Mixed-methods Delphi study builds digital teaching competency model for medical faculty

Medical educators around the world have spent the past decade racing to keep pace with digital transformation in the classroom and the clinic, but a persistent problem has been the absence of a competency framework designed specifically for them. Existing tools such as the European Framework for the Digital Competence of Educators, widely known as DigCompEdu, offer a broad foundation for teacher digital competence across disciplines, yet they fall short when it comes to the distinctive demands of medical education: guiding clinical reasoning, protecting patient privacy, embedding medical ethics into digitally delivered instruction, and managing practice-oriented teaching responsibilities. A new multi-stage mixed-methods study, published in BMC Medical Education, addresses that gap with a rigorously validated model of digital teaching competency tailored to medical faculty, built through policy analysis, in-depth interviews, and two rounds of modified Delphi consultation with experts spanning medical teaching, clinical practice, and educational management.

The research team, led by Zheng Liu and corresponding author Fan Yang of Chengdu University of Traditional Chinese Medicine, together with colleagues Yong Li, Chengbei Lu, and Rui Kang, set out to construct a model that would not merely catalog digital skills but capture the dynamic, developmental nature of how medical educators acquire and refine those skills over time. The researchers grounded their approach in action research theory, which served as the meta-framework for the entire study. This choice is significant because action research emphasizes cyclical processes of planning, acting, observing, and reflecting, which aligns naturally with the idea that digital competence is not a static credential but a continuously evolving capability that faculty must develop, apply, evaluate, and improve throughout their careers.

To build the initial version of the model, the team systematically mined three complementary sources of evidence. They analyzed 16 research articles drawn from the academic literature and 8 policy documents related to digital education and faculty development. In parallel, they conducted semi-structured interviews with 15 participants, gathering firsthand accounts of how medical educators experience the digital transformation of their teaching. Through this triangulation of published research, institutional policy, and lived professional experience, the researchers extracted candidate competency items and organized them into an initial model comprising four first-level dimensions, twelve second-level indicators, and forty-eight third-level observation points. This three-tier architecture is typical of competency modeling in education research: the top level defines broad domains, the middle level breaks those domains into measurable constructs, and the bottom level specifies concrete, observable behaviors that can be assessed in practice.

The next stage subjected this initial model to expert scrutiny using a modified Delphi approach, a well-established consensus methodology in which a panel of specialists independently rates items across successive rounds, with anonymized feedback from each round informing the next. For this study, 17 experts in medical teaching, clinical practice, and educational management were recruited, and their engagement was exceptional: the expert response rate was 100 percent in both rounds of consultation, an outcome that itself signals the perceived importance of the topic within the professional community. The technical quality indicators reported by the team support the credibility of the consensus process. The expert authority coefficient, a composite measure reflecting both the experts’ familiarity with the subject and the basis of their judgments, reached 0.833, a value conventionally interpreted as high authority. Agreement among experts, measured by Kendall’s coefficient of concordance, rose from 0.284 in the first round, which was statistically significant at P less than 0.05, to 0.316 in the second round, significant at P less than 0.001. The increase in Kendall’s W between rounds indicates that the panel converged toward greater consensus as the model was refined, which is precisely the behavior a well-functioning Delphi process is designed to produce.

The revisions that emerged from the two consultation rounds were substantive rather than cosmetic. After the first round, the experts recommended adding a new second-level indicator called “initiative in technology learning,” recognizing that proactive, self-directed engagement with new technologies is a distinct and essential component of digital teaching competence in medicine, where tools such as virtual reality surgical simulators, learning management systems, and artificial intelligence applications evolve rapidly. The panel also recommended relocating and reconstructing an indicator originally framed as “self-review of digital teaching competency” into the broader and more developmental construct of “development and adjustment of digital teaching competency,” a change that mirrors the study’s underlying action research philosophy by shifting the emphasis from a one-time self-audit to an ongoing cycle of growth and recalibration. In addition, thirteen third-level observation points were refined to sharpen their clarity and measurability, and two duplicated indicators were removed to eliminate redundancy within the model.

