Medical adherence is traditionally defined by the World Health Organization (WHO) as the extent to which a person’s behavior, including taking medication, following a diet, and executing lifestyle changes, corresponds with agreed-upon recommendations from a health care provider [1]. This definition marked a deliberate shift from earlier, medication-centered views by emphasizing 2 core dimensions. First, adherence encompasses a broad range of therapeutic behaviors, including self-management activities, follow-up attendance, and lifestyle modification. Second, it presupposes a collaborative relationship between patients and health care providers, in which treatment plans are negotiated rather than unilaterally prescribed, and adherence is actively discussed, supported, and monitored over time [1,2].

This behavioral and relational foundation has proven conceptually robust in conventional health care settings, particularly where treatment regimens are clearly defined in terms of dosage, frequency, and duration. However, the rapid diffusion of patient-facing digital health applications (DHAs) challenges the direct transferability of this framework to software-mediated interventions [3-5]. Unlike pharmacological treatments, which are typically characterized by a defined therapeutic dose and schedule, DHAs rarely have a universally accepted or empirically established notion of “optimal use” [4]. Although DHAs generate detailed usage data by design, the mere availability of digital logs (metadata) capturing frequency, duration, sequence, and intensity of use does not intrinsically define what counts as adequate adherence. In the absence of a shared reference standard for therapeutic “dose,” these usage metrics are analytically rich but conceptually underdetermined [3].

To address this gap, several authors have proposed the concept of intended use, defined as using the technology in the way it was designed to be used, and for a duration or intensity presumed necessary to achieve benefit [3,4,6-8]. This concept acknowledges that different DHAs may require distinct patterns of use to produce health outcomes. However, intended use thresholds are rarely standardized, consistently justified, or empirically validated, and often remain implicit, intervention-specific, or defined post hoc [4,7]. Consequently, adherence metrics across DHAs are highly inconsistent, limiting comparability between studies and complicating the generation and interpretation of adherence-related evidence in clinical, economic, and regulatory contexts.

These conceptual and methodological challenges continue to grow in relevance as DHAs increasingly transition from experimental settings into standardized health care pathways. A notable case is Germany, where the Digital Healthcare Act (Digitale-Versorgung-Gesetz), enacted in November 2019, established a regulatory and reimbursement framework under which patient-facing, lower-risk digital medical devices (Digitale Gesundheitsanwendungen [DiGA]) can be prescribed by physicians and reimbursed within the statutory health insurance system following assessment by the Federal Institute for Drugs and Medical Devices (Bundesinstitut für Arzneimittel und Medizinprodukte) [9,10].

Within this framework, recent legislative reforms elevate the relevance of adherence by mandating that, from 2026 onwards, at least 20% of DiGA reimbursement will be linked to performance-related indicators. Under the Digital Act (Gesetz zur Beschleunigung der Digitalisierung des Gesundheitswesens), performance is to be assessed using measures such as the duration and frequency of use, patient satisfaction, and patient-reported health status during DiGA use. Although adherence is not defined as a stand-alone endpoint, these criteria rely directly on adherence-related usage patterns and implicitly operationalize adherence as a determinant of reimbursable performance [11].

In this evolving landscape, adherence becomes a regulatory and economic parameter with direct implications for reimbursement and market access. Yet, the underlying construct remains conceptually unsettled and methodologically heterogeneous. Existing reviews [3,4,6,7,12,13] have examined adherence within specific clinical domains, explored its association with intervention outcomes, identified determinants of adherence, and analyzed how intended use has been operationalized in electronic health technologies. However, none has provided a comprehensive and comparative mapping of how adherence is currently conceptualized, operationalized, and measured in contemporary patient-facing DHAs, nor assessed how these approaches align with the emerging regulatory and reimbursement frameworks in which such applications are increasingly embedded.

