This study was approved by the Biomedical Ethics Committee of Peking University (IRB00001052-21162). Verbal informed consent was acquired from all participants prior to the interview and documented via audio recording. No patient or public involvement occurred in the study design, conduct, or reporting of this research. Participants’ privacy and data confidentiality were strictly protected via deidentification of personal information. All research data were properly preserved for academic use only. Every participant received reasonable compensation in accordance with ethical requirements.

A qualitative approach featuring semistructured cross-sectional interviews was used in this study. The study protocol and semistructured guide were jointly developed by a working group consisting of GX, JH, YW, CW, and WM.

A qualitative approach using maximum variation purposive sampling was used to recruit experts in clinical data management. Initially, 133 professionals from 103 organizations, including 47 (45.6%) pharmaceutical companies, 47 (45.6%) contract research organizations (CROs), and 9 (8.7%) academic institutions, were invited via 3 main channels: meeting attendance lists from the Clinical Research Institute of Peking University (n=70, 52.6%), membership directories of the Chinese Society of Clinical Data Management (n=29, 21.8%), and the official website of the Clinical Research Institute of Peking University together with its official WeChat Moments (n=4, 3%). Overall, 47 professionals from 47 independent organizations (n=22, 46.8% pharmaceutical companies, n=16, 34% CROs, and n=9, 19.1% academic institutions) agreed to participate in the study.

In the second stage, 44 organizations with dedicated data management teams or departments were selected, including 22 (50%) pharmaceutical companies, 14 (31.8%) CROs, and 8 (18.2%) academic institutions. Eligible experts were invited according to the following inclusion criteria: (1) holding a position as a data management leader or senior expert within their organizations and (2) providing informed consent to participate in the study. Six (13.6%) experts were unavailable or unreachable, yielding a final analytical sample of 38 (86.4%) data management experts from 38 (86.4%) distinct organizations: 13 (34.2%) pharmaceutical companies, 19 (50%) CROs, and 6 (15.8%) academic institutions.

Semistructured interviews were conducted between April and August 2022. Due to geographical constraints and COVID-19 pandemic-related restrictions, the interviews were administered by trained interviewers using a mixed-mode approach (online and offline), including telephone calls, WeChat video interviews, and face-to-face meetings. The interview guide covered organizational background, participant demographics, current CDISC implementation status (including KAP), barriers encountered, and recommendations for addressing these challenges. The complete interview guide is provided in Multimedia Appendix 1.

Experts’ knowledge of the CDISC was primarily assessed through the following question: “Do you think it is important that the National Medical Products Administration of China (NMPA) specifying CDISC standards used in submitting data is important for clinical trials?” Their attitudes toward the CDISC were evaluated using the question: “Do you support the National Medical Products Administration of China (NMPA) specifying CDISC standards used in submitting data of clinical trials?” Meanwhile, experts’ practice of the CDISC was measured via the question: “Does your institute/company use CDISC standards in submission of clinical trial data?”

Prior to each interview, participants were informed that their responses would be kept confidential and anonymized. All interviews were conducted in private, quiet environments to ensure participants could communicate openly and comfortably, free from external distractions.

Audio files were professionally transcribed into text using the online platform “iFLYREC.” The resulting text files were systematically coded using NVivo software (Lumivero). A trained interviewer generated initial codes and deductively developed a coding framework, documenting potential connections between codes that could inform subsequent theme development.

The development of the coding framework and subsequent data interpretation were guided by the Consolidated Framework for Implementation Research, which provided a clear theoretical foundation. Three core domains were used to structure the coding scheme: (1) KAP toward CDISC use, (2) barriers to CDISC adoption, and (3) recommendations for CDISC promotion. This structure ensured a systematic and consistent approach to identifying KAP, implementation barriers, and promotion recommendations. This theoretically grounded coding strategy enhanced the rigor and transparency of the analysis, enabling the identification and interpretation of key themes within a well-established implementation science framework.

