Introduction

According to the World Health Organization (WHO), adolescence (aged 10 to 19 years) is the transition between childhood and adulthood1. Due to melodramatically fast-tracked growth and advancement, adolescence is a decisive time in a person’s life after infancy2. Better diet quality is essential for supporting the rapid growth and developmental changes associated with puberty and brain development. Adolescents are particularly vulnerable to malnutrition due to their accelerated growth and increased physiological demand for nutrients. Insufficient or excessive nutrient intake during this period can lead to various forms of malnutrition, including undernutrition, overnutrition, and micronutrient deficiencies3,4.

The Dietary Diversity Score (DDS) is an indicator used to assess the quality of the diet at the population level. It reflects one key aspect of diet quality, which is a micronutrient adequacy. The most important micronutrients associated with good dietary quality include eleven essential vitamins and minerals. Specifically, the vitamins include vitamin A, the B vitamins (thiamine, riboflavin, niacin, vitamin B-6, folate, vitamin B-12), and vitamin C. The minerals covered are calcium, iron, and zinc. These nutrients are vital for the growth and cognitive development of adolescents5,6.

Diet quality encompasses multiple dimensions that are critical in understanding the causes of malnutrition. These dimensions include nutrient adequacy, food variety, and balance of food groups, all of which contribute to the overall health and nutritional status of individuals. Addressing these dimensions is essential for preventing malnutrition and promoting optimal growth and development in adolescents. Thus, improving diet quality is a key strategy in combating malnutrition7,8,9,10.

Poor diet quality, whether inadequate or excess, combined with monotonous eating habits, is one of the modifiable and immediate driving forces behind the double burden of malnutrition (DBM). It is a significant shared factor contributing to both undernutrition and overnutrition. Therefore, improving diet quality is a critical focus in efforts to combat and prevent the double burden of malnutrition11,12,13,14,15.

Evidence suggests a negative association between inadequate DDS and the DBM. Low DDS is a leading cause of undernutrition16,17,18, while the likelihood of overweight and obesity increases with higher DDS10,16,19,20. Additionally, consuming only one meal per day is associated with an increased risk of overweight and obesity21. Therefore, promoting appropriate dietary diversity is crucial for DBM prevention22.

The Ethiopian healthcare system offers nutrition education to adolescents, both in schools and communities to combat malnutrition due to poor diets. However, the current efforts have been unsuccessful in improving adolescents’ nutritional behavior, likely due to inadequate implementation of double-duty nutrition interventions and the lack of integration with health behavior models23,24.

Double-duty interventions (DDIs) have the potential to improve nutritional outcomes by addressing various forms of malnutrition through integrated actions, programs, and policies. In this study, selected DDIs were enhanced through robust nutritional education and behavior change communication (NBCC). These are focused on promoting healthy diets (adequate adolescent nutrition, dietary diversity), physical activity, preventing the consumption of energy-dense foods (e.g., junk food, sugary drinks), and regulating marketing foods12,25,26.

Evidence shows that health promotions supported by behavioral models and theories can improve dietary practices by shifting unhealthy behaviors and enhancing the delivery of nutrition interventions27,28,29. In this study, adolescents were guided by a double-duty intervention packages based on Health Belief Model (HBM) constructs24,27,30,31,32. The HBM, an interactive model, effectively encourages positive health actions and behaviors. It includes six theoretical constructs: perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to action, and self-efficacy. The model’s effectiveness in modifying health behavior aligns with previous findings23,24,33,34,35,36,37,38,39.

Regarding the research gaps, there is a notable gap in studies focusing on the application of double-duty interventions (DDIs) using behavioral models, as well as a scarcity of interventional studies in Ethiopia40,41. This study aimed to address these gaps by applying selected DDIs to improve diet quality by enhancing dietary diversity scores. Since diet quality is a modifiable and immediate factor significantly influencing the occurrence of malnutrition, a cluster-based interventional study design was chosen to emphasize intervention than observational studies26,42.

