This paper is in the following e-collection/theme issue:

Digital Health Reviews (1289) Engagement with and Adherence to Digital Health Interventions, Law of Attrition (491) Innovations and Technology for Healthy Eating Education (414) mHealth for Wellness, Behavior Change and Prevention (2871) Web-based and Mobile Health Interventions (3297)

Published on in Vol 26 (2024)

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/58735, first published .
Framework Development for Reducing Attrition in Digital Dietary Interventions: Systematic Review and Thematic Synthesis

Framework Development for Reducing Attrition in Digital Dietary Interventions: Systematic Review and Thematic Synthesis

Framework Development for Reducing Attrition in Digital Dietary Interventions: Systematic Review and Thematic Synthesis

Authors of this article:

Jian Wang1, 2 Author Orcid Image ;   Jinli Mahe1, 2 Author Orcid Image ;   Yujia Huo1, 2 Author Orcid Image ;   Weiyuan Huang1, 2 Author Orcid Image ;   Xinru Liu1, 2 Author Orcid Image ;   Yang Zhao3, 4 Author Orcid Image ;   Junjie Huang5 Author Orcid Image ;   Feng Shi6 Author Orcid Image ;   Zhihui Li7 Author Orcid Image ;   Dou Jiang8 Author Orcid Image ;   Yilong Li8 Author Orcid Image ;   Garon Perceval2 Author Orcid Image ;   Lindu Zhao8 Author Orcid Image ;   Lin Zhang1, 2 Author Orcid Image
Article Authors Cited by Tweetations (1) Metrics

    Review

    1The School of Public Health and Preventive Medicine, Monash University, Melbourne, Australia

    2Suzhou Industrial Park Monash Research Institute of Science and Technology, Monash University, Suzhou, China

    3The George Institute for Global Health, University of New South Wales, Sydney, Australia

    4The George Institute for Global Health, Beijing, China

    5The Jockey Club School of Public Health and Primary Care, Faculty of Medicine, Chinese University of Hong Kong, Hong Kong, China (Hong Kong)

    6East China Aviation Personnel Medical Appraisal Center, Shanghai, China

    7Tsinghua Vanke School of Public Health, Tsinghua University, Shenzhen, China

    8School of Economics and Management, Southeast University, Nanjing, China

    Corresponding Author:

    Lin Zhang, MD, PhD

    Suzhou Industrial Park Monash Research Institute of Science and Technology

    Monash University

    Room 805, 8th Floor, Building 2, Sangtian Island Science and Technology Innovation Park

    No. 1, Huayun Road

    Suzhou

    China

    Phone: 86 13426423807

    Fax:86 13426423807

    Email: lin.zhang2@monash.edu


    Background: Dietary behaviors significantly influence health outcomes across populations. Unhealthy diets are linked to serious diseases and substantial economic burdens, contributing to approximately 11 million deaths and significant disability-adjusted life years annually. Digital dietary interventions offer accessible solutions to improve dietary behaviors. However, attrition, defined as participant dropout before intervention completion, is a major challenge, with rates as high as 75%-99%. High attrition compromises intervention validity and reliability and exacerbates health disparities, highlighting the need to understand and address its causes.

    Objective: This study systematically reviews the literature on attrition in digital dietary interventions to identify the underlying causes, propose potential solutions, and integrate these findings with behavior theory concepts to develop a comprehensive theoretical framework. This framework aims to elucidate the behavioral mechanisms behind attrition and guide the design and implementation of more effective digital dietary interventions, ultimately reducing attrition rates and mitigating health inequalities.

    Methods: We conducted a systematic review, meta-analysis, and thematic synthesis. A comprehensive search across 7 electronic databases (PubMed, MEDLINE, Embase, CENTRAL, Web of Science, CINAHL Plus, and Academic Search Complete) was performed for studies published between 2013 and 2023. Eligibility criteria included original research exploring attrition in digital dietary interventions. Data extraction focused on study characteristics, sample demographics, attrition rates, reasons for attrition, and potential solutions. We followed ENTREQ (Enhancing the Transparency in Reporting the Synthesis of Qualitative Research) and PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines and used RStudio (Posit) for meta-analysis and NVivo for thematic synthesis.

    Results: Out of the 442 identified studies, 21 met the inclusion criteria. The meta-analysis showed mean attrition rates of 35% for control groups, 38% for intervention groups, and 40% for observational studies, with high heterogeneity (I²=94%-99%) indicating diverse influencing factors. Thematic synthesis identified 15 interconnected themes that align with behavior theory concepts. Based on these themes, the force-resource model was developed to explore the underlying causes of attrition and guide the design and implementation of future interventions from a behavior theory perspective.

    Conclusions: High attrition rates are a significant issue in digital dietary interventions. The developed framework conceptualizes attrition through the interaction between the driving force system and the supporting resource system, providing a nuanced understanding of participant attrition, summarized as insufficient motivation and inadequate or poorly matched resources. It underscores the critical necessity for digital dietary interventions to balance motivational components with available resources dynamically. Key recommendations include user-friendly design, behavior-factor activation, literacy training, force-resource matching, social support, personalized adaptation, and dynamic follow-up. Expanding these strategies to a population level can enhance digital health equity. Further empirical validation of the framework is necessary, alongside the development of behavior theory–guided guidelines for digital dietary interventions.

    Trial Registration: PROSPERO CRD42024512902; https://tinyurl.com/3rjt2df9

    J Med Internet Res 2024;26:e58735

    doi:10.2196/58735

    Keywords

    thematic synthesis; attrition rate; dropout; behavior change theory; digital dietary intervention; digital health; mHealth; eHealth; mobile apps; email 


    Dietary behavior significantly influences health across various populations [1]. Poor dietary habits are linked to serious diseases and substantial economic burdens [2]. Afshin et al [3] reported that dietary risks were responsible for approximately 11 million deaths and 255 million disability-adjusted life years in 2017. Additionally, unhealthy diets contribute significantly to noncommunicable diseases, which are projected to cost more than US $30 trillion globally in the next decade [4].

    Implementing universally accessible dietary interventions is a common approach to improving dietary behaviors [5], and digital technology, known for its efficacy, reach, and affordability, presents promising solutions to the associated challenges [6]. However, attrition—defined as participant dropout before completing an intervention—is prevalent in digital health or eHealth [7-9]. In some formal evaluations of app-based health interventions, attrition rates have reached as high as 75%-99% [7,9]. Many factors contribute to this high attrition rate. For example, digital dietary interventions—dietary programs implemented via digital technology—involve factors such as insufficient motivation [10], lack of interest [11], time constraints [12], inadequate guidance [13], financial constraints [14], limited health care services [15], doubts about efficacy [13], health issues [16], technical problems [17], and overwhelming demands [18]. If attrition remains high, it significantly compromises the validity and reliability of such interventions [9]. Moreover, this influence is not limited to the individual level but also potentially exacerbates health disparities across different social groups—a manifestation of digital health inequity.

    From the perspective of behavior theory, interventions aim to achieve behavior change, while attrition represents an interruption in this process. Behavior formation and development involve a multitude of factors, including individual factors (such as attitudes, self-efficacy, skills, and knowledge) and environmental factors (such as health care facilities, social networks, and policies) [19]. These factors are dynamic, arising from both fluctuations in the environment and the internal instability inherent within individuals [19]. When certain essential factors are lacking or insufficient in strength, it becomes impossible to maintain behavior change, leading to attrition [19,20]. Take dietary interventions for type 2 diabetes as an example: this is a long-term process aimed at promoting behavior change, requiring individual belief [21], self-efficacy [22], emotional support from family [23], nutrition advice from dietitians [24], accessible food environments [25,26], and supportive government policies [27]. Inadequate support from these factors can lead to discontinuation and participant attrition. When such attrition arises from common factors, like limited access to professional dietary guidance, it can result in population-level disruptions in behavior and health disparities, exemplifying health inequality [28,29]. Therefore, by viewing attrition as a multifactorial behavior disruption, and using behavior theories to identify the contributing factors and analyze their specific mechanisms, we can provide a novel perspective for understanding and addressing attrition and health disparities.

