The Three Waves: Visualizing America's Opioid Epidemic with CDC Data

In 1999, roughly 8,000 Americans died from opioid overdoses. By 2023, that figure was approaching 80,000 – a tenfold increase in a single generation. This post uses CDC Vital Statistics data to trace the three distinct waves of the epidemic: prescription painkillers, heroin, and synthetic fentanyl – each deadlier than the last.

The U.S. opioid epidemic has unfolded in three distinct waves since the late 1990s — each deadlier than the last. This post uses live CDC data to trace that arc: from a flood of prescription painkillers, to a heroin surge, to the synthetic fentanyl era that now kills ~75,000 Americans a year. The drug itself changed, the source changed, and the geography changed. The death toll never stopped climbing.

In 1999, roughly 8,000 Americans died from drug overdoses involving opioids. By 2023, that figure was approaching 80,000 — a tenfold increase in a single generation. Drug overdose has become the leading cause of accidental death in the United States, surpassing car crashes. Yet unlike a natural disaster, this catastrophe unfolded in slow motion, one prescription pad, one syringe, one fentanyl-laced pill at a time.

What makes the opioid crisis distinctive is that it mutated. Three separate epidemics, each triggered by a different drug, have layered on top of one another over 25 years. Understanding the mechanics of that mutation — and seeing it in data — is essential for anyone trying to understand where it goes next.

All charts on this page draw from the CDC's Vital Statistics Rapid Release (VSRR) dataset, which tracks provisional drug overdose death counts monthly across all U.S. states. The numbers use 12-month rolling totals, which smooth out seasonal noise and reporting lags.

The epidemic's shape is most visible in a single chart: overdose deaths broken out by substance, plotted as 12-month rolling totals from 2015 to 2025. Three inflection points define three eras.

The shaded zones mark the dominant wave of each period, though in practice the waves overlap considerably. Wave 3 (fentanyl) didn't replace Wave 2 (heroin) — it subsumed it. Today, most "heroin" deaths involve fentanyl; the distinction between the two has all but collapsed in the drug supply.

The chart also reveals something important: total overdose deaths (the red line) barely paused, even as the underlying substances shifted. Each time one drug supply was disrupted or restricted, another took its place — usually more dangerous than the last.

The first wave was a pharmaceutical phenomenon. Starting in the late 1990s, OxyContin and other extended-release opioids were aggressively marketed to physicians as safe long-term pain treatments. Purdue Pharma's sales force specifically targeted high-volume prescribers in rural areas with high rates of physical labor injuries — Appalachia, the rural South, the Mountain West.

The strategy worked: by 2010, U.S. physicians were prescribing enough opioids to medicate every American adult around the clock for a month. The "natural and semi-synthetic opioids" line in the chart above — the blue line — represents this wave: hydrocodone, oxycodone, codeine.

The second wave was the market's answer to the first. When prescription opioids became harder to obtain and their street price rose, heroin was cheaper, more available, and produced by the same pharmacological mechanism. Mexican cartels recognized the demand and flooded the market.

The heroin epidemic looked different from Wave 1. Where prescription opioids had been predominantly a rural and suburban phenomenon, heroin spread into every demographic and geography. Overdose deaths climbed steeply from 2010 through 2016-2017.

Then the drug supply shifted again.

Synthetic opioids — primarily illicitly manufactured fentanyl — are now the dominant driver of overdose deaths. The following chart isolates the heroin-to-fentanyl transition, which represents the most consequential shift in the drug supply in the epidemic's history.

Seeing the raw composition by year makes the three-wave structure concrete. The bar chart below shows annual death counts for each major substance category, using December 12-month-ending data as the annual snapshot.

A few features stand out:

• Prescription opioids (blue) plateau and very slowly decline after 2016 as prescribing restrictions tightened — but they never return to pre-epidemic levels. Millions remained dependent, now often obtaining diverted pills or switching to street drugs.
• Heroin (green) peaks around 2016-2017 and then falls sharply — not because fewer people were using opioids, but because the drug supply was adulterated with fentanyl. Deaths attributed to "heroin" were already fentanyl deaths; the coding just hadn't caught up.
• Fentanyl (yellow) grows explosively, roughly doubling every 2-3 years through the peak.
• Cocaine (purple) and methamphetamine (pink) deaths rise in parallel — a phenomenon researchers call the "polysubstance" wave. Stimulant deaths often involve fentanyl contamination of the cocaine/meth supply, a grim side effect of fentanyl's ubiquity in street-level drug markets.

Drug overdose deaths are large enough in absolute terms to register as a measurable fraction of all U.S. deaths. The chart below places drug overdose deaths against total U.S. mortality — context that is often missing from epidemic reporting.

The 2020-2021 spike in total deaths is the COVID-19 pandemic. Notably, drug overdose deaths also accelerated sharply during this period — fentanyl deaths increased ~30% during the pandemic year. Lockdowns, disrupted supply chains, and collapsed treatment infrastructure compounded the synthetic opioid crisis simultaneously with COVID-19 mortality.

At peak, roughly 1 in every 30 deaths in the United States was a drug overdose — a share unthinkable in 1999.

The opioid epidemic began in specific geographies — rural Appalachia, manufacturing towns in the Rust Belt — but has since diffused broadly across the country. The chart below ranks all 50 states by most recent 12-month overdose death totals.

Absolute death counts reflect population size; large states like California, Texas, and Florida show high totals. But the epidemic isn't primarily a big-state story. West Virginia, Ohio, Pennsylvania, and Kentucky remain disproportionately affected when adjusting for population — a legacy of the prescription wave that hit those communities first and hardest. The color gradient from purple to orange reflects magnitude within the top-ranked states.

Death counts are a lagging, partial indicator. They tell us how many people died, but not:

• How many people are in active addiction — estimates range from 2-4 million Americans for opioid use disorder alone
• Rates of treatment access — fewer than 25% of people with opioid use disorder receive medication-assisted treatment (buprenorphine, methadone, naltrexone)
• Near-miss overdoses — for every fatal overdose, many more are reversed by naloxone (Narcan) or survive without intervention
• The economic cascade — lost productivity, healthcare costs, criminal justice costs, and family disruption extend well beyond the death toll

The CDC's Drug Overdose surveillance and HRSA's Health Professional Shortage Area data offer complementary views — particularly the geographic clustering of mental health provider deserts, which correlates with overdose hotspots.

All charts use the CDC's VSRR Provisional Drug Overdose Death Counts dataset (dataset ID: xkb8-kh2a), accessed via the CDC's Socrata SODA API. The API is free, requires no authentication, and updates monthly.

The data covers:

• Period: January 2015 – present (with ~6-8 month reporting lag)
• Geography: All 50 states, DC, and national totals
• Granularity: Monthly, with both "12 month-ending" rolling totals and single-month counts
• Indicators: Total deaths, drug overdose deaths, opioids, fentanyl, heroin, natural/semi-synthetic opioids, cocaine, psychostimulants, and more

The 12-month rolling total ("period" = "12 month-ending") is used throughout for stability. Single-month counts have high variance due to reporting delays from medical examiner offices.

Data source: CDC VSRR Provisional Drug Overdose Death Counts. Charts generated from live API data fetched February 2026. CDC WONDER and HRSA data referenced for contextual statistics.