For each test family, the census runs ClinicalTrials.gov's own outcome search for the family's names and reads every study first posted by the day the family was counted. These studies were read on 10 November 2026. A study counts once on a page, however many tests it names. Counts for a test cover every version of the task under any of its names, so a page can count studies that used a commercial version.
The census adds two things to each record and changes nothing in it: which outcome list names the test (primary, then secondary, then other), and the condition groups, matched on the study's condition names and MeSH terms. A study can fall in several groups; one whose only condition is cognitive impairment or decline falls in “Cognitive impairment and decline”. Studies that name only instruments outside the library count in every total and share but are not listed.
A second read on 10 November 2026 added each study's overall status, whether results are posted, its intervention types, the countries of its sites and the time frame of the outcome that names each test. A time frame counts by the latest time point it names: “Baseline, 12 and 24 weeks” is 24 weeks, a single visit is within a day, and one that names no duration, such as “end of treatment”, is left out of the timings.
A third read on 10 November 2026 added the papers the registry links to each study: those its sponsor lists as reporting results, and those PubMed links to it through the study's NCT number. A study has a results paper when one of them came out in or after the month of its primary completion and PubMed does not class it as a protocol, review or comment. A paper that never gives the NCT number is not linked, so some studies counted as reporting neither did publish.
Source: ClinicalTrials.gov, a database of the U.S. National Library of Medicine, which does not endorse these pages. Every study line is in the data release 2.1.0, and the explorer's own file is clinical-trials-studies.json.