    Nevertheless, research focusing on attrition, particularly investigations into its causes and potential solutions through the lens of behavior theories, remains sparse. This study aims to bridge this gap. Through systematic review and thematic synthesis [30], it comprehensively explores and summarizes the reasons for attrition and potential solutions. These findings are then integrated with concepts from multiple behavior theories to develop a comprehensive theoretical framework. This framework will not only elucidate the behavioral mechanisms behind attrition but also guide future work in designing and implementing more effective digital dietary interventions, thereby reducing attrition rates at the individual level and diminishing health inequalities at the population level.


    Study Design

    This study uses a systematic review, meta-analysis, and thematic synthesis to investigate participant attrition in digital dietary interventions. The study protocol is available on International Prospective Register of Systematic Reviews (PROSPERO; CRD42024512902). This review follows the ENTREQ (Enhancing the Transparency in Reporting the Synthesis of Qualitative Research) guidelines for reporting qualitative syntheses [31] and the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) standards, including the PRISMA checklist in Multimedia Appendix 1.

    Search Strategy

    A thorough search across 7 electronic databases, including PubMed, MEDLINE, Embase, CENTRAL, Web of Science, CINAHL Plus with Full Text, and Academic Search Complete, was conducted using a predefined set of search terms related to attrition in digital dietary interventions, including synonyms and British spellings, and performed as full-text searches. Examples of search terms used included disengagement rate, churn rate, turnover rate, dropout rate, noncompletion rate, attrition rate, retention rate, adherence rate, compliance rate, follow-up rate, and persistence rate. This aimed to identify relevant English-language studies published from 2013 to 2023. The search period was limited to the years 2013-2023 for 3 reasons. First, 2013 marked a pivotal year in the global mobile internet landscape with the widespread adoption of 4G LTE technology, mobile devices, and applications [32], which laid the foundation for the rapid growth of digital health technologies. Second, this timeframe ensured that the data and findings were current and reflective of the latest trends and methodologies in digital dietary interventions. Third, the volume and quality of research in this field have significantly increased in recent years, providing a robust body of literature for a comprehensive review. The search strategy (see Multimedia Appendix 2) was carefully developed and executed by our experienced research team, ensuring a systematic and thorough review of the literature.

    Eligibility Criteria

    We focused on original research that either primarily or secondarily explored attrition rates in digital dietary interventions among human populations, encompassing both randomized controlled trials (RCTs) and observational studies. These interventions typically use technologies such as text messaging, social media, web-based platforms, smartphone apps, and personal digital assistants to improve dietary behaviors and support adherence to diet-related therapies, for example, managing chronic diseases and weight control [6,33-35]. Therefore, studies involving nondigital interventions were excluded. Additionally, nonoriginal studies, such as reviews, conference proceedings, commentaries, protocols, and collections, were excluded to concentrate on empirical data. Studies with minimal or unclear relevance to attrition rates were also omitted to ensure reliable data for thematic synthesis.

    Study Selection

    Zotero 6 (Corporation for Digital Scholarship), a free, open-source research management tool, was used to assist in identifying duplicates and organizing papers. The initial screening of titles and abstracts was performed to eliminate nonoriginal research or studies not pertinent to digital dietary intervention attrition rates. Full-text assessments of potentially eligible studies were then conducted to determine their inclusion based on predefined criteria. This phase excluded studies with marginal or vague relevance to attrition rates. Two independent reviewers (FS and ZL), experienced in systematic reviews and digital health, conducted the selection. Any discrepancies were resolved through discussion or input from a third reviewer (DJ).

    Data Extraction

    A standardized data abstraction form was developed by the research team to capture specific information from the included studies (see Multimedia Appendix 3). This form comprised 3 parts: the first part focused on study characteristics, including authors, year of publication, study date, targeted dietary behavior, duration, theories or behavioral techniques, study design, and intervention strategy. The second part covered sample characteristics, such as eligibility criteria, sample size, and demographics. The third part collected information relevant to attrition, including the number of participants, number of dropouts, attrition rates, reasons for attrition, and potential solutions. The form was pilot-tested on a sample of studies to ensure clarity and comprehensiveness.

    Raw data on intervention strategies, reasons for attrition, and solutions to attrition were abstracted directly from the text of the included studies. This involved line-by-line extraction of relevant excerpts from the abstracts, results, and discussion sections of the included papers, which is the first step in thematic synthesis. In subsequent steps, these excerpts were organized and analyzed to identify descriptive and analytical themes (see “Analysis and Synthesis”).

    Data extraction, carried out independently by 2 reviewers (XL and YZ) with backgrounds in public health and behavioral science, also sought consensus in the case of disagreements, facilitated by a third reviewer (JH), if required.

    Study Appraisal

    Evaluating the quality of included studies is essential prior to thematic synthesis. This process determines each study's contribution to the synthesis process, referred to as its value to the synthesis [36]. In this study, qualitative content (data) on attrition causes and solutions was extracted from the included studies. These were consolidated into descriptive themes and subsequently synthesized into higher-level analytical themes through thematic synthesis. Therefore, assessing the strength of evidence for the descriptive themes is crucial before this process.

    Although the included studies, including RCTs and observational studies, are quantitative, the evaluation focuses on the qualitative aspects of the research and, thus, should use qualitative appraisal tools. The method of Walsh and Downe [37] provides a comprehensive, flexible, and practical framework for appraising the quality of qualitative research, and it has been widely applied. Using this framework, we developed a question checklist and an evaluation form (see Multimedia Appendix 4) to appraise the quality of each study. The checklist facilitates the efficient and clear collection of essential evaluation information, while the evaluation form organizes the responses. These responses are used to assess the trustworthiness, transferability, and usefulness of each study based on 34 evaluation criteria, with points accumulated accordingly. The total score, with a perfect score of 34 points when all criteria are met, serves as the basis for rating the study’s value to the synthesis. Due to the small sample size of scores from 21 studies and their nonnormal distribution, we used the more efficient and straightforward 3-quartile method for rating. We calculated the first and second quartiles of the scores, which are 13.0 and 15.67, respectively. Based on these quartiles, we rated the studies as low (<13.0), moderate (13.0 ≤ score < 15.67), and high (≥15.67). One reviewer (YH) conducted the initial evaluation, and a second reviewer (WH) examined the ratings, with both being experienced in designing and evaluating behavior interventions. Discrepancies were resolved through discussion.

    Analysis and Synthesis

    We conducted a meta-analysis to assess the attrition rates using RStudio (version 4.3.2; Posit) and a random effects model, treating attrition rates as the effect size. Studies from RCTs were categorized into control and intervention groups, while observational studies were classified as a separate group. Sensitivity analyses and publication bias assessments were performed separately for the intervention and observational groups to evaluate the robustness of results and detect potential reporting biases.

    Thematic synthesis, involving the systematic extraction and synthesis of qualitative data from multiple studies, can offer deeper and comprehensive insights applicable across various contexts. We used this approach to synthesize descriptions of attrition reasons and solutions with NVivo (version 12; Lumivero). The process entailed: (1) line-by-line coding of data, (2) organizing codes into descriptive themes, and (3) refining themes into overarching analytical themes that elucidate attrition factors and interventions. Two experienced reviewers (JW and YL) in qualitative methods and evidence synthesis independently conducted all steps. After completing each step, they discussed interim results to reach a consensus before proceeding. This approach helped to identify and reduce discrepancies early, ensuring the reliability of the final results. Sensitivity analysis [38] tested the findings’ robustness by excluding low-quality study data and reassessing for consistent themes. This analysis confirmed the validity and robustness of our results.

    Finally, concepts from multiple behavior theories were used to construct a conceptual framework that better explains the mechanisms of attrition. This approach introduced themes of personal agency from the integrated behavior model [20], habit or impulsive behavior from the reflective-impulsive model [39], cognition and reinforcement from social cognitive theory [40], and diverse resources from the conservation of resources theory [41]. Additionally, the perceived norms theme was expanded to include both subjective or injunctive and descriptive norms, aligning with the integrated behavior model.


