All data comes from the CDC's Wonder database. There are several things going on here in the most lethal drug categories.
This is the most lethal category, and it’s seen a dramatic increase in recent years, from ~1 death per 100k population to ~4 deaths per 100k. This category represents most of the pill-form opiates, hydrocodone and oxycodone and some others. When you hear about the increase in opioid deaths, you are mostly hearing about this category. Notice, however, that it’s flattened out since 2011. It dips down in 2012 and 2013, then in 2014 matches its 2011 level. A similar rise, then dip, happens in the Benzodiazepines trendline. I’ll explain why below in the Benzodiazepines section (and I explain more thoroughly in the link above).
Cocaine was the most lethal category of drugs until about 2006, when cocaine deaths dropped dramatically as “other opioid” deaths continued rising. The drop in deaths is matched by a drop in measured usage rates. (See here, page 10 figure 11 for a time series of "past month" usage rates.) This should be part of the “rising opioid epidemic” narrative. There’s no apparent explanation for the rise in cocaine deaths between 1999 and 2006, because usage rates were flat. One could be forgiven for thinking that the rise between 1999 and 2006 is spurious, related more to reporting and cause-of-death attribution trends than actual overdoses. At any rate, it should be reported that total drug overdose deaths supposedly increased from 1999 to 2014 despite the most lethal category of drugs staying flat. Furthermore, the number of drug overdoses increased from 2006 to 2014 despite the fact that 2006’s most lethal drug category plummeting. I think this is worthy of some commentary.
The benzodiazepine trendline matches the other opioid trendline fairly closely. (See previous post; correlations are very high, and by-state correlations are also very close.) These drugs are sedatives that are often prescribed for anxiety. In a sense, benzodiazepines are just “along for the ride” with the opioid poisoning deaths. If you look at individual death records for 2014, 52% of benzodiazepine deaths also involved the “other opioids” and 76% involved “other opioids”, methadone, *or* “other synthetic narcotics.” Overwhelmingly these are not straight-up overdoses but are actually drug interactions. If you look at it the other way, ~30% of opioid poisoning deaths also involve benzodiazepines. This warning needs to be made more clearly to the public: IF you’re going to abuse either of these substances anyway, don’t mix them. We’d spare ourselves some 5,000 deaths a year if we got people to follow this advice (again, based on the 2014 numbers I have). (For a thorough treatment of multi-drug poisonings, see my previous post here.)
Heroin deaths were on a slow increase from 1999 to 2010, and they’ve increased dramatically (a factor of ~3.5) since 2010. It’s now the second most lethal drug category, after other opioids. It’s not a simple matter of rates of heroin use increasing. If you divide the number of deaths by the number of users (admittedly this later figure is an approximation based on survey data), it looks like the habit of heroin use has become more lethal. I suspect that this is because what’s sold on the street as “heroin” is actually heroin mixed with fentanyl and other ultra-powerful opioids. I quite often see news reports of a rash of “heroin” overdoses hitting some city, often one particular neighborhood of one city. It’s usually a single bad batch of heroin that’s been spiked with something a hundred or a thousand times stronger. (See carfentanil, aka elephant tranquilizer.) This is definitely a problem created by prohibition, because in a legal market nobody would buy this stuff. The preference by dealers for super-powerful opioids is driven by the fact that these substances are easy to conceal. A kilogram brick of heroin can be replaced by a tiny vial of carfentanil. But then, these tiny volumes of carfentanil need to be mixed into a larger volume of heroin (or perhaps baby formula or some other inert base material). Obviously if the mixing is imperfect or if the mixer is an amateur who doesn’t know what he’s got, the results can be deadly for whoever consumes the final product.
The dominant narrative behind the heroin overdoses is “Freely available prescription opioids have stoked a pent-up or latent demand for other opioids. People who once had a legal prescription are turning to heroin, because it’s cheaper and easier to acquire.” I seriously doubt this narrative has much merit. I think there are two different things happening here. The prescription opioid deaths represent a tiny risk multiplied across a very large population (~100 million prescription opioid users with ~200 million legal prescriptions in 2014); the heroin deaths represent a significant risk (1.5% to 3% mortality per user per year!) across a small population of users (just shy of half a million in 2014). I think it’s wrong to pile these into the same narrative, as some news sources do. You’ll often see a story touting the total number of “opioid overdose deaths” in a year, or series of years. These stories are (in my view, incorrectly) adding the heroin deaths to the other opioid, methadone, and “other synthetic narcotics” deaths. By adding them together, the authors of these pieces are assuming their narrative is the correct one, when maybe it isn’t. When you hear the claim that 2014 was a record year for “opioid deaths”, they are improperly adding heroin deaths to these other three categories without explaining what they are doing. I think these authors need to be more explicit about their assumptions, and do a better job of explaining to their readers what’s going on.
