This study was conducted as a scoping review incorporating a bibliometric analysis in accordance with the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) statement [21]. The review protocol was not preregistered. Our primary objective was to systematically map and characterize the published literature on nurse-involved DMHIs from 2020 to 2026. The completed PRISMA-ScR checklist is provided in Checklist 1.

Data were retrieved from 3 electronic databases on April 15, 2026: Web of Science Core Collection, Scopus, and PubMed. The search was limited to peer-reviewed journal articles published in English between January 1, 2020, and April 15, 2026. The search strategy combined terms related to three conceptual domains: (1) digital interventions [22,23], (2) mental health conditions [3], and (3) nursing involvement and research paradigms [4,8]. The complete search strategy for each database is provided in Multimedia Appendix 1.

Articles were included if they met all of the following criteria: (1) focused primarily on a digital intervention targeting mental health conditions; (2) involved nursing roles (nurse-led, nurse-delivered, or nurse-involved implementation); (3) reported original empirical research with human participants; (4) were published in English in peer-reviewed journals; and (5) were published between January 1, 2020, and April 15, 2026. We excluded review articles, study protocols, commentaries, editorials, theoretical papers, and studies in which nursing involvement was peripheral.

Publications were classified using a multistage process informed by implementation science frameworks [18,24]. Implementation research was defined by a focus on real-world adoption and implementation outcomes in nursing contexts [25], and efficacy and effectiveness research was defined by a focus on clinical outcomes under controlled conditions. A conservative keyword-based classification strategy was used to prioritize specificity, providing a lower-bound estimate of implementation research prevalence. Classification followed 3 stages: automated keyword screening, manual verification, and validation against a consensus standard (κ=0.73, 95% CI 0.60-0.86).

Two reviewers independently extracted data from each included study using a standardized form. Extracted items included first author, publication year, country, study design, sample size, research paradigm (implementation vs efficacy or effectiveness), theoretical framework used, implementation outcomes measured (if applicable), and key themes. Disagreements were resolved through discussion or consultation with a third reviewer. Quality assessment was not performed as this was a scoping review focused on mapping the literature rather than critically appraising evidence quality [21].

Descriptive statistics summarized publication characteristics. Proportions with 95% CIs were calculated using the Wilson score method. Temporal trends were analyzed using linear regression. Methodological differences between research paradigms were examined using chi-square tests. Co-word analysis and thematic mapping were conducted using the bibliometrix package in R (version 4.2; R Foundation for Statistical Computing) [26].

This study involved analysis of publicly available published literature and did not require ethics approval.

Our search identified 1737 records across 3 databases (Scopus: n=1183, 68.1%; Web of Science: n=431, 24.8%; PubMed: n=123, 7.1%). To ensure a conservative estimate of unique publications, we performed deduplication using both digital object identifier (DOI) and title. Of the 1737 identified records, DOI-based deduplication removed 337 (19.4%) duplicates (resulting in n=1400, 80.6% of records), and subsequent title-based deduplication identified an additional 386 (22.2%) duplicate records with inconsistent or missing DOI assignments across databases. This 2-stage approach yielded 1014 unique records (Figure 1). All 1014 records met the inclusion criteria for original empirical research on nurse-involved DMHIs published in English between 2020 and 2026.

The final sample included studies from 44 countries. The largest contributions came from the United States (373/1014, 36.8%), the United Kingdom (161/1014, 15.9%), Australia (110/1014, 10.8%), and China (83/1014, 8.2%). The mean sample size of primary studies was 162 (SD 241) participants.

Across all 1014 included studies, randomized controlled trials represented the most common study design (n=303, 29.9%). Observational designs followed as the second most frequent category (n=129, 12.7%), followed by mixed methods research (n=106, 10.5%). Feasibility and pilot studies and qualitative studies each comprised 7.1% (n=72) of the studies, whereas process evaluation designs were the least prevalent (n=44, 4.3%).

For research paradigm classification, a conservative keyword coding approach left 53.6% (544/1014) of the total studies with an ambiguous categorization that could not be clearly assigned to either implementation- or efficacy-focused work. Among the remaining 470 clearly classified studies, implementation-focused research constituted the dominant paradigm at 69.6% (n=327), whereas efficacy and effectiveness research made up the remaining 30.4% (n=143). All mixed or hybrid methodological studies were subsumed within the implementation research grouping in line with standard implementation science classification frameworks [27]. The implementation-to-efficacy study ratio was calculated at 2.29:1, demonstrating a preponderance of implementation research within the clearly classified sample.

Publication volume demonstrated substantial growth from 2020 to 2025 (Figure 2), increasing from 102 articles in 2020 to 247 in 2025 (142% increase). Linear regression analysis revealed a significant positive trend in implementation research from 2020 to 2025 (β=8.63 articles per year, 95% CI 5.21-12.05; P=.003; R²=0.91). Implementation research studies outnumbered efficacy research studies in 2022 (47 implementation-related articles vs 15 efficacy or effectiveness articles) and retained this predominance in subsequent years, peaking at 74 articles in 2025. Efficacy and effectiveness research demonstrated no statistically significant growth across the study period (β=1.8 articles per year; P=.48). The lower count in 2026 (21 implementation articles) reflects incomplete year data and was excluded from trend analysis.