    Search and Selection Results

    Database searches yielded 442 studies. After excluding 55 duplicates and 45 nonretrieved records (due to access restrictions, copyright limitations, or incomplete records), 342 underwent eligibility screening based on predetermined criteria, leading to the exclusion of 321 studies for not meeting inclusion requirements (details in Multimedia Appendix 5). Ultimately, 21 studies were included in the review (Table 1), with the selection process depicted in a PRISMA flow diagram (Figure 1).

    Table 1. Summary of extracted study characteristics.
    AuthorTargeted dietary behaviorAttrition rate (%)Design types
    Brewer et al [11]Increasing the intake of fruits and vegetables among the participants—aRandomized controlled trial
    Browne et al [42]Reducing the rate of eating among children with obesity62.5Randomized controlled trial
    Cheung et al [43]The intervention targeted dietary behaviors by including text messages that promoted Australian dietary guidelines after pregnancy. This included advice on controlling carbohydrate intake and the use of low carbohydrate vegetables and foods to satiate hunger.52.5Randomized controlled trial
    Coa and Patrick [10]Behaviors related to healthy eating43.0Observational study
    Dawson et al [12]Improving renal dietary behaviors related to potassium, phosphorus, sodium, and fluid intake, and general healthy eating and lifestyle behaviors.10.3Randomized controlled trial
    Grutzmacher et al [15]Nutrition and physical activity14.3Observational study
    Hawkes et al [44]Improving diet as one of its main objectives, alongside increasing physical activity and achieving weight loss, to reduce the risk of type 2 diabetes.63.5Observational study
    Howarth et al [18]Focusing on resilience, movement, eating, and sleep—Observational study
    Jiang et al [45]Targeting optimal nutritional intake using ordinary food and oral nutrition supplements (ONS), tailored to individual needs, preferences, and diet restrictions.8.3Randomized controlled trial
    Kaul et al [13]Tracking dietary intake and identifying dietary factors that might influence pain symptoms38.9Randomized controlled trial
    Linardon et al [14]Eating disorders, particularly behaviors related to binge eating.66.0Randomized controlled trial
    Paxton et al [17]Increasing the intake of fiber, fruits, vegetables, and reducing saturated and trans fats.35.1Observational study
    Plaete et al [46]Increasing the intake of fruit and vegetable to promote healthier dietary habits among adults71.8Randomized controlled trial
    Rom et al [16]The intervention targeted behaviors associated with binge-eating disorder, focusing on establishing regular eating patterns, self-monitoring of food intake, and addressing thoughts and feelings related to eating and body image.26.3Observational study
    Schulz et al [8]Reducing alcohol consumption among adult problem drinkers52.7Randomized controlled trial
    Silina et al [47]Increasing physical activity and dietary recommendations for dyslipidemia and weight loss3.1Randomized controlled trial
    Springer et al [48]Increasing fruit and vegetable intake—Randomized controlled trial
    Van der Mispel et al [49]Increasing fruit and vegetable intake78.2Observational study
    Whitley et al [50]Healthy eating and active living behaviors4.6Observational study
    Young et al [51]Mediterranean diet70.3Observational study
    Yuhas et al [52]Reducing the intake of SSBsb by adolescents11.5Randomized controlled trial

    aNot applicable.

    bSSB: sugar-sweetened beverage.

    ‎
    Figure 1. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram.

    Study Characteristics and Appraisal

    The characteristics of the included studies (details in Multimedia Appendix 6 [8,10-18,42-52]) span several countries: global (2 studies, subsequent numbers denote study counts), Latvia (n=1), Australia (n=4), the United States (n=9), Ireland (n=1), Germany (n=1), Belgium (n=2), and China (n=1). In total, 5 studies did not specify gender distribution, while the majority featured predominantly female participants. The interventions used varied digital technologies such as SMS, mobile apps, web-based programs, and email, targeting diverse dietary objectives including carbohydrate intake (n=1), fruit and vegetable intake (n=5), the Mediterranean diet (n=1), oral nutrition supplements intake (n=1), microelement and fluid intake (n=1), diet improvement for type 2 diabetes (n=1), binge eating (n=2), alcohol reduction (n=1), sugar-sweetened beverages intake (n=1), dietary factors related to pain symptoms (n=1), eating rate for obesity (n=1), and general healthy eating habits (n=5). Ethical considerations varied, with 2 studies not reporting on ethics, 1 bypassing review for involving voluntary workplace co-designers, and 18 obtaining clear ethical approvals. Behavior theories or techniques were used in 17 studies. Intervention durations ranged from 28 days to 4 months in 13 studies and 6-12 months in 7 studies, with 1 unspecified. Attrition rates surpassed 20% in 14 studies, peaking at 78.2%.

    Among the studies, 12 were RCTs and 9 were observational. Attrition data collection methods varied, with 9 using existing digital systems, 7 detailing only calculation or timing methods, and 5 lacking clarification. Attrition causes and solutions were primarily derived from participant quotes, author interpretations, and trial evidence. Value to the synthesis faced challenges due to missing triangulation, theoretical saturation, representativeness exposition, alternative explanations, conceptual richness, and novel findings. Seven studies was rated high, 8 were rated moderate, and 6 were rated low, with the detailed appraisal process provided in Multimedia Appendix 7 [8,10-18,36,42-52].

    Meta-Analysis

    Meta-analyses were conducted for control and intervention groups within 12 RCTs, and observational studies were analyzed as a separate group, using a random effects model in RStudio (detailed data in Multimedia Appendix 8 [8,10-17,42-52]). Missing data necessitated the exclusion of 4 studies, resulting in the analysis of 9 RCTs and 8 observational studies. Attrition rate served as the effect size, with results in Figures 2-4. The mean attrition rates were 35% (95% CI 20-52) for control, 38% (95% CI 19-59) for intervention, and 40% (95% CI 21-62) for observational groups. Heterogeneity was assessed using I² and τ², revealing high heterogeneity with I² values of 94%, 97%, and 99%, and τ² values of 0.0604, 0.0940, and 0.0922, respectively, all with P<.001. The I² statistic indicates the percentage of total variation across studies that is due to heterogeneity rather than chance. τ² represents the between-study variance, providing an estimate of the actual variation in effect sizes across the included studies. By using both I² and τ², we gain a comprehensive understanding of heterogeneity.

    Sensitivity analyses, shown in Figures 5-6, used a random-effects model, revealing attrition rates of 34%-45% for intervention groups and 35%-47% for observational groups. The slight variances in 95% CIs suggest that no individual study significantly alters the overall estimate. τ² and τ values indicated minimal dispersion and variation, with values ranging from 0.0665 to 0.1074 and 0.2579 to 0.3277 for intervention groups, and 0.0643 to 0.1077 and 0.2536 to 0.3282 for observational groups, respectively. Similarly, I² values, clustering around 95%-97% for the intervention group and remaining at 99% for the observational group, reflect substantial heterogeneity yet confirm the limited impact of individual studies on the overall results. These findings indicate that the meta-analysis results are stable and not significantly affected by the inclusion or exclusion of particular studies.

    Publication bias was evaluated using Begg's and Egger's tests, alongside funnel plots (Figures 7-8). The intervention group's funnel plot suggested potential bias, though Begg's (P=.47) and Egger's (P=.20) tests were not statistically significant. The observational group's funnel plot showed more symmetry, implying less bias, supported by nonsignificant Begg's (P=.39) and Egger's (P=.70) test results. These findings indicate no significant publication bias, suggesting the meta-analysis relatively realistically reflect the actual situation.

    ‎
    Figure 2. Forest plot of attrition rates for control group.
    ‎
    Figure 3. Forest plot of attrition rates for intervention group.
    ‎
    Figure 4. Forest plot of attrition rates for observation group.
    ‎
    Figure 5. Sensitivity analysis of attrition rates in intervention group.
    ‎
    Figure 6. Sensitivity analysis of attrition rates in observational group.
    ‎
    Figure 7. Funnel plot of attrition rate in intervention group.
    ‎
    Figure 8. Funnel plot of attrition rates in observational group.

