Phase 3 Trial Geography: Listed Sites, Study Counts, and Recruitment Status

ClinicalTrials.gov records whose phase text contains Phase 3 list 1,513,179 location entries across 49,660 studies. The United States accounts for 639,616 entries from 18,705 studies. The next eleven countries by entry count account for 514,248 entries. Poland records 35,050 entries from 4,978 studies, while Australia records 25,275 from 4,327. Restricting the cohort to four open study statuses preserves the United States lead with 181,206 entries. Location status is missing on 1,329,054 entries, so the full geography measures registered network footprint rather than active recruitment.

1. Cohort and location rules

The cohort includes study records whose phase text contains Phase 3, including combined phases. Each location array element counts once on each study. The same facility can appear on many studies and under several spellings. Country totals therefore count study-location entries rather than unique physical sites, participants, enrolled patients, or currently recruiting locations. Records without any location entry remain in the study denominator.

2. United States entries exceed the next eleven countries

The United States accounts for 639,616 location entries, or 42.3 percent of all entries. Germany follows with 66,573, China with 64,010, Japan with 60,899, and France with 57,813. The eleven countries ranked behind the United States sum to 514,248 entries. The comparison describes registry entries under one source’s coverage rules, not national research capacity.

3. Site entries and study counts produce different rankings

The United States entries occur across 18,705 studies. Canada appears on 7,423 studies, Germany on 7,241, France on 7,224, and China on 6,465. Japan has 60,899 entries across 3,767 studies, indicating more listed locations per represented study than several peers. Facility-city strings add a third unit but remain sensitive to spelling, institution naming, and duplicate addresses.

4. Open-status sensitivity preserves the United States lead

Among recruiting, not-yet-recruiting, active-not-recruiting, and enrolling-by-invitation records, the United States has 181,206 listed entries. China has 28,901, Japan has 16,639, and France has 15,213. The next eleven countries sum to 144,544. The sensitivity removes closed study statuses at the record level, but a location within an open study can still be inactive or missing its own status.

5. Location counts per study are strongly skewed

Among studies with locations, the median is three entries and the mean is 34.2. The ninetieth percentile is 94, while the maximum is 1,745. The eleven-fold mean-to-median gap reflects a long upper tail of multinational or large-network studies. A country can accumulate many entries through fewer large studies, so entry counts and study counts must remain separate.

6. Individual location status is usually missing

Location status is missing on 1,329,054 entries, or 87.8 percent. Recruiting appears on 142,149 entries, not-yet-recruiting on 20,673, and active-not-recruiting on 9,195. Completed, withdrawn, suspended, terminated, and enrolling-by-invitation values form smaller categories. The missing share prevents a complete site-level recruitment census.

7. Supported interpretation

The registry supports three conclusions within its coverage. United States Phase 3 location entries exceed the next eleven countries under both full and open-status study cohorts. Entry counts, distinct study counts, and facility-city strings measure different aspects of registered networks. Missing location status prevents an active-site census. The analysis therefore describes registered geographic footprint and must not be converted into enrollment, recruitment performance, or national research-capacity estimates.

8. Limitations

One physical institution can appear under several names and on many studies. A location can remain listed after recruitment ends or never starts. Country values reflect ClinicalTrials.gov obligations and voluntary registration, not all national registries. Facility-city pairs are normalized only by letter case and remain imperfect. Studies without locations contribute no country entry, and later updates can change historical arrays.