Ethics approval was granted by the Medical Science Research Ethics Committee of Peking University Third Hospital (IRB00006761-M2024300). The participants provided informed consent. To ensure confidentiality, all interview data were anonymized and securely stored on password-protected, access-restricted institutional servers. Only the research team had access to the data. Participants received compensation of ¥200 (US $27.85) as a token of appreciation for their participation.

This study aimed to identify problems, propose solutions, and implement actions, which was conducted in 5 steps (Figure 2): (1) developing the interview guideline on the basis of the TDF and conducting interviews [36,37]; (2) identifying the themes mentioned in the interview, classifying them into barriers and enablers, and mapping these barriers and enablers of physicians’ adoption of telemedicine to the relevant TDF domains [23,38]; (3) mapping each TDF domain to its corresponding BCW intervention functions and policy categories, following established literature [24,39], which enables the translation of findings from behavioral analysis into actionable strategies; (4) identifying corresponding BCTs for each intervention function based on published mappings to obtain operationalizable and replicable techniques [24,35]; and (5) conducting focus group discussions to propose context-specific strategies for each technique to support increased physician uptake of telemedicine. This study is reported in accordance with the Consolidated Criteria for Reporting Qualitative Research [40] (COREQ; Multimedia Appendix 1).

The BCW is a comprehensive framework for designing behavior change interventions by systematically linking behavioral determinants to intervention functions and policy categories [24,28]. According to Michie et al [24], the literature delineates these connections, whereby each domain of the TDF can be mapped to one or more BCW intervention functions, which are subsequently mapped to relevant policy categories.

The BCTs provide more standardized and detailed intervention components, encompassing 93 evidence-based techniques that can be mapped to the 9 intervention functions of the BCW to inform the selection of targeted strategies [32] (Table 3.3 in Multimedia Appendix 3). The literature suggests that a wide array of BCTs may be considered for each intervention function, generally prioritizing the more frequently used techniques.

An example is shown in Figure 3 for greater clarification of the process. In Step 1, if a barrier or enabler was coded in the knowledge domain, it was mapped to the education intervention function (Table 2.2 in Multimedia Appendix 3). In Step 2, the education intervention function was subsequently mapped to relevant policy categories, including communication/marketing, guidelines, regulation, legislation, and service provision, with one or more suitable options selected based on the contextual needs (Table 2.9 in Multimedia Appendix 3). In Step 3, BCTs were selected based on the established association with the education intervention function that was prioritized from those most commonly associated with educational strategies, including information about social and environmental consequences, information about health consequences, feedback on behavior, feedback on outcome(s) of the behavior, prompts/cues, and self-monitoring of behavior (Table 3.3 in Multimedia Appendix 3). If none of the commonly used BCTs were deemed appropriate given the contextual constraints, less frequently used but still functionally relevant BCTs were considered as alternatives. The complete English version of the BCT Taxonomy v1 is provided in Multimedia Appendix 4 [32]. In Steps 2 and 3, we adhered to the APEASE (acceptability, practicability, effectiveness, affordability, spill-over effects, and equity) criteria [24,45], drew upon prior literature, and used focus group discussion to identify the intervention function and policy categories. Our group members included clinical physicians who provide telemedicine services, staff of the Office of Internet Hospital, and experts in related fields.

Based on the results of Steps 1 to 4, examples of potential ways to implement these intervention strategies were generated from prior literature and combined focus group discussions, enhancing their potential for successful implementation in clinical practical.

The 36 physicians came from 29 different medical institutions in Beijing, with 1 to 8 physicians from each institution. The majority were public institutions (22/29, 76%), and most were tertiary-level facilities (22/29, 76%). Regarding service volume, more than one-half of the institutions (16/29, 55%) had an annual offline service volume exceeding 1,000,000 visits. In terms of volume of telemedicine services, 45% (13/29) of the institutions provided 10,000 to 100,000 telemedicine services annually. As for the participants, most physicians were from public (29/36, 81%) and tertiary institutions (30/36, 84%). There were more women (21/36, 58%) than men (15/36, 42%). The physicians ranged in age from 31 years to older than 60 years; most (18/36, 50%) had doctoral degrees, and the largest percentage had deputy chief physician titles (21/36, 58%). Physicians were from internal medicine (ie, cardiology, endocrinology, and metabolism), surgery (ie, general surgery, orthopedics), obstetrics and gynecology, anesthesiology, and others. They mainly had 1 year to 3 years of telemedicine work (21/36, 58%). Table 2 shows the details.

A total of 50 themes were identified in relation to the adoption of telemedicine, comprising 27 barriers and 23 enablers and encompassing 11 of the 14 TDF domains. No viewpoint aligned with the goals, beliefs about capabilities, and behavioral regulation domains; we received relevant responses to the pertinent interview questions. Throughout the coding and analysis, the replies were more appropriately categorized under other TDF domains. The barriers referred to 8 of the 14 domains, and enablers were involved in 7 of the 14 domains (4 domains had both barriers and enablers). See Figure 4. Tables 3 and 4 also present the themes and frequencies of the domains.

