Introduction

Sharing individual participant data (IPD) from clinical trials helps advance scientific knowledge, promote transparency in reporting, and fulfill ethical obligations to trial participants1. Data sharing from cardiovascular disease (CVD) trials is of substantial importance, especially given CVD as the leading cause of global disease burden and death in the past three decades2. The International Committee of Medical Journal Editors (ICMJE) required that trial investigators must include a data sharing statement in their trial reports since from July 2018, aiming to enhance data sharing3. Widespread support has been subsequently achieved to promote publication of data sharing statements in clinical trials and in the hopes of enhancing the actual sharing of IPD eventually4,5,6.

Journals’ requesting and publishing data sharing statements, as the first essential step towards encouraging trial investigators and sponsors to consider sharing data, has been reported to effectively advance actual data sharing7,8,9. We previously observed that 73% (2345/3229) of biomedical journals requested data sharing statements upon trial report submission10. As found from our post hoc analysis, a similar proportion (76%, 96/126) was detected in CVD journals that had data sharing statement request. Nevertheless, it was unclear whether these journals would comply with their declared request to indeed publish data sharing statements in clinical trial reports. On the other hand, editors-in-chief are well-known to play a pivotal role in generating and implementing journal requests (including data sharing statement request) in clinical trials. Soliciting their opinions can help with our in-depth understanding and thus enhance the current practice of publishing data sharing statements in cardiology trials.

Therefore, we conducted a quantitative and qualitative analysis to explore current practice of publications of data sharing statements in clinical trials in CVD journals that declared to request data sharing statements, and to survey perspectives from their editors-in-chief on this topic. Our study findings may advance knowledge on data sharing statement publications in CVD journals and potentially help promote the actual data sharing in clinical trials.

Methods

We performed the quantitative analysis to evaluate the practice of publishing data sharing statements for clinical trials by CVD journals, and conducted the qualitative analysis to explore the in-depth perspectives of editors-in-chief. This quantitative and qualitative analysis could complement each other by providing a comprehensive assessment of both current practice and underlying mechanisms of publishing data sharing statements in CVD trials.

This study received an ethical waiver from the Ethical Review Committee of Guangdong Second Provincial General Hospital (2024-KY-KZ-046-01) because all data were deidentified and no personal or sensitive information could be linked. We used the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline as a framework for our study due to its aim at transparent and comprehensive reporting of observational research11. We adhered to the STROBE guideline to report the selection of journals, data extraction, and statistical analysis.

We included the CVD journals if they were categorized as “Cardiac & Cardiovascular Systems” defined by the Journal Citation Reports (JCR, as of June 2023), if they published any clinical trial report on IPD from Jan 2019 to Dec 2022, and if they requested data sharing statements upon trial report submission. The inclusion of CVD journals publishing clinical trial reports from January 2019, was determined to align with the ICMJE data sharing policy (effective from July 2018). This 6-month gap allowed sufficient time for journals to implement and comply with this policy, which was in line with previous studies10,12. We searched the keywords (“trial” and “clinical trial”) in PUBMED and journals’ websites to obtain the publication of trial reports with IPD between 2019 and 2022. The assessment of data sharing statement request for the included journals had been described previously10. In brief, we searched journals’ submission instructions about their statement request through utilizing the keywords “data sharing”, “data availability”, “data accessibility”, “research data”, “data deposition”, and “data deposit”. If journals explicitly requested trial authors to provide a data sharing statement with their submitted manuscripts in journals’ submission instructions, they were grouped to have a request for statements. Some journals do not explicitly request statements in submission instructions, but may require authors to upload a statement separately to the submission system or ask about authors’ intent to share data in the submission system. These journals were also grouped to have a request for statements. To avoid potential misclassification, we ran mockup submissions for these journals by going through all steps in their submission systems, and found none requested statements from their submission systems. There were 96 CVD journals identified to request statements upon trial report submission.

To compare the journals’ request with their subsequent publications of data sharing statements, we further restricted those journals that had published clinical trial reports between Sep 2023 and May 2024, because we extracted data on journal request for data sharing statements from Sep 2023 (Fig. 1 displays the selection process and Supplemental Tables 1–2 show the detailed journal list). Therefore, a total of 78 journals were included for analysis. Subsequently, there were 78 editors-in-chief from these journals invited to our online survey. We used Google Translate to translate into English for those journals that were not in English or Chinese language.