Given the conceptual ambiguity, methodological heterogeneity, and increasing policy relevance of adherence in digital health interventions broadly, and in DHAs in particular, this scoping review pursues 3 interrelated objectives:

A scoping review was selected as the most appropriate methodological approach for addressing the objectives of this study. According to Munn et al [14], scoping reviews are particularly appropriate when the objectives are to clarify key concepts and definitions in the literature, identify key characteristics or factors related to a concept, and examine emerging evidence in fields where the conceptual and methodological landscape is still insufficiently settled to support a more focused systematic review. All three of these methodological indications apply directly to this review. In line with this methodological approach, this protocol has been developed using an adapted application of the PRISMA-P (Preferred Reporting Items for Systematic Review and Meta-Analyses—Protocols) guidelines (Checklist 1) [15]. In the absence of a dedicated, stand-alone reporting guideline for scoping review protocols, PRISMA-P items are selectively applied and tailored to reflect the objectives and methodological features of scoping reviews, in line with recommendations from the Joanna Briggs Institute [16]. This protocol was not registered in PROSPERO because scoping reviews are not currently accepted by the registry. To ensure transparency and prospective documentation, the protocol was instead registered on the Open Science Framework. Any deviations from the registered protocol will be documented through updates to the Open Science Framework registration and reported in a supplementary document accompanying the final review.

The conduct of the review will follow established scoping review methodology [17], and the reporting of the completed review will adhere to the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines (Checklist 2) [18], ensuring transparency, reproducibility, and consistency in reporting.

The literature search will use a combination of structured database queries and citation-based retrieval techniques. Electronic searches will be conducted in PubMed (MEDLINE), Scopus, and Web of Science. These databases were selected to provide broad and complementary coverage of biomedical, interdisciplinary, health services, and digital health–related literature relevant to patient-facing DHAs and adherence research, consistent with prior systematic reviews [3,4].

Backward citation tracking, defined as the systematic screening of reference lists, will be applied to all included studies to minimize the risk of overlooking relevant studies.

In addition to peer-reviewed literature, targeted searches of gray literature will be conducted through Google and relevant organizational websites, including those of regulatory authorities (eg, the European Commission and the US Food and Drug Administration), international organizations (eg, the WHO and the Organisation for Economic Co-operation and Development), and industry associations (eg, the National Association of Digital Health Care, Germany, and MedTech Europe, Belgium). The same conceptual keywords and overall eligibility framework used for the scientific literature searches will also guide the identification and selection of gray literature sources.

The search strategy has been developed iteratively by the research team, drawing on prior systematic reviews in this context [3,4,6,7]. The search strategy is structured around two conceptual blocks combined using Boolean operators: (1) adherence and adherence-related constructs, and (2) DHAs.

The first block includes “adherence” and related terms, which are often used interchangeably in the literature, to maximize search sensitivity.

The second block captures DHAs using terminology commonly used in the literature. The term “patient-facing” was not incorporated into the search query, as it is not consistently used in indexing or titles and could unduly restrict retrieval. Restriction to patient-facing applications will be applied during the screening process. Database-specific syntax and controlled vocabulary were used where appropriate.

The complete search strategies, as implemented in each database, are reported in Textbox 1, which presents the full queries for PubMed (MEDLINE), Scopus, and Web of Science.

Data will be collected at the level of the individual study. When a single study reports findings for multiple patient-facing DHAs, each DHA will be extracted as a separate entry in the data extraction table to ensure that data are extracted distinctly for each application. Conversely, when multiple studies refer to the same DHA, data will be extracted separately for each study, with the possibility of synthesizing findings across studies referring to the same application where relevant.

The data items are designed to capture how adherence to patient-facing DHAs is conceptualized, operationalized, and measured, along with key characteristics of the applications and their regulatory or implementation context. Extracted information will include study characteristics, features of the DHA, contextual information relevant to regulation or reimbursement, conceptual definitions of adherence and related constructs, as well as operationalization and measurement approaches.

The data extraction framework presented in Table 1 reflects the preliminary set of predefined items developed to address the review objectives. This framework may be refined following pilot testing to ensure conceptual clarity and the adequate capture of relevant information.

The findings of this scoping review will be synthesized using a narrative synthesis approach, consistent with its exploratory and mapping objectives. Descriptive summaries will be structured according to the main dimensions of the data extraction framework. Where appropriate, simple descriptive statistics (eg, frequencies and proportions) will be used to summarize patterns across studies.

Given the exploratory nature of the review, the expected heterogeneity in study designs and measurement strategies, and the focus on conceptual and methodological mapping rather than effectiveness evaluation, we will not conduct a quantitative synthesis or meta-analysis. Consistent with established guidance for scoping reviews, we will not conduct a formal assessment of the risk of bias or the methodological quality of individual studies [16].