The second round of consultation produced only one further refinement to a third-level observation point, indicating that the panel had reached sufficient stability in its judgments and that the model had converged. The final product is a four-dimension framework for digital medical teaching organized around awareness, design, implementation, and reflection, containing 13 second-level indicators and 50 third-level observation points. Each of these dimensions deserves attention for what it says about how the authors conceptualize digital teaching in medicine.

The awareness dimension addresses the cognitive and attitudinal foundation of digital competence, encompassing educators’ understanding of digital technologies, their recognition of the pedagogical opportunities these tools offer, and, critically, their initiative in learning them. The design dimension covers the capacity to plan digitally enhanced learning experiences appropriate to medical curricula, where decisions about simulation-based instruction, online case discussions, and digital assessment must account for clinical accuracy and ethical constraints. The implementation dimension concerns the actual enactment of digital teaching, including the operational fluency to deploy platforms and technologies effectively during instruction, the ability to guide students’ clinical reasoning through digital media, and the vigilance required to safeguard patient privacy when real clinical materials are used in teaching contexts. The reflection dimension, anchored by the reconstructed indicator on development and adjustment, institutionalizes the habit of evaluating one’s own digital teaching practice and adjusting it in response to evidence, feedback, and technological change.

What distinguishes this model from generic digital competence frameworks is the deliberate strengthening of medical education characteristics throughout these dimensions. Clinical reasoning guidance appears as an explicit expectation, acknowledging that medical teachers must not only transmit content digitally but scaffold the diagnostic and decision-making thought processes that define clinical expertise. Patient privacy protection is embedded as a core competency rather than an afterthought, reflecting the reality that medical educators routinely work with identifiable clinical data, imaging, and case narratives. The integration of medical ethics into digitally mediated teaching is likewise woven into the framework, ensuring that as medical curricula migrate online and into virtual environments, ethical formation is not lost in translation. These additions respond directly to the limitations the authors identified in frameworks like DigCompEdu, which, for all its international influence, was not designed with the clinical classroom, the teaching hospital, or the ethics of patient-centered education in mind.

The practical implications of the finalized model are considerable. The authors position it as a content framework for faculty self-assessment, allowing individual medical educators to locate their strengths and gaps across the 50 observable behaviors and prioritize their professional development accordingly. Institutions can use the model as a blueprint for structured training programs, aligning faculty development offerings with the specific indicators the expert panel validated. Researchers and psychometricians can use the 50 third-level observation points as the basis for developing standardized evaluation instruments, whether rating scales, rubrics, or digital badges, that measure digital teaching competency with reference to a consensus-validated structure. Finally, the model provides a scaffold for subsequent empirical research in medical schools, enabling studies that test the framework’s predictive relationships with teaching quality, student outcomes, and faculty adoption of emerging technologies such as artificial intelligence and virtual reality.

The study also carries broader significance for the global conversation on educational digital transformation. The COVID-19 pandemic accelerated the shift to digital and hybrid instruction in medical schools worldwide, and the subsequent rise of generative artificial intelligence has only intensified questions about what medical educators need to know and do. Yet faculty development efforts have often proceeded without a shared vocabulary or validated structure for describing digital teaching competence in medicine. By combining action research theory as an organizing meta-framework with a transparent, statistically documented Delphi process, the Chengdu team has delivered both a conceptual architecture and a demonstration of method that other national and institutional contexts could adapt. The funding acknowledgments, which include the National Social Science Fund of China and Sichuan Province higher education reform programs, underline the institutional priority that medical education digitization now commands in China and, by extension, the relevance of the model to comparable efforts elsewhere.

The authors are careful to frame the model as a dynamic one, reflecting the progressive enhancement of faculty abilities in the digital era rather than a fixed checklist. That orientation is embedded in the framework’s own structure: the reflection dimension explicitly commits educators to continuous development and adjustment, echoing the cyclical logic of action research. In an environment where the tools of medical teaching can change substantially within a single academic year, a competency model that assumes ongoing growth may prove more durable than one that treats digital skill as a one-time acquisition. With its 100 percent expert participation, rising consensus statistics, high authority coefficient, and medically grounded content, the framework offers medical schools a credible starting point for assessing, training, and supporting the educators on whom the next generation of physicians will depend. The study is published as an open access article, making the full model and its supporting materials freely available to institutions and researchers worldwide.