Psychosocial Determinants of mHealth Application Adherence for Healthy Eating: A Systematic Literature Review

Literature search strategy

A systematic literature search was conducted using Scopus, Google Scholar, and ScienceDirect to identify peer-reviewed studies related to the psychosocial determinants of mobile health (mHealth) application adherence for healthy eating. The search deployed strategic combinations of keywords and Boolean operators associated with human eating behaviors, digital health ecosystems, user engagement, and psychological drivers. Only empirical studies and original research articles published between January 2020 and May 2026 were considered. The initial database search yielded a total of 566 records. Titles and abstracts were screened for topical relevance following the removal of duplicate records, which narrowed the pool down to potentially eligible studies. These remaining papers underwent a rigorous, full-text evaluation based on predefined selection rules to assess their eligibility. Reference lists of the selected articles were also manually examined to identify additional relevant publications. Ultimately, 14 primary core research articles met all strict inclusion criteria and were included in the final qualitative synthesis.

Inclusion and exclusion criteria

Study selection was conducted in two stages, beginning with title and abstract screening followed by a full-text assessment based on predefined inclusion and exclusion criteria. Studies were included if they investigated the psychosocial determinants influencing user adherence and engagement with mobile health (mHealth) applications designed for healthy eating. Specifically, included papers had to report empirical outcomes connected to internal psychosocial drivers—such as self-efficacy, intrinsic motivation, cognitive load, and social support—or system-user interactions within digital health platforms. Only peer-reviewed, primary original research articles containing sufficient methodological transparency, empirical findings, and objective behavioral, lifestyle, or related clinical outcomes published between 2020 and 2026 were considered. Conversely, studies were excluded if they focused solely on macro-level healthcare systems, app development technical architectures, or pharmacological treatments without exploring human behavioral metrics and app adherence. Additionally, papers were excluded if they evaluated secondary literature such as systematic or narrative reviews, book chapters, commentary notes, and conference abstracts, or if they omitted explicit data regarding psychosocial determinants and provided insufficient methodological frameworks that prevented reliable data extraction.

Study selection process (PRISMA Flow)

The study selection process followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines, progressing systematically through the stages of identification, screening, eligibility, and inclusion. The overall process of article selection is illustrated in the PRISMA flow diagram ( Figure 1). A total of 566 records were initially identified through comprehensive database searches across Scopus, Google Scholar, and ScienceDirect. After removing 136 duplicate records, 430 unique records remained for title and abstract screening, of which 360 were excluded due to a lack of direct relevance to the psychosocial determinants of mHealth application adherence for healthy eating.

Figure 1. PRISMA flow diagram of study selection.

The full texts of the remaining 70 articles were successfully sought for retrieval and assessed for eligibility based on the predefined inclusion and exclusion criteria. During this stage, 56 articles were excluded because they did not isolate behavioral or psychological determinants of app engagement, lacked sufficient methodological transparency, or did not link app adherence to objective dietary and lifestyle outcomes. Ultimately, 14 primary core research articles met all strict eligibility criteria and were included in the final qualitative synthesis for the PRISMA flow diagram ( Figure 1). Additionally, to ensure a comprehensive thematic foundation and robust contextual discussion throughout the manuscript, 10 closely related supplementary studies were cross-referenced, bringing the total bibliography to 24 integrated references.

The characteristics of the 14 primary core studies and the 10 integrated supplementary references included in this systematic review are summarized in Table 1. This comprehensive matrix highlights the specific psychosocial determinants assessed, the targeted digital health applications or clinical contexts, and the main behavioral adherence or lifestyle outcomes reported across each investigated population.

The selected studies report the investigation of diverse dietary determinants, including internal psychological drivers, socioeconomic constraints, and digital health ecosystems, for modifying consumer food choices, maximizing mHealth platform adherence, and improving long-term behavioral and metabolic health.

