The Opioid Crisis (United States, 1995 to 2026)

After FDA approved OxyContin in 1995, prescription and then illicit opioids drove U.S. overdose deaths from 25,785 in 2003 to 107,941 in 2022; courts, regulators and states responded with pleas, bankruptcy and settlements, while the public learned of key internal facts years later.

The record

Approval and early marketing

• In December 1995 FDA approved OxyContin, the first oxycodone product dosed every 12 hours. FDA's own timeline says the agency believed the controlled-release design would carry less abuse potential, partly because MS Contin showed no significant abuse reports since 1987. (FDA timeline of selected activities; opened, in part, page text was truncated)
• The original label said the controlled-release characteristics "were believed to reduce its potential for abuse." GAO reported that FDA later called this wording "unfortunate." (GAO-04-110, December 2003; opened, first 100,000 characters)
• GAO found Purdue had nearly 700 of its own sales representatives by 2000 plus about 300 from Abbott, promoted the drug for noncancer pain, and saw noncancer prescriptions rise from about 670,000 in 1997 to about 6.2 million in 2002. Nearly half of OxyContin prescribers were primary care physicians by 2003. FDA cited Purdue for misleading advertising twice: an untitled letter in 2000 and a warning letter in 2003. (GAO-04-110)
• In July 2001 FDA added a boxed warning that crushing the tablet "leads to rapid release and absorption of a potentially fatal dose of oxycodone," narrowed the indication and obtained a risk management program. (FDA timeline; GAO-04-110)

The 2007 guilty plea

• On May 10, 2007, The Purdue Frederick Company pleaded guilty in federal court in Abingdon, Virginia, to felony misbranding of OxyContin with intent to defraud or mislead. Three executives (Michael Friedman, Howard Udell, Paul D. Goldenheim) pleaded guilty to a misdemeanor misbranding charge. Total payments were $634,515,475. (Department of Defense copy of the Justice Department release; opened)
• The statement of facts says Purdue marketed OxyContin as less addictive, less subject to abuse and less likely to cause withdrawal without supporting research or FDA approval of those claims. It also says a Purdue study showed about 68% of the oxycodone could be extracted from one 10 mg tablet. (same release)

Regulators and enforcement

• DOJ Inspector General, October 1, 2019: DEA responded slowly to the rise of opioid misuse and diversion from the early 2000s. DEA "authorized a 400% increase in oxycodone production between 2002 and 2013" and did not significantly cut the oxycodone quota until 2017, when it fell 25%. Only 8 of about 1,400 registered manufacturers and distributors filed suspicious-order reports. DEA issued 45 immediate suspension orders in 2012 and 43 across 2013 to 2017 combined. DEA agreed with all 9 recommendations. (DOJ OIG announcement; opened. The full report was not opened.)
• FDA actions over time: class-wide extended-release and long-acting labeling changes announced September 10, 2013; boxed warnings on immediate-release opioids in March 2016; the Opioid Analgesic REMS approved in September 2018. (FDA timeline)

Prescribing guidelines

• CDC published its opioid prescribing guideline on March 18, 2016. It stated that nonpharmacologic and nonopioid therapy are "preferred for chronic pain." It cited 259 million opioid prescriptions in 2012 and more than 165,000 deaths from prescription opioid overdose from 1999 to 2014. (CDC MMWR 65(1); opened in part, the recommendations section on dosage was not read)
• CDC published an update on November 4, 2022, covering acute, subacute and chronic pain. It states an aim "to discourage the misapplication of opioid pain medication dosage thresholds as inflexible standards" and says some policies attributed to the 2016 guideline went beyond its recommendations. (CDC MMWR 71(3); opened in part)

Deaths, with data vintage

All figures below come from one source: NCHS Data Brief No. 522, published December 2024, using National Vital Statistics System final mortality files. Rates are age-adjusted per 100,000 using the 2000 standard population. (NCHS Data Brief 522; opened)

| Year | All drug overdose deaths | Rate | Synthetic opioids other than methadone | |---|---|---| | 2003 | 25,785 | not read | not read | | 2021 | 106,699 | 32.4 | 70,601 (21.8) | | 2022 | 107,941 | 32.6 | 73,838 (22.7) | | 2023 | 105,007 | 31.3 | 72,776 (22.2) |

• The brief's series starts in 2003, so a 1999 count is not given here. The brief notes that rates fell 2022 to 2023 for ages 15 to 54 and rose for adults 55 and older. "Synthetic opioids other than methadone" includes fentanyl, fentanyl analogs and tramadol; the brief does not isolate fentanyl alone in the text we read.
• Provisional data, a different vintage: the CDC provisional 12-month-ending count was 65,351 (predicted 66,478) for the period ending May 2026, from the data.cdc.gov dataset xkb8-kh2a, queried 2026-10-25 for the national row. CDC's dashboard says it is based on data available October 4, 2026, and warns that flat or falling reported numbers "could be due to incomplete data, true decreases in the number of deaths, or a combination of the two." (CDC VSRR provisional drug overdose data; data.cdc.gov API; both opened)

Settlements and bankruptcy

(DOJ announcement via oversight.gov; Pennsylvania AG, July 21, 2021; Harrington v. Purdue slip opinion; New York AG, 2025; all opened)

