PopHIVE Data Source Documentation

Abcs

CDC monitors invasive bacterial infections that cause bloodstream infections, sepsis, and meningitis in persons living in the community through Active Bacterial Core surveillance (ABCs). ABCs conducts laboratory- and population-based surveillance for invasive pneumococcal disease (IPD). ABCs serotype data are used to measure the impact of vaccine use in the United States on vaccine-type IPD. This table reports IPD case counts in the ABCs catchment area by serotype for years 1998 through 2022. Cases are grouped into the following mutually exclusive age groups: age <2 years old, age 2-4 years old, age 5-17 years old, age 18-49 years old, age 50-64 years old, and age >=65 years old. ABCs methods and surveillance areas reporting IPD cases has changed over time. Given these changes, trends in serotype distribution by year and age group should be interpreted with caution. The all-site summary presented here is calculated based on the 8 sites that consistently report to ABCs and differs from the All-site measure provided by the source. Additional information on ABCs methods and surveillance population is available at https://www.cdc.gov/abcs/methodology/index.html. Analyze and visualize data using the ABCs Bact Facts Interactive Data Dashboard at https://www.cdc.gov/abcs/bact-facts-interactive-dashboard. ABCs IPD Isolates were serotyped by Quellung, PCR, or whole genome sequencing (WGS). Cases without an isolate available or with mixed serotypes reported are listed on the table as MISS. Additionally, non-typeable IPD cases are shown as NT. Zero cell rows were not included in this dataset. Minor changes to previous years serotype data can occur as additional isolates and serotype data become available. Cases were excluded from this dataset if the ABCs site did not perform surveillance in the catchment area for a full calendar year. As a result, cases were excluded from the following sites: TN, 11 counties, Jul-Dec 1999; CO, 5 counties, Jul-Dec 2000; CA, 2 counties (aged <5 years), Oct-Dec 2000.

Sources

• Data Source | Centers for Disease Control and Prevention | API/Data Location
• Data Source | University of Louisville / Pfizer | API/Data Location
• Data Source | Centers for Disease Control and Prevention | API/Data Location
• Data Source | Centers for Disease Control and Prevention | API/Data Location

Variables

data.csv.gz

gas_emm.csv.gz

gas_syndromes.csv.gz

gbs_alph.csv.gz

gbs_serotypes.csv.gz

gbs_syndromes.csv.gz

strep_counts.csv.gz

strep_rates.csv.gz

strep_resistance.csv.gz

uad.csv.gz

ACS Estimates

Extraction and formatting for these data is provided by Metopio. The American Community Survey (ACS) is an ongoing survey conducted by the U.S. Census Bureau that provides detailed demographic, social, economic, and housing data for communities across the United States. The 5-year estimates pool data over a 60-month period, providing greater statistical reliability for small geographic areas (including ZIP Code Tabulation Areas) at the cost of timeliness. Variables cover a broad range of social determinants of health including poverty, housing, education, employment, insurance coverage, and racial/ethnic composition. Data are retrieved via the Census Bureau's public API using the censusapi R package.

Sources

• Data Source | U.S. Census Bureau | API/Data Location
• Data Source | U.S. Census Bureau | API/Data Location

Variables

data_county.csv.gz

data_state.csv.gz

Arbonet

ArboNET is the national arboviral surveillance system managed by CDC together with state and territorial health departments. In addition to human disease, ArboNET maintains data on arboviral infections among presumptive viremic blood donors, veterinary disease cases, mosquitoes, dead birds, and sentinel animals. Human cases are reported by county of residence rather than location of exposure, which is not necessarily the county or state where exposure occurred, and are assigned to the year of illness onset. Several caveats apply: under-reporting is inherent to any surveillance system that depends on clinicians suspecting the disease, ordering the appropriate test, and reporting confirmed results; non-neuroinvasive cases are substantially more under-reported than neuroinvasive cases, so total case counts should not be used to compare disease activity between locations or over time; and there is a lag in reporting to CDC. Values are the counts CDC publishes for public use and are not suppressed or imputed here. Connecticut reported by county through 2022 and by planning region from 2023 onward, so its geography codes change mid-series. Records that CDC assigns to an unallocated placeholder geography (state FIPS plus '999') because the county of residence is unknown are excluded from the county file but retained in the state and national totals. Reporting years differ by disease and the most recent year is preliminary and subject to change. Data that were not collected in a given year are coded as NA, so cells with count 0 indicate that disease was tested for but not reported.