Data extraction and analysis

Data from the 14 primary core studies and 10 integrated supplementary references were extracted into a predefined spreadsheet covering primary author details, study design, targeted population context, core psychosocial determinants, theoretical frameworks, and objective health, behavioral, or lifestyle outcomes. Additional information on digital readiness, interactive app features (such as gamification or digital coaching), and clinical biomarkers—including glycemic control, blood pressure dynamics, or weight fluctuations—was also recorded to better interpret the broader biomedical and behavioral implications of mHealth utilization.

The extracted data were then synthesized narratively into three distinct thematic areas based on the socio-ecological framework. Specifically, the analysis first addresses the internal psychosocial drivers and cognitive heuristics governing user consumption intent and intrinsic motivation. It then evaluates the influence of socioeconomic constraints, developmental milestones, and life-stage vulnerabilities, such as maternal or pediatric needs, on platform interaction. Finally, the synthesis examines the effectiveness of systemic digital health ecosystems, user-interface simplicity, and human-in-the-loop coaching in bridging the behavioral intention-adherence gap to secure long-term clinical and metabolic outcomes.

Internal psychosocial drivers and cognitive heuristics in mHealth engagement

The success of mobile health (mHealth) interventions in fostering healthy eating patterns is fundamentally governed by internal psychological dynamics. Within this systematic review, intrinsic motivation and high self-efficacy emerge as the primary psychological facilitators of long-term application adherence. When individuals possess a high baseline of nutritional knowledge and internal drive, as observed in low-income pregnant women participating in structured dietary programs, their initial interactions with digital platforms are highly proactive.¹⁰ This proactive engagement is often fueled by the cognitive heuristic of health empowerment, where users perceive the digital tool not merely as a monitoring device, but as an instrument of personal autonomy. According to Self-Determination Theory (SDT), when an app supports a user's autonomy and competence, the behavior transitions from extrinsic compliance to intrinsic habit formation. For instance, in web-based personalized nutrition platforms like GENIE, high initial curiosity and tailored dietary feedback create a strong surge in user onboarding.²²

However, this internal momentum is highly susceptible to cognitive heuristics and psychological fatigue. The cognitive load required to consistently log meals, decipher complex caloric metrics, and track macronutrients introduces significant psychological friction. In studies examining patients managing Type 2 Diabetes, immediate glycemic feedback initially acts as a powerful reinforcement mechanism, driving high early compliance.¹ Yet, over extended periods, this continuous tracking frequently induces “diabetes burnout,” where the mental burden of carbohydrate counting overrides the user's initial self-efficacy.¹ This highlights a critical behavioral paradox: the very features designed to inform users can become cognitive barriers that accelerate app abandonment.

Furthermore, psychological drivers are deeply intertwined with emotional states and perceived health threats. For patients navigating early-stage multiple sclerosis, the proactive desire to control symptoms and mitigate future disability functions as a highly resilient facilitator, sustaining logging behavior even when user interfaces are sub-optimal.⁸ Conversely, negative emotional heuristics such as anxiety, stress, and depressive symptoms act as severe psychological barriers. In populations vulnerable to nutritional deficiencies or those managing concurrent food insecurity and binge eating disorders, the emotional toll of resource scarcity severely impairs executive functioning.¹⁶ In these contexts, apps like FoodSteps demonstrate that high initial adherence decays sharply when users face acute emotional stress or food scarcity anxiety, as the immediate psychological need for emotional coping overrides long-term dietary goals.¹⁶ Therefore, mHealth adherence is not a static trait but a fluctuating psychological state that is heavily dependent on the user's immediate cognitive bandwidth and emotional stability.

Socioeconomic constraints and vulnerabilities in digital dietary adherence

While internal psychological factors dictate a user's intent, socioeconomic constraints and systemic vulnerabilities establish the boundaries of actual behavioral execution. This review reveals that digital health interventions do not operate in a vacuum; rather, their efficacy is strictly constrained by the user's socioeconomic reality. Marginalized and low-income populations face severe structural barriers that frequently render standard mHealth applications ineffective. High smartphone data costs, unstable internet connectivity, and language barriers represent primary environmental barriers that trigger rapid user attrition.⁵ For these communities, a digital application that requires continuous cloud synchronization or high-bandwidth video streaming creates immediate financial strain, shifting the app from a supportive tool to an economic burden.

