A fall in heroin “initiates” is good, but tells us less than meets the eye
A recent report regarding the newest National Survey on Drug Use and Health finds a decline in “new initiates” of heroin from 2016 to 2017.
This positive public health indicator could signal progress in a crisis that claimed over 70,000 lives through overdose alone in 2017, but there are ways in which the data could be improperly spun to suggest that there is no ongoing harm to those patients who have historically received opioids and have been forced off of them. As reported in OpioidWatch by a very credible reporter, Roger Parloff:
He invited me to comment at the bottom of his article, and I did so briefly, but it’s easier to provide graphic illustrations here, so that’s what I am doing.
In reality, this statement conflates a few separate issues that deserve unpacking, or we could misestimate the addiction and pain care challenges before us today.
One is whether contractions of the prescription opioid market have affected heroin use ( the TL;DR version of my view is: “very likely, yes, but not necessarily in the last year, so lets talk about that more”).
The other is whether widely pushed efforts to taper or discontinue long-term opioids in patients with chronic pain have transpired without those patients themselves moving to the heroin market ( the TL;DR version of my view is “most are probably not going to heroin, but these data don’t tell you much about that. More importantly, if we want to understand what is happening for this vulnerable population of Americans, then simply asking whether they are consuming heroin amounts to asking the wrong question”)
What do the new data say? What’s an “Initiate” anyway?
There is some unfortunate, but longstanding wordplay going on in the use of the word “initiate” here. Semantically, because of our popular conception of drug use, we assume an “initiate” to a drug is someone who will continue to use that drug, much like a novitiate entering a monastery, or an initiate in a social organization.
But the epidemiology of drug use, including countless longitudinal and survey studies tells the opposite tale. Most “new initiates” don’t continue use. In that way the word “new initiate” is a very misleading.
What studies show is that most people who try hard drugs dabble in them, and most give them up. Most people identify more compelling priorities. I show that in this from my own published analysis of a 3500 person cohort of young adults from 4 US communities, followed prospectively over 20 years. They were asked, every 3–5 years, about recent use of “hard” drugs like cocaine, amphetamines and heroin, with elaborate confidentiality protections. Statistically the largest group (85.8%) never used those drugs. But the next largest group (Early Occasional Users, see the graph) are the folks who used them a few days a month between ages 18 and 25 and then let them go. Persisting in what was “initiated” is the least common outcome of all.
Technically, “new initiate” in the NSDUH is this: a count of persons who avow a single use in the year, having not used before, for whatever substance. Certanly we do hope that if first time use goes down, that could lead to a later decline in long-term use or addiction, but it doesn’t always play out that way. More first-time users might “stick to it” in one era, compared to another, for any number of reasons, ranging from product potency to the availability of alternative sources of reward such as community, job, money, etc.
Thus, the new decline in heroin “initiates” is a favorable development, but it doesn’t tell us exactly how many will “stick to it” or not. It is a weak indicator, albeit hopeful.
Finally, to keep this sober, the same federal survey also shows sky-high current (i.e. “past-month”) use of heroin. We have a rather large problem to solve, and it’s not going away anytime soon.
A wrinkle: Cocaine
But what if heroin dropped while other drugs rose? In this regard, we see something of greater interest, a 4 year upward trend in “new initiates” of cocaine!
And lest anyone misconstrue: no, I don’t think that “stopping opioid prescribing caused people to seek cocaine”.
That’s the kind of simplistic thinking we should avoid. Rather, it may well be that there are social and market factors leading record numbers of people to try different drugs, cocaine among them.
Two questions on prescribing reductions: often conflated, to our detriment
The declaration in the Opioid Institute report that the new data “undercuts a common narrative” regarding patients with pain migrating to heroin is narrowly true. It is true insofar as some might propose that prescription termination in 2016–17 substantially and greatly enriched the population of heroin users in 2017–18. But that’s all.
The “undercut” statement, however, does fully conflate two separate questions. Wores, it drives us toward partially incorrect conclusions on both of them.
One question is epidemiology. It’s this: does the overall availability of prescription opioids for illicit use affect the overall level of heroin initiation or heroin addiction?
The second is clinical and patient-oriented. It’s this. Does taking a particular long-term chronic pain patient, or many of them, forcibly off the prescribed medicine induce black-market heroin-seeking by any number of them?
