Childhood is a critical stage for growth and cognitive development, with nutrition playing an essential role in supporting physical, mental and cognitive well-being. Breakfast is particularly fundamental for children’s health and academic success, as it provides the energy and nutrients needed for concentration, memory and physical activity throughout the school day(Reference Kadosh, Muhardi and Parikh1). Studies have shown that a well-balanced breakfast can lead to better academic performance, higher physical activity levels and improved overall nutrition status(Reference Prangthip, Soe and Signar2,Reference Kingshipp, Raghavan and Gungor3) . However, many children start their day either without a breakfast or with a breakfast that lacks essential nutrients, potentially compromising their learning, energy levels and health outcomes. In Australia, for example, it is reported that 45 % of schoolchildren reported skipping breakfast at least once a week, with 9·5 % always skipping breakfast(Reference Sincovich, Moller and Smithers4). Even higher rates of skipping breakfast at least once a week were found in the USA (72·6 %), with 17·9 % of youths skip breakfast everyday(Reference Sliwa, Merlo and Mckinnon5). Meanwhile in Malaysia, 36 % of children experience skipping meals at least once a week(Reference Yeo, Lee and Wong6). In Indonesia, the prevalence of schoolchildren who do not eat breakfast reached 65 %(Reference Kautsar7). Moreover, another study found that the majority do not have adequate and good quality breakfast (90·8 %)(Reference Harahap, Widodo and Sandaja8). The issues related to inadequate breakfast consumption is an unfortunate global trend that needs serious attention to address.
While there are many factors affecting breakfast consumption among children, it is also known that nutrition education plays a crucial role in fostering healthy eating habits, such as regular, balanced breakfast consumption. It empowers children and their families with knowledge about nutrient-rich foods and their benefits, encouraging lifelong habits that support cognitive function and overall well-being. While research supports the link between a nutritious breakfast and improved cognitive and academic outcomes(Reference Prangthip, Soe and Signar2), and nutrients such as plant and marine n-3 and n-6 fatty acids having beneficial effects on cognitive health are well documented(Reference Djuricic and Calder9), there remains a gap in the knowledge on the impact of nutritional education and quality breakfast on school-aged children.
Currently, there is lacking evidence on the effect of a coupled nutritious breakfast provision and nutrition education intervention on schoolchildren’s health and nutrition outcomes. There are, however, several studies that have demonstrated the benefit of either breakfast consumption or nutrition education on various outcomes alone. On breakfast effect, a randomised controlled trial among adolescents aged 11–13 years found that 100 % of participants in the breakfast group reached the highest level of the Paired Associates Learning, a measurement of short-term memory capacity(Reference Arndt and Seel10), compared with only 41·7 % in the no-breakfast group. The study demonstrated that breakfast consumption significantly reduced total errors and improved reaction time and sustained attention(Reference Adolphus, Lawton and Champ11). Meanwhile, the impact on nutrition education in Nepal through a quasi-experimental study demonstrated that a 12-week school-based nutrition education intervention significantly improved students’ nutrition knowledge (mean difference = 1·8 points, 95 % CI: 1·1, 2·5) and eating behaviours, including reductions in emotional and uncontrolled eating and increases in cognitive restraint(Reference Raut, Kc and Singh12). Another cluster-randomised controlled trial among Indonesian adolescents in Makassar demonstrated that a multi-strategy school-based nutrition education programme led to significant improvements in breakfast nutrient intake, with mean energy intake increasing by 73 kcal post-intervention and remaining 49 kcal higher at follow-up (P = 0·02), and protein intake rising by more than 2 g at endline (P < 0·05)(Reference Indriasari, Nadjamuddin and Arsyad13). While these studies show promising impact of either quality breakfast provision or nutrition education alone, there is lack of evidence on the impact of both interventions coupled as a comprehensive programme.
In Indonesia, a country with a rich cultural and dietary landscape, the importance of healthy breakfast habits has yet to be widely recognised or adopted among schoolchildren(Reference Angkasa, Pratiwi and Jus’at14,Reference Putri, Briawan and Baliwati15) . The nutritional landscape shows the dual challenge of nutrient deficiencies coexisting with overconsumption of certain unhealthy foods(Reference Rachmi, Agho and Li16–18). Many children face inadequate intake of essential nutrients like n-3 and n-6 fatty acids, which are crucial for brain health, while simultaneously having diets high in sugars and saturated fats(Reference Neufingerl, Djuwita and Otten-Hofman19). In addition, Indonesia does not have a mandatory nutrition education programme imbedded in the school curriculum(Reference Rachman, Vipta and Mauludyani20). Hence, exacerbating the problem as these nutrient inadequacies may not be addressed systematically by schools, as a potential source of information and supportive food environment. This imbalance points to the need for targeted nutrition interventions and educational programmes that can support sustained behaviour change.
While schools have been known to be an ideal setting to establish healthy eating patterns, the involvement of parents, mothers, or caregivers in school-based programmes has also been well documented to enhance sustained impacts(Reference Göbel, Ercan and Bayram21–Reference Kovács, Kovács and Bacskai23). Schoolchildren primarily spend most of their time at home in which parents or caregivers have a strong influence in determining dietary habits of children, built from cultural and inter-generational food habits, knowledge, and experiences of their own(Reference Albardan and Platat24,Reference Lopes, Vilella and Moreira25) . A study assessing caregiver’s roles in the effectiveness of two Dutch nutrition education programmes demonstrated that there was a significant positive association between caregiver’s health promotion behaviours with children’s healthy eating behaviours(Reference Verdonschot, de Vet and van Seeters26). A similar integrated school-based nutrition programme involving mothers in the Philippines also yielded increments in mothers Knowledge, Attitude, Practice (KAP) that led to significantly improved children’s nutritional status(Reference Angeles-Agdeppa, Monville-Oro and Gonsalves27). Highlighting the importance of considering parental or caregiver involvement to leverage impact of school-based nutrition programmes.
