Articles

Feasibility and impact of school-based nutrition education

interventions on the diets of adolescent girls in Ethiopia:
a non-masked, cluster-randomised, controlled trial
Sunny S Kim, Celeste Sununtnasuk, Hanna Y Berhane, Tamirat Tafesse Walissa, Abdulaziz Ali Oumer, Yonas Taffesse Asrat, Tina Sanghvi,
Edward A Frongillo, Purnima Menon

Summary
Lancet Child Adolesc Health 2023; Background Adolescence is a critical period of physical and psychological development, especially for girls, because
7: 686–96 poor nutrition can affect their wellbeing as well as that of their children. We aimed to assess the feasibility and impact
Published Online of a package of nutrition education interventions delivered through public primary schools on the diets of adolescent
September 1, 2023
girls in Ethiopia.
https://doi.org/10.1016/
S2352-4642(23)00168-2
Nutrition, Diets, and Health Methods In this non-masked, cluster-randomised, controlled trial, primary schools (clusters) in the Southern
Unit, International Food Policy Nations, Nationalities, and People’s Region and Somali region of Ethiopia were randomly allocated to the
Research Institute (IFPRI), intervention group (nutrition information provided during flag ceremonies, classroom lessons, school club
Washington, DC, USA
meetings, peer group mentoring, BMI measurement and counselling, and parent–teacher meetings) or the control
(S S Kim PhD,
C Sununtnasuk MSc); Nutrition group (standard academic curriculum on health and nutrition) by use of computer-generated pseudo-random
and Behavioral Sciences numbers. Duration of the school-based interventions was 4 months, and the key messages were related to dietary
Department, Addis Continental diversity (eating a variety of foods), energy adequacy (eating breakfast and healthy snacks), and healthy food
Institute of Public Health,
choices (avoiding junk foods). Adolescent girls were eligible for participation if aged 10–14 years and enrolled in
Addis Ababa, Ethiopia
(H Y Berhane PhD); FHI grades 4–8 in a study school. Data were collected with two independent cross-sectional surveys: baseline before the
Solutions, Addis Ababa, start of implementation and endline 1·5 years later. The primary outcome of impact was dietary diversity score,
Ethiopia (T T Walissa MPH, defined as the number of food groups (out of ten) consumed over the previous 24 h using a list-based method, and
A A Oumer PhD, Y T Asrat MSc);
minimum dietary diversity, defined as the proportion of girls who consumed foods from at least five of the ten
FHI Solutions, Washington, DC,
USA (T Sanghvi PhD); food groups, in the intention-to-treat population. We also assessed intervention exposure as a measure of feasibility.
Department of Health We estimated intervention effects using linear regression models for mean differences at endline, with SEs
Promotion, Education, and clustered at the school level, and controlled for adolescent age, region, household food security, and wealth. The
Behavior, Arnold School of
Public Health, University of
trial is registered with ClinicalTrials.Gov, NCT04121559, and is complete.
South Carolina, Columbia, SC,
USA (Prof E A Frongillo PhD); Findings 27 primary schools were randomly allocated to the intervention group and 27 to the control group. Between
Food and Nutrition Policy March 22 and April 29, 2021, 536 adolescent girls participated in the endline survey (270 in the intervention group
Department, IFPRI, New Delhi,
India (P Menon PhD)
and 266 in the control group), with median age of 13·3 years (IQR 12·1–14·0). At endline, the dietary diversity score
was 5·37 (SD 1·66) food groups in the intervention group and 3·98 (1·43) food groups in the control group (adjusted
Correspondence to:
Dr Sunny S Kim, Nutrition, Diets, mean difference 1·33, 95% CI 0·90–1·75, p<0·0001). Increased minimum dietary diversity was also associated with
and Health Unit, IFPRI, the intervention (182 [67%] of 270 in the intervention group vs 76 [29%] of 266 in the control group; adjusted odds
Washington, DC 20005, USA ratio 5·37 [95% CI 3·04–9·50], p<0·0001). 256 (95%) of 270 adolescent girls in the intervention group were exposed
[email protected]
to at least one of the five in-school intervention components.

Interpretation Integrating nutrition interventions into primary schools in Ethiopia was feasible and increased dietary
diversity incrementally among adolescent girls, but could be limited in changing other food choice behaviours, such
as junk food consumption, based on nutrition education alone.

Funding Bill & Melinda Gates Foundation.

Copyright © 2023 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0
license.

Introduction health in adulthood. Nutrition during adolescence is

Adolescence is a crucial period of physical and especially important for girls because poor nutrition can
psychological development and for achieving human affect their wellbeing as well as the survival, health, and
potential. Rapid physical, psychosocial, and cognitive wellbeing of their children.1 Data and evidence on the
growth and development are coupled with increased nutritional status of adolescents in low-income and
energy and nutrient requirements.1,2 Poor nutrition during middle-income countries (LMICs) are scarce, even though
adolescence can have adverse consequences that affect this population includes about 1·8 billion individuals.

686 www.thelancet.com/child-adolescent Vol 7 October 2023

 Articles

Research in context
Evidence before this study Added value of this study
Adolescent nutrition has been largely overlooked in This study evaluates a package of school-based nutrition
intervention and policy research. We searched PubMed from education interventions coupled with community and health
database inception to Aug 31, 2022, using the search terms (in platform-based interventions and capacity building of school
various combinations) “adolescent”, “nutrition”, “intervention”, staff and service providers to improve the diet of adolescent girls
“diet”, “school”, “school-based”, and “education”, for in Ethiopia. Our sample includes adolescent girls aged
intervention studies with adolescent diet or nutrition 10–14 years, an age group for which data and research on
outcomes. We had filters for English language and human nutrition challenges and determinants are scarce, but that
studies, and cross-referenced intervention studies cited in the represents a key phase of development and habit formation. Our
2022 Lancet Series on adolescent nutrition. We found that few study adds to the limited literature on the effects of school-based
randomised controlled trials and controlled before-and-after nutrition interventions on adolescent nutrition in LMICs. The
studies of school-based nutrition interventions have been results showed that developing a package of adolescent nutrition
conducted in low-income and middle-income countries interventions delivered primarily through school-based platforms
(LMICs). The existing evidence has largely been limited to was feasible and improved adolescent girls’ dietary diversity and
micronutrient supplementation and fortification, with few meal frequency, even with a short implementation duration
studies assessing the impact of nutrition education or (about 4 months). Junk food consumption, however, remained
counselling and macronutrient supplementation on health and largely unchanged. Despite an intervention effect on the
nutrition outcomes among adolescents. Nutrition consumption of other sweets, more than half of adolescent girls
interventions within the education sector—arguably the best reported consuming junk foods in the past 24 h.
researched and most used platform—have focused on the
Implications of all the available evidence
provision of school meals, but evaluations and reviews suggest
The findings of our study, together with those of previous studies,
that school meals without nutritious foods and nutrition
reflect the importance of nutrition education and the enabling
education based on behavioural science principles have
environment for influencing nutrition behaviours among
minimal effect on dietary change and nutrition outcomes. In
adolescents. Reinforcing messages about eating diverse foods
sub-Saharan Africa, evaluations of school-based nutrition
and eating more often resulted in an incremental behaviour
interventions indicate that, although nutrition education can
change related to dietary diversity and meal frequency; however,
improve nutrition knowledge, it does not necessarily translate
advising adolescents not to eat junk foods without changing
into healthy nutrition behaviour. Important gaps in evidence in
their food environments appeared to be largely unsuccessful in
the adolescent nutrition literature include data and research on
curbing consumption of junk food. These findings suggest that
nutrition challenges and determinants disaggregated by age,
nutrition knowledge is likely to have little effect without an
and effects of interventions that are multifaceted and address
enabling environment.
multiple challenges.