Sources

• Data Source | Centers for Disease Control and Prevention, Division of Vector-Borne Diseases | API/Data Location

Variables

data_county.csv.gz

data_state.csv.gz

Area Health Resource File

The Area Health Resource File (AHRF) is an annual county-level database produced by the Health Resources and Services Administration (HRSA). It aggregates data from over 50 sources including the AMA Physician Masterfile, the American Hospital Association Annual Survey, the Bureau of Health Workforce, CMS Medicare claims, the Census Bureau, and EPA air quality monitoring. The AHRF covers all U.S. counties and territories with measures of health workforce supply, health facility counts, demographics, socioeconomic conditions, health expenditures, and environmental factors. It is widely used in health services research to characterize county-level resource availability and identify shortage areas.

Sources

• Data Source | Health Resources and Services Administration (HRSA) | API/Data Location

Variables

data.csv.gz

Atlas AMR

No standard data files found.

Beam

The BEAM Dashboard is an interactive tool built on the System for Enteric Disease Response, Investigation, and Coordination (SEDRIC) that provides timely data on enteric pathogen trends and serotype details to support prevention of illnesses from food, water, and animal contact. This source uses the state-level 'Report Data' export (dataset jbhn-e8xn) rather than the HHS-region export, so isolate counts can be represented with standard state FIPS codes. Raw records are aggregated across source type (human, food, animal, environment), source site, and serotype/species to produce a monthly isolate count per state and pathogen. National values are computed by PopHIVE/Ingest as the sum across all reporting states. Isolate rates per 100,000 population use a single 2021 Census population denominator for all years. Counts reflect isolates sequenced and reported by participating state and local public health laboratories, which varies by jurisdiction and pathogen, so trends should not be interpreted as complete disease incidence.

Sources

• Data Source | Centers for Disease Control and Prevention | API/Data Location

Variables

data.csv.gz

Bls Laus

LAUS produces monthly and annual employment, unemployment, and labor force estimates for states, counties, metropolitan areas, and cities by place of residence. County-level estimates are model-based, combining Current Population Survey, Current Employment Statistics, and Unemployment Insurance data, and are not seasonally adjusted. Used here as the direct source for county annual-average unemployment rate.

Sources

• Data Source | U.S. Bureau of Labor Statistics | API/Data Location

Variables

data_county.csv.gz

data_state.csv.gz

BRFSS

The Behavioral Risk Factor Surveillance System (BRFSS) is the nation's premier system of health-related telephone surveys that collect state data about U.S. residents regarding their health-related risk behaviors, chronic health conditions, and use of preventive services. Established in 1984 with 15 states, BRFSS now collects data in all 50 states, the District of Columbia, and three U.S. territories, completing more than 400,000 adult interviews each year. BRFSS provides state-specific data on health conditions including obesity, diabetes, depression, and health behaviors such as heavy drinking, physical activity, and tobacco use. Data are available by age, sex, race/ethnicity, and education level. BRFSS is a critical resource for public health surveillance and policy-making at both state and national levels.

Sources

• Data Source | Centers for Disease Control and Prevention | API/Data Location

Variables

data.csv.gz

data_survey.csv.gz

CDC Cfa Rt

Weekly estimates of the effective reproduction number (Rt) for COVID-19, Influenza, and RSV at the US national and state level. Rt represents the average number of secondary infections generated by one infectious person and is a key indicator of epidemic trajectory. Values above 1 indicate growth; below 1 indicate decline. Data include the median Rt estimate, 95% credible intervals, and the probability that Rt exceeds 1.0. Through May 2026, estimates were produced using EpiNow2 with independent state-level models; from June 2026 onward, estimates use hierarchical generalized additive models (HGAM) with spatial partial pooling across states. Coverage begins September 2024.