    Thematic Synthesis

    The thematic synthesis yielded 29 descriptive and 7 analytical themes, along with 19 subthemes, elucidating attrition reasons, as illustrated in Table 2. Furthermore, it identified 20 descriptive themes and 8 analytical themes, with 7 subthemes, addressing potential attrition solutions (details in Table 3). Evidence for each theme was extensively documented in Multimedia Appendix 9 [8,10,13-18,42-52], which also included descriptions not assigned to specific themes due to being nonspecific, unclear, or inefficient. Combining the themes of reasons and corresponding solutions from 2 tables, such as “motivation” with “boost and maintain motivation,” resulted in the formation of 13 merged themes. The finalized themes corresponded with the summary of Eysenbach [9] of factors influencing attrition rates (details in Multimedia Appendix 10).

    Subsequently, we drew on concepts from multiple behavior theories, resulting in 15 integrated themes. These themes were conceptually organized to elucidate their interconnections, as shown in Figure 9. This figure illustrates an explanatory framework where participant attrition is influenced by 2 main systems: the driving force system and the supporting resource system. For clarity, this integration is termed the force-resource model. The driving force system encompasses the inherent motivation or tendencies that determine the necessity of behavior execution. In contrast, the resource system provides the essential support required for the feasibility of implementing behaviors. This dichotomy leads to 2 primary causes of participant attrition: firstly, the failure of intervention strategies to generate sufficient motivation or to counteract risky habit or impulsive behaviors; secondly, the presence of inadequate or poorly matched resources. The discussion sections will provide a detailed exploration of these themes and their interplay within this conceptual framework.

    Table 2. Themes of attrition reasons from included studies.
    Analytical theme and analytical subthemeDescriptive theme
    Motivation

    		High motivation
    • Higher autonomous motivation
    • Motivation waning
    Attitude

    		Negative experiential attitude
    • Lake of interest
    • Strict timeline

    		Negative instrumental attitude
    • Limited usefulness
    • Doubt regarding efficacy

    		Positive instrumental attitude
    • Goal-connection feeling
    • Clearer expectation
    Subjective or injunctive norm

    		Lack of subjective or injunctive norm
    • No direct contact

    		With subjective or injunctive norm
    • With patient-provider relationship
    Cue

    		Distraction cue
    • Triggered by stop messages
    Reinforcement

    		Delayed reinforcement
    • Delayed feedback

    		Positive or immediate reinforcement
    • Positive feedback
    • No reimbursement
    • Immediate feedback
    Resources

    		With cognitive load
    • Overwhelmed tasks


    		Low usability
    • Technical or usability issues


    		Limited knowledge or skills
    • Limited guidance
    • Limited technical literacy

    		Service resource
    • Lack of social support
    • More health services
    • Few health services

    		Financial resource
    • Financial barriers
    • Higher income

    		Time resource
    • Time constraint

    		Personal state
    • Health or life issue
    Individual differences

    		Cultural factor
    • Cultural barriers

    		Education level
    • Low education level
    • High education level
    Table 3. Themes of attrition solutions from included studies.
    Analytical theme and analytical subthemeDescriptive theme
    Boost and maintain motivation
    • Enhance autonomous motivation
    • Enhance self-affirmation
    Improve attitude

    		Improve experiential attitude
    • Make interventions fun

    		Improve instrumental attitude
    • Educate on intervention
    Offer subjective or injunctive norm
    • Using health practitioners’ referrals
    Provide immediate reinforcement
    • Set progress markers
    • Provide immediate information
    Provide matching resources

    		Decrease cognitive load
    • Make interventions easy
    • Improve usability
    • Provide guidance

    		Provide financial resource
    • Address financial barriers

    		Improve personal state
    • Improve emotional state
    Provide social support
    • Using peer encouragement
    Personalization strategy
    • Use targeted strategies
    Personalization strategy

    		Based on feedback
    • Refine text-messages
    • Solicit user-feedback

    		Attend to individual difference
    • Screen participants
    • Understand the impact of participants’ characteristics
    • Address cultural barrier
    Dynamic intervention
    • Provide tailored follow-up
    ‎
    Figure 9. Force-resource model.

    Principal Findings

    We identified 15 interconnected theoretical themes and integrated behavior theory concepts to construct the force-resource model. As shown in Figure 9, the model comprises 2 subsystems that interact to influence behavior and contribute to attrition. The first subsystem, the driving force system, includes themes of motivation, perceived norms, personal agency, attitude, and habit or impulsive behavior, which collectively guide behavioral directions and trends. Participants are driven by this force system to engage in dietary interventions to improve health. Their behaviors are influenced by attitudes toward the diet-health connection and the efficacy of the intervention, as well as perceived norms and personal agency. This behavioral tendency underpins their initial participation and intention to persist.

    The second subsystem is the supporting resources system. The core concept of this system is resource, defined as entities either intrinsically valued or instrumental in achieving valued ends. These resources include physical and psychological states, financial support, time availability, and accessible health services [41]. Adequate resources alongside the force system likely facilitate behavior change, whereas insufficient resources obstruct it. Interestingly, an excess of resources can also contribute to attrition, as it may diminish the perceived value of the intervention. This is attributable to the diverse and competitive nature of motivation; when resources are abundant, previously unattainable desires become attainable, leading participants to pursue more appealing activities, necessitating greater cognitive resources to overcome them. For instance, in food-rich environments, pursuing weight control goals as a self-regulation process demands more cognitive resources than pursuing eating enjoyment [53]. Therefore, ensuring participants have access to appropriate and ample supporting resources is critical to prevent attrition.

    In addition to the 2 subsystems, the model includes other key components such as cues, which refer to specific environmental stimuli that trigger actions [54]; reinforcement, which increases the likelihood of a behavior by delivering a rewarding stimulus after the behavior [55]; participants’ knowledge and skills; the usability of digital interventions; and individual differences, primarily referring to stable personal attributes [56], including demographic backgrounds, personality traits, and cultural values.

    For a more detailed elucidation of the components and mechanisms of the force-resource model, refer to Multimedia Appendix 11 [8,13,15,17,18,20,39,41-43,46,52-78].

    Implications for Digital Dietary Interventions

    Overview

    The force-resource model provides a resource-matching perspective for the design of digital dietary interventions, thus, forming multifaceted behavior intervention strategies tailored to the individual, based on the characteristics and functions of the components, and their interrelationships within the model.

    User-Friendly Design

    The design of digital interventions should ensure that processes are easy to understand and use, not only by avoiding excessive tasks, such as extensive questionnaires and record-keeping, but also by fully leveraging various technological tools to help simplify these tedious tasks and reduce cognitive load. For example, digital dietary interventions can use image processing and pattern recognition to streamline dietary recording [79], and use artificial intelligence to assess daily dietary quality automatically [80].

    Behavior-Factor Activation

    Based on behavior theory, behavior change can be achieved by activating corresponding components within the force system. Taking motivation as an example, improving diet is a gradual process that necessitates persistent adherence to achieve long-term health benefits, which presents significant challenges for behavior interventions based on health motivations. We advocate for the integration of health interventions with other activities such as gaming and social interaction that provide immediate feedback, thereby potentially enhancing sustained involvement and reducing attrition. Additionally, applying immediate reinforcement can increase the likelihood or probability of the behavior, while avoiding disruptive cues and highlighting beneficial ones can foster sustained engagement.

    Literacy Training

    The aim of providing targeted training programs is to enhance 2 types of literacy in individuals. The first is digital literacy, enabling them to use digital health resources more effectively with minimal cognitive load. The second is health literacy, which facilitates the activation of behavior factors such as attitudes and self-efficacy, thereby increasing receptiveness to health interventions. Additionally, there is a complementary relationship between literacy training and user-friendly design: the former helps people better operate digital tools with varying levels of user-friendliness, while interventions with good user-friendly design can reduce the demands on literacy training. Ideally, digital tools that effectively integrate cognitive psychology, behavioral science, and human-computer interaction are designed to be intuitive, engaging, and easy to use without requiring prior training.

    Force-Resource Matching

    Regarding the resource system, it is vital to ensure the provision or conservation of resources that are compatible with the force system. For instance, the availability of professional dietary counseling significantly influences the success rate of interventions for diabetic patients. Conversely, it is pragmatic to align motivation levels with available resources, recommending the setting of achievable dietary goals accordingly. For example, individuals with limited financial resources and access to medical advice should target a balanced diet as an intervention rather than pursuing antiaging diets or precision nutrition.