Globally, the health and nutrition status of adolescents Intervention studies on adolescent nutrition in Ethiopia
have improved only modestly over the past 50 years.3 are scarce and have largely focused on supplementation
In Ethiopia, where 39% of the population is younger of single micronutrients, such as iron or folic acid;14,15
than 15 years,4 adolescent nutrition is an important however, global evidence shows that multisectoral and
public health concern. The pooled prevalence of multifaceted strategies offer the most promise.16
adolescent stunting is estimated to be 21% and that of Programmatically, schools provide an important
underweight (ie, low BMI for age) is estimated to be environment to support adolescent health and nutrition
28%, with increased risk among adolescents in rural because adolescents spend a large proportion of their
and food insecure households.5 In southern Ethiopia, time at school. School feeding programmes are globally
the prevalence of stunting among adolescent girls is the most common and long-standing interventions but
more than 27%,6,7 and a study conducted in the Damot have shown minimal effects on dietary change or
Gale district found the prevalence of inadequate nutrition outcomes except when coupled with nutrient-
intake of nutrients—a primary determinant of dense foods and nutrition education.15 Holistic school-
undernutrition—to be greater than 80% for six key based nutrition programmes that incorporate students
nutrients, including calcium and folate.8 Stunting and and teachers as well as parents and the community have
thinness among girls are also high in northern,9 been effective in increasing consumption of nutritious
northwest,10,11 and northeastern Ethiopia.12 Anaemia foods and snacks such as fruits and vegetables, while
affects 20% of adolescents aged 15–19 years nationwide, reducing consumption of sugar-sweetened beverages and
and its prevalence is likely to be higher in pastoralist frequency of skipping breakfast.17 In sub-Saharan Africa,
regions, such as Afar and Somali, where anaemia school-based interventions can have a positive effect on
among women is highest.13 nutrition knowledge and micronutrient status of

www.thelancet.com/child-adolescent Vol 7 October 2023 687

 Articles

students,18 but little evidence exists for adolescent health across all study schools, was conducted to examine
and nutrition interventions.19,20 In Ethiopia, two school- balance between study groups and to inform intervention
based adolescent nutrition education interventions design and context. A smaller sample size was included
suggest that peer-led behaviour change communication at baseline due to budgetary constraints, and the study
can improve dietary diversity among girls21 and that impact was intended to be estimated at endline only.
nutrition lessons, group discussions, and recipe Adolescent girls were eligible for inclusion in the
demonstrations focused on the consumption of pulses surveys if they were enrolled in a study primary school,
among adolescent girls can improve knowledge and were aged 10–14 years of age and enrolled in grades 4–8,
intake.22 and had received parental consent and given assent.
Alive & Thrive is a global initiative that supports Either parent of the adolescent girl (mother or father
the scaling up of nutrition interventions to save lives, present at the time of interview) was invited to participate
prevent illnesses, and contribute to healthy growth and in the survey, and informed consent was sought from the
development. In Ethiopia, Alive & Thrive has been parent before the baseline and endline surveys.
supporting the government to achieve targets set out in Ethical approval for this study was granted by the
its National Nutrition Program II (2016−20), and, in 2019, Institutional Review Boards of the International Food
Alive & Thrive developed a package of locally tailored Policy Research Institute (IFPRI) in Washington, DC,
adolescent nutrition interventions to be delivered in USA, and the Addis Continental Institute of Public
For the protocol and statistical school, household, and community settings. The core Health (ACIPH) in Addis Ababa, Ethiopia. The study
analysis plan see https://classic. interventions involved nutrition education targeted to protocol and the statistical analysis plan are available
clinicaltrials.gov/ct2/show/
NCT04121559
adolescent girls through school-based activities such as online.
classroom lessons, flag ceremonies, school clubs, peer
mentoring sessions, and BMI measurement and Randomisation and masking
counselling sessions, and to their parents through The samples across the two regions were pooled to
parent–teacher meetings and take-home messages. The estimate intervention effects. We used stratified
school-based interventions were delivered by existing randomisation by woreda, to balance potential co-
cadres of school principals and teachers, who received interventions and confounders, and by school size. Then,
training about adolescent nutrition and how to schools were randomly allocated to the intervention or
implement classroom dialogue and hands-on activities to the control group by use of computer-generated pseudo-
engage adolescents and parents. Secondarily, health random numbers before the baseline survey.
extension workers and influential community actors The evaluation study was non-masked for all
were encouraged to deliver information about adolescent participants, programme implementers, and the research
nutrition to families and the broader community during team, including analysts. Adolescents and parents in the
health facility and home visits for other health services intervention group were not informed about the results of
and at community gatherings. the randomisation, but there was no masking of the
We present the results of the core package of school- interventions at schools or other points of contact. At the
based nutrition education interventions delivered time of enrolling and interviewing participants, assessors
through public primary schools in two regions (Southern were not informed about intervention allocation, and the
Nations, Nationalities, and People’s Region [SNNPR] and intervention allocation variable was added into the datasets
Somali) in Ethiopia in terms of feasibility and impact on at the time of data analysis after data cleaning.
dietary diversity, meal frequency, and consumption of
unhealthy (junk) foods in adolescent girls. Procedures
The baseline and endline surveys included independently
Methods selected study samples. Adolescent girls were selected
Study design and participants using a school-based sampling strategy. Lists of eligible
This non-masked, cluster-randomised, controlled trial girls were taken from primary school enrolment
took place in two regions: SNNPR, a primarily agrarian registries. Enrolment records (available at the time of the
region, and Somali, a region with a high population of surveys) were used to verify the birth date or age of
pastoralists, which were selected with the Federal eligible students, which were recorded from birth
Government of Ethiopia and regional health bureaus. certificates or parent reported. Then, adolescent girls
Alive & Thrive selected seven woredas (districts; four in were randomly selected by simple random sampling
SNNPR with 31 schools and three in the Somali region until the required sample size was reached, and they
with 23 schools) as potential intervention areas on the were visited at home to be recruited and interviewed,
basis of having adequate security and primary school along with their parents. Data were collected using
access and infrastructure. structured questionnaires, administered through
The study included primary schools with grades 1–8 computer-assisted personal interviewing (CAPI) by
within the government education system. A baseline trained enumerators; CAPI forms were programmed
survey, covering a small subsample of adolescent girls and data were collected via SurveyCTO. At baseline and