As many countries without an established school food programme also experience, a lack of financial resources may constrain the implementation of such interventions(Reference Colombo, Elinder and Patterson28,Reference Almutairi, Burns and Portsmouth29) . While it is a mutual understanding that ideally, sustainable long-term interventions or programmes will result in the most desired outcomes(Reference Guan, He and Shi30,Reference Fang and Zhu31) , little is known about short-term and longer-term impact of a 4-week coupled breakfast and nutrition education intervention. Studies on the impact of a shorter-term intervention could be beneficial for programme implementers with limited resources to develop cost-effective interventions.
In response to these issues, the GESIT (Gerakan Sarapan Bergizi Berprestasi or Nutritious Breakfast for Excellence Campaign) programme, was designed and aimed at introducing and promoting balanced breakfast habits rich in n-3 and n-6 fatty acids among Indonesian schoolchildren. This programme delivered nutritional education for students, and parents, focusing on the benefits of a nutritious breakfast for academic performance, physical activity, and health, alongside practical advice on preparing balanced breakfasts. The objective of this study was to evaluate the short-term and longer-term impact of GESIT on nutrition knowledge, nutrient intake from breakfast consumption, and short-term memory among elementary schoolchildren aged 9–11 years in Bogor, Indonesia.
Materials and methods
Study design
A pre-post quasi-experimental design was conducted to evaluate the short-term impact of a 4-week nutrition intervention on elementary schoolchildren, including their mothers or main caregivers. The first group received both breakfast intervention and nutrition education (BreakfastEdu group), the second group received nutrition education only (Education group), and the third group served as a control which did not receive any intervention (Control group). Longer-term or follow-up evaluation was conducted 3 months after the intervention. This article focuses on children’s outcomes, while intervention effect towards mothers or main caregivers will be published elsewhere.
The duration of the intervention was adopted and modified from Schmitt et al. (Reference Schmitt, Bryant and Korucu32) to accommodate the time constraints of the study, align with the school academic calendar and ensure completion prior to examination and holiday periods. Participants received an equivalent total exposure of 12 sessions, but with increased intensity (three sessions per week).
This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects were approved by the Health Research Ethics Committee of the National Research and Innovation Agency of Indonesia (No: 033/KE.03/SK/02/2024). Written informed consent was obtained from parents of all subjects, and written informed assent was obtained from the students after providing them with clear and detailed information about the study’s objectives, methods and potential risks or benefits.
Sample size calculation
Sample size calculations were done using Cochran’s sample size formula for pre-post intervention studies, based on the four main outcomes of the intervention: nutrition knowledge(Reference Nuryanto, Pramono and Puruhita33), n-3 and n-6 intake(Reference MacIntosh, Ramsden and Honvoh34), nutrient intake(Reference Dehdari, Rahimi and Aryaeian35) and breakfast intake(Reference Au, Whaley and Rosen36). The highest minimum sample size needed to capture the four main outcomes of this study used an effect size of 0·40 for n-3 intake with 90 % power was sixty-four subjects for each of the three groups: BreakfastEdu, Education and Control groups. To account for the risk of dropout, the number in each group was increased by 20 % to eighty students, for a total of 240 students. Total sampling was used to include all children in grade 4 and 5 of the selected schools. Because the number of participating schools (six schools) was fixed due to limited resources and the intervention was delivered at the school level, clustering was not incorporated into the a priori sample size calculation. Instead, potential clustering effects were addressed analytically.
Recruitment of schools and participants
The target population of the study were grade 4–5 elementary schoolchildren (age 9–11 years). A publicly available online school database was used to identify all public elementary schools in Bogor(37,38) . Schools were first screened for eligibility to determine if they met the following criteria: (1) having approximately 40–45 students enrolled across grades 4 and 5; and (2) located in the Bogor region representing urban and rural areas to ensure that the data collected are representative of the region and socio-economic conditions. From the pool of eligible schools, the authors purposively selected potential sites based on accessibility, ensuring that both urban and rural intervention schools could be reached by vehicle within 1 h to facilitate operational logistics. Identified schools were then contacted and visited for the second screening, which includes the following criteria: (3) have not received a nutrition intervention or programme before and (4) willing to collaborate throughout the entire duration of the study (May–September 2024). Out of nine approached schools, six schools met the inclusion criteria and were systematically allocated to study arms according to the order of site visits. Eligible schools were enrolled consecutively and assigned sequentially to the BreakfastEdu, Education and Control groups, ensuring that each arm included one urban and one rural school. Detailed information on sampling procedure of schools and participants is described in Figure 1.
Sampling procedure of schools and participants in urban and rural areas. The figure illustrates the selection process from eligible schools to those included at baseline, endline and follow-up data collections. Note: At the baseline, all participating students were included. At the baseline-endline phase, the reduced number of participants reflects dropouts due to student absence during either baseline or endline data collection. The further decrease at the baseline-endline-follow-up stage indicates additional attrition resulting from student absence during the follow-up phase. Superscript a denotes the number of participants matched across the baseline-endline-follow-up for nutrition knowledge and related data, superscript b refers to participants matched for macro- and micro- nutrient intake data and superscript c refers to participants matched for n-3 and n-6 fatty acids data. The difference in the number of matched cases between datasets occurred because the assessments were conducted on multiple days, and some students were absent on one of those days.
All schoolchildren in grades 4 and 5 who were registered and active students in the selected schools were included in the study. Students in grade 6 were not included due to preparations for their final elementary school evaluations and difficulties for follow-up. Participants without parental consent, identified as ill during data collection or having incomplete data were excluded from the data analysis.
