Paper
Phase structure changes the story. Early development and the NA bucket behave differently from later phases.
Does reporting failure in ClinicalTrials.gov differ systematically by trial phase? We analysed the March 29, 2026 full-registry snapshot and grouped interventional studies by reported phase before calculating eligible two-year no-results rates and absolute missing-results stock. The analysis used 441,191 interventional studies overall, with 290,524 closed interventional studies and 249,507 eligible older studies driving the primary comparisons. Phase I had the highest eligible two-year no-results rate at 76.7 percent, followed by the large NA phase bucket at 65.5 percent and early phase I at 64.3 percent. By absolute count, the NA bucket contributed 96,605 unresolved two-year no-results studies, far exceeding later phases because of its scale, which shows phase structure shapes reporting behavior even before sponsor mix and therapeutic area are considered jointly. Phase III and phase IV performed better than phase I, but still remained far from transparent at 45.5 percent and 52.4 percent respectively. These phase estimates describe registry-visible non-disclosure rather than confirmed regulatory violation.