Preliminary results of this scoping review will be presented at relevant scientific conferences to facilitate early dissemination and community engagement. In addition to the scientific publication, the full data extraction table will be made publicly available in an accessible and navigable format on the DEKODE (Digital Health Adhärenz: Konzept und datenbasierte Evidenz) [28] project website to facilitate transparency and support researchers, clinicians, policymakers, and other stakeholders seeking to engage with the underlying evidence. This dissemination strategy is intended to maximize both the scientific impact and the practical accessibility of the review’s findings.

This scoping review is conducted as part of the DEKODE project, funded by the Innovation Fund of the German Federal Joint Committee (Gemeinsamer Bundesausschuss). The overarching objective of DEKODE is to develop a comprehensive adherence toolkit for patient-facing DHAs, and specifically for DiGAs, that presents adherence factors from the perspective of all relevant stakeholders, including patients, health care providers, manufacturers, and health insurance funds, and outlines strategies to increase evidence-based adherence. To support the early stages of this objective, the project also aims to develop and test measurement criteria and standards for adherence, and to derive implementation recommendations for regulation and medical practice that reward adherence-promoting initiatives.

Achieving these objectives requires, as a necessary foundation, the establishment of a robust conceptual and empirical framework for adherence in the context of patient-facing DHAs. This framework is being developed through 2 complementary literature reviews. This scoping review constitutes the first component, mapping how adherence is currently conceptualized, operationalized, and measured across patient-facing DHAs, and examining whether these approaches differ between applications embedded within formal regulatory or reimbursement frameworks and those operating outside such settings. The second review will focus specifically on the demographic and structural factors that may influence adherence to patient-facing DHAs at the patient, disease, and intervention levels. Together, these 2 reviews are intended to provide the conceptual and empirical grounding that will inform the design of qualitative interviews with patients, health care providers, DiGA manufacturers, and health insurance fund representatives, as well as the development and validation of quantitative survey instruments planned for later stages of the project.

While the DEKODE project is primarily oriented toward the German health care context and the DiGA framework, this review applies no geographical restrictions and encompasses patient-facing DHAs irrespective of their regulatory or reimbursement setting. This deliberate choice reflects the view that a general, global understanding of how adherence is conceptualized, operationalized, and measured across diverse contexts is a necessary precondition before narrowing the focus to the specific German setting. Indeed, efforts to harmonize the assessment of DHAs across European countries, which have been initiated over recent years, reflect the timeliness of an international approach [29]. Furthermore, establishing this broader evidence base first allows for a more grounded and rigorous contextualization of the DiGA-specific findings within the international landscape. At the same time, the insights generated may serve as a foundation for country- or system-specific analyses in other national or regulatory settings.

Several limitations of this review should be acknowledged. First, the use of an ML-assisted screening approach introduces a degree of methodological uncertainty. Although the sampling-based stopping criterion adopted is supported by empirical evidence, it cannot fully exclude the possibility that a small number of eligible studies may be missed. However, acceptable accuracy thresholds may vary depending on the type of review, its research question, and the intended use of its findings. In scoping reviews, the influence of the small residual proportion of unidentified studies beyond the active learning stabilization point is likely to be comparatively limited relative to evidence syntheses focused on quantitative aggregation or comparative effectiveness. Moreover, a residual degree of screening error may also occur in fully manual screening processes, particularly in large-scale reviews involving several thousand records, where reviewer fatigue may affect consistency over time. The conduct of screening across 2 independent parallel arms was intended to mitigate these risks, although residual uncertainty remains inherent in any screening approach. The ASReview workflow was described in detail to maximize methodological transparency and reproducibility. However, ML-assisted screening based on active learning does not guarantee fully deterministic reproducibility of record prioritization across different runs or reviewer interactions, as residual variability may arise from the dynamic and iterative nature of the active learning process. Detailed reporting of model configuration, stopping criteria, prior knowledge settings, and screening procedures may nonetheless support procedural transparency and facilitate methodological replication.

Second, restricting the search to 3 major databases—PubMed (MEDLINE), Scopus, and Web of Science—may result in some relevant literature not being captured. The inclusion of targeted gray literature searches and backward citation tracking is intended to partially offset this limitation.

Third, the temporal restriction to publications from 2020 to 2025, while methodologically justified, means that potentially relevant earlier literature is not systematically captured.