FormalPara Key Summary Points

This qualitative study utilized semi-structured concept elicitation interviews and a supplementary real-time data capture task to explore the perceptions of people with obesity or type 2 diabetes (T2D) around appetite, eating behaviors and drivers/triggers of food choices

Participants identified 22 drivers/triggers (including health, culture/heritage, location and stress) and discussed the associations between drivers/triggers of food choice and aspects of eating behaviors (such as appetite, hunger, cravings, satiety and motivation/determination)

Findings confirm prior research regarding impact of drivers/triggers on food choice in people with obesity or T2D and indicate underlying disease state does not appear to influence eating behaviors in people with stable body weight

The study highlights the importance of future research to assess eating behaviors in the context of clinical trials, since understanding how individuals with obesity and T2D perceive eating behaviors could be critical to the success of chronic weight management and T2D treatments

Introduction

According to the Centers for Disease Control and Prevention (CDC), the estimated prevalence of obesity in adults in the USA in 2020 was nearly 42% [1]. Obesity or overweight (defined as accumulation of excess fat that is detrimental to health) result from an imbalance between the intake and expenditure of calories [2]. It may be related to any number of biological, lifestyle or environmental factors, including genetic predisposition, irregular or unhealthy eating, limited physical activity, inadequate or poor-quality sleep, chronic stress and medications that promote weight gain [1]. People with obesity or overweight (hereafter referred to as people with obesity) have increased risk of type 2 diabetes (T2D), heart disease, stroke, osteoarthritis, obstructive sleep apnea and certain types of cancer [1, 3]. In 2019, the estimated prevalence of diabetes in adults in the US was approximately 37.1 million, with 90–95% having T2D [4]. Risk factors for T2D include obesity, age > 45 years, family history, inadequate physical activity and history of gestational diabetes, among others [5]. Diabetes-related complications can have a significant impact on individuals’ overall health and quality of life [6].

Depending on comorbid conditions, patient-specific factors and disease management needs, therapy for obesity or T2D combines lifestyle changes (e.g., diet and physical activity), pharmacologic approaches and metabolic surgery [7,8,9].

Gaining an understanding of the impact of eating behaviors and appetite can be crucial to the success of obesity and T2D treatments. Several studies have explored the influence of eating behavior concepts, environment, emotions, social norms and appetite on food choice [10,11,12,13,14,15,16]. For example, self-reported hunger was associated with irritability and anger among healthy adults including those with body mass index (BMI) > 25 kg/m2, even after controlling for BMI, dietary behavior and trait anger [11]. Emotional triggers such as work-related stress and relationship or family issues contributed to increased food consumption and weight gain in people with obesity [10]. Other studies which included people with overweight or obesity have identified factors that can prevent people from making healthy food choices, including lack of interest in nutrition, easy availability of unhealthy food, lack of time or motivation, and social influences [13, 14]. However, the impact of drivers or triggers on outcomes of food choice (i.e., dietary intake and amount of food eaten) and on eating behaviors in people with obesity or T2D is less clear.

The current research aimed to substantiate and build on the existing literature by: (1) seeking to understand and characterize the perceptions and experiences of people with obesity and/or T2D regarding appetite and eating behaviors and (2) identifying drivers/triggers of food choices.

Methods

Study Design and Participants

This non-interventional, cross-sectional, qualitative study included 45 participants (Fig. 1): Group A (people with T2D): n = 20; Group B (people with obesity, BMI ≥ 30.0 kg/m2): n = 15; Group C (people with overweight, BMI 25.0 to < 30.0 kg/m2): n = 10. Perceptions of appetite and eating behaviors were explored through in-depth, open-ended concept elicitation interviews. Prior to being interviewed, a subset of 21 participants (Group A: n = 10; Group B: n = 6; Group C: n = 5) completed a series of real-time data capture (RTDC) tasks to understand eating and food choice experiences using a smartphone app over a 5-day period.