    Social Support

    Rooted in social networks [81], social support offers emotional, instrumental, and informational assistance [82]. Guidance and education fall under the category of informational support, which can alleviate cognitive processing demands while positively influencing attitudes, self-efficacy, and motivation. Emotional support has a wide-ranging impact, as it can enhance subjective well-being and cognitive functioning [83,84], as well as influencing attitudes [57,58], perceived norm [85], self-efficacy [86], and motivation [39,57,83,87,88]. Instrumental support facilitates access to financial and service-related resources, while fostering community and motivation, thus, reducing attrition risk [18,50].

    Personalized Adaptation

    There are significant variations in the characteristics and attributes of resource and force systems among individuals, arising not only from their personal state and circumstances but also from individual differences. Digital interventions should, therefore, tailor strategies to accommodate these variations, thereby enhancing resonance [89]. For instance, adapting dietary messages to reflect personal emotions and cultural eating habits can maintain engagement across diverse demographics [43]. Personalized adaptation is integral to multiple stages of digital interventions, including design, implementation, evaluation, and optimization, and artificial intelligence holds great potential in addressing this challenge.

    Dynamic Follow-Up

    The resource and force systems are dynamic, arising from fluctuations in both the environment and individuals’ internal states [57,90]. As mentioned earlier, environmental cues can easily trigger distractions and competing motivations, contributing to participant attrition. This implies that static interventions struggle to accommodate such variability. To address this issue, interventions should incorporate real-time adaptability, providing timely and tailored actions [91]. For instance, automated prompts encouraging re-engagement could redirect disengaging users back into the program if metrics indicate disengagement [45]. Additionally, for long-term monitoring of large populations, the rapid data processing capabilities of artificial intelligence can be fully used.

    Implications for Digital Health Equity

    Participant attrition arises from mismatches between individual resource and force systems. When this phenomenon expands to population level, it essentially creates a form of digital health inequity due to disparities in access to digital health resources. Digital health equity strives for equitable access to and use of resources such as digital health technologies, training programs, digital health care systems, and community support structures, all designed to improve health outcomes universally [92,93]. In promoting digital health equity, reducing attrition rates is a key strategy [92-96]. This encourages us to broaden the goal of reducing attrition to encompass a larger population and to design solutions at more comprehensive levels based on the multilevel determinants in digital health [92,93]. We have proposed several strategic recommendations based on the force-resource model, as detailed in Multimedia Appendix 12 [92-96].

    Complementary Findings

    Through a meta-analysis of attrition rates in digital dietary interventions over the past decade, we found that the average attrition rates ranged from 35% to 40%, representing a significant barrier to the efficacy and generalizability of such interventions, irrespective of the study design or the presence of an active intervention component. Notably, the intervention group exhibited a marginally higher attrition rate compared to the control group, with overlapping CIs, suggesting a lack of effectiveness of the investigated interventions in mitigating participant attrition in the included studies. This observation highlights the need for more potent and tailored strategies to promote sustained engagement.

    The high degree of heterogeneities underscores the inherent complexity and diversity of factors influencing attrition rates in digital dietary interventions, including variations in study designs, intervention components, participant characteristics, and implementation contexts. This indicates significant room for improvement in standardization of digital dietary interventions. Taking participant characteristics as an example, some may join interventions out of curiosity without a genuine understanding or interest, making them unsuitable for the target group and likely to drop out quickly. These varied and mixed participant characteristics. The “run-in and withdrawal” strategy can mitigate this challenge by including an initial phase where all participants start the intervention [9]. This run-in phase helps identify those less likely to adhere. Participants not fully engaged or committed can then be excluded, leaving a more homogeneous and dedicated group for the remainder of the study. By ensuring a more homogeneous and committed participant group, and standardizing selection criteria, this strategy effectively reduces heterogeneity, thereby leading to more consistent and reliable assessments of intervention effects.

    Strengths

    Most research on digital interventions prioritizes efficacy, often treating attrition rates as a mere data point rather than a subject of in-depth analysis. This review stands as the first to scrutinize attrition rates within digital systems through the lens of behavior theories, introducing a force-resource model to explore underlying causes and identify possible solutions. It also expands these findings to enhance digital health equity. The insights gained provide a foundational understanding and innovative strategies for developing more effective digital dietary interventions and promoting digital health equity.

    Limitations and Future Direction

    First, despite performing a thorough search across 7 electronic databases, we acknowledge the limitation in capturing the entire body of literature related to attrition in digital dietary interventions due to the challenge of matching search terms precisely with indexed keywords. To mitigate this, we expanded the search terms and used a comprehensive full-text search strategy. However, there remains the potential for missing studies, as many do not specifically emphasize attrition, leading to its underrepresentation in indexed keywords. This limitation could affect the completeness of our review and underscores the need for more refined search methodologies in future research.

    Second, the observed heterogeneities in meta-analysis outcomes highlight a critical issue stemming from significant study variances, emphasizing the need for enhanced standardization and harmonization of protocols and components in digital dietary interventions. Establishing behavior theory–guided guidelines and best practices for designing, implementing, and evaluating these interventions could lead to more consistent and replicable results, thereby improving their generalizability and impact.

    Third, this study developed a theoretical framework to clarify attrition mechanisms and guide digital dietary interventions. However, due to the aforementioned limitations and the continuous introduction of new digital tools and strategies in this field, the framework's ability to encompass and explain currently undiscovered and future emerging scenarios remains uncertain. This necessitates further empirical validation and exploration in future research.

    Last, reducing attrition rates is a key strategy for promoting digital health equity, encouraging us to expand the findings related to attrition reduction to the population level of digital health equity. However, some unique factors at the population level, such as cultural diversity, social structures, and communication patterns, with significant influence on digital health equity, have not been fully explored in this study. This highlights potential directions for future research.

    Conclusions

    High attrition rates compromise the effectiveness and sustainability of digital dietary interventions. This review has pioneered the examination of participant attrition in such interventions through the lens of behavior theories, introducing the force-resource model. This framework conceptualizes attrition via the interaction between the driving force system and the supporting resource system, offering a nuanced understanding of participant attrition, summarized as insufficient motivation and inadequate or poorly matched resources. It highlights the critical necessity for digital dietary interventions to dynamically balance motivational components with available resources, ensuring interventions are both compelling and practically feasible for participants. Key recommendations include user-friendly design, behavior-factor activation, literacy training, force-resource matching, social support, personalized adaptation, and dynamic follow-up. These strategies are crucial for promoting sustained engagement and reducing attrition rates. Additionally, by expanding these strategies to a population level, this study contributes to the broader goal of promoting digital health equity.

    The comprehensive and reliable validation of this framework requires further empirical investigation, especially with the continuous emergence of new digital tools and strategies in this field. Concurrently, developing behavior theory–guided guidelines for the design, implementation, and evaluation of digital dietary interventions is imperative to enhance their standardization and effectiveness. Furthermore, research should explore the impact of cultural diversity, social structures, and communication patterns on digital health interventions to ensure they are inclusive and effectively address the needs of diverse populations. This comprehensive approach will help overcome barriers to successful intervention outcomes and improve overall public health equity.

    Authors' Contributions

    JW drafted the manuscript. L Zhang and JW were responsible for the concept and design of the study. FS, ZL, and DJ were involved in the search and study selection. XL, YZ, and JH extracted all data. YH and WH conducted the study appraisal. JW and YL performed the analysis and synthesis. Meta-analysis was conducted by JM and JW. GP conducted a comprehensive English language review and polishing. L Zhao coordinated and managed the organization of the study. All authors participated in developing the review’s methodology, contributed to multiple manuscript drafts, and gave their approval for the final version. Authors L Zhao (ldzhao@seu.edu.cn) and L Zhang (lin.zhang2@monash.edu) are co-corresponding authors for this article.

    Conflicts of Interest

    None declared.

    Multimedia Appendix 1

    PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) 2020 checklist.

    DOC File , 94 KB
    Multimedia Appendix 2

    Search strategies.

    DOC File , 55 KB
    Multimedia Appendix 3

    Standardized Data Abstraction Form.

    DOCX File , 18 KB
    Multimedia Appendix 4

    Question Checklist and Evaluation Form for Study Quality Appraisal.

