Designing Mobile Phone Text Messages Using the Behavior Change Wheel Framework to Influence Food Literacy in Adults With Type 2 Diabetes in Kenya: Protocol for a Systematic Development Study

Background: The worldwide prevalence of type 2 diabetes (T2D) has increased in the past decade, and it is projected to increase by 126% by 2045 in Africa. At the same time, mobile phone use has increased in Africa, providing a potential for innovative mobile health interventions to support diabetes care. Objective: This study aimed to apply the Behavior Change Wheel (BCW) framework to develop text messages to influence food literacy in adults with T2D in urban Kenya. Methods: The 8 steps of the BCW framework guided the development of text messages: (1) Define the problem in behavioral terms; (2) select target behaviors; (3) specify the target behaviors based on who needs to perform the behaviors, what needs to change, and when, where, how often, and with whom; (4) identify what needs to change; (5) identify intervention functions; (6) select policy categories; (7) select behavior change techniques (BCTs); and (8) select the mode of delivery. Recent exploratory studies in Kenya and other low-and middle-income countries provided information that was used to contextualize the intervention. Results: In step 1, the behavioral problem was defined as unhealthy


Introduction
Worldwide, the prevalence of diabetes has been on the rise in the past decade, with more than half a billion people with diabetes in 2021, of which more than 90% had type 2 diabetes (T2D) [1].Currently, in Africa, 24 million people have diabetes, and it is projected that this prevalence will rise by 126% to 55 million in 2045 [1].This projected rise in the African region will be the highest compared to all other worldwide regions [1].In Kenya, 3% of adults aged 20-79 years have diabetes, with a higher prevalence in urban (3.4%) compared to rural (1.9%) areas [2].Further, in Kenya, older age (60-69 years) is associated with a higher incidence of diabetes [2].In addition to the rising prevalence of T2D in Kenya, only 36.6% of patients achieve glycemic control [3].Glycemic control in diabetes is defined as glycated hemoglobin (HbA 1c ) levels of <7% (53 mmol/mol) [4].Poor glycemic control increases the risk of the development and progression of micro-and macrovascular complications in people with diabetes [5].As such, achieving glycemic control soon after diagnosis is an important goal in the management of diabetes.In addition to diabetes self-management medical actions, a recent prospective analysis showed that optimization of dietary patterns is an effective strategy to achieve glycemic control [6].Poor dietary quality has been assumed to be associated with a lack of food-related knowledge and skills [7].However, food literacy has been used to improve dietary behavior associated with healthy dietary patterns [8,9].Food literacy enables the application of information about food choices and critical reflection on the effect of food choices and has the potential to prevent disease, promote optimal health, and sustain the environment [9,10].
As the prevalence of T2D is on the rise, mobile phone subscription in low-and middle-income countries (LMICs) has been on the rise in the past 2 decades.Recent data show that mobile cellular subscriptions per 100 people in LMICs and Kenya rose from nearly 0% in 2000 to 104% and 123%, respectively, in 2021 [11].This increase in mobile phone subscriptions in LMICs has resulted in a commensurate rise in the use of mobile health (mHealth) in behavior change interventions [12,13].Optimization of glycemic control in adults with T2D through a food literacy intervention requires behavior change.Therefore, a systematic approach to developing an intervention and its subsequent implementation and evaluation was used [14].In this study, we used the Behavior Change Wheel (BCW) framework [15] to systematically develop intervention components.Given that the BCW framework has been used widely to design and evaluate several behavior change interventions [16][17][18][19][20], we used it to develop text messages.Specifically, the BCW framework helps in the systematic identification of target behaviors that need to be changed to improve health outcomes and to select appropriate intervention functions, policy categories, and behavior change techniques (BCTs) based on the analysis of the sources of behavior (capability, opportunity, and motivation) [14].
The BCW is a 3-stage comprehensive framework for designing complex interventions (Figure 1) that integrates behavior theory by using expert consensus and a validation process [14].The BCW framework has been designed to help understand and select relevant mechanisms of action for an intervention [15].The BCW framework uses the 3-layered capability, opportunity, motivation for behavior (COM-B) model (Figure 2) to analyze and diagnose behavior.The first, core layer of the COM-B model is further expanded to physical and psychological capability, social and physical opportunity, and automatic and reflective motivation.According to Michie et al [15], core domains are needed to increase the likelihood of performing the target behavior.COM-B is supported by the Theoretical Domains Framework (TDF), which describes 14 constructs from 33 behavior change theories [21].The second layer of the BCW framework is composed of 9 intervention functions: education, persuasion, incentivization, coercion, training, enablement, modeling, environmental restructuring, and restrictions.These intervention functions indicate how an intervention changes behavior and are linked to a taxonomy of 93 replicable BCTs (Behavior Change Technique Taxonomy version 1 [BCTTv1]) [22].The third, outermost layer of the BCW framework is composed of 7 policy categories that can be applied to support the delivery of the intervention functions.Although text messaging is increasingly used in health care and the demand for interventions based on theory, existing text messaging interventions often overlook the theoretical basis for their development [23,24].This study therefore aimed to apply the BCW framework to develop text messages to influence food literacy in adults with T2D in urban Kenya.The specific objectives of this study were to (1) define the problem affecting adults with T2D in behavioral terms, (2) define the intervention and implementation options of the identified target behaviors, and (3) develop text messages to influence food literacy in adults with T2D in Kenya.

