The Food4toddlers study - study protocol for a web-based intervention to promote healthy diets for toddlers: a randomized controlled trial

Primary outcomes of the study are child diet quality and food variety assessed at baseline and after the intervention.

Secondary outcomes include the food and eating environments conceptualized as: parental feeding practices, family meal settings (frequency of meals, meal distractions), food choice, awareness of the food environment (at home and in the grocery store), availability and accessibility of food at home, food preparation and planning, and child weight and length.

This study is a randomized controlled trial to evaluate the effect of the Food4toddlers intervention, in which the intervention group has access to the Food4toddlers intervention website and the control group does not, see Fig. 1. Children in the intervention and control groups receive their usual care at community child health centers, which normally includes three consultations with a health care nurse for children between 12 and 18 months of age. The study started in August 2017 and is ongoing.

Participating parents complete questionnaires at baseline, post-intervention (end of intervention, after 6 months), and at two follow-ups (i.e., when their child turns 2 and 4 years old). The intervention runs in waves, and the first group started the intervention in September 2017. New groups were started every month through February 2018.

The study population comprised children close to 12 months and one of their parents. To be included in the study, parents had to have a child born between August 2016 and April 2017 and the parents had to be literate in Norwegian. Participants were recruited via Facebook. A short video was launched on Facebook with a link to the project website containing information about the project and the opportunity to sign up. The recruitment period lasted 5.5 months from mid-August 2017 to January 2018. In total 404 parents were recruited. The month before the child reached age 12 months, the enrolled parent received an e-mail with a link to a questionnaire. Three reminders on e-mail were sent to non-responders the following weeks. We included a total of 298 parents who responded to more than half of the survey questions. After they had completed the baseline questionnaire, participants were randomized according to an SPSS-generated randomization list prepared of NCØ, of 500 to the intervention and control group (SPSS version 24.0). The first author was the one who enrolled participants and assigned participants to the intervention group and control group. Among the participants who answered baseline and post intervention questionnaires, ten participants were selected to receive a gift card of 1000 Norwegian kroners. Demographic characteristics of the sample are provided in Table 1.

Sample size was calculated for one of the primary outcomes, child diet quality. As we have no data on healthy eating score for the Norwegian toddler population, we used the data of Angelopoulos and colleagues [43], which showed a mean healthy eating score of 60.5 among children (SD 9.0). We considered a 3-point difference in such a score between the intervention and control groups to be relevant from a public health perspective. We calculated that 142 children in each group would be required to demonstrate statistical significance with a statistical power of 80% and α of 5%. Assuming loss to follow-up of 40%, we aimed to recruit 237 parents in each group.

This study was developed using the basic steps from the Model of Planned Promotion for Population Health, which recognizes the importance of evidence- and theory-based intervention planning [44]. The model builds on the Theory of Planned Behavior [45] and Social Cognitive Theory [46]. Health behavior theories have had a major focus on cognitive determinants, but newer models are addressing the relationship between behavior and environment [47, 48]. As suggested by Brug and colleges [44], a focus on how to promote action rather than mere motivation is emphasized. The present intervention is in line with the social cognitive theory, targeting the interwoven relationship between the person, the behavior, and the environment [49], with an emphasis on environmental factors. Research suggests that environments influence us at a basic level of which we are unaware and that we do not monitor [29, 50]. In this project, we aim to make parents aware of how the environment influences them and render them more conscious about the over 200 food choices they make on behalf of their children throughout their daily routines [29].

With the Food4toddlers intervention, we aim to influence child diet quality and food variety by targeting the main caregivers, the parents, and their awareness of the food and eating environments. The intervention outline was developed based on a literature review, extensive user involvement, and in-depth thematic discussions among the project group. Users in the development phase were parents of toddlers and health care nurses who were involved in several steps of the development of Food4toddlers. The first step in development was to contact public health nurses to get an overview of the questions that parents tend to ask about diet and nutrition, potential challenges, and how parents might perceive the potential need for online information. Three interviews were conducted (one face to face, two by telephone). We further conducted a focus group interview with health care nurses at their workplace, followed by an individual telephone interview to further elaborate on what health care nurses perceived as the most customary questions asked by parents regarding diet. One of the nurses worked in a disadvantaged community that included a large non-native population with low SES.