Intervention
Balanced breakfast was provided for the duration of the intervention (20 school days). Based on the experiences of school meal composition in several countries as reported by FAO, the common composition provided was between 20 and 35 % of energy(39). Our study provided the higher end of the range for breakfast provision, which was designed to meet 25–30 % of the children’s daily nutrient requirements based on the Indonesian RDA for that age group(40) (see online Supplemental Figure 1), considering the poor quality of breakfast of Indonesian schoolchildren found in a previous study(Reference Harahap, Widodo and Sandaja8). Additionally, the breakfast contained 10 g of fat spread with n-3 and n-6. The 20-d menu was co-developed by the researchers and a nutritionist employed by the catering service to ensure that nutrition standards are met. Meal planning was done using a formulated Excel document linked to the Indonesian Food Composition Table(41) to calculate nutrient content. Professional catering services, trained by researchers, prepared and delivered the breakfasts every morning for the duration of the intervention period. Breakfast distribution was accompanied by two research assistants per school to supervise the breakfast delivery and consumption. Meals were individually packed in bento boxes provided by the catering service. The visual Comstock(Reference Comstock, St Pierre and Mackiernan42) method was used to evaluate how much of the breakfast was consumed by each student every intervention day. The results were evaluated together with student testimonials to adjust the menu weekly across all schools. A detailed menu is described in online Supplemental Table 1.
Nutrition education was provided for schoolchildren by homeroom teachers for twelve sessions (three sessions per week). Prior to the intervention, teachers received a 1-d training on nutrition education and media on the following topics: (1) three functions of food, (2) balanced nutrition, (3) MyPlate, (4) benefits of a healthy breakfast, (5) fats in food, (6) n fats, (7) balanced nutrition pillars, (8) nutritious snacks and (9) clean and healthy lifestyle. Topics 4, 5 and 6 were repeated to ensure understanding of the vital topics of the programme. Educational materials included flipcharts, posters, a teacher manual and student games (flashcards and a MyPlate puzzle) to reinforce learning. Sessions lasted 15–20 min before class, led by the homeroom teacher and supervised by two research assistants to ensure quality, track progress and evaluate student engagement.
Additionally, mothers or primary caregivers also received nutrition education through WhatsApp, along with a cooking demonstration on preparing balanced breakfast meals. WhatsApp class sessions were conducted every week for four sessions which lasted approximately 1 h. Classes were led by researchers and facilitated by research assistants. The following topics were delivered: (1) importance of healthy breakfast, (2) balanced breakfast and MyPlate, (3) benefit of n-3 and n-6 fatty acids and (4) healthy snacks and clean and healthy lifestyle. Cooking demonstration was conducted once at the beginning of the programme and led by professional chefs who demonstrated how to cook three easy-to-prepare breakfast meals. A WhatsApp Bot was also developed as an optional source of nutrition information for mothers. Detailed information on the nutrition education packages is shown in online Supplemental Figure 2.
Study variables
Information on student’s age, sex, daily allowances, presence of chronic diseases, parent’s education level, parent’s occupation, family income, family type and number of family members was collected by trained enumerators through a structured questionnaire addressed to students and verified by mothers or main caregivers.
Dietary intake was assessed using two methods, namely a repeated 24-h recall and semi-quantitative FFQ (SQFFQ). To assess breakfast intake of macronutrients and selected micronutrients, a 24-h recall was administered to all participants accompanied by their mother or main caregiver to capture participants’ diet at home. Food consumed after waking in the morning until 09.00 was considered as breakfast. A 24-h dietary recall was selected to assess breakfast intake because it offers open-ended flexibility while objectively capturing the timing of consumption. This method also minimises response bias compared with direct self-reports of breakfast habits. A trained enumerator conducted the recall using the four-step multiple-pass technique(Reference Gibson43). Portion size was estimated using a photo picture food book(44), containing photos of commonly consumed foods in different portion sizes using household food measures, developed by the Ministry of Health, Republic of Indonesia, for the previous Total Diet Study(Reference Saputri and Widyastuti45). A repeated food recall was collected for 25 % of the population for validation. This approach was used to resemble participants’ usual intake under time and resource constraints(Reference Gibson43,Reference Deitchler, Arimond and Carriquiry46) . The study focused on certain nutrients, such as energy, protein, total fat, carbohydrate, vitamins A, B1, B2, C and D, Ca, Fe, Zn and potassium due to their important roles in supporting growth and development of schoolchildren(Reference Heymsfield and Shapses47–Reference Hong49). Specific focus was also given to measure habitual n-3 and n-6 fatty acids (α-linolenic acid (ALA), DHA, EPA, PUFA and linoleic acid (LA)) which was one of the routine components of the breakfast intervention. These fatty acid intakes were assessed using a validated SQFFQ capturing usual intake during the past month. The use of an SQFFQ in our study was deemed appropriate as it aims to capture the habitual intake of specific nutrients over a period of time, however with less time and resources(Reference Gibson43). The time frame of 1 month was to capture the duration of the intervention. The food list of the SQFFQ was first developed through a desk review of foods containing high n-3 and n-6 fatty acids commonly consumed by Indonesian schoolchildren. The SQFFQ was then pre-tested and validated in children grade 4–5 in one of the unselected intervention schools. The Spearman correlation test demonstrated positive associations between SQFFQ and 3-d 24-h recalls (ALA r = 0·868; EPA r = 0·895; DHA r = 0·775; PUFA r = 0·306; LA r = 0·950). Bland–Altman analysis showed that most differences fell within the limits of agreement, indicating good consistency between the two methods. Reliability testing using Cronbach’s α yielded a coefficient of 0·683 across the five fatty acid groups.