Study Design
Developing this intervention involved 3 stages of the BCW framework: (1) understanding the behavior and user preferences, (2) identifying intervention options, and (3) identifying content and implementation options.These stages were further subdivided into 8 steps and are further explained in detail later (Figure 1).The process of intervention development was iterative and was conducted by the research team (authors MM, FK, PY, RV, TB, JLK and CM).The team consisted of dieticians, nutritionists, public health specialists, and an endocrinologist.

Stage 1: Understanding the Behavior and User Preferences
Step 1: Define the Problem in Behavioral Terms Defining the problem in specific behavioral terms included an in-depth description of the target population and the specific behavior [15].We used the recent and contextually relevant literature [25][26][27] to refine our understanding of the target population's dietary behavior to optimize glycemic control from health providers' and patients' viewpoints.
Step 2: Select Target Behaviors Target behaviors were selected by applying a systems analysis approach, as recommended by Michie et al [15].First, we used the list of facilitators of and barriers to healthy dietary behavior in adults with TDM from our previous phenomenological qualitative study conducted in Kenya [28].In that study, 30 male and female respondents were interviewed through mobile telephones, revealing that facilitators of and barriers to healthy dietary behavior are related to food literacy.We ranked the various facilitators and barriers based on the number of thematic nodes (collection of references about a specific theme) generated by NVivo software (Lumivero).Table 1 shows the ranking of thematic nodes.Second, we used the criteria recommended by BCW guidelines to prioritize the target behaviors.The criteria include the following considerations: (1) the relative impact of the behavior, (2) the likelihood of changing the behavior, (3) the potential for spilling over into other relevant dietary behaviors, and (4) the ease of measurement of the behavior.Finally, we categorized each selected behavior as very promising, promising, unpromising but worth considering, or unacceptable based on guidance by Michie et al [15].These decisions were made by consensus by the expert panel and then tabulated for each selected behavior.
Step 3: Specify the Target Behavior Step 3 involved specifying the context in which the target behavior will occur by asking specific questions: Who needs to perform the behavior?What needs to be done differently to achieve the change?Where and when do they need to perform the behavior?How often and with whom did they perform the behavior?To answer each of these questions, we used the findings from our qualitative study conducted on the target population [28].

