Clinical Trial Results Disclosure: Compliance with Ethics & Law

Description

Dataset description This dataset contains 14,614 cancer-related clinical trials registered in ClinicalTrials.gov and sponsored by 2,061 organizations, including pharmaceutical firms, universities, and other organizations, across 112 countries. The unit of analysis is the individual clinical trial. The dataset was constructed to examine whether clinical-trial sponsors disclose trial results within 12 months of trial completion and whether disclosure varies across ethical and legal regulatory periods. Only completed trials were included. Ongoing trials and trials without a valid completion date needed to establish the 12-month disclosure window were excluded. Completed trials without results posted within 12 months were retained and coded as non-compliant. The main dependent variable is timely disclosure, coded 1 when trial results were disclosed within one year of completion and 0 otherwise. The dataset includes variables identifying the regulatory period in which each trial was completed, including the period surrounding the 2013 Helsinki Declaration and the 2017 FDAAA enforcement change. It also includes trial-level, sponsor-level, technological, and country-level characteristics used as explanatory and control variables. Key variables include sponsor type, trial completion year, trial phase, intervention versus observational design, enrolment size, number of sponsors, age and sex focus of the trial, healthy-volunteer status, NIH funding, other U.S. federal funding, host-country population, intellectual-property protection, and whether the host country is an OECD member or China. The empirical analysis associated with the dataset uses multilevel mixed-effects logistic regression, accounting for clustering at the organizational, technological, and national levels. Alternative hierarchical specifications are also used as robustness checks. Suggested repository keywords Clinical trials; ClinicalTrials.gov; result disclosure; transparency; regulatory compliance; pharmaceutical firms; universities; oncology; Helsinki Declaration; FDAAA; NIH funding; institutional logics; multilevel logistic regression Short version This dataset contains 14,614 completed cancer-related clinical trials from ClinicalTrials.gov, sponsored by 2,061 organizations across 112 countries. It examines whether trial results were disclosed within 12 months of completion and how compliance varies across sponsor types and the 2013 Helsinki Declaration and 2017 FDAAA regulatory periods. The dataset includes trial, sponsor, technology, funding, and host-country characteristics used in multilevel logistic-regression analyses.

Steps to reproduce

Steps to Reproduce Obtain the source data. Download cancer-related clinical-trial records from ClinicalTrials.gov for the study period. Restrict the sample to completed trials. Exclude ongoing trials and records without a valid completion date needed to establish the disclosure window. Construct the compliance variable. Code a trial as 1 if results were disclosed within 12 months of completion and 0 otherwise. Completed trials without results posted within 12 months remain in the sample and are coded 0. Code sponsor type. Distinguish university-sponsored trials from firm-sponsored trials according to the main sponsor classification. Construct the regulatory-period variables. Use trial completion year to identify the relevant periods: pre-2013, 2013–2016, 2017 onward. Construct event-time variables for robustness analysis. For the Helsinki analysis, calculate completion year − 2013. For the FDAAA analysis, calculate completion year − 2017. In each case, use the year immediately before the intervention (−1) as the reference category. Construct control variables. Prepare the trial-, sponsor-, and country-level controls described in the Methods, including trial phase, enrolment size, intervention type, number of sponsors, NIH funding, other U.S. federal funding, demographic focus, host-country population, intellectual-property protection, and OECD-plus-China location. Estimate the main models. Run multilevel mixed-effects logistic regressions with timely disclosure as the dependent variable and report odds ratios. Estimate interaction effects. Interact university sponsorship with the regulatory-period variables to test whether universities and firms respond differently to the 2013 and 2017 changes. Run robustness checks. Re-estimate the models using alternative temporal coding, event-time specifications, alternative disclosure windows where applicable, and alternative multilevel nesting structures. Compare results. Confirm that the direction and statistical significance of the main sponsor-period effects are consistent across the principal and robustness specifications. Verify sample consistency. Ensure that the final number of trials and sponsoring organizations is identical across descriptive statistics, regression models, tables, and appendices.

Institutions

• Liaoning University

  Liaoning, Shenyang
• Manchester Metropolitan University

  England, Manchester

Categories

Licence