How Our Data Is Measured, Modeled, and Verified

Many metrics on this site — especially mortality (leading causes of death, stroke, Alzheimer's) and chronic-disease prevalence — are reported as age-adjusted rates rather than crude counts.

Age-adjustment applies the 2000 U.S. Standard Population distribution to each group's age-specific rates, then sums them into a single rate per 100,000 people. This removes the effect of one population being older or younger than another, so you can compare Delaware directly against national figures or compare counties without age skewing the result.

Always compare age-adjusted rates to other age-adjusted rates — never to crude rates. CDC WONDER and NVSS (the source for our stroke and Alzheimer's datasets) report rates this way.

The Behavioral Risk Factor Surveillance System (BRFSS) is an annual telephone survey. Delaware-specific estimates for smoking, obesity, physical activity, and preventive care come from a sample of residents, then weighted to represent the state population. Sample sizes are small at the county level, so margins of error widen.

CDC PLACES takes BRFSS responses and uses small-area estimation models to project prevalence down to the zip-code and county level. The 2025 release uses 2023 BRFSS data for 35 measures and 2022 data for 5. These are modeled estimates, not direct counts — they carry confidence intervals and should be read as 'approximately X%'.

When a figure seems precise (e.g., '14.8% uninsured'), remember it is an estimate with a confidence interval. For policy or research conclusions, check the source's published margin of error.

Where a metric has enough historical data points, the dashboard offers a 'projection' toggle. Projections use ordinary least-squares linear regression on the historical series to extrapolate a forward trend, with a shaded confidence band based on the regression's standard error.

Projections assume the recent trend continues unchanged. They do not account for policy changes, shocks (like a pandemic), or seasonal cycles. A flu-season projection, for example, is illustrative only.

We deliberately avoid trend charts for series with too few data points or non-annual cadence, because linear extrapolation on sparse data produces misleading results. For such metrics we use point-in-time comparison bar charts (Delaware vs. national) instead.

Health data is published with a lag. Hospital discharge data typically releases 12–18 months after the close of the reporting period. Mortality data (vital statistics) lags roughly one year. BRFSS publishes state estimates several months after the survey year ends.

Each dataset card shows a 'Last Updated' date (when the source last refreshed the data) and a 'Verified Date' (when our team last confirmed the link and figures). The Freshness Indicator on each card summarizes how current the source is.

Some series include preliminary or estimated recent points (clearly marked with inline source comments in the dashboard) pending official state publication. These are flagged so you know which figures are provisional.

Delaware has three counties: New Castle, Kent, and Sussex. Most datasets report at the state or county level. Zip-code-level data (e.g., PLACES chronic disease, COVID vaccinations) is available for selected indicators but small-population zip codes may be suppressed to protect privacy.

County-level survey estimates (BRFSS, ACS) are often based on 5-year pooled data rather than a single year, because annual county samples are too small. Read county figures as multi-year averages unless the source states otherwise.

When comparing Delaware to other states, use state-level figures. Cross-state comparisons on the dashboard draw from sources that publish a consistent national baseline.

Every dataset links to its original source — typically a Delaware state agency (DHSS, DPH, DMMA, DSAMH, Division of Professional Regulation), the CDC, the U.S. Census Bureau, HRSA, SAMHSA, or the Robert Wood Johnson Foundation's County Health Rankings.

We do not alter source data. We catalog it, add context, and provide tools to explore it. The 'Notes' field on each dataset card records caveats, suppression rules, and known limitations straight from the source.

When a data source changes location or methodology (e.g., COVID hospitalizations migrating to the Open Data Portal), we note the transition so historical series remain interpretable.

This is an unofficial, community-driven catalog. It is not affiliated with any government agency, and figures should be verified against the original source for official use.

Datasets use different definitions, time periods, and geographic units — never combine two datasets in one chart without confirming they measure the same thing the same way.

Some data collection has been suspended by the federal government (e.g., CDC PRAMS). Those datasets are flagged so users know no new data is expected until collection resumes.

Every dataset now has a 'Cite This Dataset' button in its detail view that generates APA 7th, AMA, and Chicago citations from the dataset's metadata (source, title, year, URL).

Always include the original source as the author/publisher — not this site — unless you are citing a visualization or analysis we created. For our charts and dashboards, cite the Delaware Health Data Vault with the access date.