To ensure accuracy in the audio-to-text conversion and coding processes, the entire workflow was conducted by a single coder (JH) and independently reviewed and validated by a panel of 6 senior data managers. Any discrepancies were communicated to the coder (JH), followed by collaborative discussion to reach a consensus.

The sample size was determined using the data saturation method, a widely adopted approach in qualitative research to ensure the comprehensiveness of collected data [11]. Recruitment was terminated once theoretical saturation was achieved [13], which referred to the point at which no new information or themes emerged from additional interviews. Data saturation was formally assessed after each of the 3 interview batches. In the first batch, 23 experts were interviewed, with only minimal new information emerging from the final 2 participants. The second batch included 11 interviews, where no new insights were generated from the last 2 experts. To further validate the achievement of saturation, a third batch of 4 interviews was conducted, which confirmed that no additional new information could be obtained. Ultimately, 38 experts completed the interviews, forming a comprehensive sample that met the criteria of theoretical saturation and ensured the adequacy of data for subsequent thematic analysis.

Qualitative data were analyzed using the 6-step thematic analysis method [12], which is rooted in constructivist and deductive approaches. This study adopted a constructivist paradigm, which posits that social reality is subjectively constructed via individual experiences and contextual interactions rather than being fixed or universal. This paradigm was selected to capture participants’ context-specific perspectives, a key element in addressing the research questions. In the process of analysis and interpretation, constructivism emphasized the researcher’s role in meaning-making, prioritized emergent themes over broad generalizations, and grounded findings in participants’ experiences; thus, the findings represented coconstructed, context-embedded meanings rather than absolute truths. Rejecting predetermined frameworks and causal inference, this approach centered participants’ narratives to unpack subjective factors, framing results as context-bound insights. Reflexivity, defined as reflecting on biases to preserve participants’ voices and ensure holistic data interpretation, was integrated throughout the process to guarantee the authenticity and depth of the findings, as well as alignment with the study’s purpose.

Quantitative categorical variables (eg, CDISC KAP) were described using percentages, while continuous variables were reported as mean, SD, minimum, and maximum values. This study adhered to the Consolidated Criteria for Reporting Qualitative Research (COREQ) checklist [13].

A total of 38 data management experts from 38 distinct organizations participated in the study. Demographic characteristics of the experts and their affiliated companies and institutes are presented in Table 1. The experts had a mean age of 39.9 (SD 6.3) years, with 19 (50%) being male participants, 21 (55.3%) employed by international companies, 38 (100%) holding at least a bachelor’s degree, and 36 (94.7%) serving in data management leadership or higher positions within their organizations.

Among the 38 experts, 36 (94.7%) mentioned that NMPA specifying CDISC standards used in submitting data is important for clinical trials, 32 (84.2%) supported this policy, and 29 (76.3%) had prior experience using the CDISC for data submission (Table 2). KAP rates varied by organizational type (Figure 1): 100% (13/13) of pharmaceutical company experts considered CDISC submission important, of whom 84.6% (11/13) supported the requirement and 100% (13/13) had implemented it; among CRO experts, 94.7% (18/19) supported the requirement and 84.2% (16/19) had implemented the CDISC; among academic experts, 83.3% (5/6) considered CDISC submission important, 50% (3/6) supported the requirement, and 0% (0/6) had implementation experience.

We summarized the CDISC implementation barriers, their underlying mechanisms, and proposed improvement strategies into a conceptual figure (Figure 2), and described the details in Tables 3 and 4 and Multimedia Appendices 2 and 3.

Of 38 experts, 17 (44.7%) identified barriers related to policy, regulatory, and sectoral factors (Table 3). Additional details can be found in Multimedia Appendix 2.

Regulatory guidelines governing CDISC application were described as lacking actionable details.

Divergent internal standards across sponsor organizations hinder unified adoption of the CDISC.

Notable differences also exist in data submission expectations between sponsor-led and investigator-initiated trials.

