CT.gov Evidence Visibility Gap

Abstract

When posted results and linked publications are read together for older interventional trials, how visible are they? We examined 249,507 closed interventional studies drawn from the full 578,109-study registry snapshot taken on 29 March 2026. Using the ClinicalTrials.gov API and a Python program, we sorted each study into one of four evidence states: results with a publication, results without, publication without results, or neither. We found that 42.7 percent showed neither posted results nor a linked publication, whereas only 13.7 percent showed both. Publication-only visibility was 30.0 percent, and sponsor classes differed significantly, with other-government worst at 49.1 percent ghost protocols while federal sponsors led on full visibility at 33.5 percent. Reading results tabs and linked papers together shows that older registry evidence is more often partially or wholly invisible than fully visible. This captures registry-visible evidence using internal publication links, not exhaustive external bibliographic matching.