Text messages to improve child diets: Formative research findings and protocol of a randomised controlled trial in Nepal

Abstract

Given the role of malnutrition in childhood morbidity and mortality, the prioritisation of maternal and child nutrition programmes has grown significantly in the 21st century. Policies and programmes aim to improve infant and young child feeding, but questions persist about the most effective combination of interventions to achieve desired behaviour change. There is increasing interest in mobile‐based interventions globally, but scant evidence exists to guide donors, policymakers and programme implementers on their effectiveness. Formative research was conducted to assess the feasibility and acceptance of text message‐based interventions and to guide the final design of the text message intervention. This protocol is for a cluster‐randomised controlled trial to test the effectiveness of adding text messaging to other ongoing SBC interventions to promote egg consumption, dietary diversity and other ideal dietary practices, particularly among children 12–23 months of age in Kanchanpur, Nepal. The trial findings will contribute to the emerging body of evidence on the effectiveness of using text messages for behaviour change, specifically for young child dietary outcomes in South Asia. Recent studies have suggested that mobile‐based interventions alone may be insufficient but valuable when added to other social and behavioural interventions; this trial will help to provide evidence for or against this emerging theory. This trial was registered at ClinicalTrials.gov on 11 March 2019 (ID: NCT03926689) and has been updated twice.

Key messages

• Multiple rounds of formative research can facilitate higher‐quality interventions including optimal wording, timing and reach of short message service to target audiences to improve nutrition‐related behaviours.

• Experimental studies are needed to guide decision‐making around the use of technology‐based behaviour change interventions to improve diets.

• This trial to test the effectiveness of text messaging to improve young child diets, and its process evaluation, will inform future mNutrition interventions.

1. INTRODUCTION

Childhood undernutrition remains highly prevalent in low‐ and middle‐income countries (LMICs) with estimates suggesting that 178 million children less than 5 years of age are stunted, 112 million are underweight and 55 million are wasted; 90% of this global burden is concentrated in sub‐Saharan Africa and South Asia (Black et al., 2013; Wali et al., 2019). The World Health Assembly set a global target of a 40% reduction in childhood undernutrition by 2025, but this target will be missed by an estimated 30 million children if the current trends continue (de Onis et al., 2013). This is a serious global concern as undernutrition is known to be the underlying cause of more than one‐third of childhood deaths globally (Black et al., 2013).

The time between a child's birth and 2 years of age is a critical window of opportunity to ensure a child's development, in part through proper feeding practices. Even mild or moderate undernutrition during this period can cause irreversible damage. Improper infant and young child feeding (IYCF) practices are a main factor contributing to high rates of child undernutrition in many countries. Recommended IYCF practices have the potential to prevent an estimated 19% of all under‐5 deaths, more than any other single preventive intervention (Bhutta, 2013; Bhutta et al., 2008, 2013; Jones et al., 2013). Therefore, it is recommended that countries track indicators of core IYCF practices for the proper growth and development of children (World Health Organization, 2008).

Although significant reductions in under‐nutrition among children over the last several decades in Nepal have been well documented, among Nepalese children under 5 years of age, 36% are stunted and 10% are wasted (Nepal Ministry of Health, 2017). Complementary feeding practices, known to be important for nutrition outcomes, remain poor in Nepal: only about one‐third of children 6–23 months of age receive a minimum acceptable diet and less than half (47%) eat foods from at least four of seven food groups to attain minimum dietary diversity. Eggs, a food well‐known to be highly nutritious, are only consumed by 10% of these children (Nepal Ministry of Health, 2017).

A recent randomised controlled trial conducted among young undernourished children in the Ecuadorian Andes showed significant effects on the reduction of stunting after the children consumed eggs for just 6 months (Iannotti et al., 2017). Eggs are considered a complete food and arguably more accessible and affordable to resource‐poor populations than other nutritious complementary foods, specifically fortified foods (Iannotti et al., 2017). They are also relatively simple to store and prepare. Eggs are not widely consumed, however, among children and women of reproductive age, particularly in LMICs (Lutter & Morris, 2018). In low‐income settings, low consumption of eggs among young children and other poor dietary practices, are in part the result of inaccurate caregiver perceptions about the dietary needs of young children, the complexity of preparing nutritionally adequate complementary foods, household financial constraints and lack of access and availability in local markets (Waters et al., 2018). Inadequate parental knowledge, although not the only factor, is a main contributor to inappropriate IYCF and egg consumption practices in Nepal, specifically in the lowland plains (Gautam et al., 2016; Joshi et al., 2012).