Nutrition knowledge was assessed using a validated structured questionnaire (Cronbach’s α of 0·748) consisting of eighteen multiple-choice questions related to breakfast habits, food source of nutrients, nutrients function and the benefits of a balanced breakfast. Each correct answer was given a score of 1, while incorrect or ‘don’t know’ responses were scored 0. The total knowledge score was obtained by summing the number of correct answers, dividing by the total number of questions (18) and multiplying by 100. The resulting value was expressed as a knowledge score ranging from 0 to 100 points, with higher scores indicating better nutrition knowledge. The scores were then categorised into three levels according to Khomsan(Reference Khomsan50), which were low (<60), moderate (60–80) and high (>80). Breakfast habit assessed the frequency of breakfast on weekdays and weekends and also the type of breakfast consumed.
Short-term memory was measured using the Digit Span subtest from the Wechsler Intelligence Scale for Children-Revised (WISC-R) which included the Digit Span Forward (DSF) and Digit Span Backward (DSB)(Reference Petrosko51). DSF assesses short-term auditory memory by presenting a sequence of numbers aloud at 1-s intervals, which the child is then asked to repeat in the same order. In contrast, DSB evaluates the child’s capacity to mentally reorganise verbal information held in temporary memory, requiring them to repeat the number sequence in reverse order. For both tasks, the number sequences become progressively longer as the child continues to respond accurately(52). The WISC-R Digit Span subtest was selected for its validated use as a standalone measure of short-term auditory memory and attention(Reference Franzen and Franzen53). Unlike newer versions that embed Digit Span within broader cognitive indices and introduce executive demands, such as in WISC-IV(Reference Grizzle, Goldstein and Naglieri54) and WISC-V(Reference Olivier, Mahone, Jacobson, Kreutzer, DeLuca and Caplan55), WISC-R allows for focused, efficient assessment in school-based settings. Its historical use in nutrition and cognitive research further supports its appropriateness for this intervention(52,Reference Franzen and Franzen53,Reference Beauchamp and Samuels56) . Participants were tested individually by a trained enumerator in a closed classroom within the first hour of the school schedule (07.00–08.00), which is 1 to 2 h post-breakfast consumption either at home or provided at school. All enumerators underwent standardised training prior to data collection, including detailed instruction on test administration procedures, scripted instructions, scoring rules and mock assessments to ensure consistency. The results are interpreted at the group level only, to assess relative differences between study arms, and not as clinical cognitive outcomes, as has been done in previous population-based surveys(57).
Data analysis
Macronutrient and micronutrient data were analysed using NutriSurvey 2007 software updated with Indonesia’s most recent Food Composition Table(58). Due to the absence of n-3 and n-6 fatty acids in the Indonesian Food Composition Database, the authors developed a new database for these nutrients, based on foods listed in the SQFFQ, derived from the Fatty Acid Composition of Indonesian Food(Reference Sulaeman and Setiawan59), Malaysian Food Composition Database(60) and Singapore Nutrient Composition Database(61). Composite foods were broken down into common recipes derived from the average of at least three recipes found through online search.
Prior to analysis, data were cleaned and matched for complete datasets over the three time frames (baseline, endline and follow-up) for the main variables. Statistical analyses were carried out using SPSS (Statistical Package for the Social Science, version 29) for Windows. Differences between the three groups were analysed using the Kruskal–Wallis test for non-normally distributed data, while pairwise differences between two groups were assessed using the Mann–Whitney test. For categorical variables, the χ 2 test was applied to test for differences between the three groups. Additionally, within-group analyses were carried out using the paired t test (for normally distributed variables) and Wilcoxon signed-rank test (for non-normally distributed variables). ANCOVA was also performed to examine whether baseline differences influenced the intervention outcomes.
To evaluate the impact of school-level clustering on study outcomes, post hoc mixed-effects sensitivity analyses were conducted using linear mixed-effects models with a random intercept for school. For each outcome and time contrast (Δ1: baseline to endline; Δ2: baseline to follow-up; Δ3: endline to follow-up), change scores were modelled as dependent variables with the intervention group as a fixed effect. Intracluster correlation coefficients (ICC) were estimated from variance components. These analyses were conducted to assess robustness and interpret precision, not as primary tests of intervention effectiveness.
Results
General and socioeconomic characteristics
Of the 249 children enrolled in the programme, 212 children were included in the final analysis of the study. Exclusion of students was primarily due to incomplete data from one of the three time frames (baseline, endline and follow-up). Different numbers of participants (n) are shown across the findings due to different timing of data collection. All children included in each respective analysis had complete data for the three measurement points. A detailed description of the composition of the participants across the three time frames is described in Figure 1.
The study involved children aged 9–12 years, with a balanced sex distribution across the three groups. The average age of children was 11·3 years, with the Education and Control groups tending to be younger (P < 0·001). Mother’s education level was found to be higher in the BreakfastEdu group, having the highest percentage of mothers graduating from higher education (19·4 %) among the three groups (P = 0·003). Meanwhile, the Education group had the most percentage of mothers who do not work (66·2 %) (P = 0·007). In terms of father’s occupation, the Control group had higher percentage of labourer/farmer (41·3 %) (P = 0·030). In addition, mother’s age was significantly older in the BreakfastEdu group compared with other groups (P = 0·029). No differences were found across groups for daily allowance for food, father’s education level, family type, number of family members, household income, breakfast frequency and type of breakfast (Table 1). In our study population, it was found that the breakfast frequency at baseline was high in all groups, with an average of six times per week. There were also no significant differences in school setting among groups. Although baseline differences in participant’s age, mother’s education level, occupation, and age, and father’s occupation were observed across the three groups, ANCOVA indicated that these factors did not significantly affect the intervention outcomes (all P > 0·05).
General and socio-economic characteristics of participants
Different superscript letters indicate difference statistical significance between groups based on Mann–Whitney test.
*Difference statistical significance between groups based on χ 2.