Fig. 1
figure 1

Flow diagram showing the selection of journals and editors-in-chief for this quantitative and qualitative study.

Full size image

Data extraction

All the data on journals was extracted independently by study authors from Sep 2023 to Jan 2024. Any disagreements were resolved through discussion among the authors, with consultation from a senior author (GL) if consensus could not be reached. We extracted data on journal characteristics, including publication language, percentage of open access (the percentage of open access items among all citable items published in the journal in the previous three years [https://journalcitationreports.zendesk.com/hc/en-gb/articles/28351462652433-Open-Access], sex of editor-in-chief, country of editor-in-chief’s affiliation, publisher, journal impact factor in 2022 (released in June 2023), JCR quartile, whether the journal endorsed the Consolidated Standards of Reporting Trials (CONSORT), whether the journal was on the ICMJE list, and number of trials published between 2019 and 2022.

We categorized publishers into three groups (most common, common, and others) based on their frequencies of the included CVD journals (Supplemental Fig. 1). Specifically, the publisher Elsevier was grouped as most common given its highest frequency that was far more than the remaining publishers. Publishers with frequencies >5 but less than the highest, were categorized as common, while publishers with frequencies ranging from 1 to 5 were grouped as others. There were no duplicate editors-in-chief included for analysis, because no multiple journals shared the same editors-in-chief. For those journals having >1 editors-in-chief, we selected the first one shown in the journal webpages to avoid double counting13. This approach ensured that each journal was represented by a single editor-in-chief, as collecting data from multiple editors-in-chief from one journal would lead to over-counting and thus inadequate representation of this specific journal among all the editors’ perspectives.

We sent the online survey by email to 78 editors-in-chief to collect anonymous data on their opinions. We conducted this survey from May to Jun 2024, during when we sent the first invitation and then two reminders to all editors-in-chief (each with a one-week waiting time). No personal information (including names, sex, ages or affiliations) or Internet Protocol addresses were collected, because in our pilot test, collecting any personal information would substantially lower the response rate and thus the feasibility of this survey. All the invited editors-in-chief were voluntary to participate. In the survey, we first provided detailed information on the research context and the importance of soliciting opinions from editors-in-chief. Next, we asked whether the participants consented to the survey by raising the question “Are you aware of and willing to proceed with the survey?”. Only those selecting the “Yes (continue)” option were allowed to answer the subsequent questions (Supplemental Table 3).

Outcomes

For the quantitative analysis, our outcome was whether journals published data sharing statements in their published trial reports. We retrieved all published trial reports between Sep 2023 and May 2024, by searching PubMed for each included journal. After thoroughly reading the trial reports, the webpages where the journals published these reports, and their supplemental materials, we determined whether these published trial reports had any data sharing statement. Journals were therefore dichotomized into publishing any data sharing statements in their published trial reports (Yes or No).

Regarding the qualitative analysis, the outcome was the opinions of editors-in-chief on publishing data sharing statements, including their perspectives on reasons/importance, and obstacles/difficulty for data sharing statement publications. We also collected data on whether the editors-in-chief had any initiatives to promote the publications of data sharing statements in their journals. The questionnaires used for this qualitative analysis were initially developed from group discussion among all the study authors, and modified and finalized after pilot test. Supplemental Table 3 shows the questionnaires used in this study.

Statistical analysis

Continuous variables were described using medians and their lower and upper quartiles (Q1, Q3), while categorical variables were presented as counts and percentages. Medians were used to dichotomize continuous variables if necessary. To compare journals with and without publications of data sharing statements, we used t-tests and Chi-square tests for continuous and categorical variables, respectively.

Among the included CVD journals requesting data sharing statements upon trial report submission, they would differ in the strength of their wording. Therefore, based on our group discussion, we classified the included journals using terms “encourage” or “recommend” as having Weak data sharing statement request, while those using terms “require”, “should”, “request”, “mandate” or “must” were grouped as having Strong request. Subsequently, we performed a subgroup analysis to compare journals with and without publications of data sharing statements by their strength of request (Weak or Strong).