    DOC File , 77 KB
    Multimedia Appendix 5

    Papers Excluded From Analysis.

    DOC File , 182 KB
    Multimedia Appendix 6

    Study Characteristics.

    DOC File , 95 KB
    Multimedia Appendix 7

    Study Quality Appraisal.

    DOC File , 306 KB
    Multimedia Appendix 8

    Data for Meta-analysis.

    DOC File , 73 KB
    Multimedia Appendix 9

    Thematic Synthesis.

    DOC File , 122 KB
    Multimedia Appendix 10

    Analytical Themes and Eysenbach's Attrition Factors.

    DOC File , 46 KB
    Multimedia Appendix 11

    Force-Resource Model.

    DOC File , 43 KB
    Multimedia Appendix 12

    Implications for Digital Health Equity.

    DOC File , 36 KB
    1. Brownell KD, Cohen LR. Adherence to dietary regimens. 2: Components of effective interventions. Behav Med. 1995;20(4):155-164. [CrossRef] [Medline]
    2. Peeters A, Blake MRC. Socioeconomic inequalities in diet quality: from identifying the problem to implementing solutions. Curr Nutr Rep. 2016;5(3):150-159. [CrossRef]
    3. GBD 2017 Diet Collaborators. Health effects of dietary risks in 195 countries, 1990-2017: a systematic analysis for the global burden of disease study 2017. Lancet. 2019;393(10184):1958-1972. [FREE Full text] [CrossRef] [Medline]
    4. Jacka FN, Sacks G, Berk M, Allender S. Food policies for physical and mental health. BMC Psychiatry. 2014;14:132. [FREE Full text] [CrossRef] [Medline]
    5. Sjöström M, Stockley L. Toward public health nutrition strategies in the European union to implement food based dietary guidelines and to enhance healthier lifestyles. Public Health Nutr. 2001;4(2A):307-324. [CrossRef] [Medline]
    6. Chen Y, Perez-Cueto FJA, Giboreau A, Mavridis I, Hartwell H. The promotion of eating behaviour change through digital interventions. Int J Environ Res Public Health. 2020;17(20):7488. [FREE Full text] [CrossRef] [Medline]
    7. Meyerowitz-Katz G, Ravi S, Arnolda L, Feng X, Maberly G, Astell-Burt T. Rates of attrition and dropout in app-based interventions for chronic disease: systematic review and meta-analysis. J Med Internet Res. 2020;22(9):e20283. [FREE Full text] [CrossRef] [Medline]
    8. Schulz DN, Candel MJ, Kremers SP, Reinwand DA, Jander A, de Vries H. Effects of a web-based tailored intervention to reduce alcohol consumption in adults: randomized controlled trial. J Med Internet Res. 2013;15(9):e206. [FREE Full text] [CrossRef] [Medline]
    9. Eysenbach G. The law of attrition. J Med Internet Res. 2005;7(1):e11. [FREE Full text] [CrossRef] [Medline]
    10. Coa K, Patrick H. Baseline motivation type as a predictor of dropout in a healthy eating text messaging program. JMIR Mhealth Uhealth. 2016;4(3):e114. [FREE Full text] [CrossRef] [Medline]
    11. Brewer D, Dickens E, Humphrey A, Stephenson T. Increased fruit and vegetable intake among older adults participating in Kentucky's congregate meal site program. Educ Gerontol. 2016;42(11):771-784. [FREE Full text] [CrossRef] [Medline]
    12. Dawson J, Campbell KL, Craig JC, Tong A, Teixeira-Pinto A, Brown MA, et al. A text messaging intervention for dietary behaviors for people receiving maintenance hemodialysis: a feasibility study of KIDNEYTEXT. Am J Kidney Dis. 2021;78(1):85-95.e1. [CrossRef] [Medline]
    13. Kaul U, Scher C, Henderson CR, Kim P, Dyhrberg M, Rudin V, et al. A mobile health + health coaching application for the management of chronic non-cancer pain in older adults: results from a pilot randomized controlled study. Front Pain Res (Lausanne). 2022;3:921428. [FREE Full text] [CrossRef] [Medline]
    14. Linardon J, Shatte A, Rosato J, Fuller-Tyszkiewicz M. Efficacy of a transdiagnostic cognitive-behavioral intervention for eating disorder psychopathology delivered through a smartphone app: a randomized controlled trial. Psychol Med. 2022;52(9):1679-1690. [CrossRef] [Medline]
    15. Grutzmacher SK, Munger AL, Speirs KE, Vafai Y, Hilberg E, Braunscheidel Duru E, et al. Predicting attrition in a text-based nutrition education program: survival analysis of Text2BHealthy. JMIR Mhealth Uhealth. 2019;7(1):e9967. [FREE Full text] [CrossRef] [Medline]
    16. Rom S, Miskovic-Wheatley J, Barakat S, Aouad P, Kim M, Fuller-Tyszkiewicz M, et al. The acceptability, feasibility, and preliminary efficacy of a supported online self-help treatment program for binge-eating disorder. Front Psychiatry. 2023;14:1229261. [FREE Full text] [CrossRef] [Medline]
    17. Paxton RJ, Hajek RA, Newcomb P, Dobhal M, Borra S, Taylor WC, et al. A lifestyle intervention via Email in minority breast cancer survivors: randomized parallel-group feasibility study. JMIR Cancer. 2017;3(2):e13. [FREE Full text] [CrossRef] [Medline]
    18. Howarth A, Quesada JA, Donnelly T, Mills PR. The development of 'Make One Small Change': an e-health intervention for the workplace developed using the person-based approach. Digit Health. 2019;5:2055207619852856. [FREE Full text] [CrossRef] [Medline]
    19. Wang J, Shao J, Zhang S, Wang L, Zhang L. Comprehending health behavior change and maintenance: a systematic review and meta-synthesis of behavior theories. Am J Health Educ. 2024:1-15. [CrossRef]
    20. Glanz K, Rimer BK, Viswanath K. Health Behavior and Health Education: Theory Research and Practice. Hoboken, New Jersey, US. John Wiley & Sons; 2008.
    21. Sharifirad G, Entezari MH, Kamran A, Azadbakht L. The effectiveness of nutritional education on the knowledge of diabetic patients using the health belief model. J Res Med Sci. 2009;14(1):1-6. [FREE Full text] [Medline]
    22. Strychar I, Elisha B, Schmitz N. Type 2 diabetes self-management: role of diet self-efficacy. Can J Diabetes. 2012;36(6):337-344. [CrossRef]
    23. Albanese AM, Huffman JC, Celano CM, Malloy LM, Wexler DJ, Freedman ME, et al. The role of spousal support for dietary adherence among type 2 diabetes patients: a narrative review. Soc Work Health Care. 2019;58(3):304-323. [FREE Full text] [CrossRef] [Medline]
    24. Siopis G, Colagiuri S, Allman-Farinelli M. Effectiveness of dietetic intervention for people with type 2 diabetes: a meta-analysis. Clin Nutr. 2021;40(5):3114-3122. [CrossRef] [Medline]
    25. Sarkar C, Webster C, Gallacher J. Are exposures to ready-to-eat food environments associated with type 2 diabetes? A cross-sectional study of 347 551 UK biobank adult participants. Lancet Planet Health. 2018;2(10):e438-e450. [FREE Full text] [CrossRef] [Medline]
    26. Cradock KA, ÓLaighin G, Finucane FM, McKay R, Quinlan LR, Martin Ginis KA, et al. Diet behavior change techniques in type 2 diabetes: a systematic review and meta-analysis. Diabetes Care. 2017;40(12):1800-1810. [CrossRef] [Medline]
    27. Ley SH, Hamdy O, Mohan V, Hu FB. Prevention and management of type 2 diabetes: dietary components and nutritional strategies. Lancet. 2014;383(9933):1999-2007. [FREE Full text] [CrossRef] [Medline]
    28. McCartney G, Popham F, McMaster R, Cumbers A. Defining health and health inequalities. Public Health. 2019;172:22-30. [FREE Full text] [CrossRef] [Medline]
    29. Gakidou EE, Murray CJ, Frenk J. Defining and measuring health inequality: an approach based on the distribution of health expectancy. Bull World Health Organ. 2000;78(1):42-54. [FREE Full text] [Medline]
