Article type
Abstract
Background: Data sharing improves the value, synthesis, and integrity of research through access to the 50% of studies or outcomes that are unpublished, secondary analyses, or gold-standard individual patient data analyses, the harmonization of outcomes, and improved quality and integrity assessments. However, rates of data sharing are low. Data sharing might be improved if data sharing policies of journals were prominent and actionable at every stage.
Objective: To systematically assess researcher adherence to data sharing policies of journals.
Methods: This is a cross-sectional meta-research study. We included all original research published in 2022 in high-impact clinical medicine journals (the five journals with the highest impact factor for all 59 fields of clinical medicine) with a data sharing policy that either recommended or required data sharing. We manually collected data on the initial data sharing plans in study registration records and published study protocols and the final data sharing decisions in published original research.
Results: We have included 58 journals, from which we have included a total of 8,842 studies that met the inclusion criteria (original research published in 2022). Few studies had a study registration record or published study protocols (11.1%, n = 981). Of those which did, half stated their data sharing intentions in their publicly available protocol (54.7%, n = 537), and of these, less than a third intended to share data (31.5%, n = 169). Most studies stated their data sharing decisions in their publication (58.3%, n = 5,154). Of those that did, most intended to share data (89.3%, n = 4,605), and this did not differ by the presence of industry involvement. Those that did not intend to share data (10.6%, n = 545) most commonly did so for no reason (30.1%, n = 164) because they were not the data custodian (20.2%, n = 110) or to protect participant privacy (13%, n = 69). Compared with studies published in journals that recommended data sharing, studies published in journals that required data sharing were more likely to have a data sharing statement (82% vs 49.4%) and were more likely to intend to share data (75.1% vs 43.4%).
Conclusion: Preliminary results suggest suboptimal researcher adherence to, and insufficient journal enforcement of, data sharing recommendations or requirements by journals.
Objective: To systematically assess researcher adherence to data sharing policies of journals.
Methods: This is a cross-sectional meta-research study. We included all original research published in 2022 in high-impact clinical medicine journals (the five journals with the highest impact factor for all 59 fields of clinical medicine) with a data sharing policy that either recommended or required data sharing. We manually collected data on the initial data sharing plans in study registration records and published study protocols and the final data sharing decisions in published original research.
Results: We have included 58 journals, from which we have included a total of 8,842 studies that met the inclusion criteria (original research published in 2022). Few studies had a study registration record or published study protocols (11.1%, n = 981). Of those which did, half stated their data sharing intentions in their publicly available protocol (54.7%, n = 537), and of these, less than a third intended to share data (31.5%, n = 169). Most studies stated their data sharing decisions in their publication (58.3%, n = 5,154). Of those that did, most intended to share data (89.3%, n = 4,605), and this did not differ by the presence of industry involvement. Those that did not intend to share data (10.6%, n = 545) most commonly did so for no reason (30.1%, n = 164) because they were not the data custodian (20.2%, n = 110) or to protect participant privacy (13%, n = 69). Compared with studies published in journals that recommended data sharing, studies published in journals that required data sharing were more likely to have a data sharing statement (82% vs 49.4%) and were more likely to intend to share data (75.1% vs 43.4%).
Conclusion: Preliminary results suggest suboptimal researcher adherence to, and insufficient journal enforcement of, data sharing recommendations or requirements by journals.