Opioid "Epidemic": Some Relevant Time Series

The standard narrative for the opioid epidemic is superficially plausible when expressed in a sloppy narrative form, but it starts to fall apart when you look at the actual data. The overall story that "prescriptions went up, and subsequently overdose deaths went up" is clear enough. But the notion that increasing prescriptions inherently produces more addicts or drug abusers is wrong. Someone might infer from this simple story that a crackdown on prescriptions is warranted: "If prescriptions go back down, overdoses will go back down, too." Plainly this is wrong, too, given our experience in the 2010 - present period. So what's been happening over the past few decades?

Below are prescription opioid lifetime abuse charts from the book Lies, Damned Lies and Drug War Statistics.

The light circles represent lifetime non-therapeutic (recreational) use of prescription drugs. What's weird is that it's flat (maybe even declining) for the 1990s, the period over which doctors' attitudes about prescription painkillers were changing and opioid painkiller prescriptions were increasing. It suddenly spikes in the 1998 to 2002 period, then flattens out again. I'd say the spike is probably more likely a reporting bias or some kind of methodology change. It's likely that the true value was increasing over the period 1990-2002, since this is the first time large quantities of these drugs were widely in circulation. I wish that I could find "past month user" data for this period, comparable to SAMHSA data for the 2002 to present period. I don't know if it just doesn't exist or (implausibly) nobody has bothered to chart this.

See the charts below. These are SAMHSA survey data, 2002 - present, drug misuse and drug abuse disorder. I focus on the triangles, because this is the most inclusive grouping. Different age demographics move differently, but as a whole this is flat or declining for the 2002 - 2014 period.

The Monitoring the Future graphs tell the same story for the 2002 to 2014 period, but fills in details for the earlier period. It looks like there was an increase in past year use from 1990 to 2002, then a leveling off and a decline in very recent years. The discontinuity is due to a change in the survey question, in which they changed the examples of drugs in this category. I would probably "correct" the 1990 to 2002 numbers by drawing a straight line from 1990 to 2002 and saying this was probably the "real" trend. Note that availability seems to be declining. These surveys are aimed specifically at youths (8th, 10th, and 12th-graders), so they aren't directly comparable to the SAMHSA data. Still, roughly speaking the corroborate the story that illicit use of these substances was increasing in the 1990 to early 2000s and flat or declining from the early 2000s to present.

Directly below are opioid-related deaths, 1978-1998. This was under the ICD9 cause of death coding. It is not directly comparable to ICD10, which has different codes and allows for much more detail about the substances on the death certificate. At any rate, the trend is what we'd expect. It starts increasing steadily around 1990 or so.

The deaths for the 1999 - 2016 period are coded under the more detailed ICD10 system. We can see trends by various substances (or categories of substances). Again, there is a dramatic increase in "other opioids" (encompassing most opioid painkiller pills) from 1999 to 2016, a period over which recreational use rates were flat. Heroin overdoses started skyrocketing around 2010, when the "opioid epidemic" started getting widespread attention. I wonder if this is where a crackdown on pills started to take place, and perhaps some recreational Oxycontin users switched to heroin. Or maybe illicit fentanyl made street heroin cheaper (because it's so easy to smuggle/distribute), and it's an independent phenomenon.

The "standard narrative" of the opioid epidemic is that loose prescribing practices turned a bunch of unwitting patients into drug addicts. They either began abusing their own legal prescriptions or buying them on the street to use recreationally, and this led to an increase in overdoses. This story is inconsistent with the SAMHSA and MTF survey data, at least for the 2002 to present period. The volume of opioids prescribed roughly quadrupled in the 1999 to present period, leveling off in recent years (level or declining slightly since ~2010). The standard narrative would predict that this should have increased the number of illicit users. Certainly this can be said for the 1990-2000 period. But apparently you can massively increase the volume of opioids prescribed without creating new addicts. And apparently a crackdown (or perhaps just a leveling-off?) can lead to an even faster increase in drug poisoning deaths.

I get annoyed with sloppy journalism on the opioid subject. It usually reports the correlation between deaths and prescriptions in the 1999 to present period, and sloppily implies a neat causal relationship. Such stories usually feature some poor addict with a terrible drug problem, perhaps telling the reporter that they got hooked on a legal prescription and implying to the reader that this is the social phenomenon driving the increase in deaths.

Sam Quinones' book Dreamland has a long exposition about the "Xalisco boys", a Mexican drug cartel operating in the US and selling heroin. But the timeline is all wrong. Much of his story takes place in the 1990s, and the dramatic increases in heroin deaths didn't start until 2010 or so. His book came out in early 2015, so he would have at the very latest had 2013 mortality data (the CDC data for a given year comes out in December of the following year). Supposing he even looked at the CDC data, he would have seen just a couple of years' worth of the recent heroin epidemic. He never seems to tell his reader that it took over 20 years of relaxed opioid prescription practices to cause the heroin epidemic. Indeed, he gives his readers the perception of a steady rise. (Near the end of the book, he even concedes that a local crack-down on prescription opioids led to an increase in heroin overdoses. Instead of doing the requisite soul-searching and revising his priors, he simply insists that the crack-down was necessary. I can't place my finger on the passage at the moment, but it was so jarring it seared itself into my memory.)

I have a few other very long posts attacking the standard narrative, notably here and here. The broken link in their chain of reasoning is the flat recreational use rates, which according to the standard narrative should be rising.