A sequential, five-step, mixed-methods approach, determined a priori, was followed for designing and refining messages that support behaviors and choices consistent with the BDG recommendations (see Fig. 1). These messages were designed to be applicable to a wide demographic and were aimed at urban adults. This research methods included:
-
1.
Step 1: Content extraction
-
2.
Step 2: Audience analysis
-
3.
Step 3: Expert review panel
-
4.
Step 4: Message development, theoretical underpinning and message refinement
-
5.
Step 5: Test of content validity
Step 6 tested the face validity of the messages with the target audience. Results from this phase of the study are reported in a subsequent article.
The study was granted ethical approval on the 27th of September 2017 by the Ethics Committee of the University of São Paulo (69,039,117.0.0000.5421) and data was collected between October 2017 and August 2018. This study was conducted on behalf of the Brazilian Ministry of Health. The methodological approach was informed by previous literature but also took into consideration time and financial constraints.
Step 1: Content extraction
Content extraction was fundamental for identifying excerpts from the BDG that would make appropriate material for the messages. The BDG content broadly encompasses guidelines/recommendations for a healthy diet along with quantitative and qualitative evidence to justify why these guidelines are important to follow. For this exercise, only recommendations from the BDG were extracted. Two researchers (NK and PdMS), working independently, read through all chapters of the BDG to identify, extract and record relevant sections, verbatim, into an Excel spreadsheet along with information on the chapter and the themes that the excerpt addressed. The two extraction sheets were compared to include any missing information, remove redundancies and to compile a final list of relevant content.
Step 2: Audience analysis
A formative literature review was undertaken to map beliefs, knowledge and perceptions of adult Brazilians living in urban and semi-urban settings, with regards to healthy eating. Awareness, current dietary patterns, perceived barriers and opportunities towards healthy eating were also assessed. The aim of this step was to better understand the target population of these messages and use this information to design the messages that informed them about healthy practices and provided justification for adopting these practices while countering misconceptions. The databases LILACS, FSTA, Ebsco, SCIELO and Pubmed were searched from their inception to October 2017 for quantitative and qualitative articles published in either English or Portuguese using the key words food practices, beliefs about food, beliefs about processed food, attitudes towards food. The search was conducted by a study researcher (JRGN) and was geographically restricted to Brazil. A narrative review was conducted to summarize the findings.
Step 3: Expert review panel
A panel of experts was invited for a three-hour discussion to help: (i) determine the completeness of the extracted BDG content, (ii) assess the hierarchy of importance of the content, (iii) provide feedback on results from the audience analysis, and (iv) share suggestions for organizing and presenting messages. The four-member panel was composed of researchers who were part of the scientific committee tasked with the development of the original BDG. The experts studied the socio-cultural dimensions of eating practices, and had previously developed various interventions and data collection instruments based on the BDG and therefore were extremely familiar with its content. Study objectives and results from the audience analysis were presented by the study team at the start of the session along with a list of the content extracted from the DG. Experts first recorded their opinions on the relative importance of the content and followed that up with a more detailed discussion to elaborate on their opinions, highlight new information to be included and illustrate ways to contextualize the messages that incorporated learnings from the audience analysis. The discussions were audio-recorded and JRGN served as the notetaker, capturing all suggestions made and decisions reached.
Step 4: Message development, theoretical underpinning and message refinement
Steps 1–3 set the foundation for the creation of a message library that broadly aimed at improving access and comprehension of the BDG and encouraging and facilitating behavior change. Messages were designed to improve awareness and knowledge about healthy dietary patterns, highlight practices and skills, and present solutions and opportunities to improve food choices.
The development of the messages was guided by best practices related to message length, content, structure, style, tone, framing and wording [19, 21,22,23,24]. Message development was also informed by the Theoretical Domains Framework (TDF). The TDF is an integrated theoretical framework that synthesizes 14 domains and 128 theoretical constructs from 33 theories of behavior and behavior change [25]. The domains include: Knowledge, Skills, Social/Professional Role and Identity, Beliefs about Capabilities, Optimism, Beliefs about Consequences, Reinforcement, Intentions Goals, Memory, Attention and Decision Processes, Environmental Context and Resources, Social influences, Emotion, Behavioral Regulation. The TDF encompasses the cognitive, affective, social and environmental influences on behavior and messages were designed to target these domains. Message development went through several iterations that helped refine and simplify the content. In some cases, two or more very similar messages were developed that relayed the same content. This was done intentionally with the aim of identifying the most persuasive message in Stage 5. All messages were developed in Portuguese.
A health communications specialist reviewed all messages to further refine their wording, integrate colloquial terminology without compromising on the precision of the information they communicated, and condensed message length or divided the information communicated into two or more messages to make them simpler to read.
Step 5: Test of content validity
Messages were tested for their content validity through an online survey created using Qualtrics. This exercise was undertaken to help narrow down the list of messages and to capture feedback on message content and clarity. Messages were presented in the survey in four blocks, organized thematically. A total of 47 experts from across Brazil were invited to give their anonymous feedback on one of four randomly selected blocks. Each block comprised of approximately 10 messages - block one had nine messages, block two comprised of eight messages and blocks three and four had 11 messages each. Experts were chosen from among the contributors listed in the DG, and professors and researchers who had previously published on the DG or were known to use it extensively in their research.
Input was solicited on the importance and clarity of each message [‘In your opinion, how important is the information contained in this message?’; ‘In your opinion how clear is this message?’]. Response options were recorded on a 7-point Likert scale and ranged from very important (or very clear) to not at all important (not at all clear). Suggestions for revisions to the messages were also invited [‘Would you change this message in any way to make it clearer or more convincing? You could add or exclude information or choose to divide the message.’]. Multiple messages on the same theme, with near identical content but different wording were displayed together. Participants were requested to first choose the message (or messages) that they thought were most persuasive and then rate the importance and clarity of each of their choices. Providing feedback on the messages was voluntary and no reimbursement was given to incentivize participation in the 20-min long survey.
Feedback was captured in an Excel spreadsheet and the mean values of message importance and clarity were calculated. Any suggested modifications to message wording were incorporated as appropriate. Since this was a qualitative exercise to help prioritize and shortlist messages, response to questions were compared and messages were selected based on their relative scores and after extensive discussion with members of the study team.