“And Pharaoh’s heart grew hard, and he did not heed them, as the Lord had said.”

— Exodus 7:13, NKJV

It is easy, in ethical thinking, to ask catechism questions: questions whose answers are already waiting for us.

What in individuals leads to addiction? Unprocessed childhood trauma.

Why did the United States get stuck in Iraq? George W. Bush and his advisers were arrogant idiots.

Why did Boeing ship the MCAS system without proper warning, training, or even clear advice to airlines? They were greedy.

All of these pseudoexplanations flatter us. They are as useful for guiding moral behavior as the Goofus and Gallant cartoons in Highlights magazine for children. But if we want to become better morally, we need more. We need to ask better questions.

A good way to start is by examining cases where we, as a society, assume we know the answers already. Cases like Iraq. Cases like Boeing. And cases like the Sackler family, Purdue Pharma, and OxyContin.

I am choosing OxyContin for the beginning of this moral philosophy essay series because the case is well documented, morally serious, and full of decision points that the public story usually flattens into indictment. The indictments matter. The prosecutions and settlements matter. But “they were greedy” does not tell us enough. It does not tell us how the people inside Purdue interpreted warnings. It does not tell us how uncertainty became certainty. It does not tell us how success hardened judgment. It does not tell us how to avoid becoming the kind of person who can receive evidence of harm and whose first thought is “how do I profit from this.”

To bring out those decision points, we need to tell the story looking forward, not reconstruct it backward.

Not: “Here is the opioid crisis; who caused it?”

But: “Here is a plausible drug, a plausible regulatory judgment, a plausible business opportunity, a plausible sales campaign, a plausible response to early warnings. At what point did plausibility become evasion?”

That is where the moral story begins.

Thanks for reading A Flood of Ideas! Subscribe for free to receive new posts and support my work.

Decision Point 1: Look Gift Horses in the Mouth — The FDA’s Gift to Purdue

In the mid-1990s, the Purdue Frederick Company was a privately-held, midsized pharmaceutical company. Their primary business was selling opioid medications. They were coming off a successful 1980s selling MS Contin, an extended-release formulation of morphine sulfate. The FDAs approval of the drug in 1987 was a critical precedent: the Food and Drug Administration (FDA) had permitted the marketing of an extended-release drug based on the safety profile of an immediate-release formulation, provided the total daily dose did not exceed established levels. But that patented formulation was expiring, competitors were already preparing generics to compete with it, and the company needed a new drug.

It is worth pausing here to talk about a clinical problem opioids have as pain relievers. They are incredibly potent and effective; after centuries, they remain the gold standard for severe post-surgical pain, cancer pain, and palliative care. But opioids have a timing problem as well as a potency problem. A short-acting painkiller does not simply “work” and then “stop working.” It moves the patient through a cycle: relief after the dose, rising drug levels, sedation or euphoria in some patients, then falling levels, returning pain, anxiety about the next dose, and sometimes withdrawal-like distress before the next scheduled tablet. This matters clinically because a patient whose pain control rises and falls sharply may spend much of the day either overmedicated or undertreated. It also matters psychologically and pharmacologically because rapid onset and noticeable peaks are part of what make a drug reinforcing. The body learns the relationship between dose and relief. In the extreme case, this is why smoking, snorting, and injecting drugs can be so powerfully habit-forming: the route of administration creates a fast, legible reward.

So the controlled-release promise was not absurd. A pain medication that entered the bloodstream more slowly, maintained steadier plasma levels, and avoided dramatic peaks and troughs might plausibly give patients better pain control while reducing the “rush” associated with abuse. MS Contin did this, effectively, for cancer-related pain. Purdue thought their new product, Oxycontin, could do the same. And critically, if a drug could avoid abuse potential, and was safer to take than short-acting opioids, then it could be sold to the larger market of non-cancer chronic pain sufferers.