• The DOJ page we opened shows no publication date. Search results dated a secondary report to October 21, 2020 (Maryland Association of Counties). The date is secondary only.
• Harrington: Justice Gorsuch wrote for Thomas, Alito, Barrett and Jackson. Justice Kavanaugh dissented with Roberts, Sotomayor and Kagan, writing that the decision is "devastating for more than 100,000 opioid victims." The majority wrote that the Sacklers "seek to pay less than the code ordinarily requires and receive more than it normally permits." The opinion says the Sacklers withdrew about $11 billion from Purdue between 2008 and 2016, and the earlier proposal was about $4.325 billion over a decade plus an added $1.175 to $1.675 billion offer. (slip opinion)
• Whether the 2025 plan has taken effect, and its effective date, was not confirmed. A search result said it "goes effective," but we did not open that page. Unverified.
• The $26 billion figure is "up to" and depends on how many states and localities join. The Pennsylvania release says states had 30 days and localities up to 150 days to sign on. How much of the money is actually spent on treatment and prevention was not verified; the release says a large majority must go there.

Positions in the record

• Company (Purdue, Dec. 12, 2017): the "delayed absorption" sentence was "added to the label at FDA's suggestion, not Purdue Pharma's"; Purdue "has always sold OxyContin with approval from FDA"; abuse "was not expected" beyond other Schedule II opioids; it describes a May 2001 "10-point plan." (Purdue statement as reproduced by Marketplace; opened. This is a secondary reproduction.) Purdue Frederick, the corporate entity in 2007, admitted the misbranding facts above.
• Regulator (FDA): approved on a belief the design reduced abuse; later added warnings (2001) and class labeling (2013 to 2018). GAO found FDA's ad oversight acted after materials were public and relied on 39 staff positions in fiscal 2002.
• Regulator (DEA): DOJ's inspector general found slow response and quota growth. DEA's own explanation of its quota decisions was not opened.
• Prescriber and public-health: CDC's 2016 guideline urged caution; its 2022 update warns against rigid dose limits. Clinician organizations' statements and patient-group concerns about tapering were not opened. Unverified.

What this case shows

1. Key facts were not visible to prescribers or patients when decisions were made. The 2007 statement of facts shows internal knowledge (reports of abuse by March 2000, a 68% extraction figure) that sat alongside sales claims. Ordinary people could not weigh what they were not told. (DOJ release copy)
2. The public rules were set at moments most people never saw. An FDA label sentence, a DEA annual quota and a prescribing guideline each shaped supply. The inspector general found that quotas rose 400% from 2002 to 2013. No vote or hearing most voters followed set those numbers. (DOJ OIG)
3. The bankruptcy route narrowed who could object. The Supreme Court held that victims' claims against nondebtor family members could not be released without consent, so the people harmed were placed at the center of the legal question. The dissent warned this could leave victims worse off. Both readings are in the record. (Slip opinion)
4. Who paid and who is covered is uneven. The settlements pay governments. Individual victims and families recover through plan trusts or not at all, and the 2020 DOJ resolution contained no criminal release for individuals. (DOJ announcement; NY AG)
5. Harm shifted to a different supply. Deaths rose from 25,785 in 2003 to more than 100,000 a year, with synthetic opioids other than methadone accounting for about 70 to 74 thousand in 2021 to 2023. People managing pain, people with addiction and people who never used prescriptions were affected by different parts of the system. (NCHS)

The strongest objections, and our answers

1. Objection: FDA approved the label, so blame sits with the regulator. Purdue says the key sentence came from FDA. Mitigation: we state that position. The 2007 plea records marketing claims FDA had not approved, and GAO says FDA called the original wording "unfortunate." We do not assign a share of fault.
2. Objection: Prescribers and patients also chose these drugs, and pain was undertreated before. Mitigation: Agreed that the record includes legitimate pain care. CDC's 2022 update says the 2016 guideline was misapplied through rigid dose limits. Harm from under-treatment is not quantified in sources we opened.
3. Objection: Most deaths now involve illicit fentanyl, not prescriptions. Mitigation: Correct. We show synthetic opioids in the NCHS table. The brief does not identify the source of the synthetic opioids, so we do not claim the prescription period caused the later shift; that is an open question.
4. Objection: The $26 billion headline overstates actual money. Mitigation: We write "up to," show the 18-year and 9-year schedules and flag that final totals depend on participation.
5. Objection: Settlements without trial leave facts unproven. Mitigation: The 2007 and 2020 matters are guilty pleas, which are admissions. The distributor and J&J agreements are civil settlements; we do not state those companies' conduct as proven findings.

Open questions: what would change our minds

• Whether the 2025 Purdue plan is in effect and how much has been paid. Not confirmed.
• How settlement money has actually been spent, by state. No spending reports were opened.
• Fentanyl's share of deaths by year. We have only the "synthetic opioids other than methadone" category.
• The 1999 and 2010 death counts were not opened; we start the table at 2003.
• DEA's reasons for quota levels, and the agency's response to the inspector general in full.
• Evidence on pain patients harmed by tapering after 2016. Not opened.
• The text of the criminal plea agreement for 2020 and its date; DOJ page had no date.
• Whether provisional 2026 figures will be revised; recent months are known to be incomplete.

Sources

Evidence status and correction history

Key claims

Correction history

Found an error or a better source? We keep every correction on the record with its date and reason. Nothing here is quietly rewritten.