688 www.thelancet.com/child-adolescent Vol 7 October 2023

 Articles

endline, school food environments were assessed via Participants in the control group received standard
direct observation (appendix p 6). Adolescent girls were academic curriculum on health and nutrition, provided See Online for appendix
interviewed separately from their parents to provide as part of the government education system. School
privacy and reduce cues for response bias. principals and teachers were not instructed, supervised,
In February, 2020, following the baseline survey, Alive & or followed up about adolescent nutrition activities. No
Thrive conducted training for principals, teachers, and additional guidance or support was provided to the
health extension workers participating in the intervention. control schools.
Intervention materials (instruction manuals; adolescent
nutrition passports; BMI measurement guides; and Outcomes
posters on eating breakfast, healthy snacks, and dietary The primary outcome of impact was dietary diversity,
diversity) were also distributed. However, in March, 2020, defined as the number of food groups out of ten (grains,
the first COVID-19 case was reported in Ethiopia, and the white roots, and tubers; pulses such as beans, peas, and
country declared a state of emergency and closed all lentils; nuts and seeds; dairy; meat, poultry, and fish;
schools for the remainder of the 2019–20 academic year. eggs; dark green leafy vegetables; other vitamin A-rich
All programme activities were halted, and interventions fruits and vegetables; other vegetables; and other fruits)
were not delivered to students and parents. Schools consumed by adolescent girls over the previous
reopened for the 2020–21 academic year, and classes 24 h, and minimum dietary diversity.23 Adolescent girls
resumed in the study schools by October. In were asked about consumption of each of the ten food
November, 2021, Alive & Thrive reinitiated the programme groups during the day and night in the previous 24 h
with refresher training for all principals, teachers, and using a list-based method, which included local food
health extension workers in the intervention group. The examples from the Diet Quality Questionnaire adapted
duration of the school-based interventions was one school for Ethiopia.24 Minimum dietary diversity, or the
semester (about 4 months). proportion of girls who consumed foods from at least
Alive & Thrive’s core package of school-based five of the ten food groups, was also calculated as part of
interventions included six components: weekly flag the primary outcome to assess diet quality.23 Although
ceremonies, weekly classroom lessons, weekly school minimum dietary diversity is a proxy indicator for higher
club meetings, weekly peer group mentoring, BMI micronutrient adequacy for women aged 15–49 years, we
measurement and counselling (once per semester), and have applied this indicator for our study sample of girls
monthly parent–teacher meetings (appendix p 7). An aged 10–14 years in the absence of a specific indicator for
action point included in all intervention components this age group.
targeted to adolescent girls was to share what they heard Secondary outcomes were meal frequency and
or learned with their parents, so that parents’ meetings consumption of junk foods. Meal frequency (score of 0–6)
were not the only source of nutrition information for was calculated as the number of times meals and snacks
parents. Each intervention component included key were consumed during the day and night in the previous
nutrition messages related to dietary diversity (eating a 24 h both before the last school day and the last non-school
variety of foods), energy adequacy (eating breakfast and day. Adolescent girls were asked whether they consumed
healthy snacks), and healthy food choices (avoiding any food during six meal and snack times: breakfast,
unhealthy or so-called junk foods). Intervention materials snacks after breakfast but before a mid-day meal, lunch,
were designed to reflect locally appropriate language and afternoon snacks, dinner, and after-dinner snacks. Junk
images. Given that promoting all the diverse food groups food consumption included four categories: baked sweets,
might have been be too much information, interventions other sweets (eg, candy and chocolate), fried and salty
reinforced some locally available food examples that were foods, and soda and sugar-sweetened beverages.24
less frequently consumed, such as roasted mix of beans Adolescent girls were asked whether they consumed food
and nuts, banana, mango, milk, egg, and meat or fish. examples from each of these four categories in the previous
In addition to the school-based interventions, health 24 h; junk food consumption was defined as consumption
extension workers (working in the same area as an of any one of these categories. Indicators of nutrition
intervention school) were instructed to provide the key knowledge were estimated as the percentage of adolescents
intervention messages about adolescent nutrition during and parents who could name at least five different food
routine home visits for other health services at households groups (dietary diversity), knew the recommended number
in which adolescent girls lived. Health extension workers of meals and snacks per day (meal frequency), and
and influential community actors such as community and identified avoiding junk foods as part of good nutrition
religious leaders also provided adolescent nutrition practice (junk food consumption).
messages during community gatherings. These secondary To assess feasibility, exposure to each of the
interventions were not considered core interventions of six intervention components (flag ceremonies, classroom
the programme, were conducted irregularly, and reached lessons, school club meetings, peer group mentoring,
minimal exposure, and were, therefore, excluded from the BMI measurement and counselling, and parent–teacher
results in this paper. meetings) was calculated as the proportion of adolescent

www.thelancet.com/child-adolescent Vol 7 October 2023 689

 Articles

girls or parents who reported receiving or participating construct a wealth index using principal component
in the interventions in the past 3 months (secondary analysis.25 Study households were then separated into
outcome). We identified barriers and opportunities wealth terciles based on their wealth rank. The Household
related to intervention implementation based on the Food Insecurity Access Scale26 was used to measure the
exposure results. access component of household food security and define
The original protocol was amended (on March 9, 2021) households as food secure, mildly food insecure,
to remove a quantitative 24 h recall dietary assessment, moderately food insecure, or severely food insecure.
the corresponding outcome indicators on adequate The primary analysis was performed in the intention-to-
intake of micronutrients and macronutrients, and a sub- treat population at endline. We estimated intervention
sample panel of adolescent girls at endline due to effects using linear regression models for mean
budgetary and time constraints. differences, with SEs clustered at the school level obtained
No adverse events were anticipated or monitored actively, using a sandwich estimator (unadjusted model). For
and none were reported by teachers or other school staff binary outcomes, we reported intervention effects as odds
during routine supervision of intervention activities. ratios (ORs) and 95% CIs estimated from logistic
regression models. In adjusted models, we controlled for
Statistical analysis adolescent age, region, household food security, and
The sample size was estimated based on a mean dietary wealth. We examined a dose–response effect in the
diversity score of 3·26 (SD 1·40) using the women’s dataset intervention group only by using multivariable regression
from the Alive & Thrive phase 1 endline survey for SNNPR, analyses to test associations between exposure to the
assuming an intra-cluster correlation (ICC) of 0·18, total number of intervention components or the individual
cluster number of 54, 80% power to detect differences, and components and the primary and secondary outcomes,
an α value of 0·05. We calculated that a sample size of adjusted for adolescent age and geographical clustering.
189 girls per group (seven per cluster) would be able to Two-sided t tests were used to compare and infer
detect a minimum difference of 0·6 food groups. The significant differences between study groups at endline.
sample size was rounded to 270 girls per group so that For categorical variables, Pearson’s χ² tests were performed.
ten girls per school (two per grade) would be sampled, for All statistical analysis was done using Stata (version 17.0).
a total sample size of 540 adolescent girls. This trial is registered at ClinicalTrials.Gov,
Based on our baseline data (small subsample), we NCT04121559, and is complete.
observed a mean dietary diversity score of 3·69 (SD 1·56)
and an ICC of 0·015; using this information, we estimated Role of the funding source
that this sample size for the endline would detect a The funder of the study had no role in study design, data
minimum difference of 0·5 food groups with 93% power. collection, data analysis, data interpretation, or writing of
Descriptive analyses were conducted for sample the report.
characteristics and exposure to intervention components.
Control variables used in the analyses included household Results
wealth and household food security. Data on household 27 primary schools were randomly allocated to the
asset ownership and housing characteristics were used to intervention group (with a median of 625 [IQR 424–1004]