Step 4: Identify What Needs to Change
Step 4 involved 2 substeps: (1) behavioral analysis using the COM-B model and (2) identification of what needs to change using the TDF.
In substep 1 of behavioral analysis, we mapped the specified target behavior from step 3 to the COM-B component and further explored what needed to happen for the target behavior to occur and whether there was a need for change to select and eat food known to contribute to glycemic control.
In substep 2, after determining what needs to change, we linked the behavior that needed to change based on BCW and TDF guidelines [15,29].The TDF consists of 14 domains: knowledge; skills; memory, attention, and decision processes; behavioral regulation; social/professional role and identity; beliefs about capabilities; optimism; beliefs about consequences; intentions; goals; reinforcement; environment context and resources; social influences; and emotion.The domains identified in this step were tabulated and summarized to include the following aspects for the specified behavior: (1) COM-B component, (2) what needs to happen for the target behavior to occur, (3) whether there is a need for behavior change, (4) the domain linked to the COM-B component, and (5) relevance of the domain.

Stage 2: Identifying Intervention Options
Step 5: Identify Intervention Functions Intervention functions were mapped onto each of the theoretical domains identified in step 4 [15].The intervention functions that would most likely affect behavior change were selected based on the COM-B and TDF behavior analyses conducted in step 4. The relevant intervention functions were then assessed using APEASE (acceptability, practicability, effectiveness, affordability, side effects, and equity) criteria of the BCW framework [15].The assessment included checking how the selected intervention functions meet the 5 components of the APEASE criteria.

Step 6: Select Policy Categories
The policy categories included communication/marketing, guidelines, fiscal measures, regulation, legislation, environmental/social planning, and service provision, which guide decisions made by authorities that help support and enact interventions [15].We mapped the policy categories onto the intervention functions identified in step 5 using the APEASE criteria [15].The selection of relevant policy categories was based on the matrix of links between intervention functions and policy categories described by Michie et al [15] and the APEASE criteria.The decisions on the APEASE criteria were informed by our understanding of the context, as revealed by our qualitative study [28].

Stage 3: Identifying Content and Implementation Options
Step 7: Identify BCTs Based on the intervention functions identified in step 5, we selected BCTs from the BCTTv1 [22] and the APEASE criteria.The BCTTv1 is a standardized terminology used to specify the active ingredients of behavior change interventions and consists of 93 unique BCTs [22].

Step 8: Select the Mode of Delivery
The mode of delivery was guided by the taxonomy of the models of delivery for intervention functions that involved communication.Based on the taxonomy of the models of delivery, we selected the mode of delivery by using the findings from our qualitative study [15,30].The taxonomy is structured into binary options, where we selected the most practical option using our understanding of the context and the target population [28].

Testing the Feasibility of the Intervention
The BCW guidelines recommend that developed behavior change interventions be tested for feasibility.As such, we were guided by the Medical Research Council (MRC) framework that illustrates the steps to be followed in the development and evaluation of behavior change interventions [31].

Ethical Considerations
This being a study on the development of text messages to enhance food literacy in adults with T2D, human subjects were not directly involved.However, studies [25,27,28,30] that provided evidence in the development of this study sought ethical approval.