Social and behaviour change (SBC) theory can be a useful component of nutrition interventions. Studies have repeatedly shown that when target populations are reached with greater frequency and intensity, there is a greater likelihood of behaviour change. Recent analyses of USAID funded Suaahara (Integrated Nutrition Program) data confirmed this for nutrition interventions in Nepal (Suresh et al., 2019). Evidence from a recent trial shows the effectiveness of culturally tailored behaviour change strategies, particularly when combined with addressing resource constraints, to improve young child egg consumption and reduce malnutrition (McKune et al., 2020). Timely and important information increases knowledge and may also serve as an important reminder or highlight the importance of a behaviour that may not be traditionally valued, even for those who are already knowledgeable (Timmers et al., 2020). In both high‐income settings and LMICs, the use of mobile platforms for providing health services (mHealth) or spreading critical information and motivation for behaviour change is growing. It is widely claimed that sending short message service (SMS) via mobile phone may be an inexpensive, non‐labour‐intensive channel of communication to reach households residing in remote areas with important health and nutrition information (Mildon & Sellen, 2019). MHealth interventions have demonstrated their effectiveness for smoking cessation, diabetic education and promotion of diet and physical activity (Kitsiou et al., 2017; Müller et al., 2016; Regmi et al., 2017). Some studies in high‐income settings have looked at text messages and other digital platforms as an avenue for improving breastfeeding rates. Evidence to date, however, for using mHealth to promote ideal IYCF practices is scant, particularly experimental evidence or evidence from interventions implemented at scale (Graziose et al., 2018).

The rapid increase in mobile phone ownership in Nepal, even among remote populations, offers a potential mechanism for delivering nutrition messages to target groups in timely and  cost‐effective manner, relative to interpersonal counselling and mass media campaigns. Given the combination of sub‐optimal IYCF practices, the limited reach of the health system and fast‐growing rates of mobile phone ownership, the Government of Nepal as well as donors and implementers are interested in whether SMS campaigns can be used effectively for behaviour change. Thus, within the context of Suaahara II, a large‐scale integrated programme aiming to reduce under‐nutrition among women and children in the 1000‐day period (between conception and a child's second birthday) in Nepal, qualitative formative research was conducted to test the feasibility and acceptability of an SMS intervention and specific SMS messages. The majority of the SMS messages in this intervention focused on the consumption of eggs among children 6 months of age and older but other messages focused on other optimal maternal and child health and nutrition‐related behaviours, such as washing hands with soap and water, overall dietary diversity for mothers and children and participation in other intervention platforms.

1.1. Formative research

The formative research was carried out in Kailali, a geographically and demographically similar neighbouring district to the trial district, Kanchanpur. During the first round of formative research, in May 2019, Suaahara sent messages for 1 week to 1000‐day mothers residing in four municipalities (one ward per municipality), selected purposively based on having a high population of 1000‐day mothers, high mobile phone ownership, and easy road access. In total, 24 SMS messages were sent to explore the clarity, understanding, trustworthiness and likeability of the messages. Four focus group discussions (FGDs) were conducted, one in each study site. Age and stage‐specific messages were sent to the four sub‐categories of 1000‐day mothers (pregnant, child less than 6 months, child at least 6 but less than 12 months and child at least 12 but less than 24 months). A total of 37 mothers participated in the study, with 8–10 women per FGD (Krueger & Casey, 1988; Stewart & Shamdasani, 2014).