Fig. 1
figure 1

Study design. BMI body mass index, n number of participants in a specified subgroup, N total number of participants, T2D type 2 diabetes. Group A: participants with T2D. Group B: participants with BMI ≥ 30.0 kg/m2. Group C: participants with BMI 25.0 to < 30 kg/m2

Full size image

Eligible participants were identified by referrals from endocrinologists, bariatricians or obesity medicine specialists and general practitioners and were recruited through a third-party recruitment agency. Adult (≥ 18 years) US residents fluent in English who had a stable body weight (± 5 kg) in the 3 months preceding participation were included in the study. Participants were excluded if they: (1) had a diagnosis of an eating disorder or type 1 diabetes (T1D), (2) had previously received or had upcoming surgery for obesity, (3) had participated in a clinical trial in the last 3 months or were currently enrolled in a clinical trial or (4) had received treatment for weight loss (prescription and/or over the counter) within the 3 months prior to enrollment. Participants in Group A had a clinician-confirmed diagnosis of T2D and, if taking insulin treatment, had been doing so for ≥ 6 months. Group B participants had a BMI of ≥ 30 kg/m2; Group C participants had a BMI of 25–29.9 kg/m2 and one or more weight-related comorbidities (including hypertension, dyslipidemia, obstructive sleep apnea or cardiovascular disease). Participants in Groups B and C also had a history of unsuccessful dietary effort to lose body weight and no T2D diagnosis. The RTDC subset included participants who had a smartphone with a working camera and audio-recording function and who were willing to download and complete the app-based task.

The study was conducted in accordance with the Declaration of Helsinki. Eligible participants provided informed consent prior to study start, including consent for the publication of study data, and provided demographic data before the interviews. The study was approved by the Western Copernicus Group Institutional Review Board (June 13, 2021; study number: 1310521). All study data were accessed in compliance with the US patient confidentiality requirements, including the Health Insurance Portability and Accountability Act of 1996 regulations.

Qualitative Interview Procedure and App-based RTDC Task

The 60–90-min semi-structured qualitative interviews explored perceptions and eating behaviors in participants with obesity or T2D as well as drivers/triggers that influenced their food choices. The interviews began with broad, open-ended questions to encourage spontaneous responses about participants’ perspectives on appetite and eating behaviors, followed by probing questions to ensure all topics of interest were explored.

Prior to the qualitative interviews, the participants in the RTDC subset completed a series of tasks on a smartphone-based app over a 5-day period, including both a weekday and a weekend day. Participants submitted short videos, audio clips, pictures or text responses to describe their eating and food choice experiences on each task day and completed a daily snack diary. Using Likert scales or a numeric rating scale (NRS) from 0 (low) to 10 (high), participants rated how their food choices were influenced by sleep, cravings, mood, appetite, stress and hunger on different task days. Findings from the study interviews and app-based RTDC tasks were used to develop a conceptual model presenting the key eating behaviors and related drivers/triggers of food choice in people with obesity and T2D.

Sample Size Considerations and Statistical Analysis

In qualitative research, sample size is typically determined based on the goal to achieve ‘concept saturation’ (the point at which no new concepts are likely to emerge with further interviews). Research suggests, in a relatively homogeneous population, approximately 85% of all concepts will be elicited after ten interviews [17,18,19]. As such, a minimum sample size of ten participants for each subgroup was planned.

All interviews were audio-recorded and transcribed verbatim. Transcripts were fully de-identified and anonymized and entered into ATLAS.ti (a data and content analysis software; version 9) for analysis. Thematic analysis involved tagging codes to segments of textual data within each transcript to facilitate comprehension of the data themes within and between each participant transcript. An iterative coding process was implemented, which involved creating an initial coding scheme that was subsequently updated following regular team discussions and applied throughout the iterative analysis process.

All statistical analyses were descriptive. Participant socio-demographic and clinical characteristics were summarized for each subgroup and the total sample. Continuous variables (e.g., age and BMI) were summarized using total values, means and range. Categorical variables (e.g., sex or weight-related comorbidities) were summarized using frequencies and percentages. Subgroup analyses were conducted to understand the experiences of participants in the three study groups. Differences or trends in findings across the total sample according to socio-demographic/clinical characteristics (e.g., clinical subgroup, sex, race, ethnicity or BMI) were qualitatively analyzed. The audio, video and written free text responses provided in the app-based RTDC tasks were transcribed verbatim and included in the analysis. Descriptive analysis was conducted on data from the Likert scale and NRS responses.