Stage 1: Understanding the Behavior and User Preferences
Step

1: Define the Problem in Behavioral Terms
In defining the problem in behavioral terms, we considered findings from existing evidence in LMICs [25,27] and our previous studies [28,30].The existing literature, however, is limited by the fact that it was conducted in different settings and that most of these studies lacked the rigor for replication in other settings [26].Our qualitative study of facilitators of and barriers to healthy dietary behaviors among adults with T2D revealed the following facilitators: knowledge of healthy food choices, gardening, self-efficacy, food preparation skills and eating at home, education by health care workers, food availability, proximity to food-selling points, and family support.The barriers included tastes and preferences of food, health conditions barring the intake of certain foods, random eating of unhealthy food, socioeconomic status, seasonal unavailability of fruits, food safety concerns, and inaccurate beliefs and information about food and diet.In a systematic review of self-management of diabetes in sub-Saharan Africa, which included 43 studies, most of which were observational, there were misconceptions about and gaps of knowledge in what entails healthy food [27].Further, the review found that interventions on dietary behavior led to relevant improvements in healthy eating habits.Another systematic review of adherence to diabetes self-care behaviors in LMICs found that adherence to dietary recommendations ranged from 29.9% to 91.7%, while consumption of fruits and vegetables averaged 3 times per week, although the portion sizes were not revealed [25].In summary, the key gaps identified in these studies were related to our qualitative study [28], showing that dietary behavior is associated with food literacy.Food literacy is "the interrelated combination of knowledge, skills and self-efficacy on planning for meals, selecting foods, preparing food, eating and evaluating information about food with the ultimate goal of developing a lifelong healthy, sustainable and gastronomic relationship with food" [32].Food literacy enables the application of information about food choices and critical reflection on the effect of food choices and has the potential to prevent disease, promote optimal health, and sustain the environment [9,10].As such, the fundamental components of food literacy can be applied to positively influence behaviors required for healthy diet patterns.In practical terms, food literacy comprises planning, selection, preparation, and eating of food [33].Based on these findings, we defined the problem in behavioral terms as unhealthy dietary patterns among adults with T2D.
Step 2: Select Target Behaviors Based on a system analysis, we selected the following as target behaviors: evaluation of reliable sources of information about food, selection of healthy food from all food groups, preparation of good-tasting meals from any locally available food items, and control of the portion size of food (Table 2).
Based on the findings of our qualitative study on the same population [28], we prioritized the target behaviors in Table 3. a Likely impact if the behavior were to be changed.
b Ease of likelihood to change dietary patterns.c Likelihood of having an impact on other behaviors that may support a change in dietary patterns.
d Measurability either by routine data or through new data collection procedures.
e Plate model: a visual method for teaching meal planning, where the dinner plate serves as a pie chart to illustrate the proportions of the plate that should be covered by various food groups [34].

Step 3: Specify the Target Behavior
The target behavior was specified by answering the following questions, as illustrated in Table 4:

Description Specification
Adults with T2D a to plan, select, prepare, and eat healthy meals Target behavior Adults with T2D Who needs to perform the behavior?Plan, select, prepare, and eat healthy meals What is to be done differently to achieve the desired change?
Whenever eating When is it to be done?
At home, at work, or away from home Where is it to be done?Every day How often do they need to do it?Alone or with family or friends To whom do they need to do it?a T2D: type 2 diabetes.

Step 4: Identify What Needs to Change
Identification of what needs to change was informed by findings drawn from the opinions in our qualitative study [28], in addition to recent and contextually relevant findings from Uganda [26].Table 4 summarizes what needs to change based on the behavioral diagnosis using the COM-B model.A total of 10 TDF domains were identified: skills; knowledge; memory, attention, and decision processes; behavioral regulation; environmental context and resources; social influences; goals; intentions; reinforcement; and emotion.The behaviors diagnosed were further expanded to illustrate their relevance (Table 5).Call health care workers when searching for information.
Understand that some cultural values are unhealthy.

Social influences Yes Challenge cultural beliefs regarding food choice.
Social support in the selection, preparation and eating of healthy food.

Social influences
Yes: some families provide support to adults with T2D. a Awareness to family members involved in meal preparation on the preparation of healthy meals.
Change perceptions on the amount of food to be eaten.

Social influences Yes
Change cultural habits regarding food portion size.

Reflective motivation
Encourage the intention to search for information through reliable sources.c Findings from our qualitative study [28].

Intentions
d N/A: not applicable (because there is no need for change).

Stage 2: Identifying Intervention Options
Step 5: Identify Intervention Functions Based on the APEASE criteria, a total of 4 intervention functions were mapped onto the 10 TDF domains identified in step 4. The identified intervention functions included education, training, environment restructuring, and enablement (Table 6).

Physical opportunity
Adding objects to the environment Enable environmental restructuring to use a smaller plate to serve meals.

Social opportunity
Restructuring the social environment Enable environmental restructuring to modify cultural beliefs on food that affect the choice of food.

Social influences
Practical social support Enable to challenge sociocultural habits regarding food portion sizes.Social influences Practical social support Enable awareness creation among family members involved in meal preparation on how to prepare healthy meals.

Prompts and cues Educate how to clarify information, when needed. Intentions
Behavioral goal setting Enable establishment routines and habits to eat healthy meals.Goals Behavioral goal setting Enable establishment routines and habits of observing optimal portion sizes when serving food.