54 primary school clusters identified and assessed

for eligibility in 7 woredas in Southern Nations,
Nationals, and People’s Region and Somali region

54 primary school clusters randomly allocated

27 allocated to the school-based 27 allocated to the standard school

nutrition education interventions curriculum, no nutrition-specific
(intervention group) interventions (control group)

27 primary school clusters, including 27 primary school clusters, including 27 primary school clusters, including 27 primary school clusters, including
81 adolescent girls, included in 270 adolescent girls, included in 81 adolescent girls, included in 266 adolescent girls, included in
intention-to-treat analysis for intention-to-treat analysis for intention-to-treat analysis for intention-to-treat analysis for
2019 baseline 2021 endline 2019 baseline 2021 endline

Figure: Trial profile

690 www.thelancet.com/child-adolescent Vol 7 October 2023

 Articles

pupils) and 27 to the control group (with a median least five food groups; 182 [67%] of 270 in the
of 641 [375–1141] pupils). 162 in-school adolescent girls intervention group vs 76 [29%] of 266 in the control
participated in the baseline survey between Oct 8 and group; adjusted OR 5·37 [95% CI 3·04–9·50], p<0·0001).
Nov 4, 2019, and 536 adolescent girls (median age A breakdown of the individual food groups consumed
13·3 years [IQR 12·1–14·0]) participated in the endline reflects an intervention effect on seven of the ten food
survey between March 22 and April 29, 2021 (figure). All groups, including those foods promoted by the
54 primary schools were included at baseline and interventions.
endline, with no clusters lost to follow-up. At baseline, For meal frequency in the past 24 h, we observed an
demographic characteristics were similar between adjusted mean difference of 0·84 (95% CI 0·58–1·09,
study groups (table 1). At endline, 518 (97%) of the p<0·0001) meal or snack times between study groups
536 adolescent girls surveyed were students returning to (4·04 [SD 0·92] in the intervention group vs 3·17 [0·88]
the same schools, so very few were newly enrolled in in the control group) at endline (table 2). Improvements
schools at endline, and households in the study areas were observed in the breakfast and morning and mid-
were likely to have remained in place over the previous afternoon snacks times. There was no significant
year. difference in consumption of junk foods in the past 24 h
For the primary outcome of dietary diversity, we between study groups (131 [49%] of 270 in the intervention
observed an adjusted mean difference of 1·33 (95% CI group vs 146 [55%] of 266 in the control group; adjusted
0·90–1·75, p<0·0001) food groups in the dietary OR 0·73, 95% CI 0·46–1·14, p=0·17). A breakdown of the
diversity score between the intervention (mean 5·37 specific categories of unhealthy foods showed an
[SD 1·66]) and control (3·98 [1·43]) groups at endline intervention effect on the consumption of other sweets
(table 2). The intervention was also associated with an such as candies and chocolates only (38 [14%] of 270 girls
increase in minimum dietary diversity (consuming at in the intervention group vs 61 [23%] of 266 girls;

2019 baseline 2021 endline

Intervention group Control group Intervention group Control group p value*
Primary schools 27 27 27 27 ··
Primary school characteristics
Number with a toilet facility available 24 (89%) 24 (89%) 27 (100%) 25 (93%) 0·16
Number of classrooms per school 12·0 (6·2) 11·3 (5·5) 14·0 (9·8) 12·4 (5·5) 0·48
Number of teachers per school 18·4 (13·1) 16·3 (10·0) 19·2 (13·4) 17·0 (10·6) 0·49
Adolescent girls surveyed 81 81 270 266 ··
Adolescent dietary diversity
Dietary diversity score (0–10 food groups) 3·80 (1·59) 3·58 (1·52) 5·37 (1·66) 3·98 (1·43) <0·0001
Minimum dietary diversity (≥5 food groups) 24 (30%) 19 (23%) 182 (67%) 76 (29%) <0·0001
Meal frequency in past 24 h (0–6) ·· ·· 4·0 (0·9) 3·2 (0·9) <0·0001
Junk food consumption in past 24 h 50 (62%) 48 (59%) 131 (49%) 146 (55%) 0·29
Age, years ·· ·· ·· ·· 0·27
Mean (SD) 12·6 (1·2) 12·8 (1·2) 12·9 (1·2) 13·0 (1·2) ··
Median (IQR) 12·9 (12·0–13·4) 13·1 (12·0–13·9) 13·3 (12·1–14·0) 13·4 (12·3–14·0) ··
BMI, kg/m² 17·2 (2·5) 17·4 (2·2) 17·4 (2·7) 17·3 (2·5) 0·84
Grade level ·· ·· ·· ·· 0·99
4 8 (10%) 12 (15%) 53 (20%) 57 (21%) ··
5 15 (19%) 20 (25%) 55 (20%) 53 (20%) ··
6 19 (23%) 18 (22%) 56 (21%) 55 (21%) ··
7 18 (22%) 22 (27%) 53 (20%) 51 (19%) ··
8 21 (26%) 9 (11%) 53 (20%) 50 (19%) ··
Languages spoken
Amharic 47 (58%) 38 (47%) 156 (58%) 128 (48%) 0·45
Gurage 25 (31%) 22 (27%) 81 (30%) 76 (29%) 0·90
Somali 32 (40%) 36 (44%) 109 (40%) 116 (44%) 0·81
Other 22 (27%) 22 (27%) 77 (29%) 72 (27%) 0·90
Residence of parents
Father currently resides with adolescent 63 (78%) 67 (83%) 201 (74%) 218 (82%) 0·046
Mother currently resides with adolescent 78 (96%) 79 (98%) 251 (93%) 252 (95%) 0·47
(Table 1 continues on next page)

