Journal policy drives cardiology data sharing statements, but promises still outpace access
Data sharing has become a credibility test for modern clinical research, yet cardiology still trails behind the most transparent fields. In a cross-sectional analysis spanning 2,941 clinical articles published from 2020 through 2023, only 1,004 papers, or 34.14%, included a data sharing statement. The gap was not uniform across specialties. In cardiology, 52% of articles carried a statement, far below the 96% observed in general medicine and above the 12% and 14% recorded in emergency medicine and orthopedic surgery, respectively.
Policy enforcement mattered. Articles published after journal policies took effect were significantly more likely to include a statement, a finding that places editorial rules squarely at the center of research transparency. The pattern is easy to interpret at the bedside and at the policy level: when journals require disclosure of data availability, investigators comply more often. When the requirement is weak or absent, the statement often disappears altogether.
Conditional access dominated the language authors used
The content of the statements was less reassuring than their prevalence. Thematic analysis showed that conditional availability and gatekeeping were the dominant themes. In practice, this means many papers did not promise open access to data; they framed access as something contingent on request, review, or additional permissions. Those conditions may be reasonable in some studies, especially where privacy constraints, data-use agreements, or proprietary systems are involved. Even so, conditional language creates a narrow funnel through which real-world reproducibility must pass.
A single sentence can create an illusion of openness without guaranteeing actual access.
That gap became clear when investigators contacted corresponding authors who had indicated that data were available upon request. Only 31% ultimately provided access. The rest did not follow through, despite the original statement suggesting availability. In clinical research, such a discrepancy is not a minor administrative flaw. It affects the ability of other teams to verify analyses, reproduce endpoints, pool data across cohorts, and evaluate whether secondary findings hold up outside the original manuscript.
Cardiology sits between stronger and weaker transparency cultures
The discipline-level differences were striking. General medicine was near universal in its inclusion of data sharing statements, while cardiology landed in the middle of the pack. Emergency medicine and orthopedic surgery were far less consistent. Those disparities point to differences in journal culture, editorial enforcement, and perhaps the norms of each specialty’s research ecosystem.
Cardiology occupies a particularly important place in this discussion because its studies often shape treatment pathways, device adoption, and prevention strategies across large populations. The field also carries substantial commercial, multicenter, and registry-based research activity, each of which can complicate data release. A statement about data sharing is therefore not just a compliance checkbox. It is a signal about whether a journal expects authors to make their evidence auditable after publication.
Funding status, study design, article access, and impact factor also influenced whether a statement appeared. Those associations suggest that transparency is not driven by a single editorial lever. Open access policies, journal prestige, and the structure of the study itself all seem to shape how often authors acknowledge data sharing. The finding is clinically relevant because readers often assume that high-impact publications have already cleared the bar for openness. These results show that assumption is too generous.
The gap between stated availability and usable data remains the central problem
The most consequential result may be the simplest one: only 31% of authors who said data were available upon request actually delivered them. That outcome exposes the difference between a policy statement and an operational system. A journal can require a disclosure, but it cannot force investigators to maintain organized datasets, respond promptly, resolve institutional barriers, or decide in advance what can be shared.
In clinical research, this matters because data sharing is often invoked to support reproducibility, meta-analysis, independent verification, and secondary hypothesis generation. Those goals depend on timely, complete, and usable datasets, not just permissive wording. When access is delayed or denied after a manuscript has already been published, the promise of transparency becomes hollow.
Logistical and financial barriers likely play a large role. Preparing a dataset for release can require de-identification, harmonization, documentation, storage infrastructure, and staff time. Smaller teams may lack the resources to make sharing routine. Larger groups may face institutional review constraints or contractual restrictions. None of those realities excuse poor follow-through, but they do explain why a policy alone does not produce true data availability.
Editorial enforcement appears more effective than voluntary disclosure
The strongest lever in this analysis was journal policy itself. Post-policy publications were significantly more likely to include a data sharing statement, which suggests that editorial expectations can alter author behavior quickly and measurably. This is the most actionable part of the study for journals seeking to improve transparency: require the statement, enforce it consistently, and verify that the language reflects what authors are actually prepared to provide.
For investigators, the message is equally direct. If a paper says data are available, the dataset must be organized in a way that makes release realistic. Otherwise, the statement functions more like a gesture than a commitment. In an era when clinical evidence is under constant scrutiny, that difference carries real consequences for credibility.
The study leaves one practical question hanging over the field: how many published datasets are genuinely accessible when another researcher needs them? The answer, at least in this cardiology-focused sample, is fewer than the statements suggest, and the most common failure point is not policy wording but execution after publication.