Sources

• Data Source | CDC Center for Forecasting and Outbreak Analytics (CFA)

Variables

data.csv.gz

CDC Vssr

National-level provisional maternal mortality rates based on a current flow of mortality and natality data in the National Vital Statistics System (NVSS). A maternal death is the death of a woman while pregnant or within 42 days of termination of pregnancy, from any cause related to or aggravated by the pregnancy or its management (ICD-10 codes A34, O00-O95, O98-O99). Rates are reported as 12-month backward rolling periods, overall and by age and race/Hispanic origin. Death counts of 1-9 are suppressed per NCHS confidentiality standards; rates based on fewer than 20 deaths are considered unreliable and not reported. Updated quarterly by NCHS; previously released estimates are revised as additional records are received.

Sources

• Data Source | National Center for Health Statistics (NCHS)

Variables

data.csv.gz

CDC Wonder Natality

Counts of live births from the National Vital Statistics System birth certificate file, as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program and published through CDC WONDER. Coverage is national, state and county level. County data is published by CDC only for counties of 100,000 or more people; within each state every smaller county is pooled into a single "Unidentified Counties" row under the pseudo-FIPS code SS999, which is retained in the county file but is not a joinable FIPS code. CDC WONDER provides no API for sub-national data (its official API is national-only by policy) so these files are produced by driving the live query form with the Selenium scraper in PopHIVE/wonder-scraper-drowning (scraper/natality_scraper.py), one query per year x geography level x demographic breakdown. Every download is validated against the 'Query Parameters:' footer of the export before it is kept. WONDER's non-informative category rows ("Unknown or Not Stated", "Not Available", "Not Reported", "Not Stated") are collapsed into a single "Unknown" level rather than dropped. Limitations: cells representing 1-9 births are suppressed for confidentiality; breakdowns are single-dimension, not cross-tabulated; and the pull requests counts only, so no birth or fertility rate is included.

Sources

• Data Source | Centers for Disease Control and Prevention, National Center for Health Statistics | API/Data Location

Variables

data_county.csv.gz

data_state.csv.gz

data_state_detail.csv.gz

Census

The Population Estimates Program (PEP) produces annual postcensal estimates of the resident population for the nation, states, and counties by age, sex, race, and Hispanic origin, benchmarked to the most recent decennial census and updated using birth, death, and migration data. Used here as the Census Bureau's own source for county population totals and demographic composition, rather than pulling those figures from a downstream redistribution.

Sources

• Data Source | U.S. Census Bureau | API/Data Location
• Data Source | U.S. Census Bureau | API/Data Location
• Data Source | U.S. Census Bureau | API/Data Location
• Data Source | U.S. Census Bureau | API/Data Location

Variables

data_oqm.csv.gz

data_pep.csv.gz

data_sahie.csv.gz

data_saipe.csv.gz

CMS Birth

A list, with geocoded addresses, of all Medicare-registered hospitals that meet CMS's criteria for the 'Birthing Friendly' designation, published on the CMS Provider Data Catalog. The designation identifies hospitals that participate in a statewide or national perinatal quality improvement collaborative and have implemented recommended patient safety practices for maternal health, drawn from CMS's Hospital Inpatient Quality Reporting (IQR) Program 'Maternal Morbidity Structural Measure.' This ingest aggregates the geocoded facility list to a count of designated hospitals per county, state, and nationally as of the most recent CMS update, assigning each hospital to a county via a point-in-polygon match of its latitude/longitude against Census county boundaries.

Sources

• Data Source | Centers for Medicare & Medicaid Services (CMS)

Variables

data_county.csv.gz

data_state.csv.gz

CMS Mmd

The Mapping Medicare Disparities (MMD) by Population Tool is an interactive map that displays chronic disease prevalence, costs, hospitalization, and preventive care utilization data for Medicare Fee-for-Service beneficiaries. Data are available at the national, state, and county levels, stratified by age, race/ethnicity, and sex. Condition prevalence rates are calculated using ICD-10 diagnosis codes in Medicare claims data, following the Chronic Conditions Warehouse (CCW) definitions. The tool covers over 30 chronic conditions including diabetes, hypertension, COPD, heart failure, and mental health conditions, as well as preventive service utilization metrics. Data are updated annually.

