We conducted the survey in Beijing, a leading Chinese metropolis in the development of telemedicine. This study focused on tertiary public hospitals, which deliver the majority of telemedicine services in China and are highly trusted and sought after by patients [30,31]. Peking University Third Hospital, a tertiary public hospital, serves as the governing institution of the Beijing Telemedicine Quality Control Center [32]. It ranks among the top 10 hospitals in Beijing in terms of telemedicine user volume, with patients coming from dozens of provinces nationwide over recent years. The hospital provides both synchronous and asynchronous telemedicine consultations, covering the 2 predominant service models in China. Its extensive online services span the entire care process, including prediagnosis, diagnosis, and postdiagnosis.

We used convenience sampling to select patients engaged in real-time video and asynchronous graphic consultations at Peking University Third Hospital. We accounted for patients’ varied geographic locations, disease classifications, and socioeconomic statuses to ensure that the sample accurately represented the telemedicine-using patient community. The inclusion criteria were as follows: (1) at least one prior use of telemedicine at Peking University Third Hospital within the past 6 months, (2) willingness to participate in the interview, and (3) adequate communication skills to engage in the interview. Previous literature suggests that thematic saturation in qualitative interview studies is generally achieved before reaching 21 interviews [33]. Accordingly, the intended sample size was 20 to 22 participants in the preliminary phase. When the data collection process yielded no new information about the enablers and barriers influencing access to telemedicine services, data saturation was reached [34].

We applied the TDF throughout the interview design, data collection, and analysis. The TDF guided the development of the interview guide, which included 1 to 3 questions for each of the 14 domains (eg, goals and intentions) [35]. For example, one of the questions was, “What do you know about telemedicine services?” Investigators with expertise in implementation science provided guidance to develop the interview guide. Moreover, we consulted with experts in hospital management, health policy management, and physicians to refine the outline. We conducted 2 preexperiments after the outline was initially finalized. The interview outline was refined slightly after the pretests, and the samples from the pretests were not included in the study due to slight changes. Table 1 presents the TDF domains and corresponding interview questions.

The formal interview was performed via telephone and approved in January 2024. We secured the participants’ informed consent and subsequently audio recorded the interviews using the software’s exclusive features. Each interview lasted approximately 10 to 20 minutes. The interviews were conducted in Mandarin Chinese by the authors KL and YY, both of whom had professional training in qualitative interviewing and extensive experience in qualitative research (n=2, one male and one female, health management practitioners). Before each interview, the interviewer provided a brief self-introduction and overview of the interview content to establish rapport with the participant. Only 2 of the interviewers were present during the interview to create a conducive environment for communication. The recordings were transcribed into textual materials within 48 hours after the interviews, and the researcher recorded and sorted the data to ensure consistency. After each interview, participants were sent a prearranged text message, and their data were collected.

Interview transcripts were transcribed, organized, and imported into the software NVivo (version 12.0; QSR International Pty Ltd) for coding. We conducted a deductive thematic analysis guided by the TDF and followed 3 steps [36,37]. First, the 14 TDF domains served as an a priori coding framework. Second, the transcripts were reviewed in detail, with relevant content identified and coded into themes. Third, these themes were categorized as either barriers or enablers within the corresponding TDF domains. When a theme was relevant to more than 1 TDF domain, it was assigned to the domain judged to be more closely aligned. Two authors (KL and YY) independently coded the transcripts. Coding discrepancies were resolved through discussion with a third author (YW) until consensus was reached. The coded κ for each level was greater than 0.90. To prepare the results, we tabulated the themes within each TDF domain and tabulated more prominent themes into barriers and enablers separately. A detailed description of the coding process, including example excerpts and coding framework, was provided in Multimedia Appendix 1.

Following encoding and analysis, we assembled a multidisciplinary panel of hospital experts, including specialists in information technology, clinical practice, and hospital administration leadership. This collaborative endeavor integrated technical, clinical, and operational viewpoints via monthly interdepartmental meetings and methodically structured brainstorming sessions employing implementation science frameworks. Using the TDF model, the panel systematically formulated various intervention options aimed at addressing the identified barriers and enablers.

This study was approved by the Medical Science Research Ethics Committee of Peking University Third Hospital (IRB00006761-M2024029). Participants provided informed consent. We stored the survey data, which were strictly managed to remain confidential. We offered a compensation of 100 yuan (US $13.8) as a token of appreciation to all participants for their involvement in the study.

This study followed established reporting standards for qualitative research and was reported in accordance with the COREQ checklist (Checklist 1).