Concept saturation was examined once all interview transcripts had been analyzed. The principle of concept saturation was applied to the total interview sample (N = 45) and to each clinical subgroup (Groups A, B and C) to ensure all important concepts had been elicited and sufficiently explored. Saturation analysis was conducted for the appetite/eating behavior concepts and drivers/triggers reported by participants during the interviews. This involved grouping the interviews chronologically into equal sets of 11 or 12 participants and comparing the concepts emerging from each additional set of interviews using a saturation grid. In each set, any new spontaneous concept that had not emerged from previous interviews was highlighted. Saturation was achieved where no new spontaneous concepts emerged in the final interview set.

Results

Socio-Demographic and Clinical Characteristics

Overall, the mean age (range) of the participants was 49.4 (24–74) years (Table 1). Sixty percent of the participants were female, 51.1% lived with a spouse/partner and 46.7% lived in the Midwest region of the US. A good representation of race was achieved across the full sample and clinical subgroups: African American/Black: 31.1%; White, 28.9%; Asian: 22.2%; other races: 17.8%. A majority (71.1%) were non-Hispanic, although some representation of Hispanic participants was obtained across the three clinical subgroups (Group A: 25.0%; Group B: 33.3%; Group C: 30.0%; Table 1). Of the total sample, 60.0% worked outside their home > 4/5 days a week, and the majority (68.9%) worked fixed/consistent hours. Most of the participants (88.9%) were omnivores. At the time of the study, 66.7% did not follow a specific diet plan or program. The average BMI (range) was 32.7 (21.5–55.6) kg/m2 for Group A, 41.2 (30.3–55.2) kg/m2 for Group B and 28.5 (26.5–29.8) kg/m2 for Group C. Participants across all subgroups most frequently reported hypertension and dyslipidemia as weight-related comorbidities (60.0% and 40.0%, respectively; Table 1).

Table 1 Participant socio-demographic and clinical characteristics
Full size table

Participant Experiences of Eating Behaviors

Participant experiences of hunger: Most participants (43/45; 95.6%) reported hunger as a factor influencing their food choices (Fig. 2). Many described their perceptions of physical or emotional experiences of hunger (35/45 each; 77.8%), for which they used terms such as “growling stomach” or “emptiness in the stomach;” others described feeling a physical need to eat, feeling irritable or angry with hunger, i.e., “hangry,” or eating when stressed (Table 2). Five of 21 participants (23.8%) who completed the app-based RTDC task noted that hunger had influenced their meal choice on the task day, opting for convenient/quick options or leftovers when hungry (Table S1). Fifteen of the RTDC app-task participants (71.4%) mentioned that hunger influenced their decisions while grocery shopping, such as buying more food or making less healthy choices when hungry (Table S1).

Fig. 2
figure 2

Participant experience of hunger, cravings and satiety. N number of participants. Note: each concept is stratified by the number of participants who discussed the concept either spontaneously or when probed by the interviewer

Full size image
Table 2 Participant perceptions and experiences of eating behaviors
Full size table

Participant experiences of satiety: Most participants (44/45; 97.8%) considered that satiety influenced their food choices and experienced satiety as a physical sensation (Fig. 1, Table 2). Nearly half (21/44; 47.7%) of the participants described satiety as “not being able to eat any more” (Table 2). Of the 29 participants who were asked, 19 (65.5%) described satiety as an emotional experience and mentioned “feeling upset” from being “too full.” Seven out of 12 participants who responded (58.3%) in the RTDC subset mentioned feeling full after eating (Table S1). When asked to rate their hunger on the NRS from 0 to 10 after finishing a meal on the task day, 19 participants (90.5%) rated their hunger as low (between 0 and 3) from having eaten enough food or feeling satisfied or full.

Participant experiences of cravings: Of the 44 participants who were asked about cravings, 43 (97.7%) perceived them to influence their food choices, with nearly two-thirds (28/43; 65.1%) describing cravings as a strong desire for a specific food (Table 2). Sweet foods, salty foods or specific meals or snacks most frequently induced cravings (Table 2), and participants listed not having eaten the food for some time or exposure via media as some of the reasons for their cravings. Two-thirds of the participants in the RTDC subset (14/21; 67.0%) reported having a craving that day, specifically for unhealthy, high carbohydrate or fatty foods, such as chocolate, potato chips and sweets (Table S1).

