Data & Methodology
Data
Sources & export
Every figure in this dashboard traces back to a public federal dataset. Build a custom extract below, or read how each source was used.
Build an export
CSV · County levelWhere the data comes from
Each source below is color-keyed to how it appears in the maps and trend charts.
Observed mortality
CDC WONDER
Wide-ranging ONline Data for Epidemiologic Research is the CDC's public health data portal, where users can query reports and quantitative data for their own applications.
We use WONDER for observed opioid overdose deaths and mortality rates. "Observed" means every death in the set has been investigated and confirmed as an opioid overdose. Overdose death codes are X40–44, X60–64, X85, and Y10–Y14.
Modeled estimates
National Center for Health Statistics
A branch of the CDC that collects and shares data on birth and death records, medical records, and patients and providers.
We draw on the NCHS data visualization gallery, specifically a Bayesian spatiotemporal analysis using multiple cause-of-death files to model opioid-related mortality at the county level.
Provisional counts
National Vital Statistics System
The CDC system tracking births and deaths nationwide. We use its provisional county-level drug overdose death counts.
Provisional means the counts are incomplete, many deaths remain under investigation, so the values shown are underestimates. The trade-off is timeliness: provisional data runs through 2025, while the latest observed counts stop at 2024.
Policy & access
Opioid Environment Policy Scan
An open data warehouse covering the factors that shape opioid use and related health outcomes, with an interactive map, pre-cleaned data, and resources.
We use OEPS data in the trend graphs, particularly travel time to reach different medications for opioid use disorder.
Survey data
PLACES
The CDC's PLACES dataset provides chronic disease and health-related measures for US counties, including results from the Behavioral Risk Factor Surveillance System, a telephone survey of more than 400,000 Americans a year.
We use the BRFSS mental health measures to examine spatial overlap between mental health challenges and opioid addiction.
Census estimates
American Community Survey
An ongoing annual survey from the US Census Bureau covering population, housing, and socioeconomic change.
Our tool uses a selection of 5-year estimates for health-related variables at the county level.
Composite index
Social Vulnerability Index
The SVI is built from 15 US census variables and created by the CDC to identify the communities most vulnerable to natural disasters, based on social determinants of health. Its four themes and contributing variables:
Overall vulnerability
Socioeconomic status
Below 150% poverty
Unemployed
Housing cost burden
No high school diploma
No health insurance
Household characteristics
Aged 65 & older
Aged 17 & younger
Civilian with a disability
Single-parent households
English language proficiency
Racial & ethnic minority status
Hispanic or Latino (of any race)
Black or African American
Asian
American Indian or Alaska Native
Native Hawaiian or Pacific Islander
Two or more races
Other races
Housing type & transportation
Multi-unit structures
Mobile homes
Crowding
No vehicle
Group quarters
Racial and ethnic categories are all "not Hispanic or Latino" except where noted. Source: SVI 2022 Documentation.
How it was built
Methodology
Data were pulled from publicly-available, online sources with the permission of the providers. All mapping data were pre-processed in ArcGIS Pro to ensure data completeness and, where necessary, suppression. To calculate travel time to the nearest OTP, we created a centroid in each county, and utilized ArcGIS Pro's OD Cost Matrix tool to calculate the average time from that centroid to each OTP in each county. OTPs were pulled from SAMHSA's treatment locator in February 2026. Map data were then pulled into R, converted to TopoJSON format, and visualized in Leaflet. Scatterplots, trend lines, and distribution curves were created using D3 in Javascript. Trend data were processed and visualized in Stata.
We are not authorized to share our code, but are happy to provide resources and further documentation upon request, particularly to students who are interested in learning about the creation of tools like ours. Please contact our team if you're interested in learning more about our code and methodology.
From source data to what you see
01
ArcGIS Pro
Pre-processing, suppression, and the OD Cost Matrix drive-time calculation.
02
R
Map data converted to TopoJSON format.
03
Leaflet
Map visualization.
04
D3 in Javascript
Scatterplots, trend lines, and distribution curves.
05
Stata
Method
How the maps are binned
Several of these measures have a very large range, so classification choices materially change what the maps show.
01 · Quintiles
Outliers excluded from bin breaks
Values more than 1.5× the IQR below the first quartile or above the third quartile are excluded when calculating bins. This keeps bins from being skewed by extreme values, which matters most for provisional overdose counts. Outliers are still drawn, folded into the highest or lowest quintile.
02 . Tertiles on both axes
Tertiles on both axes
Each dataset is split into tertiles, then the two bins a county falls into are compared to determine its fill color. Outliers are excluded when making tertile bins here as well.
03 · Diverging
Split at the mean
Data above and below the mean are each divided into tertiles. Unlike the other two schemes, outliers are included when calculating these tertiles.