We found willingness to receive SMS, content understanding, and interest in receiving these types of SMS messages to be high. A major concern, however, was that only 22% of study participants (n = 37) had received the messages despite nearly all of the participants owning phones. Therefore, it was clear that households would need to register and have their mobile numbers updated more regularly; SMS messages would need to be sent for a longer period of time and, given the intervention and trial intent of including other adult household members, formative research would also need to be conducted with other adults in the household. During the FGDs, mothers also provided their feedback to facilitate revising the SMS messages to improve clarity, understanding and perceived effectiveness. For example, mothers sometimes found it difficult to understand complex Nepali words, as their native language was different and many participants felt that some suggestions such as feeding eggs ‘every day’ were unrealistic.

A second round of formative research was conducted in September 2019, following intensive efforts by Suaahara II staff to update registries of names and numbers. The revised SMS messages were sent to mothers and other adult household decision‐makers in two municipalities for a month. The two municipalities were selected purposively based on the same criteria: high population of 1000‐day mothers, high mobile phone ownership and easy road access. In the first municipality, two FGDs were conducted, the first with pregnant women and mothers with a child less than 6 months and the second with their family members. Similarly, in the second municipality, two FGDs were conducted, the first with mothers with a child at least 12 but less than 24 months and the second with their family members. Therefore, four FGDs, two with mothers (n = 24) and two with other adult household members (n = 18), with 7–12 participants per group (Krueger & Casey, 1988; Stewart & Shamdasani, 2014), were conducted to test 23 revised messages.

Following the more intensive intervention, the reach of SMS messages increased from less than a quarter to almost all mothers and half of other adult household members. We again found willingness and interest in receiving informative SMS messages and a high degree of understanding of the nutrition information. Additional Nepali words of medium difficulty, however, were identified as challenging to understand because of their first language not being Nepali. Findings also showed that opening SMS messages, owning a mobile phone and being able to read SMS messages were more common among the younger generation (including 1000‐day mothers, their husbands, brothers‐in‐law and sisters‐in‐law) than the older generation (mothers‐in‐law and fathers‐in‐law).

Given the programmatic interest in improving IYCF practices, specifically increasing egg consumption among young children, Suaahara II used the findings from these two rounds of qualitative formative research to improve the SMS intervention design and implement a cluster‐randomised controlled trial (C‐RCT) to test the effectiveness of using SMS to promote egg consumption and other ideal IYCF practices among children 12–23 months of age in Nepal, in the context of Suaahara II's other ongoing SBC interventions.

2. METHODS

2.1. Trial design

This C‐RCT involved 134 clusters, randomly allocated into two study arms: control (standard Suaahara II interventions) and SMS intervention (standard Suaahara II interventions plus SMS intervention) and uses a repeat cross‐sectional design. The cluster unit of randomisation was old wards (pre‐federalism), with population size ranging from 720 to 5924 (Central Bureau of Statistics, 2014), as the new wards after the federalist restructuring in 2018 are now the smallest administrative units but too large to be survey clusters. To reduce heterogeneity, four clusters were excluded for the C‐RCT (the district headquarter; two clusters in the middle of Suklaphanta National Park with less than 150 landless households without access to government resources; and one cluster where 85% of households are one specific caste/ethnic group). A total of 3350 households with a child 12–23 completed months of age at the time of the survey were to be recruited at baseline and at end line; the child's mother, father, grandmother and an additional adult male household member (preference for grandfather), when available, will be recruited for the survey. At baseline, after interviewing available grandmothers in the first 812 households (n = 327), we stopped the recruitment of grandmothers due to overall challenges in maintaining the data collection timeline, which was magnified by the grandmothers' poor comprehension and communication in Nepali. The same limited recruitment process for grandmothers used at baseline was also followed at the end line and grandmother data will only be used for secondary analyses. The C‐RCT design and randomisation details are included in Figure 1.

Figure 1.

 Study design.

2.2. Outcomes

The primary study outcome was the prevalence of egg consumption among children 12–23 completed months of age, measured at the individual level. Secondary outcomes for the study were (1) prevalence of egg consumption among other household members; (2) prevalence of obtaining minimum dietary diversity and mean dietary diversity among household members, including children 12–23 completed months of age; (3) prevalence of correct knowledge of core infant and young child feeding practices; (4) prevalence and frequency of participation in health mothers' group (HMG) meetings in the previous 3 months and (5) prevalence and frequency of listenership to the Suaahara II radio programme, Bhanchhin Aama, in the previous 3 months.