Moreover, food insecurity introduces a profound disconnect between digital dietary recommendations and practical purchasing power. In community-based participatory trials evaluating programs like “Eat, Learn, Grow,” low-income families often demonstrate an intermittent engagement pattern with mHealth platforms.⁴ Even when parental motivation is high and educational modules are successfully completed, the lack of financial access to fresh, high-quality, or specialized ingredients prevents users from executing the app's nutritional advice.⁴ This structural misalignment creates a sense of behavioral helplessness, where users abandon the application because its recommendations are fundamentally incompatible with their economic limitations.

The intersection of financial volatility and psychological stress creates a compounding barrier to digital adherence. Pregnant or postpartum women facing socioeconomic disadvantages often balance competing household priorities, making the time investment required for detailed digital logging unfeasible.¹¹ In these vulnerable demographics, the immediate demands of child-rearing and economic survival displace the structured behavioral milestones dictated by digital platforms. Consequently, mHealth platforms designed with a “one-size-fits-all” framework often fail because they assume a level of baseline stability, financial flexibility, and digital literacy that low-income or marginalized users simply do not possess. To achieve equitable health outcomes, digital dietary interventions must actively account for these structural vulnerabilities by incorporating offline functionalities, low-data interfaces, and economically realistic dietary alternatives.

Developmental milestones and life-stage vulnerabilities

The psychosocial determinants of mHealth adherence vary drastically across different lifecycle stages, demanding highly tailored digital strategies for pediatric, adolescent, maternal, and geriatric cohorts. In pediatric interventions, such as lifestyle programs for preschool children, app adherence is entirely mediated by parental proxies. The primary facilitators in this demographic are parent self-efficacy and the integration of family-centered gamification features, which transform dietary adjustment into a shared household activity.¹³ However, the corresponding barrier is parental time scarcity and household multitasking, which frequently leads to inconsistent logging when family routines become disrupted.¹³

As users transition into adolescence, the psychological drivers shift dramatically from parental oversight to peer dynamics and social identity. Adolescent nutrition platforms, such as SmartAPPetite, face the unique challenge of navigating short attention spans and a highly pervasive commercial environment saturated with fast-food marketing.¹⁴ For teenagers, traditional health-centric messaging fails to motivate; instead, adherence is driven by short-term spikes in engagement stimulated by competitive peer leaderboards, highly visual gamified badges, and social media integration.¹⁴ Without these continuous, behaviorally aligned updates, adolescent engagement decays almost immediately due to lifestyle inertia and alternative digital distractions.

At the opposite end of the developmental spectrum, older adults face distinct physical and cognitive barriers to digital adherence. In digital interventions aimed at preventing frailty in geriatric populations, technological friction—such as complex navigation paths, small font sizes, and vision impairments—presents an immediate barrier.¹⁵ Older adults require simplified user interfaces, transparent privacy assurances, and direct family or caregiver involvement to maintain digital trust and engagement.¹⁵ Meanwhile, maternal cohorts navigating prenatal and postpartum phases experience extreme lifestyle disruptions. In clinical trials like GROWell, women managing gestational or postpartum weight demonstrate that tracking adherence is highly vulnerable to postpartum exhaustion and child-rearing schedules.¹¹ These distinct life-stage realities prove that age and developmental milestones dictate not only how users interact with digital tools, but also the specific psychosocial support structures required to keep them engaged.

Adherence dynamics in specialized clinical and oncology contexts

In clinical and oncological environments, mHealth adherence ceases to be an elective lifestyle choice and becomes an integral component of medical therapeutic management. In these contexts, the psychological facilitators of adherence are exceptionally strong, driven by survival motivation and the desire for clinical security. Patients undergoing active, intense medical regimens, such as individuals treated for head and neck cancer (HNC), view digital applications as a vital lifeline to communicate severe side effects like dysphagia or extreme weight loss to their medical team.⁶ The integration of real-time clinical tracking and professional oversight acts as a powerful compliance mechanism. However, this engagement is highly volatile and tightly bound to the patient's immediate physical health status; during peak phases of chemotherapy or radiation, severe physical discomfort and profound fatigue cause app adherence to drop sharply, as survival and pain management eclipse digital tracking behaviors.⁶