We also conducted multivariable logistic regression analysis to examine the association between journal characteristics and journals’ publications of data sharing statements. Given this analysis with an exploratory nature, only three journal characteristics (publisher, whether on the ICMJE list, and CONSORT endorsement) were adjusted for in multivariable analysis, because they were previously reported to associate with data sharing statements or adherence to ICMJE requirement12,14,15. Results were presented as odds ratios (ORs) with corresponding 95% confidence intervals (CIs), where an OR > 1 indicated an increased odds of publishing data sharing statements. All statistical tests were conducted with a two-sided significance level of 0.05. The analyses were performed using R software version 4.1.1 and SAS software version 9.4.

For the qualitative analysis, because in this study we aimed to describe editors’ opinions empirically from a small sample of participants, we adopted a similar method to inductive thematic analysis to identify and group analogous perspectives of surveyed editors-in-chief for description. This approach was in line with other previous studies16,17. Specifically, participants’ responses for each survey question (reasons/importance, obstacles/difficulty, and initiatives to promote publication of data sharing statements) were first grouped based on the keywords extracted. Some response may be placed into ≥ 2 groups due to its multiple keywords. The keywords were subsequently manually coded and expanded independently by two study authors (YL and GL) to represent data from opinions of editors-in-chief naturally. We then counted the frequencies of codes to present the information on perspectives of surveyed editors-in-chief18. Results were subsequently displayed in descending order of the counts. For instance, the keyword “reporting” was extracted from 8 participants’ responses and coded as “transparent reporting”, shown as the first reason/importance of publishing data sharing statements. Nevertheless, due to the heterogeneity of the information collected from the survey, no synthesized analysis could be performed.

Consent for publication

All participated editors-in-chief consented to participation in the survey. The Ethical Review Committee of Guangdong Second Provincial General Hospital waived consent for publication of the survey data because all the data were deidentified and no personal or sensitive information could be linked.

Results

Among the 215 CVD journals initially pooled for screening, we included 78 journals for analysis (Fig. 1). All the included CVD journals were published in English language (Supplemental Table 2 displays the complete journal list). As shown in Table 1, they had a median journal impact factor of 3.4 (2.1, 6.3) and published a median 62 (25, 105) trials between 2019 and 2022. There were 33 (42.3%) journals for which the editor-in-chiefs’ affiliations were from USA (Supplemental Fig. 2). The majority of journals had male editors-in-chief (87.2%) and endorsed CONSORT (76.9%). Only 23.1% (n = 18) were on the ICMJE list.

Table 1 Descriptions of all included journals and comparisons between journals with and without publications of data sharing statements*.
Full size table

There were 54 (69.2%) journals publishing data sharing statements in clinical trials (Table 1). Significant differences in publishers and CONSORT endorsement were observed between journals with and without publications of data sharing statements. There were 60 and 18 journals with weak and strong data sharing statement request, among which 39 (65.0%) and 15 (83.3%) journals published statements respectively (Tables 2, 3). No significant difference in the proportion of journals publishing statements was found between journals with weak and strong request (P = 0.14). For analysis limited to journals with weak request, significant differences were detected in publishers and CONSORT endorsement when comparing journals publishing data sharing statements to those that did not. Among the journals with strong request, journals publishing statements had a significantly higher journal impact factor and published more trials between 2019 and 2022 than journals that did not publish statements.

Table 2 Descriptions of journals with weak request and comparisons between journals with and without publications of data sharing statements*.
Full size table
Table 3 Descriptions of journals with strong request and comparisons between journals with and without publications of data sharing statements*.
Full size table

Figure 2 displays results for the relationship between journal characteristics and journals’ publications of data sharing statements. Neither being on the ICMJE list (OR = 0.59, 95% CI: 0.17–2.03) nor CONSORT endorsement (OR = 1.67, 95% CI: 0.47–5.78) was significantly associated with journals’ publication of statements.

Fig. 2
figure 2

Multivariable logistic regression analysis results for the association between journal characteristics and publication of data sharing statements.

