Follow Us:
Opioid 101:
Contact us:
Thank you! We’ve sent you an email to confirm your address.
Have drug overdose deaths peaked?
By ROGER PARLOFF|September 5, 2018
The CDC's Provisional Drug Overdose Death Counts
CDC's National Center for Health Statistics, National Vital Statistics System
Quick Takeaway
  • The CDC's latest provisional statistics suggest that total drug overdose fatalities are starting to drop.
  • The data show the same trends for total opioid-related deaths, heroin deaths, and deaths associated with common prescription opioids.
  • To find another 12-month period in which heroin overdose deaths were as low, you’d have to go back to the one that ended in June 2016.

[See update at bottom.]

With provisos, the CDC’s data over the last four to six months suggest that drug overdose deaths, including opioid-related deaths, are starting to drop.

Just over a year ago, the Centers for Disease Control and Prevention began publishing monthly provisional overdose fatality figures for rolling 12-month periods, which it had been collecting for several years. (The data are not final, and are subject to adjustment as additional data come in—the first proviso.)

For the close to three years for which the CDC has been collecting these data, those rolling 12-month death tolls rose every single month, until the count reached the horrendous figure of 73,157 for the period ending September 30, 2017. (This total death count includes overdose fatalities from all drugs, including both opioids—like fentanyl, heroin, and oxycodone—and nonopioids, like cocaine and methamphetamine.)

Latest CDC Provisional Drug Overdose Death Count (Interactive version available at https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm)

For the period ending October 31, 2017, however, the count registered a very slight drop. Since then, the number has continued to fall gently in each reported period—four consecutive declines. For the latest 12-month period reported, which ended on January 31, 2018, the predicted toll was 71,568, or 2.2 percent below the figure for the period ending September 30, 2017.

Opioid use disorder vastly undercounted: Massachusetts study

Most of that decline appears to reflect a drop in deaths associated with fatal opioid overdoses. That subcategory of drug deaths, which reached a high of 49,552 for the 12-month period ending on September 30, 2017, has also gently declined in every reported period since, dipping to 48,612 for the span ending January 31, 2018—a drop of 1.9 percent from its peak. (Again, until last September, this figure had risen in every single rolling 12-month period for which the CDC has data, going back almost three years, to the period ending January 31, 2015.)

CDC Provisional Overdose Death Counts, broken out by subcategory. (Interactive version available at https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm)

For two key subcategories of opioid-related death, the pattern of small declines actually extends back slightly further still—six consecutive months—with the apparent crest having occurred during the 12 months ending on July 31, 2017. Total deaths in which victims were found to have had heroin in their system amounted to 16,576 during that period. Since then the toll has gradually fallen 5.8 percent, to 15,611, for the period ending January 31, 2018.

To find another 12-month period in which heroin overdose fatalities were as low, you’d have to go back to the one that ended in June 2016.

Another way to look at what’s happening is this: To find another 12-month period in which the total number of overdose fatalities associated with heroin were as low as they were in the latest published period (ending January 31, 2018), you’d have to go back to the one that ended in June 2016.

A similar pattern characterizes the category the CDC calls “natural and semi-synthetic opioids,” which includes many—though not all—of the most commonly prescribed opioid pills, including hydrocodone (like Vicodin), oxycodone (like OxyContin), hydromorphone (like Dilaudid), and oxymorphone (like Opana). While 15,631 fatal overdose victims had such drugs in their system during the year ending July 31, 2017, the number fell 5.0 percent, to 14,848, for the period ending January 31, 2018.

Or, again: To find another 12-month period in which the total number of overdose fatalities associated with these prescription pills was as low as it was in the one most recently recorded, you’d have to go back to the one ending November 30, 2016.

(The CDC’s subcategories of addictive drugs associated with overdose death are not mutually exclusive. Many people who overdose are found to have had multiple drugs in their system. Obviously, when “prescription” drugs are found in a decedent’s blood system, coroners and medical examiners cannot tell where or how the victim obtained them, be it from a pharmacy via a doctor’s prescription; a rifled medicine cabinet; a street dealer; or an illicit cryptomarket on the darknet.)

What about fentanyl—unquestionably the biggest driver of overdose death since 2013? The CDC tracks fentanyl deaths within a broader category called “synthetic opioids.” Though this category encompasses both lawful and unlawful drugs, the CDC recognizes that, since 2013, deaths associated with synthetic opioids have overwhelmingly involved illicitly manufactured and trafficked fentanyl (which, according to the DEA, is mailed or smuggled into the country, mainly from China).

Fentanyl, which is up to 50 times more powerful than heroin, is often laced into other drugs, frequently without the knowledge of the drug user. It is increasingly also being mixed into nonopioid drugs, like cocaine and methamphetamine. Overdose fatalities among coke and meth users have risen more rapidly since 2013 than any other subcategory of drug user tracked by the CDC, except for users of “synthetics” themselves (i.e., fentanyl), probably due in part to being laced with fentanyl.