Purdue chose oxycodone, the opioid in Oxycontin, strategically: it was a semi-synthetic opioid, with a short half life, high oral bioavailability, and strong lipophilicity (see footnote for pharma nerd facts).1 Unlike morphine sulfate, it was already approved by the FDA for pharmacy dispensing in products like percocet (5 mg oxycodone with aspirin). Purdue developed a proprietary delivery system, known as Acuhyst, which allowed for the controlled release of oxycodone over a twelve-hour period. This technological shift enabled the company to pack significantly higher amounts of the active ingredient into a single tablet. While traditional oxycodone products contained only 5 mg, the new formulation, OxyContin, would eventually be marketed in strengths ranging from 10 mg to as high as 160 mg.

Purdue submitted their evidence to the FDA in 1995, with data from six clinical trials - significantly more than the standard two for such products (new formulations of existing approved medications). On December 12, 1995, the FDA approved OxyContin in all strengths for the treatment of moderate to severe pain where use of an opioid analgesic was appropriate for more than a few days.

This approval gave Purdue an unexpected, unasked for gift. The FDA suggested the claim for the package insert “delayed absorption as provided by OxyContin tablets, is believed to reduce the abuse liability of a drug.” This had never been proven by clinical trials, nor even investigated. If anything, it was a statement of a research hypothesis. But it became a belief shared between Purdue and the FDA. not just permission to sell oxycodone in a new formulation, but a medically respectable hypothesis that this formulation might be safer from an addiction standpoint than ordinary immediate-release opioids. The crucial word is might.

Decision Point: Do you treat the hypothesis as something to investigate, or do you treat it as permission?

The responsible thing to do, lacking evidence - and Purdue knew there was no existing clinical evidence for this belief - is to treat it as an opportunity for post-market surveillance, one of the indicators to track along with thefts, diversions, product failures, and unexpected adverse reactions. Or, more ambitiously, a reason for an additional clinical trial to demonstrate the hypothesis.

Purdue leadership chose to treat it as permission. Once “might reduce abuse liability” had been converted into “is believed to reduce abuse liability,” Purdue could treat caution as already satisfied.

That meant the rollout did not need to be narrow, specialist-led, and watchful. It could be national, ambitious, and directed toward the much larger population of non-cancer chronic pain patients.

Independently of Purdue and the FDA, a transformation had been taking place in American medicine. In the late 1990s, the American Pain Society introduced the concept of “Pain as the Fifth Vital Sign,” arguing that pain should be measured and monitored as systematically as blood pressure or heart rate. In 2001, the Joint Commission on the Accreditation of Healthcare Organizations (JCAHO) implemented new standards for the assessment and management of pain in all hospitalized patients. While the Joint Commission later argued that their standards did not explicitly mandate the use of opioids, the institutional pressure to demonstrate “compliance” with pain management metrics led many healthcare providers to rely more heavily on pharmacologic interventions.

The standards did not order physicians to prescribe opioids. But they changed the institutional weather. Pain assessment became something hospitals had to document, administrators had to monitor, and clinicians could be criticized for neglecting. In that environment, undertreated pain became newly visible as a professional failure, while the long-term risks of aggressive opioid prescribing were still easier to discount.

This was the first hinge. Purdue had received a claim that was commercially valuable precisely because it was medically plausible and regulatorily blessed. The company could have treated that claim as fragile. Instead it treated it as load-bearing.

Decision Point 2: How to Process Counterevidence

One of the most critical periods in the history of OxyContin occurred between 1999 and 2001, when Purdue’s leadership received increasingly clear signals regarding the drug’s potential for physical dependency and withdrawal. But increasingly clear signals were received through a filter of growing revenues and the not proved but FDA blessed addiction-avoiding properties of the extended release mechanism.

In 1999, an analysis from a Purdue-affiliated company in the United Kingdom identified eight patients in an osteoarthritis study whose symptoms were likely related to opioid withdrawal. This analysis explicitly stated that some patients had become physically dependent on OxyContin tablets. In May 2000, Purdue executives learned that their Medical Services department had received multiple calls from patients and doctors reporting that patients were unable to discontinue even low-dose (10 mg) OxyContin without experiencing significant withdrawal symptoms.