Full size image

In the qualitative analysis, 43 editors-in-chief participated in the survey (response rate: 55.1%) (Fig. 3). All respondents agreed on the importance of publishing data sharing statements in clinical trials. There were 41 editors-in-chief who provided perspectives on publishing data sharing statements, among whom 33 editors provided reasons/importance and 12 provided obstacles/difficulty in publishing statements. Specifically, among the 33 respondents providing reasons/importance for publishing statements, 8 (24.2%) editors-in-chief believed the publications of statements could promote transparent reporting and advance discovery and hypothesis. Among the 12 editors who provided perspectives on obstacles/difficulty in publishing statements, 4 (33.3%) concerned that the data sharing statements published may be incongruent with authors’ willingness to share data.

Fig. 3
figure 3

Survey results of editors-in-chief of the included CVD journals regarding publications of data sharing statements in clinical trials.

Full size image

A total of 9 editors-in-chief responded that they had initiatives to promote journals’ publications of data sharing statements (Fig. 3). The most frequent initiative mentioned was to ask editorial office, reviewers, or handling editors check statements before the final acceptance of submissions (n = 4, 44.4%), followed by reminding the production office before trial report publications (n = 3, 33.3%). Other initiatives mentioned included developing a statement template that had to be submitted, providing more detailed guidance about statements on journal submission instructions, and uploading education video for statements to journal submission instructions, among others.

Discussion

In this quantitative and qualitative study, we explored the data sharing statement publications by CVD journals and surveyed perspectives from their editors-in-chief. Among the 78 journals that declared to request data sharing statements upon trial report submission from journal submission instructions, over 30% journals (24/78) indeed did not publish any statement in their published trial reports. Among the 33 editors-in-chief providing reasons/importance for publications of data sharing statements, approximately one quarter (n = 8, 24.2%) believed that journals’ publications of data sharing statements would promote transparent reporting and help enhance discovery and hypothesis. The most frequent obstacle/difficulty in publishing statements, as mentioned by the editors, was that the published statements might not accurately capture whether trial authors would be willing to share.

With the evolving impact of ICMJE requirement, data sharing statements in clinical trials as a critical avenue of promoting actual data sharing, have been increasingly published by journals19. Nevertheless, we were not aware of any evidence about the actual publications of data sharing statements in CVD journals, especially in those journals clearly requesting a statement from trial authors upon trial submission. Notably, our results revealed a substantial inconsistency between journal request from their submission instruction pages and the actual publications of data sharing statements. This demonstrated a big gap between declared and implemented data sharing statement request in CVD journals. No significant difference in proportion of journals publishing statements was observed when comparing journals with strong request to those with weak request, indicating a suboptimal practice of publications of statements in CVD journals regardless of their strength of request.

Being on the ICMJE list was related with 41% decreased odds of journals’ publications of data sharing statements, albeit non-significantly. This was contrary to the journals’ claimed commitment to endorse ICMJE requirement for enhancing the quality of medical science and its reporting20. By contrast, CONSORT endorsement was non-significantly associated with increased odds of publications of data sharing statements. Clinical trials published in journals endorsing CONSORT were reported to have higher quality of reporting and evidence, when compared with trials from journals without CONSORT endorsement21. Given the extensive endorsement of CONSORT in biomedical journals, efforts to incorporate provision of data sharing statements into CONSORT iterations for trial reports may therefore help enhance journals’ publications of data sharing statements.

From our qualitative survey, all responded editors-in-chief agreed on the importance of publishing data sharing statements in clinical trials, indicating widespread acceptance of the ICMJE requirement. However, among the 12 editors-in-chief providing obstacles/difficulty in publishing data sharing statements, one third (n = 4, 33.3%) were concerned about the inconsistency between declared data sharing statements and the actual data sharing as the main obstacle to journals’ publications of statements. Previous studies have reported this inconsistency7,22,23,24. For example, a study based on clinical trials published in JAMA, Lancet, and the New England Journal of Medicine found that among the 89 trials declaring their IPD would be stored in repositories, only 19% (n = 17) actually deposited data7. Another study that included trial reports from 333 open access journals published in 2019 by BioMed Central, also revealed that their data sharing statements could not ensure actual data sharing22. As the editors-in-chief pointed out from our survey, besides trial authors’ concern (over data protection, patient privacy or copyright), the lack of infrastructures, databases or platforms to store and manage IPD, was also one of the main obstacles that led to this inconsistency. Therefore, endeavors are required to progress to an integrated system for adequately storing, managing and eventually sharing IPD, which could help enhance publications of data sharing statements by journals and reduce this inconsistency25,26.