Analysis of studies with unambiguous classification revealed substantial methodological divergence (Table 1 and Figure 3). Implementation studies used significantly more mixed methods designs (105/327, 32.1% vs 7/143, 4.9%; χ²₅=32.4; P<.001) and qualitative approaches (92/327, 28.1% vs 2/143, 1.4%; χ²₅=41.2; P<.001) compared to efficacy and effectiveness studies. In contrast, efficacy research remained predominantly reliant on randomized controlled trials compared to implementation studies (97/143, 67.8% vs 82/327, 25.1%, respectively; χ²₅=24.8; P<.001).

Co-word analysis of keywords from all 1014 publications revealed a clear thematic evolution toward implementation science (Figure 4). Implementation-focused constructs including “implementation science,” “nursing workflow,” “task-shifting,” and “nurse-led implementation” emerged as motor themes characterized by high centrality and density. Nursing-specific implementation barriers such as “time constraints,” “training needs,” and “workflow integration” showed increasing prominence in recent publications (2024‐2025). Notably, implementation-related keywords showed the highest annual growth rate among all thematic categories (β=2.1 appearances per year; P=.008).

Our bibliometric analysis of 1014 nurse-involved DMHI studies provides robust empirical evidence of a decisive paradigm shift toward implementation science. Three convergent lines of evidence support this conclusion. First, implementation research constituted the largest proportion of clearly classified studies (327/470, 69.6%), exceeding traditional efficacy research (143/470, 30.4%). Second, implementation research demonstrated a significant annual growth rate of 8.63 articles per year from 2020 to 2025 (P=.003; R²=0.91), surpassing efficacy research in 2022. Third, fundamental methodological differences characterize these paradigms, with implementation studies using substantially more mixed methods (105/327, 32.1% vs 7/143, 4.9%) and qualitative designs (92/327, 28.1% vs 2/143, 1.4%).

Our findings align with broader trends in implementation science but extend them to the specific context of nursing digital health [28,29]. Previous research has documented increasing attention to implementation science in health care generally [30], but our study is the first to quantify this shift specifically in nurse-involved DMHI research using a multi-database bibliometric approach. The methodological divergence we observed—with implementation studies favoring mixed methods approaches—reflects the complex, context-dependent nature of implementation challenges in nursing practice [27].

Recent studies have begun to identify effective nurse-specific implementation strategies. Targeted, blended training combining online learning and in-person coaching markedly elevates nurses’ adoption of digital mental health tools, with intervention groups demonstrating more than double the uptake relative to untrained control staff [31]. Similarly, research confirms that “implementation champions” drive long-term sustainability; grassroots nurse-led digital interventions maintain stable clinical use far longer than institution-mandated top-down rollouts [32].

The documented shift has direct implications for creating more nurse-friendly digital health ecosystems.

First, for digital health developers, our findings underscore the necessity of “implementation by design.” DMHI development must move beyond user experience for patients to include clinician experience for nurses, explicitly considering workflow integration, time burden, and interoperability with clinical systems from the earliest stages [22,23].

Second, for nursing informatics leadership, chief nursing informatics officers and other leaders must advocate for and allocate resources toward implementation infrastructure. This includes dedicated time for nurse training and adaptation, the creation of role-specific implementation playbooks, and the evaluation of DMHIs based on implementation outcomes (eg, adoption, feasibility, and sustainability) alongside clinical outcomes [33].

Third, for education and training, nursing curricula and continuing professional development must evolve to build digital health implementation competency. This includes skills in evaluating digital tools for implementability, leading practice change, and using data from DMHIs to inform care—a core component of modern nursing informatics literacy [34,35].

Our findings suggest several directions for future nursing research. First, there is a need for more hybrid effectiveness-implementation designs that simultaneously examine clinical outcomes and implementation processes [27]. Second, research should explore nurse-specific implementation strategies such as tailored training programs and workflow integration tools [36]. Third, future studies should examine how implementation outcomes vary across nursing roles and settings [37]. Finally, given the geographic concentration of research in high-income countries, there is an urgent need for implementation research in low- and middle-income settings, where nurses often serve as the primary mental health care providers [38].

Strengths of this study include its comprehensive multi-database search strategy (Web of Science, Scopus, and PubMed), rigorous classification system with validation (κ=0.73), updated search up to April 2026, and integration of multiple bibliometric techniques, analyzing 1014 eligible studies.

Several limitations warrant acknowledgment. First, reliance on title and abstract text for initial classification may have misclassified some studies, although manual verification and consensus validation mitigated this risk. Our conservative classification strategy prioritized specificity, resulting in a higher proportion of unclear classifications (544/1014, 53.6%); this provides a lower-bound estimate of implementation research prevalence. Second, the exclusion of non–English-language publications introduces language bias, potentially missing relevant research from non–English-speaking countries. Third, we did not include CINAHL due to institutional access limitations, which may have underrepresented nursing-specific literature. Future reviews should include this database. Fourth, the rapid evolution of digital health technologies means that findings from older studies (2020-2022) may not fully reflect current capabilities, although our temporal trend analysis addresses changing patterns over time. Fifth, the lack of quality assessment, consistent with scoping review methodology, means that we cannot comment on the methodological rigor of the included studies.

This scoping review and bibliometric analysis documents a decisive and accelerating paradigm shift toward implementation science in nurse-involved DMHI research. Implementation research now constitutes the dominant paradigm (327/1014, 32.2% vs 143/1014, 14.1% for efficacy research) and demonstrated a significant annual growth rate of 8.63 articles per year from 2020 to 2025 (P=.003; R²=0.91), surpassing efficacy research in 2022. The field is expanding beyond a primary focus on internal validity to systematically investigate the complex determinants and processes of real-world implementation in nursing contexts.