Sources

• Data Source | Centers for Medicare and Medicaid Services (CMS)

Variables

data_state_county_age.csv.gz

data_state_county_age_by_race.csv.gz

data_state_county_age_by_sex.csv.gz

County Health Rankings

The County Health Rankings & Roadmaps (CHR&R) program, a collaboration between the Robert Wood Johnson Foundation and the University of Wisconsin Population Health Institute, ranks the health of nearly all counties in the nation. The annual rankings measure vital health factors including high school graduation rates, obesity, smoking, unemployment, access to healthy foods, the quality of air and water, income inequality, and teen births that influence how long and how well we live. Data are released annually and cover all U.S. counties. Raw data are publicly available through the Zenodo repository (https://doi.org/10.5281/zenodo.18157681) and span release years 2010 through the present.

Sources

• Data Source | University of Wisconsin Population Health Institute | API/Data Location

Variables

data_county.csv.gz

data_state.csv.gz

Countybuddy

County Buddy is a companion dataset built to give geographic and demographic context to socioeconomic analyses. It identifies U.S. counties where the incarcerated, university student, active-military, or Native American population is an outlier -- exceeding 2 standard deviations above the national mean share of total population. The original source also names the specific prisons, universities, military bases, or reservations responsible for each outlier; this ingest keeps only the numeric population counts and percentages, not those institution names. Only outlier counties appear as rows; within a row, any of the four categories that did not meet the threshold is zeroed out. Underlying figures are drawn from several federal sources at different vintages: total population and Native American population from the 2020 Census, incarcerated and university-student population from the Census Bureau's 2020 Group Quarters survey, prison facilities from a 2017 Oak Ridge National Laboratory boundary file (capacity >= 200), universities from a 2019/2020 NCES/IPEDS listing (non-junior, non-profit, 1,000+ students), military bases from a 2023 Department of Transportation dataset, and Native American reservations from a 2021 Department of the Interior dataset. Published by researchers at Georgia Tech's College of Computing (Vu, Baniassad & Andris).

Sources

• Data Source | Georgia Institute of Technology, College of Computing

Variables

data.csv.gz

Delphi Doctors Claims

The Delphi outpatient claims signal estimates the percentage of outpatient doctor visits with COVID-19 diagnoses based on claims data from health system partners. CMU Delphi receives de-identified medical insurance claims data covering a significant fraction of United States healthcare visits. The signal tracks outpatient visits with COVID-related ICD-10 diagnosis codes (U071, U072, B9729, J1281, Z03818, B342, J1289). This signal provides near-real-time insight into COVID-19 activity at the community level based on actual healthcare encounters. Data are published as daily values at national, state, and county levels with approximately 3-4 day lag from the date of service. Those daily values are already smoothed over a trailing 7-day window (they show no weekday profile and a lag-1 autocorrelation of 0.998), so PopHIVE keeps the value reported on each week-ending Saturday rather than averaging the week a second time. This signal replaces the retired COVIDcast doctor-visits signal (smoothed_cli), which was smoothed with a Gaussian linear smoother.

Sources

• Data Source | Carnegie Mellon University Delphi Research Group | API/Data Location

Variables

data.csv.gz

Delphi Hospital Claims

The Delphi inpatient claims signals estimate the percentage of new hospital admissions with COVID-19 or influenza diagnoses based on electronic medical records and claims data from health system partners. CMU Delphi receives de-identified hospital admission data covering a significant fraction of United States hospitals. The signals track inpatient admissions with relevant ICD-10 diagnosis codes for COVID-19 and influenza. This signal provides near-real-time insight into severe respiratory illness at the community level based on actual hospitalizations. Data are published as daily values at national, state, and county levels with approximately 3-4 day lag from the date of admission. Those daily values are already smoothed over a trailing 7-day window (they show no weekday profile, and runs of non-zero days bottom out at exactly seven), so PopHIVE keeps the value reported on each week-ending Saturday rather than averaging the week a second time. Geography-weeks with no Saturday observation are therefore absent, which affects about 11% of county-weeks. These signals replace the retired COVIDcast hospital-admissions signals (smoothed_covid19_from_claims and smoothed_flu_from_claims), which were smoothed with a Gaussian linear smoother.

