A Date With a Model

Finding Reality in Questions About Public Health

In 2000, a pharmacist named Jaymie Mai started working for the State of Washington’s Department of Labor & Industries, overseeing cases of injured workers receiving care under workers’ compensation claims. Sam Quinones tells her story in Dreamland, a deeply reported book on the opioid crisis in America.

Over the next few years, Mai was surprised to see reports of deaths crossing her desk for cases that didn’t seem to suggest the possibility of it: sprained back resulting in death a few months later, knee injury resulting in death a few months later. Carpal tunnel syndrome leading to death. She began to pull autopsy reports and death certificates, and discovered that people with fairly minor injuries had died of drug overdoses — a few months after being prescribed OxyContin, a powerful opioid. In 2004, Mai began calling state officials who were responsible for other populations, like Medicare patients, and asking them if they were seeing similar overdose deaths among people who shouldn’t have been on particularly serious pain medication. They began trying to answer Mai’s question, and began finding a hidden surge of opioid deaths.

In December of 2005 — note the leisurely speed at which these developments happen, which is important in our current circumstances — a team of officials from the State of Washington, including Mai and a state epidemiologist named Jennifer Sabel, presented the information they had discovered to a roomful of prominent pain specialists called in from around the state. It wasn’t a come-to-Jesus meeting; rather, it was an effort by public health officials to open a discussion, asking doctors if they knew what was happening and could provide information. They described their own efforts to dig into autopsy reports and add up a trend that no one had noticed: 24 opioid overdoses in 1995, 386 in 2004. They had also tracked prescription data, so they saw that numbers of opioid deaths were climbing as the number of opioid prescriptions were climbing. It was a whaddya think, guys? meeting, regulators querying practitioners: We’re seeing this, but are you seeing this? Would it make sense to start working up regulations limiting opioid prescriptions?

The doctors were initially unimpressed, and told the state officials that they were mistaken. After all, they explained, Purdue Pharma — the manufacturer of OxyContin — had assured doctors that they had studies proving that the drug wasn’t addictive, and couldn’t possibly lead patients into dangerous overuse. It was all very careful: trained and credentialed medical professionals dispensing a product developed by pharmaceutical industry scientists. Purdue Pharma soon followed up with correspondence to the state regulators, assuring them that studies proved patients in pain management treatment could safely take nearly 300 mg of OxyContin a day without dangerous consequences. Everyone knew that it was perfectly safe to prescribe large quantities of opioids to large numbers of patients — because my goodness, are you questioning science?

Now comes Alex Berenson, whose book Pandemia I’m reading this week — Merry Christmas! — and who experienced a comparable moment in which he compared the actual numbers to the thing that everyone knew. You’ll find this story in Chapter 10, on pages 97 to 99. Berenson describes the moment in which American policymakers and journalists were in thrall to the pandemic models developed by the Institute for Health Metrics and Evaluation at the University of Washington. The IMHE began releasing the numbers projected by their models at the end of March, 2020, and their projections immediately began to produce panic.

Take New York, where state officials relied on the IHME to tell them what was coming. “On March 30,” Berenson writes, “IHME forecast that the state would have a shortage of almost sixty thousand hospital beds, including ten thousand intensive care beds, within ten days.” Apocalyptic headlines followed.

But Berenson noticed that the IHME models started in the past, and could be tracked through the current day as they sailed off into the mists of projected mass death. He began to simply compare the numbers the IHME was projecting for whatever today was with the numbers that the model had projected for that day. The numbers weren’t even close:

For example, on April 3, 2020, IHME said that on April 3 (that same day) New York would have 61,000 hospitalized Covid patients, including 11,500 in intensive care. In reality, the state had 14,810 coronavirus patients, including 3,731 in intensive care.

In other words, the IHME model was reporting the number of people hospitalized in New York on April 3 at four times the actual number of patients. And remember, IMHE hadn’t made this model years or even months earlier. It had released the model barely a week before.

He checked the actual numbers. That’s it.

We’re living through a crisis of cognition, a conflict in which the kinds of people who are inclined by habit and temperament to check the numbers are crazy conspiracy theorists. What used to be considered the journalist’s mind — skeptical, inquisitive, driven to ask questions and check the answers — has become an outlier mentality, mocked and marginalized. What Berenson has done, in a book that I urge you to read and to read soon, is simple, direct, and necessary: he did the work of a reporter. He read the documents, he checked the numbers, he asked around, and he followed the evidence. Like Jaymie Mai, he noticed, and he asked about it.

We’ve been here before: I assure you that Soviet nuclear reactors cannot explode, you fool, and studies absolutely prove that OxyContin isn’t addictive.

Truth wins. Slowly.

I’m with family for Christmas, gathered with a mix of vaccinated and unvaccinated people who don’t care because we’re family. I wish you peace and strength for the days ahead.