Drivers or Triggers of Food Choice

Participants reported a total of 22 drivers/triggers of food choices in the interviews, both spontaneously and when probed (Fig. 3). Nearly all participants (43/45; 95.6%) considered health to be a key driver for their food choices (Table 3). Participants discussed the influence of medical diagnoses (such as obesity or T2D) on food choices, in addition to other health considerations such as food intolerances or allergies, current age and cholesterol levels. Most participants (42/45; 93.3%) discussed the influence of culture and heritage on food choices; many (34/42; 81.0%) noted that food provided a connection to their culture or occupied a central role in their family lives (Table 3). On the other hand, a few participants (6/42; 14.3%) did not consider food to be important to their identity or culture. Most (41/45; 91.1%) perceived that their location (such as being at a restaurant, work or home) influenced their food choices and mentioned eating less healthy food while visiting other people’s houses or while traveling (Table 3).

Fig. 3
figure 3

Drivers/triggers of food choice. N number of participants. Note: each driver/trigger is stratified by the number of participants who discussed the concept either spontaneously or when probed by the interviewer

Full size image
Table 3 Key drivers/triggers and impacts influencing food choice
Full size table

Of 45 participants, 40 (88.9%) noted that stress from work, family life or being busy influenced their food choices and the amount of food they ate (Table 3). Stress tended to increase participants’ cravings and the likelihood of eating unhealthy foods that might be fried, sweet, salty or carbohydrate rich while reducing motivation to make healthier food choices. Among the 21 participants in the RTDC subset, 3 (14.3%) reported severe or moderate stress and consequently selected convenient foods or ate larger servings than usual. Mood was perceived to influence food choice, appetite and motivation by 37 participants (82.2%; Table 3). Both negative (feeling depressed, bored or anxious) and positive (being in a good mood) moods led participants to make less healthy food choices. In the app-based RTDC task, one participant noted that feeling relaxed or happy on the task day helped them make healthier food choices, while another opted for comfort food because of stress.

Most of the participants (36/45; 80.0%) perceived that social environment, such as being around family members, friends, partners, co-workers or being alone, influenced their food choices (Table 3). Participants noted that being with health-conscious friends or family members motivated them to eat healthy foods. Many of the participants who were asked (33/43; 76.7%) mentioned that work influenced their food choices; some participants opted for quick and convenient foods when they had limited time to prepare or eat food when working (Table 3). Similarly, the overall convenience or ease of availability of food influenced food choices for 33 participants (73.3%), with some preferring to eat unhealthy food if it was easy or quick to prepare. Of the 43 participants who were asked, 32 participants (76.7%) perceived that engaging in physical activities influenced their food choices; exercise motivated them to eat healthier or made them feel more hungry afterwards (Table 3). Similar findings were highlighted in the app-based RTDC task, with many participants (9/21, 42.9%) noting that physical activity increased their appetite or promoted healthier eating. More than half of the participants (25/45; 56.0%) believed poor quality of sleep influenced their food choices, often leading them to choose quick or convenient foods (Table 3). Nineteen participants (43.2%) perceived that their medication either reduced (e.g., diabetes medication) or increased (e.g., birth control) their appetite (Table 3). Participants reported additional drivers or triggers that influenced their food choices, including holidays and special occasions (19/45, 42.2%), enjoyment (i.e., foods they enjoyed eating [12/45, 26.7%]), seasons (11/45, 24.4%), financial reasons (10/45, 22.2%) and routine (4/45, 8.9%).

Some differences were observed between the clinical subgroups. A total of 14 participants in Group A (70.0%), 7 in Group B (46.7%) and 1 in Group C (10%) considered that their obesity and/or T2D had influenced their food choices, such as eating healthier foods or avoiding foods high in sugar or carbohydrates (Table 3). Furthermore, more participants in Group A (14/20, 70.0%) and Group B (11/15, 73.3%) reported convenience or availability as a driver/trigger of food choice relative to Group C (2/10, 20.0%).