Moreover, previous studies on improving the dietary quality of underprivileged adolescents often did not fully incorporate double-duty intervention packages and lacked support from behavioral theories43,44,45,46. To overcome these limitations and enhance the acceptance and effectiveness of the intervention packages, this study integrated constructs from the HBM. Therefore, the main objective of the study was to estimate the effect of selected double-duty interventions on the minimum dietary diversity score of adolescents using a health belief model in Debre Berhan Regiopolitan City, Ethiopia, 2022/23.

Methods and materials

Study area, period, and design

This study was conducted in Debre Berhan Regiopolitan City (DBRPC), Central Ethiopia, which is 130 km from Addis Ababa, the capital of Ethiopia. It is the capital city of the North Shoa Zone of the Amhara region and was built by Emperor Zara Yakoob. It is an emerging and industrialized city. As an emerging and industrializing city, urbanization is expected to increase the population’s exposure to both sides of malnutrition, particularly to overweight and obesity. The details of the study area, period, and design were published previously47. Considering schools as a unit of randomization, a nine-month two-arm parallel design school-based cluster randomized controlled trial was conducted from October 13, 2022, to June 30, 2023.

Population, inclusion, and exclusion criteria

All adolescents in the secondary school of the city were the source population of the study, whereas adolescents in the selected school of the city were the study population during the study period. Regarding the inclusion criteria, adolescents aged 10–19 years and who had followed their teaching–learning in the selected schools and adolescents who had no intention of leaving the schools until the end of the study during the study period were included. The exclusion criteria were adolescents who were unable to respond to an interview and who had physical disabilities, including deformities such as kyphosis, scoliosis, or limb deformities that prevent them from standing erect.

Sample size determination

The Gpower 3.1.9.2 program was applied to calculate the sample size using a power of 80.5% for two independent groups, a double proportion formula, a precision (alpha) of 5%, a 10% loss to follow-up, a design effect of 2, and proportions of indicators of malnutrition (P0 = 15.8%48 and P1 = 6.2%49). Thus, the maximum calculated sample size was 648, with 324 children in the intervention group (IG) and 324 in the control group (CG), achieving a 1:1 allocation ratio. However, due to cluster randomization, the final sample size was 745 for baseline data collection and 708 for endline data collection.

Intervention allocation, randomization, and recruitment

The intervention allocation, randomization, and recruitment of the study participants were performed in five sub cities of the DBRPC. Ten secondary schools were found in these sub cities. To prevent contamination of the message of the intervention as well as for better logistical suitability, these schools were considered clusters for the unit of randomization instead of individuals. Moreover, to avoid such information contamination, nonadjacent kebeles which contain these schools were included, and these schools were randomly assigned to the intervention or control group using a simple random method. The randomization of the schools was conducted by an independent third party to ensure that the process was free from bias. This third party was responsible for implementing a randomization protocol that ensured the intervention and control groups were assigned fairly and without any influence from the study team.

Sampling techniques and procedures

To select adolescents from the respective schools in the city, a multistage cluster sampling technique was applied. This study was conducted considering schools in the city as clusters. Of the ten schools in the city, eight are government schools and two are private schools. Of the ten schools, six schools were selected randomly and allocated to the intervention and control groups using a randomization table. The number of adolescents in each selected cluster was taken from all these school documentations. For the data collection and provision of the interventions, sections were used in the selected schools. When an eligible study participant was absent during the first visit, a revisit was conducted three times. Participants were considered nonresponsive if they missed three consecutive visits.

