Beyond Data — Investigative Data Journalist

LIVE DASHBOARD

The uncertainty in overdose data

Every month, the CDC publishes two numbers for drug-overdose deaths — the count that's actually been reported, and a model-adjusted estimate that corrects for how long death certificates take to be written. The gap between them is how much of the most recent toll we haven't counted yet.

Sergio Fetter Refreshes monthly from CDC NCHS

What gets reported, vs what's modelled

The solid line is the deaths that have made it into the system — a rolling 12-month count. The dashed line is the CDC's model-adjusted estimate, which factors in known reporting delays. The band between them is the current best guess of the gap: deaths that happened but whose certificates haven't been processed yet.

Line chart of U.S. drug-overdose deaths, reported vs model-adjusted, showing the gap widen for recent months

The gap is almost invisible at the left of the chart — that's 2015–2023, where reporting is fully mature. It widens toward the right. That widening is not an increase in deaths; it's a measure of what the latest reported number is leaving out.

Which states are still catching up

The lag isn't uniform. Some states file death certificates faster than others, which means their reported counts are closer to reality sooner. Below are the fifteen states whose reported count for the latest 12-month period is furthest behind what the model expects.

Horizontal bar chart of the fifteen states with the largest gap between reported and model-adjusted overdose deaths

A big bar doesn't mean a state has worse overdose outcomes — it means its paperwork is slower. For a journalist comparing states using the headline reported number, this gap is the thing that quietly biases the comparison.

What drugs are driving it

The reported totals, broken out by drug class. This chart uses only the reported counts (no uncertainty band), because four series with overlapping bands becomes a tangle.

Line chart of U.S. overdose deaths by drug class over time: synthetic opioids, psychostimulants, cocaine, heroin

Why this page exists

A single reported number carries no sense of how much it's actually measuring. For fast-moving phenomena with slow reporting pipelines — overdoses, pandemics, elections being audited in real time — the headline count and the real count can be months apart. Showing the model-adjusted estimate alongside the reported one is the least a chart can do to flag that.

Data: CDC NCHS, VSRR Provisional Drug Overdose Death Counts.