Gaps in talent, financial strength, and management experience also exist between large and small organizations. Smaller companies may face industry attrition risks if they cannot recruit personnel or establish technical systems capable of supporting the CDISC.

Data management capabilities among domestic Chinese pharmaceutical companies and CROs require further improvement.

Table 4 summarizes key promotional strategies from participants to advance CDISC adoption and address identified barriers. Additional details can be found in Multimedia Appendix 3.

First, of 38 experts, 33 (86.8%) proposed one or more suggestions in the policy, regulatory, and sectoral challenges facet: (1) expedite the development of detailed rules and regulations to support CDISC implementation, (2) enhance industry awareness of the CDISC through targeted outreach and education, (3) establish a CDISC-aligned standardized review system, and (4) strengthen cross-sector collaboration.

Second, out of 38 experts, 29 (76.3%) proposed two talent-related strategies: (1) expand CDISC training and advocacy to build specialized expertise, with an emphasis on affordable, high-quality programs and (2) implement professional certification programs to standardize workforce competencies.

Third, of 38 experts, 29 (76.3%) proposed four technology-focused solutions: (1) integrate CDISC requirements from the clinical trial design phase, (2) develop tools in line with the CDISC, (3) advance CDISC localization, and (4) continue developing the CDISC. Establishing a standardized CRF library and corresponding standard operating procedures can improve efficiency and reduce barriers to CDISC-compliant data generation.

Fourth, of 38 experts, 7 (18.4%) proposed cost-related adjustments: (1) avoid mandating specific software and (2) leverage long-term efficiency gains to offset initial investments.

As CDISC implementation matures, overall research and development costs may decrease.

Among the abovementioned suggestions, expertise development and strengthening policy and environmental support were most frequently emphasized by pharmaceutical industry experts. Cost control and technical problem-solving were prioritized by academic respondents.

In this qualitative study involving 38 Chinese clinical data management experts, 36 (94.7%) regarded the specification of CDISC standards for regulatory data submission by the NMPA as important for clinical trials, 32 (84.2%) supported CDISC adoption for data submission, and 29 (76.3%) reported prior experience with CDISC-compliant submissions. Key barriers to CDISC implementation and dissemination included increased implementation costs, shortages of specialized expertise, and insufficient technical, socioenvironmental, and policy supports. Corresponding strategies should focus on mitigating cost impacts, improving CDISC implementation efficiency, strengthening publicity and training, and improving regulatory frameworks. To our knowledge, this is the first study exploring KAP, barriers, and promotional strategies for CDISC implementation among Chinese data management professionals. It provides valuable insights for advancing CDISC adoption within the clinical data management sector in China and other countries with similar situations.

This study yields several important findings. First, it represents the first systematic characterization of perspectives among Chinese data management experts toward CDISC promotion and application, with most expressing support. As data management experts and their teams represent the primary end users of CDISC standards, these findings indicate favorable societal and professional support for CDISC implementation in China. Second, through qualitative expert inquiry, we identified core challenges hindering CDISC uptake, providing clear targets for finding measures to overcome these difficulties. Third, we derived actionable policy and operational strategies for promoting CDISC dissemination, which can inform future regulatory and industry guidance. Fourth, while barriers and promotional strategies were broadly consistent across the pharmaceutical industry, CRO, and academic experts, relative emphasis differed: pharmaceutical industry experts prioritized cost, policy, and environmental factors, whereas academic respondents focused more on expertise and technical challenges. These differences likely reflect industry reliance on commercial drug development revenue compared with the diversified funding sources of academic institutions, including government grants and industry sponsorship. We believe these findings provide useful promotion strategies to support CDISC implementation in China and other countries in similar situations.