Goals
Behavioral goal setting Enable planning to serve correct portion sizes.Goals Action planning Enable planning to cook healthy meals.Goals Self-monitoring of behavior Educate to prioritize the purchase and eating of healthy food.Intentions Information about health consequences Educate on foods to avoid for optimal glycemic control.

Automatic motivation
Practical social support Persuade to look for information when in doubt.Reinforcement Practical social support Enable to establish meal plans to ensure the intake of healthy meals.Goals a COM-B: capability, opportunity, motivation for behavior.Step 6: Identify Policy Categories Policy categories were selected based on our qualitative study after analysis using the APEASE criteria: (1) Most participants (90%) had an income of less than 400 euros (US $ 422.27) per month, and this had limited the control of fiscal decisions; (2) all participants owned a basic mobile phone; and (3) the process of developing guidelines by the Ministry of Health in Kenya is structured and involves multiple stakeholders, making it likely to be limited by time.Based on this, we identified communication and regulation as the most practical policy categories.

Stage 3: Identifying Content and Implementation Options
Step 7: Identify BCTs BCTs are the "irreducible, observable, and replicable components of an intervention designed to redirect behaviour" [35].In this study, we selected 9 BCTs from the taxonomy of behavior change [36] to develop intervention content aimed at influencing dietary behavior (Table 5).These BCTs included behavioral goal setting, action planning, self-monitoring of behavior, practical social support, instruction to perform the behavior, information about health consequences, prompts and cues, restructuring the social environment, and adding objects to the environment.

Step 8: Select the Mode of Delivery
Based on findings from our qualitative study [28], most of the adults with T2D in the target population owned and used mobile phones.Additionally, our systematic review revealed that using text messages in LMICs is associated with a clinically significant effect on HbA 1c levels, in addition to being cheap and easy to use irrespective of socioeconomic status, and is not affected by racial disparities [30,37,38].As such, we selected mobile text messages as the mode of delivery [28].
The text messages were developed based on Abroms et al's [39] 4 steps for developing a text messaging program: (1) formative research, (2) design, (3) pretest, and (4) revision.In this study, we modified steps 1 and 2, as shown in Figure 4. Specifically, we modified step 2 to include 3 substeps: (2a) linking the identified intervention function to relevant Kenyan and international guidelines and recommendations for a healthy diet [42][43][44][45][46][47], (2b) structuring guideline content to the relevant BCT, and (2c) gain-or loss-framing the text message.Gain-framed health promotion messages emphasize the benefits of engaging in a certain behavior, while a loss-framed message emphasizes the consequences of failing to participate in the behavior [46].The messages were developed through a consultative process that included MM, CM, and FK.Table 7 illustrates a sample of the developed text messages.Table 7. Sample text messages to be used in the mHealth a intervention.

Physical capability
Instructions on how to perform the behavior Be able to select healthy food

•
To meet nutrient needs within calorie limits, choose a variety of nutrient-dense foods across and within all food groups in recommended amounts.Energy-dense foods include vegetables, fruits, whole grains, beans, nuts and seeds, and lean protein when prepared with little or no saturated fats, added sugars, and sodium.

•
Most processed carbohydrate foods release glucose more quickly than whole-grain carbohydrates.Eat whole-grain carbohydrates, which produce a slower rise in blood glucose levels, which are called low-glycemic-index (GI) foods and result in better blood glucose control.

Information about health consequences
Know healthy food.

•
The amount of carbohydrates in your meal has the greatest effect on blood glucose levels.Avoid big servings of carbohydrates in your meals and ensure most of your energy-giving foods are composed of whole meal or high fiber.Make vegetables and fruits take the larger portion of your plate.

•
There are 2 main types of fat: saturated and unsaturated.Excessive saturated fat in foods, such as fatty meat, sausages, and butter, can increase the amount of cholesterol in the blood, which increases your risk of developing heart disease.

Information about health consequences
Use advanced planning skills to ensure the selection of healthy food.
• Select whole or minimally processed foods, which help control your blood glucose.

•
Make an eating plan each week: this is the key to fast and easy meal preparation.This will also ensure that you plan for healthy diets and better-controlled blood glucose.