Impact on nutrition knowledge
Table 2 presents the effects of the different interventions on students’ nutrition knowledge scores across three groups. The BreakfastEdu and Education groups showed significant improvements, with mean scores rising from 61·1 at baseline to 72·2 at endline (P < 0·001). However, a notable decline occurred at follow-up, resulting in the same mean scores as at baseline. Meanwhile, the Control group showed significant smaller improvement between baseline and endline, from 61·1 at baseline to 63·9 at endline, with a slight decline occurring between endline and follow-up (–2·8 points). Despite larger changes observed in the BreakfastEdu and Education groups during the intervention, no significant difference was observed between the intervention groups and Control group regarding their breakfast knowledge before and after the intervention (at endline and follow-up).
Student’s nutrition knowledge score across groups
The table presents the median (25th–75th percentiles) knowledge scores for each study group across three phases: baseline, endline and follow-up.
Within-group analysis was measured using Wilcoxon signed ranked test. Significance in changes between baseline and endline is indicated using *, between endline and follow-up is indicated using ** and between baseline and follow-up is indicated using † (P-value < 0·05).
The differences in changes between the Control, Education and BreakfastEdu groups were analysed using the Kruskal–Wallis test. Δ1 represents the changes from baseline to endline, Δ2 represents the changes from baseline to follow-up, while Δ3 represents the changes from endline to follow-up.
Although non-parametric analyses indicated differences in knowledge change across study groups, mixed-effects sensitivity analyses revealed substantial clustering at the school level (ICC ranging from 0·30 to 0·39), indicating that changes in knowledge were strongly influenced by school context. After accounting for this clustering, group-level effects were attenuated (online Supplemental Table 2).
Student’s nutrient intake
Table 3 presents changes in nutrient intake from breakfast for the three study groups. The analysis shows that the BreakfastEdu group exhibited the most significant improvements across various nutritional parameters compared with the other groups from baseline to endline. In the BreakfastEdu group, energy intake from breakfast significantly increased by 138 kcal from baseline to endline (P < 0·001). While a slight decrease was observed at follow-up (–22 kcal), the difference from baseline remained significant (P < 0·001). Conversely, the Education and Control groups showed insignificant increases in energy intake within the groups. These short-term changes (from baseline to endline) in the BreakfastEdu group were significantly higher compared with the other groups. A similar trend was also found for protein intake, where significant increments from baseline to endline were only found in the BreakfastEdu group (2·7 g) and remained higher than baseline at follow-up. However, these changes were not significantly different compared with the other groups. For total fat intake, the BreakfastEdu group recorded a significant 7 g increase from baseline to follow-up (P < 0·001). While there was a decrease at follow-up (14·6 g), the improvement from baseline was still significant (P < 0·001). These changes were significantly different compared with other groups from baseline-endline (Δ1) and endline-follow-up (Δ3). For carbohydrate intake, the BreakfastEdu group did not show short-term improvements; however, it showed a significant increase from baseline (36·2 g) to follow-up (45·8 g; P = 0·005), as well as from endline (35·5 g) to follow-up (45·8 g; P < 0·05), whereas changes in the other groups were not statistically significant. The increment from endline to follow-up was significantly different compared with other groups (P = 0·021).
Changes in breakfast nutrient intake across groups
The table presents the median (25th–75th percentiles) nutrient intakes for each study group across three phases: baseline, endline and follow-up.
Within-group analysis was measured using Wilcoxon signed ranked test. Significance in changes between baseline and endline is indicated using *, between endline and follow-up is indicated using ** and between baseline and follow-up is indicated using † (P-value < 0·05).
The differences in changes with respect to baseline between the Control, Education Only and BreakfastEdu groups were analysed using the Kruskal–Wallis test. Δ1 represents the changes from baseline to endline, Δ2 represents the changes from baseline to follow-up, while Δ3 represents the change from endline to follow-up.
aA significant difference (P-value 0·05) between the groups.
The breakfast intervention resulted in significant increases in micronutrient intake such as vitamins A, B1, B2, C and D, Ca, Fe, Zn, and potassium from baseline to endline in the BreakfastEdu group. However, these values declined at follow-up compared with endline (Δ3, P < 0·05), and the differences between baseline and follow-up (Δ2) were not significant. Aside from the BreakfastEdu group, potassium intake was also observed to have a significant increment from endline to follow-up in the Control group.
The estimated total daily intake of essential fatty acids is described in Table 4. In the BreakfastEdu group, significant improvements from baseline to endline were shown for PUFA (+1·4 g; P = 0·012), n-3 ALA (+0·1 g; P < 0·001) and n-6 LA (+0·9 g; P = 0·015). Meanwhile, in the Education group, a significant increase was observed in n-3 EPA intake from baseline to endline. In the Control group, significant increases were found in ALA intake (from baseline to endline and baseline to follow-up) as well as in EPA intake (from baseline to follow-up). Although significant within-group increases in PUFA, ALA and LA were observed in the BreakfastEdu group, these improvements were not significantly different when compared across groups. Instead, between-group differences were only found for EPA, between baseline and follow-up.
Changes in total daily essential fatty acids intake
The table presents the median (25th–75th percentiles) nutrient intakes for each study group across three phases: baseline, endline and follow-up.
Within-group analysis was measured using Wilcoxon signed ranked test. Significance in changes between baseline and endline is indicated using *, between endline and follow-up is indicated using ** and between baseline and follow-up is indicated using † (P-value < 0·05).
The differences in changes with respect to baseline between the Control, Education Only and BreakfastEdu groups were analysed using the Kruskal–Wallis test. Δ1 represents the changes from baseline to endline, Δ2 represents the changes from baseline to follow-up, while Δ3 represents the changes from endline to follow-up.
aA significant difference (P-value 0·05) between the groups.
After accounting for school clustering through the sensitivity analysis described in online Supplemental Table 2, changes in breakfast nutrient intake indicate minimal to moderate clustering at the school level (0·05 < ICC < 0·001) for most macro- and micro-nutrients. Vitamin C, however, showed a large clustering effect with an ICC of 0·18 for Δ1 and 0·11 for Δ3. Effects were in the same direction as the primary analyses, although statistical significance was attenuated for some outcomes. Additionally, the clustering effect of daily essential fatty acid intakes was much larger than other nutrients (ICC about 0·39), indicating high school-level clustering.