In February 2001, a formal internal review of withdrawal data confirmed these concerns, finding multiple instances where adverse experiences were directly linked to withdrawal. Despite this clear feedback, Purdue’s supervisors and employees made a conscious decision on March 28, 2001, not to write up or disseminate these findings. The rationale was explicitly to avoid “adding to the current negative press” surrounding the drug. This decision represents a profound ethical breach: the choice to prioritize the drug’s commercial reputation over the obligation to inform the medical community of its risks.

Furthermore, Purdue’s sales representatives were actively promoting OxyContin using misinformation about its physical properties. They told healthcare providers that it was difficult to extract the active ingredient from the tablets for intravenous abuse, even though Purdue’s own internal studies showed that 68% of the oxycodone could be extracted from a crushed 10 mg tablet simply by stirring it in water. The company also leveraged the medical community’s misconceptions. Sales supervisors were aware that many physicians incorrectly believed oxycodone was weaker than morphine. Rather than correcting this dangerous misunderstanding, Purdue instructed its staff to protect the drug’s “unique position” in the market by not making doctors realize it was “stronger than or equal to morphine.”

Journalists and opinion writers and prosecutors like to follow prosecutorial rhetoric: present the decision point to the reader or jury, ask “how would you have acted” either directly or by implication, then present the “look what the terrible person did.” And that’s easy to do in corporate wrongdoing cases like this one. Especially when the evidence of harm starts to show up.

Decision Point 2: When evidence against your preferred thesis shows up, do you check, or do you excuse?

But prosecutors are trying to convince a jury, and journalists and opinion writers are trying to tell flattering stories to readers. All of them ignore how consequential decisions are never one offs, isolated Prisoners Dilemmas or Stag Hunts presented once to an unexposed audience. In the real world, every major decision comes weighted by the decisions behind it.

If Decision Point 1 was a balanced scale between options, Decision Point 2 came with the weight created by their choice at Decision Point 1.

By the time reports of addiction and abuse became hard to ignore, Purdue was no longer evaluating them from the position of a company with a tentative hypothesis. It was evaluating them from inside a success machine. The FDA language had not merely permitted sales; it had structured interpretation. If delayed absorption reduced abuse liability, then addiction appearing around OxyContin was less likely to be treated as product evidence and more likely to be treated as patient evidence. The drug was not producing addicts; addicts were finding the drug. The harm was not a signal from the market; it was contamination of the market by abusers, diverters, and bad doctors.

This is why the missing trial matters morally even if one cannot know in advance what it would have shown.

If Purdue had tested the abuse-liability claim and found it unsupported, the company would have faced the truth before the revenue machine had fully formed. It might still have sold OxyContin, and perhaps sold it profitably, but it could not honestly have built the same broad campaign around the same safety story. The trial would have forced the company to learn before scaling.

If Purdue had tested the claim and found it supported, the result would still have mattered. Evidence would have entered the company’s concept of their product at the beginning. Purdue could have said, truthfully, “We believe this because we tested it.” But that also would have created an evidentiary standard for later decisions. If abuse reports accumulated, the question would not be whether to defend a blessed premise, but whether new evidence had changed the balance of risk. A company that begins by testing its favorite claim is better positioned to keep testing it.

Decision Point 3: Defining the Problem as “Their Problem”

The third decision point is the point of no return, morally. After this point, every decision is optimizing wrongdoing, because wrongdoing is the path that has been chosen.

Post-market surveillance of OxyContin showed Purdue that tbe extended release formulation was not only prone to abuse - crushing the tablets and mixing them with tap water could recover 70% of the oxycodone at once - but that it was also failing in patients taking it as prescribed. Rather than twelve hours of relief, patients’ oxycodone levels were dropping at eight hours leaving them with four hours of pain and, often, withdrawal symptoms. When doctors told Purdue reps about these issues, Purdue deployed the concept of “pseudo-addiction”: this only looks like drug-seeking behavior, it really means you have not prescribed a high enough dose of OxyContin to control the patient’s pain effectively. And, Purdue believed, extended release OxyContin was a protection against addiction formation.