We provided relative intervention functions, policy categories, and strategies for each barrier and enabler. There were 7 of 9 intervention functions, 6 of 7 policy categories, and 26 of 93 BCTs selected. No barrier nor enabler mapped to the intervention functions of restriction and modeling nor mapped to the policy categories of legislation. We retained two TDF domains that involved the most barrier and enabler themes, along with the corresponding intervention functions, policy categories, and proposed strategies in Table 5. For instance, the barrier of no well-established pricing mechanism (inadequate medical insurance payment mechanisms) was identified in the environmental context and resources domain. To address this, the enablement intervention function was selected, as it is suited to overcoming external constraints and facilitating behavior change by removing barriers or increasing means or resources. To effectively deliver this function, the fiscal measures policy categories were identified as the most suitable channels. Subsequently, the relevant BCTs of “1.3 Goal setting (outcome)” and “3.2 Social support (practical)” were selected to operationalize this strategy. Accordingly, a proposed implementation strategy was setting pricing standards and adjusting the charging scope, linking professional titles to medical service fees. Multimedia Appendix 5 provides the complete results.

This study provides novel evidence and enhances the literature on physicians’ viewpoints, particularly by applying behavioral science frameworks to understand the adoption of telemedicine. The identified factors were comprehensive, and importantly, they were modifiable and amenable to intervention—offering a solid foundation for designing targeted behavior change strategies. These frontline health care practitioners directly influence the effectiveness of telemedicine and the potential success of the digital transformation of medical institutions. To our knowledge, this was the first attempt to apply interlinked frameworks of the TDF, BCW, and BCTs to identify the influencing factors for physicians’ adoption of telemedicine and provide strategies. It enriches the understanding of individual-level factors to include wider contextual drivers, contributing to a conceptual framework for understanding multilevel influences on digital health behavior change.

There were two main findings. First, 11 of the 14 TDF domains have an impact on telemedicine adoption. The barriers comprise difficulty balancing online and offline work, technical issues, potential risks to medical quality and safety with telemedicine, and others. Enablers included that the advantages of telemedicine could guarantee the continuity of medical service. These results align with those of prior research [36,48-50]. Compared with past literature, the emotion domain and social influences domain–related factors have rarely been considered before. We added evidence of physicians’ behaviors and psychology about telemedicine adoption. Second, we presented the intervention functions, policy categories, BCTs, and potential suggestions for each barrier or enabler to ensure the proposed intervention would be feasible with high fidelity. Implementing these measures requires efforts from health administrative departments, medical institutions, departments, and health care personnel. We discuss these issues in the following sections.

On the one hand, the barriers to physicians’ appropriate adoption of telemedicine involved 8 TDF domains. Two domains were particularly richly described. The most evidenced domain was environmental context and resources. It comprised 9 specific barriers, among which the technique issues were the key mentioned barriers. A systematic review that researched the facilitators of and barriers to accessing telemedicine during the COVID-19 pandemic also revealed that a lack of necessary technological conditions or equipment was the main barrier to adopting telemedicine [51]. Other studies have highlighted barriers such as the lack of availability of adequate resources [37], the absence of dedicated personnel, and the paucity of professional guidelines [12]. These findings may suggest that existing mechanisms and resource conditions of telemedicine are not ready to respond to the rapid transition to a new reality of medical care delivery based on physical distancing.

Another frequently mentioned domain was memory, attention, and decision processes. The key barrier that most physicians referred to was the inefficiency caused by turnaround times in asynchronous consultations. The inability of real-time communication between physicians and patients, the inefficiency of physician-patient interactions, and poor experiences seem to be inherent shortcomings of telemedicine, which have also been mentioned several times in prior studies [52,53]. This aligns with recent research showing that approximately 70% of surveyed health care providers believe that communication difficulties between patients and health care providers result in insufficient online consultations and poor online experiences [52]. Other research has indicated that asynchronous consultations have downsides that include risks of missed urgent issues, loss of human contact and empathy, fragmentation of care, and the potential length of time one must wait for a reply. However, asynchronous consultations could be useful for some aspects of care, such as triage, monitoring the health care needs of people with long-term conditions, or reaching people who have difficulty traveling to hospitals [54]. Therefore, informed choice and flexibility are needed. It is essential to improve the design and promotion of synchronous consultations to meet the needs of different physicians and patients.

On the other hand, we also identified enablers of physicians’ adoption of telemedicine that involved 7 TDF domains. The skills and beliefs about consequences domains were particularly richly described. With respect to the skill domain, physicians believe that communication skills are the most essential for online work. Most physicians were concerned about the importance of communication, information gathering, and system operation, and they were skilled in this area of the study. They are favorable enablers of adoption. Physicians with excellent communication skills can assist with building physician-patient trust, obtaining a more accurate patient medical history and symptom information, reducing medical disputes and complaints caused by misunderstandings, and providing patient-centered health care [55,56]. However, another prior study had different findings, indicating that information access and communication challenges are particularly prominent in pediatric care, as children may struggle to express their feelings accurately [57]. Another study showed that communication skills on the screen were a challenge for some physicians in the beginning but become a natural way to communicate with patients with increasing online work experience [37].