A similar pattern is observed among survivors of non-small cell lung cancer (NSCLC), where digital lifestyle applications offer a sense of proactive recovery and health reclamation.⁷ For these patients, clinical security cues and explicit validation from their oncology team serve as vital psychological anchors that combat post-treatment depression and physical exhaustion, thereby sustaining behavioral retention.⁷ In non-oncological chronic conditions, such as inflammatory bowel disease (IBD), Crohn's, or ulcerative colitis, app adherence exhibits a symptom-dependent pattern. Using platforms like Nootric^®, patients demonstrate a significant spike in dietary logging compliance during active symptom flare-ups, utilizing the app as a coping mechanism to map food triggers and alleviate gastrointestinal pain.¹² Yet, once symptoms enter remission, users frequently experience dietary restriction burnout, leading to a sharp drop-off in application engagement.¹²

This symptom-driven adherence cycle underscores the necessity for adaptive clinical app interfaces that transition smoothly between acute clinical monitoring and long-term wellness maintenance. For older clinical cohorts, such as prostate cancer survivors navigating post-operative telehealth interventions, structural professional oversight remains the ultimate facilitator.¹⁷ While these patients often struggle with age-related digital friction and technological anxiety, the inclusion of scheduled tele-consultations effectively counteracts these barriers, reinforcing behavioral compliance through a sense of clinical accountability.¹⁷ Ultimately, clinical mHealth systems must be structurally resilient enough to absorb the physical and emotional vulnerabilities of patients, shifting from rigid data repositories into empathetic, clinically responsive interfaces.

Systemic digital health ecosystems and technological paradigms

The structural and technological architecture of an mHealth application plays a decisive role in converting psychological intent into sustained behavioral adherence. This systematic review highlights that user interface simplicity and low cognitive friction are universal technological facilitators. In the context of cardiovascular disease (CVD) prevention, platforms like “Growing Healthy Hearts" demonstrate that automated, gamified goal setting and strategically timed push reminders can effectively overcome initial user inertia and guide individuals through the gradual stages of lifestyle modification.² When the technological ecosystem minimizes manual input through automation, users are far more likely to maintain consistent tracking behaviors. Conversely, platforms that demand exhaustive initial data entry or feature highly fragmented navigation menus, such as the multidisciplinary OneSTOP obesity program, experience high rates of early programmatic dropout due to immediate user frustration and technological overload.¹⁸

The integration of advanced technologies, such as Artificial Intelligence (AI) and personalized nutrition algorithms, represents a major paradigm shift in digital health. Personalized nutrition platforms like GENIE capitalize on user curiosity by generating highly individualized dietary blueprints.²² However, the long-term adherence of these systems is heavily contingent upon digital literacy and user attitudes toward automated decision-making. Young, tech-savvy adults frequently exhibit high initial adoption rates due to optimistic attitudes toward AI capabilities.²¹ Yet, this engagement is easily undermined by growing concerns regarding data privacy, algorithmic transparency, and a perceived lack of human empathy from purely automated systems.²¹

This algorithmic skepticism is particularly pronounced in sensitive clinical domains, such as AI-driven micronutrient supplementation for maternal health, where pregnant users express heightened safety anxieties regarding automated recommendations.²⁰ To counteract this algorithmic alienation, digital ecosystems are increasingly leveraging “human-in-the-loop" architectures. The inclusion of digital health coaching within mHealth applications, as seen in lipid management trials, dramatically boosts adherence metrics by superimposing human accountability and emotional empathy over automated data tracking.²⁴ Furthermore, the delivery format itself dictates engagement; comparative trials reveal that highly interactive, mobile-optimized gamified edutainment applications vastly outperform traditional web-based platforms by delivering snackable, visually engaging content that fits seamlessly into the user’s daily digital routine.²³