The pseudo-addiction point is not, though it is often portrayed that way, cynical drug pushing. The correct solution, pharmacologically, would be to wean the patient off OxyContin and perhaps start them on a different analgesic, or change the dosing schedule to eight hours rather than twelve. But to do the latter would have broken the unique, patented marketing protection Purdue had: an eight-hour extended release formula was not clinically distinct from existing generic products. And they believed their product worked.

The wrong lesson to draw is that pharmaceutical companies should never market to physicians. That is too blunt. Physicians are busy, medicine changes, and within limits, industry education can be genuinely useful. The morally distinguishing fact about Purdue was not that it sent representatives to doctors. Every drug company does that. The distinguishing fact was that Purdue sent them with a safety story whose most commercially valuable component had not been clinically proven: that delayed absorption reduced abuse liability. They sent them with a readymade explanation for patients’ complaints of breakthrough pain after 8 hours rather than 12: they need higher doses. Purdue’s sales force was not merely introducing a drug. It was distributing certainty. And certainty sells.

Decision Point 3: When you accept there is a problem, do you own it or do you label it someone else's problem?

The third decision point was not whether Purdue would admit that OxyContin was being abused. By the early 2000s, abuse was too visible to deny. The decision was how the company would classify that abuse. Was it evidence about OxyContin as Purdue had designed, dosed, promoted, and defended it? Or was it evidence about other people: addicts, criminals, diverters, pill mills, irresponsible patients, and bad doctors? Purdue chose the second frame. This was not wholly false. There were addicts. There were criminals. There were diverters. There were pill mills. But the presence of external wrongdoing did not answer the central question. It only helped Purdue avoid asking it.

The central question was whether the legitimate market and the illegitimate market were really separable. Purdue’s public story required a clean line: proper patients on one side, abusers on the other; medical use on one side, criminal diversion on the other. But OxyContin’s actual history blurred that line. High-dose tablets prescribed in ordinary medical channels could become objects of dependence, diversion, resale, crushing, snorting, and overdose. The same commercial features that made OxyContin attractive as a pain product, its potency, high oral bioavailability, large dose strengths, and broad distribution, also made it attractive as an object of misuse.

Once Purdue defined the problem as “not their problem,” later conduct followed. The company could intensify sales while deploring abuse. It could blame diverters while profiting from high-volume prescribing. It could speak the language of responsibility while protecting the premise that the product, properly understood, was not the source of the crisis. This is where the escalation hardens character. The company no longer merely had a profitable belief. It had a moral vocabulary that protected the belief.

After this point, even legal sanctions for false marketing could not change Purdue’s trajectory. The company could be punished without being converted. It could update labels without updating itself. It could acknowledge abuse while preserving the deeper innocence claim: OxyContin was medicine; the crisis came from other people.

The rest of the story is now familiar: lawsuits, settlements, bankruptcy, Sackler family withdrawals, congressional anger, documentaries, podcasts, and the national search for villains adequate to the scale of the dead. But the moral lesson is mostly earlier than the spectacle. By the time a case becomes famous, the decisive choices have usually disappeared into the past.

In my earlier essay on Carly Gregg, I wrote about the literal smoking gun everyone saw and almost no one treated as decision-relevant. Purdue is the institutional version of that pattern. The warnings were there. They were not hidden in a locked room. They were received, categorized, explained, minimized, and routed around.

So, the purpose of this series is not to produce better scolding. It is to recover the moments before scolding became easy. The useful question is not “How could they?” but “Where did the next wrong step become the easiest step?” That is where moral understanding begins: not at the final catastrophe, but at the earlier hinge, where someone could still have chosen to know.

1

(Infographic courtesy of ChatGPT Images 2.0)