Background: Individual participant data (IPD) meta-analysis is often regarded as the gold standard for evidence synthesis. An IPD approach can substantially improve data quality by offsetting inadequate reporting and allowing more flexible and detailed analyses such as participant-level subgroup analysis (Tierney, Stewart, & Clarke, 2019). The International Committee of Medical Journal Editors declared data sharing an ethical obligation and since 2019 requires all clinical trial registrations to contain a data sharing plan (Taichman et al., 2017). Yet, IPD meta-analyses are often limited by the reluctantance of many investigators to share their data. Data sharing statements submitted to clinical trial registries offer a unique resource to explore barriers and facilitators to data sharing (Tierney et al., 2019).

Objectives: Our aim was to determine data sharing willingness reported in recently registered clinical trial records, how this relates to clinical trial characteristics, and principal investigators’ attitudes, motivations and barriers to data sharing.

Methods: We conducted a retrospective cohort study and an online cross-sectional survey of all interventional trials registered on the Australian New Zealand Clinical Trial Registry (ANZCTR) from 1 December 2018 to 30 November 2019.

Results: In the retrospective cohort study (n=1517), data sharing willingness was low (22%) and, as shown in the Figure, was significantly lower for trials with industry involvement (OR 0.52, 95% CI 0.37-0.72) and in non-randomised (OR 0.69, 95% CI 0.52-0.91) or phase 1 trials (OR 0.38, 95% CI 0.21-0.65, χ2 (5) = 19, p = 0.002). Data sharing willingness was not associated with health condition studied, presence of a control group, trial purpose, sample size or primary sponsor country. In the online cross-sectional survey (n=281, 23% response rate), investigators showed strong support for the concept of data sharing (77%) but a substantially lower intention to share data for their own trials (40%). Major concerns included inappropriate data use by other investigators (97%), obtaining study participant informed consent to share (97%), preventing misinterpretation or misleading secondary analyses (97%) and spending undue time or effort preparing the data for sharing (96%).

Conclusions: There is insufficient willingness to share data, particularly amongst trials with industry involvement. Addressing the identified barriers would support the reuse and replication of research, strengthen the transparency and reliability of trials, and facilitate statistical techniques such as IPD meta-analyses.

Patient or healthcare consumer involvement: Healthcare consumers will be invited to comment on this research project.