In the latest 12-month period, ending January 31, 2018, fentanyl-related deaths remained flat—the first time they have not risen in at least three years. (Technically, they actually fell by two deaths, from 29,418 to 29,416, though this margin could easily vanish when the figures are finalized.) Still, no encouraging pattern has yet emerged regarding fentanyl deaths, which have otherwise been dramatically and steadily climbing since 2013.

Similarly, cocaine deaths very slightly declined in the latest reporting period for the first time in almost three years, dropping 0.7 percent, from 14,614 to 14,510. Again, it’s too early to tell what, if anything, that means.

Finally, the CDC category that includes methamphetamine deaths (“psychostimulants with abuse potential”) continued to worsen alarmingly, jumping 2.4 percent in the latest period, from 10,708 to 10,968. That toll has risen 140 percent in the past three years (since the 12-month period ending in January 2015), rising from 4,569 to 10,708.

Now let’s get to the “provisos” I mentioned in the first sentence of the article. First, this article is about what the CDC calls “provisional” overdose death figures. So if the provisional figures are later revised upward—which used to routinely happen—the declines I am highlighting could theoretically vanish by the time the figures are finalized.

However, this past April the CDC began reporting not just a provisional “reported” death figure, but also a higher, provisional “predicted” total. (For the latest period, for instance, the reported number was 69,703 deaths, while the predicted value was 71,568 deaths.) The predicted value attempts to statistically project, based on historical patterns, where the number will stand after likely adjustments. These predicted figures are retroactively fine-tuned each month, as additional data come in. All numbers used in this article are the higher, predicted values. (By my observations since April, in several instances these predicted figures have actually been adjusted downwards after more data came in.)

Given this new methodology, and the fact that certain declines have now been sustained for at least four consecutive months, I believe the data have reached the point where they are appropriate to report.

“I hope that next year is better, but it could easily be just as bad or worse.”—Keith Humphreys

At the same time, even if the data hold up, they could still turn out to be a temporary blip.

“About five years ago,” warns Keith Humphreys in an email, “opioid overdoses appeared to be cresting, and we all breathed a sigh of relief. And the next year they rose again.” Humphreys is an addiction specialist and professor of psychiatry and behavioral sciences at Stanford University.

“I would love to believe we are through the worst of this,” he continues, “but remember that fentanyl is only now starting to become a presence in the Western U.S. heroin market, the federal government response to the epidemic remains anemic and confused, and we still prescribe opioids far more than any nation on earth. I hope that next year is better, but it could easily be just as bad or worse.”

Though I do not want to undermine a sense of urgency about the epidemic, or foster complacency, I believe the data are worth reporting for two reasons:

First, the trend appears to be real and few people know about it—one definition of “news.”

Second, I often read or hear criticisms of the various steps society has taken to curb the opioid epidemic—whether through legislation, practice guidelines, insurance reforms, or whatever. Those criticisms often take the form: “We have tried X, but the overdose death rates continue to skyrocket, so clearly X isn’t working and may even be exacerbating the situation.”

There are multiple potential logical fallacies with that type of argument, but if its factual premise also happens to be wrong—overdoses aren’t, in fact, continuing to skyrocket but, rather, starting to drop—then people need to know that.

“What may be more accurate is that in some counties or states we’ve turned a corner. In other parts, things may still be getting worse.”—Andrew Kolodny

In an interview, Andrew Kolodny, co-director of Opioid Policy Research at Brandeis University’s Heller School, says it’s difficult to say if overdose deaths have crested or are cresting.

“I hope it’s true,” he says. “What may be more accurate is that in some counties or states we’ve turned a corner. In other parts, things may still be getting worse.”

The latest figures do show, in fact, that in certain very hard-hit states, like Massachusetts and Rhode Island, overdose deaths actually declined in the latest 12-month period (by 1.1 percent and 7.1 percent, respectively). At the same time, in other hard-hit states, like New Jersey, West Virginia, Ohio, and Pennsylvania, rates continue to rise at or near double-digit rates (21.1 percent; 11.2 percent; 9.3 percent; and 8.1 percent, respectively).

The broader point, Kolodny continues, is that “it’s very important not to define the opioid epidemic by just deaths. Deaths are the most serious outcome, but there are other problems,” he says, alluding to hospitalizations, foster care, neonatal opioid dependency, crime, and the devastating impacts of addiction on families and communities.

“We’ll know we’ve turned the corner when incidence of addiction is declining”—meaning the number of new cases of addiction arising each year. “I think that’s happening. Fewer Americans will become addicted this year than last year,” he says, alluding to the evidence of a recent Blue Cross Blue Shield report, which found that the rate of diagnosis of opioid use disorder among its members dropped slightly from 2016 to 2017.

Prevalence of the disease, however—the total number of cases, when new cases are added to existing cases—continues to rise, he says.

“For the rest of our lifetimes, this is going to be a major public health problem.”

[After publication, the downward trend continued. Overdose deaths associated with all drugs; all opioids; heroin; and “natural and semi-synthetic opioids” dropped again, according to the most recent CDC provisional data for the 12-month period ending February 28, 2018. Fentanyl-related deaths, however, resumed their upward trajectory.]

If readers have interpretations of what, if anything, the data described above most likely mean, please send to [email protected] Or, as always, you may comment directly below.]

Filed under: Addiction/ Research