Drawing Conclusions

What the clinical-trial data can tell us—and what it cannot.

What the data suggests

The dataset contains 602,514 registered studies, including 34,295 marked TERMINATED. The clearest conclusion is that termination is not evenly distributed: some study characteristics are associated with noticeably higher or lower termination proportions.

Phase II studies have the highest termination proportion among the main development phases at 11.1%. Oncology studies are at 9.3%, while healthy-volunteer studies are at 3.0%. Planned duration also matters descriptively: studies planned for 2–3 years are at 6.8%, compared with 3.5% for studies planned for less than six months.

These comparisons help identify where prevention efforts deserve attention. They do not prove that phase, disease area, or duration causes termination, and they should not be read as individual-trial predictions.

A few comparisons to remember

The chart below puts six useful reference points on one scale. Hover over a point to see the number of terminated studies and the total number of studies behind each percentage.

Reading the result carefully

The chart describes registered records, not the future outcome of a new trial. It includes studies with active, completed, unknown, and other statuses in the denominator. Planned duration may only be known after a study ends, and the categories can overlap with other design and sponsor characteristics. The strongest conclusion is therefore practical: use these patterns to ask better feasibility and monitoring questions, not to label a trial as destined to fail.

The underlying data comes from the Aggregate Analysis of ClinicalTrials.gov (AACT), a structured version of ClinicalTrials.gov. The processed tables and analysis scripts are documented in the project repository’s data/ and scripts/ directories.