US Health Data APIs¶

A Python toolkit for querying CDC surveillance and public health datasets — no API keys needed for most sources.

The money behind the data¶

Almost none of the data here is generated at the CDC — it is produced by state and county health departments and reported upward, under the cooperative agreements that fund them. Pick a state to trace the flow: federal dollars down, surveillance back up as the datasets this toolkit collects.

See Funding & Governance for the funding mechanisms behind each dataset and the scraped roster of every state and local health department.

Quick Start¶

git clone https://github.com/boris/health
cd health
cp .env.sample .env          # add ANTHROPIC_API_KEY for WONDER LLM queries
uv sync

Data Sources¶

Federal surveillance mostly stops at the state line. Two research catalogs cover what is available below it, and nothing in either is collected yet. State & Local Sources goes deep on the endpoints that have been verified against live data: county and census-tract figures for cancer, COVID, flu, tickborne disease and foodborne outbreaks. State Health Data Portals goes wide, listing the official portal for all 50 states and DC and grouping them by the software behind them. Funding & Governance traces how CDC dollars reach those state and county departments — and how the surveillance they fund becomes the datasets collected here — with an interactive money-flow graph and a scraped roster of every state and local health department.

One-liners¶

# WONDER — natural language → CDC API
uv run python -m wonder query "drug overdose deaths by state 2020-2023" -f csv
uv run python -m wonder query "maternal mortality by race 2018-2023, age-adjusted rate"

# GRASP — FluSurv-NET hospitalizations
uv run python -m grasp flusurv by-season --location network_all -f table
uv run python -m grasp fluview ili data --region nat --epiweeks 202001-202520

# WISQARS — firearm deaths by state
uv run python -m wisqars state --intent FA_Deaths --year 2023 -f table

# NSSP — ED visit % for influenza
uv run python -m nssp hhs influenza --region 4

# NIS — childhood vaccination coverage
uv run python -m nis rates child 2022 -f table

# CDC Open Data — leading causes of death
uv run python -m cdc_open query bi63-dtpu --where "year='2021'" -f csv

# SEER — cancer mortality by site/sex/race
uv run python -m seer mortality --site 55 --sex female -f csv

Setup Notes¶

Most sources need no credentials. WONDER's LLM query builder requires an Anthropic API key (ANTHROPIC_API_KEY in .env). The WONDER XML API itself is unauthenticated.

WONDER enforces a 15-second rate limit between requests — scripts that loop over many queries must sleep between calls. See Multi-Year Queries for a production-ready pattern.