All interviews were conducted from January to March 2024, covering a duration of 3 months. Twenty-five telephone interviews were conducted, of which 21 were considered legitimate for analysis. Of the 21 interviewees, 13 (62%) were women, and 19 (91%) held a bachelor’s degree or higher. The majority (18/21, 85%) of the participants had full-time jobs. The mean interview duration was 13.7 (SD 5.7) minutes. All participants had used telemedicine services at least once, with a mean of 3.46 (SD 2.8) visits as specified in Table 2.

A total of 190 principal assertions were documented, covering 5 domains and 28 themes, including 14 barriers and 14 enablers. Tables 3 and 4 display the themes within each domain and their frequencies.

The barriers to theoretical domains were referenced a total of 71 times. Specifically, “environmental context and resources” was referenced 33 times; “memory, attention, and decision processes” was referenced 23 times; “beliefs about consequences” was referenced 8 times; “skills” was referenced 6 times; and “beliefs about capabilities” was referenced once. In contrast, the enablers of theoretical domains were referenced a total of 119 times. “Beliefs about consequences” was referenced most frequently, occurring 71 times”; “environmental context and resources” was referenced 37 times; “beliefs about capabilities” was referenced 8 times; and “skills” was referenced 7 times.

On the basis of these findings, we designed intervention measures addressing barriers and enablers through expert group discussions (Table 5). We assessed the importance of every domain by determining the frequency of citations throughout all transcripts. The domains “beliefs about consequences” and “environmental context and resources” were the most frequently cited, with 75 and 70 mentions, respectively, accounting for 40% and 37% of the total citations.

In the “beliefs about consequences” domain, convenience emerged as a key factor encouraging patients to use telemedicine. Enhancing the experiences of first-time users was considered essential. It was advisable to offer supplementary services for patients using telemedicine for the first time, thereby enhancing their experience of its convenience.

Furthermore, in the “environmental context and resources” domain, the complexity of using the patient application represented the most prominent barrier factor. It was suggested to simplify the patient interface and limit the proliferation of complex features, hoping to facilitate patients’ focus on critical functions to meet the needs of the majority.

This study conducts an in-depth exploration of the relatively underresearched area of telemedicine among Chinese patients. It is the first instance of using the TDF model for conducting interview-based surveys with patients regarding telemedicine, thereby enhancing the scientific rigor of the research. This study found that telemedicine services adoption is jointly shaped by 14 barriers and 14 enablers, including 5 of the 14 TDF domains. This study has found more behavioral and psychological influencing factors in patients compared to other studies. Barriers arose from operational complexity, concerns about privacy and efficacy, asynchronous communication, and so on. Enablers include high convenience and accessibility, simplified system functionalities, and external support. On the basis of identified barriers and enablers, we developed targeted intervention measures and prioritized them based on their relative significance as expressed by the interviewees. In conclusion, this study advances a comprehensive understanding of the determinants influencing this adoption and establishes a theoretical foundation for developing targeted interventions in telemedicine services.

Our exploratory analysis of usage frequency indicates that barriers and enablers change alongside patients’ cumulative experiences with telemedicine services. First-time or infrequent users were more inclined to highlight challenges associated with accessing the platform and operating system, indicating the significant learning curve of initial usage. In contrast, frequent users recognized constraints in the scope and profundity of accessible system functionalities while concurrently emphasizing convenience and efficiency as primary enablers. This transition suggests a possible positive feedback loop: as patients become accustomed to the system, operational obstacles lessen, and the perceived advantages of telemedicine become more prominent, hence encouraging further utilization.

This corresponds with previous research [13,38]. Similarly, in the TAM, it is believed that “perceived usefulness” influences behavioral intention [39]. Our research delineates a self-perpetuating feedback loop: utilization of telemedicine services leads to increased satisfaction and reinforced intention to use. This phenomenon highlights the importance of the first usage experience in telemedicine service adoption. Similarly, the anchoring effect in initial interactions profoundly influences long-term commitment. Facilitating smooth navigation and practical convenience during initial interactions might create positive behavioral anchors, promoting regular use of online telemedicine services. Previous research indicates that Michigan Medicine (Ann Arbor) has established a system that incorporates template instructions and anticipatory advice, which can enhance the patient utilization process [40]. Consequently, we advocate for the deployment of multimodal guiding systems, including artificial intelligence–driven triage and a human-assisted hotline.