Intervention packages of the study and their mode of delivery

The WHO 2017 policy brief report and Hawkes et al., 2020 paper were used to modify and adopt the selected packages of the double-duty interventions. The selected packages were promoting healthy diets (adequate adolescent nutrition, dietary diversity); physical activity (do moderate intensity physical exercise, avoiding sitting for a long time); undue harm from energy-dense foods (avoiding junk processed foods, avoiding fizzy sweetened drinks, street fast foods, chips, salt, sugar, fats, etc.); and regulations on marketing foods from the customer side point of view (e.g., avoiding buying packed foods frequently). These intervention packages were given in a school-based setting through multimodal approaches within the selected intervention schools, particularly in the selected section of the students, using an NBCC tool and a theory-based HBM for at least twelve contacts. The multimodal approach was NBCC through direct contact with the students using in each school, brochures, message texts, and phone calls for their parents during the intervention period.

The intervention group was administered nutritional behavior change communication every two months, text messages every two months, and phone calls every two months for seven months duration, while the control group received the standard care provided by the health care system. The number of participants who attended the NBCC, received the message, received the phone call, and received the leaflet and had the expected frequency and attendance in the education sessions were used to measure compliance with the intervention. The details of the selected intervention packages were published previously elsewhere47.

Data collection methods, procedures, and measurements

Data collection methods and procedures

A digital data collection system was implemented using Kobo Toolbox, with a pretested and structured interviewer-administered questionnaire adapted and modified from various sources. The questionnaire was first prepared in the English language and subsequently translated into language spoken in the study area (Amharic) by language experts. Then, the questionnaire was translated back to English by an independent language expert for both languages to maintain consistency. Twelve nurses and health officers served as data collectors, while six nutritionists supervised the study. Sociodemographic variables, adolescent health, dietary practices, and HBM theoretical construct-related questions were included in the questionnaire. At baseline, sociodemographic factor data were collected, while adolescent health, dietary practice, and HBM theoretical construct-related data were measured at both baseline and end-line.

Measurements

The minimum dietary diversity score (MDDS) developed by the FAO was utilized to assess the dietary diversity score of the adolescents5. To meet the minimum dietary diversity requirement, teenagers were expected to consume at least five distinct food groups within the previous day or evening. A useful indicator of enhanced diet quality and sufficient intake of micronutrients is the percentage of teenagers in a population who fulfill this minimum criterion of consuming five food groups out of a total of ten food groups. In essence, a higher prevalence of DDSs serves as a proxy for improved micronutrient adequacy within a given population.

This study examined fourteen distinct food groups, which included cereals, white tubers, roots, legumes, seeds, and nuts; vitamin A-rich vegetables and tubers; dark green leafy vegetables; other vegetables; vitamin A-rich fruits; flesh meat; eggs; fish and seafood; and milk and milk products. These dietary categories were further divided into ten categories during the analysis period: grains, white roots and tubers; pulses (beans, peas, and lentils); nuts and seeds; milk and milk products; flesh foods (organ meat and fish); eggs; vitamin A-rich dark green leafy vegetables; other vitamin A-rich vegetables and fruits; and other vegetables and fruits5. The DDS for each adolescent was calculated based on their eating habits over the preceding 24 h. The score, which is based on 10 food groups, was categorized as high (≥ 5 food groups) or low (< 5 food groups). During the analysis, DDSs ≥ 5 food categories were categorized as “1”, and DDSs < 5 food groups were categorized as “0”.

Study variables

Improving the DDS through double-duty interventions was the dependent variable of the present study, while sociodemographic status, environmental health status, factors related to adolescent healthcare practice, factors related to adolescent diet and lifestyle, and HBM theoretical constructs were the independent variables.

Data quality control and intervention fidelity

With respect to the data quality control measures, a pretest, translation, and contextualization of the adapted and further developed English language version questionnaire into the Amharic language version were performed. Again, retranslation back to the English language was performed by a person who obtained a Master of Arts in the English language. To maintain the consistency of the two versions, a comparison was made. Translated and pretested Amharic versions of the questionnaire were used to collect the data. To test the tool (questionnaire), a pretest was administered, which was further modified next to the pretest. During the data collection period, all the details of the study were well-versed by the interviewers. All the data collectors and supervisors were trained in how to interview and supervise the data collection, respectively. During the interviews, the respondents were invigorated to feel free. The confidentiality of the responses was conserved for respondents because no information was provided publicly. Adolescents who were willing to partake and retain the informed consent document from their parents/guardians and the assent form were interviewed.

