Health estimates: Drug overdose deaths, county estimate (CDC NCHS).
Display approved under decisions/0020-sponsor-report-counsel-review-approved-2026-11-02.md (with decisions/0011), building on the sponsor’s public-domain determination (decisions/0011). Public domain does not remove attribution, vintage or limitation requirements. Export, API and commercial use are not approved.
- Attribution
- Source: CDC, National Center for Health Statistics, NCHS - Drug Poisoning Mortality by County: United States, 2003-2021 (model-based estimates), data.cdc.gov dataset rpvx-m2md (retrieved 2026-11-03). Public domain U.S. Government work. Use does not imply endorsement by NCHS, CDC, HHS or the U.S. Government.
- Period
- Calendar year 2021 model-based estimate (file covers 2003 to 2021; retrieved 2026-11-03)
- Geography vintage
- NCHS county geography (largely 2005-2007 bridged-race counties, some merged); Connecticut former counties are not matched to planning regions
- Rights basis
- “Display only, noncommercial public HTML, under decisions 0011 and 0020; display rights reviewed by the sponsor under decision 0110. The counsel statement is relayed by the sponsor and no counsel document is in the repository. Export, API and commercial use remain review_required.” Dwight Ringdahl (sponsor): display-rights review recorded in decision 0110, building on the sponsor-reported counsel review of decision 0020; recorded by the engineering delegate
- Reading notes
- Model-based drug overdose deaths per 100,000 residents with the 95 percent credible interval. Estimates, not counts.
- Raw evidence SHA-256 (combined over 3 files)
- e033a6f94246b6e26d072fb3da586c1a8cb42f3f78b27a8b4976af606b88b520 (sha256 of the sorted file checksums joined by newlines)
- Raw evidence SHA-256: NCHS drug poisoning mortality by county (rpvx-m2md)
- 857650f8070a90af3843b75f95e34340d2518f3c3af27d271b86a366fc7b8c96 · retrieved 2026-11-03 · https://data.cdc.gov/api/views/rpvx-m2md/rows.csv?accessType=DOWNLOAD
- Raw evidence SHA-256: data.cdc.gov metadata for rpvx-m2md
- f85fd49f6e850c5b61d5c18a5b6613194590d5782f49e1b3a417e1a3fb0d6f84 · retrieved 2026-11-03 · https://data.cdc.gov/api/views/rpvx-m2md.json
- Raw evidence SHA-256: NCHS drug poisoning technical notes page
- 82999c08bae73d341f57550a3f960dcd809d43034389d258859452d8b5e18694 · retrieved 2026-11-03 · https://www.cdc.gov/nchs/data-visualization/drug-poisoning-mortality/
Limitations
- These are county figures, not figures for any city or place within the county. They are model-based estimates, not counts and not observed rates: NCHS fits a hierarchical Bayesian model with spatial and temporal effects to county death counts, so a county estimate borrows information from neighboring counties and other years. NCHS says model-based estimates may over- or under-estimate the true rate and may not match CDC WONDER, and they must not be used to rank or compare places.
- The range is NCHS’s 95 percent credible interval: given the data and the model, a 95 percent probability that the true rate falls inside it. A wide range means the estimate is uncertain. Ranges that overlap say nothing about which county is higher.
- The rate is deaths per 100,000 residents for the stated year, shown to one decimal (the published value has more places). NCHS’s page describes the county estimates as model-based crude rates, so they are not adjusted for age and counties with older or younger residents differ for that reason. Deaths counted are drug overdose deaths, ICD-10 underlying cause X40-X44, X60-X64, X85 and Y10-Y14: unintentional, suicide, homicide and undetermined intent.
- NCHS cautions that rates for some counties and years may be low because many cases were still pending or the cause was filed as ill-defined (code R99), so a low estimate may understate overdose deaths. Each year’s estimates come from one model run; they are not meant to be read as a trend, and NCHS notes they are not directly comparable with its earlier model-based estimates.
- The county file uses NCHS’s own county geography, largely the 2005-2007 bridged-race counties, with some counties merged into neighbors where boundaries changed, and Connecticut’s eight former counties (FIPS 09001 to 09015). This release shows Connecticut by planning region, so Connecticut county pages carry no figure here; a Connecticut planning region is never given a former county’s number. Counties are matched by FIPS code only; a county in the file but not in this release is not shown, and a release county missing from the file is Not reported.
- A value NCHS did not publish is shown as Not reported, never as zero.
Source documentation · Source terms
This page is the same for every geography that cites this source. Return to the page you came from with your browser’s back button, or start at Follow the sources.