www.thelancet.com/child-adolescent Vol 7 October 2023 691

 Articles

2019 baseline 2021 endline

Intervention group Control group Intervention group Control group p value*
(Continued from previous page)
Parents surveyed 81 81 270 266 ··
Parent respondent type ·· ·· ·· ·· 0·84
Father 16 (20%) 10 (12%) 65 (24%) 66 (25%) ··
Mother 65 (80%) 71 (88%) 205 (76%) 200 (75%) ··
Father’s age, years 44·8 (13·8) 46·7 (11·6) 48·9 (10·7) 48·8 (11·3) 0·98
Father’s education level† ·· ·· ·· ·· 0·72
Never attended school 3 (19%) 1 (10%) 18 (28%) 21 (32%) ··
Primary school (grades 1–8) 11 (69%) 5 (50%) 35 (54%) 36 (55%) ··
Secondary school or higher 2 (13%) 4 (40%) 12 (18%) 9 (14%) ··
Father’s occupation ·· ·· ·· ·· 0·0075
Farmer 9 (56%) 7 (70%) 46 (71%) 59 (89%) ··
Other 7 (44%) 3 (30%) 19 (29%) 7 (11%) ··
Mother’s age, years 38·3 (9·4) 39·7 (8·1) 38·6 (7·9) 39·8 (7·9) 0·13
Mother’s education level† ·· ·· ·· ·· 0·045
Never attended school 31 (48%) 38 (54%) 101 (49%) 115 (58%) ··
Primary school (grades 1–8) 29 (45%) 33 (46%) 81 (40%) 75 (38%) ··
Secondary school or higher 5 (8%) 0 (0%) 23 (11%) 10 (5%) ··
Mother’s occupation ·· ·· ·· ·· 0·12
Housewife 33 (51%) 37 (52%) 94 (46%) 96 (48%) ··
Farmer 12 (18%) 18 (25%) 46 (23%) 58 (29%) ··
Other 20 (31%) 16 (23%) 64 (31%) 46 (23%) ··
Household Food Insecurity Access Scale ·· ·· ·· ·· 0·016
Food secure 35 (43%) 36 (44%) 159 (59%) 132 (50%) ··
Mildly insecure 7 (9%) 8 (10%) 23 (9%) 15 (6%) ··
Moderately insecure 27 (33%) 29 (36%) 60 (22%) 71 (27%) ··
Severely insecure 12 (15%) 8 (10%) 28 (10%) 48 (18%) ··
Wealth tercile ·· ·· ·· ·· 0·0048
Low 29 (36%) 25 (31%) 85 (31%) 94 (35%) ··
Middle 25 (31%) 29 (36%) 78 (29%) 101 (38%) ··
High 27 (33%) 27 (33%) 107 (40%) 71 (27%) ··
Data are n, mean (SD), or median (IQR) unless otherwise stated. *Differences between study groups at endline, accounting for clustering, were calculated using two-sided t tests.
For categorical variables, Pearson’s χ² tests were performed. †At endline, parents (two fathers and four mothers) who attended religious school are included in the “Never
attended school” category.

Table 1: Sample characteristics

difference 8·8 percentage points; adjusted OR 0·49, girls and their parents. 256 (95%) of 270 adolescent girls
95% CI 0·29–0·83, p=0·0085). in the intervention group were exposed to at least one of
Both adolescent girls and their parents in the the five in-school intervention components (table 4).
intervention group had higher knowledge at endline Adolescent girls in the intervention group most commonly
than those in the control group about dietary diversity, received nutrition information via flag ceremonies (230
meal frequency, and consumption of unhealthy foods [85%]), classroom lessons (226 [84%]), and BMI
(table 3). Adolescent girls had more knowledge about measurements (197 [73%]). 145 (54%) of 270 parents of
meal frequency than about dietary diversity or avoiding adolescent girls in the intervention group were also
unhealthy foods; similar patterns were observed among exposed to discussions about nutrition at parents’
their parents. meetings at schools.
At endline, free school meals were not provided, nearly There was an association between exposure to multiple
all study schools had at least one food vendor within 1 min intervention components and improved outcomes
walking distance (52 of 54 schools), and about a quarter of (appendix p 8). We observed that exposure to at least
all adolescent girls had seen or heard a food advertisement two intervention components was associated with
in the past 3 months (appendix p 6). The school-based substantially higher odds of achieving minimum dietary
intervention was highly successful in reaching adolescent diversity and high meal frequency among adolescent

692 www.thelancet.com/child-adolescent Vol 7 October 2023

 Articles

Endline Unadjusted Adjusted*

Intervention Control β or odds ratio p value β or odds ratio p value
group (n=270) group (n=266) (95% CI) (95% CI)
Dietary diversity
Dietary diversity score† (0–10 food groups) 5·37 (1·66) 3·98 (1·43) 1·39 (0·90–1·88) <0·0001 1·33 (0·90–1·75) <0·0001
Minimum dietary diversity† (≥5 food groups) 182 (67%) 76 (29%) 5·17 (2·94–9·10) <0·0001 5·37 (3·04–9·50) <0·0001
Food groups consumed in the past 24 h
Grains, white roots, and tubers 264 (98%) 264 (99%) 0·33 (0·07–1·55) 0·16 0·25 (0·06–1·05) 0·058
Pulses (beans, peas, and lentils) 199 (74%) 148 (56%) 2·23 (1·39–3·60) 0·0010 2·39 (1·51–3·76) 0·0002
Nuts and seeds 68 (25%) 35 (13%) 2·22 (1·02–4·84) 0·044 3·10 (1·52–6·32) 0·0018
Dairy 146 (54%) 97 (36%) 2·05 (0·95–4·41) 0·066 3·09 (1·53–6·27) 0·0017
Meat, poultry, and fish 67 (25%) 31 (12%) 2·50 (1·16–5·40) 0·019 3·42 (2·02–5·80) <0·0001
Eggs 113 (42%) 45 (17%) 3·53 (1·79–6·98) 0·0003 4·14 (2·06–8·30) <0·0001
Dark green leafy vegetables 130 (48%) 121 (45%) 1·11 (0·55–2·26) 0·77 1·03 (0·56–1·89) 0·94
Other vitamin A-rich fruits and vegetables 99 (37%) 37 (14%) 3·58 (2·04–6·30) <0·0001 3·44 (1·97–6·02) <0·0001
Other vegetables 252 (93%) 237 (89%) 1·71 (0·81–3·61) 0·16 1·53 (0·75–3·11) 0·24
Other fruits 112 (41%) 44 (17%) 3·58 (1·93–6·62) <0·0001 3·32 (1·84–6·00) <0·0001
Meal frequency in past 24 h (0–6)† 4·04 (0·92) 3·17 (0·88) 0·87 (0·61–1·13) <0·0001 0·84 (0·58–1·09) <0·0001
Meal or snack times, %
Breakfast 238 (88%) 199 (75%) 2·50 (1·12–5·57) 0·025 2·32 (1·05–5·15) 0·038
Morning snack 141 (52%) 44 (17%) 5·51 (2·85–10·69) <0·0001 6·45 (3·57–11·66) <0·0001
Lunch 253 (94%) 242 (91%) 1·48 (0·61–3·54) 0·38 1·30 (0·56–2·98) 0·54
Afternoon snack 168 (62%) 84 (32%) 3·57 (2·23–5·72) <0·0001 3·60 (2·26–5·75) <0·0001
Dinner 262 (97%) 259 (97%) 0·89 (0·31–2·52) 0·82 0·70 (0·23–2·10) 0·52
Evening snack 28 (10%) 15 (6%) 1·94 (0·88–4·26) 0·10 2·07 (0·93–4·58) 0·074
Consumed any junk foods in past 24 h† 131 (49%) 146 (55%) 0·77 (0·49–1·23) 0·28 0·73 (0·46–1·14) 0·17
Junk foods consumed in the past 24 h
Baked sweets 36 (13%) 38 (14%) 0·92 (0·45–1·88) 0·83 0·91 (0·47–1·79) 0·79
Other sweets (candy or chocolate) 38 (14%) 61 (23%) 0·55 (0·32–0·95) 0·03 0·49 (0·29–0·83) 0·0085
Fried and salty foods 29 (11%) 37 (14%) 0·74 (0·42–1·33) 0·32 0·72 (0·41–1·28) 0·27
Soda and sugar-sweetened beverages 88 (33%) 92 (35%) 0·91 (0·57–1·47) 0·71 0·92 (0·61–1·40) 0·70
Data are mean (SD) or n (%) unless otherwise stated. Differences between study groups are by intention-to-treat analyses, accounting for clustering using linear regression
models for mean differences, with SEs clustered at the school level obtained using a sandwich estimator. For binary outcomes, intervention effects are reported as odds ratios
and 95% CIs estimated from logistic regression models. *Linear or logistic regression, adjusted for adolescent age, region, household food security, and wealth, clustered by
school. †Intra-cluster correlation coefficients: dietary diversity score (0·392), minimum dietary diversity (0·284), meal frequency in past 24 h (0·350), and consumed any junk
foods in past 24 h (0·102).