On the first question, my view (and it’s an assessment of sorts, but commonly held, and it’s one I have regularly published): a vast expansion of prescribing built the opioid user“market”. As a result, more people would eventually come to use heroin too. Moreover, it does appear that the contraction of that pill supply likely pushed many illicit users toward heroin.
In treating this issue, it’s important to recognize that a contraction in opioid prescribing did not begin after the 2016 CDC Guideline on the matter, but earlier. It began in 2010–12, although recent data suggest it did accelerate in 2016.
The way we profile the potentially misusable prescription-type opioid market in the USA is to look at two statistics:
(a) total prescribing figures, and
(b) the DEA’s collected drug seizure summary (it combines local, state and federal seizures, and thus offers a kind of profile of “what’s for sale on the illicit drug market”).
Both tell the same story.
I’m showing here the simplest summary of US opioid prescribing, a screen shot from Bohnert et al, 2018. with highlights by me:
The contraction in the pill market seems to have begun in 2010, as reflected in this DEA report. Here is what DEA’s report shows for seizures of illegally marketed hydrocodone, in green squares.
You can see oxycodone below (red circles), and you can also observe a rise in fentanyl (mostly illicit) in the years that follow:
And heroin rose in tandem
The illicit pill market contraction was followed by rising heroin seizures and a few years later, fentanyl seizures. Here’s heroin, in beige triangles.
As you compare graphs, don’t get tricked into thinking that the rise in heroin is somehow more gradual than the fall in hydrocodone or oxycodone. The y-axis for heroin seizures is on a different scale.
As a result, if anyone does want to make a case for or against changing opioid prescribing altering heroin use prevalence, then the timeline for that discussion needs to begin properly in 2010–12, and not by looking at the changes from 2016 to 2017.
In an article (free, online!) I wrote with Dr. Adam Gordon for Addiction, we lean toward the view that contracting pill supply, starting in 2012–12, did shift would-be illicit users to heroin. However, we did so tentatively because we were aware that these relationships can be complicated, and there is a temptation to oversimplify. Thus we avoided drawing a conclusion that sometimes implies that every time a person is pushed off prescribed opioids, a new heroin user is born.
On the more clinical and patient-centered question, we should ask if forcible discontinuation of opioids has shifted any long-term pain patients to illicit heroin. That’s a legit question.
But since the population of long-term pain patients is diverse, including many older adults and fewer who are young, we will learn little about what’s happening to many of the older patients with pain by studying national profiles of illicit drug use “initiation”.
Even if a modest percentage of previous high-dose opioid patients did go start buying heroin, it will likely be a small number. The actions of that small group will not shift the overall epidemiology you see in the National Survey on Drug Use and Health. Most “initiates” for any drug will be young adults, and most will stop, as has always been the case.
As a matter of anecdotal observation, I have seen a greater number of older adults shift to alcohol, cocaine (if they also had used cocaine earlier in life, specifically), or develop social withdrawal, suicidal ideation and health care seeking through Emergency Department. than shift to heroin. I have not seen many older disabled adults adjust to loss of pills by buying heroin, although I have heard that from colleagues.
Anyway, key messages from these new data are rather limited.
What might be a very good decline in first time trying-out of heroin is still a good thing. If most of those first-time users are merely trying out cocaine instead, or committing to more alcohol intake, then maybe it’s not so good.
But for our reporting on such data, I suggest considering “new intiate” data in the context that sky high heroin use continues, and we have a steep rise in “initiates” for cocaine.
None of this speaks to the wisdom or lack of wisdom in what is being done to pain patients who have historically received opioids and are being taken off of them forcibly. Elsewhere, my colleagues and I have written and spoken on the weak scientific data for this practice (while recognizing that some tapers do prove helpful), and the early indicators of traumatic harm. Sadly, few health systems or payers have committed to measure health outcomes for these patients, including the essential matter of whether these patients live or die as a result of changes to care enacted upon them. I have witnessed, described, and heard about extraordinary harms, as have my colleagues.
These eminently measurable health outcomes are what should guide policy toward a vulnerable group of patients. The lack of commitment to measuring these outcomes says something deeply unfortunate about the values governing our collective response at this time.
Views expressed here are my own and do not represent formal positions or views of the United States Department of Veterans Affairs or the State of Alabama.
I am professor of medicine at the University of Alabama at Birmingham School of Medicine, and I focus my clinical and research work on homelessness, primary care for vulnerable populations, addiction and opioid/pain issues.