Implications for practice, app design, and public health policy

The collective findings of this systematic review yield profound, actionable implications for digital health developers, clinical practitioners, and public health strategists. For app developers, the evidence clearly signals the end of the “one-size-fits-all" design philosophy. To mitigate the steep attrition rates dictated by the law of attrition, developers must transition toward human-centered, adaptive user interfaces that dynamically adjust to the user's fluctuating cognitive and emotional states. For instance, instead of demanding perpetual, detailed macronutrient logging—which invariably leads to tracking fatigue and burnout¹—applications should feature “flexible tracking modes" that automatically simplify reporting requirements during periods of high user stress or disease remission.¹² Furthermore, developers must actively integrate robust privacy-by-design frameworks and transparent data policies to alleviate user anxieties surrounding AI algorithms, thereby securing long-term trust and platform retention.²¹

For clinical practitioners, mHealth applications must be treated as formal extensions of the clinical ecosystem rather than isolated consumer software. Clinicians should strategically prescribe applications that feature synchronous or asynchronous health coaching, as the integration of professional human accountability is the single most effective facilitator for sustaining user engagement across diverse patient populations.²⁴ When managing chronic conditions like heart failure or hypertension, clinicians must ensure that the digital platform facilitates collaborative value co-creation, actively engaging both the patient and their immediate family support network to ease the cognitive burden of strict dietary restrictions, such as sodium monitoring.^9,3

From a public health policy perspective, governments and healthcare authorities must address the digital and economic divide to prevent mHealth from inadvertently widening health inequities. Public health initiatives should focus on subsidizing data access for health applications and deploying culturally tailored digital interventions via community health advocates.⁵ Moreover, digital health policies must incentivize the co-creation of applications that directly address resource-constrained realities—such as offering low-cost grocery maps and alternative ingredient suggestions for food-insecure families.^4,16 By mandating that digital health technologies are designed with structural vulnerabilities in mind, policymakers can leverage mHealth as a scalable, cost-effective weapon to combat the global burden of non-communicable diseases and advance systemic health equity.

Limitations and research gaps in current mhealth adherence literature

Despite the valuable insights generated by this review, several critical limitations and research gaps persist within the current body of literature. A primary methodological limitation across many of the reviewed studies is the reliance on short-term intervention timelines and small, homogeneous sample sizes. A significant portion of current empirical evidence captures app adherence during acute, highly monitored trial phases lasting from 4 weeks to 6 months, failing to provide data on long-term behavioral maintenance over multiple years. Consequently, the true trajectory of the “law of attrition" in real-world settings—outside the artificial encouragement of active research trial environments—remains insufficiently documented. Furthermore, there is a distinct lack of standardized, universally accepted metrics for defining and quantifying “adherence” and “engagement” within digital health research; some studies evaluate adherence based on the frequency of app log-ins, while others measure the depth of data entry, making cross-study comparisons highly problematic.

A profound geographic and socioeconomic research gap is the overwhelming concentration of mHealth adherence studies within high-income nations possessing robust digital infrastructures. Very little empirical research explicitly isolates and models the psychosocial facilitators and barriers unique to low-and-middle-income countries (LMICs) or marginalized sub-populations facing severe structural food insecurity.^4,5 Consequently, current digital behavior-change models heavily reflect Western, educated, industrialized, rich, and democratic (WEIRD) demographic archetypes, leaving a vacuum of knowledge regarding how digital dietary tools interact with diverse cultural identities, traditional food ecosystems, and severe digital literacy limitations.

Finally, a major technological research gap lies at the intersection of behavioral psychology and advanced artificial intelligence. While platforms are rapidly deploying AI and automated personalized nutrition algorithms, ^22,21 there is a distinct shortage of longitudinal research investigating how users form long-term psychological relationships with AI health entities. Current literature fails to adequately explain how algorithmic mistrust, automated paternalism, and data privacy fears dynamically evolve over time, or how these advanced systems can be designed to deliver genuine cognitive empathy without relying constantly on costly human intervention.^20,24 Bridging these methodological, geographic, and technological gaps represents the critical next frontier for digital health research.