Cognitive performance (short-term memory)
Table 5 depicts the distribution of scores for the short-term memory test across three groups during baseline, endline and follow-up. The BreakfastEdu group showed a significant increase in scores from baseline to endline (P < 0·05). These scores were sustained during follow-up, while the other two groups showed a non-statistically significant slight reduction. There were no significant differences observed in the changes between groups.
Changes in short-term memory scores across groups
The table presents median (25th–75th percentiles) short-term memory scores for each study group across three phases: baseline, endline and follow-up.
Within-group analysis was measured using Wilcoxon signed ranked test. Significance in changes between baseline and endline is indicated using *, between endline and follow-up is indicated using ** and between baseline and follow-up is indicated using † (P-value < 0·05).
The differences in changes with respect to baseline between the Control, Education Only and BreakfastEdu groups were analysed using the Kruskal–Wallis test.
For short-term memory, mixed-effects sensitivity analyses indicated substantial school-level clustering (ICC of 0·28). After accounting for this clustering, group-level fixed effects were estimated imprecisely and did not reach statistical significance, although the direction of effects was broadly consistent with non-parametric analyses (online Supplemental Table 2).
Discussion
This study identified the short-term (pre- and post-) and longer-term (3 months post-intervention) impact of a comprehensive school breakfast programme which provided both breakfast and nutrition education for children and nutrition education for mothers, aimed at assessing nutrition knowledge, nutrient intake and short-term memory. To the best of our knowledge, our study is the first in Indonesia to emphasise the impact of such intervention when comparing students who received the full programme package (BreakfastEdu), the partial programme (Education group) or none (Control), thus providing a better understanding of the type of intervention underlining the most effective results. In addition, the inclusion of rural and urban schools provides a more comprehensive representation of the population.
Our study shows that a structured nutrition education programme delivered routinely 2–3 times a week by trained teachers was able to significantly increase the nutrition knowledge of students in both the BreakfastEdu and Education groups from baseline to endline. However, the observed increases were not significantly different between control and intervention groups at endline and after follow-up. Smaller increases were also found in the Control group. This is assumed to be caused by a retesting bias where participants have already experienced the same set of questions before(Reference Polit62). Indicating that nutrition education intervention may result in positive short-term increments in knowledge scores despite not receiving a balanced nutritious breakfast. Our initial hypothesis was that the BreakfastEdu group would gain more increment in knowledge scores, as compared to the Education Only group. This was based on the understanding that food provision provides hands-on experiences and exposure that would enhance internalisation of nutrition education provided, as demonstrated in previous studies(Reference Bisset, Potvin and Daniel63,Reference Colley, Myer and Seabrook64) . However, this effect was not observed in our study. A possible explanation is the variation in the delivery style of nutrition education by trained teachers in the intervention schools. Although all educators received the same training and were supervised throughout the sessions, differences in instructional approach may have led to improved comprehension among students in the Education Only group, thereby attenuating the relative impact of the BreakfastEdu intervention.
Our short-term findings yielded comparable results with other studies that applied a longer duration of nutrition education activities. For example, a 12-week nutrition education intervention for adolescents in Nepal increased nutrition knowledge scores by 1·8(Reference Raut, Kc and Singh12). Another study with a shorter duration of intervention (6 weeks) also resulted in significantly higher scores in the group provided with nutrition education(Reference Schmitt, Bryant and Korucu32). Despite the shorter duration of our study, positive outcomes were observed. This may be explained by active parental involvement during our study, which is supported by Murimi et al. (Reference Murimi, Moyeda-Carabaza and Nguyen65) in a systematic review. In the review, interventions among preschool children are encouraged to be structured as short but frequent sessions to accommodate their limited attention span, incorporate hands-on activities and engage parents, ideally through face-to-face interaction. This principle underscores that intervention effectiveness is not solely determined by duration but also by the frequency, intensity and quality of participant and parental engagement. In our case, the inclusion of mother’s nutritional education and more frequent sessions for students may have influenced positively towards our findings.
While our study showed promising short-term increases in the intervention groups, they were not found to be significantly different from the Control groups. Moreover, the scores decreased during follow-up, returning to their original state, suggesting that the intervention may only have had a short-term impact on nutrition knowledge. In previous similar studies, Lier van I et al. and Hahnraths et al. (Reference Hahnraths, Jansen and Winkens66,Reference van Lier, van Mil and Havermans67) also did not find a sustained effect in knowledge scores 3 months after a 3-week school-based experiential nutrition education intervention, indicating that without post-intervention reinforcement, knowledge gains are likely to fade, especially if students are not regularly exposed to nutrition topics in their daily activities, as demonstrated in other studies(Reference Schmitt, Bryant and Korucu32,Reference Murimi, Moyeda-Carabaza and Nguyen65) .
In the systematic review conducted by Murimi et al. (Reference Murimi, Moyeda-Carabaza and Nguyen65), the ideal duration for effective results of nutrition education was found to be at least 6 months. These findings highlight the need for ongoing educational support or curriculum integration to maintain and strengthen long-term knowledge retention, as also mentioned by Chatterjee et. al.(Reference Chatterjee and Nirgude68) in a different systematic review of school-based nutrition interventions. A longer duration of the intervention, extending the duration and the frequency of nutrition topics delivered, may result in a more significant and longer impact.