Sources

• Data Source | Carnegie Mellon University Delphi Research Group | API/Data Location

Variables

data.csv.gz

Delphi ILI Fluview

Influenza-Like Illness (ILI) surveillance data from the CDC's ILINet network, accessed via the CMU Delphi Epidata API. ILINet is a collaborative effort between the CDC, state and local health departments, and approximately 3,000 outpatient healthcare providers across all 50 states, Puerto Rico, the District of Columbia, and the U.S. Virgin Islands. Providers report the total number of patients seen and the number presenting with ILI (defined as fever >=100F plus cough and/or sore throat without a known cause other than influenza). Data are available from 1997 week 40 onward at national, HHS regional, and state levels. Both weighted (population-adjusted) and unweighted percentages are provided. This long-running surveillance system is the primary source for tracking seasonal influenza activity in the United States.

Sources

• Data Source | Carnegie Mellon University Delphi Research Group | API/Data Location

Variables

data.csv.gz

Delphi NHSN

Weekly hospital respiratory data reported to CDC's National Healthcare Safety Network (NHSN), accessed via the CMU Delphi Epidata API. NHSN is the nation's most widely used healthcare-associated infection tracking system, collecting data from hospitals across the United States. The data includes COVID-19, influenza, and RSV associated hospital admissions aggregated to national and state levels. Data collection became mandatory for hospitals in November 2024; prior to this, reporting was voluntary with variable participation rates.

Sources

• Data Source | Carnegie Mellon University Delphi Research Group | API/Data Location

Variables

data.csv.gz

Epic Chronic

Epic Cosmos is a collaborative research platform containing de-identified patient data from over 300 million patients across more than 1,600 hospitals and health systems using Epic electronic health record systems. Data is accessed via SlicerDicer, a self-service analytics tool. The dataset includes emergency department visits, diagnoses, immunizations, laboratory results, and other clinical data. Due to privacy protections, counts fewer than 10 are suppressed and imputed. Coverage extends across all U.S. states and territories. Note that county-level and city-level stratifications could differ markedly in total sample size due to high levels of missingness of county data in some states.

Sources

• Data Source | Epic Systems

Variables

county_no_time.csv.gz

county_year.csv.gz

state_no_time.csv.gz

state_year.csv.gz

Epic Concussions

Epic Cosmos is a collaborative research platform containing de-identified patient data from over 300 million patients across more than 1,600 hospitals and health systems using Epic electronic health record systems. Data is accessed via SlicerDicer, a self-service analytics tool. The dataset includes emergency department visits, diagnoses, immunizations, laboratory results, and other clinical data. Due to privacy protections, counts fewer than 10 are suppressed and imputed. Coverage extends across all U.S. states and territories. Note that county-level and city-level stratifications could differ markedly in total sample size due to high levels of missingness of county data in some states.

Sources

• Data Source | Epic Systems

Variables

data.csv.gz

Epic Diarrhea

Epic Cosmos is a collaborative research platform containing de-identified patient data from over 300 million patients across more than 1,600 hospitals and health systems using Epic electronic health record systems. Data is accessed via SlicerDicer, a self-service analytics tool. The dataset includes emergency department visits, diagnoses, immunizations, laboratory results, and other clinical data. Due to privacy protections, counts fewer than 10 are suppressed and imputed. Coverage extends across all U.S. states and territories. Note that county-level and city-level stratifications could differ markedly in total sample size due to high levels of missingness of county data in some states.

Sources

• Data Source | Epic Systems

Variables

data_weekly.csv.gz

weekly_tests.csv.gz