Every county, district and state, as a file you can fetch.

Addresses

Base URL: . Districts use the 119th Congress maps (2024 elections); the plan field says so in every district file.

One line of code

// JavaScript, in any web page
const r = await fetch("api/district/TX-34.json").then(r => r.json());
console.log(r.name, r.pct_lt20_trials_within_60rdmi + "% of residents 55+ short of trials", r.member.name);

# Python
import requests
d = requests.get("api/county/48215.json").json()
print(d["name"], d["people_lt20_trials_within_60rdmi"], "residents 55+ with fewer than 20 trials within 60 road-miles")

# R
d <- jsonlite::fromJSON("api/all_counties.json")
head(d[order(-d$people_lt20_trials_within_60rdmi), c("name", "pop55", "pct_lt20_trials_within_60rdmi")])

Fields

Every file carries the same core fields, so county, district and state records can be handled alike. Percentages are of residents aged 55 and older; distances are road miles on the highway network from where people live. Full definitions and limits are on Methods & data.

A value of null means not computable for that place (for example, a median road distance where no residents have a road route).

The map's other layers

Since November 2026 the files also carry the data behind every map layer, so a county's rural code, road distances, incidence, death rates and oncologists can be read from the same file. The cut-offs these use (the county median incidence, the top-third death-rate cut, the periods, site labels) are in index.json under thresholds.

License and citation

The data files are free to use, share and build on under Creative Commons Attribution 4.0: use them for anything, including commercial work, and credit the source. The underlying sources (ClinicalTrials.gov, the Census Bureau, NCI, CMS) are public data. The site's code is on GitHub.

If you use the data in a publication, a dashboard or a news story, a note to the issue page is welcome; it helps show where the work is being used.

Good to know

• Freshness. Files are rebuilt with each weekly registry refresh; data_as_of in every file is the registry date the figures come from. Trials change weekly, so cache with that in mind.
• Stability. Field names will not be renamed or removed without a version note here. New fields may be added. November 2026: the map-layer fields above were added (rural codes, road distances from the population center, incidence, death rates, oncologists, senators); nothing was renamed, and all_counties.json grew to about 9 MB, so fetch one county file when you need one place. Later in November 2026, two corrections and one addition: trials_listed_here in district and state files had been the place's census-tract count and is now the number of distinct recruiting trials with a site in the place (district: in any county that overlaps it, so a county shared by two districts counts for both); wmean_trials_within_60rdmi had been null for 48 dense districts and for New Jersey where the true value is 1,000 or more, and is now the number; and county files gained districts, the ids of the congressional districts that overlap the county (the inverse of the district files' counties), also carried as the last column of ask/counties.json. County figures and every other field are unchanged.
• Counties. County shares are computed from census tracts weighted by residents 55+, the same way as districts and states, so the three levels add up.
• Not medical advice. A listed trial may not be enrolling at a given site today, and being listed does not mean a patient is eligible.