Information about health consequences
Know the composition of the plate model.

•
Start with a 9-inch dinner plate.Fill half with nonstarchy vegetables, such as managu, Sukuma wiki, green beans, broccoli, cauliflower, cabbage, and carrots.Fill one-quarter with a lean protein, such as chicken, fish, legumes, beans, or eggs.Fill one quarter with whole-meal carbohydrate foods: grains, starchy vegetables (eg, potatoes and peas), rice, pasta, beans, fruit, and yoghurt.A cup of milk also counts as a carbohydrate food.

Social opportunity
Restructuring the physical environment Change cultural habits on the portion size.
• Controlled portion sizes of food are important to reduce calorific intake.To control your portion sizes, plan to use a smaller plate size when serving meals.

•
Your health is a personal responsibility.When eating, try to avoid places that may entice you to eat excessive food.Eat most of your meals at home for better control of healthy food and your blood sugar.

Information about health consequences, prompts/cues
Intend to select a healthy diet.
• Make your shopping list ahead of time and do not go shopping while hungry.This helps you buy healthier items but also saves money and helps you select healthier foods.

•
Instead of purchasing processed grade 1 maize flour or Unga that is processed, use posho-milled Unga or whole-meal flour.Posho-milled flour contains fiber and other plant extracts that help in blood sugar control.

Information about health consequences, prompts/cues
Intend to prioritize eating a healthy diet.
• Small changes to more nutrient-dense, single food and beverage choices combine to make a nutrient-dense meal and can lead to a whole day of nutrient-dense meals and snacks, increasing your blood sugar control.

•
Place fresh and locally available low-carbohydrate fruits on the table or a place that is easy to reach to increase intake of fruits.This will enable you to eat healthy and avoid unhealthy snacking.Instructions on how to perform a behavior Establish routines and habits to select a healthy diet.

Automatic motivation
• Replace white rice with whole-grain foods or brown rice.This ensures better blood sugar control.
Prompts/cues Establish routines and habits to select a healthy diet.

•
Eat smaller meals more often.Eat at least 3 meals a day, with snacks in between.When you wait too long to eat, you are more likely to make unhealthy food choices.

Instructions on how to perform behavior
Establish routines and habits to serve and eat a healthy diet.
a mHealth: mobile health.
b COM-B: capabilities, opportunities, and motivation for behavior.
c BCT: behavior change technique.

Testing the Feasibility of the Developed Intervention
According to the Medical Research Council (MRC) guidance on developing and evaluating complex interventions, the development of an intervention in this study is the first of 4 steps [31].The other subsequent and iterative steps include feasibility, evaluation, and implementation.Based on this guidance, the intervention has been tested for feasibility through an exploratory trial in the target population, and the findings will be reported elsewhere.

Principal Findings
This paper described the systematic development of an mHealth intervention using mobile text messages to optimize glycemic control in adults with T2D in Kenya following the BCW framework.Preparatory analysis preceding this study revealed that adults with T2D in Kenya have unhealthy dietary patterns and lack reliable sources of information to enhance food literacy [28].The diagnosis also revealed that there are barriers to the target population's capability, opportunity, and motivation relating to food literacy.Based on the diagnosis, 4 intervention functions and 9 BCTs were selected to promote food literacy.Specific BCTs were selected to be integrated into this intervention: for example, adding objects to the environment (eg, using a smaller plate to serve food for portion control) and prompts or cues (eg, replacing white rice with whole-meal foods).The identified mHealth mode of delivery was mobile text messaging based on the practical applicability in the older population of adults with T2D.
When designing this study, we identified food literacy as a channel to address the problem of poor glycemic control.Food literacy has the potential to facilitate dietary behavior change through the connection of food and nutrition-related knowledge, and cooking skills [47].Considering that food literacy explicitly focuses on health literacy skills in a food context [48], a positive correlation between health literacy and diabetes knowledge has been reported in patients with T2D [49].Overall, the appropriate application of food literacy is associated with positive health outcomes in adults with T2D.
In this study, we used the BCW framework and contextual evidence to develop text messages for adults with T2D.The approach of basing the development on theory and the content of the text messages on the needs of the target population has been used variedly in recent studies [50][51][52][53].In a recent 2-country African study [50], the BCW framework was used to develop text messages in 4 phases that included exploration of primary and secondary data and focus group discussions (FGDs), which were followed by pretesting through telephone interviews.Bartlett et al [51] developed text messages for patients with diabetes in the United Kingdom based on BCTs in a process that included health care specialist workshops, FGDs, and acceptability and fidelity surveys.MacPherson et al [53] developed text messages guided by BCTs, the motivational interviewing counselling style, and the Small Steps for Big Changes approach [54].However, unlike our study in which we specifically developed text messages to facilitate food literacy, these studies [50,51,53] have focused on the wider multicomponent approach to diabetes management and prevention.The focused approach in our study was based on evidence showing that food literacy is an important problem in the target population [28], while a multicomponent approach to diabetes care is effective in optimizing glycemic control [55,56].