Concept Saturation

For the total interview sample (N = 45), all eating behavior concepts and most of the drivers/triggers emerged in the first two sets of interviews (13/15; 86.7%), with the majority emerging in the first set of interviews (12/15; 80.0%; Table S2). Furthermore, most concepts and drivers/triggers were reported spontaneously in all four saturation sets (11/15; 73.3%), except medication, culture/heritage, sleep and mood. Medication was reported less frequently than some of the other drivers/triggers and was first reported spontaneously in the third set of interviews by one participant only. Additionally, culture/heritage was probed with a direct question at the start of the interview and therefore did not have any spontaneous counts across the sample. However, probed counts for this driver/trigger appeared in all sets of interviews, suggesting that this experience was fully explored (Table S2). The findings were similar at the clinical subgroup level. The results of this saturation analysis provide confidence that the eating behavior concepts and drivers/triggers of food choices were fully explored, and saturation was achieved within the total sample.

Conceptual Model

Based on the findings of the concept elicitation interviews, a conceptual model of eating behaviors and related drivers/triggers in obesity and T2D was developed (Fig. 4). The model presents the relationships between core eating behaviors and the drivers/triggers of eating as described by participants. The concept elicitation interviews identified seven core eating behavior concepts reflecting two levels of eating behaviors (Fig. 4). The first level contains underlying concepts which can prompt eating (i.e., appetite, hunger, cravings and satiety) as well as motivation/determination, which participants discussed in relation to making healthier food choices or following a meal plan. The second level relates to direct outcomes of food choices, including dietary intake (i.e., what food was chosen for a meal or snack) and amount of food eaten.

Fig. 4
figure 4

Conceptual model of eating behaviors and related drivers/triggers in obesity and T2D. n number of participants, N total number of participants, T2D type 2 diabetes. Drivers/triggers endorsed by five or more participants are presented in the conceptual model. Each driver/trigger is presented with the proportion of participants (n/N, %) who reported it to influence eating behaviors. Descriptors of a driver/trigger are provided with the number of participants who reported each during the interviews. Superscript numbers indicate participant-reported relationships/associations between drivers/triggers and eating behavior concepts

Full size image

The 22 drivers/triggers perceived by participants to impact these eating behaviors were organized into health-related, environmental or situational, and emotional categories within the model (Fig. 4). Overall, the drivers/triggers most frequently reported by individuals with T2D and obesity were health considerations, location and stress. Probed discussions around culture/heritage during the interviews also identified this as an important driver/trigger for food choices and the role that food plays in forming a sense of identity.

To supplement and provide additional context to the conceptual model, the number of participants reporting relationships between individual drivers/triggers and specific eating behavior concepts was examined (Table S3). The most prominent driver/trigger and eating behavior associations included those between health conditions and impact on diet composition and quality (n = 26), stress and the amount of food eaten (n = 22) and the impact of stress (n = 21) and social environment (n = 21) on diet composition and quality.

Discussion

This study sought to understand the perceptions and experiences of people with obesity and T2D regarding appetite and eating behaviors, including the influence of drivers/triggers on food choices. The study findings confirmed previous research regarding the key drivers and triggers of eating behaviors. Most study participants discussed the influence of hunger, cravings and satiety on food choices, supporting these as key concepts of interest for understanding eating behaviors in people with obesity and T2D. This is consistent with previous findings highlighting the impact of hunger, cravings and satiety on food choices and the amount of food eaten [11, 12, 15]. The study findings help understand people’s relationship with eating behaviors, overweight and obesity, which can inform future studies of pharmacologic and non-pharmacologic treatments in weight management and provide insights into the concepts related to patient experience that are important to assess in clinical studies for weight management.

Participants in the present study most frequently discussed health, culture/heritage, location and stress as drivers/triggers of food choices. Several participants perceived that their medical diagnosis and being physically active led them to make healthier food choices, possibly reflective of the participants’ desire to improve their health condition. Interestingly, many participants mentioned that being around health-conscious people motivated them to eat healthy food, suggesting that social support could help patients with obesity or T2D achieve their health-related goals [20]. Participants associated health conditions, stress and social environment with diet and considered that stress influenced the amount of food eaten. These findings build upon existing models of factors that influence food choices including physiologic factors (e.g., hunger, appetite, weight status), psychologic factors (e.g., personality, motivation, emotion) and personal experiences [21].