With regard to intervention fidelity, criteria were established according to the National Institutes of Health Behavioral Change Consortium recommendations50, which include checklists assessing the double-duty intervention design, training of educators, process of communication, receipt of the interventions, and enactment of skills gained from the double-duty intervention51,52,53,54,55,56,57. Equal numbers of clusters were taken from each cluster to balance the variations between different clusters in the city for both the intervention and control arms. Moreover, nonadjacent kebeles and buffer zones were used between adjacent clusters to overcome the possibility of information contamination occurring between clusters. In addition, to compare the effectiveness of the double-duty interventions, the study included a control group. In addition, for the application of the actual trial, the intervention process was pretested in an area other than the study area with a similar set of setups.

Data management and analysis methods

The lead investigator visually verified the consistency and completeness of each interview questionnaire. After the gathered data were exported to SPSS version 25, missing values and outliers were examined. Measures of central tendency, variability, percentages, and simple frequency distributions were used to characterize the respondents’ demographic, socioeconomic, and adolescent-related traits. Graphical methods (e.g., histogram), numerical methods (mean), and a statistical test (the Kolmogorov‒Smirnov test because the sample size is greater than fifty) were used to verify the normality of continuous data. A chi-square test was used to compare the intervention and control groups’ baseline characteristics. The final analysis was performed with R software. A per-protocol analysis was used to include respondents who followed the established protocol in the final analysis.

The goodness of fit (or model fitness) was assessed using the Quasilikelihood under the Independence Model Criterion (QIC), which is based on a "smaller-is-better" assumptions. The QIC for the model was 1382.388 and the Corrected Quasilikelihood under the Independence Model Criterion (QICC) was 1407.631. The QIC was chosen for evaluating model fitness due to its lower value.

The difference-in-difference (DID) between the intervention and control arms was analyzed using the McNemar test, a chi-square statistical test that assesses paired nominal data, particularly for evaluating changes in responses before and after the intervention. Furthermore, a generalized estimating equation (GEE) model with a binary logit function was employed to investigate the difference in changes between the intervention and control groups to assess the impact of interventions on changes in the DDS of adolescents over time. Because of repeated measurements (before and after interventions) and clustered data, this model can account for the correlation of various observations within patients. When fitting the model, the exchangeable correlation structure was considered along with the possible effects of confounding variables.

Time, intervention, time and intervention interaction, teenage sex, home garden practice, wealth index, place of residence, nutritional knowledge, ability to make decisions about food-related matters, and alcohol consumption were all considered when adjusting the model. The interaction between time and intervention was used to evaluate the intervention’s impact on the DDS. The 95% confidence intervals (CIs) for the adjusted odds ratios (ORs) were calculated. A P values less than 0.05 indicated statistical significance.

Ethical consideration

The study was conducted in accordance with the Helsinki Declaration and the principles of Good Clinical Practice (GCP)58. The research protocol was approved by the Institutional Review Board (IRB) of Jimma University, with ethical clearance obtained on August 10, 2022 (Reference: JUIH/IRB/104/22), prior to the start of data collection. Written informed consent was obtained from each participant’s parent or guardian, and assent was secured from the adolescents after explaining the study’s purpose and benefits. Additionally, the trial was prospectively registered on ClinicalTrials.gov (Registration Number: NCT05574842) and first posted on October 12, 2022.

Results

Sociodemographic characteristics of the respondents

During the baseline data collection, 745 study participants were enrolled. At the baseline, 742 (IG = 375, CG = 367) adolescents were provided complete information and randomly assigned to the intervention or control group, whereas at the endline, 708 (IG = 356, CG = 352) respondents strictly adhered to the protocol. During the baseline data collection, there was no significant difference between the intervention and control groups in terms of baseline sociodemographic characteristics (P > 0.05). The detailed baseline sociodemographic characteristics of the respondents are described elsewhere15. The Consolidated Standards of Reporting Trials (CONSORT) guidelines were used to report the results59. A schematic of the sampling procedure for participant selection was generated based on the CONSORT guideline criteria and is described elsewhere47.