Our study reveals that the utilization of telemedicine services is directionally modulated by coexisting enablers and barriers within the same theoretical domains. The 2 domains, “environmental context and resource” and “the beliefs about consequences,” involve many barriers and enablers. Within the identical structural domain, both barriers and enablers arose, and the correlation between the two was notably strong.

The initial point is the discrepancy between functional complexity and operational simplicity. Within the “environmental context and resource” domain, patients simultaneously demand operational simplicity and comprehensive functionalities. To address patients’ requirements of operational simplicity, we must reduce the challenges to assure accessibility. This aligns with the findings of our and other research from the viewpoint of physicians [25,41,42]. The complexity of system operation significantly affects both physicians and patients. However, excessive feature expansion paradoxically increases interface complexity, hindering core functionalities. This phenomenon aligns with intrinsic cognitive load theory [43]. In the development of telemedicine services, it is imperative to address this equilibrium. This is a critical aspect that we meticulously consider while devising potential intervention strategies.

The second aspect is the trade-off between convenience and privacy. The “beliefs about consequences” domain encompasses opposing factors: perceived convenience and privacy concerns. In the study, the convenience of telemedicine services is the most frequently mentioned enabler by patients. However, this depends on patients supplying comprehensive personal information, medical history, and privacy. This also results in an elevated danger of personal privacy breaches. Although the issue of privacy breaches is only mentioned once in this study, it is frequently cited as a barrier in many studies on the use of telemedicine services [44]. Patients tend to resist using telemedicine services due to the risk of privacy breaches [45]. This also results in a simultaneous presence of dependence on convenience and sensitivity to contradictions, leading to conflicting and hesitant feelings in patients during the usage process [46].

In skills, digital literacy is a significant influencing element. Numerous interviewers highlighted the concern of digital literacy for older adults, despite the interviewers in the study being youthful. Older patients, despite increasing demand resulting from mobility restrictions and multimorbidity, face barriers due to limited digital literacy and adaptability [47,48]. In contrast, younger groups with technical skills demonstrate reduced demand. This generational mismatch hinders the use of telemedicine services by patients.

In establishing intervention strategies for telemedicine services, it is essential to address the requirements and features of both generations, rather than focusing on only one group. The alteration of text size was frequently referenced during the interview. Consequently, we devised an app modification strategy for the older people, offering extensive assistance while preserving the core functionality. We have also developed the “family affection account” feature, which enables children to assist the older adults in completing telemedicine-related processes.

In research primarily focusing on Western populations, privacy concerns are often cited as a major barrier to adoption [41,49-51]. Yet, this barrier was not prominent in our study, with only one participant explicitly raising privacy-related concerns. We speculate on the possible reasons from both macro levels and micro levels.

At the macro level, the social discourse and privacy awareness in China and Western countries differ markedly. In Europe and the United States, privacy protection is widely discussed [52], whereas in China, discussions regarding medical data privacy are limited, and most patients have low awareness, leading to infrequent expression of privacy concerns [52,53].

At the micro level, Chinese patients typically exhibit a high degree of trust in health care institutions, largely due to their public and government-regulated nature [54], which may further reduce concerns about potential data misuse [55].

This study has 3 main limitations. First, the sample consisted only of individuals who had already used telemedicine and were recruited from a single tertiary public hospital in Beijing, which may limit the generalizability of the findings. Future studies should consider including more diverse populations. Second, using a deductive approach to map participants’ statements to TDF domains may result in inconsistencies across coders. To mitigate this inherent limitation of deductive analysis, we used double independent coding, with any discrepancies resolved through adjudication by a third coder. Given the high consistency of participant narratives regarding barriers and enablers, we believe any residual impact is minimal. Third, as this study represents the first paper in a planned series, its primary goal was to identify barriers to and enablers of telemedicine adoption, thereby laying the foundation for intervention development. Thus, this study did not assess the effectiveness of proposed interventions. In future work, we will incorporate considerations of feasibility and other factors to prioritize specific interventions and rigorously evaluate their effectiveness.

This study offered vital insights by identifying barriers to and enablers of patients using telemedicine services based on the TDF model. These findings not only genuinely represent patients’ experiences but also reflect the behavioral and psychological factors in using telemedicine services. This study revealed barriers, including operational difficulties, extended waiting times, and doubts concerning therapeutic effectiveness. Simultaneously, the enablers included the convenience of telemedicine, time conservation, and support from hospitals. Considering these criteria, we developed intervention strategies to enhance patients’ access to telemedicine. This research offers a solid foundation for developing more targeted intervention strategies and holds substantial practical implications for improving the telemedicine experience for patients.