This study carries important practical implications. First, it can encourage technical advances in automated CDISC data conversion, including low-cost or open-source CDISC-compliant EDC systems [14], such as OpenEDC; semiautomated legacy data conversion (clinical trial legacy data into CDISC Study Data Tabulation Model [SDTM] standards format) using supervised machine learning; and standards-based electronic regulatory submission frameworks [15,16]. Second, it supports timely and high-quality translation and localization of the CDISC, including alignment with Chinese electronic medical records and TCM standards. Although Chinese versions of SDTM, Analysis Data Model, and Controlled Terminology are available via the CDISC website, Chinese terminology coverage remains less than 1% of that of the English version. Chinese TCM experts are actively exploring CDISC application in this domain [17-21]. Third, the findings can inform the refinement of CDISC-related regulations. The United States and Japan have established comprehensive regulatory frameworks to enforce the CDISC [2,3], and the European Union has formally announced plans to adopt the CDISC for regulatory review [22]. While Chinese regulators have recommended CDISC-aligned submission, automation in clinical data management remains limited due to the lack of detailed submission guidelines. As China joins ICH and aligns drug research and development with international markets, CDISC implementation is increasingly inevitable. On the basis of these findings, we recommend that Chinese regulators adopt a phased rollout (eg, SDTM first, followed by Analysis Data Model; voluntary adoption preceding mandatory compliance), establish pilot submission programs, publish contextually appropriate technical conformance guides, and develop regulator-sponsored validation tools.

The sample was determined using the principle of theoretical saturation, whereby no new themes or conceptual insights emerged with additional interviews. Interviews were conducted in 3 batches (23, 11, and 4 participants, respectively) using a data-driven adaptive sampling strategy to evaluate saturation progressively rather than using a fixed recruitment plan. Saturation was confirmed after the third batch, with no new themes identified, supporting the adequacy of the final sample size and ensuring rigor in the qualitative analysis. Determining the sample size based on theoretical saturation not only meets research needs but also avoids unnecessarily expanding the sample size and increasing the research burden.

The study has several methodological strengths. First, the study was designed scientifically and rigorously by a multidisciplinary team with expertise in the CDISC, data management, and clinical research methodology. Second, sample size was determined using the information saturation method, with data reaching thematic saturation. Third, strict quality control procedures were implemented to ensure research credibility. Fourth, the study was carried out under the supervision of an ethics committee to ensure the privacy and rights of participants. Finally, reporting followed the COREQ checklist for transparent qualitative research reporting.

Several limitations should be acknowledged. First, representativeness was constrained by the unequal distribution of organizational types in the final sample (n=38): 19 (50%) CROs, 13 (34.2%) pharmaceutical companies, and only 6 (15.8%) academic institutions. Because none of the academic experts reported CDISC submission experience, the relatively small academic subsample may have led to overestimation of overall CDISC implementation rates. Similarly, because academic experts prioritized technical and expertise-related barriers over cost, the underrepresentation of academic experts may have resulted in underestimation of technical and expertise barriers and overestimation of cost-related challenges. Second, participants were predominantly senior data management professionals, and many from large or international organizations, chosen for their in-depth understanding of CDISC barriers and strategic recommendations. However, this participant composition limited data collection on KAP regarding the CDISC among junior data managers from smaller domestic organizations, who may have lower CDISC literacy. This may have led to overestimation of sector-wide CDISC-related KAP. Nevertheless, maximum variation purposive sampling was applied, and data reached theoretical saturation; thus, the findings remain highly informative, particularly regarding CDISC implementation barriers and promotional strategies.

The CDISC is already in use among many participating experts and their organizations, but the depth of implementation remains limited. Challenges to CDISC adoption in China span 4 domains: financial constraints, workforce skill gaps, inadequate technical support, and underdeveloped policy and regulatory environments. Strategies to promote the CDISC in China include mitigating cost burdens, expanding professional training, integrating the CDISC from the study design stage, developing automated software tools, improving localization and translation, supporting continued evolution of CDISC standards, refining regulations and implementation guidelines, establishing CDISC-aligned regulatory review systems, and strengthening cross-organizational collaboration.