Our next step was to invite the parents of toddlers to a focus group interview to share and discuss the information that they were lacking and would find useful for improving the diet and food environment for their children. In total, five focus group interviews for parent groups were conducted, one at the university, one in a home setting, and three in settings where parents meet for other reasons (e.g., baby singing class). Two telephone interviews with mothers were conducted separately. Both parents attended the interviews conducted in the home. The remaining interviews were conducted among mothers only. Approximately 40% of focus group participants were non-native individuals. The most common questions and comments in interviews with users and health care nurses confirmed the main topics that were already planned for incorporation in the intervention. However, in line with the results of the interviews, the intervention was changed to include more focus than originally planned on spicy and exotic foods for the whole family, the right amount of different foods, self-eating skills, and family meal settings.

Based on a review of the literature, feedback from users, and discussions among the project group, the intervention was framed based upon three concepts: “the plate” (i.e., the food that is actually offered to the child), “the house” (referring to food that is available and accessible at home and parental feeding practices and food preparing skills), and “the grocery store” (parental awareness of the influence of environmental cues and how to make healthy choices).

A prototype website was developed, and pilot tested with 14 participants in February 2017. The content of Food4toddlers was further refined based on this pilot test, recommendations from health authorities, and updated research in the field.

The website was developed using NEO Learning Management System. Two Masters students in Multimedia and Educational Technology at the University of Agder created a technical layout of the website, and the project group produced the content. The information provided on the website all relates to creating healthy food and eating environments for toddlers.

The homepage of the website contains an informational video about the website and information on why small changes in diet during the early years of a child’s life may be important in the long term. This page also gives some practical information on how to navigate the website and how to use the same information on a smartphone app. There is no difference in usability between the website and the smartphone app.

The website comprises four main elements: modules covering an introduction and seven topics on promoting healthy food and eating environments for the child, recipes, a discussion forum, and general information about food and beverages (the “Good to know” section), as shown in Table 2. When participants first accessed the website, not all content of the modules was visible to them, only the first two chapters. During the intervention period, access was expanded regularly (20 times) to include new content on the website; at this time, all the participants in the same wave received an e-mail with a link to the newly available information.

The first module contains information about the website and the project. The other seven modules contain two to four chapters. For each chapter, general information and tips and strategies to promote healthy behaviors are provided. In addition, one or two recipes, usually thematically linked to the topic, are recommended. The chapters also contains a video about unconscious choices while shopping [51], a game, eight quizzes, six explanatory figures, and some links to recommended websites (e.g., http://​www.​matportalen.​no).

Out of a total of 31 recipes, 30 recipes were developed by three Masters students in Public Health at the University of Agder, in cooperation with the project group (Table 2). The focus for recipe development was to inspire the preparation of healthy meals for the whole family. The age span covered in the intervention is the period in which children should be able to eat the same foods as the rest of the family. The ingredients used should be available at a local supermarket. It was possible to print the recipes. For nine of the recipes, short instructional videos were developed to inspire parents to prepare the foods and to make the preparation process easier. The videos lasted from about 1 to 3 min and were produced by undergraduate students in Multimedia Technology and Design at the University of Agder.

It was possible for participants to post questions and share information (e.g., recipes) with each other on a discussion forum. A project worker answered questions, usually within 3 working days. Participants who joined the same group had access to the same forum.

In the interviews with health care nurses and end-users, some issues about special nutrients and dishes where discussed, including salt, nitrites, cinnamon, and foreign foods (such as sushi). We listed information about these issues together with information on honey, potatoes, food additives, and cod liver oil. The information given was based on National Health Authority recommendations.