A 24‐h food recall method was used to collect the primary outcome data. An observation of egg availability in the household as well as a food frequency questionnaire to assess the frequency of egg consumption was also used to triangulate findings for the primary outcome. Similarly, the prevalence of egg consumption among other household members, the prevalence of obtaining minimum dietary diversity and mean dietary diversity among household members and children of 12–23 completed months of age were all measured using a 24‐h food recall method (World Health Organization, 2008). Mean dietary diversity was estimated for children who consumed at least four out of seven food groups and for adults who consumed at least five out of ten food groups using the Women's Dietary Diversity Score (FAO & FHI 360, 2016).

Using the WHO recommendations for the measurement of eight core IYCF practices, the corresponding prevalence of correct knowledge of IYCF practices were measured. Direct questions were asked regarding participation in HMG meetings and listenership to the Suaahara II radio programme, Bhanchhin Aama to measure prevalence and frequency in the previous 3 months.

The survey questionnaire also had modules to identify intervention exposure and other important factors that may potentially clarify linkages between exposures and outcomes or measure confounding factors. Exposure to Suaahara II interventions, sharing Suaahara II interventions (including SMS messages) and discussing health and nutrition information received with other members of the household or social networks were assessed. The survey questionnaire measured exposure to information and related knowledge and practices, using standard instruments (i.e., WHO/UNICEF/USAID FANTA) where possible.

2.3. Settings and study population

This trial was conducted in the Sudhurpaschim province, where the median years of completed education for women are 4.3, not even half (38%) of households are food secure and the prevalence estimates of stunting, wasting and underweight children under 5 years of age are 36%, 9% and 28%, respectively (Nepal Ministry of Health, 2017). Specifically, this trial was carried out in Kanchanpur (Figure 2), one of Nepal's 77 districts located in the south‐eastern corner of the Nepal terai (plains), bordered by India to the south and west and two other Suaahara II intervention districts, Kailali and Dadeldhura, to the east and north respectively. The literacy rate in Kanchanpur is 61% and 81% for women and men, respectively. The mother tongue varies widely with Doteli for about 40% of the population, Tharu for 25% and Nepali for 16% (Central Bureau of Statistics, 2014).

Figure 2.

 Map of Kanchanpur with municipality and ward boundaries. Source: Electoral Constituency Delineation Commission.

2.4. Interventions

Standard Suaahara II interventions spanning maternal and child nutrition; health and family planning; agriculture, food security and resilience; water, sanitation and hygiene and gender equality and social inclusion continued to be implemented throughout the trial. These standard Suaahara II interventions included collaboration and support to the government at federal and provincial levels; capacity enhancement of community leaders and frontline workers at the district and community levels and household‐ and individual‐level SBC activities. These SBC activities were primarily: interpersonal counselling via household visits by Suaahara II frontline workers community mobilisation such as HMG meetings, food and handwashing demonstrations and key life events to celebrate pregnancy, a new birth and a child turning 6 months of age; and mass media involving a locally produced weekly radio programme with a call‐in component known as Bhanchhin Aama and its airing on Facebook and YouTube.

The SMS campaign complemented these standard Suaahara II interventions in the trial intervention arm. The SMS campaign was designed based on extensive formative research and concept testing of individual messages with different types of respondents so that messages could address misconceptions and other barriers to ideal behaviours rather than be simply didactic. The final SMS campaign included 51 messages encouraging various optimal maternal and child health and nutrition‐related practices that are important during the 1000‐day period (Supporting Information: Table 1). At least one message per month was sent to mothers and heads of households in the 1000‐day period residing in the intervention areas, who provided their mobile numbers to Suaahara II frontline workers.

In addition to Suaahara II interventions, standard government facility‐ and community‐based health and nutrition interventions continued without difference by study arm. Additionally, there were other nongovernmental actors working in Kanchanpur, but not with district‐wide coverage and not with a nutrition or multi‐sectoral approach.