Along with “other opioids” and “other synthetic narcotics”, methadone is one of the three categories that are usually tabulated together as prescription painkillers. I believe the CDC uses this definition (citation needed). Take a look; methadone follows its own trend. It rises from 1999 to 2006, then starts to fall. The thing to keep in mind about methadone is that it has a very long half-life, so it stays in the body for a very long time after the user has taken a dose. It can thus build up to toxic levels for a naïve user who hasn’t had time to develop a tolerance, or hasn’t figured out how to titrate his own dose. Most methadone overdoses are from users who have just started methadone maintenance for heroin or have recently acquired a prescription for methadone as a painkiller (citation needed here, but I’m speaking from memory based on the books Buzzed and Karch’s Pathology of Drug Abuse.)
See Wikipedia’s description of what happened up to 2006:
In recent years, methadone has gained popularity among physicians for the treatment of other medical problems, such as an analgesic in chronic pain. Methadone is a very effective pain medication. Due to its activity at the NMDA receptor, it may be more effective against neuropathic pain; for the same reason, tolerance to the analgesic effects may be lesser compared to other opioids. The increased usage comes as doctors search for an opioid drug that can be dosed less frequently than shorter-acting drugs like morphine or hydrocodone. Another factor in the increased usage is the low cost of methadone.
On 29 November 2006, the U.S. Food and Drug Administration issued a Public Health Advisory about methadone titled "Methadone Use for Pain Control May Result in Death and Life-Threatening Changes in Breathing and Heart Beat". The advisory went on to say that "the FDA has received reports of death and life-threatening side effects in patients taking methadone. These deaths and life-threatening side effects have occurred in patients newly starting methadone for pain control and in patients who have switched to methadone after being treated for pain with other strong narcotic pain relievers. Methadone can cause slow or shallow breathing and dangerous changes in heart beat that may not be felt by the patient." The advisory urged that physicians use caution when prescribing methadone to patients who are not used to the drug, and that patients take the drug exactly as directed.
I don’t have prescription-count data for methadone, but I suspect physicians became more squeamish about prescribing a painkiller with such a low margin for error. But no doubt stiffer warnings for patients had an effect on the death totals, too.
I don’t know why this is never explained in any of the news stories about the “opioid epidemic.” The numbers are supposedly rising despite a big drop in one of the three categories that define “prescription opioids.” And, more to the point, a simple cautioning of physicians and patients about the dangers seems to have brought down the death rate. This is a useful lesson in public health if there ever was one.
I don’t see this point mentioned often, but methadone deaths are very likely to be multi-drug poisonings, which is true of *every* category of drug poisoning in the CDC death data.
Other Synthetic Narcotics
This category includes fentanyl and other synthetic opioids, some of which are super-potent. As stated above, it is one of the three categories that are summed to make up the “prescription opioids” in overdose death tabulations. Fentanyl is something like 80 times as powerful as morphine. Notice its trendline. It grows by a factor of ~3 from 1999 to 2013, then it shoots up (bad pun) in 2014. As with the heroin story, this makes a lot of sense when you consider the recent rash of tainted heroin deaths. I actually suspect that many of the so-called heroin overdoses belong in this category, and it was only in 2014 that anyone caught on to this trend. Medical examiners don’t often know what kind of opioid was taken when there is a rash of overdoses; if they can determine that at all, it requires sending blood work to a lab that takes weeks to do a work-up (citation needed, but I'm speaking from memory based on a recent NPR story). So there could be some “heroin” deaths that are miscategorized, because the decedent actually (unwittingly) took fentanyl or something else in this category.
The musician Prince’s recent tragic death would probably be coded in this category; he died of a fentanyl overdose.
Psychostimulants with Abuse Potential
This is methamphetamine and its legal prescription analogues. That’s right, in official government tabulations, methamphetamine, Adderall, and Ritalin are recognized as being the same category of substance. I have minimal commentary on this, other than to note that the apparent recent increase in deaths is not matched by an increase in use in recent years. (See this again, page 9 figure 9 for stimulant use rates. They are basically flat for each age group.) So either the habit has become more deadly, or perhaps there is a spurious trend in the death totals.
Less Lethal Drug Categories
I spell this out in a previous post. There appears to be a growth in almost all categories of drug poisonings tracked by the CDC. Even for really obscure drugs that only kill a few people, the rise is similar in magnitude as for the opioids. Maybe some of this is other drugs "going along for the ride" with opioid deaths, as in someone took some cough syrup along with a fatal dose of prescription painkillers, so the coroner or medical examiner marked "antitussives" on the death certificate. But this can't be the whole story, because even when you filter out the seven most lethal drug categories (listed above) you still see a sharp rise in overdose deaths. As I say in my previous post, a likely explanation is that we're just testing dead bodies for more substances. It's not necessarily the case that more people are actually overdosing. So beware any news story that tells you the total number of drug overdoses and then spins a good yarn about the opioid epidemic. They are inflating their numbers to puff up their story, and papering over something that casts doubt on their narrative.
So Don't Be Fooled
So keep all this in mind when you read the next perfunctory piece on the “opioid epidemic.” The author of the piece is probably conflating several very different things and cramming them into a narrative that might or might not be true. He’s probably also omitting some important details.