Table 2: Dietary diversity, consumption of junk foods, and meal frequency among adolescent girls at endline

girls. Although we observed some associations between The diets of adolescent girls in our study had low diversity
exposure to individual intervention components and overall. Low dietary diversity among adolescent girls aged
improved outcomes, there was too much overlap in the 10–19 years (broader than the 10–14 year age range in our
exposure to individual components to draw conclusions study) was corroborated in a recent observational study in
on their comparative effects (appendix p 9). SNNPR, one of our study regions; the mean dietary
diversity score was 3·6 food groups, and only 28% of
Discussion participants consumed at least five food groups.27 Each of
In this non-masked, cluster-randomised, controlled trial, the Alive & Thrive intervention components was designed
a package of school-based nutrition education to address information about locally available foods that
interventions delivered during one school semester make up a diverse diet, particularly specific food groups
(about 4 months) improved practices and knowledge with large room for improvement based on formative
related to dietary diversity and meal frequency among study, and provide motivation to acquire and consume a
young adolescent girls. High reported exposure to the diverse diet. After the interventions, we observed an impact
intervention components was achieved in the intervention of about one food group difference between study groups,
group. We observed lower consumption of sweets in the and minimum dietary diversity increased from 30% at
previous 24 h, but no impact on consumption of any baseline to 67% at endline in the intervention group. A
other junk foods. study in Jimma zone, Oromia region, showed that peer-led

www.thelancet.com/child-adolescent Vol 7 October 2023 693

 Articles

Endline Unadjusted Adjusted*

Intervention Control Odds ratio (95% CI) p value Odds ratio (95% CI) p value
group (n=270) group (n=266)
Adolescent girls
Knows ≥5 food groups of a diverse diet 111 (41%) 26 (10%) 6·44 (3·56–11·66) <0·0001 6·43 (3·54–11·68) <0·0001
Knows to eat ≥4 times per day, ≥3 meals per day, 230 (85%) 120 (45%) 7·00 (3·73–13·13) <0·0001 6·80 (3·70–12·50) <0·0001
and ≥1 snack per day
Knows avoiding junk food and sugar-sweetened 118 (44%) 36 (14%) 4·96 (2·79–8·80) <0·0001 4·93 (2·74–8·86) <0·0001
beverages is good nutrition
Parents
Knows ≥5 food groups of a diverse diet 95 (35%) 35 (13%) 3·58 (2·16–5·95) <0·0001 3·70 (2·21–6·17) <0·0001
Knows to eat ≥4 times per day, ≥3 meals per day, 218 (81%) 120 (45%) 5·10 (2·82–9·21) <0·0001 4·71 (2·63–8·46) <0·0001
and ≥1 snack per day
Knows avoiding junk food and sugar-sweetened 89 (33%) 40 (15%) 2·78 (1·71–4·51) <0·0001 2·84 (1·69–4·75) <0·0001
beverages is good nutrition
Data are n (%) unless otherwise stated. Differences between study groups in intention-to-treat analyses, accounting for clustering using logistic regression models. *Logistic
regression, adjusted for adolescent age, region, household food security, and wealth, clustered by school.

Table 3: Prevalence of correct knowledge about nutrition among adolescent girls and parents at endline

Endline Unadjusted Adjusted*

Intervention Control Odds ratio (95% CI) p value Odds ratio (95% CI) p value
group group
(n=270) (n=266)
Intervention
Heard about nutrition during flag ceremony 230 (85%) 41 (15%) 31·55 (15·98–62·32) <0·0001 33·43 (16·93–66·00) <0·0001
Heard about nutrition in the classroom 226 (84%) 61 (23%) 17·26 (9·20–32·40) <0·0001 18·66 (9·58–36·33) <0·0001
Participated in a club about nutrition for girls 121 (45%) 7 (3%) 30·05 (10·48–86·16) <0·0001 30·43 (9·74–95·09) <0·0001
Participated in a mentorship programme for girls 120 (44%) 6 (2%) 34·67 (13·18–91·18) <0·0001 38·20 (13·89–105·11) <0·0001
BMI measurements taken at school 197 (73%) 1 (<1%) 715·14 (97·24–5259·20) <0·0001 830·46 (108·54–6353·91) <0·0001
Nutrition discussed at school parents’ meeting 145 (54%) 12 (5%) 24·55 (13·00–46·37) <0·0001 27·19 (13·66–54·11) <0·0001
Exposed to ≥1 interventions 256 (95%) 85 (32%) 38·94 (11·29–134·24) <0·0001 42·58 (12·92–140·40) <0·0001
Data are n (%) unless otherwise stated. Differences between study groups in intention-to-treat analyses, accounting for clustering using logistic regression models. *Logistic
regression, adjusted for adolescent age, region, household food security, and wealth, clustered by school.