In terms of breakfast intake, this study underscores the pivotal role of targeted school-based interventions in improving students’ dietary intake. The significant improvements observed in the BreakfastEdu group for energy, macronutrient intake and intake of several micronutrients during the intervention phase highlight the efficacy of direct dietary strategies. Unlike other studies that highlighted poor breakfast habits among schoolchildren in Indonesia(Reference Kautsar7,Reference Harahap, Widodo and Sandaja8) , our study showed that breakfast habit (in terms of frequency) was well established in students at all selected schools even before the intervention. However, the nutrient quality was the main issue in our study population, with the most common breakfast composition containing only staples and one protein-source side dish. Fruits and vegetables are rarely consumed during breakfast. In populations with higher prevalence of skipping breakfast, the impact of a school meal programme, such as the GESIT programme, may yield more significant results. Nevertheless, our study showed, even with prior well-established breakfast frequency, significant improvements were shown due to the provision of structured meals with enhanced dietary quality. This aligns with previous findings that structured school meal programmes, particularly those incorporating educational elements, have the potential to enhance dietary quality and address nutrient gaps in children’s diets(Reference Micha, Karageorgou and Bakogianni69). For example, interventions that focus on fruit and vegetable consumption, similar to the one described, have shown consistent improvements in intake and overall diet quality, even when the effect sizes are modest(Reference Kristjansdottir and Thorsdottir70).
The significant improvements in PUFA, n-3 (ALA) and n-6 (LA) in the BreakfastEdu group when compared with the other two groups are particularly noteworthy given their association with cognitive and neural development in children. Djuricic and Calder(Reference Djuricic and Calder9) detailed the critical role of ALA and cognitive performance, highlighting the importance of these nutrients in school performance. Research indicates that these fatty acids play a crucial role in brain function, improving memory, attention and cognition, all of which support learning and knowledge retention(Reference van der Wurff, Meyer and de Groot71,Reference Lundqvist, Vogel and Levin72) .
Despite the initial success, the modest decline in nutrient intake during the follow-up phase in the BreakfastEdu group underscores the limitations of short-term interventions. As Scaglioni et al. (Reference Scaglioni, De Cosmi and Ciappolino73) noted, environmental and familial reinforcement is critical for embedding positive dietary behaviours into children’s daily routines. The observed reversion may reflect the broader challenge of sustaining dietary improvements beyond the controlled intervention environment, a concern also echoed in a literature review by Hartline-Grafton and Levin(Reference Hartline-Grafton and Levin74), who found limited long-term impacts of breakfast programmes on academic and dietary outcomes without sustained reinforcement.
The current study also found that breakfast interventions had minimal impact on students’ cognitive performance. The results show an increase in cognitive performance after the intervention only in the BreakfastEdu group. Compared with the Education and Control groups, it did not show to be significantly different. The statistically significant one-point increment in the total Digit Span test in the BreakfastEdu group raises the question on whether the change is substantially meaningful. Although short-term memory was not included as a primary outcome in the sample size calculation due to a lack of intervention studies using similar tests, the Digit Span test results were notably comparable to those of the US population in 1979, around when the WISC-R was first established. The study examined cross-national differences in Digit Span scores between American and Canadian children and reported a statistically significant mean difference of 1·73 points, underscoring the test’s sensitivity to population-level cognitive variation(Reference Beauchamp and Samuels56). The Digit Span test has also been used in the US National Longitudinal Survey of Youth from 1994 until 2010, with the most recent report showing an average score of 9·8 across age groups 7–11 years(75), demonstrating the potential biological significance of our findings.
Newer versions of the complete WISC may demonstrate more comprehensive impact of interventions towards cognitive performance, especially those recently validated in specific population settings(Reference Yudiana, Hendriks and Suwartono76). However, despite limitations associated with using an older version of the Digit Span test, the authors argue that newer versions of the WISC Digit Span subtest may not be suitable for use as standalone measures aimed to assess the effect of a short-term intervention. While our study showed minimum impact of the BreakfastEdu intervention towards short-term memory, previous studies have demonstrated a direct effect of breakfast on cognitive assessments. A breakfast trial providing commercially ready-to-eat cereal and milk resulted in significant improvements in cognitive performance compared with no breakfast, 70- and 215-minute post-intervention(Reference Adolphus, Hoyland and Walton77).
It is noted that longer intervention durations might enhance cognitive effects. A previous study assessing the effect of a daily breakfast intervention containing adzuki beans over an 8-month period among primary students in Sudan demonstrated substantial increments in cognitive assessments, with the percentage of students scoring average or above increasing from 30·6 % to 67·7 %. Academic performance also showed significant improvements(Reference Ahmed, Ali and Nour78). A longitudinal study in China also observed that regular breakfast consumption beginning from 6 years of age showed significant increases in IQ scores at 12 years of age(Reference Liu, Wu and Um79). In addition, several systematic reviews also supported regular breakfast consumption improving academic performance and cognitive functions, such as attention and memory(Reference Adolphus, Lawton and Champ11,Reference Muth and Park80) . Meanwhile, deficiencies in specific nutrients like n-3 fatty acids can negatively affect cognition and contribute to conditions like depression(Reference Spencer, Korosi and Layé81). Hence, longer duration of breakfast provision, containing essential fatty acids, may potentially provide more significant outcomes towards cognitive performance.
When taking into consideration the post hoc sensitivity analysis, the clustering effect was shown in variance. Nutrition knowledge and short-term memory demonstrated large ICC, indicating that 27–39 % of the results were attributed to school clustering. A possible reason for this is that both variables are highly influenced by teachers’ instructional and pedagogic strategies when delivering nutrition education and building students’ competencies. A systematic review on the effect of teacher-delivered nutrition programmes supported this argument, in which teachers have critical roles in influencing students’ outcomes(Reference Cotton, Dudley and Peralta82). Despite several strategies that have been conducted by the research team to eliminate this confounder, such as applying the same training, materials and educational tools for all intervention schools, as well as providing a supervisor during all nutrition education sessions, teachers’ individual pedagogic abilities cannot be disentangled from the results of the intervention. The high school-level clustering in this study demonstrates the existence of a distinct gap in the education systems where teachers may not have the same teaching quality standard and therefore must be assessed prior to programme trainings to provide better strategies that meet their needs.
