Exploring scalable assessment methods for terminated trials in ClinicalTrials.gov: A cohort analysis of German and Californian trials

BackgroundClinical trials can terminate early for many reasons, including non-scientific reasons. We aimed to develop scalable semi-automated methods to characterize terminated trials and explore a methodology to estimate the risk of experiencing serious adverse events (SAEs) in trials terminated due to non-scientific reasons. MethodsTwo cohorts of clinical trials registered in ClinicalTrials.gov were investigated: (1) a cohort of clinical trials affiliated with German university medical centres (reported as completed between 2009-2017) and (2) a cohort of clinical trials affiliated with Californian university medical centers (reported as completed between 2014-2017). We used these cohorts to explore scalable assessment methods and compare terminated trials to completed ones regarding trial characteristics, including therapeutic focus. In a subset of trials terminated for non-scientific reasons with tabular summary results and a parallel, randomized design, we estimated the additional risk for SAE. For the German cohort, if results were missing from ClinicalTrials.gov, results from the EU Clinical Trials Register (EUCTR) were included if available. ResultsOf 2,253 German and 1,091 Californian trials, 217 (10%) and 150 (14%) were terminated, respectively. The majority (German: 65%, Californian: 67%) cited non-scientific reasons for termination, primarily low accrual. Compared to completed trials, terminated trials showed lower rates of results reporting: 35% vs. 78% in Germany and 72% vs. 87% in California. Of 242 trials terminated for non-scientific reasons, 14 (6%; 11 from ClinicalTrials.gov, 3 from EUCTR) could be included in the SAE risk assessment. No significant difference in SAE risk was observed between the intervention and control arms (RR 1.05, 95% CI: 0.76-1.44). ConclusionsResults of terminated trials were less frequently reported, limiting opportunities for knowledge generation. Broader adoption of harmonized result reporting standards across registries, structured templates, and improved trial registry logic checks could enable scalable assessment approaches and enhance the utility of terminated trial data for clinical research transparency.