Table 4: Exposure to interventions in the past 3 months among adolescent girls and their parents at endline

behaviour change communication through school media compared with knowledge that at least five food groups
and health clubs increased minimum dietary diversity make up a diverse diet and that avoiding junk food and
among school adolescents aged 10–19 years from 35% at sugar-sweetened beverages is part of good nutrition.
baseline to 75% at endline in the intervention group.21 However, meal frequency was three meals per day on
Additionally, a 6-month school nutrition education average in the control areas at endline; this was similar to
intervention (bi-monthly lessons with recipes and tastings) findings from an observational study in Damot Sore
that specifically focused on pulse consumption among district, SNNPR, in which 76% of 719 adolescent girls
adolescent girls in Halaba district, SNNPR, improved aged 10–19 years consumed three or fewer meals per
knowledge and intake of pulses.22 Both of these studies day.7 We observed an intervention effect of nearly one
included predominantly farming and food insecure meal or snack consumed, mainly due to an increase in
households,21,22 similar to our study households. These the consumption of breakfast and mid-morning or mid-
studies as well as our study indicated that, if designed to afternoon snacks. Improvement in these meal or snack
address behavioural determinants such as knowledge, periods aligned with the key intervention messages of
attitudes, and beliefs, in-school nutrition education consuming breakfast before school and eating a healthy
interventions might be effective in improving diversified snack every day.
food or specific food consumption among adolescents, Despite an intervention effect on the consumption of
even in rural and food insecure contexts. other sweets, more than half of the adolescent girls
Meal frequency knowledge was higher among reported consuming junk food, with no differences in
adolescent girls and their parents in both study groups the consumption of fried and salty foods, baked sweets,

694 www.thelancet.com/child-adolescent Vol 7 October 2023

 Articles

or soda and sugar-sweetened beverages between study low cost to develop and implement. Total estimated
groups. Knowledge about avoiding junk food as part of programme costs (including programme staff salaries,
good nutrition improved among adolescent girls and materials and other resources, transportation, non-
their parents between study groups, but knowledge did governmental organisation budget, and school staff
not translate into practice. This result might not be salary) for programme reach per beneficiary was
surprising, considering that consumption of unhealthy roughly US$43·22 per adolescent girl (grades 4–8) in
foods is influenced by a complex set of factors. Junk the intervention schools, or, if considered school wide,
foods are designed to be tasty, convenient (ready-to-eat $13·79 per student (boys and girls in grades 1–8).
packaging), and easily available, and are subject to Study data collected were based on self-reports, which
commercial advertisements. At endline, nearly all study are subject to recall and social desirability bias. To
schools had at least one food vendor within 1 min minimise these biases, measures for the primary and
walking distance, mainly selling sweets and soda or secondary outcomes used locally available food examples
sugar-sweetened beverages, and about a quarter of or specific meal or snack times throughout the day with a
adolescent girls had seen or heard a food advertisement short 24 h recall period. Adolescent girls were interviewed
in the past 3 months. Therefore, nutrition education separately from their parents and the variation in
and telling adolescents not to eat junk foods that they intervention effects across outcomes also suggests that
crave or enjoy without changing their food environments adolescents might not be reporting socially desirable
might not be successful in curbing junk food responses. No hard clinical outcomes were included in
consumption. this study because the endpoints were dietary behaviours.
Although it was not possible to disentangle the effects In relation to adherence among study participants, we
of individual components within our intervention assessed implementation fidelity by measuring exposure
package, exposure to at least two components was to programme inputs, delivery of interventions by school
associated with improved outcomes. A previous study staff, and exposure to each intervention component
on the association between the combination of among adolescent girls and their parents over the
behaviour change interventions or number of contacts intervention period. Moreover, the primary analysis to
and child feeding practices showed that results are assess effectiveness was prespecified as intention-to-treat
context specific;28 however, using multiple social and analysis at endline; therefore, all participants remained
behaviour change communication activities and in the study sample for analysis, regardless of adherence
channels to change behaviours might be more effective to the protocol. Our study was conducted during the
than using one.29 In our study, the frequency of the COVID-19 pandemic, which affected school schedules
six core intervention components varied from weekly and activities; however, there were no differences in the
to monthly, and the effect of multiple components school conditions by study group. Selection bias is
reinforcing the key intervention messages probably possible given that the schools were randomly allocated
contributed to the improved dietary practices. and unmasked before participants were enrolled in the
Parents had an essential role in the diets of young endline survey, but we consider this bias to be minimal
adolescents in our study. Although the adolescent girls given that interventions were not rolled out to students
spent time away from their parents throughout the and parents before the pandemic and took place after
school day, parents were still the primary source of and school enrolment at the start of the new academic year.
gatekeepers to their foods. Free school meals were not In conclusion, a package of nutrition education
provided in any of our study schools, and only one interventions delivered in primary schools in Ethiopia
intervention school had a canteen but no food items were was feasible to implement and effective in improving
observed, so adolescents were limited to foods from adolescent girls’ dietary practices. Reinforcing messages
home, foods received from peers or others, or foods about eating diverse foods and eating more often resulted
purchased from around the home or school premises. in incremental behaviour change related to dietary
Nutrition education targeted to, and active involvement diversity and meal frequency; however, informing
of, parents are crucial components of adolescent adolescents to avoid junk foods, without addressing their
nutrition interventions. Evidence of improved food food environments, was not effective in reducing junk
preparation by parents for their adolescent daughters food consumption. Other intervention strategies might
and their home food environment were observed in the be required to change food environments, such as the
intervention group at endline. provision or product placement of healthy food or snack
Our study addresses the important evidence gap on options at or near schools, and behaviours particularly
the effectiveness of nutrition education interventions to related to consumption of junk foods, which is a growing
improve adolescents’ dietary practices, particularly in concern among adolescents even in the rural areas in
Ethiopia. Our randomised design lends confidence Ethiopia, as in many other contexts.
in attributing the observed impact and changes in Contributors
outcomes to the interventions. Furthermore, the SSK, CS, TTW, AAO, TS, EAF, and PM designed the study. SSK, CS,
package of school-based interventions was relatively HYB, and PM developed the study protocol. SSK and CS designed the