The small school-level clustering effect for breakfast nutrient intake was expected as macro- and micronutrients are highly varied at the individual level. A recent study by Singh et al. (Reference Singh, Verest and Salathé83) demonstrated that individual variability is common in dietary data, especially in data taken from 1 to 2 d of dietary assessments. Vitamin C, however, was an exception in this study. Essential fatty acids were also found to be largely influenced by school clustering. An explanation for this phenomenon is the fact that both vitamin C and essential fatty acids were critical components of the breakfast intervention in this study. Due to the study design, in which fruits (main source of vitamin C) and fat spread containing n-3 and n-6 fatty acids were provided in the schools receiving the breakfast intervention, such clustering effect cannot be separated from the results. Within-group analysis for vitamin C, ALA and LA intake showed significant increments from baseline to endline in the BreakfastEdu group, demonstrating this notion. Moreover, the availability of fruits and food rich in essential fatty acids may highly differ between urban and rural schools, which may also lead to the significant school-level clustering in these outcomes. For vitamin C, this is particularly relevant in Indonesia, where 97·6 % of schoolchildren and adolescents lack fruits and vegetables (consuming less than 5 servings a day)(84).
Our study demonstrates several notable strengths, one of which provides a comprehensive understanding of both the short-term and longer-term impacts of a school breakfast programme on nutritional knowledge, breakfast intake and short-term memory that is important for shaping effective nutrition intervention programmes and policies. Additionally, the inclusion of both rural and urban settings ensures that the findings are broadly applicable and inclusive, addressing the diverse needs and circumstances of students across different environments. These strengths enhance the relevance and applicability of the recommendations, supporting the development of equitable and impactful nutrition policies in schools. The findings of our study also support the potential short-term benefit of the recently launched ‘Free Nutritious Meal’ or ‘Makan Bergizi Gratis’ programme in Indonesia which covers the provision of one free meal per day (breakfast or lunch) as a blanket approach for all schoolchildren, under-five children, pregnant and lactating women. However, it is highly suggested that nutrition education should be embedded in the school curriculum to address issues of compliance and sustainability.
This study, while providing valuable insights, has several inherent limitations. First, although the number of subjects collected exceeded the minimum sample size for most variables, matching datasets for all three time points of data collection resulted in a relatively small sample size, particularly for n-3 and n-6 fatty acids, which may affect the generalisability of the findings. The study may have been underpowered to detect very small differences in fatty acid changes between intervention arms. Second, the quasi-experimental design of the study, as opposed to a randomised controlled trial, warrants caution in extrapolating results in a different population. Third, the rollout of the interventions in all schools may not be identically the same in terms of quality of delivery by teachers (nutrition education) and time of delivery (breakfast provision) as each school varies in terms of academic calendar, which may influence the results favouring schools with a more consistent schedule. Fourth, school clustering could not be accounted for in the study design due to fixed clusters (six schools), each representing a different arm; therefore, a high clustering effect was found in the results of some variables. Given the small number of participating schools, the study was underpowered to disentangle intervention effects from school effects for some of the outcomes. Moreover, statistical differences found in age, mother’s and father’s occupation, mother’s education, and mother’s age and education across the three groups may influence the findings. However, ANCOVA analysis confirmed no effect was found. These limitations should be considered when interpreting the results and suggest areas for future research.
Our study concludes that the GESIT programme was able to improve short-term nutritional knowledge within the BreakfastEdu and Education groups, despite no differences in delta scores (baseline-endline, endline-follow-up and baseline-follow-up) between the three groups. Our study also confirmed that the provision of a coupled breakfast and nutrition education programme enriched with n-3 and n-6 was effective in increasing energy, protein, fat (total fat, PUFA, ALA and LA) and few micronutrients (vitamins A, B1, B2, C and D, Ca, Fe, Zn, and potassium), as well as short-term memory within the BreakfastEdu group in the short term. However, very limited longer-term effects were shown without further reinforcement post-intervention, highlighting the need for integrative approaches that combine routine nutrition education, meal provision and policy-level interventions. To encourage sustainable eating behaviours, we strongly urge that future programmes incorporate nutrition education in the school curriculum. This will enable routine sessions, engage parents or caregivers and leverage policy frameworks to ensure consistency and long-term adherence.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114526106916
Acknowledgements
The authors would like to acknowledge the schools that have participated in the GESIT study and provided full support during the data collection.
The study was funded by Flora Food R&D B.V. and Flora Food Global B.V. Flora Food Group manufactured and provided the n-3 and n-6 fat spread during the intervention; however, participants were not exposed to any brand of food contained in the study throughout the intervention. Flora Food Group was neither involved in the recruitment of participants nor in the intervention or the final set of results.
Conceptualisation: P. H. R., D. B., E. A., M. N. H. S., A. P. R., Y. S. H., E. S. and A. L. H.; data curation: P. H. R. and E. A.; formal analysis, P. H. R., E. A., M. N. H. S., W. W. O., A. P. R. and Y. S. H.; investigation: P. H. R., E. A., M. N. H. S., R. N., W. W. O., A. P. R. and Y. S. H.; methodology: P. H. R., D. B., E. A., M. N. H. S., R. N., A. P. R. and Y. S. H.; project administration: P. H. R. and W. W. O.; supervision: P. H. R. and D. B.; validation: P. H. R., E. A., M. N. H. S., A. P. R. and Y. S. H.; visualisation: A. P. R.; writing – original draft: P. H. R., E. A., M. N. H. S., W. W. O., A. P. R. and Y. S. H.; writing – review and editing: P. H. R., E. A., E. S. and A. L. H.
E. S. and A. L. H. are employed by Flora Food Group. All authors declare no conflict of interest.
Open access