Dietary diversity scores among adolescents

Prior to the enactment of the intervention at baseline, there was no statistically significant difference in the dietary diversity score between the intervention and control groups. After the intervention, the dietary diversity score improved more in the intervention group than in the control group at the endline period. According to the DID analysis, the reduction in the proportion of adolescents with a low DDS was greater in the intervention group compared to the control group (24.6% vs. 5.8%, P < 0.001). Similarly, the increment in the proportion of adolescents with a high DDS in the intervention group was greater than in the control group (22% vs. 3.6%, P < 0.001). Moreover, according to the overall DID analysis, the proportion of adolescents with a low DDS decreased by 30.4%, while the proportion of adolescents with a high DDS increased by 18.4% in the intervention group compared to the control group (Table 1).

Table 1 Applications of differences-in-difference analysis to evaluate the effect of the double-duty intervention packages on dietary diversity scores based on time of measurements and treatment groups among adolescents in Debre Berhan City, Ethiopia, 2022/23.
Full size table

The DDS of adolescents was assessed using ten food group indicators. The consumption of these ten food groups over the previous day and night was compared by treatment group and time of measurement. At the end of the trial, the intervention group consumed higher-quality food groups compared to the control group (Fig. 1).

Fig. 1
figure 1

Food groups by treatment group and time of measurement among adolescents, Ethiopia, 2022/23.

Full size image

Effect of double duty intervention packages on dietary diversity score

To investigate the effect of the double-duty interventions on DDS, a GEE model for binary logistic regression was applied. After controlling for all possible confounding factors, the DDS showed improvement in the intervention group compared to their counterparts. In this model, adolescents in the intervention group were 1.91 times more likely to have a high DDS compared to those in the control group [AOR = 1.91, 95% CI (1.85, 1.96)]. Similarly, adolescents who completed the endline measurements were 1.28 times more likely to have a high DDS compared to those measured at baseline [AOR = 1.28, 95% CI (1.19, 1.37)]. Additionally, the interaction between the time of measurement and treatment group (time*treatment) showed a significant association with the improvement in DDS among adolescents. Adolescents in the interaction category were 11.59 times more likely to have a high DDS compared to their counterparts [AOR = 11.59, 95% CI (11.18, 12.03)] (Table 2).

Table 2 Applications of the GEE for binary logistic regression model1 to estimate the effect of double-duty intervention packages on the dietary diversity score of adolescents in Debre Berhan City, Ethiopia, for the year 2022/23.
Full size table

Discussion

To improve diet quality and achieve a better dietary diversity score, selected double-duty intervention packages were implemented using HBM. These interventions were enhanced through robust nutrition behavior change communication (NBCC) with message texts, phone call, and leaflets.

According to the DID analysis, adolescents in the intervention group had a higher DDS than adolescents in the control group did. In this analysis, the overall proportion of low DDS was decreased by 30.4% and the overall proportion of high DDS was increased by 18.4% among the intervention than control group. This finding is consistent with the existing evidence that utilized a nutrition education and nutrition behavior change communication intervention, along with one or more packages of the double-duty intervention, to enhance the dietary status of adolescents. The effectiveness of these interventions has been supported by various studies conducted in different countries, including in Croatia on diet quality60, Bangladesh on dietary diversification61, Ghana on iron and iron-rich food intake practices62, Ethiopia on DDSs63, Ethiopia on the diet quality of children64, Ethiopia on the dietary practices of women65, Ethiopia on optimal dietary practices46, and Southwest Ethiopia on the DDS of women66. However, the present study extends beyond previous research by incorporating a more diverse range of nutritional interventions alongside physical activities specifically tailored to improve the DDS and overall dietary quality of adolescents. By implementing these comprehensive approaches, the study aims to achieve better nutritional status and effectively prevent the double burden of malnutrition.