www.thelancet.com/child-adolescent Vol 7 October 2023 695

 Articles

data collection instruments with input from all authors. SSK, CS, and 13 Central Statistical Agency Ethiopia. Ethiopia Demographic and
HYB implemented and supervised the fieldwork. SSK, CS, EAF, and Health Survey 2016. 2016. https://dhsprogram.com/pubs/pdf/
PM developed the analysis strategy. SSK and CS analysed the data and FR328/FR328.pdf (accessed Aug 30, 2022).
wrote the first draft. All authors reviewed, made inputs to, and approved 14 Salam RA, Hooda M, Das JK, et al. Interventions to improve
the final paper. All authors had full access to all de-identified adolescent nutrition: a systematic review and meta-analysis.
anonymised data in the study and had final responsibility for the J Adolesc Health 2016; 59: S29–39.
decision to submit the manuscript for publication. The underlying data 15 Lassi ZS, Moin A, Das JK, Salam RA, Bhutta ZA. Systematic review
were also accessible to and verified by CS. on evidence-based adolescent nutrition interventions.
Ann NY Acad Sci 2017; 1393: 34–50.
Declaration of interests 16 Hargreaves D, Mates E, Menon P, et al. Strategies and interventions
We declare no competing interests. for healthy adolescent growth, nutrition, and development. Lancet
2022; 399: 198–210.
Data sharing
De-identified participant data, data dictionaries, and questionnaires from 17 Wang D, Stewart D. The implementation and effectiveness of
school-based nutrition promotion programmes using a health-
the endline survey have been submitted for open access publication and
promoting schools approach: a systematic review. Public Health Nutr
will be available on Harvard Dataverse. 2013; 16: 1082–100.
Acknowledgments 18 Kyere P, Veerman JL, Lee P, Stewart DE. Effectiveness of school-
We are grateful for the collaboration with Addis Continental Institute of based nutrition interventions in sub-Saharan Africa: a systematic
Public Health (ACIPH) and its staff for data collection. We thank all the review. Public Health Nutr 2020; 23: 2626–36.
adolescent girls and their families for time spent in this study. This 19 Bhutta ZA, Lassi ZS, Bergeron G, et al. Delivering an action agenda
study was supported by funding from the Bill & Melinda Gates for nutrition interventions addressing adolescent girls and young
Foundation (grant number OPP1171874), through Alive & Thrive, women: priorities for implementation and research.
managed by FHI Solutions. Ann NY Acad Sci 2017; 1393: 61–71.
20 Salam RA, Das JK, Irfan O, Ahmed W, Sheikh SS, Bhutta ZA.
References Effects of preventive nutrition interventions among adolescents on
1 Das JK, Salam RA, Thornburg KL, et al. Nutrition in adolescents: health and nutritional status in low- and middle-income countries:
physiology, metabolism, and nutritional needs. Ann NY Acad Sci a systematic review. Campbell Syst Rev 2020; 16: e1085.
2017; 1393: 21–33. 21 Tamiru D, Argaw A, Gerbaba M, Nigussie A, Ayana G, Belachew T.
2 Spear BA. Adolescent growth and development. J Am Diet Assoc Improving dietary diversity of school adolescents through school
2002; 102 (suppl): S23–29. based nutrition education and home gardening in Jimma Zone:
3 Akseer N, Al-Gashm S, Mehta S, Mokdad A, Bhutta ZA. Global and Quasi-experimental design. Eat Behav 2016; 23: 180–86.
regional trends in the nutritional status of young people: a critical 22 Dansa R, Reta F, Mulualem D, Henry CJ, Whiting SJ. A nutrition
and neglected age group. Ann NY Acad Sci 2017; 1393: 3–20. education intervention to increase consumption of pulses showed
4 United Nations Population Fund. World Population Dashboard improved nutritional status of adolescent girls in Halaba special
Ethiopia. 2022. https://www.unfpa.org/data/world-population/ET district, Southern Ethiopia. Ecol Food Nutr 2019; 58: 353–65.
(accessed Aug 17, 2022). 23 Food and Agricultural Organization of the UN. FHI 360. Minimum
5 Berhe K, Kidanemariam A, Gebremariam G, Gebremariam A. dietary diversity for women: a guide for measurement. 2016.
Prevalence and associated factors of adolescent undernutrition in https://www.fao.org/nutrition/assessment/tools/minimum-dietary-
Ethiopia: a systematic review and meta-analysis. BMC Nutr 2019; diversity-women/en/ (accessed Aug 17, 2022).
5: 49. 24 Global Diet Quality Project. DQQ Tools. 2022. https://www.
6 Handiso YH, Belachew T, Abuye C, Workicho A, Baye K. globaldietquality.org/dqq (accessed Aug 20, 2022).
Undernutrition and its determinants among adolescent girls in low 25 Demographic and Health Surveys. Wealth Index Construction.
land area of Southern Ethiopia. PLoS One 2021; 16: e0240677. https://dhsprogram.com/topics/wealth-index/Wealth-Index-
7 Gagebo DD, Kerbo AA, Thangavel T. Undernutrition and associated Construction.cfm (accessed Oct 20, 2020).
factors among adolescent girls in Damot Sore District, Southern 26 Coates J, Swindale A, Bilinsky P. Household Food Insecurity
Ethiopia. J Nutr Metab 2020; 2020: 5083140. Access Scale (HFIAS) for measurement of food access: indicator
8 Yilma B, Endris BS, Mengistu YG, Sisay BG, Gebreyesus SH. guide (v.3). August, 2007. https://www.fantaproject.org/sites/
Inadequacy of nutrient intake among adolescent girls in south default/files/resources/HFIAS_ENG_v3_Aug07.pdf (accessed
central Ethiopia. J Nutr Sci 2021; 10: e90. Aug 11, 2022).
9 Berhe K, Gebremariam G. Magnitude and associated factors of 27 Halala Handiso Y, Belachew T, Abuye C, Workicho A. Low dietary
undernutrition (underweight and stunting) among school adolescent diversity and its determinants among adolescent girls in Southern
girls in Hawzen Woreda (District), Tigray regional state, Northern Ethiopia. Cogent Food Agric 2020; 6: 1832824.
Ethiopia: cross-sectional study. BMC Res Notes 2020; 13: 59. 28 Kim SS, Nguyen PH, Tran LM, Alayon S, Menon P, Frongillo EA.
10 Tariku A, Belew AK, Gonete KA, et al. Stunting and its Different combinations of behavior change interventions and
determinants among adolescent girls: findings from the Nutrition frequencies of interpersonal contacts are associated with infant and
Surveillance Project, Northwest Ethiopia. Ecol Food Nutr 2019; young child feeding practices in Bangladesh, Ethiopia, and
58: 481–94. Vietnam. Curr Dev Nutr 2019; 4: nzz140.
11 Kebede D, Prasad RPCJ, Asres DT, Aragaw H, Worku E. Prevalence 29 Lamstein S, Stillman T, Koniz-Booher P, et al. Evidence of effective
and associated factors of stunting and thinness among adolescent approaches to social and behavior change communication for
students in Finote Selam Town, Northwest Ethiopia. preventing and reducing stunting and anemia: report from a
J Health Popul Nutr 2021; 40: 44. systematic literature review. August, 2014. https://www.spring-
12 Hadush G, Seid O, Wuneh AG. Assessment of nutritional status nutrition.org/sites/default/files/publications/series/spring_sbcc_
and associated factors among adolescent girls in Afar, Northeastern lit_review.pdf (accessed Aug 18, 2022).
Ethiopia: a cross-sectional study. J Health Popul Nutr 2021; 40: 2.

696 www.thelancet.com/child-adolescent Vol 7 October 2023