Of the beliefs about consequences domain, the widely recognized advantages of telemedicine are enhancing accessibility and fairness. These are beneficial for the health care system, physicians, and patients [58]. Telemedicine improves physician-patient medical efficiency by reducing patient wait times in traditional hospitals, thus addressing the shortage of medical resources in China [59]. Physicians could improve their personal influence, attract more patients through the online platform, and facilitate patient management, all of which were facilitators in this study. Another advantage of telemedicine is providing patients who face geographical or time constraints with convenient access to medical services. Patients can quickly have a consultation with a physician online from the comfort of their homes, saving time and reducing costs [60], and all patients can enjoy high-quality medical services.

In addition to these benefits, compared with prior literature, we also provided further views on aspects that influence physician use, such as the emotion domain and social influences domain–related factors. These factors have rarely been considered in previous studies. Some physicians expressed negative emotions about online work. To be specific, the concerns regarding privacy breaches and the risk of disputes, anxiety stemming from fragmented communication that precludes real-time responses, and a sense of detachment fostered by an inability to see the patient dampen physicians’ enthusiasm for telemedicine, influencing their willingness to adopt it. Second, of the social influences domain, social support [61], organizational culture, and atmosphere [62] will significantly impact physicians’ thoughts, sense of responsibility, work satisfaction, and even the quality of medical service [63]. For example, several physicians mentioned that medical institutions and colleagues actively support the implementation of telemedicine services. The IT department is helpful when encountering issues with the operation of the information system. All these factors have created an atmosphere that supports and encourages the development of telemedicine, enabling physicians’ participation.

BCTs further refine the intervention functions and policy categories of the BCW, playing a significant role in formulating specific strategies. We selected the appropriate BCTs to ensure that the proposed intervention would be feasible with high fidelity. To address these barriers and promote telemedicine participation, BCTs such as goal setting, social support, demonstration of behavior, and prompts and cues are crucial. BCTs have been widely applied in various fields of behavioral intervention, such as physical activity intervention [64] and behavioral intervention for health professionals [65]. This study demonstrated the availability and implementation of BCTs in telemedicine environments, adding to the growing body of evidence for the utility of BCTs in telemedicine health care systems.

All barriers and enablers should receive attention, and making changes requires efforts from health administrative departments, medical institutions, departments, and health care personnel. To promote active engagement in telemedicine, three key areas must be priority. First, taking fiscal measures and upgrading the technical infrastructure through hardware and software improvements are essential for a seamless telemedicine experience for both physicians and patients. Second, establishing clear guidelines and regulations that define the rights, responsibilities, and standardized operation procedures and liabilities of medical professionals in telemedicine services will instill confidence, address concerns, mitigate legal risks, and promote active involvement in telemedicine. Finally, a positive organizational atmosphere and social support play essential roles in physicians’ use, and the support of medical institutions and departments should be provided to address motivation.

There were three limitations to consider in this study. First, the participants were all from China, which may restrict the generalizability of the findings to other health care populations and telemedicine systems. For instance, an increased workload due to telemedicine was mentioned by only a few Chinese physicians, but it is quite prominently discussed in other regions [48,66,67]. Therefore, the barriers or enablers may also vary under different health care systems or work cultures. However, we used service providers in China’s health care system as our sample, expanding the literature in this area. The findings we present are quite insightful, as we recruited participants from across the different levels of medical institutions. Future research could broaden the participant scope and conduct a comparison study. Additionally, an important strength of this study is that participants were physicians with direct experience with delivering telemedicine services. This allowed us to gather context-rich insights grounded in actual clinical practice, thereby enhancing the validity and practical relevance of the results. Second, although this study provided rich qualitative insights, data were collected solely through interviews, which may be subject to recall or social desirability biases. Future studies could consider combining interviews with observations or system usage data. Third, the TDF domains face challenges regarding definition ambiguity and domain overlap, which may impact consensus, but this study achieved a high level of agreement in coding-related barriers and enabling factors and can minimize the influence of this factor on the results to the greatest extent. As thematic saturation was achieved, the results are comprehensive in terms of coverage of the issues. Even so, a notable strength of this study is the TDF framework behavior change analyses that added evidence of physicians’ behaviors and psychology about telemedicine adoption. This approach deepened our understanding of individual-level determinants to include wider contextual drivers and facilitated the development of targeted implementation strategies.

As an innovative medical service model, telemedicine has excellent potential for future development. This study focuses on medical service providers’ adoption, which directly influences the efficacy of telemedicine and the potential success of the digital transformation of medical institutions. We offered vital insights into this significant topic by doing an exploration that used TDF, BCW, and BCTs. Based on these comprehensive and sequential frameworks, this study identified barriers and enablers, encompassing administrative challenges and certain clinical circumstances, including technical difficulties and suboptimal management. Notably, the barriers posed by emotional and social effects were not addressed in previous investigations. From a theoretical perspective, our study enhances the comprehension of individual-level elements and broader contextual influences. From a practical perspective, our findings are significant for promoting telemedicine implementation through focused interventions, including technological optimization, the establishment of clear guidelines, and the cultivation of a healthy organizational atmosphere.