In addition, after all possible confounding factors were controlled, the binomial generalized estimating equation (GEE) model showed an improved DDS in the intervention group compared with the control group. Adolescents in the intervention group were nearly twofold more likely to have high DDS than were those in the control group. This result aligns with previous findings, despite differences in study populations. The consistency in results suggests that, regardless of population variations, the observed effects hold true across different contexts66.

Similarly, adolescents who completed the endline measurements were nearly twofold more likely to have a high DDS than adolescents in baseline period. Similarly, the interaction between the time of measurement and treatment group had a significant association with the improvement in the DDS. Adolescents in the time of measurement and treatment group interaction category were nearly twelve times more likely to have DDS than their counterparts. A similar finding was reported in previous studies conducted among women24,66.

To enhance acceptance and uptake of the interventions, the double-duty intervention packages were guided by the HBM. As a result, the DDS of adolescents improved at endline measurements and intervention group compared to both baseline measurements and the control group, respectively. This is supported by existing evidence23,24,67,68. One possible reason for this outcome could be attributed to the utilization of the HBM constructs, which effectively enhance the acceptance and adoption of the intervention. The incorporation of the HBM constructs in the intervention could have positively influenced individuals’ perceptions, motivations, and beliefs, thereby increasing their willingness to accept and engage with the intervention.

The associations between DDS and HBM scores were compared both within and between the intervention and control groups. All HBM constructs showed significant positive correlations with the DDS. The positive correlation between the HBM constructs and DDS was supported by previous findings from Ethiopia23,24,27,36,37,39. One possible reason for the improvement in the intervention could be the utilization of the HBM. By addressing key factors such as perceived susceptibility, severity, benefits, barriers, and self-efficacy, the HBM may have played a crucial role in encouraging individuals to embrace the intervention and actively participate in it. The HBM provides a framework for understanding individuals’ beliefs and motivations related to health behaviors.

Ethiopia’s Food and Nutrition Policy (FNP) was approved to improve dietary habits, including those of adolescents69. However, it does not fully incorporate double-duty interventions aimed at enhancing adolescent dietary quality through improved diversity. Therefore, updating the policy to include these interventions is recommended to address inadequate dietary diversity among adolescents, both in the study area and nationally.

The strengths of this study include the integration of a theory-based Health Belief Model with double-duty interventions, which increased the likelihood of success. Additionally, the intervention was delivered using a multimodal approach, including text messages, phone calls, and behavior-change communication. However, limitations include recall bias, social desirability, and reporting bias were expected. These biases were minimized by thoroughly explaining the process to participants, probing for details, and ensuring accurate reporting.

Conclusion

In this study, compared with those in the control group, the dietary diversity scores of the intervention group of adolescents significantly improved. This improvement can be attributed to the implementation of the double-duty intervention accompanied by the utilization of the health belief model. Compared with the control group, the intervention group led to a substantial decrease of nearly one-third in the overall proportion of respondents with a low DDS and a significant increase of almost one-fifth in the overall proportion of respondents with a high DDS among the intervention group. These findings indicate that the intervention had a considerable impact on improving the dietary diversity of the participants. Furthermore, when potential confounding factors were controlled for using the generalized estimating equation model, adolescents in the intervention group demonstrated a substantial improvement in their DDS at the end of the measurement period. This finding suggested that the observed improvement in dietary diversity was directly associated with the intervention and not influenced by other variables. Based on these findings, double-duty interventions, accompanied by the incorporation of the health belief model, are recommended to be integrated into national guidelines as part of the country’s overall strategy and nutrition policy. By doing so, the potential benefits of these interventions can be extended to a larger population, leading to improved dietary diversity and better nutritional outcomes among adolescents on a national scale.