Research Letter
December 28, 2023
Lifting of Embargoes to Data Sharing in Clinical Trials Published in Top Medical Journals
Maximilian Siebert, John P. A. Ioannidis
JAMA. 2024;331(4):354-355. doi:10.1001/jama.2023.25394
Data sharing in clinical trials enhances trust and reproducibility and allows secondary analyses.1 However, sharing of clinical trials data is the exception rather than the rule2 and is often limited by embargoes, which constrain data accessibility for a specific duration for various reasons (eg, to safeguard priority rights in exploring the data). Among trials published in 3 leading medical journals in 2018-2020, 32% mentioned some embargo in data sharing statements.2 We investigated the data sharing status 3 to 5 years after publication in these originally embargoed trials.
Methods
A previous study3 found 487 randomized trials published in JAMA, Lancet, and The New England Journal of Medicine between July 1, 2018, and April 4, 2020. Here, we focused on 158 trials that claimed an embargo for data sharing in their original data sharing statement.
One researcher (M.S.) used a wildcard search with registry identifier numbers (eg, National Clinical Trials [NCT]) on Google to identify any repositories/archives in which the data might be stored. Whenever individual participant data (IPD) could not be located, the researcher contacted via email the contact person (eg, principal investigator) listed in the data sharing statement. Up to 3 emails were sent with 2 weeks between reminders between July 18 and August 23, 2023.
The primary outcome was whether IPD were made available to analyze for independent researchers. Secondary outcomes were the data sharing mechanism and differences from the original mechanism. If the embargo had not been lifted, we noted the duration of the embargo and reasons for its continuation. We also explored post hoc whether data sets were freely available and the relationship between sponsorship (industry, National Institutes of Health [NIH], nonindustry and non-NIH, mixed) and embargo lifting. For the analysis we used R version 4.0.5.
Results
Of 158 trials with embargoes, 104 trials (65.8%) had lifted their embargo (Table 1); most commonly, 46.2% (48 of 104) used repositories and 27.9% (29 of 104) required direct requests to authors. Sharing mechanisms differed depending on funding, with high use of repositories for industry-funded (67.8% [21 of 31]), NIH-funded (66.7% [12 of 18]), and mixed sponsored trials (60.0% [9 of 15]) and frequent requests to authors for nonindustry and non-NIH funded trials (50% [20 of 40]). The mechanism of sharing had changed from what was listed in the initial statements in 19 trials and was unclear in another 8. Of 19 trials, 16 (84.2%) switched from author, company, or group requests to data repository requests.
For trials for which IPD could not be located online, we received no response (or received ambiguous response) for 42 trials, and for 12 trials we received refusals to share data (eg, for 2 trials IPD became ineligible for sharing due to change in the overseeing pharmaceutical company and unapproved indications; detailed reasons appear in Table 2).
Of 28 data sets (17.8%) that had a specified end date for the data sharing, 13 had end dates after September 2023. For the remaining 15 studies in which the window for sharing ended before September 2023, we received no reply for 10 trials, and for 5 we received a reply that the data were nevertheless available.
Discussion
Within 3 to 5 years from their publication, two-thirds of clinical trials published in the top medical journals lifted the embargoes for data sharing that they had stated in their original data sharing statements. This shift toward greater openness is a positive development in the landscape of clinical trials and suggests that embargoes, while initially restrictive, can be justified in certain contexts, particularly if they eventually lead to increased data sharing.
The present study has limitations. For trials for which information could not be found or response from authors could not be obtained, it is possible that the data still may be available elsewhere. Furthermore, lifting the embargo does not mean that data necessarily will be shared, which also depends on whether further approvals are received by companies, authors, or other entities.4
While progress is evident, challenges in data accessibility exist and there is a clear need for continued efforts to enhance data transparency.