The theory of change for this intervention (Figure 3) was adapted from the diffusion of innovations theory (Valente & Fosados 2006). We hypothesised that the SMS messages would motivate behaviour change and perhaps be shared with others in the recipient's household or social network helping to shift social norms. Thus, the pathways to impact were multiple, including both directly influencing egg consumption but also motivating greater participation in other interventions (e.g., community events or listening to Bhanchhin Aama) which also promote ideal health and nutrition behaviours, including egg consumption.

Figure 3.

 Theory of change for Suaahara II short message service campaign.

2.5. Participants

For this C‐RCT, the intervention was started after the baseline survey. Between 1 April 2017, and 14 December 2018, Suaahara II frontline workers conducted the first round of a community mapping census (CMC), a door‐to‐door enumeration of households including their family members and relevant socio‐economic and demographic factors for all households in Suaahara II implementation districts. This was to identify 1000‐day households for programme targeting, collect mobile phone numbers and create a sampling frame for monitoring. Due to migration, births, deaths and other events, demographics change regularly, and thus, each Suaahara II frontline worker continuously updates these data for his/her communities during their daily household and community level activities including home visits for counselling, key life events, food demonstrations, HMG meetings and so on ensuring newly pregnant women and new births are added, phone numbers updated and any other personal details are updated.

For inclusion into this study, at the time of the survey, the household must have met the following eligibility criteria: (1) reside in a study cluster; (2) have at least 1 child 12–23 completed months of age who were not acutely ill the previous day and does not have a disability preventing normal feeding; (3) have at least one adult residing in the household and owning a mobile phone and (4) be able and willing to register their mobile number with Suaahara II and into the trial.

2.6. Timeline

The SMS intervention in Kanchanpur was initiated on 24 October 2019 and continued throughout 2020 and early 2021 through data collection. Baseline data collection was carried out from 20 August to 19 October 2019, and analyses were completed in winter 2020 to understand baseline level prevalence and means and whether there are any imbalances at baseline that would influence the planned trial analyses. End‐line data collection initially planned for these same dates in 2020, was carried started on 1 April 2021, after being put on hold due to the global Covid‐19 pandemic. Due to a surge of Covid‐19 cases, end‐line data collection ended early on 27 April 2021; data were, however, collected from 2525 households across 103 clusters and among the 31 clusters remaining 17 were in intervention areas and 14 in comparison areas.

The trial analyses of primary and secondary research questions are underway.

2.7. Sample size

Data from Suaahara II annual monitoring surveys from 2017 to 2018, which showed an increase in the prevalence of egg consumption among children 12–23 completed months of age from 12% to 17% in the terai, were used for sample size calculations. The intraclass correlations were 0.01 and 0.07 for 2017 and 2018, respectively. Kanchanpur CMC data indicated that there were 134 old wards in the district, with about 40 children aged 12–23 completed months in each.

Considering an intraclass correlation of 0.04 (average of 0.01 and 0.07), power at 90%, a significance level of 0.05, a roughly equal number of households per cluster, a minimum detectable difference of 5% points and 25 participants per cluster, the required sample size was calculated at 3350 children 12–23 completed months of age.

2.8. Randomisation and Blinding

As this was a trial related to SBC, blinding of the intervention to the participants was not possible. The allocation of clusters to specific arms, however, was concealed to the principal investigators (PIs), Co‐PIs, and Suaahara II and partner non‐governmental organization staff in Kathmandu and Kanchanpur and will remain so until the primary analysis is completed. All data collectors were also blinded as to which survey clusters were allocated to which study arm.

To maintain the blinding of the Suaahara II staff, the external PI did the randomisation and returned the random allocation directly to the survey firm. The external PI was also blinded as the trial manager had only provided random cluster numbers and no information about the identity of the clusters. The clusters were randomised to the two arms in blocks of four based on the rank of cluster size to help ensure that cluster size would not be a confounder.

