Between 1999 and 2010 US opioid prescribing roughly quadrupled, propelled by fraudulent marketing — Purdue Pharma pleaded guilty twice to federal charges over OxyContin — alongside the 'pain as the fifth vital sign' campaign, satisfaction-score incentives, and pill mills operating as retail businesses. Prescription opioids drove the first wave of overdose deaths, and the medical profession's participation, sometimes naive and sometimes paid, is the settled and shameful part of this story. Nothing in what follows relativises it: the overprescribing era was real, industry-engineered, and lethal.

The 2016 CDC guideline recommended caution above certain morphine-equivalent doses — advisory thresholds that insurers, pharmacies, regulators and prosecutors promptly hardened into de facto law. Stable long-term pain patients were force-tapered or abruptly cut off; physicians abandoned pain management under legal fear; and cohort studies subsequently associated involuntary tapering with increased overdose, mental health crises and suicide. By 2019 the FDA had issued a formal warning against sudden discontinuation and the guideline's own authors wrote in the New England Journal that their work was being misapplied. The 2022 CDC revision removed the hard thresholds outright — a rare explicit regulatory admission of overcorrection.

The central inconvenient fact of the second phase: US opioid prescribing fell by more than half from its 2011-2012 peak, while overdose deaths climbed to record levels — because illicitly manufactured fentanyl had replaced pills as the driver, implicated in roughly 70% of opioid deaths by the 2020s. Restricting prescriptions was a defensible response to wave one and did reduce new iatrogenic addiction; it did nothing against a synthetic street supply, and for some dependent patients, cutting legal access accelerated the transition to illicit drugs. Supply-side prescription policy and the fentanyl death curve now move almost independently — a fact both political narratives routinely ignore.

The underlying clinical question remains unsettled: long-term opioid therapy for chronic non-cancer pain has weak trial support — the SPACE trial found no advantage over non-opioid regimens at 12 months for back and arthritis pain — yet a minority of patients report sustained meaningful benefit, and forced discharge of legacy patients demonstrably harms them. How to distinguish the patient who benefits from the patient who is merely dependent, how to deprescribe humanely at scale, and what to offer instead in systems that reimburse pills better than physiotherapy or psychology — these are open problems, and pretending either 'never prescribe' or 'prescribe freely' answers them is the pendulum talking.

The pendulum's least-visible casualties are patients with unambiguous indications: cancer pain, sickle-cell crises, palliative and post-surgical care. Studies documented declining opioid access among terminal cancer patients through the crackdown years; sickle-cell patients — disproportionately Black, with a disease whose crises are among the most painful in medicine — report systematic undertreatment and suspicion in emergency departments; and globally, the WHO estimates most of the world's population lacks adequate access to opioid analgesia at all, an inversion of the American problem that international policy built on American fear has made worse.

Both phases carry the same lesson: policy that treats a heterogeneous clinical population as one risk category produces mass harm in whichever direction it points. For acute pain, the evidence supports the smallest effective supply for the shortest time; for chronic pain, non-opioid multimodal care first, with opioids a considered minority option under review rather than either default or taboo; for legacy patients on stable doses, no involuntary tapering — the 2022 CDC language — and slow, consensual reduction where reduction is right. And for policymakers: prescription metrics are no longer a proxy for overdose deaths, and have not been for a decade.