There were two editors noting that trial authors often responded ambiguously with “it depends” among the 12 editors-in-chief providing obstacles/difficulty in publishing statements. This highlighted more specific guidance needed from journals for publishing data sharing statements to deliver explicit information and avoid confusion. Furthermore, lack of enforcement by publishers was also emphasized by two respondents, which indicated joint efforts from journals and publishers required to improve journals’ publications of statements. Of note, while we could not extract specific and direct information on publisher’s publications of data sharing statements in clinical trials, this qualitative analysis expanded our knowledge of understanding the obstacles and potential mechanisms regarding the current suboptimal practice of publishing data sharing statements in CVD journals.

Previous studies have explored the gap between journal policies of data sharing statements and their actual implementation in published trials. E.g., two studies reported that only 49–52% of published trial reports included data sharing statements27,28. Nevertheless, these two studies only focused on the top 10 journals in urology27 and ophthalmology28. Thus they could not comprehensively capture the state of the practice in their research fields. Also, little was known about the practice of CVD journals and trial reports. By contrast, we included all eligible CVD journals to assess the state of the practice of data sharing statements and enforcement in CVD journals, and provided empirical evidence by surveying editors’ perspectives. Therefore, our results could generate novel evidence to advance journals’ publications of data sharing statements and to help promote the actual data sharing in cardiology trials eventually.

The most frequent strategy for CVD journals to address the gap between their declared request for data sharing statements and the actual publication, was to have the relevant parties (editorial office, reviewers, or handling editors) verify the statements before final acceptance of submissions. Developing an education program and uploading a short video to journal guidance pages, was also noted to help make trial authors realize the importance of data sharing statements promptly and enhance their provision of and even adherence to statements. This approach was indeed similar to strategies for effectively promoting patients’ medication adherence29,30. Thus, these qualitative results provide in-depth insights into practically improving journals’ publications of data sharing statements in clinical trials by CVD journals.

Strengths and Limitations

This analysis comprehensively examined the current practice and generated new evidence regarding journals’ publications of data sharing statements in clinical trials by CVD journals. The qualitative survey from editors-in-chief also strengthened our study by providing informative perspectives and potential measures for improvement.

Several limitations existed in this study. First, although we aimed to include all CVD journals, some general journals that also frequently published cardiology trials (for instance, JAMA and the New England Journal of Medicine), were not grouped as “Cardiac & Cardiovascular Systems” by the JCR and thus not included for our analysis. Moreover, we excluded the CVD journals that declared request for data sharing statements but did not publish trial reports between Sep 2023 to Jan 2024; therefore it was unclear whether a different practice would exist for these journals. Our findings could only reflect the practice of journals’ publications of data sharing statements and the perspectives of editors-in-chief in 2023–2024, because we did not take into account the dynamic changes in journals’ request for data sharing statements and their editors across time. The relatively small number of included journals impeded our further analysis. Likewise, no multivariable analysis could be performed for journals by their strength of request. Results from the quantitative analysis should be interpreted with caution, given the observational design with potential confounding and biases31,32. For journals with multiple editors-in-chief, we only invited the first one, which may introduce potential selection bias to an unknown extent. Due to concerns over response rate and the feasibility, no information on participants’ characteristics was collected from the qualitative survey, thereby precluding our analytical exploration. For instance, no analysis examining whether publisher affiliation was a contributing factor to the observed differences in opinions of editors-in-chief, could be performed.

Conclusions

Among the CVD journals declaring to request data sharing statements upon trial report submission from journal submission instructions, over 30% (24/78) indeed did not publish any statement in their published trial reports. The most frequent obstacle/difficulty in publishing statements, as mentioned by the participated editors-in-chief, was that the published statements may be incongruent with trial authors’ willingness to share data. Improving journals’ publications of data sharing statements in clinical trials may be an important step towards promoting the actual data sharing and transparency in CVD trials.