2.9. Data collection, management and analyses

An external survey firm was hired to collect the baseline and end‐line survey data. All field staff including quality controllers, supervisors, and data collectors, were hired based on their similar prior work experience and understanding of local languages. All field staff were trained from 6 to 9 August 2019, including a pilot test, with the aim of familiarising them with study objectives, methods, and tools. Before this main training, a 2‐day training was provided to quality controllers and supervisors. All quantitative data were collected electronically, using android devices and a questionnaire programmed on Open Data Kit (ODK). The trial manager pre‐programmed skip patterns and logic checks to help avoid data entry errors. All questionnaires were developed in English, translated to Nepali, and back‐translated to English before data collection. All questionnaires were also pretested before transferring them into ODK.

During data collection, to select the sampled households, enumerators used Suaahara II CMC data to identify potential households with children between 12 and 23 completed months of age. This list was shown to the FCHVs in the cluster for confirmation of the eligibility of the households. FCHVs also notified the enumerators of any additional household names for households with a child in the study age range and their location. Household verification visits were conducted with all households on this updated list to confirm which of these households met the study inclusion criteria. During the household visit, a snowball method was also used to identify any other households with children 12–23 completed months of age that were not on the list. Once all verification was completed and a final list made, the required number of households in each cluster was drawn randomly.

2.10. Data management and analyses

Helen Keller International used its ONA.io enterprise license for data management, starting with creating mobile data collection tools. Suaahara II maintained administrative power to control the uploading and downloading of forms and data. Enumerators had access to upload forms/data and were required to submit their data by syncing to the server daily once the supervisor had verified their data. Quality controllers were then responsible for running ado file to check for outliers and consistency of data submitted the day before.

As dietary data was collected on paper, survey firm staff in Kathmandu entered the data and then merged the main data set and the dietary data set using a composite unique identification number. During the survey firm's cleaning of the data, they also converted answers provided as text, such as for 'other specify' response options, into numeric response options. Once the data were submitted to Suaahara II, further data cleaning such as checking all variables for outliers, missing values and patterns was done.

For analyses, imputation for missing values will be made as necessary, and normality assumptions and skew statistics for all continuous outcomes will be checked using histogram and univariate statistics. For primary and secondary study outcomes, we will estimate differences between arms in changes between baseline and end‐line prevalence or means. Intervention effects will be analysed as intent‐to‐treat (ITT) using multivariable regression with the dependent variable being the binary primary outcome or binary or continuous secondary outcomes. The multi‐level models will specify old wards as random effects and will specify the study arm, visit (baseline vs. end line), and their interaction as fixed effects. The interaction will estimate the differences between arms in changes from the baseline to the end line of the study. We will evaluate differences in changes using one‐tail p values consistent with the directional hypotheses for the outcome variables. Clustering at the old‐ward level will be accounted for in all analyses. Analyses will be performed using the diff (i.e., reg with cluster option) Stata version 14. This procedure in Stata provides the proper degrees of freedom for the tests.

Potential confounders (e.g., age, marital status, wealth, and location) will be identified. At first, all the statistical models will be analysed without adjustment for potentially confounding factors at the individual, household and community level and then we will adjust the models for these if they differed at baseline, differed over time or helped explain variation in outcomes.

Furthermore, because the proposed ITT analyses will underestimate the effectiveness had the SMS intervention been more extensively implemented and received, an analysis will be conducted to examine the impact of the trial intervention on primary and secondary outcomes among only those households who participated in the SMS intervention by providing their consent and mobile number and received and opened the SMS messages.

Additional analyses are also planned. First, pathway analyses will be performed to assess the potential paths between exposure to SMS messages and outcomes as outlined in the theory of change above. Second, regression modelling will be carried out to estimate whether individual or multiple adult household members receiving the messages simultaneously has any effect on study outcomes. Third and fourth, to identify the effect of the number of SMS messages and to identify the effects of intervention types, that is, inter‐personal communication (IPC), community events, mass media and SMS on study outcomes, a dose–response relationship will be analysed. Dose–response analyses will be conducted with the use of logit regression models while using exposure variables constructed from single and multiple platforms, the different platforms used will be IPC, community events, mass media and SMS.