    30. Thomas J, Harden A. Methods for the thematic synthesis of qualitative research in systematic reviews. BMC Med Res Methodol. 2008;8:45. [FREE Full text] [CrossRef] [Medline]
    31. Tong A, Flemming K, McInnes E, Oliver S, Craig JC. Enhancing transparency in reporting the synthesis of qualitative research: ENTREQ. BMC Med Res Methodol. 2012;12:181. [FREE Full text] [CrossRef] [Medline]
    32. Cisco. Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2014-2019 [Internet]. San Jose, CA. Cisco Systems; 2015. URL: https:/​/www.​cisco.com/​c/​en/​us/​solutions/​collateral/​service-provider/​visual-networking-index-vni/​white-paper-c11-520862.​html [accessed 2024-07-03]
    33. Allman-Farinelli M, Gemming L. Technology interventions to manage food intake: where are we now? Curr Diab Rep. 2017;17(11):103. [CrossRef] [Medline]
    34. Fakih El Khoury C, Karavetian M, Halfens RJG, Crutzen R, Khoja L, Schols JMGA. The effects of dietary mobile apps on nutritional outcomes in adults with chronic diseases: a systematic review and meta-analysis. J Acad Nutr Diet. 2019;119(4):626-651. [CrossRef] [Medline]
    35. Rose T, Barker M, Maria Jacob C, Morrison L, Lawrence W, Strömmer S, et al. A systematic review of digital interventions for improving the diet and physical activity behaviors of adolescents. J Adolesc Health. 2017;61(6):669-677. [FREE Full text] [CrossRef] [Medline]
    36. Greaves C, Poltawski L, Garside R, Briscoe S. Understanding the challenge of weight loss maintenance: a systematic review and synthesis of qualitative research on weight loss maintenance. Health Psychol Rev. 2017;11(2):145-163. [FREE Full text] [CrossRef] [Medline]
    37. Walsh D, Downe S. Appraising the quality of qualitative research. Midwifery. 2006;22(2):108-119. [CrossRef] [Medline]
    38. Carroll C, Booth A, Lloyd-Jones M. Should we exclude inadequately reported studies from qualitative systematic reviews? An evaluation of sensitivity analyses in two case study reviews. Qual Health Res. 2012;22(10):1425-1434. [CrossRef] [Medline]
    39. Strack F, Deutsch R. Reflective and impulsive determinants of social behavior. Pers Soc Psychol Rev. 2004;8(3):220-247. [FREE Full text] [CrossRef] [Medline]
    40. Bandura A. Annals of child development. Six theories of child development. In: Vasta R, editor. Social Cognitive Theory. Stamford, Connecticut, USA. JAI Press; 1989.
    41. Hobfoll SE. Conservation of resources. A new attempt at conceptualizing stress. Am Psychol. 1989;44(3):513-524. [CrossRef] [Medline]
    42. Browne S, Kechadi M, O'Donnell S, Dow M, Tully L, Doyle G, et al. Mobile health apps in pediatric obesity treatment: process outcomes from a feasibility study of a multicomponent intervention. JMIR Mhealth Uhealth. 2020;8(7):e16925. [FREE Full text] [CrossRef] [Medline]
    43. Cheung NW, Blumenthal C, Smith BJ, Hogan R, Thiagalingam A, Redfern J, et al. A pilot randomised controlled trial of a text messaging intervention with customisation using linked data from wireless wearable activity monitors to improve risk factors following gestational diabetes. Nutrients. 2019;11(3):590. [FREE Full text] [CrossRef] [Medline]
    44. Hawkes RE, Miles LM, Ainsworth B, Ross J, Meacock R, French DP. Engagement with a nationally-implemented digital behaviour change intervention: usage patterns over the 9-month duration of the national health service digital diabetes prevention programme. Internet Interv. 2023;33:100647. [FREE Full text] [CrossRef] [Medline]
    45. Jiang X, Chen J, Yuan X, Lin Y, Chen Y, Li S, et al. Feasibility of an individualized mHealth nutrition (iNutrition) intervention for post-discharged gastric cancer patients following gastrectomy: a randomized controlled pilot trial. Nutrients. 2023;15(8):1883. [FREE Full text] [CrossRef] [Medline]
    46. Plaete J, Crombez G, Van der Mispel C, Verloigne M, Van Stappen V, De Bourdeaudhuij I. Effect of the web-based intervention MyPlan 1.0 on self-reported fruit and vegetable intake in adults who visit general practice: a quasi-experimental trial. J Med Internet Res. 2016;18(2):e47. [FREE Full text] [CrossRef] [Medline]
    47. Silina V, Tessma MK, Senkane S, Krievina G, Bahs G. Text messaging (SMS) as a tool to facilitate weight loss and prevent metabolic deterioration in clinically healthy overweight and obese subjects: a randomised controlled trial. Scand J Prim Health Care. 2017;35(3):262-270. [FREE Full text] [CrossRef] [Medline]
    48. Springer A, Venkatakrishnan A, Mohan S, Nelson L, Silva M, Pirolli P. Leveraging self-affirmation to improve behavior change: a mobile health app experiment. JMIR Mhealth Uhealth. 2018;6(7):e157. [FREE Full text] [CrossRef] [Medline]
    49. Van der Mispel C, Poppe L, Crombez G, Verloigne M, De Bourdeaudhuij I. A self-regulation-based eHealth intervention to promote a healthy lifestyle: investigating user and website characteristics related to attrition. J Med Internet Res. 2017;19(7):e241. [FREE Full text] [CrossRef] [Medline]
    50. Whitley MD, Payán DD, Flórez KR, Williams MV, Wong EC, Branch CA, et al. Feasibility and acceptability of a mobile messaging program within a church-based healthy living intervention for African Americans and Latinos. Health Informatics J. 2020;26(2):880-896. [FREE Full text] [CrossRef] [Medline]
    51. Young CL, Mohebbi M, Staudacher H, Berk M, Jacka FN, O'Neil A. Int Rev Psychiatry. 2021;33(3):266-279. [CrossRef] [Medline]
    52. Yuhas M, Brock DP, Ritterband LM, Chow PI, Porter KJ, Zoellner JM. Retention and engagement of rural caregivers of adolescents in a short message service intervention to reduce sugar-sweetened beverage intake. Digit Health. 2023;9:20552076231160324. [FREE Full text] [CrossRef] [Medline]
    53. Stroebe W, van Koningsbruggen GMV, Papies EK, Aarts H. Why most dieters fail but some succeed: a goal conflict model of eating behavior. Psychol Rev. 2013;120(1):110-138. [CrossRef] [Medline]
    54. Flagel SB, Robinson TE. Neurobiological basis of individual variation in stimulus-reward learning. Curr Opin Behav Sci. 2017;13:178-185. [FREE Full text] [CrossRef] [Medline]
    55. Kinyanjui MW, Aloka PJ, Mutisya SK, Ndeke FN, Nyang’ara NM. Classroom instruction reinforcement strategies and factors that influence their implementation in Kenyan primary schools. JESR. 2015;5(3):267-278. [CrossRef]
    56. Sackett PR, Lievens F, Van Iddekinge CH, Kuncel NR. Individual differences and their measurement: a review of 100 years of research. J Appl Psychol. 2017;102(3):254-273. [CrossRef] [Medline]
    57. West R, Brown J. Theory of Addiction. UK. Addiction Press; 2013.
    58. Hall PA, Fong GT. Temporal self-regulation theory: a model for individual health behavior. Health Psychol Rev. 2007;1(1):6-52. [CrossRef]
    59. Mohr DC, Cuijpers P, Lehman K. Supportive accountability: a model for providing human support to enhance adherence to eHealth interventions. J Med Internet Res. 2011;13(1):e30. [FREE Full text] [CrossRef] [Medline]
    60. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis Process. 1991;50(2):179-211. [CrossRef]
    61. Shafi AA, Khemka M, Roy Choudhury S. A new approach to motivation: four-drive model. J Hum Behav Soc Environ. 2015;26(2):217-226. [CrossRef]
    62. Louro MJ, Pieters R, Zeelenberg M. Dynamics of multiple-goal pursuit. J Pers Soc Psychol. 2007;93(2):174-193. [CrossRef] [Medline]