Several behavioral change methods where included on the website, to improve the child’s diet through parental awareness of the child’s food and eating environments [52]. One method was belief selection. The messages on the website were designed to strengthen positive beliefs, weaken negative ones, and introduce new beliefs (i.e. reinforce the importance of family meals and highlight the importance of repetition of new foods) that are in line with the theory of planned behavior [53]. The active learning method included in this intervention are activity-based experiences; i.e. use of videos as a way to enhance cooking skills, as well as different quizzes [49]. Persuasive communication can include messages created in such a way as to be familiar and not too discrepant for participants [54]. The importance of small changes was highlighted on the website, and familiar settings were discussed (e.g., sitting as a family at the dining table). As mentioned, not all of the information on the website was immediately available to participants in the intervention group from the beginning. Revealing information gradually on the website over a span of time can enhance retention through repetition as well as the level of interest in and persuasiveness of the information [54]. Modeling is a method that can reinforce the desired action [49]. The website features videos with actors, who are in the same age group as participants, modeling desired behaviors. Our aim was to highlight barriers and facilitators and empower parents to make changes in their environment. The outcome might be that the environment for the child is created in a way that makes it easier to take action or reduces barriers to action [49].

The primary outcome of this trial is the child’s overall diet and food variety (Table 3). Parents reported frequencies of intake of a variety of food normally eaten in Norway. Categorical scales ranging from 1) “never/less than every week” to 8) “five times a day” were used for food items from a national Food frequency questionnair (FFQ) [12] and 1) “never” to 6) “three times a day” for items from the MoBa study [55]. The secondary outcomes include parental feeding practices, family meal setting, food choice, awareness of the food environment, availability and accessibility of food in the home, food preparation and planning, and child weight and length. See Table 3 for specification about continuous and categorical variables. Most of the instruments used have previously been used in Norway or other countries and have been validated and in addition, some new questions were added. The new items about meal distraction are categorical variables with response alternatives from “disagree” to “agree” on a five-point scale: i.e. “I often look at the mobile phone during meals”. To measure the food environment three different categories of questions were used. The first type of questions relates to how available different foods are in the nearby shop and at home (i.e. fruit or whole grain biscuits) with response alternatives on a four-point scale from “not available” to “very available”. The second type of questions are statements on why they chose the way they do. The parents should respond on a 5 point-scale of “disagree” to “agree” on statements like: “I buy more if the shop is tidy and neatly organized”. The third category of questions include where in the house food is stored (“very accessible i.e. on the shelf”, “accessible i.e. in a cupboard/freezer/fridge”, “not accessible i.e. stored away in the basement/freezer/cupboard”). A test-retest was conducted in 2018 among 30 parents in kindergarten responding to these new questions twice with 2–3 weeks apart to test the reliability. The results showed a mean correlation of r = 0,551.

The website contains information about which modules participants have used, how many times they have entered the modules, and the date and duration of each session. This information will form a part of the descriptive measurements in the study.

For this protocol paper we present descriptive statistics of sample characteristics. All analyses were performed by using IBM SPSS Statistics 25.0.

Intervention effects on child diet, will be examined by use of mixed models (Linear Mixed Models and Generalized linear mixed models) with time as within factor (differences between baseline and post-test, follow-up 1, 2 and 3, respectively) and condition (intervention group, control group) as between-group factor. All models of pre-post outcomes will be adjusted for baseline values to account for regression to the mean effects. Using mixed models allows for use of incomplete data at the different follow-ups and thereby increase statistical power. We will present both crude and adjusted results. Intervention effects on both primary and secondary outcomes will be adjusted for the following variables: parental SES, BMI and age, and child gender and age (variables known from previous research to potentially confound such associations). To examine potential moderating effects such as parental SES (lower versus higher education level), a three-way interaction effect (time*condition*moderator) will be investigated for each outcome. As loss to follow up is expected, loss to follow-up-analyses will be performed, analyzing those lost to follow up compared to those remaining in the study. This will be done to identify if there are characteristics specific of those lost to follow up important to interpret the results.

The data will be stored securely on a password-protected computer with no connection between the data and personally identifiable information. The data will be available after project completion.