The drivers/triggers (and associated experiences) discussed by participants with either obesity or T2D were generally consistent across the sample, suggesting that underlying disease state does not seem to influence eating behaviors differently in people with stable weight. Nevertheless, notable patterns were observed for certain subgroups. For example, participants in Group A (with T2D) were more likely to discuss how health concerns relating to their diagnosis led them to eat healthier foods compared to those with obesity (Groups B and C). This finding may reflect participants’ overall awareness of the chronic nature of T2D and the perception that obesity is not a chronic condition.

The qualitative study design provided direct insights into eating and appetite behaviors from the perspective of individuals with obesity and T2D. However, qualitative interviews typically rely on spontaneous feedback from participants recalling recent or past experiences, which may not completely or accurately reflect their day-to-day experience. A subset of participants was therefore asked to participate in an app-based RTDC task over a 5-day period prior to the interviews to provide additional insights into eating behaviors and supplement the discussions during the interviews. Daily diaries allow real-time data capture and are believed to increase ecologic validity and circumvent recall bias [22, 23]. Being able to complete the app-based task at home on a smartphone possibly offset the impact of participants wanting to provide socially desirable responses during the interviews. The increasing use of this methodology in food-based research studies reflects the shift toward patient-oriented data capture, as daily diaries may help monitor changes in eating behaviors over time [24].

Previous studies have proposed conceptual models of eating behaviors and concepts [15, 16, 21, 25]. For example, Chen and Antonelli proposed a framework of factors that determine food choice, such as social and physical environments, biologic factors, physiologic and psychologic aspects, attitudes and preferences towards food, among others [21]. Rozjabek and colleagues illustrated the association between eating concepts (hunger, appetite, satiety and cravings) and their influence on both quantity and choice of food [15]. The conceptual model proposed in the current study adds to the previous models and relates the eating behaviors that drive food choice (appetite, hunger, cravings, satiety and motivation/determination) with the direct outcomes of food choice (dietary intake and amount of food eaten). The model also includes associated drivers/triggers that participants perceived to influence eating behaviors and food choice. Whereas the previous studies included healthy adults [21] or people with overweight or obesity [15, 16], the current study reports perceptions of participants with T2D in addition to those with obesity or overweight. The proposed conceptual model therefore provides a comprehensive picture of perceptions of eating behaviors and outcomes of food choice among people with obesity and T2D. The heat map provides an additional level of detail and complexity to the conceptual model by demonstrating the most prominent associations between drivers/triggers and eating behaviors.

A strength of this study is that interviewers were trained to facilitate discussions and promote a judgment-free environment for participants to feel comfortable talking about food and food choices. Broad, open-ended questions were asked up front to encourage discussion and to allow the participants the freedom to talk about their experiences, considering the sensitivity of some of the topics discussed in the interviews. The use of a smartphone-based app task helped supplement the discussions with participants during the interviews. However, the study has its limitations. A sampling quota approach was taken to obtain a representative study sample. As there was no quota for the US region, the Southern US states were not represented because of recruitment feasibility at site locations. Furthermore, qualitative studies necessitate a smaller sample relative to quantitative studies. These factors potentially limit the generalizability of the study results. More than half (60.0%) of the Group A sample with T2D had a BMI of ≥ 30 kg/m2. Therefore, the experiences and perceptions of individuals with T2D with lower BMI may not be sufficiently represented in the study sample. However, the majority of individuals with T2D within the general population are considered to have overweight or obesity [26,27,28], suggesting that the study sample is likely to be reflective of a typical T2D population.

Conclusion

Overall, the findings from the qualitative interviews and app-based RTDC task confirm the relevance and importance of hunger, cravings and satiety as key eating behavior concepts. The study addresses gaps in the current literature and provides in-depth insights into the influence of various drivers/triggers on outcomes of food choice (i.e., dietary intake and amount of food eaten) in people with obesity and T2D. The findings also highlight the importance of future studies to assess eating behaviors in the context of clinical trials, since understanding how individuals with obesity and T2D perceive eating behaviors could be critical to the success of chronic weight management and T2D treatments. Future studies with a larger study size may provide valuable insights adding to these findings, tease out differences at the subgroup level and be more generalizable to a broader population.