, A., Armstrong, R., & Bhutta, Z. A.
PAHO/WHO. (2003). Guiding principles for complementary feeding of the (2014). Effect of integrated responsive stimulation and nutrition inter-
breastfed child. Washington D.C: Division of Health Promotion and ventions in the Lady Health Worker programme in Pakistan on child
Protection. development, growth, and health outcomes: A cluster‐randomised fac-
Penny, M. E., Creed‐Kanashiro, H. M., Robert, R. C., Narro, M. R., Caulfield, torial effectiveness trial. Lancet, 384, 1282–1293.
L. E., & Black, R. E. (2005). Effectiveness of an educational intervention
delivered through the health services to improve nutrition in young
children: A cluster‐randomised controlled trial. Lancet, 365, 1863–
SUPPOR TI NG INF ORMATI ON
1872.
Prado, E. L., & Dewey, K. G. (2014). Nutrition and brain development in Additional Supporting Information may be found online in the
early life. Nutrition Reviews, 72, 267–284. supporting information tab for this article.
Rademeyer, V. K. M., Jacklin, L., & Maxeke, C. (2013). A study to evaluate
the performance of black South African urban infants on the Bayley
Scales of Infant Development III. South African Journal of Child Health,
7, 54–59. How to cite this article: Muhoozi GKM, Atukunda P, Diep LM,
Santos, I., Victora, C. G., Martines, J., Goncalves, H., Gigante, D. P., Valle, N. et al. Nutrition, hygiene, and stimulation education to improve
J., & Pelto, G. (2001). Nutrition counseling increases weight gain among
growth, cognitive, language, and motor development among
Brazilian children. Journal of Nutrition, 131, 2866–2873.
infants in Uganda: A cluster‐randomized trial. Matern Child
Schreiner, M. (2011). A simple Poverty Scorecard for Uganda. http://www.
microfinance.com/English/Papers/Scoring_Poverty_Uganda 2009_EN. Nutr. 2017;e12527. https://doi.org/10.1111/mcn.12527
pdf (accessed July 31, 2013).