2.11. Ethics and trial monitoring

This trial has been registered at clinicaltrials.gov. This study was also reviewed and approved by both the Nepal Health Research Council and the University of South Carolina Institutional Review Board. Written informed consent was collected before each instance of data collection.

We established a Trial Steering Committee whose overall responsibilities included: reviewing and providing technical inputs to trial protocols and any amendments; monitoring the trial to ensure it is on track to meet objectives; overseeing the trial and reviewing the data safety monitoring processes; reviewing and sharing information of relevance to the trial from other sources and resolving issues that may arise during the trial. This committee provided advice on trial design and guidance during implementation, ensuring that the rigorous standards set out in the CONSORT guidelines and agreed to in the ethics protocols were followed.

3. RESULTS AND DISCUSSION

This C‐RCT aimed to determine the effectiveness of using SMS messages, in the context of ongoing SBC interventions, to deliver nutrition messages to promote egg consumption and other ideal IYCF practices among children 12–23 months of age in Nepal. To our knowledge, this was the first large trial to focus on text messaging intervention to promote egg consumption and recommended IYCF practices among households with young children.

This intervention of SMS messages leverages the power of mobile phone technology that is undergoing a rapid improvement in coverage across all of Nepal. Given the increased availability of cheap smartphones and inexpensive subscriber identity module (SIM) cards, along with improved coverage of 3G network in Nepal, mobile‐based SMS messaging interventions are feasible. Not only do 9 of 10 households own a phone but 83% of mothers of children less than 2 years of age own a mobile phone. Also, the literacy rate among adults and school attendance rates among children both high, making it easy for households to read text messages they receive.

A key strength of the study was the use of an RCT design to assess the impact of SMS messages on improving egg consumption and IYCF practices. Although the trial could not be fully blinded, research team members who recruited participants and conducted the data collection were blinded to group allocation. Incorporating feedback from the mothers during the formative research before the main trial was critical to designing an intervention with the potential for success. Furthermore, this trial contributed to emerging mHealth research findings by ensuring that the intervention included targeted rather than blanket messages and that SMS messages were complemented by other SBC approaches rather than a stand‐alone method.

A study limitation is that the trial was only conducted in one district, which reduces the generalisability of the results, particularly in hill and mountain regions of Nepal. The district selected for recruitment, however, has a population from diverse ethnic and socioeconomic backgrounds and is quite like other terai districts. Another limitation may have been that some participants had technological challenges such as being unfamiliar with the use of mobile phones, not being regular text message users, frequently changing their phone number/SIM card changes to receive better network and promotional prices or switching off their phones for long periods of time, and so on. These factors may have interfered with the success of the intervention, making measurement of trial outcomes difficult.

As the first RCT to test the effectiveness of text messaging to improve child diets, the results will help guide donors, implementers and governments about the value of these types of interventions for improving child nutrition. The forthcoming results from this trial and its process evaluation will also inform further research on mNutrition.

AUTHOR CONTRIBUTIONS

All authors designed the research study collaboratively. Aman Sen Gupta, Mohammad Masudur Rahman and Shraddha Manandhar conducted mixed‐methods research. Mohammad Masudur Rahman and Shraddha Manandhar conducted the qualitative analyses and Kenda Cunningham and Aman Sen Gupta will lead the quantitative analyses. Kenda Cunningham, Mohammad Masudur Rahman and Aman Sen Gupta drafted the manuscript. All authors reviewed and provided input into multiple drafts and reviewed and approved the final version of this manuscript.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

ETHICS STATEMENT

This study received ethical approval from both the Nepal Health Research Council and the University of South Carolina Institutional Review Board. Written informed consent is collected before each instance of data collection.

Supporting information

Supporting information.

Cunningham, K. , Pandey Rana, P. , Rahman, M. M. , Sen Gupta, A. , Manandhar, S. , & Frongillo, E. A. (2023). Text messages to improve child diets: Formative research findings and protocol of a randomised controlled trial in Nepal. Maternal & Child Nutrition, 19, e13490. 10.1111/mcn.13490

DATA AVAILABILITY STATEMENT

Data available on request from the authors.