Strengths and Limitations
We acknowledge some strengths and weaknesses in using the BCW framework in this study.First, this study used contextual data to conduct behavioral diagnosis, which included our qualitative study of facilitators of and barriers to healthy dietary behavior in the target population, as recommended by Buchanan et al [57].Second, the BCW framework provides a systematic method for designing a theory-based intervention, starting from a behavioral diagnosis of what needs to change, followed by linking the diagnosis to intervention functions, policy categories, and BCTs to change the target behavior.This approach enables the intervention to be contextualized to the needs of the target population, which increases its chances for success [58].Third, the BCW framework uses a harmonized language of theoretical constructs and BCTs, which are important for the replication and synthesis of research and evidence [59,60].
However, this study was limited by the BCW framework's lack of a structured framework for operationalizing BCTs into mHealth methods of delivery [61], leaving it open to the creativity of the intervention developers and the context of the target population.However, to address this limitation, we modified previous guidance on the development of text messages [39] and developed messages based on target behaviors following guidelines on healthy diets that are linked to BCTs.

Conclusion
This study reported the systematic use of the BCW framework, the COM-B model, the TDF, and BCTs based on the contextual needs of the target population to develop text messages to influence food literacy in adults with T2D.The efficacy of the text messages will be evaluated through an exploratory trial in adults with T2D in the target population.

Figure 1 .
Figure 1.Stages and steps in the BCW.Adapted from Michie et al [15].BCW: Behavior Change Wheel.

Figure 3
Figure 3 summarizes findings from the 8 steps of the BCW framework.

Figure 3 .
Figure 3. Summary of findings from the steps in the Behavior Change Wheel.

•••
Who? (Adults with T2D) • Where?(At home or work) • When? (When planning, selecting, eating, or preparing food) How often?(Every time or most of the time during consumption) With whom?(Either alone or with the family) is accessible in the target population.c Be aware of where to get healthy foods (shops, markets, home garden), and establish or maintain kitchen gardens.N/A N/A No: the proportion of the target population using Use the mobile phone or computer to search for information, where possible.smartphones or computers is low.c N/A N/A No: patients commonly cook at home.c Have the necessary cooking equipment.Search and use a smaller plate to serve food.Environmental context and resources Yes Use smaller plates when serving food.
b BCT: behavior change technique.c TDF: Theoretical Domains Framework.

Table 1 .
Ranking of thematic nodes.

Table 2 .
Identification of target behaviors.

Table 3 .
Prioritization of the target behaviors.

Table 4 .
Specification of the target behavior.

Table 5 .
Behavioral analysis and application of the TDF a in diagnosis.

Table 6 .
Mapping intervention functions to corresponding COM-B a components and BCTs b .
Keep a food diary for a few days to evaluate what you eat every day.Note how you were feeling when you ate: hungry, not hungry, tired, or stressed?Create a list of cues by reviewing your food diary to become more aware of when you're triggered to eat for reasons other than hunger.Is there anything else you can do to avoid the cue or situation?If you cannot avoid it, do something different that would be healthier.Replace unhealthy habits with new, healthy ones.