    63. Baumeister RF, Bratslavsky E, Muraven M, Tice DM. Ego depletion: is the active self a limited resource? J Pers Soc Psychol. 1998;74(5):1252-1265. [CrossRef] [Medline]
    64. Bandura A. Self-efficacy. In: Ramachaudran VS, editor. Encyclopedia of Human Behavior. London, United Kingdom. Academic Press; 1994:71-81.
    65. Rimal RN. Modeling the relationship between descriptive norms and behaviors: a test and extension of the theory of normative social behavior (TNSB). Health Commun. 2008;23(2):103-116. [CrossRef] [Medline]
    66. Legros S, Cislaghi B. Mapping the social-norms literature: an overview of reviews. Perspect Psychol Sci. 2020;15(1):62-80. [FREE Full text] [CrossRef] [Medline]
    67. Rivis A, Sheeran P. Descriptive norms as an additional predictor in the theory of planned behaviour: a meta-analysis. Curr Psychol. 2003;22(3):218-233. [CrossRef] [Medline]
    68. Ramanathan S, Menon G. Time-varying effects of chronic hedonic goals on impulsive behavior. J Mark Res. 2018;43(4):628-641. [CrossRef]
    69. Pfitzner-Eden F. Why do i feel more confident? Bandura's sources predict preservice teachers' latent changes in teacher self-efficacy. Front Psychol. 2016;7:1486. [FREE Full text] [CrossRef] [Medline]
    70. Parkinson J, David P, Rundle‐Thiele S. Self-efficacy or perceived behavioural control: which influences consumers' physical activity and healthful eating behaviour maintenance? J Consum Behav. 2017;16(5):413-423. [CrossRef]
    71. Vancouver JB, Thompson CM, Tischner EC, Putka DJ. Two studies examining the negative effect of self-efficacy on performance. J Appl Psychol. 2002;87(3):506-516. [CrossRef] [Medline]
    72. Skinner BF. Science and human behavior. In: The B. F. Skinner Foundation. New York, NY, United States. Free Press; 2014.
    73. Bandura A. Self-efficacy: toward a unifying theory of behavioral change. Psychol Rev. 1977;84(2):191-215. [Medline]
    74. Hofmann W, Friese M, Wiers RW. Impulsive versus reflective influences on health behavior: a theoretical framework and empirical review. Health Psychol Rev. 2008;2(2):111-137. [CrossRef]
    75. Bjorklund DF, Harnishfeger KK. The resources construct in cognitive development: diverse sources of evidence and a theory of inefficient inhibition. Dev Rev. 1990;10(1):48-71. [CrossRef]
    76. Cabeza R, Nyberg L. Imaging cognition II: an empirical review of 275 PET and fMRI studies. J Cogn Neurosci. 2000;12(1):1-47. [CrossRef] [Medline]
    77. Baumeister RF, Muraven M, Tice DM. Ego depletion: a resource model of volition, self-regulation, and controlled processing. Soc Cogn. 2000;18(2):130-150. [CrossRef]
    78. Muraven M, Baumeister RF. Self-regulation and depletion of limited resources: does self-control resemble a muscle? Psychol Bull. 2000;126(2):247-259. [CrossRef] [Medline]
    79. Probst Y, Nguyen DT, Tran MK, Li W. Dietary assessment on a mobile phone using image processing and pattern recognition techniques: algorithm design and system prototyping. Nutrients. 2015;7(8):6128-6138. [FREE Full text] [CrossRef] [Medline]
    80. Shonkoff E, Cara KC, Pei XA, Chung M, Kamath S, Panetta K, et al. The state of the science on artificial intelligence-based dietary assessment methods that use digital images: a scoping review. Current Developments in Nutrition. 2022;6:534. [CrossRef]
    81. Granqvist P, Mikulincer M, Shaver PR. An attachment theory perspective on religionspirituality. In: Vail III K, Routledge C, editors. The Science of Religion, Spirituality, and Existentialism. Cambridge, MA. Academic Press; 2020.
    82. Ryan P. Integrated theory of health behavior change: background and intervention development. Clin Nurse Spec. 2009;23(3):161-172. [FREE Full text] [CrossRef] [Medline]
    83. Bodenhausen GV. Emotions, arousal, and stereotypic judgments: A heuristic model of affect and stereotyping. In: Mackie DM, Hamilton DL, editors. Affect, Cognition, and Stereotyping. US. Academic Press; 1993:13-37.
    84. Halama P. Self-regulation capacity and decision making of Slovak managers in routine situations and in situations with possible negative outcomes. SP. 2017;59(2):156-168. [CrossRef]
    85. Bastian B, Pe ML, Kuppens P. Perceived social pressure not to experience negative emotion is linked to selective attention for negative information. Cogn Emot. 2017;31(2):261-268. [CrossRef] [Medline]
    86. Stasiewicz PR, Maisto SA. Two-factor avoidance theory: the role of negative affect in the maintenance of substance use and substance use disorder. Behavior Therapy. 1993;24(3):337-356. [CrossRef]
    87. Van DT. Motivational and affective components of workplace learning in some sectors in the Netherlands. In: Chisholm L, Lunardon K, Ostendorf A, Pasqualoni PP, editors. Decoding the Meanings of Learning at Work in Asia and Europe. Innsbruck. Innsbruck University Press; 2012:37-55.
    88. Maes S, Gebhardt W. Self-regulation and health behavior. In: Boekaerts M, Pintrich PR, Zeidner M, editors. Handbook of Self-Regulation. London, United Kingdom. Academic Press; 2000:343-368. [CrossRef]
    89. Yardley L, Spring BJ, Riper H, Morrison LG, Crane DH, Curtis K, et al. Understanding and promoting effective engagement with digital behavior change interventions. Am J Prev Med. 2016;51(5):833-842. [CrossRef] [Medline]
    90. Aunger R, Curtis V. The evo-eco approach to behaviour change. In: Gibson MA, Lawson DW, editors. Applied Evolutionary Anthropology: Advances in the Evolutionary Analysis of Human Behaviour. Midtown Manhattan, New York City. Springer; 2014:271-295. [CrossRef]
    91. Masthoff J, Grasso F, Ham J. Preface to the special issue on personalization and behavior change. User Model User-Adap Inter. 2014;24(5):345-350. [CrossRef]
    92. Lyles CR, Nguyen OK, Khoong EC, Aguilera A, Sarkar U. Multilevel determinants of digital health equity: a literature synthesis to advance the field. Annu Rev Public Health. 2023;44:383-405. [FREE Full text] [CrossRef] [Medline]
    93. Richardson S, Lawrence K, Schoenthaler AM, Mann D. A framework for digital health equity. NPJ Digit Med. 2022;5(1):119. [FREE Full text] [CrossRef] [Medline]
    94. Smith KP, Christakis NA. Social Networks and Health. Annu Rev Sociol. 2008;34(1):405-429. [CrossRef]
    95. Ewart CK. Social action theory for a public health psychology. Am Psychol. 1991;46(9):931-946. [CrossRef] [Medline]
    96. Clark MS, Reis HT. Interpersonal processes in close relationships. Annu Rev Psychol. 1988;39:609-672. [CrossRef] [Medline]

    ‎
    ENTREQ: Enhancing Transparency in Reporting the Synthesis of Qualitative Research
    PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-analyses
    PROSPERO: International Prospective Register of Systematic Reviews
    RCT: randomized controlled trial

    Edited by G Eysenbach, N Cahill; submitted 23.03.24; peer-reviewed by K Porter; comments to author 24.04.24; revised version received 09.05.24; accepted 26.07.24; published 27.08.24.

    Copyright

    ©Jian Wang, Jinli Mahe, Yujia Huo, Weiyuan Huang, Xinru Liu, Yang Zhao, Junjie Huang, Feng Shi, Zhihui Li, Dou Jiang, Yilong Li, Garon Perceval, Lindu Zhao, Lin Zhang. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 27.08.2024.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research (ISSN 1438-8871), is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.